# A grammar of Rapa Nui

Paulus Kieviet

Studies in Diversity Linguistics 12

#### Studies in Diversity Linguistics

#### Chief Editor: Martin Haspelmath

Consulting Editors: Fernando Zúñiga, Peter Arkadiev, Ruth Singer, Pilar Valen zuela

In this series:


# A grammar of Rapa Nui

Paulus Kieviet

Paulus Kieviet. 2017. *A grammar of Rapa Nui* (Studies in Diversity Linguistics 12). Berlin: Language Science Press.

This title can be downloaded at: http://langsci-press.org/catalog/book/124 © 2017, Paulus Kieviet Published under the Creative Commons Attribution 4.0 Licence (CC BY 4.0): http://creativecommons.org/licenses/by/4.0/ ISBN: 978-3-946234-75-3 (Digital) 978-3-946234-76-0 (Hardcover) 978-1-542544-79-5 (Softcover) ISSN: 2363-5568 DOI:10.5281/zenodo.235525

Cover and concept of design: Ulrike Harbort Typesetting: Paulus Kieviet, Sebastian Nordhoff Proofreading: Ahmet Özdemir, Andreas Hölzl, Annie Zaenen, Christian Döhler, Eitan Grossman, Elizabeth Zeitoun, Ezekiel Bolaji, Guylaine Brun-Trigaud, Hella Olbertz, Jean Nitzke, Julien Heurdier, Klara Kim, Matti Miestamo, Myke Brinkerhoff, Natsuko Nakagawa, Oliver Čulo, Prisca Jerono Fonts: Linux Libertine, Arimo, DejaVu Sans Mono Typesetting software: XƎLATEX

Language Science Press Habelschwerdter Allee 45 14195 Berlin, Germany langsci-press.org

Storage and cataloguing done by FU Berlin

Language Science Press has no responsibility for the persistence or accuracy of URLs for external or third-party Internet websites referred to in this publication, and does not guarantee that any content on such websites is, or will remain, accurate or appropriate.










## **Acknowledgments**

This grammar is a somewhat revised version of my PhD thesis, which I defended in July 2016. The PhD project formally started in November 2012, but the journey leading to this grammar has been much longer. In 2004, my wife Antje and I went to live in French Polynesia with our daughters Mattie and Nina, to assist language groups there with Bible translation work. After learning Tahitian, in 2005 I started to study Rapa Nui as well and became involved in checking the Rapa Nui translation of the New Testament. In 2007 we moved to Easter Island and ended up living there for three years. Among other things, I was involved in Bible translation, the edition of educational materials and the elaboration of a lexical database. In the course of time I started to collect observations on the grammar of Rapa Nui. Coming from French Polynesia, there was much of interest in a language so similar to Tahitian, yet so different in many respects.

This grammar would not have been possible without the help of many people. First of all I would like to thank Bob (Roberto) and Nancy Weber, who have devoted their lives to the Rapa Nui people and who have done a tremendous amount of work on vernacular education, Bible translation and linguistic research, as well as assisting the Rapa Nui community in anything having to do with the language. They were the ones who invited us to join them on Easter Island, made us feel welcome and helped us in many ways. Their observations, notes and suggestions helped me a great deal to learn to know the language. Over the years and decades, they have collected the texts which have served as corpus for this grammar.

I wish to thank the Rapa Nui translation team and various other people who welcomed us on the island and made their knowledge of the language available: María Eugenia Tuki Pakarati, long-time translator and linguistic worker; Alfredo Tuki Pakarati, who helped us through the visa application; Pai Hiti ꞌUira Rano Moai; María Virginia Haoa Cardinali; Victoria Hereveri Tuki; Virginia Atan Tuki.

A big thank you to my supervisors, Lourens de Vries and Ross Clark, who guided me through the process of writing a book that had to meet the requirements of a descriptive grammar and a dissertation; encouraging, correcting, fine-tuning where needed. Thanks also to the members of the reading committee for their willingness to review this dissertation: Aone van Engelenhoven, Ben Hermans, Marian Klamer, Pieter Muysken and Ger Reesink. A special thank you to Marian Klamer and Ger Reesink for their many helpful comments on earlier versions of the manuscript.

I wish to thank Kevin and Mary Salisbury, for their hospitality during my stay in Auckland; Nico Daams, Albert Davletshin, Mary Walworth, Kevin and Mary Salisbury, for our discussions about Polynesian languages; Stephen Levinsohn, for reviewing in detail my analysis of Rapa Nui narrative discourse; Steven Roger Fischer, for clearing up

many details of the transcription of Alfred Métraux' notebooks; Bob and Nancy Weber, Ross Clark, Andrew Pawley and Albert Davletshin, for supplying valuable linguistic resources; René van den Berg, for encouraging me to start this PhD project.

I am grateful to Martin Haspelmath and Sebastian Nordhoff at Language Science Press, who accepted this grammar for publication and carefully coached me through the process of revision and electronic conversion. They never failed to respond promptly to any questions I raised. Thanks also to the anonymous reviewers for their many helpful comments, and the proofreaders who took the time to get this grammar into shape.

Nico and Pam Daams were the people who invited us to join them in Bible translation work in Polynesia. Their friendship, help and encouragement over the past fifteen years have been truly invaluable.

Finally, I'd like to thank you, Antje, for your support and initiative all these years in so many visible and invisible ways. Without you this grammar would not have been written.

Paulus Kieviet *Alblasserdam, The Netherlands, January 2017*

## **List of abbreviations**

(Abbreviations for text sources are listed in Appendix B.)

### **Grammatical categories**


#### List of abbreviations


### **Language groups and protolanguages**


Map by Paulus Kieviet, licensed under the Creative Commons Attribution-Share Alike license

## **1 Introduction**

### **1.1 Rapa Nui: the island and the language**

### **1.1.1 The island and its name**

Rapa Nui, also known as Easter Island, is located at 27°05'S 109°20'W. The island is known for its giant statues (*mōai*), as well as for its extreme isolation: while the nearest islands (Sala y Gómez) are at a 400 km distance, the nearest inhabited island is tiny Pitcairn, 2100 km away. The closest population centres are Tahiti in French Polynesia (over 4200 km to the west) and Valparaíso on the Chilean coast (3700 km to the east).

The island forms a triangle, composed of three extinct volcanoes, with a surface of about 165 km<sup>2</sup> . The highest point is Mt. Terevaka (507m).

At the last census (2012), the island's population numbered 5,761.<sup>1</sup> Almost all inhabitants live in the town of Hanga Roa. Roughly half of the island's population is of Rapa Nui origin; other inhabitants include continental Chileans, as well as small numbers from other nationalities. Conversely, numerous ethnic Rapa Nui live in continental Chile, while there is also a Rapa Nui community of a few hundred people on Tahiti.<sup>2</sup>

The number of ethnic Rapa Nui does not coincide with the number of speakers of the Rapa Nui language. Wurm (2007) estimates the number of speakers at 2,400–2,500, but the actual number is probably lower. Makihara (2001b: 192) gives an estimate of 1,100 speakers, out of 1,800 ethnic Rapa Nui on the island; linguists Bob & Nancy Weber (p.c.) give a rough estimate of 1,000 speakers.

The name Rapa Nui, literally 'great Rapa', is used for the island, the people and the language.<sup>3</sup> It may have been coined in 1862, when Rapa Nui people came in contact with people from Rapa, the southernmost island of what is nowadays French Polynesia (Fischer 1993c: 64; Fischer 2005: 91); the latter is also called Rapa Iti, 'little Rapa'.<sup>4</sup>

<sup>1</sup> http://www.ine.cl/canales/chile\_estadistico/familias/demograficas\_vitales.php (accessed 27 October 2015); the projected population for 2016 is 6,600.

<sup>2</sup> It is extremely hard to estimate the total number of ethnic Rapa Nui. Estimates over the past years range from 2,600 to 7,748 (Bob Weber, p.c.). According to the 2012 census, 63.81% of the inhabitants of Rapa Nui (i.e. 3,676 out of 5,761) belong to an indigenous people group. According to the same census, 2,697 people on Rapa Nui, and 4,934 people in Chile as a whole, are able to conduct a conversation in Rapa Nui. Though these figures seem to be impossibly high, they may give an indication of the number of people on the island and on the Chilean mainland adhering to the Rapa Nui identity.

<sup>3</sup> The name is often spelled as a single word: Rapanui. In this grammar, the spelling Rapa Nui is used, in accordance with the accepted orthography (§1.4.4). The spelling sparked some debate in the *Rapa Nui Journal*: Fischer (1991; 1993b,a); Weber & Weber (1991).

<sup>4</sup> The meaning of the name Rapa itself is unknown, despite Caillot's assertion that there cannot be any doubt that it means "en dehors, à l'extérieur […] de l'autre côté" (outside, at the exterior, on the other side;

#### 1 Introduction

The island has been known by many other names (Fischer 1993c), all of them of postcontact origin. The name Easter Island and its corollaries in other European languages (Isla de Pascua, Osterinsel, Paaseiland, et cetera) dates back to 1722; it was given by the Dutch explorers who discovered the Island on Easter Sunday, April 5. No pre-contact name for the island or the people has been transmitted, and none may ever have existed.

### **1.1.2 Origins**

Linguistic, biological and archaeological data unambiguously indicate that the Rapa Nui people are Polynesians (Green 2000; Kirch 2000; Stefan, Cuny & Weisler 2002 and refs. there). In a certain sense, the early history of the island is uncomplicated. The island has a single language and there is no evidence that it was settled more than once. The date of settlement of the island is usually assumed to coincide with the date at which Rapa Nui split off from its protolanguage.

Even so, the prehistory of the Rapa Nui people is still surrounded by uncertainty, despite extensive archaeological, biological and linguistic research. The only virtually uncontested fact is, that the first settlers of the island came from somewhere in east Polynesia.<sup>5</sup> They probably arrived by a voyage of purposeful exploration rather than by chance (Bahn & Flenley 1992: 72–79; Kirch & Kahn 2007: 199). Some scholars suggest an origin from the Marquesas (cf. Bahn & Flenley 1992: 66), but the current consensus is that an origin from southeast Polynesia is much more likely, given the distance and prevailing winds. This means that the people who first discovered Rapa Nui probably arrived from Mangareva, Pitcairn and/or Henderson (Green 1998; Stefan, Cuny & Weisler 2002). Henderson Island, the closest habitable island to Rapa Nui (c. 1900 km), is deserted nowadays but was populated in the past, possibly as early as 700–900 AD (Weisler 1998; Green & Weisler 2002).

A more southern origin, from or through the Austral islands, has also been proposed (Langdon & Tryon 1983), but is generally rejected (Green 1985; 1998).

The date of initial settlement of the island is much debated. In the past, attempts were made to date the split-off of Rapa Nui from its protolanguage by means of glottochronology (using the amount of lexical change and an assumed rate of change), but these did not give satisfactory results: Emory (1963) obtained glottochronological dates as far apart as 1025 BC and 500 AD, and settled on an estimate of 500 AD on the basis of a single radiocarbon date provided by Heyerdahl & Ferdon (1961: 395).<sup>6</sup> Green (1967), Green (1985:

Caillot 1932: 69); the lexical sources for Rapa (Stokes 1955; Fischer 1996a [= James L. Green, 1864]; Kieviet & Kieviet 2006) do not list a lexeme *rapa*. In Rapa Nui there are two lexemes *rapa*: 1) 'to shine, be lustrous'; 2) 'ceremonial paddle'.

<sup>5</sup> Thor Heyerdahl's theory that the Rapa Nui came from South America, is commonly rejected (see Bahn & Flenley 1992 for an extensive critique), though Schuhmacher (1990) continues to explore the possibility of (secondary) influence of South-American languages on Rapa Nui. On the discussion about possible non-Polynesian elements in the Rapa Nui language, see Footnote 11 on p. 150.

<sup>6</sup> In general, Eastern Polynesian languages have changed vocabulary at a much higher rate than other Polynesian languages. Pawley (2009) calculates replacement rates of 0.67–2.0% per century for a number of non-EP languages, against 2.0–3.4% per century for EP languages (2.5% for Rapa Nui), based on retention of basic PPN vocabulary. This is explained by the "founder effect", i.e. rapid change in a small isolated speech community (Marck 2000: 138, Wilmshurst et al. 2011: 1818.

1.1 Rapa Nui: the island and the language

21), Emory (1972) and Kirch (1986) also give an estimate between 400 and 600 AD based on radiocarbon dates. Du Feu & Fischer (1993) and Fischer (1992) suggest a possible split between Rapa Nui and its relatives as early as the first century AD. Others give later dates: between 600 and 800 AD (Fischer 2005, based on a radiocarbon date of 690±130 given by Ayres 1971; Green & Weisler 2002), or between 800 and 1000 AD (Green 2000: 74; Spriggs & Anderson 1993; Martinsson-Wallin & Crockford 2001).

More recently, even later dates have been proposed. Re-examination of radiocarbon dates from Rapa Nui and other islands in east Polynesia, eliminating those samples not deemed reliable indicators for initial settlement, has led some scholars to date the onset of colonisation after 1200 AD (Hunt & Lipo 2006; 2007; Hunt 2007; Wilmshurst et al. 2011).

Others continue to propose dates late in the first millennium AD (Kirch & Kahn 2007; Mieth & Bork 2005; 2010).

The date of settlement of Rapa Nui is closely linked to the question of the colonisation of east Polynesia as a whole, an issue which is in turn linked to the relative chronology of the different archipelagos in east Polynesia. Here as well, a wide range of dates has been proposed. Settlement of east Polynesia started either in the Society Islands, with Tahiti at the centre (Emory 1963; Kirch 2000; Wilmshurst et al. 2011), in the Marquesas (Wilson 2012: 290, Green 1966), or in the Societies/Marquesas area as a whole (Kirch 1986: 9; Marck 2000: 138). According to Spriggs & Anderson (1993), there is solid archaeological evidence for human presence in the Marquesas from about 300–600 AD and in the Society Islands from 600–800 AD. Kirch (1986: 9) suggests that the Marquesas may have been peopled as early as 200 BC. On the other hand, Wilmshurst et al. (2011) date the initial settlement of the Societies as late as 1025–1120 AD, while all the other archipelagos in east Polynesia (including Rapa Nui) followed after 1190.

The relation between the Rapa Nui language and Eastern Polynesian is discussed in §1.2.2 below.

### **1.1.3 Snippets of history**

After its initial settlement, Rapa Nui may have maintained contact with other islands in east Polynesia, despite its geographical isolation (Clark 1983b: 424; Green 1998; 2000; Kirch & Kahn 2007).<sup>7</sup> At some point, there must even have been contact between at least one Polynesian island and South America, given the fact that the sweet potato and the bottle gourd spread from South America throughout Polynesia prior to European contact; Green (1998: 98) suggests that Rapa Nui people may have travelled to South America, returning either to Rapa Nui or to another island.

<sup>7</sup> Walworth (2015a) gives four words uniquely shared between Rapa Nui and Rapa. However, three of these (Rapa *matu* 'to advance', *kakona* 'sweet-smelling', *reka* 'happy') are also shared with other EP languages, and the fourth (*honi* 'peel') is a shared semantic innovation rather than a uniquely shared lexeme. Moreover, unique shared lexemes are not a strong indication of direct contact: Rapa Nui uniquely shares two words (*ua* 'war club', *ma ꞌa* 'to know') with Rennell in the Solomon Islands, even though direct contact between the two islands is very unlikely.

#### 1 Introduction

However, contact between Rapa Nui and other islands was probably very intermittent; Rapa Nui language and culture developed in relative isolation, an isolation which at some point became complete. This explains the high amount of lexical innovation noticed by Emory (1963), Langdon & Tryon (1983: 45) and Bergmann (1963: 36).

The history of Rapa Nui is described in Bahn & Flenley (1992), McCall (1994), Flenley & Bahn (2002) and Fischer (2005). Rapa Nui's prehistory is the tale of a society constructing hundreds of increasingly large stone statues (*mōai*) and transporting them to almost all corners of the island; a number of often feuding tribes whose names survive in legends; the gradual deforestation of an island once covered with giant palm trees;<sup>8</sup> and the 'birdman' cult, which involved an annual contest between young warriors for the season's first tern egg on one of the islets off the coast.

Rapa Nui entered written history on Easter Sunday, April 5, 1722, when it was sighted by a Dutch fleet of three ships, commanded by Jacob Roggeveen. Later in the 18th century, the island was visited by a Spanish expedition led by Don Felipe González in 1770, followed by James Cook in 1774 and Count La Pérouse in 1786. From the early 19th century on, many explorers, traders and whalers called at the island.

The repeated arrival of foreigners caused epidemic diseases, which in turn led to depopulation and a major socio-cultural upheaval. A greater trauma was yet to follow: in 1862–1863, ships raided the Pacific in search of cheap labour for mines, plantations and households in Peru. Several of these visited Rapa Nui and at least 1400 people were abducted or lured away and taken to Peru.<sup>9</sup> Most of them died of smallpox; when a few survivors were repatriated late 1863, they brought the disease with them. As a result, the population of Rapa Nui dropped even further. The events of 1863 were fatal for Rapa Nui culture, leading to the collapse of the structure of society and ultimately to the loss of old customs and traditions (Knorozov 1965: 391).

In 1870, the French trader/adventurer Dutrou-Bornier, acting for a Scottish company, managed to acquire title to most of the island and started to convert it into a giant sheep ranch. As the traditional power structure had collapsed, Dutrou-Bornier had free rein. When the situation for the remaining Rapa Nui seemed hopeless, bishop Tepano Jaussen of Tahiti formed the plan to evacuate all remaining inhabitants of the island; only the limited capacity of the vessel come to fetch them forced 230 people to stay on the island, while 275 left to settle in Mangareva and Tahiti. (In the 1880s, some of them returned, bringing with them Tahitian elements which were subsequently incorporated into the Rapa Nui language.) The number of people on Rapa Nui further decreased to 111 in 1877, after which it started to rise slowly again, doubling by 1897 and again by 1934.

In 1888, Rapa Nui was annexed by Chile. Even so, the island remained a sheep ranch under commercial control until 1953, when it passed under naval authority. During much of that time, islanders were not permitted to leave the island (presumably because of leprosy, an illness imported in the 1880s from Tahiti), so contact with the outside world was largely limited to the few foreign residents and visitors to the island.

<sup>8</sup> The causes of the deforestation of Rapa Nui (human or by rats?) and the question whether it led to a sociocultural collapse (traditionally dated around 1680) have been the subject of much debate, see Flenley & Bahn (2002); Diamond (2005); Mulrooney et al. (2007; 2009); Hunt (2007); Mieth & Bork (2010); Boersema (2011).

<sup>9</sup> In the course of these events, the name Rapa Nui may have emerged, see §1.1.2 above (Fischer 2005: 91).

1.2 Genetic affiliation

In 1966, Rapa Nui became a civil territory, a department (since 1974 a province) within the 5th region of Chile, consisting of a single municipality (*comuna*). The Rapa Nui people received Chilean citizenship. From 1960 on, Rapa Nui came out of its isolation. More and more Rapa Nui started to travel to the Chilean mainland for education and jobs; many of them settled there or emigrated to other countries. On the other hand, tourists and other visitors started to arrive in great numbers after the construction of the airport in 1967.<sup>10</sup> More jobs came available in the public sector (administration, education, health…), while the quickly expanding tourist industry also started to provide a host of job opportunities in hotels and guest houses, the building industry, retail and traditional crafts. As a result, over the past decades the island has experienced rapid economic development, but also a large influx of non-Rapa Nui residents (mainly from Chile). Tourism has continued to grow; currently the island attracts more than 40,000 people annually.

### **1.2 Genetic affiliation**

### **1.2.1 Rapa Nui in the Polynesian language family**

Rapa Nui is a member of the Austronesian language family; its complete classification according to the *Ethnologue* (Lewis, Simons & Fennig 2015) is as follows: Austronesian, Malayo-Polynesian, Central-Eastern Malayo-Polynesian, Eastern Malayo-Polynesian, Oceanic, Central-Eastern Oceanic, Remote Oceanic, Central Pacific, East Fijian-Polynesian, Polynesian, Nuclear, East. The language has no dialects.

Rapa Nui's immediate relatives are the other Polynesian languages,<sup>11</sup> which number around 35. These languages are spoken within a triangle delineated by New Zealand in the south-west, Hawaii in the north and Rapa Nui in the east; a number of Polynesian groups (known as Outliers) are located outside this area.

The basic subgrouping of the Polynesian languages was established in the 1960s. While earlier approaches used lexicostatistics and glottochronology to measure relative distance between languages (see e.g. Elbert 1953; Emory 1963), in the mid-1960s research started to focus on shared innovations: languages are likely to form a subgroup when they have a significant number of phonological, lexical and/or grammatical innovations in common. This resulted in a hypothesis which became the standard theory for Polynesian subgrouping (see Pawley 1966; Green 1966; Marck 2000), and which is represented in Figure 1.1 (based on Pawley 1966; Clark 1983b; Marck 2000). In this subgrouping, all but two languages belong to the Nuclear Polynesian (NP) branch. NP is divided in two branches: Samoic-Outlier (SO) and Eastern Polynesian (EP). Within EP, Rapa Nui forms a branch on its own, coordinate with Central-Eastern (CE) languages. CE in turn branches into Tahitic (TA) and Marquesic (MQ).<sup>12</sup>

<sup>10</sup> From 1970 on, Rapa Nui has been serviced by long-range jet airliners. As of October 2015, there were eight weekly flights to/from Santiago and one flight to/from Tahiti.

<sup>11</sup> See Krupa (1982) for a typological overview of Polynesian languages and Krupa (1973) for a history of research. More recent overviews are available for larger groupings: Lynch, Ross & Crowley (2002) for Oceanic, Blust (2013) for Austronesian.

<sup>12</sup> The evidence for EP and CE will be reviewed in §1.2.2 below.

Figure 1.1: Genetic classification of the Polynesian languages

1.2 Genetic affiliation

Though there is a wide consensus on the basic tenets of this subgrouping, various refinements and modifications have been proposed.<sup>13</sup> I will mention a few which directly or indirectly affect the position of Rapa Nui.

Firstly: within SO, there is evidence for a subgroup consisting of the Northern Outliers (NO), spoken in the northern Solomons (including the North Solomons province of Papua New Guinea) such as Takuu and Luangiua. A slightly larger group has also been suggested, consisting of the NO languages plus Kapingamarangi, Nukuoro and Tuvaluan ('Ellicean', see Howard 1981; Pawley 2009). Wilson (1985; 2012) discusses a number of innovations shared between the Northern Outliers and EP: a thorough restructuring and reduction of the pronominal system, as well as various other grammatical and lexical innovations. This leads him to suggest a NO-EP subgroup; in this hypothesis, the East Polynesians originated from the Northern Outliers, possibly migrating through the Ellice and Line Islands.

Secondly, Marck (1996a; 2000) proposes a few refinements within CE languages: nuclear Tahitic includes all Tahitic languages except New Zealand Māori; nuclear Marquesic includes Marquesan and Mangarevan, but not Hawaiian.<sup>14</sup>

Finally: more recently, the validity of Tahitic and Marquesic as clear-cut subgroups has been questioned. Walworth (2012) points out that the evidence for both subgroups is not very strong, something which has been recognised before (see e.g. Green 1966; Marck 1996a). Neither subgroup is characterised by regular sound changes or morphological innovations; the only evidence consists of lexical and semantic innovations (Green 1966) and sporadic sound changes (Marck 2000). Walworth suggests that there never was a Proto-Marquesic or a Proto-Tahitic language; rather, both branches may have developed through diffusion of features over certain geographical areas. She maintains the status of Proto-CE, which will be discussed in the next section.

### **1.2.2 Evidence for Eastern Polynesian and Central-Eastern Polynesian**

As Rapa Nui is the only language distinguishing EP from CE, it is worthwile to examine the evidence for both groupings. This evidence was collected by Pawley (1966) and Green (1985) and reviewed by Marck (1996a); while Pawley did not differentiate between EP and CE for lack of data on Rapa Nui, Green did take Rapa Nui into account, though on the basis of limited data. Marck (2000) provided further evidence on the basis of incidental sound changes. Here I will review the evidence adduced for both subgroups

<sup>13</sup> One proposal generally rejected is that by Langdon & Tryon (1983), who propose a Futunic subgroup including East Futunan, East Uvean, Rennell and Rapa Nui. The evidence for this subgroup is scant (see Clark 1983b; Green 1985).

<sup>14</sup> A different grouping is presented by Fischer (2001b), who proposes a subgroup on the basis of doublets in Mangarevan, such as *ꞌa ꞌine* 'woman' ~ *ve ꞌine* 'wife'. The first member of these doublets has not participated in the sound change *\*faf* > *\*vah*, which is common to all CE languages but does not occur in Rapa Nui (§1.2.2 below). According to Fischer, this constitutes evidence for a Proto-Southeastern Polynesian substratum, a subgroup which predates the differentiation of PCE into PTa and PMq, and which includes Rapa Nui. However, this analysis has been questioned: the doublets can also be explained as an incomplete phonological change (Rutter 2002), and even if they suggest a substratum in Mangarevan predating PCE, there is no evidence that this branch includes Rapa Nui (Marck 2002).

#### 1 Introduction

in the light of data and analysis for Rapa Nui presented in this grammar. In the list below, each proposed innovation is evaluated as valid (OK), invalid for the subgroup under consideration (**X**), or questionable (**⁇**).

For Eastern Polynesian, the following innovations have been suggested:

#### **Morphology**


**Sporadic sound changes** (Marck 2000: 131)


The following innovations are considered characteristic for Central-Eastern Polynesian:

<sup>15</sup> Lexical data for individual languages have mostly been taken from the lexical database Pollex (2009 version; Greenhill & Clark 2011).

1.2 Genetic affiliation

#### **Phonology**


#### **Morphology**


<sup>16</sup> Alternatively, *ānei* may reflect an earlier stage than PEP, as suggested by *anii* 'question marker' in Takuu (see Moyle 2011: 23).

#### 1 Introduction

#### **Syntax**

23. Loss of ergative traces. However, Rapa Nui is fully accusative (§8.4.2), so ergative traces may have been lost by PEP. **X**

#### **Lexicon & semantics**


#### **Sporadic sound changes** (Marck 2000: 41, 96–97)


<sup>17</sup> In fact, for any language X in family A, there will be a number of proto-A reconstructions for which there is no reflex in language X. This means that a subfamily B can be set up consisting of all languages of family A except language X, however implausible such a subgrouping may be on other grounds. For example, out of 710 EP+CE reconstructions, only 67 are represented in Rapa. On the basis of lexical data alone, one could thus propose a subgroup – let's call it North-Eastern Polynesian – consisting of EP minus Rapa, with no less than 643 reconstructions, while EP itself would be represented by only 67 reconstructions. 'NEP' would thus seem to be even more strongly motivated than CE. Even so, no one has ever proposed such a grouping. The small number of Rapa reflexes can be explained by a small vocabulary (i.e. widespread loss) and lack of data.

For both Rapa and Rapa Nui – and in fact for all EP languages – the total number of reflexes in EP and CE reconstructions is roughly in proportion to the total number of reflexes in Pollex as a whole.

A lexeme occurring in a branch of languages is likely to be an innovation of that branch if it can be shown to replace a lexeme with the same meaning occurring in a higher-order branch.

1.2 Genetic affiliation

#### To summarise:


We may conclude that both subgroups are reasonably well established, though on re-examination the evidence for CE is considerably weaker than has been suggested so far. This provides at least a partial solution to the challenge posed by newer theories of settlement, according to which eastern Polynesia was colonised late and rapidly (§1.1.2 above). In these scenarios, there is not much time for EP and CE to develop in isolation, so a small number of innovations for both groups is expected.

The evidence still suggests that there is a CE subgroup within EP. However, the small number of innovations and a possibly shorter chronology call into question the identity of PCE: was there ever a community speaking PCE? In other words, did all the CE innovations occur in a unified language, before subgroups (TA and MQ) and individual languages started to diverge? Or did these innovations spread over the PCE area through contact, possibly after the protolanguage had started to diverge into different dialects? Walworth (2012) proposes that innovations in Tahitic and Marquesic were not part of a unified protolanguage but spread by diffusion through different speech communities. The data above suggest that the same is true for PCE.

This also means that the first colonisers of Rapa Nui did not necessarily leave an EP homeland where PEP was spoken as a unified language. If Rapa Nui was settled from southeast Polynesia, as is the growing consensus (§1.1.2), it is conceivable that the language spoken in that area, at the time Rapa Nui split off, was already starting to differentiate from PEP towards a proto-Marquesic speech variety. This possibility is suggested by the fact that Rapa Nui shares considerably more lexemes with Marquesic than with Tahitic (Emory 1963: 94; Langdon & Tryon 1983: 42–44; Clark 1983b: 424). This scenario is not in contradiction with the standard theory (according to which Marquesic and Tahitic languages together form the CE branch): it is altogether likely that speech communities within Eastern Polynesia, especially those relatively close together such as the Societies, the Tuamotus, Marquesas and Mangareva, remained in close contact, which facilitated the diffusion of subsequent "CE" innovations. In other words, CE innovations did not necessarily predate the onset of differentiation between Tahitic and Marquesic.

#### 1 Introduction

### **1.3 The Rapa Nui language: typology and innovations**

### **1.3.1 General typology**

Rapa Nui is characterised by the following typological features, most of which are shared by the Polynesian languages in general:


### **1.3.2 Innovations and losses in Rapa Nui**

In the course of history, a number of developments took place in Rapa Nui which did not take place in PCE (though they may have taken place independently in daughter languages). In this section, only phonological and grammatical changes are listed; lexical changes are not included.

#### 1.3 The Rapa Nui language: typology and innovations


#### 1 Introduction


1.4 Sociolinguistic situation

In recent times, the following developments took place:


### **1.4 Sociolinguistic situation**

Rapa Nui has undergone profound influence from two major sources: Tahitian and Spanish.

### **1.4.1 Influence from Tahitian**

Tahitian started to exert its influence in the 1880s, when Rapa Nui speakers who had migrated to Tahiti in the 1870s started to remigrate (§1.1.3 above). After 1889, contacts between Rapa Nui and Tahiti were scarce (Fischer 2005: 141); they slowly resumed in the mid-20th century. To this day, a few hundred Rapa Nui live on Tahiti, and a weekly flight enables regular contact between the two islands.

The influence of Tahitian on modern Rapa Nui is striking. In my lexical database, which contains 5,833 lexical entries, 543 items are marked as (probably) of Tahitian origin, and another 89 as possibly Tahitian. Many of these can be distinguished phonologically, as the Tahitian and Rapa Nui consonant inventories are different, especially in the distribution of the glottal plosive (§2.2.1). Others can be recognised because of their semantics and/or recent introduction (see e.g. the discussion about *riro* 'to become'

#### 1 Introduction

in §9.6.2). Tahitian vocabulary includes a number of very common words, such as *ꞌite* 'to know', *haꞌamata* 'to begin', *ꞌī* 'full', *hāpī* 'to learn', *māuruuru* 'thank you' and the everyday greeting *ꞌiorana*.

One reason why Tahitian elements are easily adopted into the language, is their 'vernacular feel'. Tahitian words match the Rapa Nui phoneme inventory and word-forming constraints, with a few exceptions (§2.5.3.2). As a result, Tahitian borrowings are not perceived as intrusions; unlike Spanish borrowings, they are not avoided in written language and formal styles.

On historical grounds it seems plausible to date the intrusion of Tahitian elements to the 1880s (cf. Fischer 2001a: 315), when Rapa Nui remigrated from Tahiti. This remigration happened at the time when the population was at an all-time low, a situation conducive to rapid language change. Moreover, in the same period Tahitian catechists came to Rapa Nui, as well as foremen and labourers for the sheep ranch (Di Castri 1999: 101). According to Métraux (1971: 32), by 1935 many Tahitian words had entered the language; already in 1912, Knoche (1912: 65) noticed that Tahitian had exercised "einen grossen Einfluss auf Sitten und Sprache der Insulaner" (a large influence on the customs and speech of the islanders).

However, when we look at Rapa Nui texts from the 1910s–1930s (§1.6.2), the scarcity of Tahitian influence is striking, in comparison to modern Rapa Nui. The Tahitian numerals (§4.3.1) are not used, except the occasional *vaꞌu* 'eight' (though the original *varu* is much more common). The Tahitian quantifiers *taꞌatoꞌa* and *paurō* 'all' (§4.4.2, §4.4.3) do not occur either. Certain Tahitian words are commonly used in older texts (*ravaꞌa* 'to obtain', *ꞌī* 'full', *manaꞌu* 'think', *ꞌaꞌamu* 'story'), but many words common nowadays occur rarely or not at all in older texts: *ꞌite* 'to know', *riro* 'to become', *ꞌonotau* 'epoch', *haꞌamata* 'to begin', *māere* 'to be surprised', *māhatu* 'heart', *māuruuru* 'to thank; thanks', *nehenehe* 'beautiful', *ꞌe* 'and', *nuꞌu* 'people', and so on.<sup>18</sup>

This suggests that many Tahitian words common nowadays only came into use after the 1930s. The Tahitian influence noticed by Knoche and Métraux must have been less pervasive than it is today. An alternative explanation would be, that the language of the older texts is archaic and reflects a variety which was current before 1880, possibly through verbatim transmission of old legends; after all, many of these texts represent old traditions. This is not very likely, however: it would leave unexplained why certain Tahitian words are very common, while many others – equally common nowadays – do not occur at all. Neither would it explain why roughly the same picture emerges from all corpora of older texts (Egt, Ley, Mtx and MsE), including a long text which tells of post-1880 events (Ley-9-63, memories of catechist Nicolás Pakarati, recounted by his widow).<sup>19</sup> It is hard to conceive that scores of words borrowed 50 or 60 years previously would have been completely avoided in traditional stories, while others were freely used. Rather, the picture that emerges is one of two waves of Tahitian intrusions: one in the 1870s and 1880s, followed by a much bigger one after 1960, when intensive contacts between Rapa Nui and the outside world (including Tahiti) were established.

<sup>18</sup> Of these words, only *ꞌite* is found in Englert's dictionary (first published in 1948). Notice, however, that Englert does not include words known to be of recent origin.

<sup>19</sup> Only for a few words do the corpora differ mutually: *ꞌati* 'problem' occurs in Mtx and Ley, but not in MsE.

1.4 Sociolinguistic situation

### **1.4.2 Influence from Spanish**

The Spanish influence on modern Rapa Nui is likewise massive. This influence is not noticeable in the older texts, even though Rapa Nui had been a Chilean territory for almost 50 years by the time these texts were collected. Spanish influence only started to make itself felt from the 1960s on, when Rapa Nui speakers acquired Chilean citizenship, began to participate actively in government and politics, acquired jobs for which proficiency in Spanish was a prerequisite, and increasingly took part in secondary and tertiary education. Spanish is also the language of the media, the predominant language of the Roman Catholic church, and the language of the many Chileans from the mainland who moved to the island (ultimately resulting in a high proportion of intermarriage). All of this led to a gradual incursion of Spanish elements into the language.

My lexical database contains 201 lexemes of Spanish origin, but this only represents words well entrenched in the language, often with adaptation to Rapa Nui phonology (§2.5.3.1). In everyday speech, the number of Spanish words is much higher. Most of these are not considered as part of the Rapa Nui lexicon but as foreign intrusions, i.e. as instances of code mixing.

Code mixing is extremely common in modern Rapa Nui speech, involving single words, phrases, sentences or longer stretches of speech; see Makihara (2001a,b; 2004; 2007; 2009) for examples and discussion. In most modern Rapa Nui texts in my corpus, the amount of code mixing is considerably lower than in Makihara's examples. This can be explained by the fact that a large part of my corpus consists of text types for which the use of Spanish is considered less acceptable: (a) traditional stories; (b) written texts; (c) edited spoken texts.<sup>20</sup> Moreover, traditional stories make less reference to modern institutions and artifacts, so there is less need for the use of Spanish elements.

Makihara (1998; 2009) signals a growing trend of purism, in which people attempt to speak Rapa Nui free of Spanish influence. This happens especially in political discourse, but is spreading to other domains.

The extent of Tahitian influence has led Fischer (1996c: 47) to characterise modern Rapa Nui as a "Rapanui-Tahitian hybrid", a product of "language intertwining" (Fischer 2008a: 151). However, while the lexicon of modern Rapa Nui is heavily influenced by Tahitian, the grammar has not been affected to the same degree, as the following chapters will make clear (cf. Makihara 2001b: 194). Even in areas where massive replacement by Tahitian terms has taken place, e.g. quantifiers (§4.4.11) and numerals (§4.3.1), these terms have been reinterpreted into a "native" Rapa Nui syntax.

The same is true for Spanish. Spanish has certainly influenced the grammar of Rapa Nui, but Spanish borrowings have been integrated into Rapa Nui grammar without transfer of their syntactic features. For example, the Spanish noun *kampō* 'countryside' (< *campo*) became a locational (§3.6.3.3); *kā* 'each' (< *cada*) became a quantifier compatible with plurality (§4.4.8.2). The modal verbs *puē* 'can' and *tiene que* 'must' were borrowed (§11.3.6), but the third person singular of these verbs is used with all persons and numbers,

<sup>20</sup> Another reason for the discrepancy may be that much of the corpus is slightly older (1977–1990) than Makihara's data (after 1990). However, relatively high amounts of code mixing are found in some of the oldest (informal) texts in the corpus.

#### 1 Introduction

and they are used with Rapa Nui syntactic features like *mo*-complements. On the other hand, certain Spanish semantic and syntactic features have become common without borrowing of the lexical items: *kē* 'several' (§4.4.8.1), copular verbs (§9.6), the coordinating conjunction *ꞌe* 'and' (§11.2.1), the construction *oho mo* 'to be about to' (§11.3.2.4), et cetera. These elements have affected Rapa Nui grammar to a certain degree, but the same cannot be said of the numerous Spanish words and phrases interspersed in everyday speech. The fact that Spanish intrusions are avoided in certain types of discourse, confirms that these are instances of code switching and belong to the domain of language use (*parole*), without having profound effect on the linguistic system (*langue*) of modern Rapa Nui (cf. Makihara 2001b: 193).

### **1.4.3 Language use and vitality**

As indicated in §1.2.1 above, Rapa Nui does not have dialects. On the other hand, there is considerable idiolectal variation between the speech varieties of individual speakers and of different families, e.g. in the use of certain lexical items and the degree of Tahitianisation (cf. Fischer 2008a: 154).

While Rapa Nui grammar has retained its distinctive character and has not become a Rapa Nui-Tahitian and/or Rapa Nui-Spanish mix, the language is certainly endangered. The factors mentioned above which led to Spanish influence on the language (participation in Chilean civil life, education, jobs, immigration of mainland Chileans, intermarriage) also led to a gradual increase in the use of Spanish by Rapa Nui people. From the 1960s on, Rapa Nui people started to aspire to "being Chilean" (Fischer 2001a: 315), something for which proficiency in Spanish was essential. As a result, it became common for Rapa Nui people to use Spanish, initially in interaction with mainland Chileans, but then also between each other, both in public and at home. From the 1980s on, this meant that many children – even those from two Rapa Nui parents – learned Spanish as their first language. Weber & Weber (1990) found that the number of primary school children who were fully proficient in Rapa Nui (either as first language or by being bilingual) had decreased from 77% in 1977 to 25% in 1989. This can only partly be explained from an increased proportion of children from continental or mixed households. In 1997, a production/comprehension test among primary school children living on the island showed that only 49 out of 558 children (9%) were fully bilingual; an additional 80 (14%) had a reasonable level of comprehension and production in Rapa Nui (a score of 4 or higher on a scale of 0–7); 329 (59%) had virtually no proficiency at all (Weber & Weber 1998).

This trend did not go unnoticed. Various measures were taken to enhance the chances of survival of the language, many of these initiated or assisted by the *Programa Lengua Rapa Nui*. One of these was the institution of an immersion program in the local primary school, extending from kinder until year 4. This program has achieved a varying degree of success (Makihara 2009). Other initiatives include the publication of two series of textbooks (Weber, Weber & team 1990a,b) and other educational materials, the foundation of a language academy (*Academia de la lengua*) and an annual Language Day (*Día de la lengua*). At the same time, the use of Rapa Nui in public domains increased, e.g. in politics (Makihara 2001b: 204).

1.4 Sociolinguistic situation

In 2011, a new survey was conducted using the same criteria for comprehension and production as in 1997 (Calderón Haoa et al. 2011). In this survey, the same persons included in the 1997 survey were interviewed again (as far as they could be traced), as well as young people in the age 5–19. The results were as follows: out of 1338 interviewees, 138 (10.3%) were fully bilingual; another 235 (17.6%) had a score of 4–7 in comprehension and production; 721 (53.9%) had virtually no proficiency. This means that proficiency in Rapa Nui had somewhat increased since 1997, despite the fact that the proportion of children from a non-Rapa Nui background was higher than in 1997.

Ultimately, the survival of Rapa Nui will depend on whether speakers succeed in passing the language on to the next generation.

### **1.4.4 Orthography**

Even though Rapa Nui has a small phoneme inventory (§2.2), in three areas an orthographical choice needs to be made between various alternatives: the velar nasal /ŋ/, the glottal plosive /ʔ/ and vowel length.

In old word lists and lexicons, such as Roussel (1908), neither the glottal plosive nor vowel length is marked. In later sources, if the glottal plosive is marked, it is usually written as an apostrophe, either straight (ꞌ) or curled (' or '); a few sources (Fuentes 1960; Salas 1973) use the IPA glottal or a similar symbol (*ʔ ? ˀ* ).

Vowel length is represented in various ways: *aa* (Fuentes 1960; Salas 1973), *â* (Englert 1978, Conte Oliveros 1996), *á* (Du Feu 1996), or *ā* (Blixen 1972; Chapin 1978).

The velar nasal has been represented as *ng* (Métraux 1971 [1940]; Blixen 1972; Conte Oliveros 1996) or *g* (Roussel 1908; Chapin 1978). Engert was the first to use the *ŋ* symbol, a practice adopted by Fuentes (1960), Salas (1973) and Du Feu (1996).

In the *Programa Lengua Rapa Nui* (PLRN, see §1.6.2 below), the following choices were made:


These choices are presented and discussed by Weber & Weber (1985); Weber & Weber (2005).

<sup>21</sup> To prevent word processors from turning ꞌ into curly brackets (' or '), which take up more space and disrupt the visual unity of the word, a special font was used in the past containing a symbol ꞌ. More recently, the development of Unicode has obviated the need for a special font; the code point UA78C ('Latin small letter saltillo') is now available for a symbol ꞌ which is not confused with an apostrophe by word processors.

#### 1 Introduction

Another issue concerns word boundaries: should the causative marker *haka* be connected to the root (*hakaoho* 'to cause to go') or be treated as a separate word (*haka oho*)? The same question applies to nominalisers like *iŋa*: *vānaŋaiŋa* or *vānaŋa iŋa* 'speaking'? In most Polynesian languages, these elements are connected to the root, but in the PLRN orthography of Rapa Nui, they are written as separate words.<sup>22</sup>

Other grammatical elements are written as separate words as well: determiners, the proper article *a*, prepositions, aspect markers, et cetera. The same is true for phrasal proper nouns, hence *Rapa Nui*, not *Rapanui*; *Haŋa Roa* (town); *Te Moko ꞌa Raŋi Roa* (protagonist of a legend). On the other hand, certain lexical compounds are written as a single word (§5.7.2; Weber & Weber 1985: 27).

One more choice which differs from the current practice in most Polynesian languages concerns the orthography of reduplications. In most languages, these are connected to the root; in Rapa Nui, they are separated from the root by a hyphen: *riva-riva* 'good', *tē-tere* 'to run (Pl)', *vānaŋa-naŋa* 'to talk repeatedly'. This applies even to lexical reduplications, for which the base does not occur independently in Rapa Nui: *nao-nao* 'mosquito', *ꞌā-ꞌanu* 'to spit' (§2.6.3).

Over the years, the PLRN orthography has gained acceptance among the Rapa Nui community, including teachers and members of the Rapa Nui Language Academy. It is increasingly seen in publications (e.g. Gleisner & Montt 2014). In this grammar the same orthography is used, with two exceptions:


### **1.5 Previous work on the language**

### **1.5.1 Lexicon**

A good number of early visitors to the island gathered a short word list of the language. The first of these was compiled by Don Francisco Antonio de Agüera during the Span-

<sup>22</sup> This not only serves to avoid long words like *hakamāramarama* 'to cause to be intelligent' but also prevents potential spacing conflicts: both *haka* and *iŋa* may be separated from the root by certain particles or adverbs (see (52) on p. 98; (100) on p. 342).

<sup>23</sup> See http://grammar.ucsd.edu/courses/lign120/leipziggloss.pdf. Lehmann (2004) notices that there is no satisfactory solution for hyphens that do not correspond to morpheme breaks, as in *vis-à-vis*.

1.5 Previous work on the language

ish expedition in 1770 (Ross 1937; Corney 1908), followed by the German botanist Johann Forster, part of Cook's expedition in 1774 (Schuhmacher 1977). Father Hippolyte Roussel, who stayed on the island in the late 1860s, compiled a dictionary which was published posthumously.<sup>24</sup> It contains almost 6,000 Spanish lemmas with a total of about 1,800 unique Rapa Nui words; unfortunately it is heavily contaminated by Mangarevan and Tahitian vocabulary (Fischer 1992) and therefore far from reliable. Other early vocabularies include Philippi (1873), Geiseler (1883) (see also Ayres & Ayres 1995), Thomson (1889), Spanish translation Thomson (1980), Cooke (1899) and the short dictionary by Martínez (1913). The extensive vocabulary in Churchill (1912) is based on Roussel's dictionary and some of the other vocabularies.

Father Sebastian Englert, who served on the island as parish priest from 1935 until his death in 1969, was the first person to study the language in depth. His dictionary (published in Englert 1948 and revised in Englert 1978) is an invaluable resource for the language as it was spoken in the first half of the 20th century. Another extensive dictionary is Fuentes (1960). Recent dictionaries include Fedorova (1988), Conte Oliveros (2000), Hernández Sallés et al. (2001) and Hotus Chavez (2008). Publications on specific lexical domains include Gunckel (1968) and Rauch, Ibañez & Ramirez (1996) on flora, Pinochet Carte (1980) on mollusks, Randall & Cea Egaña (1984) on fish, and Bierbach & Cain (1996) on religion.

Over the past years, a number of phrase books for the wider public have been published: Haoa Rapahango & Liller (1996), Hotus Tuki (2001) and Pauly & Atán (2008).

### **1.5.2 Grammar and sociolinguistics**

The first grammar of Rapa Nui was written by Father Sebastian Englert (included in Englert 1948, revised version 1978). It is relatively short but remarkably accurate. Other grammar sketches and concise grammars include Fuentes (1960), Chapin (1978), Munro (1978), Fedorova (1988) (Russian), Conte Oliveros (1996) and Rubino (1998). The latter is a reordering of material from Du Feu (1996).

The most extensive grammar is Du Feu (1996), published in the Descriptive Grammars series.<sup>25</sup>

Several theses, articles and unpublished papers have been written on specific aspects of the language.

The phonology of Rapa Nui is described in Du Feu (1985), Guerra Eissmann et al. (1993), Salas (1973) and Weber & Weber (1982). An important landmark in Rapa Nui linguistics was the discovery that Rapa Nui preserves the PPN glottal plosive, a phoneme which has disappeared in all other EP languages. The glottal plosive was largely ignored in early

<sup>24</sup> The French original was published in Roussel (1908), a Spanish translation in Roussel (1917); the latter was republished in Foerster (2013), with a critical introduction by Bob Weber.

<sup>25</sup> This grammar suffers from some serious flaws, as pointed out in reviews by Mosel (1997) and Weber & Weber (1999). It follows the Descriptive Grammars questionnaire closely rather than presenting material in categories relevant to the language. Moreover, the analysis presented is often unclear, incomplete or incorrect. Some of the examples adduced are unnatural or even incorrect, while the glosses are often inadequate.

#### 1 Introduction

descriptions, though Englert's dictionary registers it in many words. Its phonemic status was brought to light by Ward (1961; 1964) and Blixen (1972).

The noun phrase is described in Du Feu (1987) in broad outline. Another paper on the noun phrase is Gordon (1977).

The verb phrase is discussed by Weber (1988b) (Spanish version Weber 2003b), who offers a thorough analysis of aspect marking. Papers by Fuller (1980) and Wittenstein (1978) deal with the directional markers *mai* and *atu*. Chapin (1974) analyses the use of the postverbal particle *ai*, which is difficult to define in Rapa Nui.

Grammatical relations in Rapa Nui have been the subject of several studies, especially Agent marking. The supposedly ergative traits of the case system have drawn the attention of several linguists<sup>26</sup> (Alexander 1981a,b; 1982; Finney & Alexander 1998; Finney 2000; 2001). Weber (1988b) (Spanish version Weber 2003b) argues against an ergative analysis.

Other grammatical topics include the following: modality (Du Feu 1994); interrogatives (Du Feu 1995); possession (Mulloy & Rapu 1977); reduplication (Johnston 1978), nominalisation (McAdams 1980), relative clauses (Silva-Corvalán 1978), sentence structure (Smith 1980), and negation (Stenson 1981).

Sociolinguistic aspects (language use and vitality) are discussed by Weber & Weber (1984); Weber & Weber (1990; 1998), Gómez Macker (1977; 1979) and Haoa Cardinali (2012). Makihara (1998; 1999; 2001a,b; 2004; 2007; 2009) has studied the use of Rapa Nui and Spanish in spoken language. Other studies on the influence of Spanish on modern Rapa Nui are included in Stolz, Bakker & Salas Palomo (2008).

### **1.6 About this grammar**

### **1.6.1 A corpus-based study**

This grammar is based on the analysis of a large corpus of Rapa Nui texts, in addition to observations and discussion/elicitation sessions during the time when I lived on Easter Island (November 2007 – December 2010). In addition to grammatical research, I developed a comprehensive lexical database (hitherto unpublished) based on all available lexical sources and text materials (2008–2010), and carried out an exegetical check of the Rapa Nui translation of the New Testament (2005–2013). The lexical database has served as an additional resource for this grammar, providing data for example in the area of the relation between nouns and verbs (Chapter 3).

A corpus-based approach has several advantages (cf. McEnery & Wilson 1996: 12): it is based on actual, natural data, which are not biased by the linguist's interest; a large corpus includes data from a wide range of speakers; it enables discourse analysis; the data are verifiable; and finally, a large corpus allows statistical analysis. Moreover, the corpus used for this grammar allows diachronic analysis (see below). Two possible disadvantage of corpus-based research are, that less common phenomena are harder to analyse, as they

<sup>26</sup> According to Mosel (1997: 182), "The most striking feature of Rapanui is that it shows traces of ergativity and hence similarities with West Polynesian languages."

1.6 About this grammar

are rare in texts (Chapin 1978), and that the corpus only shows what is possible, not what is impossible (Biggs 1974: 412). These problems were overcome to a certain degree (a) by using a large corpus (over 500,000 words), and (b) by supplementing corpus analysis with personal observations and elicitation/discussion sessions with a speaker of the language. The corpus is described in §1.6.2 below.

All texts in the corpus were digitised and converted to the accepted Rapa Nui orthography (§1.4.4), with consistent marking of glottal plosives and vowel length. The corpus has been formatted as a Toolbox database, which is linked to the lexical database mentioned above.

The analysis in the following chapters is based on the corpus as a whole. For certain topics (especially aspect marking and clause structure & case marking), a subcorpus of 29 texts was analysed in more detail (c. 58,000 words; see Footnote 3 on p. 314, Footnote 2 on p. 376).

This grammar also has a comparative component: for many grammatical elements and constructions, the historical derivation and occurrence in related languages is discussed, mostly in footnotes. Comparative data are mainly taken from languages for which a thorough description is available. Data from Eastern Polynesian languages (Tahitian, Māori, Hawaiian, etc.) are of primary importance; sometimes, reference is made to non-EP languages (Samoan, Tongan, Tuvaluan, etc.).

Finally, this grammar has a diachronic dimension. The corpus includes texts from the past 90 or 100 years, a period during which the language has changed considerably; this offers a certain historical perspective which has been taken into account in the analysis.

This grammar is written within the tradition of "basic linguistic theory", the approach which has become common in descriptive linguistics and which eclectically employs concepts from both traditional linguistics and various theoretical frameworks (Dryer 2001; 2006; Dixon 2010a,b; 2012).

### **1.6.2 The corpus**

The corpus used as data for this grammar contains two subcorpora: older texts (c. 1910– 1940, 124,500 words) and newer texts (c. 1977–2010, 399,000 words). In addition, there is a small collection of texts from the early 1970s (14,500 words). This section gives a description of the different parts of the corpus. The texts in the corpus are referenced with three-letter abbreviations in this grammar; a full listing is given in Appendix B. In this grammar, the term *older Rapa Nui* is used for features only found in pre-1940 texts; features only occurring after 1970 are labelled *modern Rapa Nui*. These labels are used for convenience, without implying that the pre-1940 texts reflect the pre-contact language sometimes referred to "Old Rapa Nui".

The corpus contains a wide variety of texts. Narrative texts – both spoken and written – are the largest category. Other genres include speeches, conversations, radio interviews, poetry, newspaper articles, procedural texts (e.g. descriptions of traditional customs and techniques) and expository texts (e.g. episodes of the history of the island). The sources are as follows:

#### 1 Introduction


<sup>27</sup> MsE is one of six manuscripts (labeled A–F) discovered during the Norwegian archeological expedition in 1955; see Barthel (1965), Horley & Labbé (2014). MsE is by far the most extensive of the six; the others mainly contain lists and fragmentary material. Barthel (1978: 298) considers MsE as a copy of an original written before 1914. Recently, a set of photographs of a hitherto unpublished manuscript were discovered; the ms. was written in the same hand as MsE and is now labelled Manuscript H (Horley & Labbé 2014; 2015).

<sup>28</sup> Despite the late date of publishing, most – possibly all – of these texts were collected in the 1930s. Many were published (sometimes with minor variations) in Englert (1939a,b,c); all of these were written in 1936. Of the stories not included in these publications, the majority were transmitted by the same narrators mentioned in Englert (1939a,b,c): Mateo Veriveri, Juan Tepano and Arturo Teao. Other stories were told by the wife and sons of the catechist Nicolás Ure Potahi (1851-1927).

Many of the texts in Englert (1939b) are not included in Englert (1980); these are not included in the corpus, as I only discovered this publication in November 2015.

<sup>29</sup> According to Fischer (2008b; 2009), the original notebooks were lost, though a photocopy was preserved; Davletshin (p.c. 2016) pointed out to me that the originals still exist, and are in the Thomas Barthel Nachlass in Tübingen.

#### 1.6 About this grammar

6. Finally, the largest single text in Rapa Nui is the translation of the New Testament, as well as portions of the Old Testament. This translation (as yet unpublished) was made by a number of Rapa Nui speakers, with exegetical and linguistic advice from Robert & Nancy Weber. In 2006–2012, the New Testament was meticulously checked for naturalness by a team of Rapa Nui speakers. In this grammar, the Bible translation is used as a secondary resource, especially to illustrate phenomena for which few or no clear examples are available otherwise.

Not included in the corpus are a number of other Rapa Nui texts:


### **1.6.3 Organisation of this grammar**

This grammar is organised as follows.

Chapter 2 deals with the phonology of Rapa Nui. The following topics are discussed in turn: phonemes (with special attention to the glottal plosive), syllable and word structure, stress, intonation, phonological processes, and reduplication.

Chapter 3 deals with nouns and verbs. In many analyses of Polynesian language, the existence of lexical nouns and verbs is denied; rather, the two categories are defined syntactically ("a noun is any word preceded by a determiner"). Arguments are given to show that this approach obscures various differences between nouns and verbs, and that the distinction between both should be maintained. A classification of nouns is proposed, as well as a classification of verbs. Adjectives (a subclass of verbs) and locatives (a subclass of nouns) are discussed.

<sup>30</sup> Several scholars have suggested that *rongorongo* was developed after the Rapa Nui witnessed writing in 1770, when Spanish explorers drew up a deed of cession in which the island was handed over to the Spanish crown (Emory 1972; Fischer 1996b; 1997).

#### 1 Introduction

Other word classes are discussed in Chapter 4: pronouns, numerals, quantifiers, adverbs, demonstratives and prepositions. Not treated in this chapter are words exclusively occurring as particles in the noun and/or verb phrase, such as determiners and aspect markers.

The noun phrase potentially contains a large number of elements; these are discussed in Chapter 5. Two determiners are discussed extensively: the article *te* (which marks referentiality, not definiteness or specificity) and the predicate marker *he*. Possessive relationships are also discussed in this chapter; possessives can be marked with *o* or *a*, depending on the relationship between possessor and possessee. Another area discussed here is compounding, and the difference between compounding and modification.

Chapter 6 deals with possessive constructions. Possessors occur as modifiers in the noun phrase, as predicates of nominal clauses, and in various other constructions. A common feature in Polynesian is the distinction between *o*- and *a-*marked possessors; this is discussed in detail.

Chapter 7 discusses the verb phrase. A major topic is the use of aspect markers, a set of five preverbal particles. Other common verb phrase particles include directionals and postverbal demonstratives. Finally, a section is devoted to serial verbs, a construction not found in other Polynesian languages.

Some Polynesian languages are accusative, others are (partly) ergative; at first sight, Rapa Nui does not seem to fit either pattern. In Chapter 8 on the verbal clause, I show that Rapa Nui is accusative, and that case marking of Agent and Patient is governed by an interplay of syntactic, semantic and pragmatic factors. A passive construction is shown to exist, even though it is less obvious than in related languages. Other topics in this chapter include non-canonical case marking, constituent order, comitative constructions and causatives.

Nonverbal clauses are common in Rapa Nui; these are discussed in Chapter 9. Two major types are classifying and identifying clauses, respectively. Existential clauses can be verbal or non-verbal. The chapter closes with an unusual feature in Rapa Nui (compared to other Polynesian languages): the emergence of copula verbs in classifying clauses.

Chapter 10 deals with mood (imperatives, interrogatives, exclamatives) and negation. Constructions involving multiple clauses are discussed in Chapter 11: coordination, relative clauses, clausal complements and adverbial clauses.

Appendix A provides illustrative interlinear texts. Appendix B lists the texts in the corpus used as data for this grammar.

This grammar does not contain a separate section on discourse issues. Discoursebased analysis has been applied to a number of phenomena in different sections of the grammar instead: pre- and postnominal demonstratives (§4.6), aspect marking (§7.2), directional particles (§7.5), subject and object marking (§8.3–8.4), non-canonical subject marking and non-standard constituent orders (§8.6).

## **2 Phonology**

### **2.1 Introduction**

As this grammar is primarily based on corpus research, it does not include a complete phonology; rather, what follows is a relatively brief phonological sketch. The following topics are discussed:


Rapa Nui is one of the few Polynesian languages in which the glottal plosive is a contrastive phoneme; it is discussed in detail in §2.2.4–2.2.5. The discussion will show that while the glottal plosive is clearly contrastive in lexical words, in prenuclear particles the situation is different.

Phonological processes such as metathesis and vowel shifts have profoundly affected the lexicon of Rapa Nui, perhaps more so than in other Polynesian languages. These processes are described and illustrated in §2.5.2.

Finally, §2.6 deals with reduplication. Rapa Nui has two types of reduplication; first the form, then the function of each type is discussed.

The research for this grammar does not include formal acoustic analysis (though for certain topics a speech corpus was used). This means that the pronunciation of phonemes is only indicated in general terms (§2.2.1–2.2.2). Likewise, the treatment of intonation is limited to general statements. A full analysis of the phonetics of Rapa Nui has never been carried out so far.

### **2.2 Phonemes**

The phoneme inventory of Rapa Nui consists of 10 consonants and 10 vowels (5 short and 5 long).

#### 2 Phonology

### **2.2.1 Consonants**

**Inventory** The consonant inventory of Rapa Nui is given in Table 2.1.

bilabial labiodental alveolar velar glottal voiceless plosive p t k Ɂ nasal m n ŋ voiceless fricative (f) (s) h voiced fricative v flap r

Table 2.1: Consonant inventory

/p/ is a voiceless unaspirated bilabial plosive.

/t/ is a voiceless unaspirated alveolar plosive.

/k/ is a voiceless unaspirated velar plosive. Before front vowels /e/ and /i/ it is somewhat fronted towards the palatal position.

/Ɂ/ is an unaspirated glottal plosive. It is sometimes realised as creaky voice on the surrounding vowels. It is not unusual for the glottal plosive to be elided; this happens especially in rapid speech and/or between identical vowels (e.g. *toꞌo* 'take', *nuꞌu* 'people'). Elision of the glottal plosive is more common with certain speakers than with others.

/m/ is a voiced bilabial nasal.

/n/ is a voiced alveolar nasal.

/ŋ/ is a voiced velar nasal.

/h/ is a voiceless glottal fricative. Between vowels, it may become voiced in rapid speech.

/v/ is a voiced labiodental fricative. In rapid speech it may become a labiodental approximant [ʋ].<sup>1</sup>

/r/ is a voiced alveolar flap [ɾ], not a trill [r].

The remaining two consonants only occur in loanwords:

/f/ is a voiceless labiodental fricative.

/s/ is a voiceless alveolar fricative.

Even in loanwords, /f/ and /s/ are often changed to native phonemes (§2.5.1 below). In this grammar, all phonemes are written in accordance with standard Rapa Nui orthography (§1.4.4), i.e. /Ɂ/ is written as *ꞌ*, /ŋ/ as *ŋ*.

**Contrastive sets** All consonants are contrastive both word-initially and between vowels. The following minimal sets show contrastivity for groups of similar consonants.

<sup>1</sup> Guerra Eissmann et al. (1993: 14) notice that a few of their consultants tend to pronounce it as a bilabial fricative. However, Fischer (2001a: 317–318), quoting – among others – Weber & Weber (1982), confirms that despite pervasive Spanish influence on the language, /v/ is still a labiodental.

2.2 Phonemes

Bilabials and labiodental: *p*, *m*, *v*


Alveolars: *t*, *n*, *r*


Velars and glottals: *k*, *ŋ*, *ꞌ*, *h*


Glottal versus Ø


These examples show that the glottal plosive is contrastive word-initially; however, this does not mean that it is contrastive phrase-initially (§2.2.5).

Plosives: *p*, *t*, *k*, *ꞌ*



Fricatives: *v*, *h*


#### 2 Phonology

*v* versus *u* (notice that the segmental difference in these pairs also implies a difference in syllable structure: *ꞌa.va.hi* versus *ꞌa.u.a.hi*, see §2.3.1)


*h* versus Ø


*ŋ* is relatively rare word-initially. Only about 1/6 of its token occurrences in the text corpus are word-initial, and 2/3 of these concern the plural marker *ŋā*. (Likewise, of all occurrences of *ŋ* in the lexicon, less than 1/6 is word-initial.)

**Derivation** The consonant correspondences between Rapa Nui and its ancestors (Proto-Polynesian, Proto-Nuclear Polynesian and Proto-Eastern Polynesian)<sup>2</sup> are given in Table 2.2 (data from Marck 2000: 23–24). The consonants of Proto-Central-Eastern Polynesian and Tahitian are also included, not only because Central-Eastern languages are Rapa Nui's closest relatives (§1.2.1), but also because Rapa Nui borrowed extensively from Tahitian (§1.4.1; §2.5.3.2).


Table 2.2: Derivation of consonant phonemes

As Table 2.2 shows, the PPN glottal plosive was retained in Rapa Nui but lost in PCE (though it is sporadically retained in some words in CE languages, see Wilson 2012: 335). This means that Rapa Nui is the only EP language where it was preserved. The phonemic status of the glottal is discussed in more detail in §2.2.4 below.

<sup>2</sup> See the Polynesian language tree in Figure 1.1 on p. 6.

<sup>3</sup> PPN *\*s* was still present in PEP and PCE, but Penrhyn is the only EP language to retain it; in all others, *\*s* became *h*, as in Rapa Nui.

2.2 Phonemes

PEP *\*f* and *\*s* became *h* in all environments in Rapa Nui. In fact, *\*f* merged with *\*s* in all EP languages, either in some or in all environments.<sup>4</sup> One change which occurs in all CE languages but not in Rapa Nui, is *\*faf-* > *\*vah-*: PPN/PEP *\*fafa* 'mouth' > PCE *\*vafa*, but Rapa Nui *haha*; PNP/PEP *\*fafie* 'firewood' > PCE *\*vafie*, but Rapa Nui *hahie* (§1.2.2 no. 13).

PPN*\*h* is lost in most languages. In some NP languages (including some EP languages), PPN *\*h* is reflected as *s* or *h* in a few words (Marck 2000; Rutter 2001 argues that some of these actually reflect PPN *\*s* rather than *\*h*). In Rapa Nui, it is reflected as a glottal plosive in a few words (for examples see §2.5.2, cf. Davletshin 2015).

### **2.2.2 Vowels**

The vowel inventory of Rapa Nui is given in Table 2.3.



In this grammar, vowel length is represented by a macron over the vowel, in accordance with standard Rapa Nui orthography.

The vowel system was inherited without change from PPN. All vowels are contrastive in word-initial, -medial and -final position. Table 2.4 gives minimal sets of two or more contrastive short vowels.

Vowel length is contrastive. Some examples of monosyllabic minimal pairs:


Notice that in all these pairs the short-vowel word is a prenuclear particle, while the long-vowel word is a lexical word or postnuclear particle. As a result, the two words will never occur in an identical context.

For bisyllabic words, most minimal pairs concern final vowels; in these cases, the length distinction also implies a difference in stress (§2.4.1): *haꞌi* /ˈhaɁi/ 'to embrace' versus *haꞌī* /haˈɁiː/ 'to wrap up'. Other examples include:

<sup>4</sup> In Hawaiian and Rapa Nui, *\*f* > *h* in all environments; in Mangarevan, Rapa and Rarotongan, *\*f* > *ꞌ* in all environments; in Tahitian and Māori, *\*f* > *h* medially and before round vowels, though not without exceptions (see Harlow 1998).

#### 2 Phonology


Table 2.4: Vowel contrasts


There are a few pairs of words which only differ in vowel length in the antepenultimate (hence unstressed) syllable; these words are distinguished by vowel length only.


2.2 Phonemes

### **2.2.3 Phoneme frequencies**

In the text corpus, totalling over 1.6 million segments, the token frequency of each segment is as given in Table 2.5. 5


Table 2.5: Phoneme frequencies

56.9% of all segments are vowels (52.3% short, 4.6% long), 43.1% are consonants.

The most common phonemes, in descending order of frequency, are *a e i t o ꞌ h*. If corresponding short and long vowels are considered as instances of the same vowel (i.e. figures for *a* and *ā* are added up), the order is *a e i o t ꞌ u h*. The least common phonemes, in ascending order of frequency, are *ū ē ī ō v ŋ*.

### **2.2.4 The glottal plosive**

As indicated above, Rapa Nui is the only Eastern Polynesian language which preserved the PPN glottal plosive. The glottal was lost various times in the history of Polynesian; apart from Rapa Nui, it was preserved only in Tongan, Rennell-Bellona, East Uvean and East Futunan. Within Eastern Polynesian, the presence of the PPN glottal is one of the features distinguishing Rapa Nui from the Central-Eastern languages.

The glottal plosive was not recorded in early lexical sources: neither in word lists by Philippi (1873), Geiseler (1883), Thomson (1889), Cooke (1899), nor in the lexica by Roussel (1908; 1917), Churchill (1912) and Martínez (1913). Métraux, who stayed on the island in 1934–1935, explicitly mentions that "so far as I can trust my ear there is no trace of the glottal stop on Easter Island" (Métraux 1971: 32).

<sup>5</sup> Counts are based on written data, so elision (e.g. of the glottal plosive) is not taken into account. The corpus also contains 10,600 non-Rapa Nui characters, such as *s* and *l* (both around 2000 times), which occur in borrowings and proper names. These do not affect the overall percentages.

Alternatively, phoneme frequencies could be based on a list of lexemes. However, as the PLRN lexical database collates data from all lexical sources, contains a relatively high proportion of words occurring in one or two older sources (especially Roussel 1908) which were never part of the language.

#### 2 Phonology

Englert, who lived on the island from 1935 until 1968, did notice the significance of the glottal plosive: he lists minimal pairs, where the presence or absence of the "hiato" changes the meaning of the word (Englert 1978: 16). All of the glottals he noticed occur word-medially between non-identical vowels (e.g. *vaꞌe* 'foot' versus *vae* 'choose').

The first linguist to fully recognise the glottal plosive as a phoneme in Rapa Nui was Ward (1961; 1964). Ward compared occurrences of the glottal plosive with cognates in other Polynesian languages that retain the PPN glottal, and concluded that the glottal in Rapa Nui corresponds to the original PPN glottal (apart from Tahitian borrowings, see below). An example is Rapa Nui *hōꞌou* 'new'; Tongan, East Futunan, East Uvean *foꞌou*, Rennell *hoꞌou*, but Hawaiian and Tahitian *hou*. Around the same time, Bergmann (1963: 4) included the glottal in his phoneme inventory of Rapa Nui, though he suggested that it has disappeared in the modern language.

The adoption of the glottal as a full-fledged consonant phoneme was confirmed in later phonological analyses: Blixen (1972), Salas (1973), Weber & Weber (1982) and Guerra Eissmann et al. (1993). <sup>6</sup> Despite Métraux' and Bergman's assertions to the contrary, in current Rapa Nui the glottal stop is consistently present. Only a minority of speakers (especially those for whom Rapa Nui is not their first language) tend to elide it frequently.

While most instance of the glottal plosive in Rapa Nui correspond to the PPN glottal, a second source for the glottal plosive is Tahitian. Rapa Nui borrowed extensively from Tahitian (§1.4.1); this includes words containing glottals, like *hoꞌo* 'buy/sell', *ꞌaꞌamu* 'story', *ꞌānoꞌi* 'to mix' and *haꞌari* 'coconut'. The fact that the glottal was already part of the phoneme system doubtlessly facilitated the adoption of these words without elision of the glottal (§2.5.3.2).

### **2.2.5 The glottal plosive in particles**

As shown in §2.2.1 above, the glottal plosive is contrastive both word-initially and after vowels. Now all examples given there concern full words (§3.1), i.e. content words; full words in natural speech are usually preceded by a particle, e.g. an aspectual or a determiner. They hardly ever occur at the start of a prosodic phrase.

Words which do occur at the start of prosodic phrases are prenuclear particles, such as aspect markers and prepositions. In the standard Rapa Nui orthography, some of these are written with a glottal: *ꞌe* 'and', *ꞌa* 'of'; others do not have a glottal: *e* 'ag; ipfv', *o* 'of'.

The question is, whether there is a real phonetic distinction between the presence and the absence of a glottal in these particles. To answer this question, I analysed the pronunciation of eight particles – four with orthographic glottal, four without – in an oral text corpus, spoken by a number of speakers of different genders and age groups.<sup>7</sup> For each occurrence, I determined:

<sup>6</sup> According to Marck (2000: 24, 69), the Rapa Nui glottal was lost in the environment *a\_\_a*; however, this is based partly on sources with defective orthography, such as Fuentes (1960) (e.g. RN *\*haaki* 'to inform' < PNP *\*fa ꞌaki*; the actual Rapa Nui form is *hā ꞌaki*), partly on Tahitian loans (RN *tane* 'male' < Tahitian *tāne*, cf. PNP *\*ta ꞌane*). See also Davletshin (2015).

<sup>7</sup> The corpus consists of Bible passages used for the dubbing of a Biblical movie. These texts were rehearsed recitation, partly read from paper, which may favour a pronunciation in line with the orthography; however, the passages were practiced until pronounced smoothly and naturally, which should have mitigated the "orthography effect".

2.2 Phonemes



Table 2.6: Pronunciation of glottals in particles

This yields the statistics given in Table 2.6. This table shows that particles written with a glottal are indeed overwhelmingly pronounced with a glottal (81.2%), while particles written without glottal are predominantly pronounced without glottal (67.0%). However, this effect is largely due to the distribution of these particles. At the start of a prosodic phrase, most Rapa Nui speakers almost automatically pronounce initial vowels with a sharp onset, i.e. a non-phonemic glottal plosive. As Table 2.6 shows, phrase-initial particles are overwhelmingly pronounced with a glottal: 440+315 against 33+4 without glottal, i.e. 755 out of 792 (95.3%). On the other hand, non-phrase-initial particles tend to be pronounced without a glottal: 945+75 against 42+26 with glottal, i.e. 1020 out of 1088 (93.8%). The fact that this strongly correlates with the presence or absence of the

#### 2 Phonology

*written* glottal, is because certain particles happen to occur much more frequently after boundaries than others. For example, the conjunction *ꞌe* 'and' is almost always preceded by a pause, while the proper article *a* is very often preceded by a preposition, hence non-initial. In other words, the glottal is not phonemic in these particles; it just tends to be pronounced phrase-initially and omitted otherwise.

The following example illustrates this. The first line represents the orthography, the second line is a broad phonetic transcription. | indicates a prosodic phrase break; \_ represents a vowel onset without glottal.

(13) *Te* [ te art *nuꞌu* nuʔu people | *e* **ʔe** ipfv *rerehu* rerehu faint *rō* roː emph *ꞌi* **\_i** at *te* te art *riꞌariꞌa.* riʔariʔa afraid | *ꞌI* **ʔi** at *rā* raː dist *hora* hora time *he* he ntr *takeꞌa* takea see *i* **\_i** acc *te* te art *Poki* poki child *o* **\_o** of *te* te art *Taŋata* taŋata man | *ka* ka cntg *topa* topa descend *mai* mai hither | *ꞌi* **ʔi** at *ruŋa* ruŋa above *i* **\_i** at *te* te art *raŋi* raŋi heaven | *ꞌi* **ʔi** at *ruŋa* ruŋa above *i* **\_i** at *a* **\_a** prop *ia* ia 3sg | *te* te art *pūai* puːai power *ꞌe* **ʔe** and *te* te art *ꞌanaꞌana* ʔanaʔana glory *o* **\_o** of *te* te art *ꞌAtua.* ʔatua ] God

'The people will faint from fear. At that time they will see the Son of Man descending in heaven, on him the power and glory of God.' [R630-13.010]

As this fragment shows, the preposition *ꞌi* 'in' is pronounced with glottal after a pause (3x), but without a glottal within a prosodic phrase (line 1). The conjunction *ꞌe* 'and' is pronounced with glottal after a pause (line 3), but so is the imperfective marker *e* (line 1). The preposition *i* is never pronounced with glottal in this fragment, but then, it does not occur phrase-initially. The same is true for the proper article *a* and the possessive preposition *o*.

This example also shows that the orthography is accurate as far as glottals in content words are concerned: glottals are usually pronounced where they are written, both wordmedially (*nuꞌu, riꞌariꞌa*) and word-initially (*ꞌAtua*). The same is true for longer particles (which do not occur in this example), such as the postverbal markers *ꞌā* 'cont' and *ꞌai* 'subs'; these are consistently pronounced with glottal.

We may conclude that the glottal is not contrastive in prenuclear particles. The glottal is a phonetic reality only to the extent that particles occur post-pausally.<sup>8</sup> This does not mean that the use of the glottal symbol in these particles is without justification: it helps the reader to distinguish possessive *ꞌa* from the proper article *a*, and the conjunction *ꞌe* from the many particles *e*. Yet one should keep in mind that the distinction is in a sense

<sup>8</sup> Cf. Clark (1976: 222): particles fail to follow the normal correspondences, which "is probably a result of their typically phrase-initial position". He points out that there is a universal tendency to insert a glottal after a pause, so "glottal stop in such position is of dubious value" (23). He gives several examples of initial particles which have an initial glottal stop in Tongan (a language which has preserved the PPN glottal), no glottal stop in Rennell (idem), but a glottal stop in Tahitian (which does not retain the PPN glottal).

For this reason, it is difficult to reconstruct the protoform of phrase-initial particles; there is some discussion whether the PPN possessive markers were *a* or *ꞌa*, and *o* or *ꞌo* (see Fischer 2000, Lichtenberk 200; Wilson 1982; Lynch 1997).

2.3 Phonotactics

superficial. This is especially important in the case of the prepositions *ꞌi* and *i*, which are etymologically a single preposition (§4.7.2).

### **2.3 Phonotactics**

### **2.3.1 Syllable structure**

The syllable structure of Rapa Nui is (C)V(ː). The syllable contains a single short or long vowel, optionally preceded by one consonant. A syllable cannot contain two or more vowels. This means that all sequences of non-identical vowels are disyllabic, even those often analysed as diphthongs in other Polynesian languages: words like *kai* 'to eat' and *mau* 'really' do not contain a diphthong, but two syllables.

In older descriptions of Rapa Nui, such as Englert (1978), Fuentes (1960) and Salas (1973), certain VV sequences are analysed as diphthongs; for example, Englert (1978: 16) mentions *ai*, *au* and *oi*. <sup>9</sup> This is understandable, as phonetically it is often impossible to distinguish two separate syllable nuclei in sequences like *ai* and *au*. Even so, there are several reasons to consider all VV sequences as disyllabic.<sup>10</sup>



<sup>9</sup> In other Polynesian languages the diphthong inventory may be different. For example, in Tahitian, all VV sequences in which the first vowel is more open, are considered diphthongs (Académie Tahitienne 1986: 5). The same is true in Māori (Harlow 2007a: 69).

<sup>10</sup> The first three of these are also mentioned by Weber & Weber (1982).

<sup>11</sup> In this area, the difference with Tahitian is especially obvious. While cognates of the first four items occur in Tahitian as well (*rāua* and *maika* are shared cognates, *māuiui* and *haraoa* were borrowed from Tahitian), the pronunciation in Tahitian is markedly different because of diphthongisation and stress shift: [ˈraːᵘa ˈmaⁱɁa ˈmaːᵘiᵘi faˈraːᵒa].

#### 2 Phonology

If, for example, *ai* in *maika* would be a diphthong, it would be impossible for *i* to receive stress; rather, the stress pattern would be [ˈmaika].


### **2.3.2 Word structure**

#### **2.3.2.1 Constraints on word structure**

The phonological structure of words in Rapa Nui can be described using concepts from metrical theory (Kager 1995; Hayes 1995). Words and phrases are organised in metrical units; in ascending order: morae (µ), syllables (*σ*), feet (F) and prosodic words. The following constraints apply:


These constraints are inviolable and apply to all words, including reduplications, compounds and borrowings. This means that all words in the language conform to a single rule: a heavy syllable is never followed by an odd number of morae.<sup>16</sup> In other words, when heavy syllables are followed by light syllables, the latter always occur in pairs; patterns such as the following do not occur:

<sup>12</sup> Van den Berg (1989: 24) makes the same observation for Muna. Similarly, Rehg (2007: 127) points out that the wide range of diphthongs in Hawaiian (as opposed to e.g. English) suggests that they are VV sequences rather than occupying a single V slot (though in his analysis, they are not bisyllabic).

<sup>13</sup> Coda consonants, which occur in some borrowings, are non-moraic as well. Coda consonants can occur in any non-final syllable, including the penultimate: *toro***m***po* 'spinning top', *ase***r***ka* 'chard', *ramie***n***ta* 'tool'. If these consonants were moraic, the penultimate syllable would be heavy, while the final syllable is light. They would thus violate the \*…HL constraint formulated below.

<sup>14</sup> This condition was formulated as a universal constraint by Hayes (1995: 50).

<sup>15</sup> An alternative would be to state that the initial mora can be left unparsed. However, the fact that initial syllables sometimes receive secondary stress, suggests that they are in fact parsed into a degenerate foot. Cf. Kager's principle of exhaustivity (Kager 1995: 370), which requires all syllables of a word to be parsed.

<sup>16</sup> In other Polynesian languages a similar tendency operates, though usually in a weaker form. For example, in Samoan, the penultimate vowel in trisyllabic words cannot be long (Hovdhaugen 1990: 102).

2.3 Phonotactics

#### (15) \* H L \* L H L H \* H L L L \* H L L L H

Table 2.7 lists all occurring word patterns. Certain patterns are common, while others are rare or nonexistent. Foot boundaries are indicated by dots. Column 2 gives the number of morae, column 3 lists the frequency of each pattern.<sup>17</sup>

The table shows that words containing up to six morae are common. Longer words are rare; in fact, all 7–8 mora words are either reduplications or compounds (e.g. *tōuamāmari* < *tōua* 'yolk' + *māmari* 'egg').

<sup>18</sup> Most of these have identical final and penultimate syllables, but for none of them there is clear evidence that they are reduplications.



<sup>17</sup> Counts are based on the PLRN lexical database (§1.6.1). All words in the database are included, including obsolete words (the length of which cannot be ascertained, though Englert's lexicon often records length accurately), as well as words which may never have been genuine Rapa Nui words, but which occur in less reliable sources such as Roussel (1908). Homonyms are counted separately.

#### 2 Phonology

Table 2.8: Metrical word structures (cont.)


#### 2.3 Phonotactics

Below are examples of the metrical structure of *hānautama* 'pregnant', *keretū* 'pumice' and *mauku* 'grass'. (Feet are indicated by round brackets; the strong mora within the foot is marked as x, the weak mora as a dot.)


The absence of certain structures follows straightforwardly from these constraints. For example, the Tahitian word *tāne* 'man, male' was borrowed into Rapa Nui, but with shortening of the first vowel: *tane*. The form *\*tāne* (with the non-attested pattern \*HL) would involve either a degenerate foot at the end of the word (violating constraint 4), a foot spanning a syllable boundary (violating constraint 3), or an unparsed syllable (violating constraint 2). These alternatives are illustrated below.<sup>19</sup>


We may conclude that the prosodic shape of words is determined by a set of nonviolable metrical constraints. Once these constraints are established, a number of other issues can be addressed: minimal words, vowel sequences, and the frequency of metrical patterns. These are discussed in the following sections.

#### **2.3.2.2 Minimal words**

Content words minimally consist of one bimoraic foot: *pō* 'night', *kai* 'to eat', *hare* 'house', *oho* 'to go'. Postnuclear particles are minimally bimoraic as well (in fact, most of these are bimoraic): *nō* 'just', *era* 'distal', *mai* 'hither'. The same is true for particles occurring in isolation, such as *ꞌina* 'neg' and *ꞌī* 'imm'. Only prenuclear particles may be monomoraic: *te* 'art', *e* 'ipfv', *ki* 'to'.<sup>20</sup>

<sup>19</sup> Finney (1999: 171) notes that in most Polynesian languages, words can end in V1V1CV. In that case, speakers "tend to treat the antepenult and the penult as a foot, a single long syllable, even though that violates the normal [process of right-to-left foot formation]". We may conclude that Rapa Nui differs from other Polynesian languages in that the constraints on foot formation impose absolute constraints on word formation.

<sup>20</sup> This means that prenuclear particles are not independent phonological words, which can be taken as evidence to analyse them as clitics. Another reason to consider them as clitics is the fact that they occupy a fixed (i.e. initial) position in the phrase, without being attached to a single category: a prenuclear particle precedes whatever comes next in the phrase. In the end, whether or not they are to be considered clitics may be a matter of terminological preference. (Cf. Payne 1997: 23: it is uncommon to use the term clitic for elements such as adpositions and tense/aspect markers.)

#### 2 Phonology

#### **2.3.2.3 Vowel sequences**

In the previous section, several reasons were mentioned to analyse sequences of two non-identical vowels as disyllabic sequences rather than diphthongs. The conditions on metrical structure provide another argument for a disyllabic analysis. As pointed out above, (C)Vː(C)V words such as *\*tāne* do not occur in Rapa Nui, a fact which can be explained by metrical constraints ruling out \*HL patterns. On the other hand however, (C)V1V2(C)V words are common: *mauku* 'grass', *hauꞌa* 'smell', *maika* 'banana', *koia* 'with', *paihi* 'torn', *taote* 'doctor', et cetera. Now if *au, ai, oi* and *ao* would be monosyllabic (i.e. diphthongs), these words would have an HL pattern, and it would be unclear why these words are possible while *tāne* is not. On the other hand, if these sequences are disyllabic, these words have a LLL pattern just like *makenu* 'to move' and *poreko* 'to be born', a pattern which is metrically well-formed and which is in fact very common.<sup>21</sup>

#### **2.3.2.4 Common and uncommon patterns**

For words consisting of 1, 2, 3, 4 or 6 morae, all possible patterns are attested. (Longer words are very rare overall.) Even so, some patterns are more common than others.

In general, light syllables are more common than heavy syllables. Patterns with an LL foot in a given position in the word are much more common than patterns with a H foot in the same position, e.g. L.LL (1010) versus L.H (54); LL.LL (1290) versus LL.H (70) or H.LL (453). The only exception is H.LL.LL (130), which is more common than LL.LL.LL (35).

H syllables are more common word-initially than word-finally. Not counting monosyllabic H words, there are 686 words with initial H, against 228 words with final H. Medial H is also relatively uncommon; it mainly occurs when the preceding or following syllable is also H. Of all 329 three- and four-foot words (the only ones in which medial H is possible), 164 have initial H, 47 have one or two medial H, while 35 have final H.

The patterns listed above, as well as etymological data, suggest that there is a tendency to avoid degenerate feet.<sup>22</sup> Lengthening an initial vowel turns a degenerate foot into a complete bimoraic one. If the reconstructed forms in Pollex (Greenhill & Clark 2011) are correct, the initial syllable was lengthened in words such as *hōꞌou* 'new' (PPN *\*foꞌou*), *ꞌūnahi* 'fish scale' (PPN *\*ꞌunafi*), *hōhonu* 'deep' (PEP *\*fofonu*) and *pūꞌoko* 'head' (by metathesis from PEP *\*upoko*). In longer words this tendency is even stronger: there are more H.LL.LL words in the lexicon (130) than L.LL.LL (101).<sup>23</sup> Certain reduplication patterns show a tendency to lengthen L.LL.LL to H.LL.LL (§2.6.1.2). On the other hand, the pressure toward whole feet is not sufficiently strong to prevent the occurrence of many hundreds of LLL words; in fact, this is the third most common pattern overall.

<sup>21</sup> Following a similar reasoning, Anderson & Otsuka (2006) conclude that long vowels in Tongan must be disyllabic, as they may span a foot boundary.

<sup>22</sup> Englert (1978: 17) already notices the tendency to lengthen antepenultimate vowels. Cf. Kager (1995: 399): languages employ various strategies to avoid degenerate feet, such as lengthening and reparsing.

<sup>23</sup> In actual language use the difference is even more marked: many of the L.LL.LL words in the lexicon are borrowings, some of which only occur in older sources such as Roussel (1908) and which are no longer (or never were) in use.

2.3 Phonotactics

Another issue related to metrical structure is stress assignment. This will be discussed in §2.4.1.

### **2.3.3 Cooccurrence restrictions**

#### **2.3.3.1 Between vowels**

As discussed above, all sequences of non-identical vowels are possible. This is illustrated in Table 2.9.


Table 2.9: Examples of VV sequences

Not all VV sequences are equally common, however. The text corpus contains 90,700 disyllabic VV sequences; their relative frequencies are given in Table 2.10.


Table 2.10: Frequencies of VV sequences

As Table 2.10 shows, sequences of a high and a low vowel in either order (*ai*, *au*, *ia*, *ua*) are much more common than those containing a mid vowel or consisting of two high vowels. The former four sequences together account for 62.1% of the total, while

#### 2 Phonology

the other fourteen sequences account for 37.9%. *uo* hardly occurs at all. *eo* and *eu* are rare, as well as *iu*, *ie*<sup>24</sup> and *ue*.

VVV sequences are common as well: *māua* 'we (dual excl.)', *pūai* 'strong', *tāea* 'to throw a lasso', *tōua* 'egg yolk', *tūai* 'ancient'.

#### **2.3.3.2 Between vowels and consonants**

Any consonant can be followed by any vowel, with one exception: the syllable *vu* is extremely rare. Apart from the loanword *vuto* 'sweet' (< Sp. *dulce*), it only occurs in *vuhi* (and its reduplication *vuhivuhi*) 'to whistle', a word not occurring in the text corpus.<sup>25</sup>

*vo* is not very common either; it occurs in eight lexical entries, such as *vovo* 'dear girl' and *voꞌu* 'to shout'.

#### **2.3.3.3 Between consonants**

As discussed in §2.3.1 above, consonants are always separated by a vowel; contiguous consonants do not occur. (The only exceptions occur in loanwords, see §2.5.3.1 below.) Even so, there are a few co-occurrence restrictions between consonants in adjacent syllables.<sup>26</sup>

Firstly, the co-occurrence of a homorganic nasal + stop (in that order) within a root is very rare, though not completely excluded. The co-occurrence of homorganic stop + nasal is somewhat less rare, though by no means common. The data are given in Table 2.11.

Table 2.11: Nasal + plosive cooccurrence restrictions


<sup>24</sup> Most of the occurrences of *ie* are due to Spanish influence. While *ie* is quite rare in Rapa Nui words (apart from some proper names), it is very common in Spanish and often occurs in loanwords: *fiesta, noviembre, tiene*…

<sup>25</sup> De Lacy (1997) reports similar restrictions in Māori, where *\*wu, \*wo, \*whu* and *\*who* do not occur. In Tuvaluan, *\*vu-* is unattested, while *\*vo-* is rare (Besnier 2000: 612).

<sup>26</sup> Similar restrictions operate in other Polynesian languages (see Mosel & Hovdhaugen 1992: 24 on cooccurrence restrictions between labial and labiodental consonants; Bauer, Parker & Evans 1993: 554 on a tendency towards consonant dissimilation in adjacent syllables).

<sup>27</sup> Non-initial *kVŋ-* is not uncommon in forms containing the nominaliser *-ŋa*, such as *pikoŋa* 'hiding place', but here the two consonants are separated by a morpheme break.

2.4 Suprasegmentals

Secondly, Davletshin (2015) mentions a constraint against the occurrence of glottal stops in adjacent syllables of bimoraic words.<sup>28</sup> This is confirmed by my data: in the lexicon and the text corpus there are no words of the shape *ꞌVꞌV*.

### **2.4 Suprasegmentals**

### **2.4.1 Stress**

Metrical structure as described in §2.3.2 above allows a simple description of word stress: the final foot of the word is prominent. This results in the following pattern:


The strong morae of the other feet receive secondary stress. This results in a rhythm of alternating strong and weak morae. Some examples:


Not all non-final feet have the same level of stress. For example, when the initial foot is degenerate, either the initial or the second syllable may be slightly more prominent:


More study is needed to determine which factors determine levels of lower-order stress.

In connected speech, phrase stress is more conspicuous than word stress. Stress is assigned at the level of the prosodic phrase, according to the same rule as word stress: the final foot of the phrase is prominent. In other words, stress falls on the phrase-final syllable if it is long, and on the penultimate syllable otherwise. Prosodic phrase breaks

<sup>28</sup> This constraint does not operate in words having more than two morae: glottals in adjacent syllables occur in bisyllabic words with long vowels (*ꞌīꞌī* 'slightly spoiled (food)', *ꞌuꞌū* 'to groan'), and in trisyllabic words (*ꞌa ꞌaru* 'to grab', *ha ꞌiꞌa* 'Malay apple').

#### 2 Phonology

usually coincide with breaks between syntactic constituents, but not all syntactic phrases constitute a separate prosodic phrase.

In the examples below, prosodic phrase breaks are represented by |.


(21) *ˌKi* to *a* prop *ˌkōˈrua,* 2pl | *ki* to *te* art *ˌnuꞌu* people *ˌhakaˌroŋo* listen *ˈmai,* hither | *ˌꞌī* imm *a* prop *ˌau* 1sg *he* ntr *ˌkī* say *ˈatu…* away [ˌkiaˌkoːˈrua kiteˌnuʔuˌhakaˌroŋoˈmai ˌʔiːaˌauheˌkiːˈatu] 'To you, to the people listening, I tell you…' [R630-04.063]

As these examples show, primary stress always falls on the final foot of the prosodic phrase, whether this is a lexeme (*oromatuꞌa* in (18)), a continuous marker (*ꞌā* in (18)), a nominaliser (*iŋa* in (19)), a postnuclear demonstrative (*ena* in (20)), or a directional (*mai* in (21)). All other feet potentially receive secondary stress. However, secondary stress is not always conspicuous, especially on or near long vowels, when two contiguous syllables both contain a strong mora. Not all secondary stresses are equally strong, though this has not been indicated in the examples above. A more refined analysis is needed to determine how different levels of non-primary stress are assigned. Two factors that seem to play a role are:

• semantic or pragmatic prominence. The nucleus of the phrase (often the only lexical word) tends to get relatively heavy secondary stress, especially the syllable that would be stressed according to the word stress rules; e.g. in *haꞌatura* in (19) and *haka ꞌite* in (20), the second foot receives more stress than the first. The deictic particle *ꞌī* in (21) is relatively prominent as well.

2.4 Suprasegmentals

• linear distance. Feet immediately preceding the main phrase stress are not heavily stressed. This means that the stressed syllable of content words may not receive a high degree of stress if it is immediately followed by the phrase stress: in *hakaroŋo* in (21), the initial syllable receives more stress than the penultimate one, despite the word stress on the latter.

### **2.4.2 Intonation**

This section describes a number of intonation patterns in declarative and interrogative clauses. Examples are given from basic sentences, i.e. monoclausal sentences with standard constituent order. A full treatment of intonation would require precise acoustic analysis and is outside the scope of this grammar.<sup>29</sup>

#### **2.4.2.1 In declarative clauses**

Intonation in declarative clauses is characterised by a peak on the stressed syllable of the predicate. Subsequently, the pitch may gradually drop:<sup>30</sup>

<sup>29</sup> Intonation in imperative clauses is not illustrated. Imperative clauses tend to show a high rise, followed by a gradual decline. This means that the intonation pattern is superficially identical to the intonation of declarative clauses. A more precise analysis could reveal subtle differences between declarative and imperative intonation, e.g. in the shape or timing of the rise.

<sup>30</sup> Intonation graphs were created using Speech Analyzer 3.1 (SIL International, 2012). In the examples, syllables bearing phrase stress are underlined.

2 Phonology

The final constituent may show a second peak, as on *poki era* in the next example:

Alternatively, the sentence may end in a high plateau. In the next example, there is a high rise on the second (stressed) syllable of *pāhono*; the pitch remains on this level throughout the final syllable.<sup>31</sup>

<sup>31</sup> The example is from a younger male speaker. Data from a range of speakers could show if this is pattern is limited to certain age groups.

2.4 Suprasegmentals

#### **2.4.2.2 In questions**

In polar questions (§10.3.1), there is usually a high rise on the stressed syllable of the first constituent; after that the pitch is low or falling, but on or just before the final stressed syllable the pitch quickly goes up. After a quick rise it tends to drop somewhat in poststress syllables, but not all the way back to the previous low level.

Below are two examples. In both cases there is a rise on the first constituent. The last stressed syllable of the sentence also exhibits a sharp rise; in (26) this rise is higher than the first one, while in (27) it is somewhat less high.

2 Phonology

Content questions (§10.3.2) are characterised by a high rise on the stressed syllable of the question constituent, followed by a sharp drop. There may be a moderate rise on the final stressed syllable, but the question may also end in a low pitch. Here are two examples. Both exhibit a high rise on the stressed syllable of the interrogative constituent; (28) has a falling pitch at the end of the question, while (29) has a rise to mid-range pitch.

2.5 Phonological processes

### **2.5 Phonological processes**

Rapa Nui is poor in morphology; as a consequence, morpho-phonological processes are uncommon. The only exception is found in the area of reduplication (§2.6.1). This section describes phonological processes which are not morphologically conditioned. §2.5.1 discusses three regular phonological processes: word-final devoicing, pre-stress lengthening and elision.

Other phonological processes are lexically determined and result in lexical items having a different form than expected on the basis of cognates, or having two or more alternate forms; these are discussed in §2.5.2. Finally, §2.5.3 deals with the (more or less regular) phonological adaptation of borrowings.

### **2.5.1 Regular processes**

This section discusses three regular phonological processes, i.e. processes which are not limited to certain lexical items. All three are optional. Two processes, word-final vowel devoicing and pre-stress lengthening, take place in certain well-defined phonological contexts. For a third process, elision, no specific conditions can be formulated without extensive further analysis.

#### **2.5.1.1 Word-final vowel devoicing**

Word-final short (hence unstressed) vowels are optionally devoiced after voiceless consonants. This happens especially at the end of an utterance, or at least the end of a prosodic phrase.<sup>32</sup>

<sup>32</sup> Vowel devoicing occurs in other Polynesian languages as well: Māori (Harlow 2007a: 76; Bauer, Parker & Evans 1993: 556); Niuafo'ou (Tsukamoto 1988: 23-25; De Lacy 2001), Tongan (Feldman 1978: 137). The conditions under which devoicing occurs in these languages, are different from those in Rapa Nui. In general, high vowels are affected more than low vowels.

#### 2 Phonology

In a stretch of careful speech by different speakers (about 7,400 words), I counted 80 instances of word-final devoicing. 75 of these occur at the end of a prosodic phrase, 72 of which occur at the end of an utterance. All vowels undergo the process: *a* (19x), *e*, (15x), *i* (9x), *o* (12x) and *u* (25x). Devoicing occurs after all voiceless consonants: *p* (2x), *t* (24x), *k* (17x), *ꞌ* (27x), *f* (1x, a foreign name), *h* (9x). It never occurs after voiced consonants or in non-final syllables.

Some examples: *tahataha* [tahaˈtahḁ] 'edge', *taŋata* [taˈŋatḁ] 'person,' *viꞌe* [ˈviɁe̥] 'woman', *ꞌariki* [Ɂaˈɾiki] 'king', ̥ *mōꞌoku* [moːˈɁoku̥] 'for me', *oho* [ˈoh̥o] 'go'.

#### **2.5.1.2 Pre-stress lengthening**

Occasionally, a short vowel immediately preceding the main phrase stress is lengthened.<sup>33</sup> I have noticed this phenomenon in particular with the contiguity/imperative marker *ka*: *ka tanu* [kaːˈtanu] 'bury'; *ka pure* [kaːˈpure] 'pray', *ka tuꞌu* [kaːˈtuɁu] 'arrive'. The phenomenon occurs with other particles as well: the proper article *a* in *ki a ia* [kiaːˈia] 'to him'; the negator *e ko* in *e ko pau* [ekoːˈpau] 'does not run out', the exhortative marker *e* in *e ꞌite* [eːˈɁite] '(you must) know'.

A possible explanation for this lengthening is the preference for whole feet; this preference is noticeable on the word level (§2.3.2) and could be operative on the phrase level as well. This would explain why (*ka*)*<sup>F</sup>* (*pure*)*<sup>F</sup>* – with a degenerate initial foot – is lengthened to (*ka:*)*<sup>F</sup>* (*pure*)*F*. However, this does not explain the lengthening of *a* in (*ki a*)*<sup>F</sup>* (*ia*)*<sup>F</sup>* and *ko* in (*e ko*)*<sup>F</sup>* (*pau*)*F*, which already have two complete feet.

#### **2.5.1.3 Elision**

It is not uncommon for phonemes or whole syllables to be elided. Guerra Eissmann et al. (1993: 45–47) give examples of elision of almost all consonants and vowels in a spoken speech corpus, such as *o Rapa Nui* [oˈrapaːi] 'of Rapa Nui'; *ꞌina e tahi* [inaˈtai] 'not one'; *meꞌe rivariva* [meːriːˈriːa] 'a good thing'. They do not indicate if any conditions on elision can be formulated; answering this question would require careful analysis of a corpus of spoken texts by different speakers, including different speech styles. Such an analysis lies outside the scope of the present investigation.

### **2.5.2 Lexicalised sound changes**

Even though there are regular sound correspondences between Rapa Nui and its protolanguages (§2.2.1–2.2.2), there is a considerable number of words for which Rapa Nui has an irregular reflex of the protoform, i.e. where sound changes have been at work. This includes numerous lexical items for which Rapa Nui has two or more alternate forms. These processes are productive: the same patterns can be observed in recent borrowings. Davletshin (2015), who illustrates these processes in detail, points out that they should be labeled "incomplete" rather than "sporadic" or "irregular": they are not completely unpredictable, but follow certain patterns.

<sup>33</sup> In Nukeria, monosyllabic prenuclear particles are lengthened before a bimoraic root (Davletshin 2016).

2.5 Phonological processes

Below is an overview of these sound changes. The etymology is given where known.

**Metathesis** Methathesis is very common in Rapa Nui (cf. Du Feu & Fischer 1993: 166), mostly between onset consonants of adjacent syllables, occasionally between vowels of adjacent syllables, and very occasionally between whole syllables. It is especially common between the antepenultimate and penultimate syllable of trisyllabic words, but may occur in any pair of adjacent syllables. The consonants affected are often similar, e.g. two plosives (*t*/*k*), or two glottal consonants (*ꞌ*/*h*).

	- a. *haꞌi* 'to embrace' < PPN *\*ꞌafi* 'to hold or carry under the arm'
	- b. *haꞌiꞌa* 'Malay apple' < Tah. *ꞌahiꞌa*
	- c. *kōtini* 'sock, stocking' < Eng. 'stocking'; Thomson recorded *tokin* in 1889 (Thomson 1980: 157)
	- d. *ŋaroꞌa* 'perceive' < PPN *\*roŋo* + *-na*
	- e. *takoꞌa* 'also' < PPN *\*katoa*
	- f. *tikeꞌa* 'to see' < PPN *\*kite* + *-a*.
	- g. *ꞌarīnā* / *ꞌanīrā* 'later today'
	- h. *ꞌavahata* / *ahavata* / *haꞌavata* 'box'
	- i. *ꞌavai* / *vaꞌai* 'to give'
	- j. *rava* / *vara* 'usually'
	- k. *ravaꞌa* / *varaꞌa* 'to obtain' (< Tah. *roaꞌa*)

#### (31) Vowels:

	- a. *kia*-*kia* 'seagull sp.' < PPN *\*aki*-*aki*
	- a. *taꞌoraha* < PNP *\*tafolaꞌa* shows metathesis between *ꞌ* and \**f* in non-adjacent syllables
	- b. *hōŋaꞌa* 'nest' < PPN *\*ofaŋa* (Ø C<sup>1</sup> C<sup>2</sup> > C<sup>1</sup> C<sup>2</sup> *ꞌ*); cf. PCE \**kōfaŋa*

**Vowel changes** Vowel changes are common. Most of these occur either in Tahitian borrowings or as variants alongside the original form. Most of these involve a single degree of height (*a/e*, *a/o*, *e/i*, *o/u*), but other alternations occur as well.

#### 2 Phonology

#### (34) *a/e*:


#### (35) *a/o*:


#### (36) *e/i*:

	- a. *kāhui* / *kāhoi* 'bunch' (< PEP *\*kāfui*)
	- b. *ku* / *ko* 'prf' (< PPN *\*kua*)
	- c. *tautoru* 'to help' < Tah. *tauturu*

#### (38) *i/u*:


#### (39) *a/i*:

a. *takeꞌa / tikeꞌa* 'to see' (< PPN *\*kite* + *-ꞌa*)

**The liquid** *r* This consonant alternates with either a glottal or zero in a number of words.<sup>34</sup>

	- a. *kioꞌe* 'rat' (< PNP *\*kiole*)
	- b. *tikeꞌa* / *tikera* 'to see' (< PPN *\*kite* + *-ꞌa*)
	- c. *ŋoriŋori* / *ŋoꞌiŋoꞌi* 'tiny'
	- d. *hatuꞌa* / *hatura* 'cinch, belt' (< PEP *\*fātuꞌa*)

<sup>34</sup> In Marquesan, *r* > *ꞌ* is a regular – though not exceptionless – change (Clark 2000b).

2.5 Phonological processes

(41) Ø/*r*:

a. *emu* 'to drown' < PPN *\*lemo*

b. *ꞌōhiohio* / *ꞌōhirohiro* 'whirlwind' (cf. PPN *\*siosio*)

As these examples show, in those cases where the etymology is known, the *r* is usually – but not always – secondary.

**Glottals** The glottal plosive is sometimes added, occasionally deleted.

	- a. *kaꞌikaꞌi* 'sharp' < PNP *\*kai*<sup>35</sup>
	- b. *ꞌohi* 'stem' < PPN *\*osi*
	- c. *paꞌo* 'to chop' < PPN *\*pao*
	- d. *haꞌataꞌahinu* / *haꞌatāhinu* 'to administer the last rites' (< Tah. *faꞌatāhinu*)
	- e. *pōꞌiri* 'darkness' is probably a borrowing from Tah. *pōiri*, with an inserted glottal by analogy of the synonym *pōꞌuri* (< PPN *\*pōꞌuli*).
	- f. *taꞌutini* < Tah. *tauatini*, with glottal inserted possibly by analogy of *taꞌu* 'year'
	- a. *maŋeo / maŋeꞌo* 'sour, bitter' < PPN *\*maŋeho*<sup>36</sup>
	- b. *ꞌīŋoa* 'name' < PPN *\*hiŋoa*
	- c. *ꞌai*/*ai* 'who' < PPN *\*hai* (see Footnote 3 on p. 483)
	- d. *ꞌaŋahuru* 'ten' < PPN *\*haŋafulu*
	- a. benefactive prepositions *mo* (< PPN *\*moꞌo*) and *mā* (< PPN *\*maꞌa*); the glottal was retained in the pronominal forms *mōꞌoku, māꞌana* etc (§4.2.3).
	- a. *pē* 'gone' < Tah. *pe ꞌe*
	- b. *hāpī* 'to learn' < Tah. *haꞌapiꞌi*
	- c. *haꞌamaitai* 'to bless' < Tah. *haꞌamaitaꞌi*

**The consonant** *h* In a few cases, *h* alternates with zero:

(46) a. *aŋa* 'to make, do, work' < PPN *\*saŋa* (the regular reflex would be *\*haŋa*) b. *ia / hia* 'yet'

<sup>35</sup> This word does not have a glottal in other glottal-preserving languages. The same is true for PPN *\*osi* and *\*pao* below.

<sup>36</sup> PPN *\*h* was lost in most languages. In the case of *\*maŋeho* it was not preserved in any other EP language, so the PEP form may have been *\*maŋeo*. Interestingly, the Hawaiian reflex is *mane ꞌo*, with a glottal as in Rapa Nui.

#### 2 Phonology

**Nasal consonants** Some words exhibit shifts between different nasal consonants, mostly between *n* and *ŋ*: 37

	- b. *tiŋaꞌi* 'to kill' < PNP *\*tinaꞌi*
	- c. *tumu* / *tuŋu* 'cough' (< PEP \**tuŋu*)
	- d. *norinori* / *ŋoriŋori* 'tiny'
	- e. *nako* 'fat, marrow' < PPN *\*ŋako*
	- f. *kona* 'place' < PPN *\*koŋa* 'fragment, part, place'<sup>38</sup>

In the last two examples, *ŋ* dissimilated to *n* in the vicinity of *k*.

**Monophthongisation** A number of particles exhibit monophthongisation of a VV cluster, resulting in a single short or long vowel:

	- b. *nō* 'just' < PPN *\*noa*
	- c. *rō* 'emph' < PPN *\*roa*
	- d. *hē* 'cq' < PPN *\*hea* (see Footnote 7 on p. 487)
	- e. *tū* 'demonstrative' < older Rapa Nui *tou* < *tau* (§4.6.2.1)
	- f. *ki* 'purpose marker' < PPN *\*kia* (§11.5.3)
	- g. Another possible example is *ꞌo* 'lest' <? PPN *\*ꞌaua* 'neg. imperative' (see Footnote 25 on p. 554).

The opposite process occurs in *toa* 'sugarcane' < PPN *\*tō*, and *roe* 'ant' < PPN *\*rō* (though cf. Pa'umotu *rōe*).

**Elision** Some words with identical vowels in the penultimate and final syllables have a reduced variant in which the final consonant is elided:

	- b. *rovaꞌa* / *rovā* 'to obtain'
	- c. *pūtītī / putī* 'blistered'
	- d. *ꞌana / ꞌā* 'continuity marker'

<sup>37</sup> Blixen (1972: 10) notices a few cases of *n* > *ŋ* after *i*, though none of them are certain, e.g. *mahiŋo* 'people with common bond' ~ Tongan *mahino* 'distinguished'.

<sup>38</sup> Notice that *kona*, with *n* rather than *ŋ*, is also found in Mangarevan ('bed; dwelling'); cf. also PNP *\*kona* 'nook, corner'.

2.5 Phonological processes

### **2.5.3 The phonology of borrowings**

As discussed in §1.4, Rapa Nui has incorporated numerous borrowings, especially from Tahitian and Spanish. It is well known that borrowings are often adapted to the phonological structure of the recipient language, both in phoneme inventory and in phonotactics (Tent & Geraghty 2004; Matras & Sakel 2007). The degree of adaptation may vary within a language,<sup>39</sup> depending for example on:


This also happens in Rapa Nui, as illustrated below. Borrowings from Spanish and from Tahitian will be discussed separately. Rapa Nui has also incorporated some words from other European languages (English, French); these follow the same general principles as borrowings from Spanish.

#### **2.5.3.1 Borrowings from Spanish**

This section deals with codified borrowings, loanwords which are commonly used and have become part of the language. Codified borrowings should be distinguished from spontaneous borrowings, such as the following:

(50) *Cincuenta* cincuenta *matahiti* year *o* of *te* art *hāipoipo,* marry *paꞌi.* in\_fact 'The wedding was fifty years ago, in fact.' [R415.498]

Spontaneous borrowings are instances of code switching, even though they involve just a single word (cf. Fischer 2007). They are inserted without phonological adjustments. Codified borrowings, on the other hand, tend to be adapted to Rapa Nui phonology to a greater or lesser degree. This adaptation does not follow hard and fast rules; the same word may be adjusted in various degrees and various ways. For example, *olvida* ('forgets') may be pronounced as *orvida*, *orvira*, *orovida* or *orovira* (Makihara 2001b: 195). This means that the adjustments described below may or may not apply in individual cases, depending on the factors mentioned above.

<sup>39</sup> See Sakel (2007: 17); Mosel (2004) on Samoan; Fischer (2007) and Makihara (2001b) on Rapa Nui.

<sup>40</sup> Puristic attitudes are widespread in Polynesian languages, especially where languages are perceived as endangered. This may lead to the rejection of borrowings (see Harlow 2004: 154 on Māori), or increased adaptation of borrowings to the recipient language phonology. In Tahitian, there is a tendency to remove formerly accepted non-Tahitian consonants from European borrowings; in Rapa, Tahitian borrowings are consciously adjusted to the Rapa phonological system (Kieviet & Kieviet 2006; Walworth 2015b).

**2.5.3.1.1 Phoneme level** On the phoneme level, no adjustments are needed in vowel quality, as both Rapa Nui and Spanish have a five-vowel system.

In the area of consonants, on the other hand, the two languages are considerably different. Many Spanish consonants do not occur in Rapa Nui; these tend to be adjusted to Rapa Nui phonology.

**Voiceless plosives and nasals** These consonants do not need adjustment.

**Voiced plosives** The treatment of voiced plosives can be explained from their pronunciation in (Chilean) Spanish. Word-initially and after consonants, these are pronounced as plosives in Spanish. After vowels, they are pronounced as voiced fricatives; in Chilean Spanish, these tend to be very weak: they often become approximants or almost disappear. In connected speech, word-initial voiced plosives after a vowel are pronounced as fricatives as well.

In Rapa Nui, Spanish *g* is consistently adjusted to *k*; *d* is usually either adjusted to *r* or elided (the latter only after vowels); *b* is either adjusted to *v* (word-initially) or elided (word-initially before *u*; after vowels).


**Fricatives** The fricative *s* (also spelled *c* before *i/e* and *z* before *a/o/u*) is either maintained or becomes *t*; *j* (= velar fricative [x]) becomes *k* or *h*. *f* is maintained or changed to *p*.


**Affricates** The affricate *ch* ([tʃ]) becomes a plosive *t* or a fricative *s*:


**Liquids** Spanish *rr* (= trill [r]) and *r* (= flap [ɾ]) both become *r*, which is a flap in Rapa Nui. *l* is likewise adjusted to *r*:

2.5 Phonological processes

(54) *rr* > *r karetia* 'wheelbarrow' < Sp. *carretilla*; *karo* 'jug' < Sp. *jarro l* > *r rēkaro* 'present, gift' < Sp. *regalo*; *Tire* 'Chile' < Sp. *Chile*

**Other** Spanish *ll*, which is a voiced palatal approximant [j] or fricative [ʝ] in Chilean Spanish, becomes *i*: *kaio* < *callo* 'callus', *kameio* < Sp. *camello* 'camel'. After *i* it is elided: *tapatia* < Sp. *zapatilla* 'slipper'.

**2.5.3.1.2 Phonotactics** Borrowings are also adjusted to the phonotactics of Rapa Nui; this affects the syllable structure and stress pattern.

**Final consonants** Final consonants are not allowed. This is resolved by adding a final vowel, which is either *e* or identical to the previous vowel: *ꞌavione* 'airplane' < Sp. *avión*; *kōrore* 'colour' < Sp. *color*; *tampuru* 'drum' < Sp. *tambor*. Alternatively, the final consonant is elided; this happens especially with consonants such as *d* and *s/z*, which have a weak pronunciation postvocalically in Chilean Spanish: *noverā* 'news' < Sp. *novedad*; *kapatā* 'foreman' < Sp. *capataz*.

**Consonant clusters** Consonant clusters are disfavoured. Word-initial consonant clusters are not allowed, with the exception of *pr-*. Some clusters are allowed wordmedially, especially homorganic nasal + plosive: *kampō* 'countryside' < Sp. *campo*; *atrasao* 'delayed' < Sp. *atrasado*; *rentara* 'apron' < Sp. *delantal*.

Clusters can be resolved by vowel epenthesis: *ꞌaramā* 'army' < Sp. *armada*; *karesone* 'underwear' < Sp. *calzón*; *kurua* 'crane' < Sp. *grúa*, *parata* 'silver' < Sp. *plata*. Another strategy is consonant elision; this is especially common with nasals or continuants preceding another consonant: *ꞌātā* 'until' < Sp. *hasta*; *rito* 'ready' < Sp. *listo*; *matakia* 'butter' < Sp. *mantequilla*; *tēnero* 'calf' < Sp. *ternero*.

**Word shortening** Long words are somewhat disfavoured; some words are shortened by elision of an unstressed syllable: *apenti* 'appendix' < Sp. *apéndice*; *tafate* 'dish' < Sp. *azafate*; *rentara* 'apron' < Sp. *delantal*; *pīnere* 'longline fishing' < Sp. *espinel*.

**Vowel lengthening** Sometimes, vowels are lengthened. This may serve to keep the stress in the same position: *kā*ˈ*pē* (not *\*ˈkape*) 'coffee' < Sp. *ca*ˈ*fé*; *nove*ˈ*rā* 'news' < Sp. *nove*ˈ*dad*; *pā*ˈ*rē* 'wall' < Sp. *pa*ˈ*red*. However, there are also cases where no adjustments are made to prevent stress shift: *pērī*ˈ*kura* 'movie' < Sp. *pe*ˈ*lícula*.

In other cases, lengthening may serve to avoid degenerate feet, conforming the word to a preferential metrical pattern (§2.3.2). For example, the antepenultimate vowel is lengthened in *mūseo* 'museum' < Sp. *museo*.

In yet other words, the reasons for lengthening are unclear. In four-syllable words, there is a tendency to lengthen the first two vowels, creating a HHLL pattern: *ꞌōpītara* 'hospital' < Sp. *hospital*; *ꞌāpōtoro* 'apostle' < Sp. *apóstol*. This happens even though LLLL is a common pattern in the language (§2.3.2). Lengthening may even shift the stress with respect to the Spanish original: *kara*ˈ*pā* 'tent' < Sp. ˈ*carpa*; *Kiri*ˈ*tō* 'Christ' < Sp. ˈ*Christo*.

#### 2 Phonology

#### **2.5.3.2 Borrowings from Tahitian**

Most borrowings from Tahitian do not need any phonological adjustment: all Tahitian phonemes are also part of the Rapa Nui phoneme inventory, with the exception of *f* (see §2.2.1). In fact, borrowings from Tahitian are often not perceived as borrowings at all.

In some words, *f* is retained (*fata* 'altar'< Tah. *fata*), but more commonly, it becomes *h*: *haraoa* 'bread' < Tah. *faraoa*; *hauhaꞌa* 'value'< Tah. *faufaꞌa*.

Some long vowels are shortened: *hohoꞌa* < Tah. *hōhoꞌa* 'image'. Shortening may serve to avoid an illicit metrical pattern: *tane* < Tah. *tāne* 'male' (§2.3.2).

Glottals are usually preserved, but in a number of words, they are elided: *hāpī* 'to learn' < Tah. *haꞌapiꞌi*; *haꞌamaitai* < Tah. *haꞌamaitaꞌi* 'to bless'.

Occasionally, vowels are modified: *haꞌamuri* 'to worship' < Tah. *haꞌamori*; *manaꞌu* 'to think' < Tah. *manaꞌo*; *mareti* 'plate' *<* Tah. *merēti* (§2.5.2).

Even when the phonemic content of Tahitian borrowings is exactly retained, borrowing may involve phonotactic shifts, especially because Rapa Nui differs from Tahitian in diphthongisation and stress placement (see Footnote 11 on p. 37): *māuiui* [ˌmaːuiˈui] 'sick' < Tah. *māuiui* [ˈmaːᵘiᵘi]; *pāpaꞌi* [ˌpaːˈpaɁi] 'to write' < Tah. *pāpaꞌi* [ˈpaːpɁaⁱ]; *haraoa* [ˌharaˈoa] 'bread' < Tah. *faraoa* [faˈra(ː)ᵒa] (< Eng. 'flour').

### **2.6 Reduplication**

Reduplication is a process whereby all or part of the root is copied and prefixed or suffixed to the root. The copied part of the root is called the base; the copy is called the reduplicant. In the following example, the root is placed between brackets; the base is underlined, while the reduplicant is in bold:

(55) taŋi *taŋi* 'to cry' > **ta**[taŋi] *tataŋi* 'to cry (plural)'

In Rapa Nui orthography the reduplicant and the root are separated by a hyphen, a practice not adopted in this grammar (§1.4.4).

Reduplication is very common in Rapa Nui, just as it is in Polynesian languages in general. It occurs with many verbs (including adjectives) and is productive, to the extent that it is even applied to borrowings.<sup>41</sup> Nouns are generally not reduplicated, though a nominal root may be reduplicated to form a denominal verb, while a few verbal roots are reduplicated to form a deverbal noun (§2.6.2.2 below).

The patterns of reduplication will be discussed in §2.6.1, while §2.6.2 deals with the functions of reduplication. §2.6.3 briefly discusses reduplications for which the base form does not exist independently.

<sup>41</sup> Bob Weber (p.c.) once heard someone commenting at the telephone exchange that the line was engaged all the time: *ko okuokupao ꞌā* (< Sp. *ocupado* 'occupied, engaged'). Makihara (2001b: 198) gives an example of *kamikamiare* (*kamiare* 'to change' < Sp. *cambiar*).

2.6 Reduplication

### **2.6.1 Patterns of reduplication**

Although there is a wide variety of reduplication patterns, all of these can be reduced to two types:


These two patterns will be referred to as type 1 and 2, respectively. They will be discussed in turn in the following subsections. They are analysed using concepts from prosodic morphology (McCarthy & Prince 1995; 1996; Inkelas & Zec 1995), which allows segmental content and prosodic structure to be subject to distinct processes and/or constraints. This allows reduplication to be described in terms of prosodic structure (i.e. feet, syllables and morae), even though the segmental content affected does not necessarily coincide with prosodic constituents, and may vary in size and shape.

#### **2.6.1.1 The morphology of type 1 reduplication**

Type 1 reduplication occurs with a good number of bisyllabic verbs, as well as a few trisyllabic verbs and – to my knowledge – one quadrusyllabic verb.<sup>43</sup>

Table 2.12 illustrates the different patterns in terms of light (L) and heavy (H) syllables. Each pattern will be discussed below.

As Table 2.12 shows, for most bisyllabic words the reduplicant is a short syllable, i.e. it is an exact copy of the first syllable of the root.<sup>45</sup> For some verbs, however, the vowel of the reduplicant is lengthened. The choice between the two patterns is lexically determined: there is no difference in function between both patterns, nor is there a phonological motivation for the choice.

Both patterns can be accounted for by stating that type 1 reduplication adds one mora to the root. This mora must be integrated to the prosodic structure, which means that an additional foot is added to the word. This is illustrated in the following structure:<sup>46</sup>

<sup>42</sup> This means that there is no principled distinction between full and partial reduplication; cf. Blust (2001: 39). (Davletshin 2015 does take a full/partial distinction as primary.)

<sup>43</sup> Possibly type 1 reduplication also occurs with a few monosyllabic verbs: *kīkī* 'to say repeatedly' can be analysed as reduplication of the initial mora + secondary lengthening. However, the function of this form (iterative, not plural) suggests that this is a case of type 2 reduplication. The same is true for other reduplicated monosyllabic roots.

<sup>44</sup> When the root is vowel-initial, the reduplication contains a bisyllabic VV-sequence, which is not merged to a single long vowel.

<sup>45</sup> As discussed in sec. §2.3.2, the first syllable of bisyllabic words is always short.

<sup>46</sup> For sake of conciseness, the PrWd level is not included in the structure trees in this section.

#### 2 Phonology

(a) L L > L L L *ꞌara* 'to wake up' *ꞌaꞌara* 'to wake up (pl)' *eke* 'to mount' *eeke* 'to mount (pl)'<sup>44</sup> *haꞌi* 'to embrace' *hahaꞌi* 'to embrace (pl)' *rahi* 'much' *rarahi* 'important' *rehu* 'to forget' *rerehu* 'to faint' *turu* 'to go down' *tuturu* 'to go down (pl)' (b) L L > H L L *mate* 'to die' *māmate* 'to die (pl)' *piko* 'to hide' *pīpiko* 'to hide (pl)' *tere* 'to run' *tētere* 'to run (pl)' (c) L L L > H L L *haꞌuru* 'to sleep' *hāꞌuru* 'to sleep (pl)' *haꞌere* 'to walk' *hāꞌere* 'to walk (pl)' *tahuti* 'to run' *tāhuti* 'to run (pl)' (d) L L L L > L L L L L *paŋahaꞌa* 'heavy' *papaŋahaꞌa* 'heavy (pl)'

Table 2.12: Patterns of type 1 reduplication

The initial foot only has one mora, i.e. it is degenerate. Word-initial degenerate feet are allowed in Rapa Nui, but there is pressure towards a pattern of whole feet (§2.3.2). For some words, this leads to the addition of a second mora to the initial foot, which is filled by spreading the first vowel:

#### 2.6 Reduplication

For trisyllabic roots, a mora is added to the existing degenerate foot; no additional foot is needed. Moreover, no segmental content needs to be added, as the additional mora can be filled by spreading the initial vowel of the root:<sup>47</sup>

We may conclude that, even though the surface result of reduplication is quite different for trisyllabic roots than for bisyllabic roots, both can be analysed as involving the same process: addition of a single mora to the root. Another indication that both groups of words involve the same type of reduplication, is that in both cases the most common function of reduplication is plurality: *hāꞌere* is the plural of *haꞌere*, just like *tuturu* is the plural of *turu*. This will be discussed in more detail in §2.6.2.1 below.

The only example of a quadrusyllabic word shows the same mechanism at work: a mora is added, resulting in an additional degenerate foot, which is filled with a copy of the initial syllable of the root: *paŋahaꞌa* > *papaŋahaꞌa*.

#### **2.6.1.2 The morphology of type 2 reduplication**

Type 2 reduplication has the following features:


The patterns of type 2 reduplication are listed in Table 2.13. These patterns are united by a single feature: the addition of a foot to the word, which is filled in some way by copying two morae from the root. The different patterns are discussed in turn below.

<sup>47</sup> With bisyllabic roots, spreading of the vowel (*rahi* > \**rāhi*) is not possible, as the resulting long vowel would cross a foot boundary, creating an impossible prosodic pattern (§2.3.2).

#### 2 Phonology

Table 2.13: Patterns of type 2 reduplication


**Bimoraic words (patterns a–b)** For bimoraic words (whether mono- or bisyllabic), prefixing and suffixing yield the same result. In both cases the whole root is copied, resulting in a two-foot word. Below are examples of reduplications of H and LL words (here prefixing is assumed, cf. Footnote 50 on p. 66):

**Trisyllabic words (patterns c–f)** For trisyllabic LLL and HLL words, the pattern is more intricate. The relevant data are repeated in Table 2.14. <sup>48</sup> As this table shows, regardless the length of the root vowels, in prefixing forms all vowels are short, while in suffixing forms the first vowel is long. These data can be accounted for by the following constraints:

<sup>48</sup> The same patterns of lengthening and shortening also occur in Māori; Meyerhoff & Reynolds (1996: 148) give examples of patterns d–f.

#### 2.6 Reduplication


Table 2.14: Type 2 reduplication of trisyllabic roots

#### *Non-violable:*


#### *Violable:*

4. All feet are complete.

Constraint 4 is a general soft constraint in Rapa Nui (§2.3.2) which can be fulfilled – if possible – by vowel lengthening. But in type-two reduplications, the data show that the possibilities of lengthening are limited (constraint 3): only the first vowel of the root may be lengthened (*hāꞌereꞌere*), not the first vowel of the reduplicant (*\*hāꞌehaꞌere*). Moreover, the first root vowel is lengthened only word-initially, not when it is preceded by the reduplicant (*\*haꞌehāꞌere*).<sup>49</sup> When vowel lengthening is not possible, the initial foot is

(i) (ha ꞌe)<sup>F</sup> [ (ha)<sup>F</sup> (ꞌere)<sup>F</sup> ]

(ii) \*(vana)<sup>F</sup> [ (vā)<sup>F</sup> (naŋa)<sup>F</sup> ]

Another reason not to adopt this analysis, is that some speakers put secondary stress on the second vowel: [haˌʔehaˈʔere]. This suggests a foot structure where the second syllable is prominent, i.e. foot-initial:

(iii) (ha)<sup>F</sup> (ꞌeha)<sup>F</sup> (ꞌere)<sup>F</sup>

Pattern (i) is proposed for derivations like *haapai* > *hapahapai* in Māori by Meyerhoff & Reynolds (1996: 161); in their analysis, *\*hapahaapai* would violate a correspondence constraint which requires that every element in the reduplicant has a correspondent in the base. Notice that Māori is metrically different from Rapa Nui: degenerate feet are disallowed, and main stress falls on the leftmost foot.

<sup>49</sup> An alternative option to account for *ha ꞌeha ꞌere* would be, to assume that the boundary of the root coincides with a foot boundary, so that the initial foot is complete, while the second foot is degenerate:

But this would mean assuming an otherwise unattested pattern containing a non-initial degenerate foot. Moreover, it would raise the question why the root-initial vowel of *vanavanaŋa* (based on the foot *vānaŋa*) is short, rather than long; one would expect:

#### 2 Phonology

degenerate, in accordance with the following non-violable constraint in the language (§2.3.2):

5. All non-initial feet are complete; the initial foot may be degenerate.

Though constraint 3 may seem to be somewhat arbitrary, it corresponds to a general tendency in Rapa Nui: the statistics in §2.3.2 show that long vowels are much more common word-initially than in other positions; moreover, they are very rare when surrounded by short vowels. (*\*haꞌehāꞌere* would correspond in prosodic structure to *manupātia* 'wasp', one of the few LLHLL words.)<sup>50</sup>

The constraints under discussion result in the following structures for LLL words:

For HLL words the situation is identical, except that the root consists of two complete feet. Interestingly, the length of the initial vowel is not carried over into the reduplication. This is somewhat surprising, as in other cases long vowels remain long under type 2 reduplication (see (a) above and (g) below).

<sup>50</sup> These constraints may explain why suffixing is much more common with these words than prefixing, even though in other cases where prefixing and suffixing can be distinguished (type 1 reduplication; type 2 for quadrumoraic words) only prefixing occurs: suffixing allows the initial vowel to be lengthened (constraint 3), so the word consists of three whole feet (satisfying constraint 4); on the other hand, prefixing results in a degenerate foot. In general, prefixing reduplication is much more common in Polynesian, and in Austronesian in general (Finney 1999).

2.6 Reduplication

**Quadrumoraic words (patterns g–h)** Finally, the reduplication of quadrumoraic words (LLH words like *ꞌauē* 'cry', LLLL words like *tokerau* 'wind') is illustrated below. In both cases, a complete foot is added, which is filled segmentally by copying the first two syllables of the root:

Occasionally, type 1 and type 2 reduplication are applied in sequence: the result of type 1 reduplication serves as the base of type 2 reduplication. This is only attested with a few LL roots; the process can be described as follows:

(56) µ<sup>1</sup> µ<sup>2</sup> > **µ<sup>1</sup>** [µ<sup>1</sup> µ2] > **µ<sup>1</sup> µ<sup>1</sup>** [µ <sup>1</sup> µ<sup>1</sup> µ2]

The result is a form in which the initial syllable of the root is repeated four times. A few examples:

(57) *ꞌuri* 'black' > *ꞌuꞌuꞌuꞌuri* 'black (many things)' *tea* 'white' > *tetetetea* 'white (many things)' *kikiu*<sup>51</sup> 'to shriek, squeak' > *kikikikiu* 'to shriek again and again'

#### 2 Phonology

### **2.6.2 Functions of reduplication**

The basic function of type 1 reduplication of a verb is expressing plurality of its S/A argument; the basic sense of type 2 reduplication is repetition or intensity.<sup>52</sup> However, in both cases exceptions and lexicalised meanings are not uncommon. Both types are discussed in turn below.

#### **2.6.2.1 Type 1: plurality**

The sense of type 1 reduplication is lexically determined. For most verbs, it indicates a plural (i.e. more than one) S/A argument. Some examples:


Most verbs do not have a plural form at all.<sup>53</sup> For those verbs that do have a plural form, its use is not obligatory – in other words, the base form is not limited to singular argument. In (61) the basic form *tuꞌu* is used, even though a plural form *tutuꞌu* exists.

(61) *He* ntr *tuꞌu* arrive *mai* hither *tou* dem *ŋā* pl *uka* girl *era.* dist 'Those girls arrived.' [Blx-3.053]

Some type 1 reduplications have a different sense; this is lexically determined, hence unpredictable.


As this list shows, for a few of these verbs the base form is intransitive, while the reduplicated form is transitive. Here is a pair of examples:

<sup>52</sup> The same is true in Polynesian languages in general, see Finney (1999).

<sup>53</sup> The lexical database includes 56 plural forms with type 1 reduplication, on a total of over 3500 verbs and adjectives.

2.6 Reduplication


#### **2.6.2.2 Type 2: iterativity and intensity**

Like type 1 reduplication, type 2 mainly affects verbs, though unlike type 1, there are some examples where a noun is involved; the latter mostly concerns cases of denominal verbs (see below). However, only for verbs is reduplication productive. Its function depends largely on the nature of the verb.<sup>54</sup>

**2.6.2.2.1 Repetition** Type 2 reduplication often adds an element of repetition to the event expressed by the verb:


For some verbs, reduplication indicates repetition of the parts or stages making up the event, rather than the event as a whole:<sup>55</sup>


**2.6.2.2.2 Distributive** Repetition of the event may imply a distributive reading, involving different participants: the event happens repeatedly, each time affecting a different Patient or Recipient, or performed by a different Agent.

<sup>54</sup> Johnston (1978), after a detailed lexical study, concludes that reduplication in Rapa Nui indicates 1) repetition; 2) quantification (of the subject); 3) duration; 4) the degree of vigour in which the action is carried out. I have not found any case where (3) is the sole factor involved; whenever reduplication may be taken as indicating duration, this is usually by virtue of iterativity. "quantification" may involve either the subject (usually with type 1, but occasionally with type 2) or the object; see below in this section.

<sup>55</sup> Haji-Abdolhosseini, Massam & Oda (2002), describing reduplication in Niuean, use the term *phase repetition*.

#### 2 Phonology


With plural Agents, the sense of the verb may seem to be similar to a type 1 reduplication. However, the type 2 reduplication still refers to a series of separate events: each Agent performs the action individually (possibly at different times), not as a group. In the following example, *tuꞌutuꞌu* expresses multiple events of arriving, i.e. different ships arriving at different occasions. The plural *tutuꞌu* (type 1) would imply that different ships arrived in a single event.

(68) *Mai* from *te* art *taŋata* man *nei* prox *i* pfv *haꞌamata* begin *ai* pvp *i* pfv *tuꞌutuꞌu* arrive:red *mai* hither *ai* pvp *te* art *pahī* ship *papaꞌā.* foreign

'Starting with this man (=the explorer Jacob Roggeveen), foreign ships started to arrive (on Rapa Nui).' [R111.014]

The choice between mere repetition and a distributive reading results to some degree from the semantics of the verb. Transitive verbs are more likely to have a distributive sense: repetition of a transitive event will often affect different objects. However, the precise meaning of the reduplication is not lexically specified, but may vary depending on the context. The two examples below show different uses of type-2 reduplication of *uꞌi* 'to look'. While in (69) *uꞌiuꞌi* has an iterative sense, in (70) it is distributive (and effectively reciprocal).

(69) *Pē* like *rā* dist *ꞌā* ident *e* ipfv *uꞌiuꞌi* look:red *era* dist *a* by *tuꞌa* back *koiꞌite* perhaps *e* ipfv *tute* chase *rō* emph *mai* hither *e* ag *tū* dem *ꞌamahiŋo* evil\_person *era* dist *ko* prom *Makoꞌi.* Mako'i 'Like that he kept looking behind him, to see if he was followed by that wicked

Mako'i.' [R214.038]

(70) *He* ntr *uꞌiuꞌi* look:red *ia* then *te* art *ꞌāriŋa* face *a* prop *totoru.* red:three 'The three looked at each other.' [R313.005]

2.6 Reduplication

**2.6.2.2.3 Intensity** With many adjectives, reduplication signifies increased intensity:


However, reduplication of an adjective does not always imply intensity: *rivariva* 'good', not 'very good'. See §2.6.2.2.5 below.

**2.6.2.2.4 Lexicalised meanings** For some verbs, the sense of the reduplicated form is lexicalised and unpredictable, even though it is obviously related to the meaning of the root.


In some cases the meaning of the reduplication, even though lexicalised, is clearly derived from an iterative sense. In the case of *mānaꞌunaꞌu* the specialised sense 'to be worried' developed from the iterative sense 'to think much'. (In fact, *manaꞌu rahi* 'think much' is used with a similar sense.)

**2.6.2.2.5 Reduplication as basic form** For certain words, the reduplicated form is more common than the root. In these cases, the simple form is often limited in use. This is especially common with adjectives (§3.5.1.2): *nuinui* 'big' is much more common than *nui* 'big', which is used in limited contexts. For other words, the simple form is not in use at all; these are discussed in §2.6.3 below.

**2.6.2.2.6 Conversion** A number of reduplications are denominal verbs or adjectives:


A few reduplications are deverbal nouns:


#### 2 Phonology

**2.6.2.2.7 Attenuative** Finally, reduplication may have an attenuative function, implying a certain weakening. Iterativity may mean that the event – or a phase of the event – takes place repeatedly, but each time to a small extent:


With adjectives, the reduplication may indicate a weaker, 'more or less' sense. I have found this sense only with one adjective; it is probably not accidental that in this case, an intensified sense ('very cooked') does not fit in well with the semantics of the word.

(76) *mata* 'ripe; cooked' *matamata* 'half-ripe; half-cooked'

### **2.6.3 Reduplications without base form**

There are a number of type 2 reduplications for which the base does not occur on its own. Most of these are either nouns or adjectives with a bisyllabic base: *hiohio* 'strong', *kutakuta* 'foam', *rairai* 'thin, flat', *naonao* 'mosquito', *rohirohi* 'tired', *tokotoko* 'walking stick'. Examples with a trisyllabic base are *māuruuru* 'to thank; thank you' and *māuiui* 'sick'. Sometimes there is evidence that the simple form did exist in Rapa Nui: *paka* 'dry' is found in older texts, but in modern Rapa Nui only *pakapaka* is used. Other forms (e.g. *naonao* and *māuruuru*) were borrowed as a whole from Tahitian.

Some of these reduplication-only forms have a plural of type 1, based on the root: *kaokao* 'narrow', *kakao* 'narrow (Pl)'; *kaꞌikaꞌi* 'sharp', *kakaꞌi* 'sharp (Pl)'.<sup>56</sup>

There are also verbs which have the shape of a type 1 reduplication (*σ*<sup>1</sup> *σ*<sup>1</sup> *σ*2), but for which the non-reduplicated form does not occur: *ꞌaꞌaru* 'to grab', *totoi* 'to drag', *nēneꞌi* 'to defecate', *nonoꞌi* 'to ask, beg'. For some of these, it is clear that the base form was known in the past: *neꞌi* 'defecate' occurs in Englert's dictionary, *toi* is found once in an older text, but neither is used nowadays. For other verbs such as *ꞌaꞌaru*, the base form is not attested at all. Even so, they are treated as reduplications in the accepted orthography (i.e. they are written with a hyphen), because a type 2 reduplication of the same base does exist with a typical type 2 sense such as iterativity. For example, while there is no simple form \**ꞌaru*, there is a type 2 reduplication *ꞌaruꞌaru* 'to grab several things'; hence, *ꞌaꞌaru* is considered a type 1 reduplication and written with a hyphen (*ꞌa-ꞌaru*).

In fact, most words with identical first and second syllables can be considered reduplications for one of the reasons above. Exceptions are e.g. *ꞌaꞌamu* 'story' (neither *\*ꞌamu* nor

<sup>56</sup> Interestingly, this leads to a situation where the plural is shorter than the corresponding singular (cf. Blust 2001: 40).

*\*ꞌamuꞌamu* is attested), *rarama* 'inspect' (there are no related lexemes *\*rama* or *\*ramarama*), and *tātara* 'to make a speech' (there are no related lexemes *\*tara* or *\*taratara* in Rapa Nui, though PPN *\*tala* 'to talk; story' has reflexes in many other languages).

### **2.7 Conclusions**

The preceding sections have given an overview of Rapa Nui phonology. The phoneme inventory of Rapa Nui is small (10 consonants, 5 short and 5 long vowels) and closely reflects the phoneme inventory of Rapa Nui's protolanguages. The glottal plosive is contrastive in lexical words, both word-initially and word-medially, but acoustic analysis shows that it is not contrastive phrase-initially. This means that it is not contrastive in certain prenuclear particles; the latter tend to have a glottal only when they occur at the start of a prosodic phrase.

The syllable structure of Rapa Nui is (C)V(ː). There are no (C)V1V<sup>2</sup> syllables: sequences of non-identical vowels are analysed as disyllabic. One argument for this is stress assignment: the second vowel of a VV sequence may be stressed, which shows that it does not form a syllable with the preceding vowel. Another argument is word structure. Rapa Nui has a strict constraint on the metrical structure of words: long (i.e. heavy) syllables cannot be followed by an odd number of morae; in other words, the penultimate syllable cannot be long when the final syllable is short. This means that a word like *mauku* 'grass' must be trisyllabic, as a long penultimate syllable *mau-* would be metrically impossible.

Stress – both on word and phrase level – falls on the penultimate mora; in connected speech, stress is assigned on the phrase level. Interestingly, all postnuclear elements are (minimally) bimoraic, which avoids a possible conflict between word and phrase stress.

Two phonological processes which are regular but optional, are word-final vowel devoicing and pre-stress lengthening. The former is especially common.

A wide range of sporadic sound changes can be detected in the lexicon, resulting either in variants within Rapa Nui, or irregular reflexes of protoforms. Metathesis is rampant; other sound changes especially affect vowels, glottals and the liquid /r/.

Borrowings – especially from Spanish – tend to be adjusted to the phonology of Rapa Nui, but in various ways and to varying degrees. Some non-native phonological features are more liable to be accepted (hence not adjusted) than others, especially certain wordmedial consonant clusters and the fricative /s/.

Finally, this chapter deals with reduplication. Two basic types can be distinguished: monomoraic (expressing plurality) and bimoraic (expressing repetition or intensity). Reduplication may be full or partial, but there is no principled distinction between the two: whether all or part of the root is reduplicated, simply depends on the size of the root.

Depending on the prosodic shape of the root, various processes of lengthening and shortening take place; these can be explained by metrical constraints, most of which correspond to general phonological tendencies in the language.

#### 2 Phonology

Further research could throw more light on the following areas:


## **3 Nouns and verbs**

### **3.1 Introduction: word classes in Rapa Nui**

This chapter and the next deal with the description of word classes. In this area, the most basic distinction in Rapa Nui – as in other Polynesian languages – is that between full words and particles.<sup>1</sup> Full words occur in the nucleus of a phrase and mostly form large, open classes (though certain types of full words, such as locationals, are closed classes). Particles are a closed class: they can be exhaustively listed. They occur in fixed positions before or after the nucleus, and most of them are highly frequent.

In Rapa Nui, full words and postnuclear particles have a minimal length of two morae; prenuclear particles may be one mora.

Pro-forms have an intermediate status between full words and particles. Like full words, they occur in the nucleus of a phrase and can be preceded and followed by particles. Unlike full words, they do not have a lexical meaning, and like particles, they form a closed class. Pro-forms include personal, possessive and benefactive pronouns, as well as interrogative words.

Two other intermediate categories are the negator *ꞌina* and the numerals. Both of these form a closed class, yet they function as phrase nuclei, as they can be followed by postnuclear particles, while numerals are also preceded by a particle.

Full words can be divided into word classes (parts of speech) on the basis of grammatical and semantic criteria. Some word classes can be defined by a single unambiguous criterion. These include the following:


For verbs and common nouns the situation is much less clear. In §3.2, the distinction between nouns and verbs is discussed, and various aspects of their interaction are explored.

<sup>1</sup> Buse (1965) uses these same terms for Rarotongan. Biggs (1961) uses the terms "bases" and "minor morphemes" for Māori; in Biggs (1973) the latter term has been replaced by "particles". Mosel & Hovdhaugen (1992: 71) distinguish full words, particles, proforms and interjections in Samoan.

<sup>2</sup> By contrast, quantifiers (§4.4) cannot be grouped together as a word class on the basis of distributional criteria, as different quantifiers show a different distribution.

#### 3 Nouns and verbs

The remainder of this chapter discusses other issues concerning nouns (§3.3) and verbs (§3.4).<sup>3</sup> §3.5 discusses adjectives, a subclass of verbs, while §3.6 discusses locationals, a subclass of nouns. Other – minor – word classes will be discussed in Chapter 4.

### **3.2 Nouns and verbs**

There are three types of nouns in Rapa Nui: common nouns, proper nouns and locationals. Proper nouns and locationals are easily distinguished from other types of nouns and from other word classes, as indicated above. For common nouns, the distinction with other parts of speech – especially verbs – is less obvious. This section deals with the noun/verb distinction in Rapa Nui; in this discussion, *noun* should be read as a shorthand for *common noun*. §3.2.1 deals with the question whether there is a distinction between nouns and verbs in the lexicon of Rapa Nui; the existence of this distinction has been denied for some Polynesian languages. As I will show in §3.2.1.1, there are various types of evidence suggesting that Rapa Nui does have a distinction between lexical nouns and verbs. A given word may seem to be both noun and verb, but in most cases the two are either lexically distinguished (often with unpredictable relationships between nominal and verbal meanings), or the word is primarily a verb which may enter in nominalised constructions. In §3.2.1.2 I propose an analysis in terms of prototypes; this analysis captures the fact that there is a tendency to congruence of form and function: verbal meaning tends to go together with verbal syntax, nominal meaning with nominal syntax. At the same time, various non-prototypical types also occur: words and constructions having features of both nouns and verbs. The latter are discussed in the next subsections (§3.2.2–3.2.4). Finally, §3.2.5 brings together evidence for a general tendency in Rapa Nui to maximise the nominal domain.

### **3.2.1 The noun/verb distinction**

Polynesian languages are known to be very flexible in use of nouns and verbs: many words seem to be used both as nouns and verbs. This is also true for Rapa Nui. In (1) below, *poki* 'child' occurs in a noun phrase (preceded by the article *te*) which is subject of the clause; in (2), it occurs in a verb phrase (preceded by the imperfective marker *e*) which is the clause predicate:


<sup>3</sup> See also Chapters 5 and 7 on noun and verb phrases, respectively.

3.2 Nouns and verbs

Likewise, in the following examples, *ꞌaꞌamu* is first used as a noun 'story' (in a noun phrase functioning as clause subject), then as a verb 'to tell' (in a verb phrase functioning as clause predicate):


'We told about all the problems we have on Rapa Nui.' [R649.238]

Like all Polynesian languages, Rapa Nui has hundreds of words which, like *ꞌaꞌamu*, are defined both as a noun and a verb (These will be discussed in more detail in §3.2.2). Moreover, there is no inflectional morphology in the language which would facilitate distinguishing nouns from verbs. It is therefore not surprising that the existence of a lexical noun/verb distinction in Polynesian languages has been denied.<sup>4</sup> In such an approach, the terms *noun* and *verb* are used in a purely syntactic sense: whatever occurs in the nucleus of a noun phrase is a noun, whatever occurs in the nucleus of a verb phrase is a verb. Such a distinction is workable as there is a strict distinction between nominal and verbal phrases,<sup>5</sup> a distinction which also applies in Rapa Nui.

Nevertheless, I will argue that there are good reasons to maintain a lexical distinction between noun and verbs. That is, words are defined as noun or verb in the lexicon. This does not mean that all occurrences of these words are completely and unambiguously nominal and/or verbal. Lexical verbs very commonly enter into constructions which have certain nominal features; less frequently, lexical nouns are used in constructions with certain verbal features (as in (2) above). Moreover, many words are specified as both noun and verb in the lexicon, as *ꞌaꞌamu* in (3–4) above.

§3.2.1.1 lists reasons to maintain a lexical distinction between nouns and verbs. In addition, several reasons are given why a syntactic approach to the noun/verb distinction is unsatisfactory. §3.2.1.2 proposes a definition of nouns and verbs in terms of prototypes. This approach maintains a lexical distinction between noun and verbs, while at the same time recognising that the two cannot always be unambiguously distinguished.

<sup>4</sup> See e.g. Mosel & Hovdhaugen (1992) for Samoan, Lazard & Peltzer (1991; 2000) for Tahitian, and Elbert & Pukui (1979) for Hawaiian. A similar approach recognises a large class of "generals" (Biggs 1961) or "universals" (Biggs 1973), besides smaller classes of (pure) nouns and verbs. See e.g. Buse (1963b; 1965) for Rarotongan, Tchekhoff (1979) for Tongan and Biggs (1961; 1973) for Māori. See Vonen (2000) for an overview of the different approaches.

<sup>5</sup> In some languages the distinction is not as strict. Moyse-Faurie (2005: 168) points out that in East Futunan, aspect markers and articles may co-occur.

#### 3 Nouns and verbs

#### **3.2.1.1 Reasons to maintain a lexical noun/verb distinction**

Firstly: despite the flexibility in the use of nouns and verbs, the large majority of noun phrases have a nucleus denoting an entity, while an overwhelming majority of verb phrases have a nucleus denoting an event. While all action words can be used in nominal phrases, many entity words are never used in verb phrases, or only in very specific, uncommon contexts. For example, the word *oho* 'go' is very often preceded by the imperfective marker *e*, but the word *taŋata* 'man' is never preceded by this particle in the text corpus. Other words, like *ꞌaꞌamu* in (3–4) above, are commonly used both as noun and as verb, but with a different sense. Somehow generalisations like these should be accounted for in the grammar. To assume one large class of words, which can be indiscriminately slotted into noun or verb phrases, does not do justice to actual usage.

A second reason not to abandon the notion of nouns and verbs in the lexicon, is that the semantic relationship between nouns and verbs is not always predictable. In other words, it is not always possible to derive the nominal and verbal meanings of a word from an underlying acategorial sense. This will be illustrated in §3.2.2. This could be accounted for by analysing nouns and verbs of the same form as homophones (i.e. separate lexical items), but in that case the relationship between corresponding nouns and verbs is lost: under a homophone analysis, a lexical item used in a noun phrase is unrelated to an identical-sounding item with a related meaning in a verb phrase. This is unsatisfactory from a semantic point of view, for even though the relation between nominal and verbal senses may be unpredictable, the senses are always clearly related.

A third argument that the apparent freedom of use does not imply the absence of lexical categories, comes from a rare phenomenon: very occasionally, words from other (minor) word classes are used as a noun or a verb. For example, a pronoun may occur in the nucleus of a verb phrase as in (5); demonstrative particles may be the nucleus of a verb phrase, as in (6):<sup>6</sup>


'—Is it finished? —I'm done (lit. something like 'there is your thing').' [R230.105]

These words belong to well-defined categories (pronouns, demonstratives), so it is clear that they are not acategorial; yet they occur in a noun phrase or verb phrase. This suggests that the absence of a strict boundary between word classes can be explained by freedom of cross-categorial use rather than the absence of lexical categories: the possibility of cross-categorial use is present in the grammar anyway.

We may conclude that the distinction between lexical nouns and verb should be maintained.

<sup>6</sup> Postverbal *era ꞌā* indicates a finished action (sometimes equivalent to a perfect or pluperfect).

3.2 Nouns and verbs

In addition, there are a number of reasons why the syntactic approach to nouns and verbs common in Polynesian linguistics is unsatisactory. In this approach, nounhood and verbhood depends wholly on syntactic criteria: a word is a noun when it is the nucleus of a noun phrase, and a verb when it is the nucleus of a verb phrase. This can be further reduced to a single criterion: a word is a noun when preceded by a determiner, and a verb when preceded by an aspect marker.<sup>7</sup> There are syntactic, semantic and pragmatic problems with this assumption.

Syntactic: the presence of a determiner does not necessarily mean that the phrase is entirely nominal. The nucleus of such a phrase may control verbal arguments:<sup>8</sup> in (7) below, the subject of *vānaŋa* has the agent marker *e*; in (8), *runurunu* is followed by a direct object marked with the accusative marker *i*.


The phrase may also contain other VP elements like directionals, such as *mai* in (8). These elements do not appear in ordinary noun phrases, i.e. phrases headed by entity words like *taŋata* 'man' or *hare* 'house'. In conclusion, a phrase introduced by a determiner may still have certain VP characteristics.

Semantic: despite the presence of a determiner, the nucleus may have a verbal sense, referring to an event rather than an entity. Even though it is preceded by a determiner, the verb may therefore have a different meaning from a real noun with the same form.<sup>9</sup> This can be illustrated with the word *vānaŋa*, which may denote an action 'to talk', or an entity 'word, spoken utterance'. In (9) *vānaŋa* denotes an event and occurs in a verb phrase (preceded by the aspect marker *e*); in (10) it denotes an entity and occurs in a noun phrase (preceded by the article *te*):

(9) *E* ipfv *vānaŋa* speak *rō* emph *mai* hither *ꞌā* cont *paurō* every *te* art *mahana* day *ki* to *a* prop *au.* 1sg 'Every day he speaks to me.' [R655.018]

<sup>7</sup> For examples of this approach, see Biggs (1973: 19); Mosel & Hovdhaugen (1992: 76); Lazard & Peltzer (2000: 21).

<sup>8</sup> Waite (1994), working in a generative framework, captures this insight by proposing that D (=determiner) in Māori can take not only NP complements, but VP and AdjP as well. This means that a verb may occur in a nominal context (DP) while retaining its verbal character.

<sup>9</sup> The same point is made by Besnier (2000: 511) for Tuvaluan.

#### 3 Nouns and verbs

(10) *Ka* imp *taiꞌo* read *pūai* strong *te* art *ŋā* pl *vānaŋa* word *nei:* prox *raŋi, raŋi rano, rano rapu. rapu* 'Read the following words aloud: *raŋi, rano, rapu*.' [R616.147]

Clearly, in (9) *vānaŋa* is a verb, while in (10) it is a noun. So far, so good. In (7) above however, *vānaŋa* denotes an event, even though it is preceded by a determiner. It serves as the complement of *oti*, a verb which commonly takes a nominalised event word as complement. Thus, *te vānaŋa* in (7) is not nominal in the same way as *te ŋā vānaŋa nei* in (10). Notice that this semantic difference correlates with certain syntactic differences: in (10), *vānaŋa* is preceded by the plural marker *ŋā*, a noun phrase element; in (7) it is followed by a subject marked with the agentive *e*, something to be expected of a verb.

Pragmatic: in some constructions, a nominal phrase is syntactically not a clause predicate, yet it expresses an event and functions as a predicate pragmatically. This happens in the nominalised actor-emphatic construction, in which the actor is expressed as a possessive, while the event is expressed in a noun phrase (§8.6.3). Syntactically these constructions are nominal clauses with the nominalised verb as subject; pragmatically, however, they express an event with the verb as nucleus. One example:

(11) *ꞌĀꞌana* poss.3sg.a *te* art *haka* caus *tere* run *i* acc *te* art *henua.* land

'He (was the one who) governed the country.' [R370.005]

Constructions such as (11) are only found with event words, not with entity words. If the underlined phrases were regarded as noun phrases because of the presence of the article, they would be undistinguishable from "normal" noun phrases, which never enter into this construction.

We may conclude that it would be unsatisfactory to consider a word as noun whenever it is preceded by a determiner. Event words preceded by determiners may have either a nominal sense and nominal function, or a verbal sense and verbal function (possibly with verbal syntactic trappings). This suggests that we should make a distinction between lexical nominalisation, which turns a verb into a true noun, and syntactic nominalisation, in which a verb is used as nucleus of a noun phrase, while retaining its verbal meaning and other verbal characteristics, such as the possibility to take verbal arguments. The examples above show that both occur in Rapa Nui: (10) is an example of lexical nominalisation, while (7) is an example of syntactic nominalisation. These processes will be discussed in sections §3.2.2 and §3.2.3, respectively.

#### **3.2.1.2 Prototypicality**

As discussed in the previous section, it would be unsatisfactory to deny a basic distinction between nouns and verbs: most instances are either nominal or verbal both syntactically and semantically; the semantic relation between nominal and verbal uses of a word is often predictable, hence lexical; and the fact that words from minor (and well-defined) word classess can be used cross-categorially suggests that the freedom of use of nouns and verbs can also be accounted for as crosscategorial use, rather than by an absence of

3.2 Nouns and verbs

categories. Moreover, an alternative approach, which defines nounhood in terms of the occurrence of a noun phrase (minimially defined by a determiner) does not do justice to the often very verbal nature of noun phrases: a phrase which seems to be a noun phrase because of the presence of a determiner, may yet have a strongly verbal character. It may contain certain VP elements, while certain NP elements are excluded; it may function as a predicate; it may denote an event rather than an entity.

The nominal and a verbal domain are not divided by a sharp boundary in Rapa Nui. Rather, "verbness" and "nounness" can be conceived of as a continuum, defined in terms of prototypes: at one end there are constructions which are entirely nominal (prototypical nouns), at the other end there are constructions which are entirely verbal (prototypical verbs). In between is a range of constructions which share characteristics of both.<sup>10</sup>

As the discussion above has made clear, prototypical nouns and verbs cannot be defined solely on the basis of lexical meaning, nor solely on the basis of syntactic properties. Rather, a prototypical form combines syntactic, semantic and pragmatic characteristics. I suggest the following definitions:

A prototypical verb is a word which


A prototypical noun is a word which


<sup>10</sup> See Payne (1997: 34–38) for discussion of prototypicality in word classes. Croft (2000) defines word classes as unmarked combinations of a pragmatic function and a lexical semantic class:


Other combinations are possible: an object word may be used in predication (predicate nominals), action words may be used as modifier (e.g. participles), et cetera. Croft reserves the terms *noun* and *verb* for the unmarked combinations, i.e. prototypical nouns and verbs.

#### 3 Nouns and verbs

This approach enables us to account for flexibility in use, while at the same time maintaining the basic noun-verb distinction: *taŋata* 'man' can be defined as a noun, even though it occasionally occurs in a verb phrase; the latter is simply a case of nonprototypical use.<sup>11</sup>

Between prototypical nouns and prototypical verbs lies a whole range of non-prototypical constructions, as illustrated above. Any attempt to divide this area up by drawing a line separating the "noun area" from the "verb area" is arbitrary. However, for practical reasons I will use the term *verb* for any word which is lexically (i.e. semantically) a verb, and *noun* for any word which is lexically a noun. Thus, in the examples above, the underlined lexical item in (1–3) and (10) is called a noun, while the underlined word in (9) and (11) is called a verb. *Vānaŋa* is a verb when it denotes the action 'to talk', whether it occurs in a prototypical VP or in a phrase that also has nominal properties. When *vānaŋa* denotes an entity 'word, utterance', it is a noun. As these two senses are obviously related, the relation between the two can be defined as polysemy (one lexical item having two related but not identical senses) rather than homophony (two unrelated lexical items which happen to share the same phonological form).

For many words, the semantic criterion is sufficient to classify them as either nouns or verbs: they primarily designate either an entity or an event. However, with nonconcrete words this criterion does not work as well; it is not always obvious whether a notion should be classified as an event or an entity. Take for example natural phenomena: without a syntactic context, should 'rain' be classified as an event ('it rains') or an entity ('the rain falls')? Is 'flood' a thing or an event? The same is true for abstract nouns (*haŋa* 'to love; love'; *manaꞌu* 'to think; thought'). Such words are hard or impossible to assign to a word class apart from syntax; they are by nature not prototypical nouns or verbs. For these words, therefore, syntactic criteria are needed to assign them to a word class. One possible criterion is the presence of a determiner, but as discussed above, this is not a very strong clue; the determiner is a very weak criterion for nounhood. There are other syntactic criteria, however:


<sup>11</sup> Besnier (2000: 257) takes a similar approach for Tuvaluan: each word in the language has one basic wordclass membership; use of the same word in other word classes is marked (e.g. nominals may be used as a verb, but this use is less frequent than their use as a noun and may be subject to structural restrictions). Moyse-Faurie (2005: 184) likewise argues for an approach starting from the prototypical sense and function of a lexeme (as opposed to an approach based on a syntactic dichotomy between noun phrase and verb phrase).

3.2 Nouns and verbs

These criteria tend to converge into the same direction; for concrete words, this is usually the same direction as suggested by the semantics of the word: an entity word is usually modified by NP elements and may take a possessor; an event word is usually modified by VP elements and may take canonical arguments. In other words, nouns and verbs tend to show prototypical behaviour.<sup>12</sup> The same syntactic criteria can now be used to assign abstract words (for which the semantics do not provide a strong clue) to a word class. For an example of how syntax can help to categorise an abstract word, see the discussion on (14) in §3.2.2.1 below.

In two cases, there are morphological clues for noun- or verbhood.


In the following sections, the area between prototypical nouns and prototypical verbs is further explored. §3.2.2 discusses lexical noun/verb pairs; §3.2.3 discusses syntactic nominalisation; §3.2.4 briefly discusses the use of nouns in verbal contexts.

### **3.2.2 Lexical noun/verb correspondences**

Many words in Rapa Nui are used both as nouns and as verbs, without any difference in form but with a difference in meaning. As discussed in the previous section, these are best considered as cases of polysemy, a single lexical item having both a nominal and a verbal sense.

First a note on terminology. Common terms like *nominalisation* and *deverbal noun* indicate that a noun is derived from a verb. While this is often the case, for other words the verb is derived from the noun, or the direction of derivation is undetermined. As the verb and the noun are identical in form, there are no morphological clues for the direction of derivation. For this reason the neutral term *noun/verb correspondences* is used here.

<sup>12</sup> Croft (2000: 96) points out that the meaning of words tends to shift towards the unmarked sense associated with their syntactic use: action words used in referring expressions tend to denote an object typically associated with the activity (e.g. 'learn+NOM' > 'school'); object words used as a predicate tend to denote an action typically associated with the object (e.g. 'baggage+V' > 'to pack').

#### 3 Nouns and verbs

§3.2.2.1 explores the semantic relationships between these homophonous noun/verb pairs.<sup>13</sup> §3.2.2.2 discusses the – much rarer – derivations involving a nominalising suffix.<sup>14</sup>

#### **3.2.2.1 Homophonous noun/verb pairs**

**3.2.2.1.1 Concrete entities** In many cases, the noun denotes a concrete entity (an object or person), while the verb denotes an activity in which this entity plays a certain role. Different semantic relationships can be discerned:

**Instrument** The noun indicates a physical object, while the verb denotes an action performed with that object as instrument: 'to use N, to do something with N'. Examples: *hiahia* 'saw;<sup>15</sup> to saw'; *hoe* 'paddle; to paddle'; *harihari* 'comb; to comb hair'.

Sometimes the verb is more specific in sense than the noun: *rama<sup>N</sup>* 'torch', *rama<sup>V</sup>* 'to fish with a torch' (a fishing technique done at night). In other cases the noun is more specific: *raŋo<sup>V</sup>* 'to support', *raŋo<sup>N</sup>* 'stretcher, handbarrow'; *haŋuhaŋu<sup>V</sup>* 'to pant, breathe heavily', *haŋuhaŋu<sup>N</sup>* 'bellows; forge'.

**Product** The noun denotes the product or result of the action, often a concrete object. Examples: *hohoꞌa* 'to take a picture; a picture'; *taka* 'to roll up; a roll, spool'; *tūtia* 'to sacrifice; offering'; *tarakī* 'to dry meat; dried meat'. One of the senses may be more specialised: *pū<sup>N</sup>* 'hole'; *pū<sup>V</sup>* has the underlying basic sense 'to make a hole, pierce, perforate' but is only used in several specific senses: 'to hit with a bullet, to hook a fish, to dig out tubers'. *Para<sup>V</sup>* has a wide range of senses: 'to decay, ripen, rot, rust', while *para<sup>N</sup>* only means 'rust'.

**Utterance** Similar to the preceding cases are verbs of speaking, where the corresponding noun expresses the utterance produced by the action of speaking: *vānaŋa* 'to speak; word, utterance'; *ꞌaꞌamu* 'to tell; story'; *reoreo* 'to lie; a lie'. This category also includes *manaꞌu* 'to think; thought'. It seems that all words in this semantic domain can be both verb and noun; however, the semantic relationship may be idiosyncratic: *pure<sup>V</sup>* 'to pray', *pure<sup>N</sup>* 'prayer' but also 'mass'. Notice that the nominal sense of these words is

<sup>13</sup> To obtain the data for this section, I listed all words in my lexical database that have both a nominal and a verbal definition. As this database incorporates data from all previous dictionaries and word lists of Rapa Nui, it includes many doubtful definitions, translational equivalents for which it is not clear that the word is actually used in that particular sense. Besides, the lexical resources include many words not attested in the text corpus, either because they are obsolete or because the corpus is limited in size. This leaves just over 200 words that are attested in the corpus in both verbal and nominal senses; it is from these words that the data in this section are taken.

<sup>14</sup> Apart from the nominalising suffix and the causative prefix, there are no productive derivative affixes in Rapa Nui. The lexicon does show traces of derivative suffixes, but in all cases the word was certainly or probably borrowed or inherited as a whole. For example, *tāmiti* 'to salt, cure' is obviously related to *miti* 'salt', but *tāmiti* was probably borrowed from Tahitian, where *tā-* frequently occurs as a (non-productive) factive prefix.

<sup>15</sup> English definitions not preceded by 'to' are nouns.

3.2 Nouns and verbs

not just 'the act of performing X': one can expose a lie or print a story, without being involved in the act of lying or storytelling itself.

**Patient** Other cases in which the noun is the Patient of the corresponding verb are those in which the noun is an entity undergoing the action or affected by the action: *kai* 'to eat; food'; *ꞌakaveŋa* 'to carry on the back; basket carried on the back'.

**Agent** For a number of words, the noun denotes the Agent of the corresponding action. In some cases the noun denotes a profession: *haꞌavā* 'to judge; a judge'; *tāvini* 'to serve; servant'. For other words the Agent may be anyone who performs the action, whether incidentally or regularly: *mataꞌite* 'to testify; eyewitness'; *reoreo* 'to lie; liar'.

**Location** Words indicating the place where the action happens, are rare. One example is *haka iri* 'to ascend; slope'. *Hāpī* 'to learn' may be used in the sense 'school' (*turu ki te hāpī* 'go down to school'), but more commonly this sense is expressed by *hare hāpī* 'house + learn'.

**3.2.2.1.2 Abstract words** For abstract words, it is harder to distinguish distinct nominal and verbal senses. Noun and verb often refer to the same 'thing', but with an aspectual difference: while the verb expresses an event taking place in time, the noun denotes the same event as a bounded whole. This suggests that the distinction is syntactic rather than lexical.

**Natural phenomena** Many natural phenomena (e.g. meteorological conditions) can be expressed as either noun or verb. The following pair of examples illustrate this for *aꞌa* 'flood': in (12) it is a verb with the flooded object as subject, in (13) it is a noun in idiomatic collocation with the verb *rere* 'fly'.


Other words in this category only occur as nouns: *ꞌua* 'rain'<sup>16</sup> (with *hoa* 'throw': *He hoa te ꞌua*, 'It rained', lit. 'The rain threw'); *tokerau* 'wind' (often with *puhi* 'blow' or *hū* 'roar').

<sup>16</sup> Only rarely is *ꞌua* used as a verb, without a subject: *e ꞌua rō ꞌā* 'it was raining' (R475.003).

#### 3 Nouns and verbs

**Human experiences** There is a large category of words for the expression of human experiences: feelings and propensities (*mataku* 'to be afraid; fear'; *nounou* 'to be greedy; greed'); physical experiences (*mamae* 'pain; to suffer pain'; *mare* 'asthma; to have asthma').

**Other abstract words** There are many other abstract words. Some of these express telic events, events with a natural endpoint; in that case the noun expresses a bounded entity, the event conceived as an object: *hāipoipo* 'to marry; wedding ceremony, wedding party'; *ꞌāꞌati* 'to compete; competition'; *tauꞌa* 'to fight; battle'. For other words the semantic distinction between the nominal and the verbal sense is less clear: *hāpī* 'to learn, to teach; schooling, lesson, education'; *haꞌamata* 'to begin; beginning'; *haꞌuru* 'to sleep; sleep'; *mate* 'to die; death'.

It is questionable whether abstract nouns are lexically distinct from the corresponding verbs. In a few cases, the noun has developed more specific senses: *makenu<sup>V</sup>* 'to move about', *makenu<sup>N</sup>* 'action, movement; development; party/feast'; *rē<sup>V</sup>* 'to win'; *rē<sup>N</sup>* 'victory; goal (in soccer)'. Further lexical research could show if other abstract words show subtle meaning differences between noun and verb.

As suggested in §3.2.1.2, syntactic criteria could also help to determine the existence of lexical nouns and verbs. The consistent absence of verb phrase particles could indicate that the nucleus is a lexical noun, not a nominalised verb. Another syntactic criterion is the syntactic context in which the noun phrase appears. As discussed in §3.2.3.1 below, in certain constructions nominalised verbs occur in noun phrases without a suffix, while in other nominal positions they tend to have a nominalising suffix. If a word occurs in one of the latter contexts *without* a nominalising suffix, this suggests that it is a lexical noun. In the following example, *manaꞌu* 'think' and *ŋaroꞌa* 'perceive' both occur in the direct object position, a position in which verbs usually take a nominalising suffix. *Ŋaroꞌa* does indeed have the suffix *iŋa*; *manaꞌu* however does not, which suggests that it is a lexical noun.

(14) *A* prop *au* 1sg *e* ipfv *haka* caus *ꞌite* know *atu* away *ena* med *i* acc *tōꞌoku* poss.1sg.o *manaꞌu,* think *i* acc *tōꞌoku* poss.1sg.o *ŋaroꞌa* perceive *iŋa.* nmlz 'I will make known what I think, what I feel.' [R443.013–015]

In many other situations it is hard to classify the abstract word as a noun or a verb, and for these words the existence of a lexical noun/verb distinction could be called into question. For many concrete words, on the other hand, there is a clear lexical noun/verb distinction. As indicated above, the noun often denotes a participant in the event rather than the event itself. Moreover, either the verb or the noun may have idiosyncratic senses.

Another indication that nominal and verbal senses are lexically determined is the fact that many noun/verb pairs which could be expected to exist, do not occur.<sup>17</sup> A few examples:

<sup>17</sup> Clark (1983a) presents similar observations for Māori.

3.2 Nouns and verbs


This confirms that noun/verb correspondences are – at least for certain words – defined in the lexicon.

#### **3.2.2.2 Lexical nominalisation involving a suffix**

While hundreds of words in the Rapa Nui lexicon show zero derivation, cases of lexical nominalisation involving a nominalising suffix are much less numerous. As discussed in §3.2.3.2 below, there are various nominalising suffixes, without a sharp distinction in meaning and use: *-ŋa*, *haŋa*, *iŋa*, *aŋa*, *eŋa*, *oŋa*. In the standard Rapa Nui orthography (§1.4.4), all of these are written as separate words, with the exception of *-ŋa*. All these forms can be used in lexical nominalisation as well as syntactic nominalisation, often with the same verb. An extreme example is the verb *noho* 'to sit, stay', which occurs with all suffixes: *nohoŋa*, *noho haŋa*, *noho iŋa*, *noho aŋa*, *noho eŋa*, *noho oŋa*; all of these may have the lexicalised sense 'epoch, period'.

As discussed in the previous section, lexical noun/verb pairs without suffix may have various meaning correspondences. In the same way, suffixed nominalisations may be related to the root verb in various ways. Some indicate an object associated with the event or action: *moe* 'to lie', *moeŋa* 'mat'; *hatu* 'to weave leaves', *hatuŋa* 'woven roofing'; *toe* 'to remain', *toeŋa* 'leftovers'.

Others refer to a place where the action is performed: *puhi* 'to fish for lobsters and eels at night', *puhiŋa* 'a place where lobsters and eels are caught at night'; *piko* 'to hide', *pikoŋa* or *piko haŋa* (both obsolete) 'hiding place'.

Other derivations yet have a more abstract sense. *Noho* + nmlz is mentioned above. Another example is *haka tere iŋa* 'system, culture, religion', from *haka tere* 'to lead, rule, govern'.

#### 3 Nouns and verbs

All these examples concern lexical nominalisation. The use of the nominalising suffix in syntactic nominalisation will be discussed in §3.2.3 below.

#### **3.2.2.3 Cross-categorial use of borrowings**

The Rapa Nui lexicon has incorporated a large number of Spanish borrowings (§1.4.2). These are used cross-categorially with great freedom. For many Spanish noun/verb pairs, Rapa Nui has borrowed one form, usually either the noun or the verb in the 3rd sg. present, and this form is used as both noun and verb. Below are two examples from the text corpus.<sup>18</sup> In (15), *rivuho*, originally a noun (Sp. *dibujo* 'drawing'), is used as a verb; in (16), the verb *agradece* (Sp. *agradece* 'gives thanks') is used as a noun.


§3.2.2.1 showed that there are lexical restrictions and idiosyncrasies in the cross-categorial use of Rapa Nui words. Further research could show whether similar restrictions apply in the use of borrowings.

### **3.2.3 Syntactic nominalisation**

Syntactic nominalisation refers to constructions in which a lexical verb enters into a construction which has "some of the formal trappings of a noun phrase" (Clark 1981: 65). As discussed in §3.2.1.1, no change in meaning is involved; the verb retains its verbal sense, while the phrase may retain VP characteristics. In Rapa Nui, the minimal criterion for nominalisation is that the verb is preceded by a determiner (see the inventory of determiners in §5.3.1). This is usually the article *te*, occasionally a demonstrative determiner, but it may also be the nominal predicate marker *he*: see the discussion of (23) below.<sup>19</sup>

Just like lexical nominalisation, syntactic nominalisation occurs both with and without a nominalising suffix. In the first case, the suffix is usually *iŋa* or *haŋa*, occasionally *eŋa*, *aŋa* or *oŋa*; the form of the suffix is discussed in §3.2.3.2. In the second case (zero nominalisation), the nominalised form is identical to the verb itself.

In §3.2.3.1 the use of both types of nominalisations is discussed. In §3.2.3.3, the nominalised phrase is examined in further detail, showing that this phrase retains certain verbal characteristics.

<sup>18</sup> See also Makihara (2001b), who gives many examples from a corpus of spoken texts.

<sup>19</sup> *Pace* Moyse-Faurie (2011: 136): "In Polynesian languages, only the specific article may nominalise a verb phrase".

3.2 Nouns and verbs

#### **3.2.3.1 Uses of zero and suffixed nominalisation**

The choice between zero and suffixed nominalisation depends to a large extent on the syntactic context. Generally speaking, zero nominalisations are used in more verbal contexts, while suffixed forms are used in more nominal contexts. However, there is no clear watershed between the two sets of contexts: in certain environments either one can be used. The difference between the two can be formulated as follows: zero nominalisation presents the event as an event, i.e. as something which has a temporal duration, and which may be progressive or habitual. Suffixed nominalisation noun presents the event as an object, i.e. as a bounded entity, seen as a whole.<sup>20</sup> Often it refers to one particular occasion when the event took place, or to a set of such occasions. By contrast, zero forms may refer to potential occurrences. Broadly speaking, suffixed nominalisation are realis, while zero nominalisations may be irrealis.<sup>21</sup>

The event/object distinction goes a long way towards explaining the distribution of both items. The different syntactic contexts will be listed and illustrated below, but here are some general observations. Zero nominalisations are commonly used as main clause predicate, a typical verbal context. Aspectual verbs like *haꞌamata* 'begin' refer to the temporal structure of an event, so it is not surprising that they take a zero nominalisation as complement. By contrast, in typical nominal positions (subject, possessor…) suffixed forms are more common.

When the event is negated (an irrealis context), zero nominalisations are common, while suffixed forms are extremely rare.

(17) *ꞌI* at *te* art *taꞌe* not *hakaroŋo,* listen *he* ntr *ŋaro* lost *rō* emph *atu* away *ꞌai.* subs 'Because (the sheep) did not listen, it got lost.' [R490.005]

When the event is modified by a numeral (i.e. is countable), a verbal noun is used:

(18) *He* ntr *takeꞌa* see *mai* hither *ka* cntg *teka* revolve *e* num *rua* two *haka* caus *teka* revolve *iŋa* nmlz *ꞌi* at *muri* near *o* of *te* art *motu.* islet

'I saw (the bird) making two rounds near the islet.' [R338.014]

Not all distributional facts are easily explained, though. Certain constructions take a suffixed nominalisation, even though they denote an event with temporal duration (e.g. the *ko te V* construction in (24–25) below). On the other hand, a reason clause may refer to one particular instance, yet contain a zero nominalisation (see (29)).

<sup>20</sup> Clark (1981: 79) makes a similar distinction, when he suggests "the hypothesis that unsuffixed nominalizations denote activities or processes […] whereas suffixed nominalizations denote events, which can be enumerated and located in time".

<sup>21</sup> Realis, as defined by Payne (1997: 244), asserts that an event has actually happened; the irrealis mode does not assert that the event actually happened, nor that it did not happen (§11.5.2).

#### 3 Nouns and verbs

We may conclude that the choice between the two forms is partly based on semantics, partly conventionalised (certain constructions always or usually take one form), and partly free.

Regardless the syntactic position of the phrase, suffixed forms are used when the word refers to the place, time<sup>22</sup> or manner of the event, as the following examples show:


'I have not forgotten it from my childhood (lit. smallness) until now.' [R416.936]

(21) *Pē* like *nei* prox *te* art *aŋa* make *haŋa* nmlz *o* of *te* art *rāꞌau* medicine *nei.* prox 'This is how you make (lit. the making of) this medicine.' [R313.159]

In the remainder of this section, the different contexts in which the two nominalisations are used, are listed and illustrated.

**3.2.3.1.1 Nominalised main clauses** A number of constructions involve a main clause which is nominalised, even though they express an event. In most of these, zero nominalisation is used.

**The actor-emphatic** The actor-emphatic is a very common construction, in which an Agent is preposed as a possessive pronoun or phrase (§8.6.3). In one actor-emphatic construction (there are three in Rapa Nui), the verb is nominalised (i.e. preceded by the article).

(22) *ꞌĀꞌana* poss.3sg.a *te* art *kai* eat *i* acc *te* art *meꞌe* thing *ririva* good:red *taꞌatoꞌa.* all 'He (was the one who) ate all the best things.' [R532-01.011]

*Ko S te V* Much less common is the *ko S te V* construction: a topicalised subject marked by *ko*, followed by a zero nominalisation (see (94–95) on p. 410).

<sup>22</sup> The temporal sense is found with "stage words" (see Broschart 1997: 148): certain adjectives like *ꞌitiꞌiti* 'small', and the noun *poki* 'child': *poki iŋa* 'childhood'. In the corpus, *poki* is the only noun taking the nominaliser.

3.2 Nouns and verbs

*He V te aŋa* Another common construction is *he V te aŋa* (lit. 'the doing was Ving'), which indicates an habitual action, event or attitude. This construction involves two nominalised verbs: *aŋa* 'to do' is nominalised and serves as the subject of the clause; the other verb serves as nominal predicate. An example:

(23) *ꞌI* at *rā* dist *noho* stay *iŋa* nmlz *he* pred *tuꞌu* arrive *nō* just *mai* hither *te* art *aŋa* do *o* of *te* art *nuꞌu* people *paꞌari* adult *ki* to *tōꞌoku* poss.1sg.o *koro* Dad *uꞌi.* look

'At that time the old people always came to see my father (lit. just arriving was the doing of the old people).' [R649.101]

As *he* can precede both verbs (the aspect marker *he*) and nouns (the predicate marker *he*), it may not be immediately obvious that *he tuꞌu* is nominalised. However, the further contents of the clause show that this is the case: the subject of the clause is *te aŋa*, which is not an argument of *tuꞌu*. Conversely, the Agent of *tuꞌu* is not expressed as subject of the clause, but as a genitive phrase after *aŋa*. (Another indication that the verb in this construction is nominalised is, that its object may be incorporated; see (128) on p. 262.)

*Ko te V* The construction *ko te* + verb signifies that an action or situation is ongoing or persisting. In most cases a suffixed nominalisation is used, followed by the identity particle *ꞌana/ꞌā*, as in (24). Sometimes the identity particle is left out, in which case a zero nominalisation may be used, as in (25).


**Other main clauses** Occasionally zero-nominalised verbs occur in main clauses in other cases. This happens especially with verbs like *haŋa* 'want' and *kī* 'say' (cf. §9.2.6).<sup>23</sup> Notice that their S/A argument<sup>24</sup> is expressed as a possessive.

(26) *Te* art *haŋa* want *era* dist *o* of *Malo* Malo *mo* for *ai* exist *ko* prom *Hepu* Hepu *mo* for *rē.* win 'Malo wants (lit. 'Malo's wish') Hepu to win.' [R408.064]

<sup>23</sup> Interestingly, in Māori there is also a tendency to express 'wish'-type predicates nominally, followed by a purpose clause (Bauer, Parker & Evans 1993: 459).

<sup>24</sup> Following Comrie (1978), the following terms are used in this grammar to refer to verb arguments without specifying a semantic role: S = the single argument of an intransitive verb; A = the most agentive argument of a transitive verb (typically an Agent or Experiencer); O = the least agentive argument of a transitive verb (typically a Patient or Theme).

3 Nouns and verbs

(27) *Tāꞌana* poss.3sg.a *kī:* say *taꞌe* conneg *tātou* 1pl.incl *hokotahi* alone *nō.* just 'What she said, was: we are not alone.' [R649.191]

**Reason clause** Finally, a construction with nominalised verb is sometimes used to express reasons (see (259–260) on p. 568).

**3.2.3.1.2 Subordinate clauses** In subordinate clauses, either suffixed or zero nominalisations are used, depending on the type of clause:

*ꞌO* **+ nominalised verb** In causal clauses, after the preposition *ꞌo*, nominalised verbs are common (§4.7.3):

(28) *Ko* prf *koa* happy *rivariva* good:red *ꞌana* cont *te* art *ŋā* pl *poki* child *ꞌo* because\_of *te* art *turu* go\_down *hakaꞌou* again *o* of *rāua* 3pl *ki* to *te* art *hāpī.* learn

'The children are really happy because they go back to school.' [R334.128]

*ꞌI* **+ nominalised verb** The preposition *ꞌi* followed by a verb has various usages. It may indicate a reason, in which case it is followed by either a zero or a suffixed nominalisation; the latter is more common.


*ꞌI* is also used in a temporal sense; in that case the clause usually has a suffixed nominalisation:

(31) *ꞌI* at *te* art *tuꞌu* arrive *iŋa* nmlz *hakaꞌou* again *mai* hither *era* dist *mai* from *Hiva…* mainland 'When he returned (lit. in the returning) again from the mainland….' [R487.021]

3.2 Nouns and verbs

**Temporal clauses** In temporal clauses introduced by *ki* or *ꞌātā ki* 'until', suffixed forms are used:

(32) *E* ipfv *tiaki* wait *rō* emph *atu* away *ki* to *tuꞌu* poss.2sg.o *topa* descend *haŋa* nmlz *atu.* away 'I will wait until you come down.' [R230.047]

However, *ante ki* 'before' is followed by a zero nominalisation – possibly because its sense is more irrealis than *ꞌātā ki*.

(33) *ante* before *ki* to *te* art *uru* enter *ki* to *roto* inside 'before she went inside' [R181.005]

**Circumstantial clauses** Occasionally in circumstantial clauses, after *koia ko*, a zero nominalisation is used; more commonly, however, *koia ko* is followed by a verb (§8.10.4.2).

**Purpose clauses** Purpose clauses, introduced by *mo* 'in order to', usually have a (non-nominalised) verb (§11.5.1). Interestingly, occasionally they have a suffixed nominalisation directly following *mo*. This is the only construction in which a suffixed form is not preceded by a determiner:

(34) *ꞌIna* neg *he* pred *māꞌeha* light *mo* for *uꞌi* see *iŋa* nmlz *i* acc *te* art *kai.* food 'There was no light to see the food.' [R352.070]

**3.2.3.1.3 Nominal roles** In nominal positions in general, suffixed forms are much more common than zero nominalisations.

**Subject** Suffixed nominalisations may be the subject of verbal or nominal clauses:


'His living there started in the year '35.' [R539-1.492]

For more examples, see (19) and (21) above.

#### 3 Nouns and verbs

However, the subject may also be a zero nominalisation. This tends to happen when the verb refers to a potential or general situation, rather than an event which happens at a specific time:

(37) *ꞌO* because\_of *ira* ana *te* art *oho* go *tai* sea *e* ipfv *oho* go *hai* ins *mahana* day *rivariva.* good:red 'Therefore, going to sea is done on beautiful days.' [R354.016]

Notice however, that (35) does not refer to a specific instance either, yet it involves a suffixed form.

These examples suggest that there is a certain freedom in the use of both forms.

**Direct object** In direct object position, zero nominalisations are common with two classes of verbs: aspectual verbs and certain cognitive verbs.

Aspectual verbs include *oti* 'to finish', *haꞌamata* 'to begin' and *hōrou* 'to hurry, do in a haste', as well as a few less common verbs like *haka mao* 'to terminate'. They may take a nominalised verb as complement, which may or may not be introduced by the object marker *i* (§11.3.2).


Aspectual verbs are not always constructed with a nominalised verb, however. For more details, see §11.3.2.

Cognitive verbs include, among others, *ꞌite* 'to know', *hāpī* 'to learn' and *māhani* 'to be or get used to'. These often take a zero nominalisation when the content of knowledge is a skill, a 'how to':


However, a suffixed form may also be used, possibly indicating the manner of performing an activity:

(42) *Mo* for *hāpī* learn *rivariva* good:red *ōꞌou* poss.2sg.o *i* acc *te* art *pāpaꞌi* write *haŋa* nmlz *o* of *te* art *ŋā* pl *meꞌe* thing *nei…* prox 'In order for you to learn well the (way of) writing these things…' [R617.003]

With complements of other verbs, for example verbs of perception and speech, suffixed forms are much more common:


**Possessives** When a verb is used as a possessive, suffixed nominalisations are often used:

(45) *E* ipfv *ai* exist *rō* emph *ꞌana* cont *e* num *rua* two *huru* manner *o* of *te* art *uꞌi* look *iŋa* nmlz *o* of *te* art *taŋata.* man 'There are two ways in which people see it (lit. two ways of seeing).' [R648.218]

Zero nominalisations also occur in this position, especially after temporal nouns:

(46) *Ka* cntg *rua* two *matahiti* year *o* of *te* art *poreko* born *o* of *Puakiva…* Puakiva 'Two years after Puakiva's birth…' [R229.007]

One might expect a suffixed form here, as the birth is a one-time event which has happened; yet zero forms are more common when modifying a temporal noun.

**After prepositions** Suffixed nominalisations are found after most prepositions: *mai* 'from', *hai* 'with', *pē* 'like', *ki* 'to' (often temporal 'until', see above), and after locationals. Two examples:


#### 3 Nouns and verbs

**Nominal predicates** With the nominal predicate marker *he*, suffixed nominalisations are used (except in the construction *he V te aŋa*, see §3.2.3.1.1 above, ex. (23)). This happens for example in titles as in (49), in existential clauses, and in classifying clauses as in (50).


#### **3.2.3.2 The form of the nominalising suffix**

As indicated above, there are various forms of the nominalising suffix: *haŋa*, *iŋa*, *eŋa*, *oŋa*, *-ŋa*. <sup>25</sup> In older texts, both *haŋa* (86x) and *-ŋa* (132x) are common, while *iŋa* (9x) and *aŋa* (3x) occur sporadically. In newer texts, *haŋa* still occurs (255x), but *iŋa* is now the predominant form (914x). *-ŋa* has become very rare (12x), but other forms have developed: besides *aŋa* (9x), *eŋa* (25x) and *oŋa* (14x) are found. The latter two are the result of vowel assimilation: *oŋa* occurs after *noho* 'to stay' and *oho* 'to go', while *eŋa* occurs after various verbs ending in *-e* and *-o*; in the first case *eŋa* is the result of total assimilation, in the second case it results from vowel height assimilation: *noho iŋa > noho eŋa*.

The predominant forms, then, are *haŋa* and *-ŋa*<sup>26</sup> in older Rapa Nui, and *haŋa* and *iŋa* in modern Rapa Nui. The question is, if there is a distinction between the two forms in each stage.

In older Rapa Nui, *haŋa* has a wide range of uses, corresponding to the uses of suffixed nominalisations described in the previous section. *-ŋa* often has a more nominal and sometimes lexicalised sense: *ohoŋa* 'go nmlz = trip'; *nohoŋa* 'stay nmlz = epoch'. The *-ŋa* form may refer to an object related to the event: *toeŋa* 'remain nmlz = leftovers'; *hatuŋa* 'weave nmlz = roofing'; *moeŋa* 'lie nmlz = mat'.

<sup>25</sup> The forms *-(C)aŋa* and *-ŋa* occur throughout the Polynesian languages. Originally, the initial consonant in *-Caŋa* was lexically determined; this is still the case in languages like Māori (Bauer, Parker & Evans 1993: 512) and Samoan (Mosel & Hovdhaugen 1992: 194). In other languages, the paradigm was simplified, as in Tahitian, where only *-ra ꞌa* (< *\*raŋa*) was retained. In Rapa Nui the situation is more complicated, as this section shows.

<sup>26</sup> One could ask whether forms like *nohoŋa* in older texts actually contain a long vowel (*nohōŋa*) or even a disyllabic double vowel (*noho oŋa*). The former is theoretically possible: long vowels are poorly represented in older texts, and in other languages (e.g. Samoan), the vowel before *-ŋa* may be lengthened as well. Notice, however, that Rapa Nui has an absolute constraint against long vowels in penultimate syllables (§2.3.2). Concerning the possibility of *noho oŋa* underlying *nohoŋa*, there is no positive evidence for this; on the contrary, the occurrence of \**-ŋa* in many other Polynesian languages and the rarity of *Vŋa* in older Rapa Nui texts suggest that *-ŋa* is an original form while *-Vŋa* is a recent development, even though the occurrence of *-iŋa/-aŋa* in some languages could be taken as evidence to the contrary. A possible scenario is, that a form like *uꞌiŋa* 'look nmlz' developed into *uꞌi iŋa*; the form *iŋa* was then generalised to verbs not ending in *i*, supplanting *-ŋa*.

3.2 Nouns and verbs

However, the distinction between *haŋa* and *-ŋa* is by no means clear-cut. On the one hand, *haŋa* forms are used with nominal senses, especially in the sense of place, manner and time (see (19–21) above): *noho haŋa* means 'epoch', just like *nohoŋa*; *ꞌitiꞌiti haŋa* 'small nmlz = infanthood'; *piko haŋa* 'hide nmlz = hiding place'. On the other hand, *-ŋa* forms may be used with a verbal sense, just like *haŋa* forms:

(51) *Ki* when *roaroa* long:red *te* art *mimiroŋa,* spin:nmlz *he* ntr *vīviri* roll *te* art *henua.* land

'When he has turned around for a while (lit. 'when the spinning is long'), he will get dizzy (lit. the land rolls).' [Ley-8-52.013]

In modern Rapa Nui, the distinction between different nominalisers is even harder to pinpoint. *Haŋa* (255x) is less common than *iŋa* (914x), but occurs in a wide variety of texts, in a wide variety of uses, and with no less than 82 different verbs. To give two examples:


More generally, both suffixes occur in nominalisations used as subject, object, genitive, after prepositions, and in time clauses introduced by *ꞌi*. The only construction in which *haŋa* never occurs, is the predicate construction *ko te V* (see (24–25) above). Speaker preference may play a role: it is telling that the Bible translation consistently uses *iŋa*, almost never *haŋa*. Apart from this, I have not been able to find a distinction between the two.

#### **3.2.3.3 The nominalised phrase**

In §3.2.1.1 it was pointed out that verbs preceded by a determiner may still be accompanied by certain verb phrase elements, as well as certain noun phrase elements. The noun phrase is discussed in detail in Chapter 5, the verb phrase in Chapter 7. This section is limited to a brief listing of elements occurring with nominalised verbs, which shows the hybrid character of nominalised verb phrases.

**Verb phrase elements** Some verb phrase elements never occur with nominalised verbs: aspectual and modal markers, the intensifier *rō*, and the VP-final particle *ai* or *ꞌai*. However, other elements do occur:

Nominalised verbs may be followed by an adverb (§4.5.1) like *hakaꞌou* 'again' or*takoꞌa* 'also'. Notice that *hakaꞌou* and *takoꞌa* may also occur in the noun phrase (§5.8.1). With suffixed nominalisations, *hakaꞌou* and *takoꞌa* occur after the nominalising suffix.

The adverbs *tahi* 'all' and *koraꞌiti* 'slowly' – which do not occur in the noun phrase – both occur once in the corpus with a nominaliser; interestingly, they precede the suffix:

#### 3 Nouns and verbs

(52) *Ko* prom *te* art *turu* go\_down *koraꞌiti* slowly *iŋa* nmlz *ꞌā* ident *te* art *vai.* water 'The water went slowly down.' [Gen. 8:5]

Both suffixed and zero nominalisations may be followed by a directional *mai* or *atu* (§7.5):

(53) *Ko* prf *rivariva* good:red *ꞌana* cont *tōꞌona* poss.3sg.o *rere* jump *iŋa* nmlz *mai.* hither 'His jump(ing) was good.' [R408.022]

Zero nominalisations may take the constituent negator *taꞌe* (§10.5.6):<sup>27</sup>

(54) *ꞌI* at *te* art *taꞌe* conneg *hakaroŋo,* listen *he* ntr *ŋaro* lost *rō* emph *atu* away *ꞌai.* subs 'Because (the sheep) did not listen, it got lost.' [R490.005]

*Taꞌe* hardly ever goes together with a suffixed nominalisation, possibly because the realis character of the verbal noun precludes its use with a negation.

The limitative particle *nō* 'just, still' (§7.4.1) occurs with zero and suffixed nominalised verbs (see (30) above).

**Noun & verb phrase elements** Certain particles occur in both noun phrase and verb phrase:

The demonstrative particles *nei*, *ena* and *era*. In the verb phrase, they co-occur with certain aspectual markers (§7.6); in the noun phrase, they co-occur with any determiner (§4.6.3). They do not occur with zero nominalisations, but they do occur with suffixed nominalisations, for example in (48), here repeated:

(55) *He* ntr *hati* break *te* art *vaꞌe* foot *pa* like *he* pred *hati* break *iŋa* nmlz *era* dist *ꞌā* ident *o* of *tōꞌoku* poss.1sg.o *vaꞌe.* foot

'He broke his leg, like I broke my leg (lit. like the breaking of my leg).' [R492.021]

The marker *ꞌā/ꞌana* occurs in the verb phrase as a continuous marker, co-occurring with certain aspectual markers (§7.2.5.5); in the noun phrase it serves as an identity marker (§5.9). It occurs with verbal nouns, as illustrated in (48) above. In this context, where a comparison is involved, *ꞌā* is clearly an identity marker.

**Noun phrase elements** Nominalised verbs may also be accompanied by noun phrase particles. They may be preceded by any kind of determiner: the article *te*, demonstratives like *tū* as in (29) above, possessive pronouns as in (27), and the predicate marker *he* as in (55). Suffixed nominalisations tend to denote single instances of an event, so they may

<sup>27</sup> *ta ꞌe* does not occur within the noun phrase; when it modifies a noun, it occurs before the predicate marker *he*.

3.2 Nouns and verbs

be countable: they can be modified by a numeral (see (18)), or by quantifiers like *taꞌatoꞌa* 'all'. The corpus does not contain examples of the plural marker *ŋā* with verbal nouns, but this may be accidental.

In conclusion, both zero and suffixed nominalisations retain a partly verbal character in their phrase. The latter are more nominal than the former, as they allow quantifying elements but do not allow negation.

### **3.2.4 Nouns used as VP nucleus**

Any noun (i.e. entity word) can be used as the nucleus of a verb phrase. Usually, the noun is used in a predicative sense: a verb phrase headed by noun N signifies that the subject is or becomes N; it possesses or acquires property N. These constructions are somewhat similar to nominal predicates marked with *he* (§9.2.1), yet they are different: the noun may be preceded by any preverbal marker, e.g. an aspectual as in (56) or a negator as in (57), and it may be followed by verb phrase particles such as *rō ꞌā* in (56). Also, the clause may express a process ('become'), while nominal predicates only express a state ('be').


Very occasionally, the noun does not indicate 'be/become N', but a typical action associated with N:

(58) *…i* pfv *eꞌa* go\_out *mai* hither *ai* pvp *e* num *tahi* one *rūꞌau* old\_woman *e* ipfv *tokotoko* cane *rō* emph *ꞌana.* cont '…an old woman appeared leaning on a cane.' [R437.079]

Nouns in a verb phrase are in fact rare in texts, with the exception of temporal nouns. The latter are commonly used as verbs, usually expressing that a period of time passes.

(59) *Ko* prf *tāpati* week *ꞌā* cont *i* pfv *tuꞌu* arrive *iho* just\_then *atu* away *ai.* pvp 'When a week had passed, he arrived.' [R416.515]

### **3.2.5 Nominal drift**

In §3.2.3.1.1 above (examples (22–27)), a number of constructions are listed in which a verb is nominalised, even though they are main clauses expressing an event: the actoremphatic construction, *ko te* + verb, et cetera. §3.2.3.1.1 (28–34) lists a number of nominalised subordinate constructions, e.g. *ꞌo te* + verb to indicate cause or reason. (As shown in §11.6.4, various other nominal constructions are used as well to express reason.)

#### 3 Nouns and verbs

These examples illustrate a tendency in Rapa Nui to maximise the nominal domain. This tendency reveals itself in a number of other areas as well:

**Complements** Motion verbs may be followed by a nominal Goal complement as in (60), even though the goal is semantically an event (§11.6.3). As the example shows, the event may be expressed by a verb following the nominal complement. Likewise, perception verbs may be followed by a nominal object + a verbal clause, as in (61) (§11.3.1.2).


**Compounding** In a peculiar case of compounding, an event is expressed by a verb attached as modifier to one of its arguments; the argument is syntactically the head of the construction (§5.7.2.3):

(62) *ꞌI* at *tōꞌona* poss.3sg.o *mahana* day *he* ntr *ai* exist *mai* hither *te* art *aŋa* work *he* pred *ꞌāua* enclosure *titi,* build *ꞌo* or *he* pred *rau* leaf *kato.* pick

'On certain days there were jobs like making fences or picking leaves.' [R380.084]

**Arguments as possessives** In a number of constructions, verbal arguments – especially S and A – may be expressed as possessives, even when the verb is not nominalised. This is the default way to express the S/A argument in clauses introduced by *mo* as in (63) (§11.5.1.2); it commonly happens in relative clauses as in (64) (§11.4.4); and under certain conditions it happens in main clauses as in (65) (§8.6.4.1).


3.3 Nouns

### **3.3 Nouns**

### **3.3.1 Classification of nouns**

Apart from locationals (§3.6), there are two main types of nouns in Rapa Nui: common nouns and proper nouns. Common nouns, such as *hare* 'house' or *poki* 'child', designate a class of entities characterised by certain properties; they can be used as nominal predicates, and it is only within a referential noun phrase that they acquire reference to one or more entities. Proper nouns, such as *Tiare* 'Tiare' and *koro* 'Dad', are inherently referential; they are not used as predicates and have unique reference in a given context.

These classes impose different constraints on the noun phrase of which they are the head. The most important differences are:


Prototypical common nouns denote classes of concrete, bounded entities, for example persons (*taŋata*) and objects (*hare* 'house', *toki* 'adze'). Prototypical proper nouns are names of persons. The precise extent of each category can be deduced from the syntactic behaviour of nouns, with (1) and (3) above as main criteria: nouns preceded by the proper article *a* are proper nouns; nouns preceded by determiners like the article *te* are common nouns. This will be explored in the next section.

Both common and proper nouns function as head of a noun phrase. The structure of the common noun phrase is discussed in §5.1; the structure of the proper noun phrase is briefly discussed in §5.13.

Within the class of common nouns, we may distinguish countable and non-countable nouns. Non-countable nouns include mass nouns like *toto* 'blood' and *ꞌōꞌone* 'earth, soil', and abstract nouns like *haŋa* 'love' and *mamae* 'pain'. There is no morphological or syntactic difference between countable and non-countable nouns in Rapa Nui, except that the latter cannot be combined with noun phrase elements related to quantification: plural markers, numerals and universal quantifiers.

A third group of nouns is the class of locationals, which are preceded neither by determiners nor by the proper article. This class contains a small group of locational terms like *muꞌa* 'front', as well as deictic terms like *nei* 'here, nearby'. Locationals are discussed in §3.6.

Geographical names mostly pattern with locationals, but in some situations they behave like proper nouns (§3.3.2).

#### 3 Nouns and verbs

The properties of the different types of nouns are summarised in Table 3.1.

Table 3.1: Types of nouns

### **3.3.2 Proper nouns**

The class of proper nouns contains those items which are – in the appropriate contexts – preceded by the proper article *a*. This includes the following categories:

**3.3.2.1 Proper names of persons** Some examples:


Geographical names do not take the proper article, whether they designate countries or islands, towns, mountains or any other geographical entity. Nor do they take the common noun article *te*:


3.3 Nouns

This characteristic distinguishes geographical names from both common and proper nouns, and includes them with locationals (§3.6). There are some exceptions though.

Firstly, *Tire* 'Chile' is the only geographical name which always takes the personal name in the appropriate contexts.

(70) *Ararua* the\_two *nō* just *pāꞌeŋa* side *e* ipfv *tuꞌu* arrive *mai* hither *era,* dist *mai* from *Tahiti* Tahiti *ꞌe* and *mai* from *i* at *a* prop *Tire.* Chile 'Both sides arrived, from Tahiti and from Chile.' [R539-2.221]

Secondly, other geographical names may take the proper article when topicalised (personal names and pronouns would also *a* this context):

(71) *A* prop *Rapa* Rapa *Nui* Nui *he* pred *henua* land *ꞌitiꞌiti* small:red *e* num *tahi…* one 'Rapa Nui is a small island…' [R351.001]

Finally, the proper article is used before geographical names used metonymically for their inhabitants. In this case, the presence of *a* shows that the geographical name has been transformed into a personal name:

(72) *He* ntr *aŋa* make *a* prop *Rapa* Rapa *Nui* Nui *i* acc *to* art:of *rāua* 3pl *riu* song *tuai.* ancient '(The people of) Rapa Nui made their old songs.' [R620.013]

#### **3.3.2.2 Personal pronouns**


**3.3.2.3 Kinship terms** The proper article is common with certain kinship terms, especially *koro* 'father, older man' and *nua* 'mother, older woman'. These words are used in the same way as 'Dad' and 'Mum' in English: like personal names, they have a unique referent in the context, and therefore do not need a determiner.

(75) *He* ntr *kī* say *a* prop *koro* Dad *ki* to *a* prop *nua…* Mum 'Dad said to Mum…' [R333.303]

By contrast, *matuꞌa* 'parent' is a common noun. It does not have a unique referent; in order to refer to a particular parent, its reference must be defined, e.g. by a possessive pronoun:

#### 3 Nouns and verbs

(76) *He* ntr *kī* say *ia* then *a* prop *Tiare* Tiare *ki* to *tōꞌona* poss.3sg.o *matuꞌa* parent *vahine* female *era…* dist 'Then Tiare said to her mother…' [R481.137]

**3.3.2.4 General terms referring to people** The word *māhaki* 'friend' (which has a certain compassionate connotation: 'poor one') usually takes the proper article:

(77) *Ka* imp *turu* go\_down *kōrua,* 2pl *ka* imp *uꞌi* look *i* acc *a* prop *māhaki.* companion 'Go down to have a look at (our) friend.' [Ley-2-05.011]

The same applies to a few similar, but less common words: *vērā* 'that poor one', *ꞌeꞌete* 'so-and-so'*, taureka* 'that guy'*.*

**3.3.2.5 The collective marker** The collective marker *kuā*/*koā* (§5.2) is usually followed by a proper name or another word from the categories mentioned above, but even when followed by a common noun, it may be preceded by the proper article. In the following example, *korohuꞌa* is preceded by the plural marker *ŋā*, something which only happens with common nouns. Even so, *kuā* is preceded by the proper article.

(78) *ꞌO* because\_of *ira* ana *a* prop *koā* coll *ŋā* pl *korohuꞌa* old\_man *e* ipfv *maꞌu* carry *hio-hio* strong:red *era* dist *i* acc *te* art *hahaꞌu* tie *iŋa* nmlz *o* of *te* art *pātia.* harpoon

'Therefore the old people tied the cable of their harpoons well.' [R360.020]

**3.3.2.6 Names of months** Names of months always take the proper article, regardless which names are used: the old Rapa Nui names as in (79), the modern English-based names as in (80), or Spanish names as in (81):


'In November of this same year he returned again to this island.' [R343.016]

(81) *Ki* when *oti* finish *te* art *Tāpati* Tapati *nei* prox *he* ntr *piri* join *tātou* 1pl.incl *ꞌi* at *a* prop *marzo.* March 'When the Tapati (= festival week) is finished we are close to March.' [R625.131]

3.4 Verbs

**3.3.2.7** *Hora* **'summer'**

(82) *¿Pē* like *hē* cq *a* prop *kōrua* 2pl *i* pfv *noho* stay *ai* pvp *ꞌi* at *a* prop *hora?* summer 'How were you during summer?' [R334.051]

NB *Hora* 'time', a different lexeme, is a common noun. *Toŋa* 'winter' is also a common noun.

**3.3.2.8 Definite numerals** Definite numerals like *a totoru* 'the three' are always preceded by the proper article (§4.3.4). Unlike all other elements that take the proper article, they cannot occur after prepositions.

The use of the proper article *a* is limited to certain syntactic contexts. This is discussed in §5.13.2.1.

### **3.4 Verbs**

### **3.4.1 Classification of verbs**

As discussed in §3.2.1.2, a prototypical verb is a word which denotes an event, functions as clause predicate and is the head of a verb phrase. Verb phrases will be discussed in Chapter 7, verbal clauses in Chapter 8. This section will be limited to a brief discussion of verb types.

Verbs may have zero, one, two or three arguments. Zero-argument verbs are, for example, words indicating a moment in time or the passage of time.<sup>28</sup> *ꞌŌtea* 'to dawn' in (83) and *ahiahi* 'to be evening' in (84) do not have a subject or any other argument, whether overt or implied. The bracketed clause consists of a predicate only.


'(When) it was late afternoon, they opened the earth oven.' [Mtx-7-15.030]

One-argument verbs include:

1. active intransitive verbs, i.e. verbs involving an Agent, such as *oho* 'go', *hopu* 'to bathe, swim', *piko* 'to hide oneself';

<sup>28</sup> On zero-argument verbs, see Dryer (2007a: 267). Crosslinguistically, zero-argument verbs typically involve weather conditions ('It rains'). In Rapa Nui however, weather terms are not zero-argument verbs: as discussed in §3.2.2.1.2, weather conditions are expressed by subject–predicate collocations, i.e. one-argument predicates.

#### 3 Nouns and verbs


Two-argument verbs in Polynesian languages are often divided into two groups: canonical transitives and middle verbs.<sup>29</sup> The former involve an Agent which acts voluntarily and deliberately, and a Patient affected by the action. Examples are *kai* 'to eat' and *tiŋaꞌi* 'to kill, hit'. With middle verbs, the object is not affected by the action, and the action may be spontaneous rather than voluntary. This category includes verbs of cognition, affection and perception: 'to know', 'to love', 'to see'. As discussed in §8.6.4.2, in Rapa Nui the difference has consequences for the marking of the object.

Many verbs may be either transitive or intransitive, depending on whether an object is expressed or implied.<sup>30</sup> For example, the verb *kai* 'to eat' is transitive when a certain (type of) food is mentioned or implied in the context: in (85) below it is transitive; in (86) it is transitive as well, even though the object is implicit (it has been mentioned just before); in (87) it is intransitive.<sup>31</sup>


Three-argument verbs involve an Agent, a Patient, and a participant to which the action is directed in some way; depending on the verb, this may be a Goal, Addressee,

<sup>29</sup> See e.g. Chung (1978), Hooper (1984b), Harlow (2007a). Bauer (1983) uses the term *experience verbs*. In Chung's description, the difference concerns the affectedness of the patient; Pawley (1973) and Elbert & Pukui (1979) focus on the difference between deliberate and spontaneous actions. Both classifications yield the same sets of verbs. Syntactic differences between canonical transitives and middle verbs are languagespecific: (a) in ergative languages, they take different transitive constructions (§8.2.1); (b) when nominalised, they may take different possessive markers (e.g. in Hawaiian, Elbert & Pukui 1979: 48); (c) middle verbs may take the acc marker *ki* rather than *i*. The latter is true in Rapa Nui and Māori (Bauer 1983; Bauer 1997: 267). In Hawaiian and Tahitian, the development *k >* glottal neutralises the difference between *ki* and *i*, as initial glottals in particles are usually not contrastive.

<sup>30</sup> In this grammar, any clause in which a Patient/Theme argument is either expressed or implied, is considered transitive. See also Footnote 1 on p. 375 on transitivity.

<sup>31</sup> Whether a verb is transitive or intransitive may have syntactic repercussions, even when no object is expressed. See the discussion on causativisation of transitive verbs in §8.12.3, esp. examples (244) and (245).

3.4 Verbs

Recipient or Beneficiary. Examples are *vaꞌai* 'to give', *tuhaꞌa* 'to distribute', *hāpī* 'to teach', *ꞌaꞌamu* 'to tell'. Usually the Patient is expressed as direct object, while the other argument is marked with either *ki* or *mo*. This is discussed in §8.8.2; one example:

(88) *He* ntr *vaꞌai* give *a* prop *nua* Mum *i* acc *te* art *kai* food *ki* to *a* prop *koro.* Dad 'Mum gave the food to Dad.' [R236.078]

There is one exception to this pattern: the verb *hāpī* 'teach' may take two direct objects; the first of these expresses the person taught, the second the content of teaching:

(89) *He* ntr *hāpī* teach *i* acc *te* art *taŋata* person *i* acc *te* art *pure.* pray 'He taught people to pray.' [R231.304]

Three-argument verbs also include causativisations of transitive verbs, such as *haka takeꞌa* 'caussee = to show', *haka aŋa* 'cause to make', *haka ꞌamo* 'make (someone) carry'; these are discussed in §8.12.3. One example:

(90) *He* ntr *haka* caus *tikeꞌa* see *e* ag *Te* Te *Pitu* Pitu *ki* to *a* prop *Uka* Uka *Oho* Oho *Heru* Heru *i* acc *te* art *ꞌōꞌone* soil *meamea.* red:red 'Te Pitu showed (=made see) Uka Oho Heru the red soil.' [Fel-1978.070]

### **3.4.2 Active, stative, intransitive**

Transitive and active intransitive verbs together form the class of active verbs**.** These are characterised by


The remaining verbs form the class of stative verbs. This class is well-established in Polynesian linguistics.<sup>32</sup> Criteria for this class vary per language. In Rapa Nui, statives are characterised only by the two criteria above: they do not occur in the actor-emphatic and their subject cannot be marked with *e*. In other languages, criteria may include the impossibility of passivisation and the impossibility to be used in the imperative.<sup>33</sup>

Regarding the latter criterion, the incompatibility of stative verbs with the imperative is probably semantically/pragmatically motivated: there are simply few contexts in which it is appropriate to use a property word in a command. In Rapa Nui, the word *koa* 'happy' – which is otherwise a typical adjective (§3.5.1.4) – does occur in the imperative:

<sup>32</sup> The term was introduced by Buse (1965) and adopted e.g. by Hohepa (1969a), Biggs (1973), Elbert & Pukui (1979), Chung (1978), Seiter (1980), Mosel & Hovdhaugen (1992).

<sup>33</sup> See Biggs (1973; 1974) on statives in Māori. Within this class, Biggs distinguishes between stative adjectives and stative verbs (discussed as "neuter verbs" in Hooper 1984b); the latter are a small class of verbs with inherently passive meaning, distinguished by the impossibility to enter into a nominal construction. In Rapa Nui, no such distinction can be made.

#### 3 Nouns and verbs

(91) *Ka* imp *koa* happy *mai* while *ꞌāpī* new *ꞌe* and *mai* while *nehenehe* beautiful *a* prop *koe.* 2sg 'Be happy as long as you are young and beautiful.' [R453.018]

Stative verbs in Rapa Nui are also characterised by the use of the perfect aspect *ko V ꞌā* to express a present situation; however, this use also occurs with certain categories of active verbs (§7.2.7.2).

Intransitive verbs are united by two features:


As discussed in §8.6.4.7, this mainly happens with non-agentive verbs (categories 2 and 3 in the previous section), but given the right context, it may also occur with agentive intransitives (category 1).

Adjectives can be considered as a subclass of stative verbs and will be discussed in §3.5. Even though there are no clear-cut criteria to distinguish adjectives from other verbs (especially from patientives), in §3.5.1 it will be shown that there are sufficient grounds to recognise adjectives as a separate subcategory.

Table 3.2 lists the different types of verbs with their features.

### **3.5 Adjectives**

Adjectives are words denoting properties. As Bhat (1994) points out, adjectives differ from nouns in that they refer to a single property, while nouns refer to a cluster of properties. Adjectives differ from verbs in that they denote a time-stable property, while verbs denote a transient event.

§3.5.1 discusses the question whether adjectives form a separate part of speech in Rapa Nui and how they can be distinguished from other words, especially verbs.<sup>34</sup>

<sup>34</sup> According to Croft (2000), adjectives are intermediate between verb and noun. A prototypical adjective describes a property and acts as a modifier; properties are intermediate between objects and actions (one could think of a scale of time-stability here), while modification is intermediate between reference and predication. Therefore, in a language like Rapa Nui, where there is so much interaction between noun and verb, it is only to be expected that adjectives are even harder to distinguish.

According to Dixon, it is probable that every language has a class of adjectives (Dixon 2004; Dixon 2010a: 53; Dixon 2010b: 62, 104; different from Dixon 1982), though the criteria to distinguish adjectives from either nouns or verbs may be subtle and not obvious at first sight. Dixon (2004), Dixon (2010b: 70– 73) suggests criteria to distinguish adjectives from verbs and nouns. Note however, that out of thirteen language descriptions in Aikhenvald & Dixon (2004), five authors consider adjectives as members of the verb class, even though there are differences between adjectives and (other) verbs (e.g. Hajek 2004; Hyslop 2004).

3.5 Adjectives


Table 3.2: Types of verbs

§3.5.2 discusses degrees of comparison, a grammatical category largely confined to adjectives.

### **3.5.1 Does Rapa Nui have adjectives?**

#### **3.5.1.1 Adjectives as a prototypical category**

In Polynesian languages – and in Oceanic languages in general – property words such as 'big' and 'good' tend to behave like verbs; for example, they are often preceded by an aspect marker and function as predicate of the clause. Many grammars therefore deny that adjectives are a separate word class; rather, they are considered as verbs. On the other hand, property words are sufficiently different from action words to be classified as a separate subclass of verbs. As discussed in §3.4.2 above, in Rapa Nui – as in other Polynesian languages – a class of stative verbs can be distinguished; this class includes typical adjectives such as size and colour terms, but also non-active verbs like 'die' and 'be forgotten'.

The question is, whether it is possible in Rapa Nui to distinguish a subcategory of adjectives within the stative verbs. Englert (1978: 28) remarks: "Es dudoso si en el idioma rapanui existe el adjetivo propiamente así llamado. Tal vez hay solamente adjetivos verbales o participios." (It is dubious if the adjective properly so called exists in the Rapanui language. Perhaps there are only verbal adjectives or participles.) Property words in

#### 3 Nouns and verbs

Rapa Nui behave like verbs in most respects. On the other hand, there are also significant differences, as will be shown in the following sections. These differences are sufficiently far-reaching to recognise adjectives as a separate subclass within the category of verbs. At the same time, it is impossible to draw a sharp line between adjectives and other verbs; I have not found a single criterion which sharply and clearly defines a category of adjectives. The boundary between adjectives and verbs is fluid in two ways. First, it is not possible to give an exact list of adjectives; some words are more adjectival than others.<sup>35</sup> Second, some contexts are more adjectival than others, so that a given word may show more adjectival or more verbal behaviour, depending on the context. The adjectival category can therefore best be defined in terms of a prototype (cf. the same approach for nouns and verbs in §3.2.1.2), which unites certain semantic, pragmatic and syntactic properties. A prototypical adjective


This raises the question whether less prototypical cases are also labelled as adjectives, and if so, how far the use of this label is extended. For practical reasons, in this grammar the term *adjective* is used for property words modifying a noun, and in a looser sense also for property words in other syntactic positions.

In the following sections, I will discuss adjectival characteristics and show to what extent these may serve to distinguish adjectives from other words.

#### **3.5.1.2 Morphology of adjectives**

Two things can be said about the morphology of adjectives. Firstly, some adjectives are full reduplications. This is true for


<sup>35</sup> Hohepa (1969a: 8) lists adjectives in Māori (as distinguished from stative verbs) on the basis of a number of syntactic and morphological criteria. However, as Harlow (2007a: 106) points out, other attempts to list Māori adjectives exhausively have resulted in somewhat different lists.

<sup>36</sup> Reduplications as basic colour terms are common in Oceanic languages, even though (a) the use of reduplications as basic lexemes is unusual in Austronesian; (b) it is typologically unusual to have morphologically complex words as basic colour terms (Blust 2001; Blust 2013: 304. Blust (2001: 42) suggests that reduplications originally had an intensive sense, which lost its intensity over time through frequent use.

3.5 Adjectives

For most of these, the simple form also exists, but with a marked sense and limited use. For example, even though both *ꞌiti* and *ꞌitiꞌiti* are used adnominally and adverbially, *ꞌiti* is more common as an adverb, while *ꞌitiꞌiti* is predominantly adnominal. While *rivariva* means 'good', adnominal *riva* means either 'good' or 'pretty'. The reduplicated forms may have had an intensifying sense originally, but nowadays they are the default forms in most contexts. In some case the sources exhibit a shift over time: while *rake* 'bad' occurs in old texts, in modern Rapa Nui only *rakerake* is found.

Secondly: Just like some verbs, a number of adjectives have a separate plural form, which is partially reduplicated. For example: *roaroa* 'long', *roroa* 'long (Pl)'; *rivariva* 'good', *ririva* 'good (Pl)'. The plural forms may be used when the denoted entity is plural, but their use is optional.

*ꞌItiꞌiti* 'small' has a suppletive plural *rikiriki*; the use of this form is obligatory when the adjective modifies a plural noun or is a predicate with a plural subject.

#### **3.5.1.3 Syntactic function: adnominal and other uses**

The prototypical syntactic function of adjectives, which distinguishes it from nouns and verbs, is adnominal: adjectives typically modify a head noun (Croft 2000). Now this fact alone is not sufficient to distinguish adjectives from nouns and verbs, as the latter are used adnominally as well (§5.7.1). Moreover, no adjective is used *exclusively* as a noun modifier: the same words also serve as predicates, NP heads and/or adverbs, and many also serve as a base for causativisation. The following examples of *rivariva* 'good' illustrate this:


#### 3 Nouns and verbs

Even though adnominal use as such cannot serve as an absolute criterion, the *frequency* of adnominal use may be used as a diagnostic. Words denoting events and objects (i.e. verbs and nouns) are used adnominally only occasionally, while for property words adnominal use is quite common.

The frequency of adnominal use differs considerably between different adjectives: some are mainly used adnominally, others are mainly used in other functions.<sup>37</sup> For example, *nuinui* 'big' is adnominal in 58.3% of all occurrences in the text corpus,<sup>38</sup> while *rivariva* 'good, well' is adnominal in only 24.6% of all occurrences.<sup>39</sup> Even so, for both of these, adnominal use is considerably more common than for the noun *taŋata* 'man', which is adnominal in 2.3% of all occurrences (72 out of 3120), or the verb *oho* 'to go', which is adnominal in 1.0% of all occurrences (51 out of 5011).

When adjectives are grouped in semantic categories, such as suggested by Dixon (2010b: 73), some patterns emerge, as shown in Table 3.3. <sup>40</sup> This table gives the total number of occurrences for the following categories:


Table 3.3 shows that words denoting colour, age and dimension are mostly used adnominally. For value terms, the adnominal function is the most common one as well, though it accounts for only 36.8% of all occurrences. For all other categories, less than

<sup>37</sup> In the frequency counts in this paragraph, adjectives that are part of a name are excluded. Also excluded are syntactically isolated adjectives, e.g. in lists and appositions.

<sup>38</sup> 403 occurrences in total; 25.8% are predicate, 9.7% are NP heads and 3.7% are adverbs.

<sup>39</sup> 837 occurrences in total; 19.6% are predicate, 4.3% are NP heads and 37.8% are adverbs.

<sup>40</sup> For this and the following section, I analysed a number of common adjectives from different semantic categories. See the following footnotes for a listing. In the table, values over 20% are in bold.

#### 3.5 Adjectives

20% of the occurrences are adnominal; these words are more commonly used as predicate or as noun. We may conclude that dimension, age and colour terms are the most prototypical adjectives, as far as their syntactic function is concerned; value adjectives are close to prototypical. This coincides with Dixon's generalisation (Dixon 2012: 73) that if a language has any adjectives at all, it will have at least some adjectives from (some of) these four categories.

#### **3.5.1.4 Adnominal adjectives versus adnominal nouns and verbs**

The previous paragraph showed, that adjectives show a high frequency of adnominal use compared to nouns and verbs. Apart from this, adnominal adjectives are also different in function and syntax from adnominal nouns and verbs. In the first place, modifying nouns are usually part of a compound, expressing a single concept together with the head noun, while modifying adjectives specify an additional property of the concept expressed by the head noun (§5.7.1). Modifying nouns are incorporated into the head noun; different from adjectives, they cannot be followed by modifying particles, while adjectives may be accompanied by e.g. degree markers and adverbs (§5.7.3.2).

Modifying verbs occur in two constructions. First, they may form a compound together with the head noun (§5.7.2.3); in this case, they express a single concept together with the head noun, and the same constraints apply as with modifying nouns. Alternatively, modifying verbs may be the head of a relative clause (§11.4), which consists of a verb phrase optionally followed by one or more arguments or adjuncts. The verb in a relative clause is often preceded by an aspect marker. By contrast, prototypical adjectives – such as terms of dimension, age and colour – are never preceded by an aspectual marker when used adnominally.

Table 3.3: Uses of adjectives


*<sup>a</sup>* Most nominal uses are cases of *mātāmuꞌa* 'past', which is often used as a noun 'the past, the old days', and *hō ꞌou* 'new', which is used idiomatically as a term of endearment. Without these two, figures for this category would be as follows:


#### 3 Nouns and verbs

Less prototypical adjectives (such as those of position and physical property) do occur in aspect-marked relative clauses, though only occasionally. In the following example, *hāhine* 'near'+ is used in a relative clause:

(98) *ꞌIna* neg *takoꞌa* also *o* of *Oceanía* Oceania *te* art *taꞌatoꞌa* all *henua* land *era* dist *e* ipfv *hāhine* near *era* dist *ki* to *Asia.* Asia 'Not all the islands that are close to Asia belong to Oceania either.' [R342.008]

*Hāhine* is mostly used as predicate; its adnominal use is relatively rare, which suggests that it is not a prototypical adjective.

Now Rapa Nui also has "bare relative clauses", relative clauses in which the verb is not preceded by an aspect marker (§11.4.5). One could ask whether an adnominal adjective is structurally identical to the verb in a bare relative clause. After all, there are certain similarities between both, besides the absence of the aspect marker. For one thing, adnominal adjectives may be preceded by degree markers and followed by adverbs (§5.7.3.2), elements which also occur in verb phrases (§7.3.2; §4.5.1). Adjectives may enter into the comparative construction, but verbs occasionally enter into this construction as well (see (95) in §7.3.2).

However, there are also structural differences between adnominal adjectives and bare relative clauses. Adnominal adjectives do not take the full range of postverbal particles: they are never followed by the evaluative markers *rō* and *nō*, or by directionals *mai* and *atu*. This is true for all adjectives included in Table 3.3 in the preceding section, not just the prototypical categories. Verbs in relative clauses, on the other hand, do take the full range of postverbal particles.<sup>41</sup>

When adjectives are used predicatively, these restrictions do not hold: not only are predicate adjectives preceded by an aspectual marker, they can be followed by evaluative markers, or by a directional marker as in the following example:<sup>42</sup>

(99) *Ku* prf *rikiriki* small:pl:red *atu* away *ꞌā* cont *te* art *ika* fish *nei* prox *pē* like *he* pred *tapatea* kind\_of\_eel *ꞌana.* ident 'These fish are quite small, just like *tapatea*.' [R364.015]

Another difference between verbs and adnominal adjectives is, that the latter are only followed by a limited set of adverbs, all of which express a degree: *rahi* 'much', *riꞌariꞌa* 'very, terribly', *taparahi-taꞌata* 'terribly', or *tano* 'in a moderate degree' (§5.7.3.2). With the exception of *rahi*, these adverbs do not occur in the verb phrase, while on the other hand most verb phrase adverbs do not occur in the adjective phrase (§4.5.1).

We may conclude that there are subtle but clear semantic and structural differences between adnominal adjectives and verbs. Together with the higher frequency of adnominal use of adjectives, this suggests that the prototypical adjective is different from a verb.

<sup>41</sup> See also sec. §5.7.2.3 on the difference between modifying verbs as compounds and bare relative clauses.

<sup>42</sup> Examples such as (99) are not very common, as time-stable properties are not naturally associated with directionality. In the example above, *atu* is used in the sense of extent (§7.5.1.5).

3.5 Adjectives

#### **3.5.1.5 Predicate adjectives**

Adjectives are used as verbal predicates (i.e. predicates marked with verbal particles) to express non-permanent properties, properties which characterise their argument during a moment or a period of time. Permanent properties are expressed in nominal clauses, in which the adjective modifies a nominal predicate (§9.2.7).

Adjectives and verbal predicates may take the full range of aspect markers discussed in §7.2: neutral *he*, perfective *i*, imperfective *e*, contiguity *ka* and perfect *ko*. Below are some remarks on specifically adjectival uses (or non-uses) of aspect markers.

**The contiguity marker** *ka ka* is used with adjectives in the same way as with any verb. However, there is one use of *ka* which only occurs with certain adjectives, the exclamative construction discussed in §10.4.1.

**Imperfective** *e* As discussed in §7.2.5.4, *e* with adjectives commonly occurs in the construction *e V (nō/rō) ꞌā*, but rarely in the construction *e V PVD*. *E V nō ꞌā* indicates that a state still exists, implying that it could end at some point, but has not ended yet.

(100) *Te* art *poki* child *nei* prox *e* ipfv *ꞌitiꞌiti* small:red *nō* just *ꞌā.* cont 'This child is still small.' [R532-14.007]

**Perfect** *ko V ꞌā ko V ꞌā* indicates that a state has been reached as the result of an otherwise unstated process:


Now this use of *ko V ꞌā* is not restricted to adjectives, but occurs with a much wider range of verbs, including certain types of active verbs (§7.2.7.2).

**Neutral** *he He* with adjectives expresses a state as such.

(103) *He* ntr *rivariva* good:red *tāꞌana* poss.3sg.a *aŋa* work *era* dist *ka* cntg *aŋa* do *era.* dist 'The work he was doing, was good.' [R313.116]

*He* + adjective may be used in situations where a state starts to exist, as in the following examples:

#### 3 Nouns and verbs


In such cases, the clause can be labeled inchoative; however, this is not expressed by *he* as such, but simply a feature which can be inferred from the context.

**Other preverbal markers** Just like verbs, adjectives can also be used with the modal markers *ana*, *mo* and *ki*, and be preceded by the verbal negators *ꞌina*, *kai* and *e ko*. Two examples:


#### **3.5.1.6 Nominal use of adjectives**

As pointed out in §3.5.1.3 above, adjectives can be used nominally, i.e. as heads of noun phrases. Nominal adjectives refer to a property as such, not to an object possessing the property: *rivariva* 'goodness', not 'a good one' (§5.6):


Verbs are also used nominally in a variety of constructions (§3.2.3.1). However, two nominal constructions occur only with adjectives, not with verbs.<sup>43</sup> Both have an exclamative sense.

1. Exclamative *ꞌai te X* is only found with adjectives of size, such as *nuinui* 'big' and *kumi* 'long' (§10.4.3).

<sup>43</sup> See Bhat (1994: 29): adjectives are typically able to be the basis of exclamations.

3.5 Adjectives

2. Exclamative *ko te X* is used with both nouns and a wide range of adjectives (value, physical property, size etc.) (§10.4.2).

Nominally used adjectives usually do not have a nominalising suffix; in this respect they differ from verbs. For example, in (95) in §3.5.1.3 above, *rivariva* is used as object of a verb of perception; in this context, verbs normally get a nominalising suffix (§3.2.3.1), but *rivariva* does not.

There are two contexts in which adjectives do have a nominalising suffix:

	- 2. In the construction *ko te V iŋa ꞌā* (§3.2.3.1.1):

#### **3.5.1.7 Conclusions**

The previous sections have shown that property words differ in their syntactic behaviour from event words in a number of respects:


#### 3 Nouns and verbs

This allows the conclusion that Rapa Nui has an adjective category. However, given the close correspondence with verbs, it is best to consider adjectives as a subclass of verbs, more specifically, of stative verbs.

The discussion has also shown that the adjectival category is not a monolithic one. Some adjectives – especially those denoting colour, age and dimension – are more prototypical than others.

### **3.5.2 Degrees of comparison**

#### **3.5.2.1 The comparative**

Rapa Nui has a number of different comparative constructions.<sup>44</sup> In one of these, the particle *ꞌata* serves as index of comparison ('more, -er'); it precedes the adjective expressing the parameter of comparison. This construction can be used whether the adjective is adnominal as in (113) or predicative as in (114). The standard of comparison is expressed by *ki* + noun phrase.


(114) *¿*[*ꞌAta* ] more index [*maneŋe* ] medium\_size parameter [*koe* ] 2sg comparee [*ki te poki era ai*]*?* to art child dist there standard 'Are you smaller than that boy there?' [R415.176]

*ꞌAta* also functions as a degree marker in front of event verbs (§7.3.2, where its etymology is also discussed). With verbs, it may also form a complete comparative construction, including a standard of comparison (see (95) on p. 341).

A second construction uses the verb *hau* 'to exceed, surpass, be superior', with the comparee as subject. The parameter of comparison is marked with the locative preposition *ꞌi*. The standard of comparison is expressed by *ki* + noun phrase, as in the *ꞌata*construction above.

(115) [*E hau rō atu* ] ipfv exceed emph away index [*a ia* ] prop 3sg comparee [*ꞌi te roroa* ] at art red:long parameter [*ki a au* ]*.* to prop 1sg standard 'He is taller than me (lit. he is more/surpassing in length to me).' [Notes]

<sup>44</sup> For the different elements in comparative constructions, I use the following terms:

*Susan (is) more intelligent than Mary* comparee index parameter standard

3.5 Adjectives

*Hau* can in turn be reinforced by *ꞌata*, in which case the aspectual marker before *hau* tends to be left out.

(116) [*ꞌAta hau* ] more exceed index *hoꞌi* indeed [*a Veriamo* ] prop Veriamo comparee [*ꞌi te reherehe* ] at art weak:red parameter [*ki a meꞌe ki a Eva* ]*.* to prop thing to prop Eva standard 'Veriamo was weaker than what's-her-name, than Eva.' [R416.171]

In the older language, comparisons are sometimes made without any marking; only *ki* indicates that a comparison is made:

(117) *Te* art *poki* child *nei* prox *poki* child *maꞌori* expert *ki* to *tētahi* other *poki.* child

'This child is more intelligent than the other.' [Englert 1978: 30]

Although this sentence still sounds acceptable nowadays, speakers of modern Rapa Nui would tend to add *ꞌata* in front of *maꞌori*.

#### **3.5.2.2 The superlative**

The superlative can be expressed by *hope ꞌa* 'last' (a Tahitian loan not found in old texts), followed by a genitive phrase which contains a nominalised adjective:


*Hope ꞌa* can also be used in a superlative sense without a qualifying adjective, to express that something is 'ultimate, extreme', whether in a positive or negative sense:

(120) *Te* art *tai* sea *hopeꞌa* last *mo* for *te* art *hāhaki* gather\_shellfish *he* pred *tai* sea *pāpaku.* thin 'The best tide for gathering shellfish is low tide.' [R353.018]

*Hope ꞌa* + genitive is also used for the absolute superlative: 'very'.

#### 3 Nouns and verbs


In the older language, the superlative can be expressed by the adjective as such, without any special marking; such unmarked superlatives are obsolete nowadays.

(123) *Te* art *maꞌuŋa* mountain *Terevaka* Terevaka *te* art *maꞌuŋa* mountain *nuinui* big *o* of *te* art *kāiŋa.* homeland 'Mount Terevaka is the biggest hill of the island.' [Englert 1978: 30]

#### **3.5.2.3 The equative**

The equative, 'X is as [Adj] as Y', is expressed using the preposition *pē* 'like' (§4.7.9). The quality with respect to which the two entities are compared, may be expressed as a noun modifier, such as *rikiriki* in the following example:

(124) *He* ntr *hakarē* leave *i* acc *a* prop *Tiare* Tiare *ꞌi* at *muri* near *i* at *te* art *tētahi* other *ŋā* pl *poki* child *rikiriki* small:pl:red *pē* like *ia* 3sg *ꞌā.* ident

'He left Tiare with the other children that were as small as her.' [R481.034]

But more commonly, it is expressed as a noun phrase:


3.6 Locationals

### **3.6 Locationals**

### **3.6.1 Introduction**

Rapa Nui has a set of words serving to locate entities in space. These words behave somewhat like nouns, yet are a class of their own, and are called locationals in this grammar.<sup>45</sup>

Different groups of locationals can be distinguished.

	- (127) *A* prop *nua* Mum *ꞌi* at *roto* inside *i* at *te* art *hare.* house 'Mother is in the house.' [R333.284]

*Roto* locates mother with respect to the house. The preceding preposition *ꞌi* indicates that this locative relationship is stable: there is no movement involved towards a position inside the house, or from the inside to the outside.

In this grammar, this first group is called relative locationals.<sup>46</sup>

	- (128) *He* ntr *turu* go\_down *a* prop *koro* Dad *ki* to *tai.* sea 'Dad went down to the seashore.' [R333.388]
	- (129) *E* ipfv *vaꞌu* eight *mahana* day *i* pfv *noho* stay *ai* pvp *ꞌi* at *nei.* prox

'He stayed here (=on Rapa Nui) for eight days.' [R374.005]

<sup>45</sup> All Polynesian languages have such a word class. They have been called local nouns (Churchward 1985 [1953]; Bauer 1997; Mosel & Hovdhaugen 1992; Besnier 2000), locationals (Clark 1976: 59, Clark 1986, Du Feu 1996), L-nouns (Clark 1976: 55), locative nouns (Elbert & Pukui 1979) and locatives (Biggs 1973; Bowden 1992). For the relative locationals, Harlow (2007a: 145) uses the term *relator nouns*.

<sup>46</sup> Note, however, that there is not always a second referent involved. These same locationals can also indicate a general direction:

<sup>(</sup>i) *He* ntr *rere* jump *a* by *ruŋa.* above 'He jumped up.'

#### 3 Nouns and verbs

4. Temporal locationals. There is a small group of time words belonging to the locational class, such as *ꞌaŋataiahi* 'yesterday'.

What all locationals have in common is that they can be preceded by prepositions, like common nouns. Unlike common nouns, they do not take determiners: the preposition immediately precedes the locational. Nor do they take the proper article, as proper nouns do.

Another class of lexical items commonly used in Rapa Nui discourse for spatial orientation, is the class of geographical names, such as *Tahiti* 'Tahiti'. As discussed in §3.3.2, these can be immediately preceded by prepositions; unlike personal names, they do not take the proper article. Therefore they do not belong to the class of proper nouns, but to the locationals.<sup>47</sup> Geographical names will not be discussed in further detail.

The following sections discuss relative (§3.6.2) and absolute (§3.6.3) locationals. Deictic locationals are very similar in form and function to demonstratives and are discussed in the section on demonstratives (§4.6.5). §3.6.4 discusses temporal words belonging to the class of locationals. §3.6.5 shows which modifying elements may occur in the locative phrase.

Finally, the interrogative *hē* partly behaves like a locational as well; it is discussed in §10.3.2.3.

### **3.6.2 Relative locationals**

Relative locationals, in Polynesian linguistics often simply called 'locationals', indicate basic spatial relationships. They are listed in Table 3.4.


Table 3.4: Relative locationals

<sup>47</sup> Clark (1976: 54) likewise classifies proper names of places among the locationals.

3.6 Locationals

Most of these have the same basic sense throughout the Polynesian languages, though the Rapa Nui locationals underwent some idiosyncratic developments.<sup>48</sup>

In the following subsections, these locationals are discussed in detail. Sections §3.6.2.1 and §3.6.2.2 discuss the syntax of locational constructions. §3.6.2.3 discusses the semantics of certain locationals and locational expressions. This is continued in §3.6.2.4, which discusses the temporal use of certain locationals.

#### **3.6.2.1 Adverbial expressions**

Locationals are usually preceded by one of the locative prepositions discussed in §4.7. Together with these prepositions, the locationals form adverbial expressions of location. Here are a few examples:


Adverbial expressions like these often have an absolute sense. For example, in (131) *ki raro* indicates 'down, in a lower direction'. In other cases, the locational is interpreted relative to a second referent, which is implied. In (133), the context makes clear that *o ruŋa* is to be interpreted with respect to an island.

<sup>48</sup> This becomes clear when we compare the Rapa Nui forms and meanings with their PPN equivalents, as given in Pollex (Greenhill & Clark 2011): PPN *\*muri* meant 'behind, after, to follow, be last'. Rapa Nui is the only language in which its meaning shifted to 'proximity'; the original meaning is still present in older texts. PPN *\*tupuaki* meant 'the top of the head'. In no other language did it develop into a locational. PPN *\*waheŋa* is glossed as 'division, portion, share, piece of land; middle'. In many languages it is a common noun, and Clark (1976) does not list it as a locational in PPN; however, in most EP languages it does occur as a locational: Marquesan *vaveka* (Cablitz 2006: 331–332, *k* < PPN *\*ŋ*), Māori *vaenganui* (Biggs 1973: 41), Hawaiian *waena* (Elbert & Pukui 1979: 121), Pa'umotu *vaeŋa* (Stimson 1964: 594), Mangarevan *vaega* (Tregear 2009: 118). It does not occur in Tahitian.

#### 3 Nouns and verbs

#### **3.6.2.2 Locationals with complement: prepositional expressions**

The adverbial expressions discussed in the previous section can be followed by a preposition + noun phrase to indicate a spatial relationship with respect to a second referent. The combination of preposition + locational + preposition acts as a sort of complex preposition, in which the locational indicates the spatial relationship between two referents, and the initial preposition the way in which this relationship holds. In the following example, *roto* expresses that the spatial relationship is such that referent A (the cat) is inside referent B (the house). The preposition *ki* expresses that referent A moves towards that location.

(134) *He* textscntr *uru* enter *te* art *kurī* cat *ki* to *roto* inside *i* at *te* art *hare.* house 'The cat entered into (lit. to inside) the house.' [Notes]

The second preposition does not have any semantic contribution; it serves just to provide a syntactic link between the locational and its complement. The following examples show different ways in which this preposition can be realised:


As these examples show, the second preposition may be either *i* as in (134), *o* as in (135), or a copy of the first preposition as in (136–137). When the first preposition is *ꞌi* or *o*, the analysis of the second preposition is ambiguous: in *ꞌi raro i* in (138), the second preposition may be either a default preposition *i*, or a copy of the first preposition (*ꞌi* and *i* are variants of the same preposition, see §4.7.2). The same is true for *o tuꞌa o* in (139).

3.6 Locationals

As Clark (1976: 54–55) points out, all other Polynesian languages use either *i* or *o* as second preposition;<sup>49</sup> Rapa Nui is the only language in which the second preposition may be a copy of the first.<sup>50</sup>

In older texts the copying strategy is used in an overwhelming majority of the cases. Not counting the ambiguous *ꞌi* LOC *i* and *o* LOC *o* constructions, the second preposition is a copy of the first in 93% of all PREP + LOC + PREP constructions in this corpus (768 out of 826). Thus, constructions like (136–137) are common in older texts, while constructions such as (134–135) are rare.

In modern Rapa Nui the copying strategy is still in use, as illustrated by (137) above, but it has become relatively rare, occurring in only 10% of all nonambiguous cases (175 out of 1761).<sup>51</sup> And some of these are, on a closer look, not copies at all, but prepositions introducing a new constituent. The following example illustrates this:

(140) *I* pfv *oti* finish *era* dist *he* ntr *turu* go\_down *ki* to *raro* below *ki* to *te* art *teata* cinema *mātaꞌitaꞌi.* observe 'After that they went down to the theatre to watch.' [R210.145]

This is not a case of a complex preposition 'to below N': *ki raro* is not interpreted relative to the second referent *te teata* (in that case, people would go to a location below the cinema); rather, *ki raro* and *ki te teata* are two separate, parallel constituents.

Instead of a copy of the first preposition, the second preposition is usually *i* or *o* nowadays; both are used without a clear difference in meaning.<sup>52</sup>

In general, *i* is more common in modern Rapa Nui than *o*: over the whole corpus of

(i) *He* ntr *topa* descend *mai* hither *te* art *timo* warrior *ki* to *roto* inside *ki* to *te* art *ꞌana* cave *o* of *ꞌAna* Ana *te* te *Ava* Ava *Nui.* Nui 'The warriors were dragged into the cave of Ana te Ava Nui.' (Mtx-3-03.231)

<sup>49</sup> Vaitupu (a dialect of Tuvaluan) is the only variety apart from Rapa Nui where both *o* and *i* are used, without apparent difference in meaning.

<sup>50</sup> As Clark (1976: 56–57) indicates, the copying construction could have arisen from cases like (138) or (139): the second preposition, which originally was an invariable *i* or *o*, was reanalysed as a repetition of the first one. This reanalysis could have been facilitated by constructions like the following (quoted by Clark):

While such constructions could originally have consisted of two parallel phrases: 'inside, to the cave' they could easily be reanalysed as a single phrase 'into the cave', in which the second *ki* is a copy of the first. According to Finney & Alexander (1998: 27–28), *ki … ki* also occurs in Vaitupu and, in some constructions, in Māori; however, this does not amount to a generalised copying strategy as in Rapa Nui.

<sup>51</sup> This tendency is even stronger in the Bible translation, which is more recent than most of the newer texts: in the new Testament, the preposition is *i* or *o* in over 99% of the prepositions, with *i* in the overwhelming majority (88%).

<sup>52</sup> Just like the copying construction may be the result of reanalysis (see Footnote 50 above), the choice for *i* or *o* could also have been brought about by reanalysis: in expressions like *ꞌi ruŋa i* and *o roto o*, the second preposition (which was a copy of the first) was reanalysed as default *i* or *o*, and their use was subsequently generalised. *I* lends itself to a generalised use as it is the most general locative preposition; *o* lends itself to a generalised use as it is common as genitive marker. Notice that it is not uncommon for spatial relationships to be expressed by the genitive (see Dixon 2010b: 285). Fischer (2001a: 324) considers the generalisation of *i* as second preposition as a development under Spanish influence.

#### 3 Nouns and verbs

modern texts, *i* outnumbers *o* in a proportion of 2:1.<sup>53</sup> The choice between the two is free to a certain degree, but certain tendencies can be observed:

	- 2. The choice between *i* and *o* also correlates with the choice of locational: *o* is more common with *raro*, *muꞌa* and *tuꞌa*, while *i* is more common with *roto*, *ruŋa* and *muri*. The preposition preceding the locational does not play a role.

The locational *roto* and the following article *te* are often contracted: *roto (i/o) te > rote*. This contraction is a recent development; it does not occur in older texts.

(142) *E* voc *koro,* Dad *¿e* ipfv *aha* what *ꞌā* cont *koe* 2sg *ꞌi* at *rote* inside\_art *ꞌua?* rain 'Dad, what are you doing in the rain?' [R210.097]

#### **3.6.2.3 The semantics of some locationals**

This section discusses the meaning of some individual locationals, and of some locational expressions.

**3.6.2.3.1** *Muri* in older texts means 'after': either in spatial sense ('behind'), or in a temporal sense ('afterward').


In modern Rapa Nui, *muri* indicates spatial proximity, 'close to, next to':

(145) *He* ntr *tuꞌu* arrive *ki* to *muri* near *ki* to *te* art *pahī,* ship *he* ntr *ekeeke* go\_up:red *ki* to *ruŋa.* above 'They came alongside the ship and went on board.' [R210.081]

<sup>53</sup> *Pace* Finney & Alexander (1998: 28), who claim that "*o* has largely displaced earlier *i* as right-side preposition".

**3.6.2.3.2** *Tuꞌa* refers to the back of something. *ꞌI tuꞌa* normally refers to a location behind, on the outside of something: *ꞌi tuꞌa o te hare* 'behind the house'. But in some situations it may refer to a location within, at the back side. This may occasionally lead to ambiguities:

(146) *Ka* imp *hakarē* leave *te* art *bombona* gas\_bottle *ꞌi* at *tuꞌa* back *o* of *Te* art *kamioneta.* van 'Put the gas bottle behind the van', or: '…in the back of the van.'

*ꞌI tuꞌa* in this example refers to a location either inside or outside the car.


In relation to a set of two referents it indicates a location in between the two:

(150) *ꞌI* at *vāeŋa* middle *o* of *te* art *hare* house *nei* prox *ꞌe* and *o* of *te* art *hare* house *era* dist *te* art *karapone.* shed 'The shed is between this house and that house.' [Notes]

**3.6.2.3.5** Some combinations of preposition + locational have specialised meanings: *a raro* 'on foot':

(151) *Ko* prom *koro* Dad *a* by *raro* below *ꞌā* ident *i* pfv *iri* ascend *ai.* pvp *Ko* prom *nua* Mum *a* by *ruŋa* above *te* art *hoi.* horse 'Dad goes up (to the field) on foot. Mum goes on horse.' [R184.052–053]

#### 3 Nouns and verbs

*a vāeŋa* 'in half':

(152) *Ana* irr *haŋa* want *he* ntr *ꞌavahi* divide *a* by *vāeŋa,* middle *hoa* throw *hai* ins *miti…* salt 'If you want, you cut (the fish) in half, put salt on…' [R185.007]

*o ruŋa i* 'about', in the sense of a topic of knowledge or discourse:

(153) *He* pred *ꞌuiꞌui* ask:red *nō* just *te* art *aŋa* do *o* of *ruŋa* above *i* at *te* art *ꞌaꞌamu* story *tuai.* ancient 'He was always asking about the old stories.' [R302.018]

#### **3.6.2.4 Temporal use of locationals**

While *muꞌa* 'front' and *tuꞌa* 'back' are primarily spatial terms, they are also used temporally, referring to past and future. However, the temporal dichotomy between past and future does not coincide with the spatial dichotomy between front and back – in other words, it is not the case that *muꞌa* refers to the future and *tuꞌa* to the past, or the other way around. Rather, *muꞌa and tuꞌa* acquire specific temporal senses in combination with certain prepositions.<sup>54</sup>

*Pe muꞌa* (often in the expression *pe muꞌa ka oho ena*) means 'later, in the future':


*A tuꞌa* either means 'before, ago' or 'later, afterwards':

(156) *E* ipfv *ai* exist *rō* emph *ꞌā* cont *te* art *rivuho* drawing *… meꞌe* thing *rahi* many *matahiti* year *a* by *tuꞌa* back *i* pfv *aŋa* make *ai.* pvp

'There is a drawing… made many years ago.' [R296.010–013]

<sup>54</sup> See Tetahiotupa (2005) for an equally complex situation in Tahitian. Temporal reference leads itself easily to ambiguity, as there are two fundamentally different ways to conceptualise the passage of time: either the world is seen as fixed and time moves from the future to the past, or time is fixed and we travel through it from the past to the future (see Anderson & Keenan 1985: 296). In the second case, the future is clearly ahead, while the past is behind. On the other hand, as the past is known and therefore visible while the future is unknown and invisible, the past can be conceived as being before our eyes, while the future is behind our backs.

3.6 Locationals

(157) *Ka* cntg *rima* five *taꞌu* year *a* by *tuꞌa* back *… he* ntr *manaꞌu* think *hakaꞌou* again *a* prop *ia* 3sg *ki* to *a* prop *Rokeꞌaua* Roke'aua *ararua* the\_two *ko* prom *Makita.* Makita 'Five years later he thought again of Roke'aua and Makita.' [R243.205]

*ꞌI/o muꞌa ꞌā* means 'first, in the past':

(158) *Te* art *mana* power *ꞌi* at *muꞌa* front *ꞌā* ident *meꞌe* thing *pūai.* strong '*Mana* (supernatural power) was something strong in the past.' [R634.001]

### **3.6.3 Absolute locationals**

Polynesian languages have a small set of locationals which locate a person or object with respect to a certain generally known geographical area.<sup>55</sup> These can be labelled 'absolute locationals'.

The absolute locationals are listed in Table 3.5:

Table 3.5: Absolute locationals


*Kampō* is borrowed from Spanish *campo* 'field, countryside'.<sup>57</sup> The other words are common in the Polynesian languages.<sup>58</sup>

Like the relative locationals, these words are immediately preceded by prepositions. Unlike the relative locationals, they cannot be followed by a prepositional phrase indicating a second referent with respect to which the spatial relation holds.

<sup>55</sup> Cf. Levinson & Wilkins (2006: 21): "The absolute frame of reference in ordinary language use requires fixed bearings that are instantly available to all members of the community." See Cablitz (2005) for a discussion of absolute or geocentric localisation in another Polynesian language, Marquesan.

<sup>56</sup> Based on its meaning *kōnui* would seem to belong to the category of deictic locationals (§4.6.5 below). However, syntactically it behaves like the absolute locationals, in that it can be followed by the postnominal demonstrative *era*; see section §3.6.5 about elements modifying locationals.

<sup>57</sup> Rapa Nui is not the only language in which the class of locationals has been extended with borrowings. For example, in Tongan, *uafa* 'wharf', *piliisone* 'prison' and *sitima* 'steamer' are locationals. See Clark (1976: 55).

<sup>58</sup> Most Polynesian languages have a locational *kō* 'there', often modified by deictics *nei*, *ena* or *era* to indicate the degree of distance. *Kōnui*, in which *kō* is modified by *nui* 'big', is its only Rapa Nui reflex. (Similarly, in Rapa Nui *raro nui* became lexicalised, meaning 'deep', and *ruŋa nui*, meaning 'high'.)

#### 3 Nouns and verbs

The following sections discuss each of these locationals in turn. First, however, a general note on spatial reference. As the list above shows, the main reference points for spatial orientation in Rapa Nui are related to the sea. Spatial reference in Rapa Nui reflects the geography of the environment in which the language is spoken: a single island, a closed world of limited dimensions. In this world, the coast is always close; it is either visible, or one knows at least in which direction it is. It is not surprising that orientation happens predominantly with respect to the sea.<sup>59</sup>

As the speech community is small and the area is limited, common orientation points (most of them on the island, a few outside, like Tahiti and the mainland) are generally known by name. Therefore, spatial reference in stories often happens by place names. The following is a typical example:

(159) *He* ntr *eꞌa* go\_out *ki* to *ruŋa,* above *he* ntr *tere* run *he* ntr *oho* go *mai* hither *ki* to *Maꞌuŋa* Ma'unga *Teatea,* Teatea *ki* to *Mahatua* Mahatua *… He* ntr *oho,* go *he* ntr *tuꞌu* arrive *ki* to *Vaipū…* Vaipu

'They got up and travelled to Ma'unga Teatea, to Mahatua … They went and arrived at Vaipu…' [Mtx-3-01.214–216]

Another feature of Rapa Nui geography is, that the Rapa Nui population is concentrated in one town. The rest of the island is largely uninhabited (though easily accessible) and can be designated as a whole by a couple of generic locationals: either *ꞌuta* 'inland' where agriculture takes place, or *kampō*, 'the countryside' where one goes for an outing. This will be discussed in more detail below.

The cardinal points (north, east, south, west) are not used for spatial orientation. Proto-Polynesian does have words for two of these: *\*toŋa* 'south, southern wind' and *\*tokelau* 'north, north wind'; these are reflected in many daughter languages, but in Rapa Nui they have a different sense: *toŋa* = 'winter', *tokerau* = 'wind (in general)'.

#### **3.6.3.1** *Tai* **'seashore';** *ꞌuta* **'inland'**

*Tai* indicates orientation with respect to the seashore:<sup>60</sup>

	- 'All the children gathered near the shore.' [R161.013]

(i) *Ko* prf *māria* calm *ꞌā* cont *te* art *tai.* sea 'The sea is calm.'

<sup>59</sup> A correlation between the geographical environment and grammaticalisation of spatial reference systems is crosslinguistically common; Palmer (2015) captures this generalisation as the Topographic Correspondence Hypothesis: "absolute coordinate systems are not merely anchored in, but are motivated by the environment" (210).

<sup>60</sup> There is a difference in meaning between the locational *tai*, which refers to the seashore, and the noun *tai*, which refers to the surface or condition of the sea:

3.6 Locationals


As (161) shows, the verb used for a movement in the direction of the sea is *turu* 'go down'. This verb is always used for seaward movement, even when no vertical movement is involved. Note, however, that in the hilly landscape of Rapa Nui a movement towards the sea will often involve some downward movement.

The locational *tai* is only used for movement and location on land. A movement at sea toward land is indicated with *ꞌuta* 'inland' (see the next section).

*ꞌUta* indicates orientation towards the inland, away from the coast. It may indicate a location on land (as opposed to the sea), or a place well inland (as opposed to the coastal region).

For example, *ki ꞌuta* either indicates a movement from sea to land as in (163), or a movement from a place on land to a place further inland as in (164). In the first case the verb *tomo* 'go ashore' is used, in the second case *iri* 'go up'.


*Tai* and *ꞌuta* are not only used for large-scale movement, but also for movement and localisation on a small scale. They may serve, for example, to localise people in a group, or objects on a table:


<sup>61</sup> There is some uncertainty about the meaning of the terms *hānau ꞌe ꞌepe* and *hānau momoko*. The traditional interpretation is 'long ears' and 'short ears', but Englert (1978) translates 'raza corpulenta' and 'raza delgada', respectively (see Mulloy 1993). More recently, Langdon (1994) has defended the traditional interpretation.

#### 3 Nouns and verbs

#### **3.6.3.2** *Tahatai* **'seashore'**

*Tahatai* indicates the seashore. Its meaning is similar to *tai* (§3.6.3.1), but seems to focus more narrowly on the line separating land and sea. Like the other locationals, it may be preceded by different prepositions:


Like *tai*, *tahatai* is only used for movement on land. Movement from the sea to the shore is indicated by *ꞌuta*.

#### **3.6.3.3** *Kampō* **'countryside'**

*Kampō*, from Spanish *campo*, indicates the area outside town.


'On days when in Ovahe or another place in the country the fish would come ashore, they would load it on a wagon…' [R539-1.482]

As *kampō* is principally used with reference to outings, and as outings typically take place near the shore, *kampō* usually refers to a place near the coast. In this respect it is different from *ꞌuta* 'inland', which often refers to areas inland where people grow their crops.

#### **3.6.3.4** *Kōnui* **'far'**

*Kōnui* 'far, distant' does not indicate an absolute point of reference, but any point far away from the reference point. The reference point may be the starting point of a movement as in (172), or the place where the action takes place as in (173).

3.6 Locationals


'From afar he saw a cow that was turning round and round.' [R250.137]

### **3.6.4 Temporal locationals**

There are a number of words referring to time which, like locationals, are preceded by prepositions. (They could be labelled 'temporals'.) These words all share the nonproductive prefix *ꞌaŋa-*, which indicates recent past.<sup>62</sup> They are listed in Table 3.6.

Table 3.6: Temporal locationals


Like other locationals, these words are preceded by prepositions, such as locative *ꞌi* (174) or genitive *o* (175):

(174) *Kai* neg.pfv *haꞌuru* sleep *mātou* 1pl.excl *ꞌi* at *ꞌaŋapō.* last\_night 'We did not sleep last night.' [R250.126]

<sup>62</sup> This prefix occurs with a similar meaning in many other Polynesian languages, but always as a reflex of PPN *\* ꞌana*; Rapa Nui is the only language in which *\*n* became *ŋ*.

Green (1985: 12) mentions *\*ina(a)fea* 'when (past)' as a PCE innovation; in fact, this reflects a more general shift from PEP *\* ꞌana-* to PCE *\*ina-*. This shift is not only reflected in \**inafea*, but also in Māori *inapoo*, Tahitian *inapō* 'last night'; Māori *inakuanei*, Tahitian *inā ꞌuanei*, Pa'umotu *inākuanei* 'just now'; Tahitian & Māori *inanahi* 'yesterday' (Pollex, Greenhill & Clark 2011; Bauer, Parker & Evans 1993; Académie Tahitienne 1999). The Rapa Nui forms *ꞌaŋa-*show that the shift *a* > *i* took place after Rapa Nui split from PEP. (Notice also that all reflexes of *\*ina* are from Tahitic languages, except Marquesan *inehea* 'when (past)'.)

<sup>63</sup> About the origin of these terms: *ꞌaŋapō* and *ꞌaŋahē* are transparent: *pō* 'night', *hē* 'content interrogative particle' (< PPN *\*fea*, see Footnote 7 on p. 487). For *-nīrā* and *-rīnā* Pollex does not give any cognates (only Samoan *\*analeilaa* 'earlier today' is a possible candidate). It is not clear which form is original in Rapa Nui, as both appear in older texts. For *ꞌaŋataiahi*, the only known cognate in Pollex is Māori (Eastern dialect) *tainahi* 'yesterday'. However, the second part *-ahi* is common as part of a word meaning 'yesterday': most Polynesian languages have a reflex of PPN \**nanafi* 'yesterday', sometimes preceded by *i-* or *a-*.

#### 3 Nouns and verbs

(175) *Te* art *nuꞌu* people *ruku* dive *o* of *ꞌaŋanīrā* today.past *ko* prom *ai* exist *ꞌana* cont *te* art *raperape,* swimming\_fin *te* art *hiꞌo…* glass 'Today's divers have swimming fins, goggles…' [R539-1.348]

There is also a set of three time words referring to the future; these are listed in Table 3.7.



These are not locationals but adverbs: they are not preceded by prepositions but form a clause adjunct on their own. The initial *a* in all three words reflects PPN *\*ꞌā-*, a prefix indicating near future (Pollex, Greenhill & Clark 2011), despite the variety in spelling in its Rapa Nui reflexes (*ꞌa*, *ā*, *a*).<sup>64</sup>

Some examples:


Interestingly, Rapa Nui has no generic temporal words 'now' and/or 'then'. To express these, the noun *hora* 'time' is used, in combination with a postnominal demonstrative (§4.6.3): *hora nei* indicates temporal proximity 'now', *hora era* expresses temporal distance 'then'.

<sup>64</sup> This prefix occurs in different words in several languages, e.g. Samoan *aa taeao* 'tomorrow'; Tongan *ꞌapogipogi* 'tomorrow'; Tahitian *ꞌā ꞌuanei* 'shortly, in a while', *afea* 'when (future)' (cf. *ina ꞌuanei* 'just now, a while ago', *inahea* 'when (past)'). Māori, like Rapa Nui, has a whole set of expressions sharing this morpheme: *aapoopoo* 'tomorrow', *aa hea* 'when (future)', *aaianei* 'now', *aakuanei* 'presently', *aa teeraa tau* 'next year' (see Biggs 1973: 79).

3.6 Locationals

#### **3.6.4.1** *Raꞌe* **'first'**

One more element needs to be mentioned here. *Raꞌe* 'first' is used in a variety of constructions: it can be an adjective modifying a noun (§4.3.3), a verb, or an adverb modifying a verb. It is also used as a locational, always preceded by the preposition *ꞌi*. *ꞌI raꞌe* functions as an adverbial phrase 'first, before anything else' (§11.6.2.4).

(179) *He* ntr *kai* eat *ia* then *ꞌi* at *raꞌe* first *e* num *tahi* one *ꞌapa* part *haraoa.* bread 'First I will eat a piece of bread.' [R476.031]

### **3.6.5 The locational phrase**

Like other nouns, locationals can be modified by certain noun phrase elements. The full range of possibilities is represented in Table 3.8.


Table 3.8: Structure of the locational phrase

Position 5 is only available for relative and absolute locationals, not for deictic locationals. This is not surprising, as postnominal demonstratives have (almost) the same form and function as the deictic locationals themselves. Position 7, which connects the locational to a second referent, is only available for relative locationals, not for absolute and deictic locationals.

Here are a few examples:


#### 3 Nouns and verbs

Compared to the common noun phrase (see the chart in §5.1), adjectives are absent from the locational phrase,<sup>65</sup> as well as anything related to quantification: determiners, quantifiers, numerals, plural markers and the collective marker *kuā*. The locational phrase is very similar to the proper noun phrase (§5.13.1), which also excludes quantifying elements; the main difference is, that the latter includes the proper article *a*.

### **3.7 Conclusions**

Like other Polynesian languages, Rapa Nui has no inflectional (and little derivational) morphology; moreover, many lexical items are freely used in both the noun phrase and the verb phrase. The existence of a distinction between nouns and verbs in the lexicon has been questioned for Polynesian languages. However, in this chapter I argue that there are good grounds to maintain this distinction. Approaches which conflate the two classes (or which define the bulk of the lexicon as "universals") do not do justice to the fact that the semantic relationship between the nominal and verbal uses of a lexeme is often unpredictable, and the fact that many words are either predominantly nominal or predominantly verbal in meaning and use. Rather, the occurrence of words with a typically verbal sense in the noun phrase can be regarded as cross-categorial use.

The boundary between nouns and verbs is not clear-cut; hence, the two can be defined in terms of a prototype, an intersection of certain syntactic, semantic and pragmatic features. In actual use, these features are not randomly distributed but tend to converge: a word referring to an entity tends to occur in a noun phrase, modified by noun phrase particles, and function as a referring expression.

The common cross-categorial use of nouns and verbs can be described in terms of two processes: lexical nominalisation (which turns a verb into a true noun, with a nominal sense) and syntactic nominalisation (where a verb is used in a construction which has certain nominal features). In both cases, the resulting nominal form may or may not have a suffix. While in lexical nominalisation the suffix is relatively uncommon, in syntactic nominalisation the use of the suffix depends on the construction; generally speaking, suffixed nominalisations are used when the event is presented as an object, a bounded entity, rather than as an event happening over time.

Syntactic nominalisation is in fact very common in Rapa Nui. In several constructions, a main clause predicate is constructed nominally; in addition, nominalised verbs are used in various subordinate constructions, such as causal clauses and certain complement clauses. The variety and frequency of nominal constructions are evidence of a "nominal drift", a tendency to maximise the use of nominal constructions.

Nouns can be subdivided into common nouns (which are preceded by determiners), proper nouns (which take the proper article *a*) and locationals (which take neither).

Verbs can be subdivided into several classes, based on criteria such as the number of arguments, the use of the agent marker *e* and the possibility to enter into the actor-

<sup>65</sup> There is one exception: *ruŋa* 'above' and *raro* 'below' may be followed by *nui* 'big', in both cases with idiomatic sense: *ruŋa nui* 'high', *raro nui* 'deep'. Notice that the same element *nui* has also been added to the original PEP locative *\*kō* 'there', resulting in *kōnui* 'far' (§3.6.3).

emphatic construction. Adjectives are a subclass of verbs; they are characterised by frequency of adnominal use, as well by the presence of certain modifiers and the absence of modifiers occurring with other verbs.

## **4 Closed word classes**

### **4.1 Introduction**

As indicated in §3.1, there is a basic distinction in Rapa Nui between full words (notably nouns and verbs) and particles. The previous chapter dealt with word classes that are clearly full words: nouns and verbs and subtypes thereof. This section discusses word classes which have at least some characteristics of particles: they form closed classes and do not have a lexical meaning. All of these, except personal pronouns, occur in the periphery of the noun and/or verb phrase.

However, most of these words also share characteristics of full words. Numerals and (occasionally) demonstratives may also be a clause constituent. Pronouns and numerals, and to a lesser extent quantifiers and adverbs as well, may form phrases containing preand or postnuclear particles.

Table 4.1 lists these word classes in roughly descending order of full word status:<sup>1</sup>


Table 4.1: Closed word classes

<sup>1</sup> Other word classes are not discussed in this chapter, but in sections corresponding to their respective functions. This concerns negators (§10.5), the polar question marker (§10.3.1), coordinating conjunctions (§11.2), preverbal subordinators (§11.5) and subordinating conjunctions (§11.6). Yet other words are particles occurring in fixed positions in the noun phrase and the verb phrase; these are discussed in Chapters 5 and 7: determiners (§5.3), the proper article (§5.9), the collective marker (§5.2), plural markers (§5.5), the identity marker (§5.13.2), aspect markers (§7.2), other preverbal particles (§7.3), evaluative markers (§7.4), directionals (§7.5) and the continuity marker (§7.2.5.5).

#### 4 Closed word classes

### **4.2 Pronouns**

Rapa Nui has a set of personal pronouns, two sets of possessive pronouns and a set of benefactive pronouns. §4.2.1 discusses personal pronouns; §4.2.2 discusses possessive pronouns; §4.2.3 lists benefactive pronouns. Finally, §4.2.4 discusses a few marked uses of pronouns.

NB Demonstrative particles are used as pronouns in limited contexts; this is discussed in §4.6.6.

### **4.2.1 Personal pronouns**

#### **4.2.1.1 Forms**

Personal pronouns are inflected for number (singular, dual, plural), person, and inclusiveness. The forms are given in the Table 4.2. 2


Table 4.2: Personal pronouns

The inclusive forms indicate that the addressee is included in the group referred to by the pronouns: *tāua* 'you and me', *tātou* 'we all, including you'. The exclusive forms indicate that the addressee is not part of the group referred to: *māua* 'the two of us (but not you)', *mātou* 'we (excluding you)'.

Most Polynesian languages have distinct dual and plural pronouns in all persons. As Table 4.2 shows, in Rapa Nui dual and plural are only distinguished in the first person. In the second and third person, the original dual forms *kōrua* and *rāua* extended their use to plural, while the PEP plural forms *\*kōtou* and *\*rātou* were lost.<sup>3</sup>

Personal pronouns tend to be used for animate referents only: humans and animals. Note however that possessive pronouns can be used for inanimates as well. Here is an illustration from a description of a palm tree:

<sup>2</sup> Apart from the loss of plural forms, the personal pronouns were inherited from PEP without any changes (see the reconstructed forms by Wilson 1985: 98); the singular forms are even unchanged from the PPN forms as reconstructed by Kikusawa (2003: 168). Ultimately, the dual and plural forms go back to Proto-Oceanic, where the dual forms had a suffix *\*-dua* 'two' and the plural forms a suffix *\*-tolu* 'three' (Pawley 1972: 37).

<sup>3</sup> According to Bergmann (1963: 55), in some other Polynesian languages, dual pronouns have extended uses. Thus in Tongan, the first person inclusive dual is often used with a plural sense (Churchward 1985 [1953]: 124–125). A similar process in Rapa Nui may have led to the extension in use of dual pronouns to include plurality, eventually superseding the original plural forms.

4.2 Pronouns

(1) *Tumu* tree *nei* prox *e* ipfv *ai* exist *rō* emph *ꞌā* cont *tōꞌona* poss.3sg.o *maŋa* branch *roaroa.* long:red 'This tree has long branches (lit. there are its long branches).' [R483.005]

In the same text, personal pronouns are avoided to refer to the tree; full noun phrases are used instead:

(2) *Tāꞌaku* poss.1sg.a *aŋa* do *he* pred *manava* stomach *mate* die *ki* to *te* art *tumu* tree *nei.* prox *E* ipfv *ai* exist *rō* emph *ꞌana* cont *te* art *maŋa* branch *pakapaka* dry:red *o* of *te* art *niu* palm\_tree *nei.* prox 'I always like this tree. This palm tree has dry branches.' [R483.009–010]

Personal pronouns belong to the class of proper nouns. This means that in many syntactic contexts, they are preceded by the proper article *a* (§5.13.2.1).

### **4.2.2 Possessive pronouns**

Rapa Nui has two sets of possessive pronouns.<sup>4</sup> One set is based on the article *te* and starts with *t-*; I will call this series *t-possessives*. The other set, which does not start with *t-*, will be called *zero possessives* (*Ø-possessives*). In addition, the singular pronouns in each set exhibit a distinction between *o*- and *a*-forms. This results in four forms, for example in the first person singular:

(3) *tōꞌoku tāꞌaku ōꞌoku ꞌāꞌaku* 'my, mine'

The *t*-possessives are discussed in §4.2.2.1, the Ø-possessives in §4.2.2.2.

In this chapter, only the forms of possessive pronouns are given. Their use is discussed – together with possessive phrases in general – in Chapter 6 on possession.

#### **4.2.2.1** *t***-possessives**

**4.2.2.1.1 Singular possessors** In the singular, there are two classes of possessive pronouns, characterised by the use of *o* and *a,* respectively. These classes indicate different types of relationships between possessor and possessee; the issue of *o-* and *a*-possession is discussed in §6.3.3.

The singular *t*-possessives are given in Table 4.3.

<sup>4</sup> Following Dryer (2007b: 182), I use the term *possessive pronoun* for any pronominal possessive form, whether used as a constituent on its own (English 'mine', 'yours') or as a modifier within the noun phrase (English 'my', 'your'). The latter are often called possessive adjectives, reserving the term possessive pronoun for independent forms which can function as nominal complement or predicate. In Rapa Nui, the difference between the two sets of possessive forms does not correspond to the difference between socalled possessive adjectives and possessive pronouns. Both can, for example, occur as modifier before the noun (§6.2.1). Moreover, the term *possessive adjective* would not be entirely satisfactory for Rapa Nui, as possessors do not occur in the same position in the noun phrase as adjectives (see the position chart in §5.1).

#### 4 Closed word classes

Table 4.3: Singular t-possessive pronouns


*Tuꞌu* and *toꞌu* are shortened forms of *tōꞌou*; *taꞌa* and *taꞌu* are shortened forms of *tāꞌau*. 5 In older texts the short forms are rare; the long forms are used exclusively in all contexts:


Nowadays, when *t*-pronouns are used in the noun phrase, only the shorter forms are used (*tuꞌu matuꞌa* 'your parent', *taꞌa poki* 'your child', §6.2.1). The long forms are only used nowadays in the partitive construction *Poss o te N* (§6.2.2).

**4.2.2.1.2 Plural possessors** In the plural, *a* and *o* forms are not distinguished. Even so, there are two series of *t-*possessive pronouns: one with *to*, one with *te*. Their forms are given in Table 4.4.



There is no difference in meaning between the two series. The *to*-series is older; it is still used occasionally nowadays, but has an archaic ring to it. The *te*-series is found

<sup>5</sup> According to Mulloy & Rapu (1977: 13), *ta ꞌa* and *tuꞌu* "demonstrate a relatively recent sound change" from the older forms *tā ꞌau* and *tō ꞌou*. Note however that, while the shortened forms are indeed relatively rare in older texts, they do occur in MsE and Ley (though not in Mtx).

4.2 Pronouns

occasionally in older texts (17x), but *to* is predominant in these texts (176x).<sup>6</sup> In newer texts, *te* is predominant: there are 127 *to*-forms against 1314 *te*-forms.

#### **4.2.2.2 Ø-possessives**

The singular Ø-possessives are listed in Table 4.5. They have the same form as the *t*possessives, minus the initial *t-*. The *a*-forms are spelled with an initial glottal, just like the possessive preposition *ꞌa* (§2.2.5).

Table 4.5: Singular Ø-possessive pronouns


As with the *t*-possessives, there are shortened forms in the 2nd person singular: *uꞌu* and *oꞌu* are shortened forms of *ōꞌou*, *aꞌa* and *aꞌu* are shortened forms of *ꞌāꞌau*. There is no difference in meaning between the longer and the shorter forms.

The plural forms are given in Table 4.6, t. In the plural, the Ø-possessives are identical to the personal pronouns preceded by the genitive preposition *o*, as in a genitive noun phrase. As with the *t*-possessives, the plural pronouns do not make a distinction between *a* and *o*-possession.

Table 4.6: Plural Ø-possessive pronouns


<sup>6</sup> 12 of the 17 *te*-forms in old texts are *te kōrua* in Mtx; *to kōrua* is only used once in Mtx. This may suggest that the change *to* > *te* started off as dissimilation before *o* (*kōrua* is the only plural pronoun with *o* as first vowel); subsequently this was generalised to all pronouns. In any case, the data show that the *te*-possessives are a recent innovation, not a retention from PEP as suggested by Wilson (1985: 105–106); Wilson (2012: 298).

#### 4 Closed word classes

### **4.2.3 Benefactive pronouns**

Benefactive pronouns express that something is destined/intended for the person in question. They are identical to the *t*-possessive pronouns (for the second person singular, the long form is used), but with an initial *m-* instead of *t-*. As with possessive pronouns, there is an *o*/*a* distinction in the singular, but not in dual and plural. The forms are listed in Table 4.7.

*o*-class *a*-class 1 singular *mōꞌoku māꞌaku* 2 singular *mōꞌou māꞌau* 3 singular *mōꞌona māꞌana* 1 dual inclusive *mo tāua –* 1 dual exclusive *mo māua –* 1 plural inclusive *mo tātou –* 1 plural exclusive *mo mātou –* 2 dual/plural *mo kōrua –* 3 dual/plural *mo rāua –*

Table 4.7: Benefactive pronouns

Benefactive pronouns are the pronominal counterpart of the preposition *mo/mā* + NP, and have the same uses. The use of these prepositions is discussed in §4.7.8.

### **4.2.4 Uses of pronouns**

Personal pronouns are used in the same contexts as nouns: as subjects, objects, after prepositions et cetera. The uses of possessive pronouns will be discussed in more detail in Chapter 6. In this section, a few nonstandard uses of pronouns are discussed.

#### **4.2.4.1 Generic pronouns: 'one'**

As in many languages, the second person singular pronoun *koe* can be used in a generic way, referring to persons in general.

(4) *E* ipfv *ko* neg.ipfv *takera* see *e* ag *koe* 2sg *e* ipfv *noho* sit *ꞌana,* cont *e* ipfv *riri* angry *ꞌana* cont *ꞌo* or *e* ipfv *tātake* argue *ꞌana…* cont '(describing someone's character:) You would never see him angry or arguing…' [R302.050]

4.2 Pronouns

(5) *E* ipfv *riꞌariꞌa* afraid *nō* just *koe* 2sg *ꞌi* at *te* art *kai* food *ka* cntg *hiko* snatch *era.* dist

'People (lit. you) were afraid because she would snatch the food away.' [R368.104]

*Koe* as a generic pronoun can have a distributive sense: 'each one, every one'. In the following example this is reinforced by the repeated *te kope era* 'that person':

(6) *He* ntr *oho* go *te* art *taŋata,* man *he* ntr *toꞌo* take *mai* hither *koe* 2sg *i* acc *tāꞌau* poss.2sg.a *viꞌe,* woman *te* art *kope* person *era,* dist *te* art *kope* person *era* dist *hoki* also *ananake.* together 'The men came, every one took a woman for himself, each and every young man.' [Mtx-3-01.285]

This example also shows that possessive pronouns (here *tāꞌau* 'your') may have a generic sense as well.

#### **4.2.4.2 Second person pronouns of personal involvement**

There is yet another use of the second person singular personal and possessive pronouns, which could be labeled "personal involvement". Even though no participant in the discourse is an addressee, someone – either a participant in the story or the hearer – is addressed directly, to communicate a degree of personal or emotional involvement from the part of the speaker.

Firstly: sometimes a participant in a narrative text is referred to as *koe*, followed by a vocative phrase (§8.11 on the vocative). The narrative is in the third person, i.e. no addressee is involved as a participant; yet the speaker is, as it were, addressing the participant:<sup>7</sup>


'The old man arrived on his skinny horse.' [R363.017]

As (8) shows, even when the participant is "addressed" in this way, for all other purposes it is still a third-person participant (*tōꞌona hoi*, 'his horse').

<sup>7</sup> Fedorova (1965: 400) gives examples of this same construction in Mss. A and C (see Footnote 27 on p. 24): *koe e … ē*, calling it "the article circumfix".

#### 4 Closed word classes

Sometimes the pronoun could be paraphrased as 'that dear one', but in many cases its exact connotation is hard to convey in translation.

Secondly: the second person singular possessive pronouns*taꞌa* and *taꞌu* (§4.2.2.1.1) can be used without a real possessive meaning.<sup>8</sup> This happens both in conversation and in third-person contexts. In conversation, they are used to imply that the noun is in some loose way connected to the hearer: 'your thing', i.e. the thing you were referring to, or the thing you asked about, or the thing that is of interest to you.

In (9), two people are discussing a photograph. One of them points out a woman they both know:

(9) *—ꞌAi* there *taꞌu* poss.2sg.a *viꞌe* woman *ko* prom *Eva.* Eva *—ꞌAi* there *te* art *rūꞌau* old\_woman *era* dist *ko* prom *Eva.* Eva '—Here is the (lit. 'your') woman Eva. —(Indeed), here is the old woman Eva.' [R416.461–462]

The same use of possessive pronouns is also found in narrative contexts where no second-person participant is involved. By using a second person pronoun the speaker is, as it were, addressing the listener, implying that the object or person under discussion is in some way relevant to him/her. One could say that the listener is made part of the story, a strategy which makes the story more vivid. One function of the pronoun in this construction is stressing familiarity: the person or object is already known to the listener, whether from the preceding text or from general knowledge. *Taꞌa*/*taꞌu* could thus be paraphrased as 'the one you know'.


In this loose sense, the possessive pronouns *taꞌu* and *taꞌa* have lost their possessive force; rather, they have become a sort of demonstrative, similar to demonstrative determiners like *tū* and *tau*. However, the latter require a postnominal demonstrative *nei*, *ena* or *era*, while *taꞌa* and *taꞌu* don't.

<sup>8</sup> This use is also noted by Englert (1978: 21), who distinguishes "*taa* y *taau* como artículos" (*taa* and *taau* as articles) from "el pronombre posesivo *taau*" (the possessive pronoun *taau*), and Bergmann (1963: 48).

4.3 Numerals

### **4.3 Numerals**

Rapa Nui has a decimal counting system, as is usual in Eastern Polynesia (see Lemaître 1985). As is equally usual, it has terms for several powers of ten.

Cardinal numerals are usually preceded by one of the particles *e* (the default marker), *ka* (the contiguity marker) and *hoko* (when referring to a group of persons); these will be discussed in §4.3.2. Using these particles as a criterion, the interrogative *hia* 'how many' also classifies as a numeral (§10.3.2.4).

On the other hand, the archaic form *ꞌaŋahuru* 'ten' does not qualify as a numeral in older texts, and neither do certain other forms which are obsolete nowadays (§4.3.1.3).

In this section, first the forms of the numerals are discussed (§4.3.1). §4.3.2 discusses elements preceding and following the numerals in the numeral phrase, especially the numeral particles *e*, *ka* and *hoko*. §4.3.3 discusses ordinal numerals; §4.3.4 discusses definite numerals, special forms with collective reference. Finally, §4.3.5 discusses the expression of fractions.

In the noun phrase, numerals occur either before or after the noun (§5.1); the use of numerals in the noun phrase will be discussed in §5.4. Apart from that, numerals also occur as predicates of numerical clauses; these are discussed in §9.5.

### **4.3.1 Forms of the numerals**

#### **4.3.1.1 Basic and alternative forms**

**4.3.1.1.1 One to ten** The cardinal numerals from one to ten in modern Rapa Nui are given in Table 4.8.


Table 4.8: Numerals 1–10

As this table shows, for certain numerals there are two forms: a basic form and an alternative form. The alternative numerals are used in compound numerals, i.e. as part

#### 4 Closed word classes

of numerals higher than ten. They are also used in a number of other cases, described in §4.3.1.2.

For 'ten', *hoꞌe ꞌahuru* is the most common form nowadays. (*ꞌAhuru* is never used on its own, but always preceded by *hoꞌe* 'one' or a higher numeral.) *ꞌAŋahuru* is an older form which is still in use, but rare. It is especially used as a noun 'a group of ten', and as ordinal number 'tenth' (§4.3.3).


Table 4.9: Numerals 11–100

**4.3.1.1.2 11 to 100** Numerals above 10 are illustrated in Table 4.9. As this table shows, the alternative numerals are used both for the tens (*piti ꞌahuru*, not *\*rua ꞌahuru*) and the units (*mā piti*, not \**mā rua*). Tens and units are connected by the particle *mā* 'and, with'.<sup>9</sup>

Like *ꞌahuru*, *hānere* is always preceded by another numeral, whether *hoꞌe* 'one' or a higher numeral:

(12) *E* num *hoꞌe* one *hānere* hundred *māmoe* sheep *hāpaꞌo* care\_for *ꞌāꞌana…* poss.3sg.a 'He had one hundred sheep he cared for…' [R490.002]

To indicate an unspecified number above ten, *tūmaꞌa* is used: 'something, and a bit'.

<sup>9</sup> *Mā* is common in Polynesian languages in the sense 'and, with' (< PPN \**mā*); in various languages it serves to connect to tens to units in numerals, like in Rapa Nui. In Rapa Nui, it is also used in circumstantial clauses (§11.6.8), a function shared with Tahitian (Académie Tahitienne 1986: 107, 196); possibly Rapa Nui borrowed *mā* from Tahitian.

4.3 Numerals


**4.3.1.1.3 Above 100** Table 4.10 shows numerals above 100. Just as in the numerals between 10 and 100, units as part of higher numerals are preceded by *mā*. Between hundreds and tens, and between thousands and hundreds, the particle *e* can be used, but this is not obligatory.

Table 4.10: Numerals > 100


In spoken language, high numbers are often expressed with Spanish numerals. These are not preceded by a numeral particle:


Common as this may be, speakers do not consider this to be proper Rapa Nui; Spanish numerals are not accepted in formal spoken and written language.

#### 4 Closed word classes

**4.3.1.1.4 Etymology of the numerals** The basic forms of the numerals are regular reflexes of the PPN forms, while the alternative numerals listed above (*hoꞌe, piti, maha, pae, ꞌahuru*) are borrowed from Tahitian.<sup>10</sup> The basic numerals are the original Rapa Nui forms, except *vaꞌu* (the original form is *varu*) and *hoꞌe ꞌahuru*, which are also Tahitian. The forms *toru, ono, hitu* and *iva* are common to both languages.<sup>11</sup>

*Hānere* is also a Tahitian borrowing, derived from English 'hundred'. The origin of *taꞌutini* is a little more complicated. It was probably borrowed from Tahitian *tauatini*, whereby the second *a* disappeared and a glottal was introduced between the first two vowels. Tahitian *tauatini* itself is a development from the older form *tautani*, from Eng. 'thousand'.<sup>12</sup>

We may conclude that in modern Rapa Nui, all numbers higher than seven are expressed by Tahitian numerals. The remarkable extent of lexical replacement is evidence for the widespread influence of Tahitian on the language (§1.4.1).<sup>13</sup>

#### **4.3.1.2 Other uses of the alternative numerals**

As described above, in numbers above ten, the alternative (Tahitian) numerals are used. They are also used in dates and when telling the time, and sometimes in measures. These constructions are discussed here.

**4.3.1.2.1 Days and dates** Most of the names of days of the week contain a (Tahitian) numeral:

<sup>10</sup> The Tahitian forms for 1, 2, 4 and 5 are language-internal developments, some of which may have occurred as late as the early 19th century (White 1968: 64).

<sup>11</sup> A Spanish expedition in 1770 recorded a set of numerals totally different from the usual ones and unlike any numerals known from other Polynesian languages: *coyana, corena, cogojui, quiroqui, majana, teuto, tehea, moroqui, vijoviri, queromata* (with some variation between different manuscripts; see Ross 1937). These have sometimes been considered as evidence of a non-Polynesian substrate (Ross 1936; Schuhmacher 1976; 1990; Mangor & Schuhmacher 1998). More likely, however, they represent Rapa Nui words which the Spanish transcribers mistook for numerals (Fischer 1992; Fedorova 1993; Mellén Blanco 1994).

Four years later, in 1774, Johann Reinhold Forster recorded a set of regular Polynesian numerals (Fischer 1992: 184).

<sup>12</sup> The development from *tautani* to *tauatini* probably happened under the influence of Tah. *tini* 'numerous': the second half of *tautani* was assimilated to *tini*, which had a closely related meaning.

<sup>13</sup> According to Lynch & Spriggs (1995: 37), in almost all Oceanic languages, "the basic monomorphemic numerals are well known and very frequently used". Two notable exceptions are Chamorro (Guam, see Topping 1973: 166), in which the whole numeral system was replaced by Spanish, and Anejom (Vanuatu, see Lynch & Spriggs 1995), where all numerals above three were replaced by Bislama/English forms. Clark (2004), however, suggests that numerals are quite susceptible to replacement by terms from a European language, because numerals are often used in domains of interaction with Europeans: in European culture, numbers play a much larger role than in traditional culture. For higher numerals, another reason for substitution is the length of terms: higher vernacular numerals tend to be much longer than English equivalents. This is true for Rapa Nui as well.

We also have to keep in mind that substitution by Tahitian terms is different from borrowing from Spanish. As Fischer (2007: 397); Fischer (2008a: 151) points out, Tahitian forms are considered as indigenous; they do not stand out as Spanish or English borrowings would (§1.4.1). Large-scale replacement of numerals also happened in other languages under Tahitian influence. In Mangarevan for example, all numbers higher than five are nowadays expressed with Tahitian numerals (P. Auguste Uebe-Carlson, p.c.).

4.3 Numerals

(17) *mahana* day *piti;* two *mahana* day *pae* five 'Tuesday; Friday'

For numbering the days of the month, the Tahitian numerals are used as well:

(18) *ꞌi* at *te* art *hoꞌe* one *mahana* day *o* of *Mē* May 'on the first of May' [R231.045]

**4.3.1.2.2 Telling time** 'X o'clock' is expressed as *hora X*, where X is a Tahitian numeral:

(19) *Hora* hour *maha* four *nei,* prox *ꞌe* and *hora* hour *hitu* seven *tātou* 1pl.incl *ka* cntg *tuꞌu* arrive *iho.* just\_then 'It is now four o'clock, and seven o'clock we will arrive.' [R210.198]

**4.3.1.2.3 Measuring space and time** With spatial measuring words like *mētera* 'meter', *māroa* 'fathom' and *ꞌumi* 'ten fathoms', both Rapa Nui and Tahitian numerals are used: (20) has the Tahitian term *pae*, while (21) has the Rapa Nui term *rua*.

(20) *Te* art *tumu* tree *nei* prox *tumu* tree *nikoniko* curl:red *e* ipfv *ko* neg.ipfv *oŋa* appear *te* art *pae* five *mētera* meter *o* of *te* art *roaroa.* long

'This tree is a twisted tree which doesn't surpass five meters of length.' [R478.055]

(21) *E* num *rua* two *mētera* meter *mā* plus *pae* five *o* of *te* art *roaroa.* long:red 'He was 2.05 meters tall.' [R250.177]

With time words we also find an alternation between Tahitian and Rapa Nui numerals:


4 Closed word classes

#### **4.3.1.3 Old numerals**

In older texts, only the original Rapa Nui numerals are used. The numbers one through seven and nine are identical to the basic forms still in use today, listed in §4.3.1.1. For 'eight', the old form is *varu*. 14

For 'ten', the original numeral is *ꞌaŋahuru*, which is still marginally in use today (§4.3.1.1 above). In older texts, it is usually preceded by the article *te* rather than the numeral particles *e* or *ka*. Between *ꞌaŋahuru* and the noun there is a second article:

(24) *He* ntr *here* tie *e* num *tahi* one *te* art *ꞌaŋahuru* ten *te* art *taka.* roll 'They tie ten rolls (of *mahute* fibers) together.' [Ley-5-05.002]

Thus, *ꞌaŋahuru* is more a noun than a numeral;<sup>15</sup> the counted item follows as a second noun phrase. On the other hand, it is not quite a regular noun, as the construction *te N te N* is never used with other nouns.

When *ꞌaŋahuru* is used nowadays, it behaves like any other numeral. For example, in the following example it is not preceded by an article:

(25) *ꞌI* at *ira* ana *i* pfv *noho* stay *ai* pvp *e* num *tahi* one *ꞌaŋahuru* ten *o* of *te* art *mahana.* day 'There they stayed ten days.' [R420.047]

Three other old – and obsolete – numerals are *kauatu* 'ten', *rau* 'hundred' and *pīere* 'thousand'. Like *ꞌaŋahuru*, they are preceded by the article rather than by a numeral marker.

### **4.3.2 The numeral phrase**

Cardinal numerals are always preceded by one of the particles *e*, *ka* and – less commonly – *hoko*. <sup>16</sup> These are discussed in the following subsections. §4.3.2.4 shows that numerals may be followed by a number of modifying particles.

<sup>14</sup> It is interesting to note, that the Tahitian form *va ꞌu* appears already in MsE, the oldest text in the corpus, where it is used alongside *varu*. In Ley and Mtx, however, *varu* is consistently used. Englert's grammar (Englert 1978: 58), which otherwise does not mention Tahitian numerals, states that, while *varu* is the older form, nowadays only *va ꞌu* is used.

*Va ꞌu* may have been replaced earlier than the other numerals because it is a relatively high number, and/or because the Tahitian form is close to the Rapa Nui form. Moreover, the alternation between *r* and glottal is a process which occurs within Rapa Nui as well (§2.5.2).

<sup>15</sup> It is not unusual for higher numerals to have the status of nouns; see Dixon (2012: 78).

<sup>16</sup> A prefix found in other Polynesian languages but not in Rapa Nui is the distributive prefix *\*taki-* (e.g. *takitahi* 'one each'), used e.g. in Tahitian (Lazard & Peltzer 2000: 182), Pa'umotu (Stimson 1964: 492), Māori (Bauer, Parker & Evans 1993: 498), Samoan (Mosel & Hovdhaugen 1992: 116). In Rapa Nui, the distributive is expressed by repetition of the noun phrase.

4.3 Numerals

#### **4.3.2.1 Neutral** *e*

*e* is the neutral numeral particle. In most contexts, cardinal numerals are preceded by *e*. Numerals preceded by *e* occur before or after the noun in the noun phrase (§5.4.1– 5.4.2); they also occur as the predicate of a numerical clause (§9.5). Numerous examples of *e* + numeral are provided in the referred sections, as well as in §4.3.1 above.

#### **4.3.2.2 The contiguity marker** *ka*

*Ka* is an aspectual marker indicating contiguity between two events (§7.2.6). With numerals, *ka* is used in counting, or when listing or summing up a series of things:


When used within a noun phrase, like *ka tahi mahana* in (27), *ka* + numeral always occurs before the noun, never after the noun (different from *e*).

*Ka* as a numeral marker is used when a number or quantity has been reached; it indicates an extent. It is especially common with time words, indicating that a certain time has elapsed. In the following example, *ka* + numeral indicates that the age of ten years has been reached:

(28) *E* num *tahi* one *poki* child *te* art *ꞌīŋoa* name *ko* prom *Eva* Eva *ka* cntg *hoꞌe* one *ꞌahuru* ten *matahiti.* year 'There was a girl whose name was Eva, ten years old.' [R210.001]

In this sense 'elapsed time', *ka* is used to indicate minutes after the hour (§4.3.1.2.2). As *ka* indicates a quantity which has been reached, it may emphasise the amount: 'up to, as many as'. In (29) this emphasis is further enhanced by the use of *rō atu*:

(29) *Mo* if *ai* exist *rō* emph *kona* place *hore* cut *iho* just\_then *hai* ins *ꞌārote* plough *e* ipfv *puꞌa* cover *era* dist *e* num *ono* six *ꞌo* or *ka* cntg *vaꞌu* eight *rō* emph *atu* away *ꞌuei.* ox

'When a field was ploughed for the first time, it was ploughed with six or even eight oxen.' [R539-1.110–111]

(30) *Ka* cntg *ono,* six *ka* cntg *ono* six *taŋata* man *i* pfv *mate* die *ꞌi* at *tau* dem *ꞌura* lobster *era* dist *ko* prom *tetu.* huge 'As many as six men died by that enormous lobster.' [Mtx-4-05.014]

4 Closed word classes

#### **4.3.2.3 The person marker** *hoko*

The particle *hoko* is used when counting persons: *hoko rua* 'two people', *hoko toru* 'three people' etc.<sup>17</sup> It is only used with numerals under ten.

Numerals preceded by *hoko* may occur after the noun as in (31), before the noun as in (32) (though this is relatively rare), or on their own as in (33):


'When nine men had come out of the airplane…' [R539-2.215]

(33) *He* ntr *haꞌuru* sleep *hoko* num.pers *hā,* four *hoko* num.pers *toru* three *ka* cntg *ꞌara* wake\_up *ka* cntg *vānaŋanaŋa* talk:red *nō.* just 'Four (men) slept, three were awake and were talking.' [MsE-050.005]

*Hoko rua* and *hoko tahi* have both developed certain lexicalised uses in which the sense is somewhat different from 'a group of X persons'; in these cases, they are written as one word. *Hokorua* is used as a noun 'companion' and as a verb 'to accompany'; *hokotahi* is used as an adjective 'lonely, solitary', or an adverb 'alone, on one's own':


'Ure a Reka saw that Marama was lonely, he had no companion.' [Ley-7-48.013]

<sup>17</sup> A prefix *soko* or *hoko* preceding numerals (PPN *\*soko*) is found in a smattering of languages throughout Polynesia (Pollex, see Greenhill & Clark 2011), though never exclusively referring to persons; it either means 'just, exactly' or 'one, alone, a single'; the latter sense occurs in Rapa Nui in *hokotahi* 'solitary'. A numeral prefix restricted to human reference is PPN *\*toko*, which occurs in the majority of Polynesian languages. Possibly both *\*toko* and *\*hoko* existed originally in Rapa Nui; the two were conflated because of their formal and semantic similarity, resulting in the form *hoko* with semantic features of both \**hoko* and \**toko*.

4.3 Numerals

#### **4.3.2.4 Modifiers in the numeral phrase**

Cardinal numerals may be followed by modifying elements like *mau* 'really', *nō* 'just' or *hakaꞌou* 'again, more, other', elements which also occur in the noun phrase (§5.8).


Numerals preceded by *ka* may also be followed by the verb phrase particles *rō* (§7.4.2) and *ꞌō* (§4.5.4.5). *Rō* (which may in turn be followed by *atu*) emphasises the extent or limit of the number: 'up to, as much as, even':


'When a field was ploughed for the first time, it was covered with six or even eight oxen.' [R539-1.110]

### **4.3.3 Ordinal numerals**

Rapa Nui does not have separate forms for ordinal numerals, except *raꞌe* 'first', which is an adjective, occurring after the noun. Other numerals are interpreted as ordinal numerals by virtue of their position: they are ordinal numerals when they occur before the noun and are preceded by a determiner.


The determiner can be the article *te* as in the examples above, but also a possessive pronoun as in (41), or the predicate marker *he* as in (42):

(41) *Ku* prf *aŋa* make *ꞌana* cont *i* acc *tōꞌona* poss.3sg.o *rua* two *vaka* boat *era.* dist 'He built his second boat.' [R539-1.168]

4 Closed word classes

(42) *Te* art *hare* house *pure* pray *he* pred *rua* two *hare* house *pure* pray *era.* dist 'The church (shown in this picture) is the second church.' [R412.203]

*Rua* as an ordinal numeral is also used in the sense 'the other' (out of two):

(43) *E* num *rua* two *ŋāŋata* men *hiva,* continent *e* num *tahi* one *taŋata* man *he* pred *italiano,* Italian *te* art *rua* two *taŋata* man *he* pred *harani.* French

'There were two foreigners, one man was an Italian, the other man was a Frenchman.' [Egt-02.185]

With a time word, *rua* means 'next'.

(44) *ꞌI* at *te* art *pōꞌā* morning *o* of *te* art *rua* two *mahana* day *he* ntr *ꞌara* wake\_up *a* prop *Piu.* Piu 'In the early morning of the next day, Piu woke up.' [R437.088]

### **4.3.4 Definite numerals**

To express 'the two, the three' et cetera, a special form of the numerals is used, in which the first mora is reduplicated (type 1 reduplication, see §2.6.1.1). These forms, which can be labeled definite numerals, <sup>18</sup> are always preceded by the proper article *a* (§5.13.2; possibly *a* is used because the definite numerals are functionally similar to pronouns). They are listed in Table 4.11.

Like all reduplications, these forms are written with a hyphen in standard orthography (*a to-toru* etc.). As the table shows, the original *a rurua* has evolved into *ararua*. As this is a frozen form which is not recogniseable as a reduplication, it does not have a hyphen in the standard orthography: *ararua* versus *a ru-rua*.

The definite numerals often occur on their own as in (45) below, but they are also used in the noun phrase. In the latter case they may placed either before the noun as in (46), or after the noun as in (47).


'The six young men came and arrived at the house of Kave Heke.' [Ley-4-01.007]

<sup>18</sup> Other possible terms are "proper numerals" (because of the use of *a*), or "collective numerals" (because they denote a collectivity).

<sup>19</sup> *A rurua* is used only in Ley and MsE; Mtx has *ararua*.

4.3 Numerals


Table 4.11: Definite numerals

#### (47) *He* ntr *oŋe* shortage *te* art *ꞌaro* side *a* prop *hahā* red:four *o* of *nei.* prox

'The four sides of the island (lit. of here) here suffered shortage.' [Mtx-5-02.017]

Like the cardinal numerals, definite numerals are never preceded by prepositions. They are usually found in subject position, where no preposition is needed. However, they are also used occasionally in positions that would normally require a preposition. In the following example, *a vavaru* occurs in a locative phrase, where the preposition *ꞌi* 'in' is expected; the preposition is left out.

	- '…they killed fifty (people who were hiding) in the eight holes.' [Mtx-3-01.237]

Notice that this restriction distinguishes the definite numerals from all other items preceded by the proper article: pronouns and common nouns marked with the proper article can be preceded by prepositions without a problem (§5.13.2.1).

Nowadays the definite numerals other than *ararua* are used less frequently than in the past.<sup>20</sup> Their role is partly taken over by *ananake* (§4.4.4), which used to be the universal quantifier 'all', but which nowadays has the sense 'together'. Both *ananake* and the definite numerals are mainly used pronominally nowadays, i.e. without a head noun or pronoun.

One could say that *ararua* and *ananake* form a mini-paradigm in modern Rapa Nui, with *ararua* referring to a group of two entities and *ananake* to more than two.<sup>21</sup>

<sup>20</sup> In the corpus of old texts (122,600 words), there are 73 occurrences, roughly once in 1,700 words; in the much larger corpus of newer texts (367,500 words) there are only 39 occurrences, roughly once in 9,400 words. *Ararua*, on the other hand, is common both in older and newer texts: in the former it occurs 136 times (once in 900 words), in the latter 865 times (once in 425 words).

<sup>21</sup> This is even clearer in comitative constructions: nowadays both *ararua* and *ananake* are used as comitative

4 Closed word classes

### **4.3.5 Fractions**

*ꞌAfa* means 'half'. It is only used in *ꞌe te ꞌafa* 'and a half', supplementing a whole number:

(49) *e* num *toru* three *mētera* meter *ꞌe* and *te* art *ꞌafa* half 'three and a half meters' [Notes]

The expression as a whole was borrowed from Tahitian, which in turn borrowed the word *ꞌafa* from English 'half'.

There are no common terms to express other fractions. They can be circumscribed using *ꞌapa* 'part'. In the Bible translation, where certain fractions occur, this may lead to constructions such as the following:

(50) *Ko* prf *mate* die *ꞌana* cont *e* num *tahi* one *ꞌapa* part *o* of *te* art *ꞌapa* part *e* num *toru* three *o* of *te* art *taŋata.* man 'One third of the people (lit. one part of the three parts of the people) had died.' [Rev. 9:20]

The word *ꞌapa* was probably borrowed from Tahitian, where it means 'half of a fish or animal, cut lengthwise' or 'piece of tissue, patch' (Académie Tahitienne 1999: 49). It is not used in older texts.

### **4.4 Quantifiers**

### **4.4.1 Overview**

Quantifiers are semantically similar to numerals in that they express a quantity; unlike numerals, quantifiers do not indicate an exact amount.

The quantifiers of Rapa Nui are listed in Table 4.12. As the table shows, the quantifier system has undergone significant changes over the past century. A number of new quantifiers have been introduced, while others have undergone semantic shifts.

Table 4.13 lists a few words which are syntactically different from quantifiers (i.e. they do not occur in the same positions in the noun phrase) but are discussed in this section because they have a quantifying sense.

In Rapa Nui, quantifiers are syntactically like numerals in two respects:

markers, while other definite numerals are not used as such (§8.10.3).

#### 4.4 Quantifiers


#### Table 4.12: List of quantifiers

#### Table 4.13: Quantifier-like words



There are important differences as well. Quantifiers are not preceded by the numeral particles *e*, *ka* and *hoko*. And even though they seem to occupy the same positions in the noun phrase, on closer analysis they sometimes turn out to be in a different position. In fact, quantifiers also differ from each other in the positions in which they can occur. They may be pre- or postnominal; if prenominal, they occur before or after the article or without article. Table 4.14 lists the position(s) of each quantifier in the noun phrase.<sup>22</sup>

This table demonstrates that the position of quantifiers is lexically determined. All quantifiers occur before the noun, only some after the noun. *Taꞌatoꞌa* 'all' and *rauhuru* 'diverse' occur both pre- and postnominally; the other quantifiers only occur before the noun. The position with respect to the article *te* is lexically determined as well: whereas *paurō* 'all' is always followed by the article, the other quantifiers mostly occur without article or after the article. The question whether the position of the quantifier has semantic repercussions, is discussed in the subsections on the individual quantifiers.

<sup>22</sup> Not included are *ananake* (which rarely occurs within a noun phrase in modern Rapa Nui) and the minor quantifiers *kā* and *pūra*.

<sup>23</sup> When *me ꞌe rahi* occurs before the article, it is external to the noun phrase.

#### 4 Closed word classes


Table 4.14: Distribution of quantifiers in the noun phrase

### **4.4.2** *Taꞌatoꞌa* **'all'**

The universal quantifier *taꞌatoꞌa* 'all' is the most common quantifier in modern Rapa Nui. It is a relative newcomer, borrowed from Tahitian.<sup>24</sup> It occurs in a variety of positions in the noun phrase; a difference in position does not always imply a clear difference in meaning.

*Te N taꞌatoꞌa* The most common position of *taꞌatoꞌa* is after the noun, before postnominal demonstratives (see the chart in §5.1). The noun is preceded by the article *te* or another determiner:


*Te taꞌatoꞌa N taꞌatoꞌa* may also appear before the noun, after the determiner:

(53) *ꞌI* at *te* art *mahana* day *nei* prox *te* art *taꞌatoꞌa* all *ŋā* pl *poki* child *he* ntr *porotē.* parade 'Today all the children participate in the parade.' [R334.324]

The difference between *te N taꞌatoꞌa* and *te taꞌatoꞌa N* is mainly a stylistic one: some speakers freely use *taꞌatoꞌa* prenominally, others feel that *te taꞌatoꞌa N* is less grammatical. On the whole, postnominal *taꞌatoꞌa* is much more common.

Yet there is also a slight difference in meaning: at least for some speakers, prenominal *taꞌatoꞌa* is somewhat emphatic. Compare (54) with (53) above: (53) indicates 'all without exception', while (54) is more neutral.

<sup>24</sup> It is found in Fel, Blx and newer texts.

4.4 Quantifiers

(54) *ꞌI* at *te* art *mahana* day *nei* prox *te* art *ŋā* pl *poki* child *taꞌatoꞌa* all *he* ntr *porotē.* parade 'Today all the children participate in the parade.'

As (53) and (54) show, the noun phrase containing *taꞌatoꞌa* may include a plural marker. In most cases, however, no plural marker is used; the use of *taꞌatoꞌa* itself is a sufficient indication of the plurality of the noun phrase.

*Taꞌatoꞌa te N* Very occasionally, *taꞌatoꞌa* occurs before the article:

(55) *Taꞌatoꞌa* all *te* art *taŋata* person *o* of *Rapa* Rapa *Nui* Nui *i* pfv *oho* go *ai* pvp *ki* to *te* art *pure* prayer *pāpaku* corpse 'All the people of Rapa Nui went to the funeral mass.' [R231.349]

This is merely a stylistic variant of *te taꞌatoꞌa N*. This construction is never used when the noun phrase is preceded by a preposition.

*Taꞌatoꞌa N Taꞌatoꞌa* often occurs before the noun without a determiner. This is only possible when the noun phrase is not preceded by a preposition: prepositions require a deteminer to be present (§5.3.2.1). *Taꞌatoꞌa* without determiner mostly occurs in noun phrases at the beginning of the sentence or clause, as in (56). However, it may occur further on in the sentence as well, as in (57).


This use of *taꞌatoꞌa* may indicate a more generic, less exact quantification, without establishing a precisely defined group: 'all sorts of, everything, whatever'.

*(Te) taꞌatoꞌa Taꞌatoꞌa* may occur without an accompanying noun, i.e. in a headless noun phrase (§5.6). In this case, it can be translated as 'all, everyone, the totality'. Headless *taꞌatoꞌa* may occur either with or without article:


4 Closed word classes

The choice between *taꞌatoꞌa* and *te taꞌatoꞌa* in headless noun phrases is partly syntactically determined: when the noun phrase is preceded by a preposition, there needs to be a determiner. This is the case in (58). When the context does not require a determiner, the determiner tends to be left out, as in (59). However, this is no absolute rule.

**With pronoun** Finally, *taꞌatoꞌa* may quantify a pronoun; usually it appears after the pronoun:

(60) *E* voc *koro,* Dad *ꞌī* imm *a* prop *mātou* 1pl.excl *taꞌatoꞌa* all *ia.* then 'Dad, here we all are!' [R237.051]

### **4.4.3** *Paurō* **'each'**

Like *taꞌatoꞌa*, *paurō* 'each, every, all' is a newcomer in Rapa Nui, borrowed from Tahitian *pauroa*. Interestingly, it already occurs in Mtx and Egt, but only once in each. It is much more common in recent texts.

*Paurō* usually precedes the determiner and is mostly used with temporal nouns like *mahana* 'day', *vece* 'time, turn', *matahiti* 'year'. Some examples:


Occasionally *paurō* is used with other nouns, mostly after the noun. In these cases it is equivalent to *taꞌatoꞌa*:

(64) *Te* art *aŋa* do *he* pred *riꞌariꞌa* afraid *nō* just *te* art *taŋata* man *paurō* every *ꞌi* at *tū* dem *ŋā* pl *tahutahu* witch *era.* dist 'All the people were continually afraid of those witches.' [R233.007]

4.4 Quantifiers

### **4.4.4** *Ananake* **'together'**

In old texts, *ananake* is the most common quantifier; in these texts it has a wide range of uses, much like *taꞌatoꞌa* nowadays. In modern Rapa Nui, the use of *ananake* is semantically and syntactically restricted. In the following sections, these two stages are discussed separately.

*Ananake* does not occur in other languages, but the simple form *anake*<sup>25</sup> is widespread in Polynesian (< PNP *\*anake* = 'completely, only'). *Ananake* may have developed from *anake* by analogy of the definite numerals (§4.3.4): the development *anake* > *ananake* is very similar to *(a) toru* > *a totoru*. This would explain the otherwise unattested reduplication pattern, in which the penultimate syllable of a three-syllable word is reduplicated. As discussed in §4.3.4 above, *ananake* shows similarities in use to the definite numerals.

#### **4.4.4.1 Modern use**

The modern sense of *ananake* is 'together, all together'. It is mostly used pronominally: *ananake* is not accompanied by a noun, nor preceded by an article. Its referent must be known from the preceding context. Some examples:


As these examples show, *ananake* may occur after the verb in the subject position as in (65), but also before the verb as in (66), or at the end of the clause as in (67).

Regardless of its position in the clause, *ananake* always refers to the subject. For example, (67) does not mean 'they carried all the lobsters'. Now this also has a semantic reason: *ananake* normally has human reference; it is uncommon for *ananake* to be used for animals or inanimate things.

Another current use of *ananake* is in the comitative construction (§8.10.3).

#### **4.4.4.2** *Ananake* **in older Rapa Nui**

The modern pronominal use, in which *ananake* quantifies an implied subject, already occurs in older texts. More commonly, however, *ananake* is used in these texts as a

<sup>25</sup> *Anake* also occurs in Rapa Nui, but only in older texts.

#### 4 Closed word classes

quantifier within the noun phrase. This syntactic difference between the old and the modern language coincides with a semantic difference: while in modern Rapa Nui *ananake* means 'together', in older texts it is a universal quantifier 'all', a sense nowadays expressed by *taꞌatoꞌa* and *paurō*.

Just like *taꞌatoꞌa* nowadays, *ananake* in the older language may occur after a noun or pronoun:


It also occurs before the noun; in that case it precedes the article *te*.


Prenominal *ananake* is never preceded by a preposition. Even so, the examples show that it may occur in noun phrases with a variety of functions, for example subject as in (70), locational adjunct as in (71), or temporal adjunct as in (72). But *ananake te N* is especially common with nouns denoting place or time, as in (71–72), a construction that has been taken over by *paurō te N* nowadays.

### **4.4.5** *Rauhuru* **'diverse'**

*Rauhuru* means 'diverse, manifold, many kinds'. It is a recent word, derived from *rau* 'one hundred (archaic)' + *huru* 'kind, sort'. Like *taꞌatoꞌa*, it occurs before and after the noun, with and without article, preceding and following the article.

#### *Rauhuru te N*

(73) *I* pfv *noho* stay *era* dist *te* art *oromatuꞌa* priest *ꞌi* at *nei,* prox *he* ntr *takeꞌa* see *rauhuru* diverse *te* art *meꞌe* thing *mātāmuꞌa.* past 'When the priest lived here, he saw manifold things of the past.' [R423.021]

4.4 Quantifiers

#### *Te rauhuru N*

	- art man

'Now we are going to talk about the different kinds of work of people.' [R334.203]

#### *Rauhuru N*

	- '… to do all sorts of good things which he wants' [2 Tim. 3:17]

The use or non-use of the article is partly determined by syntax: after prepositions the article is obligatory. Partly it is a matter of style; the article is more common in this construction in certain texts than in others.

*(Te) N rauhuru* The postnominal use of *rauhuru* is limited to some speakers. The article may or may not be used.


**As a noun** Finally, *rauhuru* itself can also be used as a noun, followed by a possessive phrase:

(78) *Te* art *aŋa* work *ꞌāꞌana* poss.3sg.a *he* pred *ꞌoka* plant *i* acc *te* art *kai,* food *i* acc *te* art *rauhuru* diverse *o* of *te* art *meꞌe.* thing 'His work was planting food, all kinds of things' [R444.015–016]

Nominalised *rauhuru* may or may not be preceded by the article. Again, this choice is partly prescribed by the syntax, partly free.

### **4.4.6** *Tētahi* **'some, other'**

In this section first the syntax of *tētahi* will be discussed (§4.4.6.1), then its meaning (§4.4.6.2).

#### 4 Closed word classes

#### **4.4.6.1 Syntax of** *tētahi***:** *te* **+** *tahi***?**

*Tētahi* 'some, other' is an ambiguous element. Its origin is clear: the article *te* + the numeral *tahi* 'one'. *Tētahi* still betrays this origin when it occurs after prepositions:


These prepositions are obligatory followed by a determiner (§5.3.2.1). The fact that they can be followed by *tētahi* shows that in these cases *tētahi* contains a determiner, the most natural explanation being that *tētahi* consists of the article *te* followed by *tahi*.

Yet in other cases *tētahi* does not incorporate a determiner. It can be preceded by determiners, such as the article *te* (82) or a demonstrative (83):


Also, *tētahi* may follow the preposition *hai*, a preposition which is never followed by a determiner (§4.7.10):

(84) *A* prop *Kontiki* Kontiki *takoꞌa* also *i* pfv *hāꞌūꞌū* help *mai* hither *hai* ins *tara* money *ꞌe* and *hai* ins *tētahi* some/other *atu* away *meꞌe.* thing 'Kontiki (=Thor Heyerdahl) also helped with money and with other things.' [R375.094]

We may conclude that *tētahi* has – at least in these cases – undergone a process of reanalysis and turned into a monomorphemic quantifier which no longer includes a determiner.

4.4 Quantifiers

#### **4.4.6.2 Use of** *tētahi*

*Tētahi* can be used with singular nouns in the sense 'another':

(85) *E* exh *hoki* return *mai* hither *hoꞌi* indeed *koe* 2sg *ꞌi* at *tētahi* some/other *mahana.* day 'Come back another day.' [R344.034]

More commonly, the noun has a plural sense, and *tētahi* means 'some' or 'others':


Multiple noun phrases can be conjoined in juxtaposed clauses using *tētahi … tētahi*: 'some … others':

(88) *…tētahi* some/other *ŋā* pl *poki* child *tane* male *nunui* pl:big *he* ntr *hāpī* learn *mo* for *haka* caus *taŋi* cry *i* acc *te* art *kītara.* guitar *Tētahi* some/other *ŋā* pl *poki* child *he* ntr *hāpī* learn *i* acc *te* art *ꞌori* dance *rapa* shine *nui,* crouch *tētahi* some/other *hakaꞌou* again *mo* for *ꞌori* dance *i* acc *te* art *cueca. cueca*

'…some bigger boys learn to play the guitar. Other children learn Rapa Nui dancing, yet others dancing the *cueca*.' [R334.130–131]

As the last clause in (88) shows, *tētahi* can also be used without a following noun.

### **4.4.7** *Meꞌe rahi* **and** *rahi* **'much, many'**

#### **4.4.7.1** *Meꞌe rahi***: from noun phrase to quantifier**

*meꞌe rahi*, lit. 'many things', is originally a noun phrase, consisting of the noun *meꞌe* 'thing', modified by the adjective *rahi* 'much/many'. The few times when it is used in older texts (there are only four occurrences), it is used as such. In the following example, the noun phrase *meꞌe rahi* is in initial position as the predicate of an attributive clause,<sup>26</sup> followed by the subject noun phrase.

(89) *Meꞌe* thing *rahi* many/much *te* art *manu* bird *o* of *ruŋa.* above 'There were many birds (lit. many [were] the birds) on (the island).' [Egt-02.083]

<sup>26</sup> Attributive clauses commonly have the dummy noun *me ꞌe* as anchor of the predicate adjective (§9.2.7).

#### 4 Closed word classes

This example has the same structure as the attributive clause below (§9.2.7):

(90) *Meꞌe* thing *paŋahaꞌa* heavy *te* art *kūmara.* sweet\_potato 'Sweet potatoes are heavy (food).' [Ley-5-24.008]

Nowadays *meꞌe rahi* is still used in the same way, i.e. as a predicate of an attributive clause. If this construction contains a verb, possibly with other arguments, this is constructed as a relative clause following the subject.

(91) *Meꞌe* thing *rahi* many/much *te* art *taŋata* man *[i* pfv *mate* die *ꞌi* at *rā* dist *noho* stay *iŋa].* nmlz 'Many people died (lit. many [were] the people who died) at that time.' [R250.093]

However, this is not the most common way in which *meꞌe rahi* is used nowadays. It has also developed into a frozen form which as a whole functions as a quantifier, occupying the quantifier position in the noun phrase. *Meꞌe rahi* as a quantifier is distinguished by the following characteristics:


'They distributed land to many people to plant corn.' [R424.013]

(95) *ꞌI* at *te* art *kona* place *nei* prox *i* pfv *noho* stay *ai* pvp *meꞌe* thing *rahi* many/much *mahana.* day 'In this place this stayed many days.' [R420.055]

4.4 Quantifiers

These examples show that reanalysis has taken place. As discussed above, in older Rapa Nui *meꞌe rahi* was the predicate of a nominal clause, optionally containing a relative clause:

(96) [ *Meꞌe rahi* ]NP [ *te N* ([ *i V* ]Rel) ]NP

This construction was reanalysed to a simple clause with initial subject, in which *meꞌe rahi* is a quantifier occurring before the article, by analogy of other quantifiers which may occur in the same position (e.g. *taꞌatoꞌa te N*, §4.4.2):

(97) [ *Meꞌe rahi te N* ]NP [ *i V* ]VP

Once *meꞌe rahi* is part of the noun phrase, the way is open for two developments:

	- 2. *meꞌe rahi* may occur in non-initial noun phrases with different semantic roles, as in (93–95) above.

There is still one difference with quantifiers like *taꞌatoꞌa*: *meꞌe rahi* is not preceded by the article. If the article is used, it follows *meꞌe rahi*.

#### **4.4.7.2** *Rahi* **'many, much'**

*Rahi* is used in the expression *meꞌe rahi* (see above), but also has a number of other uses. *Rahi* occurs in older texts, but not nearly as frequently as in modern Rapa Nui.<sup>27</sup> Though the word occurs throughout Polynesia, Rapa Nui may have borrowed it from Tahitian, or extended its usage under the influence of Tahitian.<sup>28</sup> Apart from the marked increase in use, another indication for Tahitian influence is the fact that *rahi* can be followed by the Tahitian nominaliser *-raꞌa*.

**Predicate** In older texts, *rahi* is mainly used as a verbal/adjectival predicate.

(99) *Ku* prf *rahi* many/much *ꞌā* cont *te* art *mamae* pain *o* of *te* art *viꞌe* woman *ꞌa* of.a *Tau* Tau *ꞌa* a *Ure* Ure *rāua* 3pl *ko* prom *tāꞌana* poss.3sg.a *poki.* child 'Tau a Ure's wife and her child were in much pain (lit. Much was the pain of…)' [Ley-9-63.019]

This usage is still common nowadays. *Rahi*, preceded by an aspectual marker, can be the predicate of either a main clause or a relative clause after the noun:

<sup>27</sup> In older texts, *rahi* (including *me ꞌe rahi*) occurs 20x (once per 6,100 words), in newer texts it occurs 896x (once per 410 words).

<sup>28</sup> The form of the word would be the same, whether inherited or borrowed.

4 Closed word classes


**Adverb** *Rahi* often serves as adverb, modifying a verb or adjective: 'a lot, very (much)'. When modifying a verb, *rahi* often implies quantification of the subject or object of the verb (in the same way as *tahi* 'all', §4.4.9). E.g. in (103), *rahi* does not refer to many acts of seeing, but quantifies the object of seeing.


**Noun** When *rahi* is used a noun, it means either 'the many, the large number' or 'the majority, most'. The counted entity is expressed as a genitive phrase after *rahi*. 29


**Adjective** Finally, *rahi* is used adjectivally, i.e. as a noun modifier. As discussed in §4.4.7, in older texts the expression *meꞌe rahi* is found occasionally, in which *rahi* is an adjective to the generic noun *meꞌe*. There is only one example in these texts of *rahi* modifying a noun other than *meꞌe*:

<sup>29</sup> Some speakers use *rahira ꞌa* in the same senses, either 'the many' or 'the majority'. (*-ra ꞌa* is the Tahitian nominaliser).

4.4 Quantifiers

(106) *He* ntr *toꞌo* take *mai* hither *i* acc *te* art *moa,* chicken *moa* chicken *rahi.* many/much 'They took chickens, many chickens.' [Ley-9-55.012]

Though one example does not carry too much weight, it is interesting to note that the adjective is not simply put after the noun *moa*. Rather, *moa* is repeated as an apposition, yielding a sort of predicate noun phrase to which *rahi* is added. (Appositions in Rapa Nui are similar to predicate noun phrases.)

The use of *rahi* as an adjective thus seems to be a recent development. *Rahi* as an adjective is relatively common nowadays, though still not quite as common as the predicate and adverbial uses of *rahi*. Speakers hesitate somewhat to use *rahi* as an adjective; when they do so, it is often in situations where a construction with *meꞌe rahi* is difficult or impossible. *Rahi* as an adjective is especially found in the following situations:

First: when the noun phrase is preceded by a preposition requiring a determiner.

(107) *He* ntr *haꞌere* walk *mo* for *haka* caus *ora* live *ꞌi* at *te* art *rohirohi* tired:red *o* of *tū* dem *aŋa* work *rahi* many/much *era.* dist 'He went to rest from the fatigue of those many works.' [R233.069]

Second: when the quantifier is negated by the constituent negator *taꞌe*.

(108) *Ika* fish *taꞌe* conneg *rahi* many/much *nō* just *i* pfv *ravaꞌa* obtain *ai.* pvp 'They caught few fish.' [R312.010]

Third: when the noun modified by *rahi* is itself a modifier:

(109) *ꞌI* at *te* art *mahana* day *tokerau* wind *rahi,* many/much *e* ipfv *ko* neg.ipfv *eꞌa* go\_out *ki* to *te* art *ika* fish *hī.* to\_fish 'On days with much wind, (people) don't go out fishing.' [R334.254]

Fourth: in predicate noun phrases, especially in attributive clauses:

(110) *Nuꞌu* people *rahi* many/much *te* art *nuꞌu* people *i* pfv *manaꞌu* think *pē* like *nei* prox *ē* thus *ko* prf *tētere* pl:run *ꞌana* cont *ki* to *Tahiti.* Tahiti

'Many people (lit. many people were the people who) thought that they had fled to Tahiti.' [R303.051]

In fact, this is the same construction as *meꞌe rahi* when used as a noun phrase (see (89–91) above).

Finally: with abstract nouns like *riva* 'goodness', *mamae* 'pain', *haŋa* 'love', *aŋa* 'work' and *manaꞌu* 'thought'. *Rahi* can be translated here as 'much, great':

4 Closed word classes


### **4.4.8 Other quantifiers**

#### **4.4.8.1** *Kē* **'some, others'**

*Kē* is common as an adjective meaning 'other, different', but in modern Rapa Nui it also serves as a quantifier in the sense 'some' or 'other(s)'. When used as a quantifier, it occurs before the noun; the noun phrase has no determiner.

*Kē* is similar in meaning to *tētahi*, but more than *tētahi* it singles out a subgroup within a larger group. Often, two subgroups are juxtaposed: *kē … kē* 'some … others'.


This use of *kē* may be influenced by Spanish, where quantifiers like *ciertos* and *algunos* (both meaning 'certain') occur before the noun.

#### **4.4.8.2** *Kā* **'every'**

*Kā* 'every' is an adaptation of Spanish *cada*. <sup>30</sup> It occurs before the noun and may be preceded by the article *te*:


<sup>30</sup> As intervocalic *d* is pronounced very weakly in Chilean Spanish, it tends to drop out completely in Rapa Nui borrowings (§2.5.3.1).

4.4 Quantifiers

Unlike Spanish *cada*, which precedes only singular nouns, *kā* is not limited to singulars: it may be followed by the plural marker *ŋā* or the inherently plural noun *nuꞌu* 'people'.


#### **4.4.8.3** *Pura* **'mere, only'**

*Pura* is borrowed from Spanish *pura* (feminine of *puro*) and means 'mere, only, pure, sheer, simple'. It may or may not be preceded by the article or the predicate marker *he*, depending on the syntactic requirements of the context.


'The Tapati Rapa Nui is a week, seven days of just dancing.' [R240.003]

(121) *Kahu* clothes *ꞌō,* really *pura* only *kahu* clothes *teatea* white:red *e* ipfv *uru* dress *era* dist *ꞌi* at *rā* dist *tiempo.* time 'As for the clothes, they wore only white clothes at that time.' [R416.1272]

### **4.4.9** *Tahi* **'all'**

*Tahi* is the numeral 'one' (§4.3.1.1), and as such it is always preceded by one of the numeral particles *e*, *ka* or *hoko*. Apart from that, *tahi* is also used as an adverb in the verb phrase, in the sense 'all'.<sup>31</sup> *Tahi* has reference not to the action itself (in which case it would indicate that the action happens completely), but to one of the arguments of the verb. This argument is usually plural (whether explicitly indicated or not) and *tahi* indicates that all of the entities referred to by the noun phrase are concerned by the action.

*Tahi* may have reference to an O argument as in (122), an S argument as in (123), or an A argument as in (124):

<sup>31</sup> This use of *tahi* does not occur in older texts and is probably borrowed from Tahitian, where *tahi* likewise occurs as a VP adverb meaning 'all'.

4 Closed word classes


When both arguments of a transitive verb are plural, the reference of *tahi* may be ambiguous. In the following example, *tahi* may quantify either the implied Agent, or the Patient 'the sweet potatoes'.

(125) *He* ntr *keri* dig *tahi* all *rāua* 3pl *i* acc *te* art *kūmara.* sweet\_potato 'They dug up all the sweet potatoes' or 'They all dug up the sweet potatoes' [Notes]

### **4.4.10 The quantifier phrase**

Unlike numerals, quantifiers are not preceded by obligatory particles. However, like numerals they can be followed by certain particles; in other words, they are the nucleus of a quantifier phrase.

Universal quantifiers are often followed by the limitative particle *nō* 'simply, just' (§5.8.2), which emphasises that the quantifier involves all people or things, without exception.


After *tētahi* 'some/others', and occasionally after *meꞌe rahi* 'many' and *rauhuru* 'diverse', the directional particle *atu* is used. After verbs this particle indicates a movement away from the speaker (§7.5), but it may also emphasise a quantity or extent (§7.5.1.5); the latter is relevant when it is used after a quantifier.

4.4 Quantifiers

(128) *He* ntr *iri* ascend *te* art *poki* child *ki* to *ꞌuta* inland *tuatua* dig:red *i* acc *te* art *kūmara,* sweet\_potato *ananake* together *ko* prom *tētahi* other *atu* away *ŋā* pl *poki.* child

'The child went to the field to harvest sweet potatoes, together with other children.' [Mtx-7-25.009]

The same quantifiers may be followed by *hakaꞌou* 'again' (§4.5.3.4), here in the sense 'more, others', which serves to single out a second or further subgroup:

(129) *Tētahi* some *ŋā* pl *poki* child *he* ntr *hāpī* learn *i* acc *te* art *ꞌori* dance *rapa* shine *nui,* crouch *tētahi* other *hakaꞌou* again *mo* for *ꞌori* dance *i* acc *te* art *cueca.* cueca 'Some children learn Rapa Nui dancing, others (learn) to dance the cueca.' [R334.131]

### **4.4.11 Conclusions**

The sections above have shown that quantifiers occur in different positions in the noun phrase: after the noun, after the article, without article, sometimes before the article. The positional possibilities are different for each quantifier, as shown in Table 4.14 on p. 160; however, there is a general tendency for prenominal placement, as well as a tendency to omit the article when the determiner is prenominal. In fact, the five most common quantifiers (*taꞌatoꞌa* and *paurō* 'all', *rauhuru* 'diverse', *tētahi* 'some' and *meꞌe rahi* 'many') all occur in the construction *QTF N*. For quantifiers occurring in multiple positions, there may be subtle semantic differences between different placements, but it does not seem to be possible to formulate general rules across the group.

As Table 4.12 on p. 159 shows, the quantifier system has undergone significant changes over the last century:


Interestingly, while *taꞌatoꞌa* 'all' and *paurō* 'all' were borrowed from Tahitian, their syntax differs from their Tahitian equivalent. In Tahitian, both quantifiers only occur after the noun or pronoun they modify (Lazard & Peltzer 2000: 172, Académie Tahitienne

#### 4 Closed word classes

1986: 148–149). They never occur before the noun, though *taꞌatoꞌa* does occur independently: *te taꞌatoꞌa* 'the totality' (Académie Tahitienne 1986: 149). Both elements also occur after verbs; in the examples given by Lazard & Peltzer (2000: 147), they quantify the subject of the verb, in the same way as *tahi* in Tahitian and Rapa Nui. By contrast, in Rapa Nui, *taꞌatoꞌa* occurs either before or after the noun or independently, but only rarely after verbs. When *taꞌatoꞌa* occurs independently in Rapa Nui, it may or may not be preceded by the article; in Tahitian, the article is obligatory.

Likewise, Rapa Nui *paurō* is quite different from its Tahitian equivalent *pauroa*: while the latter occurs after nouns and verbs, Rapa Nui *paurō* usually precedes the article and mainly occurs with temporal nouns. The differences are summarised in Table 4.15.


Table 4.15: Distribution of Tahitian and Rapa Nui quantifiers

We may conclude that, even though the form and meaning of *taꞌatoꞌa* and *paurō* were borrowed from Tahitian, they acquired a distinctive Rapa Nui syntax, which they partly inherited from *ananake*. For *tahi* a different development took place: the word already existed in Rapa Nui as numeral 'one', but came also to be used as a quantifier-like adverb in the VP. If this happened under the influence of Tahitian – as seems likely – this means that an existing word acquired a new syntactic behaviour through borrowing.

Another language-internal development in Rapa Nui is the change of *tētahi* 'some, other', originally a combination of article + numeral, into a monomorphemic quantifier which does not include a determiner.

Last of all, *rahi* 'much' has undergone a significant syntactic shift. While it used to function predominantly as an adjectival predicate, it came to be used as an adjectival modifier of *meꞌe* 'thing' (a construction already found in old texts, but only sporadically), a combination which subsequently developed into a quantifier.

To summarise: the Rapa Nui quantifier system has radically changed over the past century, partly under Tahitian and Spanish influence, partly as a language-internal development. But even borrowed elements show a syntactic behaviour which is distinctly Rapa Nui.

4.5 Adverbs

### **4.5 Adverbs**

There are two classes of adverbs in Rapa Nui: verb phrase adverbs, which are part of a verb phrase, and sentential adverbs, which form a separate constituent in the clause. These two classes are discussed in §4.5.1 and §4.5.2, respectively. The two sets are largely distinct.

In §4.5.3, a number of individual adverbs are discussed.

### **4.5.1 Verb phrase adverbs**

Adverbs in the verb phrase occur immediately after the verb (see the chart in §7.1). The following words function primarily as verb phrase adverbs:


*Takoꞌa* 'also' and *koroꞌiti* 'slowly, softly' are also used as sentential adverbs. *Takoꞌa, hakaꞌou* 'again' and *mau* 'really' also occur as adverbs in the noun phrase.

Other words occur both as adjectives and as verb phrase adverbs; this includes words like *rivariva* 'good; well', *rahi* 'much/many', *ꞌitiꞌiti* 'small; a bit', *raꞌe* 'first' (§3.6.4.1), *riꞌariꞌa* 'terrible; terribly, very', *kē* 'different(ly)', *pūai* 'strong(ly)'. The first two occur very frequently as adverbs, the others somewhat less.

Still other words occur as adverbs very occasionally; they function primarily as adjectives or verbs. Examples are *parautiꞌa* 'truth; true, truly', *hōrou* 'quick(ly)', *nuinui* 'big; in a big way, on a large scale'; *ora* 'to live; alive', *reoreo* 'to lie; lying', *tano* 'correct; somewhat (after an adj.)', *rikiriki* 'small (pl.); a bit'.

Though the verb phrase chart in §7.1 shows a single adverb slot, the verb may be followed by more than one adverb, as the following examples show:


#### 4 Closed word classes

All adverb combinations in the corpus occur in a consistent order. For example, *V rivariva takoꞌa* is attested, but *\*V takoꞌa rivariva* is not. Therefore it is possible to establish a number of ordered adverb slots, as shown in Table 4.16. 33

Table 4.16: Order of postverbal adverbs


Another indication for the existence of multiple adverb slots is found in nominalised phrases: *tahi* and *koraꞌiti* (slot 2) occur before the nominalising suffix, while adverbs from slots 3 and 4 occur after the suffix (§3.2.3.3).

### **4.5.2 Sentential adverbs**

Sentential adverbs are a constituent on their own; they are not part of a noun phrase or verb phrase.

Sentential adverbs form a small class in Rapa Nui. They include words expressing temporal notions relating to the future, like *ꞌanīrā* 'later today' and *āpō* 'tomorrow' (§3.6.4).<sup>34</sup> Apart from these, there are only a few common sentential adverbs: *takoꞌa* 'also', *koraꞌiti* or *koroꞌiti* 'slowly', *koiꞌite* 'perhaps, if perhaps', *korohaŋa* 'even' and *pēaha* 'perhaps, probably'.<sup>35</sup> Some of these are discussed individually in §4.5.3 below. Two of them, *takoꞌa* and *koroꞌiti*, also occur in the verb phrase.

Sentential adverbs occur in different positions in the clause. For example, while *tako ꞌa* 'also' as a sentential adverb is usually clause-initial (see (139) below), *pēaha* 'perhaps' occurs after the initial constituent as in (132):

<sup>33</sup> All adverbs in this table co-occur with at least one adverb in the contiguous slot(s), i.e. all adverbs in slot 2 co-occur with an adverb in slot 1 and with one in slot 3, and so on. Adverbs in the same slot do not co-occur in the corpus. Adverbs in the combined slot 1 + 2, such as *ꞌiti*, do not co-occur with any adverb in slot 1 or 2.

<sup>34</sup> Notions like 'here' and 'there' are not expressed by adverbs, but by a preposition + locational (§4.6.5.1). The same is true for temporal notions related to the past, like 'yesterday'.

<sup>35</sup> A few other sentential level adverbs are used occasionally, such as *pōrā/pōpōrā* 'quickly' and *umarā* 'hurriedly'. These will not be discussed separately.

4.5 Adverbs

(132) *Ku* prf *toke* steal *ꞌā* cont *pēaha* perhaps *to* art:of *tāua* 1du.incl *tāropa* basket *ꞌura.* lobster 'Our basket of lobsters seems to have been stolen.' [Mtx-7-28.050]

Sentential adverbs can be modified by particles and form an adverb phrase. For example, time adverbs may be followed by *mau* 'really', *nō* 'just', postnominal demonstratives, and the identity marker *ꞌā*. In the following example, *āpō* is followed by no less than three particles:

(133) *Āpō* tomorrow *mau* really *ena* med *ꞌā* ident *he* ntr *hakaroŋo* listen *koe* 2sg *i* acc *a* prop *ia.* 3sg 'Tomorrow ('Just tomorrow' or 'Tomorrow exactly') you will hear him.' [Act. 25:22]

### **4.5.3 Individual adverbs**

In this section, a number of adverbs is discussed in more detail. All of these are verb phrase adverbs, though *takoꞌa* 'also' is also used as a sentential adverb.

#### **4.5.3.1** *Iho* **'just then'**

*Iho* (< PPN *\*hifo*) is originally a directional particle 'downwards', which is widespread throughout the Polynesian languages. In all languages apart from Rapa Nui it is a directional, in the same class as *mai* 'hither' and *atu* 'away' (§7.5); additionally, in some languages it is used as a verb in the sense 'to descend'. In many languages, directional particles have additional senses besides the directional one, such as deictic, aspectual and/or reflexive.<sup>36</sup> However, only in Rapa Nui have the following two developments taken place:


*Iho* indicates that an event takes place exactly at, or just prior to, the time of reference. This often implies that something will happen only at the time specified, not earlier. In a perfective clause, this means that the event has just happened: 'recently, just'; in other aspects, *iho* can be translated as 'just at that moment, exactly then'. When *iho* occurs in a main clause with perfective sense, the aspectual tends to be left out, as (134) shows.

<sup>36</sup> See e.g. Cablitz (2006: 427) for Marquesan, Académie Tahitienne (1986: 175, 217, 330) for Tahitian, Elbert & Pukui (1979: 92–95) for Hawaiian.

<sup>37</sup> Notice that *iho* can indicate recent past in Hawaiian (Elbert & Pukui 1979: 92) and Tahitian (Académie Tahitienne 1986: 175).

#### 4 Closed word classes


Sometimes *iho* means 'for the first time':

(136) *He* ntr *piri* get\_together *iho* just\_then *mai* hither *ki* to *a* prop *au* 1sg *te* art *roe* ant *taꞌe* conneg *kai* eat *tihota.* sugar 'This is the first time I meet an ant which doesn't eat sugar.' [R214.008]

#### **4.5.3.2** *Takoꞌa* **'also'**

*Takoꞌa* (variants *tokoꞌa*, *takora*) is an additive connective: 'also, as well'. It may have developed by metathesis from PPN *\*katoa* 'all' (with added glottal): in several other EP languages (Rarotongan, Tahitian, Pa'umotu), reflexes of *katoa* have the same sense 'also'.

*Takoꞌa* is most commonly used to indicate a connection between two clauses. In this function it normally occurs as an adverb in the verb phrase. The following is an example:


In these examples, *takoꞌa* indicates that the event or action applies not only to an entity mentioned previously, but to another entity as well, e.g. a different subject as in (137), or a different object as in (138).

In modern Rapa Nui, *takoꞌa* may also occur clause-initially, sometimes preceded by *ꞌe* 'and'. This construction may be influenced by Spanish, where *(y) además* '(and) moreover' commonly occurs at the beginning of a sentence. This serves to create a link between what precedes and what follows, but unlike the examples above, there is not necessarily a constituent which is identical in both clauses.

In the following example (from a text about marriage in the past), the two sentences linked by *tako ꞌa* describe two aspects of the relationship between the families of the bride and the groom.

4.5 Adverbs

(139) *Te* ntr *huaꞌai* family *o* of *Iovani,* Iovani *ko* prf *ꞌite* know *ꞌā* cont *taꞌe* conneg *he* pred *huaꞌai* family *o* of *rāua* 3pl *te* art *huaꞌai* family *o* of *Tiare.* Tiare *Takoꞌa,* also *ꞌina* neg *a* prop *Iovani* Iovani *kai* neg.pfv *māhani* accustomed *hia* yet *ki* to *a* prop *Tiare.* Tiare 'Iovani's family knew that Tiare's family was not related to them. Also, Iovani did not know Tiare yet.' [R238.004–005]

*Takoꞌa* also occurs in the noun phrase. Just as in the verb phrase, it occurs in the adverb position, before other particles (§5.8.1). Like *takoꞌa* in the verb phrase, it signals that an entity in the clause replaces an entity in the previous clause. It occurs in the noun phrase in the following situations:

Firstly, in nominal clauses, where there simply is no verb to attach to, as in (140):

(140) *He* ntr *viri* roll *i* acc *te* art *moeŋa* mat *ꞌi* at *te* art *kahu,* cloth *ꞌi* at *roto* inside *ꞌana* ident *a* prop *Ure* Ure *takoꞌa.* also 'They rolled the mat (in which Ure was sleeping) in a cloth, Ure too was inside (the cloth).' [Mtx-7-03.016]

Secondly, when the noun phrase in question is preposed, as in (141):

(141) *A* prop *au* 1sg *takoꞌa* also *e* ipfv *hāꞌūꞌū* help *rō* emph *ꞌā* cont *ki* to *tōꞌoku* poss.1sg.o *matuꞌa* parent *hāpaꞌo* care\_for *i* acc *te* art *ꞌanimare* animal *pē* like *Mihaera.* Mihaera

'I also help my father to take care of the animals, like Mihaera.' [R334.293]

Thirdly, in elliptic clauses, where the predicate is omitted. In (142) below, only the contrastive constituent – the locative phrase – is expressed, and *takoꞌa* is added to this constituent.

(142) *Te* the *ŋāŋata* men *o* of *te* art *nohoŋa* stay:nmlz *tuai* old *era* dist *ꞌā,* ident *ꞌi* at *roto* inside *i* at *te* art *hare* hout *te* art *moeŋa* sleep:nmlz *haŋa,* nmlz *ꞌi* at *roto* inside *i* at *te* art *ꞌana* cave *takoꞌa.* also 'The people of the old times, they slept in houses, and also in caves.' [Ley-5-02.001]

#### **4.5.3.3** *Hoki* **'also'**

*Hoki* 'also' is only used in older texts,<sup>38</sup> apart from a few examples of what could be frozen usage in newer texts. It occurs at the end of a constituent; this constituent is

<sup>38</sup> *Hoki* (< PPN *\*foki* 'also') does not occur in MsE, but is common in Mtx and Ley. As *tako ꞌa* also occurs in these corpora, the question is whether there is a clear difference between *hoki* and *tako ꞌa*. As far as there is any difference, it seems to be that *hoki* indicates a stronger contrast. For example, while *tako ꞌa* occurs with implicit subjects, *hoki* never does (see 144 below). Also, while *hoki* is used in preposed (i.e. focused) noun phrases, *tako ꞌa* is not (see 145), at least not in older texts.

4 Closed word classes

typically a noun phrase or verb phrase, but may also be a quantifier phrase (within an NP). *Hoki* signals that the constituent it modifies is added to another constituent of the same kind and in some way parallel to it.

*hoki* can be used to connect NPs or to connect clauses. When it connects two noun phrases, it indicates that the NP is parallel to an earlier NP. This typically happens in lists, where a series of NPs all play the same role in a clause. In this case, *hoki* is attached to the last element of the list.

(143) *ꞌI* at *te* art *tapa* side *te* art *matuꞌa,* parents *a* prop *koro,* Dad *a* prop *nua,* Mum *te* prop *uka* girls *riva,* good *te* prop *repa* young\_man *riva* good *hoki.* also 'To the side are the parents, the fathers, the mothers, the pretty girls, the handsome boys as well.' [Ley-5-24.013]

When *hoki* connects two clauses, it signals that the second clause (in which *hoki* occurs) is parallel in some way to the first clause. Usually this means that both clauses are identical in one or two constituents,<sup>39</sup> while they contrast in one or two other constituents.<sup>40</sup>

When *hoki* connects two clauses, it is usually added to the first constituent of the second clause. This is normally the verb phrase as in (144), but it may be a preposed constituent as in (145):

(144) *He* ntr *toꞌo* take *mai* hither *te* art *nui,* big *he* art *ꞌakaveŋa.* carry\_on\_back *[He* ntr *toꞌo* take *mai* hither *hoki]* also *te* art *ꞌiti,* small *he* ntr *ꞌakaveŋa* carry\_on\_back *takoꞌa.* also

'The oldest (girl) took (the food) and carried it on her back. The youngest also took (food) and also carried it on her back.' [Mtx-7-24.041–042]

(145) *He* ntr *toꞌo* take *mai* hither *i* acc *te* art *ꞌōꞌone…* soil *[Hai* ins *moa* chicken *hoki]* also *ana* irr *toꞌo* take *mai.* hither 'He took dirt… He also took a chicken.' [Ley-5-28.002-004]

#### **4.5.3.4** *Hakaꞌou* **'again'**

*Hakaꞌou*<sup>41</sup> (older variants *hokoꞌou, hakahou*) 'again' marks various kinds of repetition. *Hakaꞌou* may indicate that an event which has happened before, is repeated:

<sup>39</sup> An exception are clauses which are only identical in their subjects; these clauses are extremely common in narrative and don't warrant the use of *hoki*.

<sup>40</sup> Levinsohn (2007: 92) distinguishes several ways in which clauses can be contrasted: "replacing focus" means that one constituent of the clause is replaced by another, while the rest of the clause is identical or synonymous; "prototypical contrast" means that clauses have one constituent in common and two points of contrast. Both of these can be indicated by *hoki*.

<sup>41</sup> *Haka ꞌou* has cognates in several EP languages: Mangarevan *ꞌaka ꞌou* 'again', Tahitian *fa ꞌahou*, Pa'umotu *hakahou*. In these languages it consists of the causative prefix (PEP *\*faka*) plus a reflex of PPN \**fo ꞌou*

(146) *He* ntr *hāŋai* feed *hakaꞌou* again *i* acc *te* art *māꞌaŋa* chick *rikiriki.* pl:small 'He raised chicks again. (He had raised chickens before.)' [Mtx-7-05.021]

More often *hakaꞌou* has a broader sense, indicating that the event has some element of repetition, without being repeated exactly. For example, the action expressed by the verb is performed again, even though the event as a whole is not the repetition of a previous event. In the following example, *oho hakaꞌou* signals that the people keep going, without implying that they had gone to Mount Pu'i before.

(147) *ꞌAi* there *ka* cntg *oho* go *hakaꞌou* again *mai* from *ira* ana *ki* to *Pua* Pua *Katiki.* Katiki 'Then they went (continued their way) from there to Pua Katiki.' [R420.047]

In a negated clause, neg + *hakaꞌou* means 'not any more' (cf. Spanish *ya no*):

(148) *ꞌIna* neg *koe* 2sg *ko* neg.ipfv *taŋi* cry *hakaꞌou.* again '(to someone who is crying): Don't cry anymore.' [R349.016]

In the noun phrase, *hakaꞌou* means 'other, another'.

(149) *…kī* say *tū* dem *taŋata* man *era* dist *ki* to *tū* dem *poki* child *era,* dist *ki* to *tū* dem *taŋata* man *hakaꞌou* again *era.* dist '…said the man to the child and to the other man.' [R102.020]

### **4.5.4 Sentential particles**

There is another small group of words which can be classified as sentential adverbs: they form a separate constituent in the clause and specify the clause as a whole. Unlike the adverbs discussed so far, these elements are not derived from lexical items: they are short, usually monosyllabic, and in this respect they are similar to particles occurring in the noun phrase and the verb phrase. Also, their sense is more general and they are less straightforward to translate. In other words, they have a more grammatical, less lexical character than the adverbs described so far. Hence their characterisation as particles, even though – different from particles occurring in the NP and the VP – they form a constituent on their own.

These elements are described in the following subsections.

<sup>&#</sup>x27;new'. The Rapa Nui reflex of *\*fōu* is *hō ꞌou*; the form *haka ꞌou* suggests that the word was borrowed from Mangarevan, with the initial glottal becoming *h* by analogy of the RN causative prefix *haka*.

#### 4 Closed word classes

#### **4.5.4.1 Deictic particles**

**4.5.4.1.1** *ꞌĪ* **'here'** *ꞌĪ* is a deictic particle expressing immediacy. It is used to point towards an object or event, expressing that it is close in space or time to the speech situation.<sup>42</sup> By drawing attention to the object or event, the latter is put in focus.

*ꞌĪ* is used to draw attention to something which is nearby.

(150) *ꞌĪ* imm *au.* 1sg 'Here I am.' [R363.078]

(151) *Ka* imp *eꞌa* go\_out *koe,* 2sg *ꞌī* imm *tuꞌu* poss.2sg.o *pāpā* father *era* dist *ꞌi* at *haho.* outside 'Go outside, here is your father outside.' [R210.094]

Clauses such as (150–151) could be labeled "presentational": *ꞌī* followed by a nominal constituent serves to signal the presence of someone or something.<sup>43</sup>

*ꞌĪ* may indicate that an event takes place immediately ('right now'):

(152) *ꞌĪ* imm *au* 1sg *he* ntr *oho* go *rō* emph *ꞌai* subs *mai* from *taꞌe* conneg *pō.* night 'I'm going now, before it gets dark.' [R153.042]

More generally, *ꞌī* expresses simultaneity with respect to a time of reference. In (152) above the time of reference is the present; in narrative discourse the time of reference is the time when events in the context take place. In combination with perfect aspect *ko V ꞌā*, *ꞌī* underlines that an event has just taken place.

(153) *ꞌĪ* imm *ku* prf *eꞌa* go\_out *hakaꞌou* again *mai* hither *ꞌana* cont *a* by *ruŋa* above *mai* from *te* art *ꞌara* look\_under\_water *iŋa,* nmlz *ka* cntg *uru* enter *mai* hither *era* dist *ꞌi* at *tū* dem *hora* time *era,* dist *ka* cntg *ŋau.* bite 'Just when he came up again from looking under water, the (shark) entered and bit.' [R361.016]

*ꞌĪ* may convey immediacy and urgency to a statement or request: 'I'm telling you, listen, look…'.

(154) *E* voc *Pea* Pea *ē,* voc *ꞌī* imm *a* prop *Kava* Kava *e* ipfv *taŋi* cry *mai* hither *nei* prox *ki* to *a* prop *au,* 1sg *mai* from *te* art *pōꞌā* morning *ki* to *te* art *hora* time *nei.* prox

'Pea, (listen,) Kava is crying for me, from morning till now.' [R229.017]

<sup>42</sup> *ꞌĪ* is similar in function to focus particles such as *idou* in Koine Greek and *hinneh* in Biblical Hebrew (see Levinsohn 2007: 58, 82; Bailey 2009).

<sup>43</sup> This does not mean that *ꞌī* is a general device to express presentational clauses, e.g. to introduce participants in narrative (on presentational clauses, see Bailey 2009: 4). The use of *ꞌī* in presentational clauses is limited to deictic contexts, where the entity presented is visible to speaker and hearer.

4.5 Adverbs

(155) *ꞌE* and *ꞌī* imm *a* prop *au* 1sg *ka* cntg *hatu* advise *rō* emph *atu* away *ki* to *a* prop *koe.* 2sg 'Look, I'm advising/warning you.' [R310.294]

Often *ꞌī* occurs in combination with a perception verb. *ꞌĪ* has the effect of putting the perceived object into focus. What the participant sees or hears is something significant or even surprising. The act of perception may also be implied, as in (157).


As (154–156) show, the subject of the clause tends to be placed straight after *ꞌī*, before the verb (§8.6.1.1. This is not obligatory, though.

**4.5.4.1.2** *ꞌAi* **'there is'** Just like *ꞌī*, *ꞌai* is a deictic particle, calling attention to an object or event; it indicates greater distance.

*ꞌAi* is used to point at things at a certain distance:


'There are your belongings, which have just been disembarked by the captain of the ship.' [R231.142]

(159–160) are presentational clauses, which indicate the presence of an entity in the distance, just like *ꞌī* presents entities nearby.<sup>44</sup> As (159) shows, initial *ꞌai* may be followed by another *ꞌai* at the end of the clause, just like *ꞌī* may be followed by clause-final *ꞌī aꞌa*.

Like *ꞌī*, *ꞌai* may have a temporal function; it marks a clause referring to a subsequent event:

<sup>44</sup> This use led Fischer (2001a: 319) to take *ꞌai* as derived from Spanish existential marker *hay*. However, the use of *ꞌai* to introduce presentational clauses already occurs in old texts.

#### 4 Closed word classes

(161) *He* ntr *haka* caus *ekeeke* go\_up:red *i* acc *te* art *taŋata* person *ki* to *ruŋa* above *ki* to *te* art *vaka,* boat *ꞌai* there *ka* cntg *maꞌu* carry *ki* to *ruŋa* above *i* at *te* art *pahī.* ship 'They made the people embark the boat, then took them on board the ship.' [R210.042]

As this example shows, *ꞌai* is usually followed by the contiguity marker *ka*. Like *ꞌī*, *ꞌai* may lend emphasis to a clause: 'I'm telling you…':

(162) *Ko* prf *mate* die *era* dist *ꞌana,* cont *ꞌai* there *koe* 2sg *ka* cntg *manaꞌu* think *nō* just *e* ipfv *haꞌuru* sleep *ꞌana.* cont 'She has died, and there you are just thinking that she is asleep!' [R229.303]

*ꞌAi* is marginally used as a deictic preposition 'there in/at' (§4.7.11). Furthermore, *ꞌai* is obviously related to the postverbal particle *ꞌai*, which occurs in the construction *he V rō ꞌai* (§7.2.3.3). It is similar in function: while deictic *ꞌai* frequently indicates sequential events, *he V rō ꞌai* marks final and culminating events in a series.

**4.5.4.1.3** *Nā* **'there near you'** The demonstrative *nā* (which indicates medial distance, see §4.6.4.4) is used as a deictic particle. Like *ꞌī* and *ꞌai* it occurs clause-initially, and usually serves to point at something in the vicinity of the hearer. Different from *ꞌī* and *ꞌai*, *nā* is used only in a spatial sense, not in a temporal sense.


As (165) shows, *nā* may be reinforced by another *nā* at the end of the clause.

#### **4.5.4.2** *Hoꞌi* **and** *paꞌi*

*Hoꞌi* and *paꞌi* are discourse particles which are very common in spoken language; their function is not easy to pin down.<sup>45</sup> They usually occur after the first constituent of

<sup>45</sup> Both particles were borrowed from Tahitian. They are very common in modern Rapa Nui discourse, but do not occur in older texts. Even in the texts collected by Felbermayer and by Blixen in the 1970s, they

4.5 Adverbs

the clause; they lend emphasis to this constituent and/or provide a connection to the previous clause.

**4.5.4.2.1** *Paꞌi* **'in fact'** *Paꞌi* is used to link clauses, indicating that the clause in some way builds upon, confirms or reinforces the preceding clause. In (166), *paꞌi* (2x) confirms what the other speaker has said. In (167), a fragment from an oral text, *paꞌi* appears to be sprinkled throughout the discourse without a very specific function.


More commonly, *paꞌi* is used in clauses providing the grounds for the previous clause: 'for, as' (Spanish *pues*):


'How could I kiss you? You are not with me.' [R228.006–007]

*Paꞌi* seems to have taken on the range of use of Spanish *pues*, which both specifies grounds or reasons ('for, as') and provides confirmation or emphasis ('well, indeed').

occur hardly or not (*pa ꞌi* 0x, *ho ꞌi* 2x). In Tahitian, *pa ꞌi* underlines a statement ('indeed'); *ho ꞌi* may have the same function, but may also connect a statement to the preceding context, for example providing a reason ('for, as'), an addition ('also') or a contrast ('however'). See Académie Tahitienne (1986: 381) and Lazard & Peltzer (2000: 117).

4 Closed word classes

**4.5.4.2.2** *Hoꞌi* **'indeed'** *Hoꞌi* gives (some) emphasis to the preceding constituent. It is used after a wide range of elements, such as deictic particles (170) and the negation *ꞌina* (171). Like *paꞌi*, it may have a confirmatory function: 'indeed' (172).

(170) *ꞌAi* there *hoꞌi* indeed *te* art *taŋata* person *e* ipfv *haꞌamata* begin *era* dist *e* ipfv *tari* carry *era* dist *ki* to *ruŋa* above *i* at *te* art *pahī.* ship

'Then *hoꞌi* the people started to be transported on board the ship.' [R210.040]


*Hoꞌi* may occur at the start of a new episode in discourse, marking a new topic or another initial constituent:

(173) *A* prop *nua,* Mum *hoꞌi,* indeed *e* ipfv *haka* caus *rito* ready *ꞌā* cont *e* num *tahi* one *ꞌavahata* box *kahu.* clothes '(In the meantime,) Mum *hoꞌi* was preparing a box of clothes.' [R210.027]

The constituent marked with *hoꞌi* may be emphasised in opposition to another constituent. In this way, *hoꞌi* may come to express contrast:

(174) *Kai* neg.pfv *ꞌite* know *mai…* hither *Ko* prom *koe* 2sg *hoꞌi* indeed *mo* for *ꞌite* know *i* acc *taꞌa* poss.2sg.a *kai* food *tunu* cook *nei* prox *paꞌi* in\_fact *e* ag *koe.* 2sg 'I don't know… You *hoꞌi* are the one who knows what food you have cooked.' [R236.029–030]

Altogether, *hoꞌi* can be characterised as a spacer: an element which marks the boundary between two constituents and indicates that the preceding constituent is special in some way (Dooley & Levinsohn 2001: 37). According to Levinsohn (2007: 74), it is not unusual for the same spacer in a given language to occur after a topic, a point of departure (such as a time phrase), or a constituent in focus.

4.5 Adverbs

#### **4.5.4.3** *Ia* **'so, then'**

The particle *ia* 'so, then' is a transition marker. It occurs in modern Rapa Nui only.<sup>46</sup> When the clause starts with a verb phrase, *ia* occurs after the verb phrase; (175) shows that it occurs after the VP-final particle *ꞌana*:

(175) *Ko* prf *koa* happy *atu* away *ꞌana* cont *ia* then *a* prop *Tamy* Tamy *ꞌi* at *tū* dem *hora* time *era.* dist 'Tamy was happy at that moment.' [R315.300]

When the verb phrase is not initial in the clause, *ia* occurs either after the first constituent as in (176), or after the verb phrase as in (177).


*Ia* indicates that the event in the clause develops from events mentioned before. It may be the result of, or dependent on, other events ('so, thus'), as in (178) below. In a weaker sense it marks events which are simply subsequent to other events ('then'), or marks the apodosis of conditional clauses as in (179).


Often *ia* marks a new step in the discourse, for example, at the start of a new scene in a story, or a new topic in an exposition. The latter can be seen in the following example:

<sup>46</sup> *Ia* may be a borrowing from Tahitian: Tahitian *ia* is "un anaphorique général qui renvoie d'une manière assez vague à ce qui précède, thème antéposé ou, plus généralement, contexte antérieur" (a general anaphoric which refers in a rather vague way to what precedes, a preposed theme or, more generally, the preceding context, Lazard & Peltzer 2000: 118).

#### 4 Closed word classes

(180) *He* ntr *oꞌo* enter *ia* then *ꞌi* at *te* art *hora* time *nei* prox *ki* to *te* art *aŋa* make *iŋa* nmlz *o* of *te* art *hiꞌo.* glass '(in an explanation of different aspects of diving:) Now let's turn to the making of goggles.' [R360.026]

Some speakers use *ia* in clauses which simply represent the next event in the discourse. Others use it sparingly, or not at all. The frequency of *ia* thus depends to a large degree on the preference of the speaker, just like 'then' in English discourse.

#### **4.5.4.4 The intensifier** *rā*

The intensifying particle *rā*<sup>47</sup> (not to be confused with demonstrative *rā*) occurs in content questions and imperatives. It is placed after the first constituent of the clause; in questions this is the question phrase, in imperatives the verb phrase.<sup>48</sup> *Rā* occurs after the verb phrase-final particle *ꞌana* as in (182), but before other sentence-level particles like *ia* 'then', as in (183).


*Rā* adds an element of insistence to the question or command.<sup>49</sup> It may be used in rhetorical questions, often adding a note of provocation or rebuke:

(184) *¿Mo* for *aha* what *rā* intens *koe* 2sg *i* pfv *kī* say *ai* pvp *i* acc *taꞌa* poss.2sg.a *vānaŋa* word *pē* like *nā?* med 'Why did you say something like that?' [R301.301]

It is also used in non-rhetorical questions, to which the speaker expects a reply. *Rā* conveys a certain vividness and inquisitiveness: the speaker is eager to get an answer. (185), for example, is spoken by a curious child. (186) is spoken by one (teenage) friend to another.

<sup>47</sup> The etymology of *rā* is unknown, but it is probably related to Tahitian *rā*, which also occurs after the first clause constituent. Tahitian *rā* is a contrastive conjunction 'but', but also serves as an intensifier in commands and conditional clauses (Lazard & Peltzer 2000: 98; Académie Tahitienne 1999: 399).

<sup>48</sup> *Rā* itself is not part of the verb phrase: in (181) it occurs after the direction *mai*; in (182) it occurs after the VP-final particle *ꞌana*.

<sup>49</sup> Du Feu (1987; 1996) labels this particle as [-REA] (as opposed to *rō* [+REA]), in line with the fact that it does not occur in statements expressing a fact. Weber (2003b) labels *rā* as dub(itative).

4.5 Adverbs


A question like (186), with its somewhat insistent tone, is only appropriate when talking to friends or close acquaintances. When speaking to strangers, one would use the more neutral ¿*Pē hē koe*? 'How are you?' (Nancy Weber, p.c.).

When *rā* is used in imperative clauses, it marks insistence as well. The context may involve a certain emotion: enthusiasm as in (187), defy as in (188):


#### **4.5.4.5 Asseverative** *ꞌō*

The particle *ꞌō* (etymology unknown, possibly from the exclamation *ꞌō* 'oh!') is asseverative. It occurs after the first constituent of the clause and underlines the truth of the clause. Often, but not always, the clause expresses something unexpected.


*ꞌŌ* is often used in exclamative constructions (§10.4.2), where it underlines that something is not according to normal expectations.

4 Closed word classes

(191) *¡Ko* prom *te* art *ꞌaroha* pity *ꞌō* really *i* at *a* prop *koe!* 2sg 'Poor you! (How pitiable you are!)' [R490.018]

*ꞌŌ* is used in rhetorical questions to which a negative answer is expected. As in other cases, *ꞌō* emphasises that the proposition expressed in the question is not in line with what one would expect.


#### **4.5.4.6 Dubitative** *hō*

*Hō* – a rather rare particle – adds an element of uncertainty or doubt to questions (whether polar or content questions): 'maybe…'. It occurs after the first constituent of the clause.


*Hō* may be used in rhetorical questions to which the expected answer is 'no'.<sup>50</sup>

(196) *¿E* ipfv *ai* exist *rō* emph *ꞌana* cont *hō* dub *te* art *meꞌe* thing *mo* for *taꞌe* conneg *rovaꞌa* obtain *e* ag *te* art *ꞌAtua* God *mo* for *aŋa?* do 'Would there be anything God is not able to do?' [Gen. 18:14]

### **4.6 Demonstratives**

### **4.6.1 Forms**

Rapa Nui does not have a single class of demonstratives. Rather, it has four classes of particles with demonstrative functions. Each class consists of three particles indicating

<sup>50</sup> This use of *hō* is only found in the Bible translation.

#### 4.6 Demonstratives

different degrees of distance: proximal (close to the speaker), medial (at some distance, often close to the hearer) and distal (removed from speaker and hearer).<sup>51</sup> The four classes are similar or even identical in form, but differ in syntactic status; besides, they exhibit certain differences in function.


The first two are what Dixon (2010b: 225) calls "nominal demonstratives", elements in the noun phrase which specify nouns for definiteness, indicate distance with respect to the speaker or hearer, and enable participant tracking in discourse. The others have a more independent status.

The forms for each class are given in Table 4.17. 52

The four classes will be discussed in the following subsections. First the neutral demonstrative determiners will be discussed (§4.6.2), followed by the postnominal demonstratives (§4.6.3), as these commonly occur together. The other demonstrative determiners are discussed in §4.6.4. §4.6.5 deals with deictic locationals, §4.6.6 with demonstrative pronouns.

Postverbal demonstratives are discussed in Chapter 7 (§7.6), as their use is closely tied to other verb phrase elements (especially aspect markers).

<sup>51</sup> Forms of all classes are glossed prox, med and dist, respectively.

<sup>52</sup> Clark (1974) reconstructs two sets of demonstratives for PPN: monomoraic unstressed forms *\*-ni/\*-na/\* ra* and bimoraic long forms *nei/naa/laa*. In Rapa Nui, as in some other languages, both sets are reflected, with the exception of *-ni* (*nī* patterns with the long forms and must have developed from *nei* by monophthongisation). Rapa Nui is the only language to have *e-* in the short form *era*; *ena* occurs in Tongan as well. However, similar forms occur in Rarotongan (Buse 1963a: 415–416) and Tahitian (pers.obs.), though less overtly. In these languages, the enclitics *na* and *ra* cause lengthening of the preceding vowel, accompanied by stress shift:

<sup>(</sup>i) [te taˈɁata] 'the man' [te taɁaˈtaː ra] 'that man' (Tahitian)

Thus, *na* and *ra* in these languages actually consist of a CV syllable preceded by an unspecified vowel (*Vna*, *Vra*), which means that they are quite similar to Rapa Nui *ena* and *era*, respectively.

#### 4 Closed word classes

demonstrative determiners postnuclear demonstratives deictic locationals demonstrative pronouns proximal *nei, nī nei nei nei* medial *nā ena nā nā* distal *rā era rā rā* neutral *tau/tou/tū, hū ira*

Table 4.17: Classes of demonstratives

### **4.6.2 Neutral demonstrative determiners**

#### **4.6.2.1 The** *t***-demonstrative: form and function**

Rapa Nui has a set of demonstrative determiners of the form *tVV*:

$$(197) \quad \text{tau}^{53} \qquad \text{to} \qquad \text{t}\bar{u}$$

These forms are semantically and syntactically equivalent; they succeed each other in the history of Rapa Nui. In older texts, *tau* is predominant; in some corpora it is the only form in use. *Tou* occurs in both older and newer texts; nowadays, *tū* is used. The sources thus show a gradual vowel assimilation *tau* > *tou* > *tū*. 54

As the three forms are diachronic variants of the same particle, they will be treated as a single "*t-*demonstrative". The *t*-demonstrative is a neutral form, which – unlike other demonstratives – is not differentiated for relative distance.<sup>55</sup> It is always accompanied by one of the following postnominal elements: either a postnominal demonstrative (PND) *nei*, *ena* or *era* or the identity marker *ꞌā* or *ꞌana*, but never both. Of these two options, the PND is by far the most common one.

In combination with a PND, the *t*-demonstrative has anaphoric function: it signals that the entity referred to has been mentioned in the preceding context (and, by impli-

<sup>53</sup> *Tau* is probably related to PEP \**taua* (see Pawley 1966: 60; Green 1985: 12), which, however, only occurs in Tahitic languages. An indication for a relationship between the two is that *taua*, like Rapa Nui *tau*, is an anaphoric determiner which co-occurs with postnominal demonstratives – obligatorily so in Tahitian (Académie Tahitienne 1986: 64–65), optionally in Māori (Bauer, Parker & Evans 1993: 152). Rigo & Vernaudon (2004: 462) consider Tahitian *taua* to consist of the article *te* + a cognate of the demonstrative *ua* which appears in Hawaiian but has no cognates in any other language. They tentatively propose that this *ua* is originally the same morpheme as the perfect aspectual *ua* which occurs in both Tahitian and Hawaiian; however, the latter is a reflex of PPN *\*kua*, while *taua* also occurs in languages which have preserved PPN \**k*, like Māori, Rarotongan and Pa'umotu (Pollex, see Greenhill & Clark 2011).

<sup>54</sup> A similar monophthongisation process may have taken place in Rapa (=Rapa Iti): the definite marker *tō* is probably derived from \**taua*, through a development *taua > tau > tou > tō* (Walworth 2015b: 183).

<sup>55</sup> According to Anderson & Keenan (1985: 280), one-term deictic systems, which do not indicate relative distance, are crosslinguistically very rare. French *ce* is another example, but like the Rapa Nui *t-*demonstrative, it usually goes together with another demonstrative element which does express distance. Notice that the *t-*demonstrative in combination with the identity marker *ꞌā*/*ꞌana* is a true one-term subsystem: in this construction no relative distance is expressed, despite the presence of a demonstrative. In such a case, as Anderson and Keenan suggest, the demonstrative is little different from a definite article.

4.6 Demonstratives

cation, is known to the hearer). In (198), there are three referents: Ure a Ohovehi, the boat and the men. All have been mentioned before, and all are referred to with the same combination of a *t*-demonstrative and a PND.

(198) *He* ntr *tikeꞌa* see *e* ag *tau* dem *kope* person *era,* dist *ko* prom *Ure* Ure *ꞌa* a *Ohovehi,* Ohovehi *tau* dem *vaka* boat *era* dist *o* of *tau* dem *ŋāŋata* men *era.* dist

'That man Ure a Ohovehi saw that boat of those people.' [Blx-3.070]

The use of the *t-*demonstrative with postnominal demonstratives is further discussed in §4.6.3.

In combination with the identity marker *ꞌā/ꞌana* the *t*-demonstrative expresses identity with an entity previously mentioned; this is discussed in §5.9.

As demonstratives are the main anaphoric device to track participants in discourse, they are much more common than English demonstratives. Example (198) would sound unnatural in translation if all the demonstratives were translated by demonstratives.<sup>56</sup>

#### **4.6.2.2 The demonstrative** *hū*

The demonstrative *hū*<sup>57</sup> is always accompanied by a postnominal demonstrative or an identity marker, just like the *t-*demonstrative. It is much less common than the *t-*forms and especially occurs in older texts, but is still in use. Like *tū*, it indicates that the referent has been mentioned before; it may indicate a more pointed deixis: 'just that, precisely that'.

(199) *—Taꞌe* conneg *ko* prom *Reŋa* Renga *Roiti* Roiti *taꞌa* poss.2sg.a *meꞌe* thing *ena.* med *—¿He* pred *aha* what *rā* intens *hū* dem *meꞌe* thing *era?* dist

'—That one is not Renga Roiti. —Then what exactly is it⁈' [Ley-9-56.092–093]

(200) *He* ntr *kī* say *ki* to *te* art *nuꞌu* people *mo* for *oho* go *a* toward *ꞌuta* inland *ꞌana* ident *mo* for *haka* caus *tau* hang *mo* for *uꞌi* look *ꞌatakea* if *ko* prom *hū* dem *ŋā* pl *io* young\_man *era.* dist

'He told the people to go ashore and lie in waiting to see whether it would be those (same) boys.' [R425.011]

<sup>56</sup> See Englert (1978: 21): "El artícula *tou-era* (a veces*tau-era*) es pronombre demostrativo que se usa frequentemente como simple artículo definido." (The article *tou-era* (sometimes *tau-era*) is a demonstratie pronoun which is often used simply as a definite article.)

<sup>57</sup> Etymologically, *hū* is more different from *tū* than its shape may suggest. As *hū* (unlike *tū*) already occurs in older texts, it cannot be derived from *tū* (e.g. by analogy of *te* and *he*). *Hū* may be related to Marquesan *hua*, which likewise serves as an anaphoric article. (Cablitz 2006: 62; Bergmann 1963: 49.) Bergmann also suggests a tentative link to the Hawaiian demonstrative *ua*.

4 Closed word classes

### **4.6.3 Postnominal demonstratives**

The postnominal demonstratives *nei*, *ena* and *era* (henceforth PND) indicate different degrees of distance:


PNDs occur towards the right periphery of the noun phrase (see the chart in §5.1).

As discussed in §4.6.2, PNDs are obligatory when the noun is preceded by a *t*-demonstrative (*tau*/*tou*/*tū*), unless the noun phrase contains the identity marker *ꞌā*/*ꞌana*. PNDs also occur in combination with other determiners: articles as in (201), possessive pronouns as in (202):


PNDs may be used either deictically or anaphorically. As deictic markers they serve to point at something which is visible in the nonlinguistic context. As anaphoric markers they refer to entities in the discourse context: entities which have been mentioned before or which are known by some other means. In practice, the anaphoric use is much more common in discourse.<sup>58</sup>

In the following sections, the PNDs are discussed in turn, starting with the most common form *era*.

#### **4.6.3.1 Distal/neutral** *era*

When *era* is used deictically, it serves to point at something at a distance from both speaker and hearer.


<sup>58</sup> Hooper (2010: 363) notices the same in Tokelauan discourse: situational (=deictic) use only plays a "very minor part" in texts.

4.6 Demonstratives

Much more commonly, *era* is anaphoric. *Era* is by far the most common postnominal demonstrative and the most neutral in sense. In its anaphoric use *era* usually does not have a connotation of distance, but is simply a general-purpose demonstrative.

*Era* is especially common with the *t*-demonstrative determiner (§4.6.2). The combination *tū/tou/tau N era* is the most general device in narrative texts to refer to participants mentioned earlier in the context. This makes its use extremely common in discourse.<sup>59</sup> In the following example, the two main characters of the story, neither of whom is mentioned by name, are referred to as *tau taŋata era* 'that man' and *tau viꞌe era* 'that woman'.

(205) *He* ntr *moe* lie\_down *rō* emph *ꞌavai* certainly *tau* dem *taŋata* man *era.* dist *He* ntr *koromaki* miss *ki* to *tau* dem *viꞌe* woman *era* dist *toꞌo* take *era* dist *e* ag *tōꞌona* poss.3sg.o *matuꞌa.* parent *He* ntr *moe* lie\_down *tau* dem *taŋata* man *era,* dist *kai* neg.pfv *kai.* eat *He* ntr *ꞌōtea,* dawn *he* ntr *pō* night *hakaꞌou,* again *tau* dem *taŋata* man *era,* dist *he* ntr *mate* die *tau* dem *taŋata* man *era,* dist *he* ntr *koromaki* miss *ki* to *tau* dem *viꞌe* woman *era.* dist

'The man slept. He longed for the woman that had been taken (back) by her father. The man slept, he did not eat. Day came, then night again; the man died, that man, out of longing for the woman.' [Mtx-5-02.057-060]

In the following example, two participants (the father and the child) and one object (the child's umbilical cord) are first introduced with the article *te.* The next time they are mentioned, all are marked with *tou/tū N era*.

(206) *He* ntr *poreko* born *te* art *poki* child *o* of *te* art *taŋata* man *e* num *tahi.* one *He* ntr *uŋa* send *mai* hither *te* art *roŋo* message *mo* for *eꞌa* go\_out *atu* away *o* of *te* art *taŋata* man *nei,* prox *mo* for *oho,* go *mo* for *hahaꞌu* tie *i* acc *te* art *pito.* navel *I* pfv *eꞌa* go\_out *era* dist *te* art *taŋata* man *nei,* prox *i* pfv *oho* go *era* dist *ki* to *tou* dem *pito* navel *era* dist *o* of *tū* dem *poki* child *era* dist *o tū taŋata era mo hahaꞌu…*

of dem man dist for tie

'A child was born to a certain man. A message was sent for this (other) man to come, to tie the navel (cord). When man had gone out to tie the navel (cord)<sup>60</sup> of the child of that man…' [Blx-2-1.001-005]

<sup>59</sup> Næss (2004: 81) notices that demonstratives in Pileni (a Polynesian outlier) are "used to an extent which appears quite extraordinary for a language of this family, perhaps for any language". The same is true for Rapa Nui: over the whole text-corpus, *era* occurs almost 15,000 times and is the seventh most common word overall (after the determiners *te* and *he* and a number of prepositions). Given the fact that demonstratives not only serve to indicate spatial deixis but mark definiteness and anaphora as well (functions carried out by definite articles in other languages), their high frequency is not as surprising as it may seem at first sight.

#### 4 Closed word classes

Another determiner-demonstrative combination is *te N era*, with the article *te* instead of a demonstrative determiner. This combination is used to refer to something which is known to both speaker and hearer, whether or not it has been mentioned in the preceding context. This means that *te N era* indicates definiteness: <sup>61</sup> it signals that speaker and hearer are both able to identify the referent of the noun phrase. It is therefore the equivalent of the English (or Spanish) definite article, rather than a demonstrative.

Like *tau/tou/tū N era*, it may be used to refer to participants in a story who have been mentioned before. In (207), *tau poki era* and *te poki era* refer to the same child:

(207) *He* ntr *oŋa* appear *mai* hither *tau* dem *poki* child *era* dist *o* of *tau* dem *taŋata* man *era* dist *ko* prom *Kava* Kava *te* te *Rūruki.* Ruruki *He* ntr *tikera* see *te* art *poki* child *era…* dist 'The child of that man Kava te Ruruki observed him. The child saw it…' [Ley-9-57.035]

(208) *He* ntr *tupu* grow *te* art *poki* child *o* of *te* art *viꞌe,* woman *he* ntr *poreko…* born *He* ntr *hāŋai,* feed *he* ntr *nuinui* big:red *te* art *poki* child *era.* dist

'A woman was with child, it was born. The child was raised and grew up.' [Mtx-7-21.004–005]

*Te N era* may also refer to entities which are generally known, or which are presumed to be present in the context. In the following example, 'the cliffs' refers to the cliffs in general (which all hearers will presumably know to be part of the Rapa Nui coastline); no specific cliff is meant.

(209) *I* pfv *naꞌa* hide *era* dist *a* prop *ꞌOho* Oho *Takatore* Takatore *i* acc *tū* dem *kūpeŋa* net *era,* dist *he* ntr *oho* go *mai* hither *ki* to *te* art *kona* place *ꞌōpata* cliff *era.* dist 'When Oho Takatore had hidden that net, he went to the cliffs (lit. the cliff place).' [R304.110]

*Te N era* may also refer to things which have not been previously mentioned, but which are definite because they are explained in the noun phrase itself: a modifying phrase or relative clause after the noun specifies what the noun refers to. In (210) below, the referent of *te haŋa era* 'the bay' is specified by the genitive phrase *o ꞌAkahaŋa*; in

<sup>60</sup> Lit. 'gone out to the navel to tie'; for this construction, see §11.6.3.

<sup>61</sup> See the discussion of definiteness in section §5.3.3. The development of demonstratives to definite markers may have taken place in Tongan as well: Clark (1974) shows how the "definitive accent" (a stress shift to the final syllable of the noun, marking definiteness) may have derived from a postposed demonstrative *\*aa*.

4.6 Demonstratives

(211) *te haꞌu era* 'the hats' is explained by the relative clause *e aŋa era hai rau toa* 'made with sugarcane leaves'.


In these contexts, where the noun phrase becomes definite by virtue of a modifier, *tū N era* is not (or rarely) used. In other words, where *Det N era* has a unique referent, *tū* is used; where *Det N era* as such does not have a unique referent but needs a modifier to pinpoint its reference, *te* is used.

To summarise:


#### **4.6.3.2 Proximal** *nei*

*Nei* indicates proximity. It is more commonly used with the article *te* than with the demonstrative *tū*. When used deictically, *nei* refers to something close to the speaker:

(212) *Te* art *kona* place *nei* prox *ꞌi* at *te* art *hare* house *nei* prox *mo* for *te* art *poki* child *māꞌaŋa* chick *nei* prox *ꞌāꞌaku.* poss.1sg.a 'This place (here) in this house is for my adopted child.' [R229.271]

The proximity indicated by *nei* may also be temporal: the event takes place close to the time of speaking. This is especially clear when *nei* is used with nouns denoting time.

(213) *ꞌI* at *te* art *hora* time *nei* prox *paꞌi* in\_fact *ku* prf *ŋaro* disappear *ꞌana* cont *rā* intens *mauku.* grass 'Nowadays (lit. 'in this time') that grass has disappeared.' [R106.050]

However, temporal proximity is not necessarily related to the time of speaking. The reference time may also be the time of other events in the same text. In the following example, *te noho iŋa nei* 'this time/epoch' refers to the time when the events in the story happened.

#### 4 Closed word classes

(214) *ꞌI* at *te* art *noho* stay *iŋa* nmlz *nei,* prox *hoꞌi,* indeed *ꞌina* neg *he* pred *mōrī,* light *ꞌina* neg *he* pred *vai…* water 'At this time there was no electricity, no water…' [R539-1.092]

*Nei* also has anaphoric uses. It may refer to something which has been mentioned just before; the referent is close in a textual sense.

(215) *'¡Ka* imp *haka* caus *kore* lack *te* art *kope* person *ena* med *ꞌe* and *ka* imp *haka* caus *eꞌa* go\_out *mai* hither *a* prop *Varavā!'* Barabbas *Te* art *taŋata* man *nei* prox *i* pfv *puru* close *ai* pvp *ꞌo* because\_of *te* art *haka* caus *tumu* origin *i* acc *te* art *ture.* quarrel ' "Away with that man, release Barabbas!" This man had been imprisoned for provoking a riot.' [Luke 23:19]

Unlike other postnominal demonstratives, *nei* is also used cataphorically, pointing forward to what follows. One such cataphoric use is at the beginning of stories: here *nei* is often used to introduce (main) participants.<sup>62</sup> An example is:

(216) *ꞌI* at *ꞌOhovehi* Ohovehi *te* art *noho* stay *iŋa* nmlz *o* of *te* art *ŋā* pl *roe* ant *nei* prox *e* num *rua.* two 'In Ohovehi was the place where these two ants lived.' [R214.001]

This sentence is the beginning of a story about two ants. The use of *nei* signals to the reader that the two ants will be playing an important role in the story that follows.<sup>63</sup> This use of *nei* can be considered as cataphoric: *nei* directs the hearer to look forward to provide more information about the indicated participant.

Another cataphoric use of *nei* is after generic nouns like *meꞌe* 'thing'. Here *nei* signals that more specific information follows:<sup>64</sup>

(217) *Te* art *meꞌe* thing *nei* prox *he* pred *ruku* dive *e* exh *ai* exist *te* art *ŋā* pl *meꞌe* thing *nei:* prox *he* pred *pātia,* harpoon *he* pred *hiꞌo…* glass 'For diving you need the following things: a harpoon, glasses…' [R360.001]

The same use of *nei* (though not in a noun phrase) is found in the expression *pē nei ē*

'like this', which introduces speech or thought (see (237–238) on p. 205).

<sup>62</sup> This use is common in newer stories, but not found at all in older texts.

<sup>63</sup> English has a similar – somewhat informal – use of *this*, to introduce a participant at the start of a story: *'*Yesterday I met this guy…*'*

<sup>64</sup> Again, English provides a parallel use of 'this': 'Listen to this: ….'; 'This is what you need…'

4.6 Demonstratives

#### **4.6.3.3 Medial** *ena*

*Ena* indicates something removed from the speaker, but close to the hearer:

(218) *ꞌIna* neg *koe* 2sg *ko* neg.ipfv *kai* eat *i* acc *te* art *meꞌe* thing *ena* med *o* of *roto* inside *o* of *te* art *kete* basket *ena.* med 'Don't eat those things in that basket (you have there).' [Blx-3.036]

However, the use of *ena* is somewhat limited: while *nei* is regularly used with first person pronouns, *ena* is not used with second person pronouns.

After temporal nouns like *tāpati* 'week' or *matahiti* 'year', *ena* signifies 'next'.

(219) *Matahiti* year *ena* med *he* ntr *hoki* return *a* prop *au* 1sg *ki* to *te* art *hāpī.* learn 'Next year I return to school.' [R210.003]

Here, *ena* signifies a referent which is in the future, one step removed from the time of speaking. To refer to a time one step removed in the past, *ena* is used in combination with the verb *oti* 'finish'. The following example occurs in a newspaper published in May, i.e. it refers to the previous month:

(220) *ꞌI* at *tū* dem *ꞌāvaꞌe* month *oti* finish *ena* med *o* of *Vai Tuꞌu Nui* April *i* pfv *haꞌamata* begin *i* pfv *keri* dig *ai* pvp *o* of *koā* coll *Jo* Jo *Anne…* Anne

'In the past month of April, Jo Anne and the others started to dig…' [R647.106]

### **4.6.4 Demonstrative determiners**

*Nei*, *nī*, *nā* and *rā* are demonstrative determiners indicating relative distance. Like the *t-*demonstrative they exclude the article, but unlike these, they are not accompanied by a postnominal demonstrative (except *nī*).<sup>65</sup> In fact, these demonstratives themselves are very similar in sense to postnominal demonstratives. They are a recent development: demonstrative determiners are rarely found in older texts. It is not unlikely that they developed under Spanish influence: *nei taŋata* 'this man' by analogy of Sp. *este hombre*. 66

As *rā* is the most common (and most neutral) form, it will be discussed first.

<sup>65</sup> Chapin (1974: 8) also mentions a demonstrative *tenei*, supposedly used in Egt-02. However, in Englert (1974), which includes this text, the form in question is printed as *to nei*. The forms *teenei*, *teenaa* and *teeraa*, which are common in Nuclear Polynesian languages (Pawley 1966: 51), do not occur in Rapa Nui (see also Langdon & Tryon 1983: 21), though they may have existed at a prior stage: *tenā* possibly appears in the old chant *e timo te akoako* (Fischer 1994: 426). The fact that the demonstrative determiners *nei*, *nā* and *rā* hardly occur in older texts, suggests that they did not develop from the PEP demonstrative determiners *\*teenei, \*teenaa, \*teeraa* through loss of *tee-*, but are an independent recent development.

<sup>66</sup> This is pointed out by Fischer (2007: 389).

#### 4 Closed word classes

#### **4.6.4.1 Distal/neutral** *rā*

*Rā* is similar in meaning to the postnominal *era* (§4.6.3.1): just like *era* is the neutral postnominal demonstrative, *rā* is the neutral, most common, demonstrative determiner.

*Rā* is used deictically, referring to something which has not been mentioned before, but which is present in the extralinguistic context and hence accessible to both speaker and hearer. It is used in conversation, for example, when pointing out something at a certain distance, or when indicating something on a picture or map:

(221) *ꞌI* at *rā* dist *hare* house *a* prop *mātou* 1pl.excl *e* ipfv *noho* stay *ena.* med '(discussing a photograph:) In that house we lived'. [R416.961]

Like *tū N era*, *rā* is also used anaphorically. In the following examples, the noun in question has been introduced in the preceding context.


This means that *tū N era* and *rā* are similar in function. Even so, there are differences between the two.

First of all, *rā* is somewhat more informal than *tū N era*. It tends to be more common in conversation and direct speech, while *tū N era* occurs more commonly in narrative texts.

Secondly, there are also collocational differences: *rā N* is especially common before words denoting a moment or period of time, like *hora* 'time, moment, hour', *mahana* 'day' and *noho iŋa* 'period, epoch', while *tū N era* is found more often with concrete nouns like *hare* 'house' and *taŋata* 'man, person'.

In the third place, the relation between *rā* and *tū N era* also has a diachronic aspect. *Rā* is extremely rare in older texts. The demonstrative *rā* does occur in these texts, but almost always as a locational (§4.6.5): *ꞌi rā* 'over there'.

In newer texts (most of which date from the 1980s), *rā* is common, but *tū* still occurs about twice as often. However, in the Bible translation – the largest part of which was done, or at least thoroughly revised, after 2000 – *rā* is about 50% more frequent than *tū*. In the Bible translation, *rā* is commonly used to track participants in discourse.

Only when the noun phrase contains a modifier (an adjective, a possessor or a relative clause), *tū N era* continues to be the default choice, even in the Bible translation:

(224) *tū* dem *taŋata* man *matapō* blind *era* dist 'that blind man' [John 9:6]

4.6 Demonstratives

Taking these facts together, we arrive at the following explanation: *rā* was originally a deictic locational (§4.6.5), used to point at things and locations: 'there, over there'. *Tau/tou* had a different role: tracking participants in discourse, i.e. referring to entities mentioned earlier in the context.

When *rā* started to be used as a prenominal demonstrative, it was initially with the same deictic role it already had, pointing to for example things and locations ('that house there', 'that place over there'), and points in time ('on that day'). Gradually it acquired a participant-tracking role as well, but until recently this role has been predominantly fulfilled by *tau/tou/tū*. This use of *rā* is becoming more and more frequent, to the point where it is now more common than *tū/tou*. Only in complex noun phrases is *tū* still preferred.

#### **4.6.4.2 Proximal** *nei*

Prenominal *nei* is similar in meaning to postnominal *nei* (§4.6.3): it indicates proximity in time, location or discourse. It may refer to something near the speaker as in (225), something just mentioned as in (226), or to a time close to the time of the preceding discourse as in (227):


'From other countries, to the north of New Zealand, they came to this island (= New Zealand).' [R346.012]

(227) *ꞌE* and *takoꞌa* also *paꞌi,* in\_fact *nei* prox *noho* stay *iŋa* nmlz *kai* neg.pfv *rahi* much *mai* hither *ꞌā* cont *te* art *meꞌe* thing *he* pred *ꞌaurī* iron *ki* to *nei.* prox

'And also, at this time (the period just mentioned), there was not much iron here.' [R353.006]

Pre- and postnominal *nei* are not completely identical in function: while postnominal *nei* may be cataphoric, referring to something which has not been mentioned yet, prenominal *nei* always refers something which has been mentioned before.

#### **4.6.4.3 Proximal** *nī*

*Nī* is a relatively rare demonstrative, which is not found in older texts. Its function is similar to *nei;* it must have arisen from *nei* by vowel assimilation. That this only

#### 4 Closed word classes

happened prenominally may be because the prenominal position is phonologically less prominent: unlike postnominal *nei*, it never receives phrase stress.

*Nī* often refers to something which has been recently mentioned. In the following example, *nī taŋata* refers back to *e te taŋata e tahi* in the previous sentence.

(228) *Pē* like *ira* ana *i* pfv *hīmene* sing *ai* pvp *e* ag *te* art *taŋata* man *e* num *tahi…* one *i* acc *te* art *hīmene* song *e* num *tahi.* one *Ko* prom *Toꞌo* To'o *Raŋi* Rangi *te* art *ꞌīŋoa* name *o* of *nī* prox *taŋata.* man 'In that way one man… sang a song. To'o Rangi was the name of this man.' [R539-1.127–128]

Unlike prenominal *nei*, *nī* can be accompanied by a postnominal demonstrative. Interestingly, the latter is not necessarily *nei*:


*Nī* tends to be used for referents which are not central participants in the discourse: minor participants, objects (*iate* above), places *(ꞌōpata* above), time words like *mahana* 'day' and *hora* 'time'.

#### **4.6.4.4 Medial** *nā*

*Nā* is occasionally used as a prenominal demonstrative. It is similar in meaning to postnominal *ena*, referring to something not close to the speaker, but close to the hearer. Therefore it typically appears in direct speech, as in the following example:

(231) *Ka* imp *toꞌo* take *mai* hither *nā* med *matā* obsidian *ka* imp *vero* throw *ki* to *rote* inside\_art *haha.* mouth 'Take that obsidian spearhead and throw it into his mouth'. [R304.020]

### **4.6.5 Deictic locationals**

#### **4.6.5.1** *Nei***,** *nā* **and** *rā* **as deictic locationals**

Deictic locationals have the same form as demonstrative determiners (§4.6.4): *nei*, *nā* and *rā*. As locationals, they are a nucleus in their own right, rather than modifiers of a head noun. Just like all locationals (§3.6.1), they can be preceded by a preposition, but not by a determiner. They usually have a deictic function. Some examples:


Preceded by *pē* 'like', the deictic locationals tend to be used anaphorically rather than deictically. *Pē rā* is similar in function to *pē ira* (see the next section): it refers back to a state of affairs mentioned before, 'like that, in the same manner'. An example:

(235) *Pē* like *rā* dist *ꞌā* ident *e* ipfv *ꞌamo* carry *mai* hither *era* dist *ꞌi* at *te* art *mahana.* day 'In the same way (as described before) he carried (food) every day.' [R372.139]

*Pē nei* is used anaphorically as well. As *nei* expresses proximity, anaphoric *pē nei* indicates what can be called discourse proximity: it refers back to something which has just been mentioned.

(236) *Pē* like *nei* prox *a* prop *Nueva* New *Zelántia* Zealand *i* pfv *noho* stay *ai* pvp *mo* for *te* the *paratane.* British 'In this way (just described) New Zealand came to belong to the British.' [R346.022]

Especially common is *pē nei ē*, which introduces the content of a speech or thought, or a piece of knowledge.


4 Closed word classes

#### **4.6.5.2 The anaphoric locational** *ira*

*Ira*<sup>67</sup> is a multipurpose anaphor (hence its gloss ana). Whereas personal pronouns serve as anaphors referring back to animate referents mentioned in the preceding context, *ira* refers back to locations or situations.<sup>68</sup>

When preceded by a locative preposition, *ira* refers to a location which has been mentioned before: 'that place, there'. In (239) *ira* refers back to 'home' in the previous clause, in (240) to 'his country':

(239) *I* pfv *tuꞌu* arrive *hakaꞌou* again *era* dist *a* prop *Makoꞌi* Mako'i *ki* to *te* art *kona* place *hare* house *era,* dist *ꞌi* at *ira* ana *a* prop *Paepae.* Paepae 'When Mako'i arrived home again, Paepae was there.' [R214.071]

(240) *He* ntr *tuꞌu* arrive *ki* to *tōꞌona* poss.3sg.o *kāiŋa* homeland *ko* prom *Maꞌuŋa* Mount *Terevaka.* Terevaka *I* pfv *tuꞌu* arrive *era* dist *ki* to *ira…* ana

'He went to his own place, mount Terevaka. When he arrived there…' [R314.159–160]

One of the contexts in which locational *ira* can be used, is in a relative clause with locative relativisation (see (101–102) on p. 532).

Preceded by other than locative prepositions, *ira* refers to a situation, a state of affairs which has been mentioned in an earlier clause. This happens with *mo ira* 'therefore, for that purpose', *ꞌo ira* 'because of that',<sup>69</sup> and the very common *pē ira* 'like that, thus':

(241) *Mo* for *ira* ana *te* art *puka* book *nei* prox *i* pfv *aŋa* make *ai.* pvp 'Therefore I have made this book.' [R531.014]

(i) *He* ntr *māta ꞌita ꞌi* observe *ararua* the\_two *i* acc *te* art *ŋā* pl *mōai* statue *era* dist *o* of *ira.* ana 'The two of them admired the statue there (lit. the statue of there)' [R478.044]

<sup>67</sup> *Ira* does not occur in any other Polynesian language. However, most other EP languages have a locational anaphor *reira/leila* ('there', referring to a place mentioned before); Rapa Nui *ira* may be a truncated reflex of the same form. This would mean that *\*leila* is not a PCE innovation as suggested by Green (1985: 12) and Pollex (Greenhill & Clark 2011), but a PEP innovation with subsequent shortening in Rapa Nui.

<sup>68</sup> Together, personal pronouns and *ira* cover a large part of the field of possible referents for anaphora. For other referents, no anaphor is available, however: 1) inanimates. To refer back to an inanimate, the generalpurpose noun *me ꞌe* 'thing' can be used: *te me ꞌe era*, lit. 'that thing'. 2) time. *ꞌI ira* can only refer to place, not to time. To refer back to a moment in time, phrases like *ꞌi te hora era* 'at that time' are used.

<sup>69</sup> *ꞌO ira* (with the reason preposition *ꞌo*, §4.7.3) should not be confused with *o ira* 'of there' (with possessive *o*), in which *ira* has a locational sense:

4.6 Demonstratives


### **4.6.6 Demonstrative pronouns**

Demonstrative pronouns are relatively rare. In order to refer to a situation in general ('this', 'that'), the dummy noun *meꞌe* is often used:

	- Nui

'This (the practices just described) was something very good for the life of the people of Rapa Nui.' [R231.314]

The demonstratives *nei*, *nā* and *rā* are also used pronominally, but only as subject of a classifying or identifying clause (§9.2.1–9.2.2). In these constructions, the demonstrative is a constituent by itself; unlike personal pronouns, it is never preceded by the proper article or *ko*, or followed by modifying particles. The constituent order is always predicate—subject. Two examples:


Just like demonstratives in the noun phrase, the demonstrative pronoun can be used either deictically (pointing at something in the non-linguistic context) or anaphorically or cataphorically (pointing back or forward to something mentioned in the text).

Certain postnominal elements belonging to the predicate noun phrase occur after the subject: genitives as in (247), relative clauses as in (248).

(247) *He* pred *toru* three *eꞌa* go\_out *iŋa* nmlz *atu* away *nei* prox *o* of *Tāpura* Tapura *Reꞌo.* Re'o 'This is the third issue of (the newspaper) Tapura Re'o.' [R649.001]

#### 4 Closed word classes

(248) *Famiria* family *hope ꞌa* last *rā* dist *oho* go *mai* hither *mai* from *kampō,* countryside *mai* from *ꞌAnakena.* Anakena 'That was the last family who came from the countryside, from Anakena.' [R413.889]

Even though the demonstratives in these examples may seem to be postnominal particles which are part of the predicate noun phrase, in reality they are pronominal, i.e. constituents in their own right. This is shown by the following evidence:


This means that the nominal predicate in (247–249) is split in two, and interrupted by the subject. Split predicates also occur with other pronominal subjects (§9.2.5).

### **4.7 Prepositions**

### **4.7.1 Introduction**

Prepositions express a semantic relationship between a noun phrase and the rest of the clause. Rapa Nui has a variety of prepositions, some of which (like *ꞌi* and *ki*) have a wide range of uses, while others are more narrowly defined. They also serve to mark case, especially the prepositions *i* (direct object) and *e* (agentive subject).

Syntactically, prepositions are characterised by the fact that they are followed by a noun phrase. When the preposition is followed by a common noun phrase, this noun phrase must contain a *t*-determiner (§5.3.2.1). Two prepositions show different behaviour, however:


With a proper noun or pronoun complement, prepositions ending in *i* (with the exception of *hai*) are followed by the proper article *a*, while others are directly followed by the (pro)noun (§5.13.2.1).

4.7 Prepositions

Most prepositions can be followed by locationals (§3.6.2.1); locationals immediately follow the preposition, without a determiner.

These patterns are summarised in Table 4.18.



Group 1 includes *i* 'object marker', *i/ꞌi* 'locative', *ki* 'to' and *mai* 'from'. Group 2 includes *a* 'by, along', *pe* 'toward', *e* 'agent marker', *ꞌo* 'because of', *o/ꞌa* 'possessive', *to/ta* 'possessive', *mo/ma* 'benefactive', *ko* 'prominence marker' and *ꞌai* 'there in/at'. Group 3 only includes *hai*, group 4 only includes *pa*/*pē*.

In the following subsections, prepositions are discussed individually, except the Agent marker *e*, the accusative marker *i* (§8.2–8.4), and the possessive prepositions *o* and *ꞌa* (§6.2–6.2.4). §4.7.2–4.7.7 deal with prepositions which are primarily locative, such as *ꞌi* and *ki*. The causal preposition *ꞌo* will be discussed together with *ꞌi* in §4.7.3, as the two are similar in function.

§4.7.8–4.7.10 discuss prepositions with other than locative functions, such as benefactive and instrumental. §4.7.11 discusses the rare preposition *ꞌai*. Finally, §4.7.12 deals with the prominence marker *ko*, which is different in function from other prepositions, but which is nevertheless a preposition syntactically.

### **4.7.2 The preposition** *ꞌi***/***i* **'in, at, on'**

The preposition *ꞌi/i* 'in, at, on' is a reflex of PPN *\*ꞌi* (Pollex, see Greenhill & Clark 2011) or *\*i* (Clark 1976: 41) – due to the instable character of glottals in particles (Clark 1976: 22), it is impossible to tell with certainty if the preposition had a glottal in PPN.

In most Polynesian languages this preposition has a wide range of functions.<sup>71</sup> In the accepted Rapa Nui orthography (§1.4.4), this preposition is written either *ꞌi* and *i*, depending on its function: certain uses of this preposition are written with glottal, others without. The inclusion or non-inclusion of glottals in particles is largely based on whether the particle occurs mainly at the start or in the middle of prosodic units (§2.2.5). This means that functions of *i/ꞌi* which tend to occur phrase-initially are written with

<sup>70</sup> In fact, animate complements of *hai* are rare. They are found e.g. in the Bible translation: *hai Ietū* 'with/by Jesus', *hai ia* 'with/by him'.

<sup>71</sup> Chapin (1978: 428) mentions "the extreme polyfunctionality of Polynesian prepositions, and of *i* in particular".

#### 4 Closed word classes

glottal, while functions mainly occurring in the middle of phrases, or at the start of phrases prosodically connected to the preceding context, are written without glottal. As a result, the preposition in a locative sense is written *ꞌi*, while the preposition occurring after locationals is *i*.

In the following sections, the uses of *i*/*ꞌi* are discussed. Because *ꞌi/i* is largely used in a locative or temporal sense, it is glossed 'at'. The causal use of *ꞌi* is discussed in §4.7.3, together with the preposition *ꞌo*, which is similar in function.

#### **4.7.2.1 Locative** *ꞌi*

*ꞌI* expresses stationary location: 'in, on, at'. In this sense it is often followed by locationals (§3.6.2). Here are examples where it is directly followed by a noun phrase: in (250–251) in a spatial sense, in (252) in a temporal sense.


Temporal *ꞌi* may be followed by a nominalised verb, making the *ꞌi*-marked constituent similar to a temporal clause.

(253) *ꞌI* at *te* art *kī* say *nō* just *o* of *Puakiva* Puakiva *ki* to *a* prop *Pea* Pea *i* acc *tāꞌana* poss.3sg.a *vānaŋa,* word *kai* neg.pfv *haka* caus *mou* silent *e* ag *Pea.* Pea 'When Puakiva said her words to Pea, Pea didn't silence her.' [R229.489]

In comparatives, *ꞌi* marks the quality with respect to which the comparison is made (§3.5.2.1).

#### **4.7.2.2 General-purpose** *i*

The preposition *i* serves as a general-purpose oblique marker. It is used to mark noun phrases which are in some way related to the action. Generally speaking, the *i*-marked constituent expresses a participant with respect to whom the event takes place; this constituent can be characterised as the locus of the event.

In some cases this noun phrase has a possessive sense:

4.7 Prepositions


Possessive *i* is also found in proprietary clauses (see (89–90) on p. 470).

With adjectives, *i* expresses the possessor of a certain quality, i.e. the entity where the quality is located. Examples of this are the *ko te X* exclamative construction (see (82–83) on p. 492) and cases such as the following:

(256) *Meꞌe* thing *ꞌaroha* pity *i* at *tū* dem *nanue\_para* kind\_of\_fish *era* dist *ana* irr *ai* exist *ko* prf *ravaꞌa* obtain *ꞌana* cont *e* ag *te* art *nuꞌu* people *hī* to\_fish *ika.* fish

'Poor *nanue para* fish (lit. A pity *i* that *nanue para*) if it is caught by fishermen' [R301.320]

In other cases the sense of *i* is hard to define more precisely; however, it is clear that the *i*-marked NP is involved in the action in some way; the event takes place with respect to the participant mentioned.


Two other uses of *i* are discussed elsewhere:


4 Closed word classes

### **4.7.3 Causes and reasons:** *ꞌi* **and** *ꞌo*

Causes and reasons may be expressed by a verbal clause (§11.6.4). More commonly, however, they are expressed by a noun phrase marked with either *ꞌi* or *ꞌo*. This noun phrase often contains a nominalised verb or an adjective.

*ꞌI* is used to express causes. These causes can be events or states as in (259–260), but also non-human entities as in (261–262). Cf. the discussion on agentive *i* in §8.6.4.7.


Causes are also expressed with the preposition *ꞌo*, but there is a difference. *ꞌI* is used in situations where cause and effect are closely linked, i.e. for direct causes which automatically lead to a certain effect. In (260), for example, laughter is not only the cause of falling down, but also accompanies the falling down: 'He fell while laughing, he fell down with laughter'. Similarly, in (259), hunger it not only the cause of death, but hunger and death go together. In some cases – such as (260) – cause and effect are so closely linked, that the *ꞌi*-marked constituent is similar to a circumstantial clause. *ꞌO*, on the other hand, is used in situations where cause and effect are less closely linked. Here are a few examples:


4.7 Prepositions

*ꞌO* is often used to express reasons, i.e. situations where cause and effect are mediated by a volitional decision:

(265) *Hora* time *kai,* eat *ꞌina* neg *he* pred *haraoa* bread *māꞌau* ben.2sg.a *ꞌo* because\_of *tuꞌu* poss.2sg.o *toke* steal *i* acc *te* art *haraoa* bread *o* of *te* art *poki* child *era.* dist 'At dinnertime, there is no bread for you, because you stole the bread of that

child.' [R245.048]

(266) *ꞌIna* neg *paꞌi* in\_fact *o* of *māua* 1du.excl *kona* place *mo* for *noho.* stay *ꞌO* because\_of *ira* ana *au* 1sg *i* pfv *iri* ascend *mai* hither *nei* prox *ki* to *a* prop *koe.* 2sg 'We don't have a place to stay. Therefore I have come to you.' [R229.210]

### **4.7.4 The preposition** *ki* **'to'**

*Ki* (< PPN *\*ki*) indicates movement in the direction of a goal. It is often found with verbs of movement:


*Ki* is often followed by a locational indicating the direction in which the movement takes place (see e.g. (134) on p. 124).

*Ki* is used when the referent makes a movement touching the endpoint, for example with the verb *tuꞌu* 'arrive':


*Ki* is used when a referent is oriented towards an object, even when no movement towards this object is involved: in (271) the tree bark is exposed to the sun, but not moved towards the sun.

#### 4 Closed word classes

(271) *Ki* when *oti* finish *he* ntr *toꞌo* take *mai,* hither *he* ntr *tauaki* dry\_in\_sun *ki* to *te* art *raꞌā,* sun *he* ntr *haka* caus *pakapaka.* dry:red 'After that they take (the tree bark) and put it in the sun to dry.' [Ley-5-04.009]

In a temporal sense, *ki* indicates the end of a stretch of time: 'until', as in (272).<sup>72</sup> In this sense, *ki* may be reinforced by *ꞌātā* (< Sp. *hasta* 'until') as in (273).


'From that time until now, not one Rapa Nui left (the island) anymore.' [R303.211]

*Ki* has a wide range of metaphorical extensions; it is the default preposition for semantic roles like Recipient, Beneficiary and Goal (§8.8.2) as in (274). *Ki* is used to indicate an opinion or point of view: 'according to', as in (275). In addition, *ki* marks the object of middle verbs (§8.6.4.2).


Finally, in comparative constructions, *ki* marks the standard of comparison (§3.5.2.1).

### **4.7.5 The preposition** *mai* **'from'**

*Mai* indicates a spatial or temporal point of origin:

<sup>72</sup> The preverbal marker *ki* has the same function (see examples (197–198) on p. 553).

4.7 Prepositions


'From this day on, Mum didn't send Tiare anymore.' [R179.046]

When *mai* is followed by a proper noun or pronoun, the proper article *a* is used (as with *ki* and *i/ꞌi*); however, the preposition *i* must be added between *mai* and the proper article, as shown in the following example:

(278) *Ararua* the\_two *nō* just *pāꞌeŋa* side *e* ipfv *tuꞌu* arrive *mai* hither *era,* dist *mai* from *Tahiti* Tahiti *ꞌe* and *mai* from *i* at *a* prop *Tire.* Chile 'The two sides came, from Tahiti and from Chile.' [R539-2.221]

The use of *mai* as a preverbal marker in subordinate clauses is discussed in §11.5.5.

### **4.7.6 The preposition** *pe* **'toward'**

The preposition *pe* indicates a general direction or orientation.<sup>73</sup> Its function is similar to *ki*, but it is not goal-oriented: to go *ki X* implies that one intends to arrive at X; *pe X* does not have this implication.


In a temporal sense, *pe* is used with the locational *muꞌa* 'front' to refer to a period of time in the future, or posterior to a time of reference (see (154–155) on p. 128).

*Pe* may also indicate an approximate location as in (282), or an approximate time as in (283):

<sup>73</sup> This preposition does not occur in other languages. It may be derived from *pē* 'like', but its syntax is different: while *pē* 'like' is followed by the predicate marker *he*, *pe* 'towards' is followed by a *t*-determiner, like most prepositions.

4 Closed word classes


### **4.7.7 The preposition** *a* **'along; towards'**

Unlike other locative prepositions, *a* is used mainly in a spatial sense, rarely in temporal expressions.<sup>74</sup>

The preposition *a* may indicate a spatial relation which is neither stationary (*ꞌi*), nor involves a movement towards (*ki*) or away from (*mai*) an object. It is used when one object moves with respect to another object in some other way: by, along or via the other object.


It is also used when a part of something is singled out as the location where an event takes place.


<sup>74</sup> The preposition *a* (different from possessive *a* or *ꞌa*) occurs in a few languages as a locative preposition (Pollex, see Greenhill & Clark 2011) and is reconstructed as PEP \**aa*.

4.7 Prepositions

*A* often indicates a general direction; this direction is expressed by a locational (such as *muꞌa* in (288)), but there is no second referent involved with respect to which this movement takes place.


*A* indicating a direction is similar to *pe* 'toward' (§4.7.6 above.) A difference between the two is, that *a* is far more common with locationals than *pe*. Another difference is, that some expressions with *pe* have a temporal rather than a spatial sense, whereas *a* is usually spatial.

Certain combinations of *a* followed by a locational have a lexicalised meaning: *a raro*, *a vāeŋa* (§3.6.2.3) and *a tuꞌa* (§3.6.2.4).

### **4.7.8 The benefactive prepositions** *mo* **and** *mā*

The benefactive prepositions *mo* and *mā* express benefactive relations in a broad sense; they are used in situations where an event or object is destined for or aimed at the participant. This pair of prepositions displays the *o*/*a* distinction between two classes of possessives (§6.3.2). This distinction is only made with proper nouns and singular pronouns: with proper nouns either *mā* or *mo* is used; singular benefactive pronouns start with *māꞌa-* or *mōꞌo-* (§4.2.3). With common nouns and with plural pronouns, *mo* is used in all situations.

Regarding the etymology of these prepositions, the PPN forms are *\*moꞌo*, *\*maꞌa*. <sup>75</sup> In Rapa Nui the glottal is still present in the singular pronouns *māꞌaku* etc.; the lengthening in these forms is the result of a general tendency to lengthen the first vowel of threesyllable words. In the prepositions as separate words, the glottal has disappeared.<sup>76</sup>

A benefactive relation is in fact a possessive relationship in which a possessee is destined/intended for the possessor.<sup>77</sup> Whether *mo* or *mā* is used, depends on the relationship between the prospective possessor and possessee, as discussed in §6.3.2: *mā* is used when the possessor has control, authority or responsibility over the possessee, *mo* in all other cases. Thus, *mā* is used for possessions over which the possessor has control, as in (290). *Mo* is used for means of transport as in (291).

<sup>75</sup> The original form of both particles, with glottal, appears in other languages that preserved the PPN glottal: East Uvean, Rennell and Tongan (Pollex, see Greenhill & Clark 2011).

<sup>76</sup> See Wilson (1985) on the loss of the glottal in *t*-possessives and benefactives. For benefactives, he uses the term *irrealis*.

<sup>77</sup> In Polynesian linguistics, these forms are sometimes characterised as "irrealis possessives" (see e.g. Clark 2000a; Wilson 1982: 48): they indicate not-yet realised possession, in contrast to the "realis possessives" starting with *n-* or Ø (see Footnote 9 on p. 290). The likely origin of the *m*-forms is an irrealis marker *m-* (Clark 1976: 115).

4 Closed word classes


With certain verbs, possessive *mo/mā* may express a Goal or Recipient, indicating that the object of the verb is destined for this participant; this is discussed in §8.8.2.

Apart from the uses discussed so far, both *mo* and *mā* have uses of their own. *Mo* may indicate the person towards whom an action or attitude is directed. This happens for example with the verbs *riri* 'be angry' and *ꞌaroha* 'be sorry':


*Mo* may also indicate a participant from whose perspective the event expressed in the clause is true: 'for X, as far as X is concerned'.<sup>78</sup> This use may have been influenced by Spanish *para*. In (294), the things described in the preceding context are news, not necessary for everyone, but for the person mentioned: as far as he is concerned, they are news. In (295), the clause expresses a point of view which is true for the person expressed with *mo*:


*Mā* marks the Agent in the imperfective actor-emphatic construction (§8.6.3):

(296) *Māꞌaku* ben.1sg.a *ꞌā* ident *a* prop *koe* 2sg *e* ipfv *hāpaꞌo* care\_for *atu.* away 'I will take care of you myself.' [R310.067]

<sup>78</sup> Cf. the use of *ki* to express a point of view (§4.7.4).

4.7 Prepositions

### **4.7.9 The preposition** *pa/pē* **'like'**

*Pē* is an equative preposition: it serves to compare two entities, expressing that one resembles the other.<sup>79</sup> Equative constructions are discussed in §3.5.2.3; in this section other syntactic and semantic particularities of *pē* will be discussed.

First of all, *pē* is usually followed by the predicate marker *he*, <sup>80</sup> not only when the compared entity is generic as in (297), but also when it is a single, identifiable entity as in (298):


Before *he*, *pē* is often dissimilated to *pa*. The choice between *pē* and *pa* is free; certain speakers favour one over the other.

(299) *ꞌArero* tongue *nei* prox *pa* like *he* pred *ꞌarero* tongue *rapa* Rapa *nui* Nui *ꞌā.* ident 'This language is like the Rapa Nui language.' [R231.272]

Occasionally *pē* is followed by a *t-*determiner as in (300), or a proper noun or pronoun as in (301):


As most of the examples above show, the comparison may be reinforced by the identity particle *ꞌā* (§5.9).

In modern Rapa Nui, *pē* also expresses the category to which someone belongs. In (302) below, *pē he ꞌōtare* does not mean that the speaker resembles an orphan, but that he is an orphan. This usage may be influenced by Spanish *como*.

<sup>79</sup> *Pē* (< PPN \**pee* 'like') occurs in Hawaiian and Māori, but only or mainly as a bound root, followed by a demonstrative (*pēnei*). It is more common in non-EP languages.

<sup>80</sup> Interestingly, the same is true for the preposition *me* 'like' in Hawaiian, Marquesan and Māori (§5.3.3).

4 Closed word classes

(302) *ꞌIna* neg *ōꞌoku* poss.1sg.o *matuꞌa,* parent *ꞌo* because\_of *ira* ana *a* prop *au* 1sg *e* ipfv *noho* stay *nei* prox *pē* like *he* pred *ꞌōtare.* orphan 'I don't have parents, therefore I live as an orphan.' [R214.013]

### **4.7.10 The instrumental preposition** *hai*

*Hai*<sup>81</sup> is an instrumental preposition, indicating the means or tool with which something is done: 'with, using, by means of':

(303) *He* ntr *puꞌapuꞌa* beat:red *hai* ins *pāoa;* club *he* ntr *mate.* die 'They beat her with a club and she died.' [Egt-01.082]

As discussed in §4.7.1, *hai* is not followed by a determiner, but by a bare noun. This correlates with the meaning of *hai*, which tends to occur with non-specific referents as in (303) above. Occasionally, however, *hai* is followed by pronouns or proper nouns as in (304), or by definite nouns (preceded by a demonstrative) as in (305):


The semantic range of *hai* is large. It may indicate the instrument or material with which an action is done, as in (303) above and the following examples:


<sup>81</sup> This preposition is not found in any other language. It may have developed from PPN *\*fai*, which occurs in several languages as a verb or prefix meaning 'have, possess'. Reflexes of PPN *\*fai* occur in many non-EP languages; the only EP language in which it occurs, is Māori (Pollex, see Greenhill & Clark 2011). The fact that Rapa Nui *hai* is followed by a bare noun suggests that it originated from a prefix *\*fai-* (itself related to the root *\*fai* just mentioned) rather than a full word. As a prefix, it occurs for example in Nukuoro, where *hai-* is – among other things – prefixed to nouns to form derived verbs: *hai hegau* 'do work' = 'to work'; *hai bodu* 'do spouse' = 'to marry' (cf. Carroll & Soulik 1973: 628). It would be a relatively small step for such a prefix to develop into a preposition taking a bare noun complement.

4.7 Prepositions

*Hai* may mark various kinds of noun phrases which are in some way instrumental to the action, such as the price paid as in (308), or the language spoken as in (309).


As *hai* expresses the means by which something happens, it may indicate a resource. Used in a more abstract way, it indicates a reason or motive: 'because of, on account of, thanks to'.


The NP marked with *hai* may also be a resource which is needed but not found yet. This sense is found with verbs of asking or searching as in (312), but also in other contexts as in (313):


Finally, *hai* may mark Patient arguments (§8.6.4.3), especially when their role is similar to Instruments.

### **4.7.11 The deictic preposition** *ꞌai*

*ꞌAi* is a deictic particle (§4.5.4.1.2). Occasionally it is used as a preposition to point at something which is at a certain distance: 'there at/in/on…'. Like other prepositions, it may be followed by locationals as in (314–315) or nouns as in (316):

4 Closed word classes


This preposition may be a contraction of the deictic particle *ꞌai* + the preposition *ꞌi*. (The glottal in *ꞌi* is not pronounced when it is not preceded by a prosodic boundary, see §2.2.5.)

### **4.7.12 The prominence marker** *ko*

The prominence marker *ko* precedes common nouns, proper nouns and pronouns.<sup>82</sup> Even though it does not mark grammatical or semantic relations in the same way as other prepositions do, it is a preposition syntactically:


*Ko* has many different uses, which can be summarised under the heading of prominence: *ko* signals that the noun phrase is in some way prominent within the context. *Ko* has two main functions:

In the first place, it marks prominent topics in verbal clauses (§8.6.2.1):

(317) *Ko* prom *ia* 3sg *i* pfv *eke* climb *ki* to *tuꞌa* back *o* of *tū* dem *hoi* horse *era.* dist

> '(He put the child on his horse, at the front.) He (himself) mounted on the back.' [R399.046]

Secondly, it marks predicates in identifying clauses (§9.2.2):

(318) *Te* art *kona* place *hope ꞌa* last *o* of *te* art *nehenehe* beautiful *ko* prom *ꞌAnakena.* Anakena 'The most beautiful place (of the island) is Anakena.' [R350.013]

<sup>82</sup> Rapa Nui also has two other particles *ko*, which should not be confused with the prominence marker: the negation *(e) ko* (§10.5.4) and the perfect marker *ko/ku* (§7.2.7).

4.7 Prepositions

As a nominal predicate marker, *ko* also marks noun phrases in focus in cleft constructions (§9.2.6):

(319) *Ko* prom *te* art *nūnaꞌa* group *era* dist *ꞌa* of.a *ꞌŌrare* Orare *te* art *nūnaꞌa* group *i* pfv *rē.* win 'Orare's group was the group that won.' [R539-3.313]

Bu *ko* has a number of other uses as well: It marks non-topicalised verbal arguments (§8.6.4.5):

(320) *He* ntr *poreko* born *ko* prom *te* art *heke* octopus *ꞌAkaverio.* Akaverio 'The octopus Akaverio was born.' [Mtx-7-14.003]

It occurs in comitative constructions (§8.10):

(321) *He* ntr *noho* stay *Rano* Rano *rāua* 3pl *ko* prom *tāꞌana* poss.3sg.a *poki,* child *ko* prom *te* art *viꞌe.* woman 'Rano lived with his child and his wife.' [Mtx-7-18.001]

It marks noun phrases in apposition (§5.12):

(322) *He* ntr *oho* go *mai* hither *era* dist *te* art *ꞌariki* king *ko* prom *Hotu* Hotu *Matuꞌa.* Matu'a 'King Hotu Matu'a came.' [Mtx-2-02.043]

It occurs in the interrogative pronoun *ko ai* 'who' (§10.3.2.1):

(323) *¿Ko* prom *ai* who *koe?* 2sg 'Who are you?' [R304.097]

It occurs in exclamative clauses (§10.4.2):

(324) *¡Ko* prom *te* art *manu* animal *hope ꞌa* last *o* of *te* art *tau!* pretty 'What an extremely pretty animal!' [R345.072]

Finally, *ko te* + verbal noun expresses continuity of action (§3.2.3.1.1):

(325) *Ko* prom *te* art *kimi* search *ko* prom *te* art *ohu* shout *a* prop *nua.* Mum 'Mum kept searching and shouting.' [R236.082]

In the following subsections, only those uses of *ko* are discussed which do not have a place elsewhere in this grammar. This is followed by a general discussion on the nature of *ko*.

#### 4 Closed word classes

#### **4.7.12.1** *Ko* **in lists and in isolation**

*Ko* is used to mark items in a list. These items may be proper nouns or common nouns with definite reference. The list may be isolated from the syntactic context as in (326), but it may also have a syntactic role in the clause: in (327) the noun phrases introduced by *ko* are direct object, yet they are marked with *ko* rather than the accusative marker *i*. 83


*Ko* also marks noun phrases used in isolation, i.e. without a syntactic context. In a running text, examples of isolated noun phrases are hard to detect, as a noun phrase which seems to be isolated, may actually be the predicate of a nominal clause with implied subject (see (16–17) on p. 455). Clearer examples of isolated noun phrases are found in titles of stories and other texts. The following examples show that isolated pronouns and proper nouns are marked *ko*, while common nouns in isolation are marked with either *ko* or *he* (§5.3.4.1).


#### **4.7.12.2** *Ko* **as a locative preposition**

Very occasionally, *ko* is used as a preposition with a locative sense. This usage only occurs before locationals. In modern Rapa Nui, it indicates immediacy: something is in a location without delay, in a flash.

<sup>83</sup> Common nouns in lists may also be marked with *he* (§5.3.4.1).

4.7 Prepositions

(330) *He* ntr *tuꞌu* arrive *ki* to *ꞌApina,* Apina *ko* prom *raro* below *te* art *rūꞌau* old\_woman *nei,* prox *ko* prom *roto* inside *i* at *te* art *hare,* house *ki* to *rote* inside\_art *piha* room *o* of *Vai* Vai *Ora.* Ora 'When she arrived at Apina, the old woman got off (her horse) straightaway, inside the house (she went), into Vai Ora's room.' [R301.111]

In older texts, its use is somewhat different. The sense of immediacy is not obvious; *ko* seems to be similar in sense to other locative prepositions like *ꞌi*.

(331) *He* pred *nunui* pl:big *ararua* the\_two *pāꞌiŋa* side *ko* prom *tuꞌa* back *ko* prom *te* art *ꞌana,* cave *ko* prom *haho* outside *ko* prom *te* art *motu.*

islet

'Both groups of children grew up, those in the back of the cave and those outside on the islet.' [Mtx-3-01.293]

(332) *Ka* imp *varu* shave *mai* hither *te* art *pūꞌoko* head *ki* to *toe* remain *ꞌitiꞌiti* little:red *ko* prom *vāeŋa* middle *nō* just *o* of *te* art *rauꞌoho.* hair

'Shave the head, so a little hair will remain only in the middle.' [Ley-6-44.033]

#### **4.7.12.3 Lexicalised** *ko*

In a number of cases, *ko* has become lexicalised, i.e. become part of a word or expression. In these expressions, *ko* is always used, even in syntactic contexts in which it would not occur otherwise. One example is the construction *ko ŋā kope* 'the people, the guys' (§5.5.2). Another example is *taꞌe ko ꞌiti*, which acts as a frozen expression meaning 'not a few, a considerable number, many':<sup>84</sup>

(333) *He* ntr *turu* go\_down *ia* then *te* art *taŋata* person *taꞌe* conneg *ko* prom *ꞌiti* small *ki* to *tū* dem *kona* place *era* dist *o* of *te* art *pahī.* ship

'Quite a few people went down to the place where the ship was.' [R250.211]

Thirdly, the word *tetu* 'huge, enormous' is usually preceded by *ko*. This combination *ko tetu* is lexicalised, that is, its use cannot be predicted from *ko* + *tetu*. <sup>85</sup> *Ko tetu* is used very flexibly: as an adjective modifying a noun, but also freestanding as in (334).

(334) *Nā,* med *te* art *vave* wave *e* num *tahi* one *ko* prf *uru* enter *mai* hither *ꞌā* cont *ko* prom *tetu.* huge 'Look, there comes a huge wave.' [R243.028]

<sup>84</sup> All other adjectives are negated by *ta ꞌe* without the use of *ko* (see (148) on p. 506).

<sup>85</sup> *Ko* in *ko tetu* may have found its origin in the exclamatory *ko*, discussed in sec. §10.4.2.

#### 4 Closed word classes

#### **4.7.12.4 What is** *ko***?**

§4.7.12 started out with the observation that *ko* is a preposition. The question remains, how the function of *ko* should be characterised in general – if this is possible at all.

The multitude of uses of *ko* discussed in various parts of this grammar make clear that *ko* is a marker with an extremely wide range of use. The most common (and probably syntactically most significant) uses are those where *ko* marks a core constituent: a topicalised subject of a verbal clause, the predicate of an identifying clause, or a noun phrase in focus in a cleft construction.

So on the one hand, *ko* marks NPs in focus, a function associated with high information load: focus highlights new and significant information. On the other hand, *ko* marks topical NPs, a function associated with a relatively low information load – topicalised NPs represents information already established in the context (cf. Levinsohn 2007: 51– 52). Several authors have pointed out this dual nature of Polynesian *ko* (e.g. Clark 1976 on PPN, Bauer 1991 and Pearce 1999 on Māori; Massam, Lee & Rolle 2006 on Niuean86).<sup>87</sup> Pragmatically, these two functions can be combined under the label *prominence*: in both functions, the noun phrase is in some way prominent or highlighted. For this reason, *ko* is uniformly glossed as prom.

However, the list in §4.7.12 above shows that the range of functions of *ko* is much wider than topic and focus. Some uses can be reduced to the categories above; for example, *ko ai* in questions is a constituent in focus; the same may be true for *ko* in exclamative clauses, while *ko* in isolated NPs such as titles may be topical. Not all uses are easy to categorise, however: it is less clear how *ko* in appositions, lists, comitative constructions ('X with *ko* Y') and with verbal nouns should be analysed as either topic or focus. The only feature connecting these functions, is that they involve a function not marked by any other preposition. The conclusion seems justified that *ko* is a default preposition for noun phrases which have no thematic role in the clause (i.e. no role marked by any other preposition), an analysis proposed by Clark (1976: 45) for Proto-Polynesian, and adopted by Massam, Lee & Rolle (2006) for Niuean. This analysis is plausible for Rapa Nui as well. Most uses of *ko* involve a noun phrase which either does not have a semantic role, or which has been moved out of its normal argument position. (The only exceptions are non-topicalised subjects marked with *ko*, see §8.6.4.5.)

In many functions, *ko* is in complementary distribution with the nominal predicate marker *he*. (This does not imply that both are structurally identical: while *ko* is a preposition, *he* is a determiner.) Table 4.19 shows how both are used in similar contexts.

As discussed in §5.3.4.1, *he* marks non-referential noun phrases, while other determiners indicate referentiality. We may conclude that noun phrases in non-thematic positions are either non-referential, in which case they are marked with the predicate marker *he*, or referential, in which case they get the default preposition *ko*. For common nouns, both

<sup>86</sup> In Niuean, *ko* has an even wider range of uses than in Rapa Nui, as it also occurs before verbs. Incidentally, Massam, Lee & Rolle (2006: 15) mistakenly assume that the same is possible in Rapa Nui, based on confusion of the prominence marker *ko* and perfect *ko*.

<sup>87</sup> According to Clark (1976: 46), the functions of *ko* can possibly be reduced to "nominal predicate" and "topic", and the two should not be confused.

4.8 Conclusions


Table 4.19: Comparison of *ko* and *he*

strategies are possible. Pronouns and proper nouns, on the other hand, are necessarily referential, so they are always marked with *ko*.

### **4.8 Conclusions**

Closed word classes in Rapa Nui can be placed on a continuum ranging from full words (= open classes of words occurring in the nucleus of a phrase which is a constituent of the clause) to particles (= closed classes occurring in the periphery of a phrase).

Pronouns are close to the full word end of the continuum: they are a closed class, but serve as clause constituents and may take some of the same noun phrase modifiers as proper nouns. They are differentiated for singular, dual, and plural, though the dual/plural distinction was lost in the second and third person.

Both numerals and quantifiers show a massive shift between older and modern Rapa Nui under Tahitian influence. All numerals above seven (or even above five) were replaced by Tahitian equivalents, and in certain contexts the Tahitian terms are used even for lower numerals. On the other hand, a set of reduplicated numerals unique to Rapa Nui (the definite numerals) was preserved, though their use is on the wane (except *ararua* 'the two', which was lexicalised).

Three quantifiers were introduced from Tahitian, while existing quantifiers underwent semantic shifts. Interestingly, the introduced quantifiers were incorporated into Rapa Nui in ways not predictable from their Tahitian origin; their syntax shows features not found in Tahitian.

Demonstratives are very common in Rapa Nui. One set of demonstratives is differentiated for distance (proximal, medial, distal); it actually consists of four subsets with

#### 4 Closed word classes

similar forms, which occur in different syntactic contexts: as determiners, pronouns, locationals and postnuclear particles. The other set consists of a single member *tū*, not differentiated for distance. Demonstratives are extremely common in discourse; in combination with articles, they serve to indicate definiteness, deixis and anaphora.

Rapa Nui has about a dozen prepositions. Prepositions impose restrictions on the following noun phrase: after most prepositions the noun phrase must be introduced by a determiner. The instrumental preposition *hai*, however, precludes the use of a determiner (perhaps reflecting its origin as a prefix), while *pē* 'like' is usually followed by the predicate marker *he*, just like its counterparts in other Polynesian languages (even when these are not etymologically related to *pē*).

The most versatile preposition is *ko*; it marks noun phrases with a wide range of functions: prominent topics, constructions in focus, nominal predicates, et cetera. It can be characterised as a default preposition, marking all noun phrases not marked otherwise.

## **5 The noun phrase**

### **5.1 The structure of the common noun phrase**

As discussed in §3.2, Rapa Nui has three types of nominal elements: common nouns, personal nouns and locationals. This chapter discusses the different elements occurring in the noun phrase. The largest part (sections §5.1–5.12) is devoted to the common noun phrase and its constituents. Proper noun phrases may only contain a subset of these constituents; these are discussed in §5.13.

A prototypical common noun phrase consists of a noun, preceded by a determiner and possibly other elements, and possibly followed by adjectives and other elements. Within the noun phrase, a large number of different positions can be distinguished. Some of these contain a single word, others may contain a phrase. Each position may be empty, including – under certain circumstances – the nucleus. Some positions are mutually exclusive; for example, of the three possessive positions, only one (occasionally two) can be filled in a given noun phrase.

Table 5.1 and Table 5.2 represent the structure of the common noun phrase.


Table 5.1: The common noun phrase: prenominal elements

Apart from these, the noun phrase may contain the following elements:


#### 5 The noun phrase

8 9 10 11 12 13 14 15 16 17 modif. quantif. phrase adverb emph. mkr limit. mkr PND ident. mkr num. phrase poss. deictic part. AdjP QtfP *hakaꞌou*; *takoꞌa mau nō nei*; *ena*; *era ꞌā*; *ꞌana* NumP Ø-poss; poss. phrase *ai* § 5.7 4.4 5.8 5.8 5.8 4.6.3 5.9 5.4 6.2.1 5.10

Table 5.2: The common noun phrase: postnominal elements

Below are examples illustrating the different noun phrase positions. The constituents are numbered according to the numbering in the tables.


In §5.2–5.12, different elements in the common noun phrase will be discussed in turn. Some elements are discussed in other chapters: quantifiers and demonstratives are discussed in Chapter 4, possessors in Chapter 6. See the references in Table 5.1 and Table 5.2. 5.2 The collective marker*kuā*

### **5.2 The collective marker** *kuā*

*Kuā* (etymology unknown; there is also a less common variant *koā*, which does not occur in older texts) indicates a human collectivity, a group of people belonging together. With a singular noun, as in examples (8) and (9) below, it has an associative sense (Dixon 2012: 50): *kuā N* means 'N and the ones around him/her, N and the others'. When the noun itself has plural reference, as in (10–13) below, *kuā* refers to 'the group of N'.

In older texts *kuā* is rare and only occurs before proper names. Nowadays its use has increased in frequency and it occurs before the following elements:

#### **Proper names**

(8) *Pē* like *ira* ana *a* prop *kuā* coll *Tiare* Tiare *i* pfv *iri* ascend *ai* pvp *ki* to *ꞌuta* inland *e* num *tahi* one *mahana.* day 'Thus Tiare and the others went to the countryside one day.' [R151.048]

**Kinship terms** Examples are *koro* 'Dad', *nua* 'Mum':

(9) *He* ntr *nonoho* pl:sit *a* prop *kuā* coll *koro* Dad *he* ntr *kakai.* pl:eat 'Dad and the others sat down and ate.' [R333.538]

**Other personal nouns** An example is *māhaki* 'that person':

(10) *E* voc *Tiare,* flower *e* exh *hāpaꞌo* care\_for *rivariva* good:red *koe* 2sg *i* acc *a* prop *kuā* coll *māhaki.* companion 'Tiare, take good care of the little ones.' [R496.015]

#### **Pronouns**

(11) *¿O* of *kuā* coll *kōrua* 2pl *i* pfv *aŋa?* make 'Did you (pl.) make it together?' [R415.808]

In short, *kuā* occurs before the same words which can also be preceded by the proper article *a*, i.e. proper nouns (§3.3.2). This may have led Du Feu (1987: 474) to classify *kuā* as a proper article as well. However, *kuā* is different from the proper article. As examples (8), (9) and (10) show, *kuā* can co-occur with the proper article. This indicates that the two do not belong to the same class of particles, but occupy different slots in the noun phrase.

In fact, the syntactic behavior of *kuā* shows it to be quite different from *a*. *Kuā* occurs in a number of contexts in which *a* is impossible:

In the first place, even though *kuā* usually occurs before proper nouns, it occasionally occurs before common nouns. *Repa* 'friend' is a common noun which never takes the proper article *a*, but it can be preceded by *kuā*:

#### 5 The noun phrase

(12) *Ka* imp *oho* go *mai,* hither *e* voc *kuā* coll *repa* young\_man *ē.* voc 'Come, my friends.' [R313.004]

Secondly, unlike the proper article *a*, *kuā* occurs after the preposition *o*, as in (11) above. Thirdly, unlike the proper article, *kuā* can be followed by a possessive pronoun:

(13) *Ko* prom *kuā* coll *tōꞌoku* poss.1sg.o *ŋā* pl *poki* child *taina* sibling *rikiriki* small:pl:red *era* dist *ko* prom *tōꞌoku* poss.1sg.o *pāpārūꞌau* grandfather *era.* dist 'We were with my little brothers and my grandfather.' [R123.014]

In the fourth place, unlike the proper article, *kuā* can be followed by the plural marker *ŋā*, as in (13) above.

Finally, *kuā* occurs in the vocative, as in (12) above, something which is not possible with *a*.

All of this shows that *kuā* not only occurs in the proper noun phrase (§5.13.1), but also in the common noun phrase. The fact that *kuā* mostly occurs before the same nouns as the proper article, may have semantic rather than syntactic reasons. The proper article *a* precedes nouns which have a unique referent, and similarly, *kuā* indicates reference to a group which is identified by a unique referent. This unique referent is either a central member of the group (*kuā koro* 'father and company', *kuā Tiare* 'Tiare and the others'), or identified with the group as such (*e kuā repa ē* 'you friends', *kuā ŋā kope* 'guys').

### **5.3 Determiners**

### **5.3.1 Introduction**

The term determiner is used for a category which includes articles, as well as other elements which occur in the same position in the noun phrase, such as demonstratives and possessive pronouns (Dryer 2007b: 161). This means that 'determiner' is a purely structural category; the function of elements in determiner positions may vary.

Table 5.3 lists the elements occurring in the determiner position in Rapa Nui.

These elements are in complementary distribution; for example, the article *te*, demonstrative determiners and possessive pronouns of the *t*-series never occur together.

There is a fundamental distinction between the predicate marker *he* and the other determiners. Categories 1–3 introduce referential noun phrases; in fact, it can be argued that the determiner serves to make the noun phrase referential (see §5.3.3). Categories 1–3 will be referred to as *t*-determiners, as most of these elements start with *t-*.

The predicate marker *he*, on the other hand, marks non-referential noun phrases, such as nominal predicates. Nevertheless, *he* is analysed as a determiner, because it occurs in the same structural position (§5.3.4.2).

5.3 Determiners


Table 5.3: Determiners

Category 4, numeral phrases, is neutral with respect to referentiality. Numeral phrases in determiner position occur both in referential noun phrases and in certain types of nonreferential noun phrases; in both cases, they provide the noun phrase with an obligatory determiner.

In the following subsections, first the distribution of *t-*determiners will be examined (§5.3.2). §5.3.3 then examines the function of the determiner *te*, leading to the conclusion that *te* marks referentiality, not specificity or definiteness. §5.3.4 discusses the use of the predicate marker *he* and shows that, despite its different character compared to other determiners, it should be analysed as a determiner. Finally, §5.3.5 shows that numeral phrases in determiner position occur in both referential and non-referential noun phrases, which shows that they are neutral with respect to referentiality.

### **5.3.2 The syntax of** *t***-determiners**

In this section I will show that the use of *te* is largely determined by syntax. In many contexts, a *t*-determiner is needed; if no other determiner is present, *te* is used as default determiner.

In the following sections the syntactic conditions are listed under which determiners are or are not used. First the conditions will be listed under which determiners are obligatory (§5.3.2.1), then contexts in which determiners are excluded (§5.3.2.2), and finally contexts in which the determiner is optional (§5.3.2.3).

According to Chapin (1974), Rapa Nui is much more flexible than other Polynesian languages in the omission of the article, and the circumstances under which the article can be omitted are not completely clear. A close look reveals, however, that determiners can only be omitted in a limited set of specific contexts.

#### **5.3.2.1 Obligatory** *t***-determiners**

In Rapa Nui discourse, the vast majority of all noun phrases is preceded by a determiner. The most neutral determiner is the article *te*; in fact, *te* is the most common word overall in Rapa Nui discourse.

#### 5 The noun phrase

In many cases, determiners are obligatory. Two main rules can be formulated:<sup>1</sup>


Rule 1 is illustrated in the following examples, both with a subject noun phrase:


Rule 2 is true for all prepositions, including the Agent marker *e* (which often marks subjects) and the accusative marker *i*, but with the exception of *pē* 'like' and *hai* 'with' (§4.7.1). It is stricter than rule 1: the determiner must follow the preposition immediately and it must be a *t*-determiner. This means that noun phrases preceded by a preposition cannot have a prenominal numeral, as the latter excludes a *t*-determiner. In some cases a preposition is omitted to allow for a prenominal numeral. In the example below, the emphasised noun phrase is a direct object, which would normally be preceded by the accusative marker *i*; in this case however, the accusative marker must be omitted, as the noun phrase does not have a *t*-determiner.

(16) *He* ntr *aŋa* make *(\*i)* acc *e* num *tahi* one *paepae.* shack 'He built a shelter.' [Blx-2-1.015]

Most other prepositions cannot be omitted, however. After these prepositions, numerals in determiner position are excluded, so numerals must be placed after the noun. See §5.4 for a fuller discussion and examples.

The rule that prepositions are immediately followed by a *t*-determiner, also has consequences for noun phrases containing a quantifier. As Table 4.14 on p. 160 shows, certain quantifiers may precede the determiner (*taꞌatoꞌa te taŋata* 'all the people') or occur without determiner (*taꞌatoꞌa taŋata* 'every man'). But when the noun phrase is preceded by a preposition, nothing may precede the determiner, so these constructions are excluded. The quantifier must either occur after the determiner (prep *te taꞌatoꞌa taŋata*), or after the noun (prep *te taŋata taꞌatoꞌa*).

<sup>1</sup> NB These rules only apply to common nouns, which have a determiner position in the noun phrase. With personal nouns, the proper article *a* is used, but not in exactly the same contexts as common noun determiners (§5.13.2.1).

5.3 Determiners

The quantifier *meꞌe rahi* 'many/much' (§4.4.7.1) excludes the use of a determiner; this means it cannot occur at all in noun phrases preceded by a preposition. (In this case, speakers may resort to using the adjective *rahi* 'many/much' as an alternative strategy, see (107) on p. 171.)

Exceptions to rules 1 and 2 are only found in a few well-defined contexts, which are described in the next two subsections. Most of these exceptions only apply to rule 1; there are very few situations in which rule 2 is violated.

#### **5.3.2.2 Contexts in which determiners are excluded**

Determiners are excluded in the following contexts:

	- 2. When the noun is followed by *aha* 'what' or *hē* 'which' (§10.3.2.2, §10.3.2.3), even when preceded by a preposition (an exception to rule 2 formulated above):
	- 3. When the noun phrase is preceded by the instrumental preposition *hai* (§4.7.10):
	- 4. In a few expressions in which the noun phrase is non-referential, even when preceded by a preposition (another exception to rule 2). These expressions are typically repeated noun phrases with a distributive sense:

<sup>2</sup> Alternatively, *me ꞌe rahi* itself could be analysed as a determiner.

#### 5 The noun phrase

In these cases the noun phrase does not refer to any house in particular: *hare era* in (20) does not refer to a certain house, but to houses in general. In other words, the noun phrase is non-referential. If the article were used (*ki te hare era*), the noun phrase would refer to a specific house.

#### **5.3.2.3 Contexts in which** *t***-determiners are optional**

In the following situations, *t*-determiners are optional:

	- 2. Occasionally, the determiner is omitted when the noun phrase contains the plural marker *ŋā* (§5.5.2).
	- 3. Sometimes the determiner is omitted in sentence-initial noun phrases which contain a postnominal demonstrative (*nei*, *ena* or *era*). The noun phrase may be the subject of a verbal (23) or nominal clause (24), or a left-dislocated constituent (25):

In these cases, the absence of the article makes no difference in meaning; the omission is a purely stylistic matter, and limited to a somewhat informal style.<sup>3</sup>


<sup>3</sup> Note that the same happens in verb phrases: aspect markers are occasionally left out at the beginning of a sentence, but only if the verb phrase has a postverbal particle (§7.2.2).

5.3 Determiners

### **5.3.3 The function of the article** *te*

The article *te* is widespread in Polynesian languages.<sup>4</sup> Older descriptions characterise it as a definite article, while *he* is described as an indefinite article. According to Du Feu (1996: 11), *te* in Rapa Nui is a [+specific] article, while *he* is [±specific]. However, in actual fact *he* and *te* are not two articles which can be substituted for each other. They occur in quite different syntactic contexts. *He* mainly introduces noun phrases which serve as predicates of non-verbal clauses (this will be discussed in §5.3.4 below). It does not occur, for example, in noun phrases serving as subject or object of a verbal clause:


This means that *te* is the only full-fledged article in Rapa Nui. As indicated in §5.3.1, it is in complementary distribution with the other *t*-determiners: demonstratives and possessive pronouns of the *t-*series.

The article *te* occurs with all common nouns, that is, all nouns which do not take the proper article *a* (§5.13.2). It is not specified for gender or case. Neither is it specified for number:<sup>5</sup> both singular and plural nouns are introduced by *te*. Number is indicated by the plural marker *ŋā*, by numerals or understood from the context.

*Te* can be used with count nouns as in (28), mass nouns as in (29), and abstract concepts as in (30):


<sup>&#</sup>x27;Great love is like a dove…' [R222.036–037]

<sup>4</sup> In fact, cognates of *te* occur in all Polynesian languages, though in some language PPN *\*te* underwent an irregular change (e.g. Tongan *he*, Samoan *le*). Interestingly, *te* as a definite or specific article is not reconstructed for any protolanguage prior to PPN; however, Clark (2015) shows that possible cognates occur in various Oceanic languages, mostly as an indefinite article. If these are indeed cognates, this article extended its use to definite NPs in PPN.

<sup>5</sup> This is different from the situation in some other EP languages, where the plural marker is in determiner position and in complementary distribution with *te*; see Footnote 17 on p. 249.

#### 5 The noun phrase

Is *te* a definite article, as older descriptions suggest? Lyons (1999) defines definiteness in terms of identifiability: the definite article signals that the hearer is in a position to identify the referent of a noun phrase. When a speaker says 'Pass me the hammer', the hearer infers that there is a single hammer that he/she is able to identify.<sup>6</sup>

In this sense, *te* cannot be considered a definite article. In many cases, *te* introduces noun phrases with indefinite reference.

(31) *Ko* prf *tuꞌu* arrive *ꞌana* cont *a* prop *au* 1sg *ki* to *ruŋa* above *i* at *te* art *henua* land *e* num *hitu.* seven '(In my dream) I arrived on seven islands.' [R420.014]

Even when not definite, *te* usually refers to a specific entity. Thus in the following example, *te taŋata e tahi* refers to a specific man; his name is mentioned straight afterwards.

(32) *ꞌI* at *te* art *noho* stay *iŋa* nmlz *tuai* ancient *era* dist *ꞌā* ident *te* art *taŋata* man *e* num *tahi* one *te* art *ꞌīŋoa* name *ko* prom *Tuꞌuhakararo.* Tu'uhakararo

'In the old times (there was) a man called Tu'uhakararo.' [R477.002]

However, *te* can also be used in non-specific contexts. This is for example the case in general statements, in which the noun phrases have generic reference:

(33) *E* ipfv *tano* correct *nō* just *mo* for *maꞌu* carry *i* acc *te* art *mōai* statue *e* num *hoꞌe* one *ꞌahuru* ten *toneladas* tons *… e* ag *te* art *taŋata* man *e* num *hoꞌe* one *hānere* hundred *vaꞌu* eight *ꞌahuru.* ten

'It is possible to transport a statue of ten tons… by one hundred and eighty men.' [R376.062]

This sentence does not refer to any specific situation involving a specific statue and specific people, but to statues and people in general.

A noun phrase is also non-specific when its referent is hypothetical. This happens for example when the item is desired or sought as in (34–35), denied as in (36), or its existence is questioned as in (37). In all these examples, the referent has not been mentioned in the preceding context, but even so, *te* is used:

(34) *…mo* for *ai* exist *o* of *te* art *moni* money *mo* for *hoꞌo* trade *mai* hither *i* acc *te* art *haraoa.* bread

'(He sells food) in order to have money to buy bread.' [R156.023]

<sup>6</sup> As the notion of identifiability is not without problems, Lyons (1999) also uses the notion of inclusiveness: the definite article signals that there is only one entity satisfying the description used, relative to the context. Thus in 'There was a wedding. The bride was radiant,' the hearer cannot identify the bride (he does not know who she is), and yet 'the bride' is definite: the article indicates that in the given situation there is only one bride.

5.3 Determiners


We may conclude that the article *te* in Rapa Nui does not indicate definiteness or specificity.<sup>7</sup> Rapa Nui does have other devices to indicate these:


This raises the question whether *te* has any meaning at all. Its role seems to be purely syntactic as a default determiner: whenever a determiner is needed and the noun phrase has no other determiner, *te* is used. However, this begs the question why the syntax requires a determiner at all in the contexts discussed in §5.3.2. To recapitulate: *te* (or another *t*-determiner) is generally required in core grammatical roles and after prepositions. On the other hand, *t*-determiners do not occur when the noun phrase serves as a predicate; in that case *he* is used (§5.3.4. This suggests that *te* does have a semantic function: the article *te* (and other *t*-determiners) signals referentiality; it turns a common noun into a referential expression.

A common noun as such is not referential; in itself it does not refer to an entity, but denotes a certain property which defines a class of entities. A determiner is needed to create an expression which refers to one or more entities belonging to this class, and only on this condition can the noun be used as a subject or object, or as complement of a preposition. The noun phrase may refer to a specific entity (whether known to the hearer or not, i.e. definite or indefinite) or to some unspecified one: referentiality is not the same as specificity.

A noun phrase in argument position or after a preposition is referential, so it needs a determiner. Any determiner will do: a demonstrative, a possessive pronoun, or – by

<sup>7</sup> One could wonder whether an element not encoding definiteness or specificity still qualifies as an article. Dryer (2007b: 157) answers this question in the affirmative.

#### 5 The noun phrase

default – the article *te*. <sup>8</sup> On the other hand, in order for a noun to function as predicate, *t*determiners are excluded: the predicate should not be referential, but denote a property.

This analysis explains why determiners are used with noun phrases in argument positions and after prepositions, while *he* is used with predicate noun phrases. It is further confirmed by a number of other phenomena.

In the first place, as discussed in the section on quantifiers, the article is often omitted with a prenominal quantifier (§4.4.2 on *taꞌatoꞌa* 'all'; §4.4.5 on *rauhuru* 'diverse'; §4.4.7.1 on *me ꞌe rahi* 'many'). While referential noun phrases presume the existence of an entity, noun phrases with the universal quantifier 'all' do not; in other words, they can be considered non-referential.

It is thus not surprising that with the universal quantifier, the article can be left out. Now this does not explain yet why the article can also be left out with *rauhuru* 'diverse' and *meꞌe rahi* 'many': unlike the universal quantifier, these do imply the existence of a set of entities. However, they do not single out a definite number of individuals from a set; rather, they denote an unspecified subset from the total set of entities denoted by the noun. 'Many people went' implies that there exists a subset from the class of 'people' for whom the predicate 'went' is true, but without being specific about the extent of this subset. While these expressions are not strictly non-referential, they appear to be lower on the referentiality scale than expressions referring to individuated entities. These quantifiers are somewhat similar to distributional expressions (§5.3.2.2), which likewise exclude the article *te*:

(38) *He* ntr *eꞌa* go\_out *rā* dist *matuꞌa* parent *Iporito* Hippolytus *ki* to *hare* house *era* dist *ki* to *hare* house *era.* dist 'Father Hippolytus went here and there (lit. to that house to that house).' [R231.282]

Secondly, as will be discussed below (§5.3.4), *t*-determiners are excluded – and the predicate marker *he* is used – not only with nominal predicates, but with other nonreferential noun phrases as well:


<sup>8</sup> Rigo & Vernaudon (2004: 467) apply the same analysis to the Tahitian article *te*. They refer to Lemaréchal (1989), who analyses such elements as translating a "nom" into a "substantif". A "nom" expresses a quality (e.g. 'doctor' = the quality of being a doctor), while a determiner converts this into a referring expression (a person who has the quality of being a doctor). Gorrie, Kellner & Massam (2010) give a partly similar analysis for determiners in Niuean: determiners are the obligatory elements which allow a noun to function as an argument. They separate this function from referentiality, which in their analysis is provided by other noun phrase elements.

5.3 Determiners

(Cablitz 2006: 136–137), Hawaiian (Cook 1999: 53; Elbert & Pukui 1979: 156), Māori (Polinsky 1992: 237; Bauer, Parker & Evans 1993: 356) and Tuvaluan (Besnier 2000: 224). In all these languages, *me* must be followed by the predicate marker *se* or *he*. The use of *he* after prepositions meaning 'like' can be explained from the nonreferential character of the noun phrase after these prepositions: the noun phrase does not refer to any concrete entity, but involves a comparison with a class of entities; hence the predicate marker *he* is appropriate rather than referential *te* (cf. Polinsky 1992: 236; cf. the discussion of *he* in §5.3.4).

Thirdly, noun phrases in interrogative and negative sentences are less referential than those in positive declarative sentences: in both cases, the noun phrase does not refer to an entity whose existence is presupposed.<sup>9</sup> Now the use of *t-*determiners is not excluded in interrogative and negative contexts per se. Two examples from the previous section are repeated here:


On the other hand, there is one negative and one interrogative construction in which *t-*determiners are excluded:


Recapitulating: *t-*determiners are excluded (and *he* used instead) in non-referential noun phrases: nominal predicates, appositions, after *pē* 'like'. They are also commonly omitted (though not excluded) with the universal quantifier *taꞌatoꞌa*. Finally, *t-*determiners are excluded in a number of contexts which are low on the referentiality scale (though not strictly non-referential): with *meꞌe rahi* 'many', in questioned noun phrases, and after the negation *ꞌina*.

<sup>9</sup> Chung, Mason & Milroy (1995: 437) explain the use of *he* in Māori in (among others) interrogative and negative constructions precisely from the non-referential character of the noun phrase in these contexts.

<sup>10</sup> Notice that the use of *he* rather than *te* in this example cannot be explained as an existential construction. This sentence is not a negation of 'there was an old woman who went down to Hanga Roa', but refers to a definite woman, as the demonstrative *nei* indicates. Even so, the negation triggers the use of the predicate marker instead of the referential article.

#### 5 The noun phrase

### **5.3.4 The predicate marker** *he*

#### **5.3.4.1 Uses of** *he*

The determiner *he* reflects PPN *\*sa* (> PNP *\*se*); its cognates occur in most Polynesian languages. In the past these have often been analysed as indefinite articles (see references in Polinsky 1992: 230). For Rapa Nui, Englert (1978: 18) already realised that *he* is something different from an indefinite article: *he* 'se emplea cuando se trata de denominaciones generales de personas u objetos' (is used when general designations of persons and objects are concerned).

The Proto-Polynesian ancestors of *te* and *he* did function as definite and indefinite (or specific and non-specific) article respectively (see Clark 1976: 47–50; Hamp 1977: 411). In Samoic and Tongic languages, *he* continued to function as an indefinite article: it is commonly used to introduce referential noun phrases functioning as verb arguments.<sup>11</sup> In Eastern Polynesian languages, however, *he* mainly functions as nominal predicate marker, though in some languages it is occasionally used to mark argument noun phrases<sup>12</sup> (see §5.3.3 on referentiality).

As explained in the previous section, Rapa Nui *he* occurs in non-referential noun phrases and is excluded in referential noun phrases (with a single exception, see Footnote 12 above). It is mainly used to mark noun phrases as predicates of a verbless clause. In the following example, *he taŋata* is the predicate of the clause: 'man' is predicated of the subject *tau manu era*. The clause is classifying (§9.2.1): it expresses that the subject belongs to the class of human beings. *Taŋata* does not refer to any man in particular, nor to a group of men or even to men in general; rather, it denotes the property of 'being man'.

(42) *He* pred *taŋata* man *tau* dem *manu* bird *era.* dist

'That bird was a human being.' [Mtx-7-12.069]

Besides classifying clauses, *he* is also used in existential clauses (§9.3.1):

(43) *He* pred *taŋata* man *ko* prom *Eŋo.* Engo 'There was a man (called) Engo.' [Mtx-7-28.001]

As a nominal predicate marker, *he* also marks the complement of the copula verbs *riro* 'become' and *ai* 'be' (§9.6).

Apart from marking the predicate of a verbless clause, *he* has the following other uses:

<sup>11</sup> See e.g. Mosel & Hovdhaugen (1992: 261–264) on Samoan, Besnier (2000: 365) on Tuvaluan, and Anderson & Otsuka (2006: 22) on Tongan.

<sup>12</sup> In Māori (Polinsky 1992; Chung, Mason & Milroy 1995) and Hawaiian (Cook 1999), *\*he* occasionally marks objects or non-agentive subjects. In Rapa Nui its use with argument noun phrases is very marginal, though not limited to non-agentives: it may mark topicalised noun phrases, regardless the nature of the verb (§8.6.2.2). Clark (1997) provides a reconstruction of the shifts in the use of *he* in PEP.

5.3 Determiners

**In appositions** *He* optionally occurs before common nouns in appositions (§5.12):

(44) *He* ntr *kī* say *e* ag *te* art *matuꞌa* parent *tane* male *era* dist *o* of *Te* Te *Rau,* Rau *he* pred *taŋata* man *pūꞌoko* head *o* of *te* art *nuꞌu* people *o* of *Kapiti…* Kapiti

'The father of Te Rau, the leader of the people of Kapiti, said…' [R347.089]

**In isolation** *He* is used before common nouns in isolation (i.e. without a semantic role in the clause), for example in titles:


**In lists** Noun phrases in enumerations or lists may also be syntactically isolated, in which case they are also marked by *he*:

(47) *Te* art *aŋa* work *nō* just *ꞌa* of.a *Reŋa* Renga *he* pred *tunu* cook *i* acc *te* art *kai:* food *he* pred *moa,* chicken *he* pred *tarake,* corn *he* pred *kūmā,* sweet\_potato *ika* fish *ꞌe* and *tētahi* other *atu.* away 'What Renga used to do was cooking food: chicken, corn, sweet potato, fish and other things.' [R363.119]

(Proper nouns and pronouns in isolation and in lists are marked with *ko*; common nouns are marked with *ko* when uniquely identifiable; see §4.7.12.1.)

**After** *pē* **'like'** After the preposition *pē* 'like' (§4.7.9), *he* is obligatory.

**After the negator** *ꞌina He* is used in noun phrases immediately following the negator *ꞌina*, whether the noun phrase is referential or not (§10.5.1):

(48) *ꞌIna* neg *he* pred *rūꞌau* old\_woman *nei* prox *he* ntr *turu* go\_down *mai* hither *ki* to *Haŋa* Hanga *Roa.* Roa 'This old woman did not go down to Hanga Roa.' [R380.006]

#### 5 The noun phrase

**In topicalisation** *He* occasionally marks topicalised subjects in a verbal clause (§8.6.2.2):

(49) *He* pred *taŋata* man *he* ntr *oho* go *he* ntr *ruku* dive *i* acc *te* art *ika* fish *mo* for *te* art *hora* time *kai.* eat 'The men went diving for fish for lunch.' [R183.019]

#### **5.3.4.2** *He* **is a determiner**

The discussion so far has shown that the distribution of *he* is quite different from that of *t*-determiners: it usually does not mark verbal arguments, does not occur after most prepositions, but instead is mainly used when the noun phrase functions as predicate or is in another non-argument position. Even so, *he* is most plausibly analysed as a determiner, i.e. an element occurring in the same structural position as the *t-*determiners. There are different reasons for doing so.<sup>13</sup>


Likewise, *he* it is excluded when the noun phrase contains a prenominal numeral. Here is an example with a noun phrase following the negation *ꞌina*, a context in which normally *he* would be used (see 48 above):

	- 2. Although *he* is precluded after almost all prepositions, there is one exception: *he* does occur – and is even obligatory – after the preposition *pē* 'like' (§4.7.9), as the following little riddle shows:

<sup>13</sup> Cook (1999) proposes the same analysis for Hawaiian *he*, based on the fact that it can be preceded by the preposition *me* 'like', cannot be followed by another determiner, and does not occur before pronouns.

5.3 Determiners

	- 4. Like other determiners, *he* does not occur before pronouns or proper names. Pronouns and proper names are rather preceded by *ko* or the proper article *a* (see §5.13.2).
	- 5. Like other determiners, *he* can signal nominalisation of a verb; see (23) on p. 91 for an example.

This evidence clearly shows that *he* is a determiner, even if its distribution is different from other determiners.

### **5.3.5 Numeral phrases in determiner position**

As indicated in §5.3.1 above, numeral phrases may occur in determiner position, thereby excluding other determiners. Interestingly, they occur both in referential noun phrases (which would otherwise contain a *t-*determiner) and in non-referential noun phrases (which would otherwise contain the predicate marker *he*).

Here are two examples of prenominal numerals in referential noun phrases, as subject (56) and direct object (57), respectively:


Prenominal numerals also occur in various non-referential (or at least less referential) constructions, in which the determiner *he* would be used otherwise (cf. the uses listed in §5.3.4.1 above):

After *ꞌina*:

(58) *ꞌIna* neg *e* num *tahi* one *kope* person *i* pfv *ꞌite* know *ko* prom *ai* who *te* art *meꞌe* thing *i* pfv *rē.* won

'No one (lit. not one person) knew who had won.' [R448.018]

5 The noun phrase

In an existential clause:

(59) *E* num *tahi* one *poki* child *te* art *ꞌīŋoa* name *ko* prom *Eva* Eva *ka* cntg *hoꞌe* one *ꞌahuru* ten *matahiti.* year 'There was a child called Eva, ten years old.' [R210.001]

In a list:

(60) *E* num *ono* six *kope* person *o* of *ruŋa:* above *e* num *hā* four *taŋata,* man *e* num *tahi* one *viꞌe,* woman *ꞌe* and *he* pred *poki* child *e* num *tahi.* one

'There were six people on board: four men, one woman, and one child.' [R231.085]

In the first version of this grammar (Kieviet 2016), prenominal numerals were analysed as being in a post-determiner position. However, on a closer analysis it makes more sense to analyse them as determiners: they are in complementary distribution with other determiners and they occur in contexts where otherwise either a *t*-determiner or *he* is required. Under the present analysis, noun phrases with a prenominal numeral are no exception to the rule that in these contexts a determiner is obligatory.

Finally, if prenominal numerals are determiners, this also explains why they are excluded after the instrumental preposition *hai*, which precludes the use of a determiner: in order to use a numeral after *hai*, the numeral must occur after the noun (see (65) in §5.4.2 below).

### **5.4 Numerals in the noun phrase**

§3.5 discusses numerals in general; in the present section, their occurrence in the noun phrase is discussed. Numerals can appear either before or after the noun; both positions will be discussed in turn.

### **5.4.1 Numerals before the noun**

Numeral phrases very often appear before the noun. In §5.3.5 I argued that these numerals are in determiner position, as they exclude other determiners and have a distribution typical of determiners.

As discussed in §5.3.2.1, prenominal numerals cannot be preceded by a preposition. As a consequence, they occur most commonly in noun phrases functioning as subject or direct object<sup>14</sup> (see (56–57) in §5.3.5 above); however, the noun phrase may also be an oblique argument (61) or adjunct (62). Without a prenominal marker, the noun phrase in (61) would be preceded by the preposition *ki* 'to', while the adjunct noun phrase in (62) would be preceded by *ꞌi* 'in, at'.

<sup>14</sup> The same constraint applies in Tahitian: with prenominal numerals, the object marker is omitted (Lazard & Peltzer 2000: 184).

5.4 Numerals in the noun phrase


Constructions like (61) above are somewhat rare, though; it is unusual for the preposition *ki* to be omitted.

### **5.4.2 Numerals after the noun**

When the noun phrase is preceded by a preposition requiring a determiner (§5.3.2.1), numerals must be placed after the noun. Here are examples with possessive *o* (63), and the preposition *i* after a locational (64):<sup>15</sup>


The instrumental preposition *hai* 'with' excludes a determiner (§4.7.10); here as well, numerals must be postnominal.

(65) *E* ipfv *ꞌauhau* pay *era* dist *ꞌi* at *te* art *ꞌāvaꞌe* month *hai* ins *māmoe* sheep *e* num *hā.* four

'He was paid four sheep (lit. with four sheep) per month.' [R250.053]

Likewise, numerals must be postnominal after the preposition *pē* 'like', which is obligatorily followed by the predicate marker *he* (§4.7.9):

<sup>15</sup> The only case in which a numeral does occur after a locative expression, is when the noun phrase is headless. In the following example, the noun phrase *e tahi o kōrua* consists of a numeral phrase and a possessive; there is no head noun.

<sup>(</sup>i) *…ꞌo* lest *topa* happen *tā ꞌue* by\_chance *rō* emph *atu* away *te* art *ꞌati* problem *a* by *ruŋa* above *e* num *tahi* one *o* of *kōrua.* 2pl 'lest a disaster may fall on one of you' (R313.010)

In this case, there is no postnominal position available (alternatively, one could assume that the numeral is in postnominal position, which cannot be distinguished from the prenominal position anyhow).

#### 5 The noun phrase

(66) *Taꞌatoꞌa* all *mata* tribe *e* ipfv *ai* exist *rō* emph *ꞌana* cont *te* art *rāua* 3pl *taŋata* man *pūꞌoko…* head *pē* like *he* pred *suerekao* governor *e* num *tahi* one *te* art *haka* caus *auraꞌa.* meaning

'All tribes had a leader… someone like a governor (lit. like one governor the meaning).' [R371.006]

In other situations, a noun phrase must contain a *t-*determiner for discourse reasons. When a participant is definite, this is indicated by a *t-*determiner + a postnominal demonstrative (§4.6.3). In such cases, numerals must come after the noun. Consider the following example:

(67) *A* prop *Makemake* Makemake *i* pfv *hakaroŋo* listen *mai* hither *era* dist *ki* to *te* art *ani* request *atu* away *o* of *tou* dem *ŋāŋata* men *era* dist *e* num *rua.* two 'Makemake listened to the request of those two men.' [Fel-40.044]

The two men have been mentioned before and are therefore known; this is signalled by anaphoric *tou N era*. The numeral *e rua* necessarily occurs after the noun.

### **5.4.3 Optional numeral placement;** *e tahi* **'one'**

§5.4.1 above describes contexts in which the numeral can be prenominal. This does not mean that the numeral must be prenominal in these contexts. Syntactically, in most of these cases the numeral can be placed after the noun as well. Here are examples with a postnominal numeral in a subject noun phrase (68) and a direct object noun phrase (69):


In cases like (68–69) the choice between pre- and postnominal numerals is syntactically free; however, there may be a subtle difference in meaning. This is especially the case with *e tahi* 'one'. In prenominal position, *e tahi* tends to have a non-specific sense. This sense is especially clear after negations, when *tahi* can often be translated as 'any':

(70) *He* ntr *hāhaki* gather\_shellfish *mai,* hither *pero* but *ꞌina* neg *kai* neg.pfv *ravaꞌa* obtain *e* num *tahi* one *meꞌe.* thing 'She went to gather shellfish, but she didn't get anything.' [R178.026]

5.5 Plural markers

Similarly, in existential clauses, *ꞌina e tahi* is used in the sense 'not any, none at all':

(71) *ꞌIna* neg *e* num *tahi* one *kona* place *toe* remain *mo* for *moe.* sleep 'There was no place left to sleep.' [R339.027]

When the numeral is placed after the noun, its sense is often specific, 'one, a certain':

(72) *He* ntr *moe* lie\_down *ki* to *te* art *uka* girl *e* num *tahi…* one 'He married a (certain) girl…' [Blx-3.002]

There is no absolute distinction between the two, though. For example, in narrative texts, both prenominal and postnominal *e tahi* are common to introduce participants at the beginning of stories:


### **5.5 Plural markers**

### **5.5.1 The plural marker** *ŋā*

#### **5.5.1.1 The position of** *ŋā*

The plural marker *ŋā* always occurs immediately before the noun:

(75) *He* ntr *tuꞌu* arrive *mai* hither *tou* dem *ŋā* pl *uka* girl *era.* dist 'Those girls arrived'. [Blx-3.053]

As this example shows, *ŋā* is not an article.<sup>16</sup> It occurs in a different position than the article *te* and often co-occurs with it. This is different from its cognates in most other Eastern Polynesian languages, which are usually plural articles.<sup>17</sup>

The fact that the plural is always contiguous to the noun, is an indication of its close syntactic association to the noun. Other indications are:

<sup>16</sup> *Pace* Du Feu (1987: 474).

<sup>17</sup> Hawaiian *naa*, Māori *nga*, Marquesan *na* and the possible cognate Tahitian *nā* are all determiners, which do not co-occur with *te*. In Hawaiian and Māori this article denotes plurality, in Marquesan and Tahitian it is used for a dual or limited plural (see Elbert 1976: 19; Biggs 1973: 20; Zewen 1987: 11; Académie Tahitienne 1986: 16). In Tahitian, according to Académie Tahitienne (1986: 18), *nā* is incompatible with the article *te*

#### 5 The noun phrase


#### **5.5.1.2 Use and non-use of** *ŋā*

*Ŋā* is not obligatory. When it is clear that the noun phrase has plural reference, *ŋā* can be left out; this happens in the following situations:

Firstly, when the noun phrase contains a numeral:

(77) *Viꞌe* woman *nei* prox *e* ipfv *ai* exist *rō* emph *ꞌā* cont *e* num *rua* two *poki* child *vahine.* female 'This woman had two daughters' [R491.008]

Secondly, when the noun is subject of a verb which has a plural form as in (78), or is modified by a plural adjective.

(78) *He* ntr *ŋaro* disappear *tū* dem *pereꞌoa* car *era,* dist *he* ntr *ŋāŋaro* pl:disappear *te* art *vaꞌehau.* soldier 'The carriage disappeared and the soldiers disappeared.' [R491.039]

Thirdly, when the noun phrase contains the collective marker *kuā* (§5.2).

Finally, when there are other indications in the context that the noun phrase has plural reference. The following example occurs in a story about a party. No plural marker is needed to indicate that a party involves more than one man and more than one woman:

(79) *He* ntr *hoki* return *te* art *taŋata,* man *te* art *viꞌe,* woman *te* art *ŋā* pl *poki…* child *ki* to *to* art:of *rātou* 3pl *hare.* house '(When the party is finished,) men, women and children go home.' [Mtx-7-30.037]

In conclusion, *ŋā* can be omitted when it is clear that reference is plural. However, this does not mean that *ŋā* is only used when there is no other clue for plurality. It may co-occur with a numeral or other quantifier as in (80) or a plural verb form as in (81):

(80) *He* ntr *eꞌa* go\_out *mai* hither *tou* dem *ŋāŋata* men *e* num *ono* six *mai* from *roto* inside *mai* from *te* art *hare* house *ki* to *haho.* outside 'Those six men came out of the house.' [Ley-4-01.023]

and composite determiners containing *te* (despite Tryon's example *tā ꞌu nā ꞌurī*, Tryon (1970: 17)), but it may co-occur with the demonstrative *taua*: *taua nā tamariꞌi a Noa*… 'those children of Noah' (Gen. 9:18).

In Rarotongan, on the other hand, the particle *ŋā* – which is most commonly used for pairs – is commonly preceded by the article or another determiner: *tōku ngā metua* 'my parents' (Buse 1963a: 405–406); *te ngā pēre pūtē* 'the two bales of sacks' (Sally Nicholas, p.c.).

5.5 Plural markers

(81) *Ko* prf *ꞌaꞌara* pl:wake\_up *ꞌana* cont *tū* dem *ŋā* pl *vārua* spirit *era.* dist 'Those spirits woke up.' [R233.026]

The only case in which *ŋā* is obligatory, is with the noun *io* 'young man', which (almost) only occurs as a plural *ŋā io*. *Ŋā io* is especially common in older stories, but is still in use. It is so much a unit that Englert (1978; 1980) writes it as one word.

#### **5.5.1.3 Semantics of** *ŋā*

In older texts, *ŋā* is almost exclusively used with nouns referring to humans: *taŋata* 'man', *viꞌe* 'woman', *poki* 'child', *matuꞌa* 'father', *taina* 'brother', et cetera.<sup>18</sup>

Nowadays, *ŋā* is frequently used with inanimate nouns as well, including abstract nouns:<sup>19</sup>


The sense of *ŋā* is very general. It can be used for small and large numbers alike:


It can be used for items forming a group as in (86), or for a plurality of separate items as in (87):<sup>20</sup>

<sup>18</sup> Englert (1978: 26) states that *ŋā* is only used for persons.

<sup>19</sup> According to Schuhmacher (1993: 170), this development occurred under influence of Tahitian *nā*; more likely, it was influenced by Spanish – where plurality is consistently marked – or a (quite natural) languageinternal development.

<sup>20</sup> *Pace* Du Feu (1996: 135); Du Feu (1987: 485).

#### 5 The noun phrase


In conclusion, *ŋā* may indicate any kind of plurality with any noun.

### **5.5.2 Co-occurrence of** *ŋā* **and the determiner**

As shown above, the occurrence of *ŋā* is independent of the occurrence of the article. However, there are some noun phrases containing *ŋā* which do not have any determiner, even though there should be a determiner according to the conditions listed in §5.3.2. Here are some examples:


'After ten years of (the reign of) the children of Miru, the children of Tupahotu revived.' [Mtx-3-07.016]

(91) *He* ntr *oho* go *au* 1sg *ki* to *ŋā* pl *hare* house *he* ntr *noꞌinoꞌi* request:red *hai* ins *kona* place *mahute.* mulberry 'I'm going to the houses to ask for mulberry fibres.' [R352.025]

Although these examples are unusual, they are grammatical and can be explained in one of several ways:

• In (88) and (89) the noun is *kope*. *(Kōrua) ko ŋā kope* is more or less a frozen expression, though (89) shows that it also occurs without *ko*. It expresses endearment: 'those dear boys'.

5.5 Plural markers

	- (91) may be an example of non-referential use. In such expressions the noun phrase does not refer to any house, but to houses in general. (See (20–21) on p. 235.)

### **5.5.3 Other words used as plural markers**

Sometimes plurality is expressed by other words than *ŋā*.

*Mau* Some speakers use the Tahitian plural marker *mau* (not to be confused with the emphatic marker, §5.8). Tahitian *mau*, like Rapa Nui *ŋā*, is a marker which occurs after the article. For speakers familiar with Tahitian, the similarity in syntax would facilitate using the Tahitian form.


Like most Tahitian borrowings, this is a relatively recent phenomenon.

*Nuꞌu Nuꞌu* 'people' (an inherently plural noun, borrowed from Tahitian *nuꞌu* 'army, collectivity') can be placed in front of another noun and play the same role as a plural marker. It implies a group of people being and/or acting together.


'What he likes is… having a good time with his friends.' [R489.003]

5 The noun phrase

This does not mean that *nuꞌu* is a plural marker like *ŋā*. Syntactically *nuꞌu* is a head noun modified by another noun. It can even be preceded by *ŋā* (*tū ŋā nuꞌu era* 'those people').

### **5.6 The noun: headless noun phrases**

In most contexts, the noun is obligatory; headless noun phrases are uncommon in Rapa Nui. They do occur, but only in certain specific contexts.


'One she gave to her aunt. The others she gave to other people.' [R168.006–007]

(98) *Taꞌatoꞌa* all *e* ipfv *tahuti* run *era,* dist *e* ipfv *tari* carry *mai* hither *era* dist *i* acc *te* art *kai.* food 'All ran, carrying the food.' [R210.155]

**5.6.2** After a *t*-possessive pronoun, in the partitive construction "possessive *o te* noun" (§6.2.2). In this construction, the noun phrase does not have a head noun; instead, the main concept is expressed by a genitive phrase:

(99) *Kai* neg.pfv *toe* remain *tāꞌana* poss.3sg.a *o* of *te* art *ika,* fish *o* of *te* art *ꞌura,* lobster *o* of *te* art *kōꞌiro.* conger\_eel 'There was no fish, lobster or conger eel left for her.' [Mtx-4-04.003]

In other cases it is also possible to leave out the noun after a *t*-possessive pronoun. The implied head noun may be expressed in a preceding clause as in (100), or not at all as in (101).


5.6 The noun: headless noun phrases

**5.6.3** Similarly, a possessive phrase may occur without head noun; the head noun is understood from the context. The noun phrase starts with *to*, which is a contraction of the article *te* and the possessive marker *o* (§6.2.3):

(102) *Ko* prom *Koka* Koka *te* art *ꞌīŋoa* name *o* of *tōꞌona* poss.3sg.o *hoi…* horse *ko* prom *Parasa* Parasa *to* art:of *te* art *rūꞌau* old\_woman *era* dist *ꞌāꞌana.* poss.3sg.a

'Koka was the name of his horse, Parasa the (name) of his old wife.' [R539-1.420]

(103) *¿Ko* prom *ai* who *te* art *ꞌīŋoa* name *o* of *te* art *taŋata?* name *¿To* art:of *te* art *viꞌe?* woman 'What is the man's name? And the woman's?' [Notes]

**5.6.4** Headless noun phrases are marginally possible in noun phrases containing adjectives. Adjectives in the noun phrase usually need a noun. If need be, a generic noun like *kope* 'person' or *meꞌe* 'thing' is used.

(104) *He* ntr *kī* say *te* art *poki* child *(kope,* person *meꞌe)* thing *nuinui* big:red *ki* to *te* art *poki* child *(kope,* person *meꞌe)* thing *ꞌitiꞌiti…* small:red '(There were two children.) The big one said to the small one…' [Notes]

But with a few adjectives in a specific idiomatic sense, the noun can be left out:

(105) *Te* art *pepe* chair *nei* prox *mo* for *te* art *hōnui.* respected *Te* art *pepe* chair *era* dist *mo* for *te* art *rikiriki.* small:pl:red 'These chairs are for the authorities. Those chairs are for the small people.' [Notes]

In the (infrequent) cases above, the nounless construction refers to someone or something possessing a quality. These should be distinguished from nominally used adjectives which refer to the quality as such. The former can be considered as ellipsis of a noun, the latter as conversion of an adjective to a noun (Bhat 1994: 96). In the following examples, *nuinui* 'big' is used as a noun in the sense 'bigness, size, greatness'. It cannot be used in the sense 'big one'.

(106) *Te* art *nuinui* big:red *o* of *Tahiti* Tahiti *e* ipfv *ꞌāmui* add *atu* away *tāua* 1du.incl *e* num *ono* six *nuinui* big *nei* prox *o* of *Rapa* Rapa *Nui.*

Nui

'The size of Tahiti altogether is six times the size of Rapa Nui.' [R348.003]

#### 5 The noun phrase

**5.6.5** Relative clauses (§11.4) can never be headless, but need to be preceded by a noun. When no other noun is available, the dummy noun *meꞌe* is used. This happens for example in clefts (§9.2.6).

To summarise: headless noun phrases occur occasionally, on the condition that the noun phrase contains either a possessor, a numeral phrase, a quantifier, or one of a small set of adjectives. They cannot occur with just any adjective; neither is the presence of a relative clause or a demonstrative sufficient to omit the head noun.

### **5.7 Modifiers in the noun phrase**

### **5.7.1 Introduction: types of modifiers**

The noun may be modified by various elements: nouns, adjectives or – less commonly – verbs.<sup>21</sup> A modifying noun may in turn be modified by another noun, verb or adjective, and so on. Modifying verbs may be followed by a direct object; modifying adjectives may be modified by various elements, such as degree markers.

At first sight, a modifying noun or verb seems to have the same status as a modifying adjective, but there are important differences between the two. Syntactically, a modifying noun or verb is incorporated into the head noun; it is a bare noun or verb, not followed by verb phrase particles. Modifying adjectives, on the other hand, form an adjective phrase, which may contain elements like degree modifiers, negators and adverbs (§5.7.3.2). This correlates with a difference in position: when a noun phrase contains both a modifying noun or verb and an adjective, the former is usually closer to the head noun.

Semantically, a modifying noun or verb tends to express a single concept together with the head noun. In other words, the combination is a compound, a single lexical item. Adjectives, on the other hand, express some additional property of the concept expressed by the head. For example, in the following noun phrase, *pūꞌoko haka tere* 'head caus run' is a compound noun with the sense 'leader, head', consisting of a noun and a modifying verb. The adjective phrase *taꞌe tano* 'not correct' = 'unrighteous' modifies this compound noun.

(107) *te* art [*pūꞌoko* head *haka* caus *tere* ] run [*taꞌe* conneg *tano* ] correct *era* dist *o* of *te* art *hare* house *ture* judgment 'the unrighteous head of the courthouse' [Luke 18:6]

We may therefore assume a distinction between modifiers as part of a compound and modifiers in a post-nominal modifier position; in other words: noun adjuncts versus noun phrase adjuncts. This coincides with another syntactic difference: the order of elements within a compound is fixed, while the relative order of adjectives is sometimes free ((135–136) on p. 264).

<sup>21</sup> Cf. Vernaudon (2011: 325), who gives examples of an adjective, noun and verb modifying a noun in Tahitian.

5.7 Modifiers in the noun phrase

Now the distinction between nominal and adjectival modifiers is not absolute. In the following example, the adjective *ꞌāpī* and the proper noun *rapa nui* both modify the noun *poki*; there is no functional difference between the two modifiers.

(108) *Poki* child *ꞌāpī* new *te* art *meꞌe* thing *era,* dist *poki* child *rapa* Rapa *nui* Nui *te* art *meꞌe* thing *ena.* med 'That one is a young child, that one is a Rapa Nui child.' [R416.238]

In the following example, the noun+adjective combination *tuki tōumāmari* is further removed from the noun than the adjective *teatea*; here the modifying noun *tuki* is obviously not incorporated into the head noun *pokopoko*, but is a modifier on the same level as adjectives.

(109) *Ka* imp *maꞌu* carry *mai* hither *taꞌa* poss.2sg.a *pokopoko* container *teatea* white:red *tuki* dot *tōuamāmari.* yellow 'Bring your white, yellow-dotted bowl.' [Notes]

Also, the fact that a certain noun+noun combination is a semantic unit does not imply that it is necessarily a syntactic unit as well, occupying the head position as a whole. Syntactic structure does not always mirror semantic structure. The underlined expressions in the following examples have an idiomatic sense, but they are not a syntactic unit. While there is a compound *vare/ŋao* 'slimy' + 'neck' = 'to crave', here the same two elements are used in a verb + subject construction:<sup>22</sup>


Thus, the fact that a collocation is a semantic unit does not imply that its parts are in a single position in the noun phrase. Moreover, some noun-adjective combinations also express a single concept, just like noun + noun compounds.


In conclusion, there is no absolute distinction between modifying nouns and modifying adjectives. However, the following things are clear:

<sup>22</sup> However, idioms like this do have a tendency to become syntactically united. In newer texts the expression *mate te manava* is not found; instead, the compound verb *manava mate* is used.

<sup>23</sup> This compound was borrowed from Tahitian as a whole. 'Word' is the Tahitian sense of *parau*; in Rapa Nui, *parau* on its own does not mean 'word', but 'paper', 'document' or 'authority'.

#### 5 The noun phrase


Because of the distinction between 3 and 4, the noun phrase chart in §5.1 places compounds as a whole in the head position, while modifying adjectives are placed in a separate slot.

In the following sections, the different types of modifiers will be discussed: §5.7.2 deals with compounds, §5.7.3 with modifying adjectives. Even though this section is part of the chapter on noun phrases, verb compounds (i.e. compounds with a verb as head and occurring in a verbal context) will be discussed in §5.7.2.4.

### **5.7.2 Compounds**

As shown in the previous section, compounds in Rapa Nui are formed by simply juxtaposing two words. The head word comes first, then the modifier. The structure may be recursive: the modifier may itself be the head to a second modifier. The modifying element may be a noun or verb. Most compounds are nouns (i.e. they have a noun as their head), but the discussion in these sections includes examples of compound verbs and adjectives as well.

A distinction can be made between lexical and syntactic compounds (see Dryer 2007b: 175). Lexical compounds have a meaning which is not predictable from the meaning of their parts, while syntactic compounds are productive constructions with a predictable meaning. Both are found in Rapa Nui and are discussed separately below. There is, however, no sharp distinction between the two. Certain compounds have a somewhat specialised, not quite predictable sense, yet it is easy to see how this sense could have arisen from the sense of their components. In fact, the distinction between lexical and syntactic compounds can be thought of as a continuum. At one end are completely predictable and productive compounds, at the other end are compounds with a completely unpredictable (e.g. figurative) sense. Table 5.4 gives examples illustrating different points along this continuum.

<sup>24</sup> For compounds written as one word in the standard orthography, the parts are separated by a slash.

#### 5.7 Modifiers in the noun phrase


Table 5.4: Syntactic and lexical compounds

#### **5.7.2.1 Lexical compounds**

As discussed in the previous section, there are various degrees of lexical compounding. Some compounds are specialised in meaning (i.e. the sense of the compound cannot be predicted from the sense of the parts), but it is still clear how their meaning is derived from the meaning of the parts:<sup>25</sup>


In the last two examples above, the relation between the two words is not that between head and modifier. In *matavai*, the second noun *vai* is semantically the head.<sup>26</sup> In *repahoa*, both components are synonyms which together yield a third synonym.

The compound may also be a verb or adjective:


Some compounds are more than specialised in meaning: their sense is to a greater or lesser degree opaque.


<sup>25</sup> In the tables in this section, the second column gives the meaning of the component parts, the third column the meaning of the whole compound.

<sup>26</sup> Another example is *motore vaka* 'motor boat', noted by Fischer (2001a: 322); this is probably a calque from English.

#### 5 The noun phrase

Opacity goes even further in compounds where one or both components do not occur at all in Rapa Nui (at least, not in the sense underlying the compound); the origin of these components may or may not be reconstructible.


These compounds either developed at a stage when both components were still in use in the sense they had in the protolanguage, or else they were inherited from the protolanguage as a whole. Opaque compounds may also have a more recent origin, being borrowed as a whole. One such word is *hare toa* 'store', borrowed from Tahitian. The first part means 'house' (Rapa Nui *hare*, Tahitian *fare*), the second part means 'store' in Tahitian (from English) but is not used in other contexts in Rapa Nui.<sup>27</sup>

In other cases, both components are known as Rapa Nui words, but one of them is no longer in use, or at least archaic.


Such compounds function practically as single words: the original sense of their parts no longer plays a role.

Near the other end of the spectrum, i.e. similar to syntactic compounds, are compounds which are quite transparent in meaning, but which are still lexicalised to a certain degree; that is, they may be a single unit in the mental lexicon of speakers of the language. Though it is impossible to say exactly whether a compound is or is not lexicalised, two indications for lexicalisation of a compound are:


Some examples are:


<sup>27</sup> *hare toa* is written as two words, because (at least some) speakers know the origin and meaning of the second part.

5.7 Modifiers in the noun phrase

#### **5.7.2.2 Syntactic compounds**

Syntactic compounds are transparent in sense: their meaning can be predicted from the meaning of the parts. Syntactic compounds are productive and may express a wide variety of semantic relations. Here are some examples:


In syntactic compounds, the plural marker *ŋā* may intervene between the two nouns:

(120) *Taꞌe* conneg *he* pred *aŋa* work *ŋā* pl *viꞌe* woman *rā.* dist 'It's not women's work.' [R347.103]

As illustrated in the previous section, the second element of a lexical compound may also be a verb. This also happens with syntactic compounds. The noun may refer to a location where the event expressed by the verb takes place (as in a and b below), or an instrument used to perform the action expressed by the verb (as in c).


Compounds may also consist of three members. The third word is a noun (122), verb (123) or adjective (124) modifying the second noun; together they modify the head noun. (On modifying verbs, see §5.7.2.3 below.)


#### 5 The noun phrase

#### **5.7.2.3 Incorporation of objects and verbs**

A verb as modifier may in turn be followed by its object. Like any modifying noun, the object is a bare noun, not marked with a determiner and/or object marker. This is a case of object incorporation: the object loses its object marking and its status as a noun phrase, and is directly adjoined to the verb.


A combination of noun and verb modifiers and object incorporation may lead to even longer compounds, as the following examples show:

(126) *ꞌi* at *te* art *mahana* day *taꞌe* conneg *noho* stay *ꞌi* at *te* art *kona* place *ꞌāua* enclosure [*ꞌoka* to\_plant *kai*] food *nei* prox *ꞌāꞌana* poss.3sg.a

'on a day when he did not stay in his garden plot' [R381.004]

(127) *Hai* ins *meꞌe* thing *potupotu* piece:red *niuniu* wire:red *taratara* spine:red [*haro* pull *ꞌāua* ] enclosure *ena* med *e* ipfv *aŋa* make *era* dist *te* art *meꞌe* thing *vivi* chain *rikiriki.* small:pl:red 'With pieces of barbed fence wire they made little chains.' [R364.005]

It is also possible to incorporate the verb into the noun which is semantically its object.

These compounds are unusual in that the noun is syntactically the head of the compound (it retains its status as a regular noun, i.e. head of a noun phrase), even though it is semantically an argument of the verb.<sup>28</sup> These compounds may appear in any nominal context, just like any noun or noun compound. (In (128–129), the compound is the predicate of a nominal clause.)

(128) *ꞌI* at *tōꞌona* poss.3sg.o *mahana* day *he* ntr *ai* exist *mai* hither *te* art *aŋa* work *he* pred *ꞌāua* enclosure *titi,* build *ꞌo* or *he* pred *rau* leaf *kato.* pick 'On certain days there were jobs like making fences or picking leaves.' [R380.084]

<sup>28</sup> For a somewhat similar mismatch between syntax and semantics, cf. the nominal purpose construction discussed in §11.6.3. There as well as here, an event is expressed by a nominal construction, with one of the arguments of the verb in question as syntactic head. Both of these are among the many instances in Rapa Nui where a nominal construction serves to express an event (§3.2.5).

5.7 Modifiers in the noun phrase

(129) *He* pred *kai* food *toke* steal *nō* just *mai* hither *o* of *te* art *taŋata* man *te* art *aŋa.* do 'Stealing the people's food was what she did all the time.'<sup>29</sup> [R368.017]

Noun + verb compounds are similar to bare relative clauses (§11.4.5): in the latter, the verb – which is always initial in relative clauses – is not preceded by an aspectual; just as in a compounds, it follows immediately after the head noun. There are two important differences, however.

In the first place, a bare relative clause is still a clause: the verb is part of a verb phrase which may contain postverbal particles, such as *iho* in (130). Moreover, arguments of the verb may be expressed by independent case-marked noun phrases, such as the subject *e ia* (with agentive marking) in (131).


By contrast, a modifying verb in a compound does not form a clause. No other VP elements can be included.

Secondly, a bare relative clause expresses an event which happens or happened at a specific time, whether once or repeatedly. By contrast, an incorporated verb denotes something which characterises the noun, irrespective of whether the event has really taken place or not. For example, a plot of land may be *ꞌāua ꞌoka kai* (garden for planting food, (125)d), even when nothing has been planted yet.

#### **5.7.2.4 Compound verbs**

Though the vast majority of compounds in Rapa Nui function as nouns, compound verbs are also found. Some of these were mentioned in §5.7.2.1, e.g. the lexical compound *tunuahi* (cook + fire) 'to roast on a fire'.

Most compound verbs consist of a verb + noun. The noun may have various semantic roles in relation to the verb; interestingly, it is usually not the direct object, but often the instrument with which the action is performed:

(132) *He* ntr *toꞌo* take *mai* hither *era* dist *he* ntr *tunu* cook *pani,* pan *he* ntr *tunuahi.* cook.fire

'They took the food and cooked it in the pan, roasted it on a fire.' [R107.049]

<sup>29</sup> The noun *kai* has a genitive modifier *o te taŋata*; this is leapfrogged over by the incorporated verb. The construction is similar to nominal purpose constructions (§11.6.3).

#### 5 The noun phrase

(133) *He* ntr *tunu* cook *māꞌea* stone *vera,* hot *haka* caus *hopu* bathe *i* acc *te* art *poki* child *hai* ins *vai* water *vera.* hot 'He cooked (the water) with hot rocks, and bathed the child with hot water.' [Mtx-1-07.016]

In the following example, the modifier *rapa nui* can also be considered as an instrument in a loose sense.

(134) *ꞌE* and *nuꞌu* people *taꞌe* conneg *rahi* many *ꞌi* at *te* art *raꞌā* day *nei* prox *e* ipfv *ꞌaroha* greet *rapa* Rapa *nui* Nui *nei.* prox 'Few people today greet each other in Rapa Nui (with this Rapa Nui greeting).' [R530.038]

That these combinations are compounds is clear from the fact that the noun is not preceded by a determiner, nor by a preposition indicating its semantic role. (For example, the instrumental role would normally be indicated by *hai*.) Also, postverbal particles follow the noun (*nei* in (134) above), showing that the noun has been incorporated into the verb phrase.

### **5.7.3 Modifying adjectives**

As discussed in §5.7.1, modifying adjectives are usually semantically different from modifying nouns. This section discusses a few issues concerning adjectives in the noun phrase.

Several elements occurring in the adjective position are discussed elsewhere:


#### **5.7.3.1 Multiple adjectives**

As (109) shows, the noun phrase may contain more than one adjective. The order of the adjectives is not fixed:


5.7 Modifiers in the noun phrase

As the translation shows, there is a subtle difference between the two examples above. The adjective closest to the noun denotes the quality that is most fundamental in the context; this noun + adjective combination is in turn modified by the second adjective. This is in line with the general principle noted in §5.7.1: elements closest to the noun are semantically closer to it as well; they form a unit with the noun which may in turn be modified by other modifiers.

Cases of multiple adjectives are uncommon, though. The contrasting examples above were given during a discussion session. An example from the text corpus is the following:

(137) *He* ntr *uꞌi* look *mai* hither *i* acc *te* art *ꞌohe* bamboo *tītika* straight *rivariva.* good:red

'You look (= one looks) for a straight, good bamboo stick.' [R360.015]

More commonly, multiple adjectives are separated by a pause or the conjunction *ꞌe* 'and'; for an example, see R215, sentence 02 in Appendix A (p. 575).

#### **5.7.3.2 The adjective phrase**

The adjective constituent which modifies the noun is not always a bare adjective, but can be a phrase containing other elements: adverbs and/or particles.

The adjective may be preceded by a modifier of degree: *ꞌapa* 'to a moderate degree, somewhat, sort of' or *ꞌata* 'to a higher degree, more'. *ꞌAta* is discussed in §3.5.1.1; here are examples of *ꞌapa*:


America.' [R350.003]

The adjective may be followed by an intensifying adverb *rahi* 'much', *riꞌariꞌa* 'very, terribly', *taparahi-taꞌata* 'terribly' (a Tahitian phrase which literally means 'killing people'), *tano* 'in a moderate degree', *mau* (in (142), emphasised by the identity particle *ꞌā*):

(140) *Meꞌe* thing *rivariva* good:red *rahi* much *te* art *meꞌe* thing *nei* prox *mo* for *te* art *oraraꞌa* life *o* of *te* art *mahiŋo* people *o* of *Rapa* Rapa *Nui.* Nui

'This was something very good for the life of the people of Rapa Nui.' [R231.314]

#### 5 The noun phrase


The adjective may also be followed by a prepositional phrase as in (143):

(143) *ꞌI* at *Haŋa* Hanga *Roa* Roa *te* art *nonoho* pl:stay *haŋa,* nmlz *ꞌi* at *te* art *kona* place [*hāhine* near [*ki* to *te* art *ꞌōpītara* hospital *tuai* ancient *era* ] ]*.* dist

'They lived in Hanga Roa, in a place close to the old hospital.' [R380.003]

### **5.8 Adverbs and** *nō* **in the noun phrase**

### **5.8.1 Adverbs**

As the position chart in §5.1 shows, after the quantifier phrase there is a position for adverbs. The only adverbs found here are *hakaꞌou* 'again', *takoꞌa* 'also' and *mau* 'really'. *Hakaꞌou* and *takoꞌa* are more common in verb phrases, but do appear in noun phrases occasionally; they are discussed in sections §4.5.3.4 and §4.5.3.2, respectively. *Mau* may co-occur with another adverb (just as in the verb phrase, §4.5.1), hence its separate slot in the noun phrase chart in §5.1.

### **5.8.2 The limitative marker** *nō*

*Nō* is a marker of limitation, which is also common in the verb phrase (§7.4.1). In the noun phrase, *nō* has a number of uses. In several constructions it serves to restrict the reference of a noun phrase, though – as will be illustrated below – not necessarily the noun phrase it occurs in. In other cases it is used in the sense 'just, simply' in much the same way as in verb phrases.

#### **5.8.2.1 'The only one'**

In initial subject NPs, *nō* indicates that the set referred to by the noun phrase has only one entity, viz. the one described in the rest of the sentence. The sentence can be paraphrased as: 'There is only one [NP], and that is [rest of sentence]', or more simply: '[rest of sentence] is the only [NP].' For example in (144): 'There was only one thing on board, and that was a piece of pumpkin', or 'A piece of pumpkin was the only thing on board.'

5.8 Adverbs and *nō* in the noun phrase


'And the only name we knew him by, was Sebastián Englert.' [R375.005]

#### **5.8.2.2 'Only that one'**

With noun phrases in other positions, *nō* signals that the rest of the sentence applies only to the entities described by the noun phrase with *nō*. The sentence can be paraphrased as: 'only for [NP] is it true that [rest of sentence]'. For example in (146): 'Only for lobster and crab is it true that they fished with it'; in other words: 'Lobster and crabs were the only (bait) they fished with.'


This is also common with *nō* in predicate noun phrases. *Nō* indicates that there is only one entity to which the subject applies, viz. the one referred to in the noun phrase containing *nō*. The sentence can be paraphrased as: 'Only [predicate] is [subject]', or more naturally: '[predicate] is the only [subject].' This happens for example in the identifying clause (§9.2.2) in (148) below: 'Only she was the new child inside' = 'She was the only new child inside.'

(148) *Ko* prom *ia* 3sg *nō* just *te* art *poki* child *ꞌāpī* new *o* of *roto.* inside 'She was the only new child inside (the class).' [R151.020]

#### **5.8.2.3 'Just'**

In all cases above, *nō* limits the reference of a noun phrase. It may also have a weaker sense: 'just, simply, no more than':

(149) *He* pred *tāvini* servant *nō* just *māua* 1du.excl *ōꞌou.* poss.2sg.o 'We are just your slaves.' [R214.015]

5 The noun phrase

(150) *He* pred *repahoa* friend *nō* just *au* 1sg *ōꞌou.* poss.2sg.o 'I am just your friend.' [R308.032]

#### **5.8.2.4 Contrastive use**

*Nō* is used in a number of expressions indicating a contrast. The noun phrase *te N nō*, placed initially in the clause, functions as a connective which signals that the following clause is an exception to what has been stated before. An appropriate translation is 'however'. The noun may express how this contrast is to be evaluated, whether negatively as in (151), positively as in (152), or neutral as in (153). In (151), the contrast is reinforced with the Spanish conjunction *pero*.


### **5.9 The identity marker** *ꞌā***/***ꞌana*

*ꞌĀ* and *ꞌana* are variant forms of the same particle.<sup>30</sup> This particle functions as a continuous marker in the verb phrase and as an identity marker in the noun phrase. This section deals with its use in the noun phrase; its use in the verb phrase is discussed in §7.2.5.5.

The choice between *ꞌā* and *ꞌana* is partly a stylistic one. *ꞌĀ* is somewhat more informal (and therefore more common in oral language), while *ꞌana* is more formal. Rhythm may also play a role: in some contexts a one-syllable particle may yield a better rhythm than a two-syllable one, or the opposite.

Other euphonic effects may play a role as well. For example, after the particle *ena*, one usually finds *ꞌā*, not *ꞌana*: the alliterating *ena ꞌana* is avoided.<sup>31</sup>

<sup>30</sup> In other Eastern Polynesian languages, cognates of *ꞌana* are used in the verb phrase, but not in the noun phrase (see Footnote 21 on p. 328).

<sup>31</sup> By contrast, after *era* both *ꞌā* and *ꞌana* are commonly used.

5.9 The identity marker *ꞌā*/*ꞌana*

Part of the difference is ideolectical, as shown by the fact that some (groups of) texts show a strong preference for one variant. For example, in Ley *ꞌā* is about six times as common as *ꞌana* (296 against 58 occurrences), while in MsE *ꞌana* is predominant (121 against 23 occurrences). One recent text (R539) shows an extraordinary preference for *ꞌana* (557x *ꞌana* against 30x *ꞌā*), while some oral texts use *ꞌā* almost exclusively. In most texts, however, the two occur in more equal proportions, though *ꞌā* is more common overall.

Concerning the use of *ꞌā*/*ꞌana*: with a pronoun it may be used when the pronoun has a reflexive sense, i.e. is coreferential with the subject of the clause. The pronoun may be, for example, the direct object or an oblique argument:


However, *ꞌā*/*ꞌana* as such is not a reflexive marker: a noun phrase containing *ꞌā* does not need to be in the same clause as its antecedent. In the following example, *ꞌā* appears with a subject pronoun, coreferential with the subject of the preceding sentence:

(156) *He* ntr *kī* say *atu* away *ia* then *e* ag *tōꞌoku* poss.1sg.o *koro* Dad *era* dist *ki* to *a* prop *au…* 1sg *ꞌAi* there *ka* cntg *kī* say *hakaꞌou* again *atu* away *e* ag *ia* 3sg *ꞌā…* ident 'Then my uncle (lit. father) said to me…. Then he himself said again…' [R230.254-6]

It is more accurate to analyse *ꞌā/ꞌana* in broader terms: it serves as a marker of identity. As such, it can be used in different ways. Sometimes it indicates that the referent of the noun phrase is identical to another referent in the same clause, as in the reflexive examples (154–155) above. In other cases it indicates that the referent of the noun phrase is identical to another referent mentioned earlier in the text, as in (156). It may also underline that the referent is identical to an entity known in some other way ('the same'). Some examples:

(157) *I* pfv *oti* finish *era* dist *te* art *kai,* eat *he* ntr *haꞌuru* sleep *rō* emph *ꞌai* subs *a* prop *Taparahi* Taparahi *ꞌi* at *tū* dem *kona* place *era* dist *ꞌā.*

ident

'When he had finished eating, Taparahi slept at that same place.' [R250.032]

#### 5 The noun phrase

(158) *I* pfv *poreko* born *ai* pvp *a* prop *ia* 3sg *ꞌi* at *te* art *motu* islet *mau* really *nei* prox *ꞌā* ident *ꞌi* at *te* art *matahiti* year *1922.* 1922 'He was born on this very same island here in the year 1922.' [R487.041]

In (157), the place where Taparahi sleeps is the same place where he has just eaten. In (158), the island where the person in question is born is the same island where the story is being told.

These examples also illustrate the syntax of *ꞌā*/*ꞌana*: when *ꞌā*/*ꞌana* follows a noun, the noun phrase also has a demonstrative: usually prenominal (*tū* in (157)), occasionally postnominal (*nei* in (158)). When *ꞌana* follows a pronoun, no demonstrative is used.

After a possessive pronoun, *ꞌā* (often preceded by *mau*) stresses the identity of the possessor: 'one's own'.

(159) *¿E* ipfv *ai* exist *rō* emph *ꞌā* ident *tuꞌu* poss.2sg.o *vaka* boat *ōꞌou* poss.2sg.o *mau* really *ꞌā?* ident 'Do you have your own boat?' [Notes]

One more nominal construction in which *ꞌā/ꞌana* is used, is *ko te V iŋa ꞌā/ꞌana* (§3.2.3.1.1).

### **5.10 The deictic particle** *ai*

The deictic particle *ai* is used when pointing at something; it can only be used when the entity referred to is visible.


As these examples show, *ai* is usually preceded by a postnominal demonstrative (*era*, *nei* or *ena*) or an identity marker (*ꞌā* or *ꞌana*).

This particle is similar in function to the sentence-initial particle *ꞌai* 'there is'. Hhe particles are phonetically different, however: NP-final *ai* has no glottal, while initial *ꞌai* does. Even so, the two could be etymologically related (§2.2.5 on glottals in particles). Another possibility is that final *ai* has developed from the existential verb *ai*. This verb is used postnominally to construct certain types of relative clauses (§11.4.3):

(162) *te* art *nuꞌu* people *ai* exist *o* of *te* art *vaka* boat 'the people who had a boat' [R200.086]

5.11 Heavy shift

It is conceivable that the deictic particle *ai* developed from a relative clause which was truncated, and of which only the verb was left.

### **5.11 Heavy shift**

Sometimes longer subphrases are placed at the end of the noun phrase. This is in accordance with a universal tendency to move long constituents to the end of the phrase or clause, a phenomenon known as heavy shift (Payne 1997: 326).

In (163) below, the noun is modified by a complex adjective phrase 'smaller than it'. The adjective itself is in its normal position, but its complement *ki a ia* 'than it', which expresses the standard of comparison, is placed after the postnominal demonstrative *era*. In (164), the whole adjective phrase is placed at the end of the noun phrase, even after the relative clause:


'At that moment Kalia saw something white, which was seen on the ship, bigger than a man.' [R345.061]

### **5.12 Appositions**

### **5.12.1 Common nouns in apposition**

Common noun phrases in apposition are never preceded by a *t*-determiner. They may be marked in several ways: without any marker (bare appositions), with the predicate marker *he*, or with the prominence marker *ko*.

**Bare appositions** Bare appositions may have generic reference, indicating that the head noun belongs to a certain class of referents. In (165), the apposition tells that Renga Roiti belongs to the class of female children.

#### 5 The noun phrase


They may also have specific reference, identifying the head noun with a certain referent. For example, the apposition in (167) tells that Papeete is the same place as the capital of Tahiti.

(167) *te* art *kona* place *ko* prom *Pape ꞌete,* Papeete *kona* place *rarahi* important *o* of *Tahiti* Tahiti 'the city of Papeete, the capital of Tahiti' [R231.045]

*He***-marked appositions** Appositions introduced by *he* may also be either specific as in (168) or generic as in (169). In the Bible translation, appositions tend to be marked with *he*, possibly because the translation employs a relatively polished/formal style.

(168) *He* ntr *oho* go *ki* to *Vērene* Bethlehem *ꞌi* at *Hūrea,* Judea *he* pred *kona* place *poreko* born *o* of *tōꞌona* poss.3sg.o *hakaara* ancestor *ko* prom *Tāvita.* David

'They went to Bethlehem in Judea, the birth place of his ancestor David.' [Luk. 2:4]

(169) *He* ntr *haka* caus *hāhine* near *rā* dist *nuꞌu* people *ki* to *a* prop *Feripe,* Philip *he* pred *kope* person *o* of *Vetetaira.* Bethsaida 'Those people approached Philip, a man from Bethsaida.' [John 12:21]

Bare and *he*-marked appositions are used as the equivalent of nonrestrictive relative clauses, clauses which provide information about a noun phrase without limiting its reference.<sup>32</sup> In Rapa Nui, relative clauses must be restrictive, and therefore they cannot be attached to nouns which already have a unique reference, like proper names. To add a clause providing more information to such a noun, a noun with generic meaning (e.g. *meꞌe* 'thing', *kope* 'person') is placed in apposition; a relative clause is attached to this apposition, limiting the reference of the generic noun:

(170) *He* ntr *oho* go *ia* then *a* prop *Vakaiaheva* Vakaiaheva *ki* to *Rano* Rano *Raraku,* Rarako *kona* place *[ꞌi* at *ira* ana *te* art *kape* boss *e* ipfv *noho* stay *era].* dist

'Vakaiaheva went to Rano Raraku, the place where the boss lived.' [R440.028]

<sup>32</sup> Andrews (2007a: 207) does not consider nonrestrictive clauses as relative clauses, as relative clauses (in his definition) delimit the reference of the noun phrase.

5.12 Appositions

(171) *He* ntr *turu* go\_down *a* prop *Rovi,* Rovi *he* ntr *taŋata* person *[hāpaꞌo* care\_for *i* acc *te* art *poki* child *ꞌa* of.a *Hotu* Hotu *ꞌariki].* king 'Rovi came down, the man who took care of the child of king Hotu.' [R422.002]

*Ko***-marked appositions** Sometimes a common noun apposition is marked by the prominence marker *ko* (§4.7.12), followed by a determiner.<sup>33</sup> This happens when the apposition refers to an entity uniquely identifiable by the hearer (cf. §9.2.1 on the distinction between *ko-*marked and *he-*marked noun phraes).

(172) *He* ntr *tuꞌu* arrive *mai* hither *te* art *ꞌavione* airplane *raꞌe* first *ko* prom *te* art *ꞌavione* airplane *ena* med *e* ipfv *kī* say *ena* med *he* pred *DC 10.*

DC 10

'The first airplane, the airplane called DC 10, arrived.' [R203.062]

(173) *te* art *mahiŋo* people *i* pfv *haka* caus *maraŋa* scattered *ena* med *ꞌi* at *ruŋa* above *i* at *te* art *henua* land *nei* prox *ko* prom *te* art *kāiŋa* homeland

'the people who spread over the land, over the homeland' [R350.016]

### **5.12.2 Proper nouns in apposition**

If the apposition is a proper noun, it is introduced by *ko*. This is to be expected, as proper nouns are inherently uniquely identifiable in a given context (§9.2.1).


The examples above illustrate the most common way to express a combination of a common noun and a name: the common noun is the head noun; the name follows as apposition, introduced by *ko*. There are exceptions though: sometimes *ko* is not used as in (176); sometimes the name precedes the common noun as in (177):

(176) *Te* art *kona* place *noho* stay *o* of *te* art *ꞌariki* king *Hotu* Hotu *Matuꞌa* Matu'a *ꞌi* at *Hiva* Hiva *Maraꞌe* Mara'e *Reŋa.* Renga 'The place where king Hotu Matu'a lived in Hiva was Mara'e Renga.' [Ley-2-01.002]

<sup>33</sup> *Ko* in appositions is common in Polynesian languages, see Clark (1976: 45).

#### 5 The noun phrase

(177) *He* ntr *turu* go\_down *a* prop *Rovi,* Rovi *he* pred *taŋata* man *hāpaꞌo* care\_for *i* acc *te* art *poki* child *ꞌa* of.a *Hotu* Hotu *ꞌariki.* king 'Rovi came down, the man who took care of the child of king Hotu.' [R422.002]

### **5.13 The proper noun phrase**

Proper noun phrases are those headed by proper nouns. As discussed in §3.3.2, the class of proper nouns in Rapa Nui not only includes names of persons, but a number of kinship terms and other nouns as well, as well as pronouns. These items are grouped together on syntactic grounds: they do not take the determiner *te*, but the proper article *a*.

What proper nouns have in common semantically, is that they refer to a unique entity. Unlike common nouns, which denote a property or class, they do not need a determiner to be referential. Anderson (2004: 456) argues that proper names and pronouns belong to the same category as determiners and deictics like *this*: while determiners turn a common noun into a referential expression, proper nouns are inherently referential. While common nouns can function as predicates, proper nouns cannot. In Rapa Nui this means that they cannot take the predicate marker *he*. And as they do not need a determiner to acquire referentiality, they do not take the common noun article *te*.

In §5.13.1, the structure of the proper noun phrase is discussed. §5.13.2 examines the distribution and structural position of the proper article *a*.

### **5.13.1 Structure of the proper noun phrase**

As Dixon (2010a: 108) points out, proper nouns usually have fewer syntactic possibilities than common nouns. In Rapa Nui, most proper noun phrases consist only of a proper noun preceded – if syntactically appropriate – by the proper article *a*. Even so, the proper noun phrase may contain a range of other elements as well. The full structure of the proper noun phrase, including the preceding preposition, is shown in Table 5.5 and Table 5.6.


Table 5.5: The proper noun phrase: prenominal elements

The head is obligatory, and so are the preposition and the proper article, if required by the syntactic context. All other elements are optional.

#### 5.13 The proper noun phrase


Table 5.6: The proper noun phrase: postnominal elements

With the exception of the proper article, all items occur in the common noun phrase as well. They have been discussed in the preceding sections (see the paragraph references in the tables).

The following examples illustrate different possibilities; each word or phrase is numbered according to the numbering in Table 5.5 and Table 5.6.


Most of these elements (except for the *kuā*, determiners, and genitive phrases) may occur with pronouns as well. (See §3.3.2: pronouns belong syntactically to the class of proper nouns.). A few examples:

(183) *ko* prom *ia* 3sg *takoꞌa* also 'he (knew it) as well' [R620.037] 5 The noun phrase


The determiner position plays a marginal role in personal noun phrases. It can only be filled by possessive pronouns, and only when the head noun is a kinship term; see (198) on p. 279. The post-nominal elements are uncommon as well.

### **5.13.2 The proper article** *a*

This section discusses the proper article *a*. <sup>34</sup> In §5.13.2.1 the contexts are listed in which this article occurs. In §5.13.2.2 the question is raised whether *a* is a determiner.

According to Clark (1976: 58), *a* occurs in almost all Polynesian languages preceding a personal noun or pronoun after certain prepositions; in a number of Nuclear Polynesian languages it is also used in the nominative case. Both are true for Rapa Nui as well, see below.<sup>35</sup> The nominative case marker *ꞌa* in Tongan reflects the same PPN particle.<sup>36</sup>

#### **5.13.2.1 Contexts in which** *a* **is used**

The proper article *a* is not the proper noun equivalent of the common noun article *te*: it is not used in the same contexts where a common noun would have the article *te*. The use of the proper article is limited to the following contexts:

**Subject** The proper article is used when the noun phrase or pronoun is subject of a verbal or nonverbal clause.

(186) *He* ntr *tutu* set\_fire *a* prop *nua* Mum *i* acc *te* art *ahi.* fire 'Mum lighted the fire.' [R232.047]

<sup>34</sup> In Polynesian linguistics, *a* is more commonly called "personal article"; I use "proper article", a term suggested by Dixon (2010a: 108), as *a* is exclusively used with the class of proper nouns. The term "proper" seems more appropriate than "personal": this class is not defined by 'personal' (i.e. human) reference, but by its 'proper', name-like character.

<sup>35</sup> In languages where *a* is only used after prepositions, it tends to be considered (and written) as one word together with the preceding *i* or *ki*: *ia*, *kia*. See e.g. Elbert & Pukui (1979: 107) for Hawaiian, Lazard & Peltzer (2000: 186) for Tahitian.

<sup>36</sup> Fischer (1994: 429) presumes that the Old Rapa Nui form was *ꞌa*, which was replaced by Tahitian *a* in Modern RN. This is based on the fact that the form reconstructed for PPN is *\* ꞌa*; the latter is based on the Tongan nominative marker *ꞌa*. Notice, however, that the form does not have a glottal in other languages which normally preserve the PPN glottal (Rennell, East Uvean and East Futunan). It is thus well possible that *a* had lost the glottal by PNP. In any case, the glottal is unstable in particles in Polynesian languages, especially in initial particles, and may disappear and (re)appear unpredictably (§2.2.5; Clark 1976: 20). Notice also that in Tahitian *a* has a more limited distribution than in Rapa Nui: it is only used after prepositions.

5.13 The proper noun phrase

(187) *ꞌI* at *te* art *ahiahi* afternoon *he* ntr *oho* go *a* prop *au* 1sg *he* ntr *tatau* squeeze *i* acc *te* art *ū.* milk 'In the late afternoon I go and milk the cows.' [R334.277]

With personal pronouns used as subject, the proper article is sometimes left out. This happens especially with the plural pronouns and *koe*, less commonly with *au*, never with *ia*.


Usually, the proper article is omitted before the subject pronoun of an imperative clause, as in (190).

(190) *Ka* imp *oho* go *koe.* 2sg 'Go.' [Notes]

**After prepositions ending in** *-i* <sup>37</sup> When a proper noun is preceded by *ꞌi/i* 'in, at', the accusative marker *i*, *mai* 'from' or *ki* 'to', the proper article is used. When the preposition is *mai*, the preposition *i* is added between *mai* and the proper article (see (278) on p. 215).


The proper article is **not** used after any other preposition: agentive *e*, vocative *e*, genitive *o*, the prominence marker *ko*, and the prepositions *mo/mā* 'for', *a* 'by', *ꞌo* 'because of', *pe* 'toward', *pē* 'like', *hai* 'with'. The proper noun or pronoun follows immediately after these markers:

<sup>37</sup> In almost all Polynesian languages *a* occurs after *i*, *ki* and *mai*, but not after other prepositions. Clark (1976: 58) suggests this can be explained by a morphophonemic rule which deleted *a* after prepositions ending in a non-high vowel. This rule must have been operative at a stage prior to Proto-Polynesian, as it affected all Polynesian languages. The fact that *a* in Rapa Nui does not occur after *hai* 'with' shows that the rule is no longer productive.

5 The noun phrase


When the noun phrase or pronoun is used in an elliptic construction, *a* is used if a context is implied where it would normally be used. In the following example, the reply *a au* implies the clause 'I climbed the crater', in which the pronoun is subject of a verbal clause, a context in which *a* would be used.

(195) *—¿Pē* like *hē* cq *koe* 2sg *i* pfv *iri* ascend *ai* pvp *ki* to *te* art *rano?* crater\_lake *—¿A* prop *au?* 1sg *A* by *raro* below *ꞌā,* ident *a* by *pie.* foot '—How did you climb the crater? —Me? On foot.' [R623.015–017]

In other contexts, isolated proper nouns are marked by the prominence marker *ko* (§4.7.12.1).

#### **5.13.2.2 Is** *a* **a determiner?**

In a number of respects, the proper article shows complementary distribution with the common noun article *te*:


It seems a logical step to analyse *a* as an article, and indeed, in Polynesian linguistics *a* is often labelled as "personal article" (see e.g. Clark 1976: 58, Bauer, Parker & Evans 1993: 109, Cablitz 2006: 62). There are, however, important differences in distribution between *te* and *a* in Rapa Nui. For one thing, after the prepositions *e*, *o*, *ko*, *a*, *o* and *pe*, the article *te* is obligatory (§5.3.2.1), but *a* is not used.

It is even questionable whether *a* is a determiner at all. For one thing, the collective marker *kuā* precedes the determiner *te*, but occurs after the proper article:

(196) *a* prop *kuā* coll *Tiare* Tiare 'Tiare and the others' [R315.227] (197) *kuā* coll *te* art *kape* captain 'the captain and company' [R416.864]

Secondly, while *a* does not co-occur with the article *te*, it does co-occur occasionally with possessive pronouns which are in the determiner position:

(198) *He* ntr *ꞌui* ask *iho* just\_then *ia* then *ōꞌoku* poss.1sg.o *ki* to *a* prop *tōꞌoku* poss.1sg.o *koro* Dad *era…* dist 'Then I asked my uncle (lit. father) again…' [R230.121]

These data show that *a* is not in the determiner position, but in an earlier position in the noun phrase. It can thus only be called "proper article" in a loose way, without implying that it occupies the same position as other articles.

*A* is not a preposition or case marker, either, as it occurs both with subject nouns/pronouns and after several prepositions, such as the accusative marker *i*.

### **5.14 Conclusions**

The preceding sections have shown that the structure of the noun phrase in Rapa Nui is complex, with no less than seventeen different slots. Apart from the head, the only element which is obligatory in most contexts is the determiner. In the determiner position, two fundamentally different elements occur: *t*-demonstratives and the predicate marker *he*. The former mark referentiality (not specificity or definiteness; the latter is indicated by demonstratives), while *he* marks a noun phrase as non-referential. Indefiniteness is sometimes indicated by the numeral *e tahi* 'one'.

In subject position and after most prepositions, the determiner is obligatory. On the other hand, the determiner cannot co-occur with prenominal numerals and certain quantifiers; this means that the latter are excluded when a determiner is needed.

Two elements which do not occur in the determiner position are the collective marker *kuā/koā* and plural markers. The proper article *a*, which precedes proper nouns and pronouns, is not a determiner either: it occurs in a different position in the noun phrase. Also, it occurs in less contexts than determiners; in many contexts, proper nouns are not marked with *a*. This means that *a* is not the proper noun counterpart of the article *te*.

The head noun is usually obligatory. There are a few constructions in which a noun phrase is headless, but all of these are relatively rare.

The noun may be modified by either a noun, verb or adjective, but these do not have the same status. Modifying nouns and verbs are incorporated into the head noun, forming a compound: they are bare words and express a single concept together with the head noun. Modifying adjectives, on the other hand, express an additional concept and may form an adjective phrase.

Modifying verbs are superficially similar to bare relative clauses; in both of these, the verb is not preceded by any aspect or mood marker. However, unlike modifying verbs,

#### 5 The noun phrase

bare relative clauses are full clauses which may contain arguments and modifiers. Also, they do not express a single concept together with the head noun, but express a specific event.

Finally, the noun may be modified by certain adverbs, the limitative marker *nō*, the identity marker *ꞌā/ꞌana* and the deictic particle *ai*.

## **6 Possession**

### **6.1 Introduction**

This chapter describes the syntax and semantics of possessive constructions.<sup>1</sup> Possessive constructions in Rapa Nui are defined by the use of the possessive prepositions *o* and *ꞌa*. They occur in a wide variety of syntactic environments: as modifiers in the noun phrase, as predicates of nominal clauses, and in several other constructions. This range of possessive constructions is discussed in §6.2. §6.3 deals with the semantics of possessives. Possessives express a wide range of relationships besides the idea of possession as such; these are described in §6.3.1.

Whether a possessive construction is marked with *o* or *a* depends on the semantic relationship between possessor and possessee. The distinction between the two classes of possession is discussed in §6.3.2–6.3.4.

The *o/a* distinction also applies to the benefactive prepositions *mo/ma* (§4.7.8),<sup>2</sup> which in turn form the basis for benefactive pronouns (§4.2.3). The semantic relationships discussed in the sections below are also valid for benefactives.

### **6.2 Possessive constructions**

As mentioned above, all possessive constructions contain a possessive preposition *o* or *ꞌa*. In certain constructions, *o* and *ꞌa* coalesce with the determiner *te* into a marker *to* or *ta*. These four forms (*o* and *ꞌa*, *to* and *ta*) in turn form the basis for possessive pronouns (§4.2.2). The *o/a* forms are labelled Ø-possessives, the *to/ta* forms *t-*possessives.

In this section, the range of possessive constructions is discussed. §6.2.1 deals with the use of possessives in the noun phrase. A peculiar noun phrase construction is the partitive; this is discussed in §6.2.2. Other possessive constructions (such as possessive

<sup>1</sup> The term *possessive* is used as a technical term here including not only relationships of possession, but any relationship expressed by possessive pronouns, possessive prepositions, or benefactive pronouns or noun phrases (cf. Lichtenberk 1985: 94). For relationships within noun phrases, the term *possessee* is used for the head, *possessor* for the possessive modifier.

<sup>2</sup> Possessive constructions marked with prepositions or possessive pronouns are the common pattern in Polynesian. In this respect, Polynesian is unusual within the wider family of Oceanic languages: the latter are generally characterised by a distinction between direct and indirect possession (Lynch 1997; Lichtenberk 1985). Direct possession is marked by suffixes on the noun; in Polynesian, traces of this system survive in a number of kinship terms ending in *-na* (Lynch 1997; Marck 1996b); in Rapa Nui: *tuakana* 'older sibling'; *taina* 'sibling', *makupuna* 'grandchild', *tupuna* 'ancestor'; possibly also *ha ꞌana* 'woman's sister's husband (obsolete)' (Métraux 1971: 99), cf. PPN *\*sa ꞌa* 'family, clan'.

6 Possession

clauses) are listed in §6.2.3; they are discussed in more detail elsewhere in this grammar. §6.2.4 summarises the use of the different possessive forms.

### **6.2.1 Possessives in the noun phrase**

Possessive noun phrases (i.e. those involving a common or proper noun) occur at the end of the noun phrase, after other particles. They are introduced by *o* or *ꞌa*: 3


Possessive noun phrases also appear in headless noun phrases, in which case *o*/*ꞌa* coalesces with the determiner to *to*/*ta* (examples (102–103) in §5.6).

When the possessor is pronominal, i.e. a possessive pronoun, it may occur in three different positions, as Table 5.1 and Table 5.2 in §5.1 show: in determiner position; before the noun but not in determiner position; after the noun.

As explained in §4.2.2, Rapa Nui has two sets of possessive pronouns: *t-*possessives (*tāꞌana, te mātou*) and Ø-possessives (*ꞌāꞌana, o mātou*). Which form is used, depends on the position of the pronoun in the noun phrase.

When the noun phrase needs a determiner (§5.3.2.1), possessive pronouns usually occur in the determiner position. In this position, *t*-possessives are used. This happens for example when the noun phrase is subject, or occurs after a preposition:

(4) *¿He* pred *aha* what *tōꞌona* poss.3sg.o *rua* two *ꞌīŋoa?* name

'What is his second name?' [R412.079]

(i) *E* pred *mea* thing *maita ꞌi* good [*tā* art:of.a *terā* dist *ta ꞌata* ] man *tipi.* knife (Tahitian) 'That man's knife is good.'

The head noun *tipi* is preceded by a genitive noun phrase *tā terā ta ꞌata*; the possessive preposition *ta* is based on the article *te* + possessive *a*. In Rapa Nui this construction occurs only in headless noun phrases (§5.6). Fischer (2000: 336) gives an example of a prenominal noun phrase possessor (*tā te taŋata poki* 'the man's children'), but such a construction does not occur anywhere in my corpus.

<sup>3</sup> In other Eastern Polynesian languages, possessive noun phrases may also occur in the determiner position, introduced by a *t-*form *to* or *ta*. See for example the following example from Tahitian (pers.obs.):

6.2 Possessive constructions


That the possessive pronoun is in determiner position, is also confirmed by the fact that prenominal quantifiers occur after the possessor, as in (4).

Possessive pronouns may also occur before the noun in noun phrases not containing a determiner. In that case, Ø-possessives are used.<sup>4</sup> This happens especially when the noun phrase contains a prenominal numeral, but also after the negator *ꞌina*. Prenominal numerals preclude the use of all determiners (§5.3.2.2), while *ꞌina* precludes the use of *t-*determiners (§10.5.1):


When plural Ø-pronouns occur before the noun, the *o* is omitted.<sup>5</sup> This means that they have the same form as the corresponding personal pronouns; only their position identifies them as possessive pronouns.

(10) *E* ipfv *ai* exist *rō* emph *ꞌā* cont *e* num *tahi* one *rāua* 3pl *poki* child *tane* male *te* art *ꞌīŋoa* name *ko* prom *Iovani.* Iovani

'They had one son (lit. there was one their son) named Iovani.' [R238.002]

(i) *E* num *rua* two *o* of *mātou* 1pl.excl *hare.* house 'We have two houses.'

<sup>4</sup> When a *t*-possessive occurs in this position, the numeral is not part of the noun phrase, but predicate of a so-called 'numerical clause' (§9.5).

<sup>5</sup> Wilson (1985: 106) gives an example from Rapa Nui in which the genitive preposition is not omitted (modified spelling & gloss):

Unfortunately, no source is given for this example; it may well be erroneous, as no examples of this construction occur in my corpus.

#### 6 Possession

(11) *ꞌIna* neg *tātou* 1pl.incl *haŋu* sustenance *mo* for *kai* eat *ko* prom *kuā* coll *nua.* Mum 'We don't have any food left to eat, me and Mum.' [R372.047]

After the noun, these pronouns do have the *o*, as illustrated in (12) below. Finally, possessive pronouns may occur at the end of the noun phrase, in the same position as possessive noun phrases (see (1–3) above). In this position, Ø-possessives are used:


These postnominal possessives occur when the determiner slot is occupied by another element. As these examples show, this especially happens when the noun phrase contains a demonstrative determiner such as *tū*. *Tū* fulfills the requirement for the noun phrase to have a determiner, but it precludes the use of a prenominal possessive, hence the possessive is placed after the noun.

Sometimes a *t*-possessive pronoun before the noun occurs together with a Ø-possessive after the noun. In this double possessive construction, the two pronouns reinforce each other:


'In your view, was that time better than the present time?' [R380.106]

Possessive doubling only happens in the second person. The *t*-possessive before the noun is always one of the shortened forms *tuꞌu* or *taꞌa* (§4.2.2.1.1).

6.2 Possessive constructions

### **6.2.2 The partitive construction**

Besides the common construction "*t*-possessive N" discussed above, Rapa Nui has a construction "*t*-possessive *o te* N". In this construction, the possessee has been demoted from the head noun position to a possessive phrase *o te N*. The construction has a partitive sense, indicating someone's share, portion: *tāꞌaku o te vai* = 'my portion of the water, the part of the water that is mine'. Some examples:


As the examples above show, the sense of 'share, portion' often implies that the item is not yet in the hands of the possessor, but destined for him or her. This construction may be emphatic: 'yours, nobody else's':


'He lifted the boy on the horse, and he mounted on his own horse.' [R105.028]

As these examples show, in this construction the long second-person pronouns *tōꞌou* and *tāꞌau* are used, even though prenominal possessive pronouns usually have one of the short forms *taꞌa, tuꞌu* etc (§4.2.2.1.1). There is another difference between prenominal possessives and partitives. While prenominal possessives can only be pronouns (§6.2.1 above), the possessive in a partitive construction may also be a full noun phrase. This noun phrase is constructed with a possessive preposition *to* or *ta*, following the *o*/*a* distinction (§6.3.2). In the following examples, just as in some of the examples above, the construction expresses something destined for the possessor.

(21) *ꞌĪ* imm *au* 1sg *he* ntr *haꞌataꞌa* separate *i* acc *to* art:of *Vaha* Vaha *o* of *te* art *kahu.* cloth(es) 'I will put apart some clothes for Vaha.' [R229.194]

#### 6 Possession

(22) *ꞌĪ* imm *au* 1sg *he* ntr *haꞌataꞌa* separate *i* acc *ta* art:of.a *Māria* Maria *o* of *te* art *kai.* food 'I will put apart some food for Maria.' [Notes]

A similar but simpler construction – which can be labelled "pseudo-partitive" – is *to te N*. In this construction, the noun phrase *te N* is introduced by *to* in a possessive/partitive sense:


This construction is reminiscent, syntactically speaking, of the headless possessive construction (§5.6), of which an example is given here:

(25) *Ko* prom *Koka* Koka *te* art *ꞌīŋoa* name *o* of *tōꞌona* poss.3sg.o *hoi…* horse *ko* prom *Parasa* Parasa *to* art:of *te* art *rūꞌau* old\_woman *era* dist *ꞌāꞌana.* poss.3sg.a

'Koka was the name of the horse he went on, Parasa the (name) of his old wife.' [R539-1.420]

There is an important difference though: while in (25) *to te rūꞌau* has a straightforward possessive sense (parallel to the possessive phrase *o tōꞌona hoi*), in (23–24) the possessive phrase occurs in a context where normally the dative preposition *ki* would be used.

Semantically, (23–24) are similar to the partitive construction discussed above. In both cases, the noun phrase refers to something which is destined for the person referred to; moreover, the sense is partitive: 'some of the food, some of the clothes'. Also, in both cases the *to*-phrase is independent: there is no head noun to which it is attached. The difference is that in the partitive construction in (16–18) above the possessee is expressed by a genitive phrase *o te kahu* which is semantically the head of the phrase (the noun phrase as a whole refers to 'clothes', not to 'Vaha'), while in (23–24) it is not expressed at all.

6.2 Possessive constructions

### **6.2.3 Other possessive constructions**

Possessive constructions occur not only as modifiers in the noun phrase, but in a range of other constructions as well. This section gives a concise listing; all of these constructions (with the exception of the elliptic construction in (34) below) are discussed in more detail elsewhere in this grammar.<sup>6</sup>

**6.2.3.1** *o*-class Ø-possessives are used to mark the S/A argument of a clause introduced by preverbal *mo* 'if; in order to' (§11.5.1.2):

(26) *Mo* if *kī* say *ōꞌoku* poss.1sg.o *he* pred *teatea,* white:red *he* ntr *rere* jump *a* by *ruŋa* above *he* ntr *kī* say *atu* away *he* pred *ꞌuriꞌuri.* black:red 'If I say it's white, he jumps up and says it's black.' [R480.003]

Occasionally, *O*-class Ø-possessives are used to mark the S/A argument of a main clause (§8.6.4.1):

(27) *He* ntr *uꞌi* look *atu* away *ōꞌoku* poss.1sg.o *i* acc *tōꞌoku* poss.1sg.o *pāpā* father *era…* dist 'Then I saw my father…' [R101.012]

**6.2.3.2** *a*-class Ø-possessives<sup>7</sup> serve to express the Agent in the actor-emphatic construction (§8.6.3):

(28) *¡ ꞌĀꞌau* poss.2sg.a *rō* emph *taꞌa* poss.2sg.a *moeŋa* mat *nei* prox *o* of *māua* 1du.excl *i* pfv *toke!* steal 'It was you who stole that mat of ours!' [R310.428]

**6.2.3.3** Ø-possessives (both *a*- and *o*-class) served as the predicate of proprietary clauses (§9.4.2).

(29) *Ōꞌona* poss.3sg.o *hoꞌi* indeed *te* art *ꞌāua* field *era.* dist 'That field is his.' [R413.228]

<sup>6</sup> Not included here are possessives expressing the subject of a relative clause in the "possessive-relative construction"; as argued in §11.4.4, these should be considered as normal noun-phrase possessors which are syntactically separate from the relative clause.

<sup>7</sup> *A*-forms only occur with singular pronouns and with proper nouns (§6.3.2); with plural pronouns and with common nouns, only the default *o*-forms are available.

#### 6 Possession

**6.2.3.4** In older Rapa Nui, the *t*-possessives serve as the predicate of possessive clauses (§9.3.3). In modern Rapa Nui, this construction is no longer in use.<sup>8</sup>

(30) *He* ntr *poki* child *tāꞌana* poss.3sg.a *e* num *tahi,* one *poki* child *tamāroa.* male 'He had a child, a boy.' [Ley-9-57.002]

**6.2.3.5** Possessives of the *o*-class may serve as the predicate of existential-locative clauses (§9.3.2); see §6.3.1.8 below on the locative use of possessives. In modern Rapa Nui, Ø-possessives are used as in (31); in older Rapa Nui, *t*-possessives were used as in (32–33).


**6.2.3.6** Finally, *to* (i.e. the *t*-possessive *o*-form) + locational is sometimes used as an elliptic noun phrase. This use is especially found in older Rapa Nui. *To* + noun refers to a group of people situated in the location indicated by the locational: 'those ones inside/outside/over there…' (cf. the headless construction 2 above). In the following example, *to haho* is a short way to refer to 'the people outside'.

(34) *He* ntr *toe* remain *e* num *toru* three *te* art *ꞌaŋahuru* ten *nō* just *toe,* remain *ku* prf *oti* finish *ꞌā* cont *to* art:of *haho.* outside 'Only thirty (men) were left, those outside were finished.' [Mtx-3-01.092]

### **6.2.4 Summary: use of possessive forms**

The occurrence of the different possessive forms is summarised in Table 6.1.

The discussion in the previous sections has shown that various possessive forms are used, depending on the construction. Three parameters play a role, which are reflected in Table 6.1:

• the choice between *t*- and Ø-possessives;

<sup>8</sup> In modern Rapa Nui, possessive clauses are constructed as verbal or verbless existential clauses; the possessor is expressed not as a predicate, but as a noun phrase modifier (§9.3.3):

#### 6.2 Possessive constructions


#### Table 6.1: Possessive constructions


#### 6 Possession

Summarising these data: *t*-possessives are used in the following environments:


Ø-possessives are used:<sup>9</sup>


### **6.3 The semantics of possessives**

As in many languages, possessive constructions express a wide range of relationships between two entities. These are listed in §6.3.1.

When the possessor is a singular pronoun or a proper noun, it can be expressed in two ways, using either *o* or *a*. The distinction between *o* and *a* is discussed in §6.3.2–6.3.4; in those sections, the range of relationships expressed by the possessive will be discussed and illustrated in more detail.

<sup>9</sup> The range of use of the Ø-possessives is remarkably similar to the use of *n-*possessives which occur in most other EP languages, e.g. Māori *nāku* 'mine', *nōna* 'his' (cf. Wilson 2012: 316). For example, in Māori and Tahitian *n*-possessors are used in the actor-emphatic construction and in proprietary clauses; in Tahitian and Hawaiian, they also occur in the noun phrase (Bauer, Parker & Evans 1993: 208–209; Harlow 2000; Lazard & Peltzer 2000: 175–176, 189; Cook 2000: 349). Māori also has Ø-possessive pronouns, which occur in the noun phrase and in negated possessive clauses (Bauer, Parker & Evans 1993: 202, 381; Harlow 2000: 359). This suggests that the Rapa Nui Ø-possessive pronouns are cognates of both the *n*-possessives and Ø-possessives in other EP languages: PEP had a set of Ø-possessives and a set of *n-*possessives; in Rapa Nui, the initial *n-* was lost, so that both sets coincided; in CE languages, *n-* was retained.

The reverse scenario, in which PEP only had the Ø forms and *n-* was added in PCE, is unlikely. First, it would leave the co-existence of Ø- and *n-*possessives in Māori unexplained: if PEP only had Ø-possessives, it would be hard to explain why *n-* was added in some contexts, while in other contexts the Ø-forms were retained. Second, there is no ready explanation for the addition of *n-* within Central-Eastern Polynesian, while on the other hand the existence of *n-* in PEP can be explained either from the past tense marker *ne*, or – more likely – from the possessive *ni* which occurs in various Outliers (see Wilson 1982: 50; Wilson 1985: 101; Clark 2000a: 263).

We may conclude that the *n*-possessives are not a PCE innovation as suggested by Green (1985: 12), but already present in PEP (§1.2.2).

6.3 The semantics of possessives

### **6.3.1 Relationships expressed by possessives**

Possessive constructions serve to express a wide range of relationships between two entities.

**6.3.1.1** Ownership:

(35) *He* ntr *haka* caus *hopu* wash *i* acc *tāꞌana* poss.3sg.a *paiheŋa.* dog 'She washed her dog.' [R168.012]

**6.3.1.2** Whole/part relations:

(36) *He* ntr *puru* close *i* acc *te* art *papae* door *o* of *te* art *hare.* house 'He closed the door of the house.' [R310.144]

**6.3.1.3** Interhuman relationships, such as kinship and friendship:

(37) *He* ntr *hokorua* accompany *a* prop *au* 1sg *i* acc *tōꞌoku* poss.1sg.o *repahoa.* friend 'I accompany my friend.' [R208.138]

(38) *¿He* ntr *aha* what *toꞌu* poss.2sg.o *tau* pretty *tuꞌu* poss.2sg.o *rivariva?* good:red 'What (use) is your beauty, your goodness?' [R372.045]

**6.3.1.5** Specification (epexegetical use), where the possessive has the same referent as the head noun:

(39) *He* ntr *eke* go\_up *ki* to *ruŋa* above *ki* to *te* art *ꞌana* cave *o* of *ꞌAna* Ana *Havea.* Havea 'He climbed above the cave (of) Ana Havea.' [Mtx-7-18.010]

**6.3.1.6** Actions, feelings and emotions, when these are expressed as nouns or nominalised verbs:

(40) *Meꞌe* thing *rahi* much *tōꞌoku* poss.1sg.o *māuruuru* thank *ki* to *a* prop *koe,* 2sg *e* voc *koro* Dad *ē.* voc 'I am very grateful (lit. much is my gratitude) to you, Dad.' [R363.112]

**<sup>6.3.1.4</sup>** Attributes:

#### 6 Possession

(41) *…ꞌo* because\_of *tōꞌoku* poss.1sg.o *kī* say *nō* just *mo* for *toꞌo* take *mai* hither *i* acc *a* art *Puakiva* Puakiva *mai* from *i* at *a* prop *ia.* 3sg

'(Kava is crying) because I said (lit. because of my saying) I would take Puakiva away from her.' [R229.017]

**6.3.1.7** Any kind of association, for example between a person and a group, or a person and a location:


**6.3.1.8** The possessor may be a location to which the possessee belongs as in (44), or a place where the possessee is located at a given time, as in (45–46).


In (45–46), *o* is close in meaning to the locative *ꞌi* 'in, at'. As these examples show, the possessive is used especially in negative or interrogative clauses. In those sentences, *ꞌi* is considered awkward.

**6.3.1.9** When the head noun expresses time, the possessive may express an event with respect to which this time applies. In (47) the event is punctual, and 'three days' is the time elapsed after the event. In (48) the event is durative, and 'eight days' is the time elapsed since the beginning of this event. (In both cases, *ka* indicates that a certain moment in time has been reached, see §4.3.2.2.)

6.3 The semantics of possessives


The possessive after the temporal noun may also refer to somebody or something which was involved in a certain event at the time specified. The event itself is expressed as a relative clause following this noun. (49) can be translated literally 'Three days of the rain which fell'.

(49) *E* num *toru* three *mahana* day *o* of *te* art *ꞌua* rain *i* pfv *hoa* throw *ai,* pvp *ko* prf *reherehe* soft:red *atu* away *ꞌā* cont *te* art *ꞌōꞌone.* soil

'When it had been raining for three days, the ground was quite soft.' [R378.040]

(50) *E* num *tahi* one *nō* just *mahana* day *o* of *te* art *pahī* ship *holandese* Dutch *nei* prox *i* pfv *noho* stay *mai* hither *ꞌi* at *nei* prox *ꞌi* at *Rapa Nui.*

Rapa Nui

'The Dutch ship only stayed one day here on Rapa Nui (lit. Just one day of the Dutch ship that stayed).' [R373.005]

**6.3.1.10** Regardless of the semantic relationship, the possessor may express something which does not yet belong to the possessor, but which the possessor intends to have:<sup>10</sup>


<sup>10</sup> Cf. Lichtenberk (2002), who gives examples of "prospective possessive relationships" in several Oceanic languages.

#### 6 Possession

### **6.3.2** *A***- and** *o***-possessives**

Like most Polynesian languages, Rapa Nui makes a distinction between two types of possessive marking, which are characterised by the vowels *o* and *a*, respectively.<sup>11</sup> In most languages this distinction is pervasive, affecting all possessive noun phrases and pronouns. In Rapa Nui, the *o*/*a* distinction is only made with the following nominal elements:<sup>12</sup>

	- 2. Names and other proper nouns:

With common nouns and plural pronouns, only *o*-forms are used:<sup>14</sup>

(56) *te* art *poki* child *o* of *te* art *ꞌariki;* chief *te* art *matuꞌa* parent *o* of *te* art *ꞌariki* chief 'the chief's child; the chief's parent'

<sup>11</sup> Besides the grammars of individual languages, see Clark (1976: 42-44); Capell (1931); Biggs (2000). On the *o/a* distinction in Rapa Nui, see especially Mulloy & Rapu (1977). In Rapa Nui, *ꞌa* (both as a preposition and at the start of possessive pronouns) is written with a glottal, while *o* is not. The main reason is, that *ꞌa* happens to occur initially more often than *o*. It is used, for example, in the active-emphatic construction (§8.6.3). See sec. §2.2.5 on the (non-)occurrence of initial glottals in particles. It is not impossible that the glottal-non/glottal distinction in these particles was inherited from an earlier stage. Clark (2000a: 259) points out that in Tongan, certain *a*-forms have a glottal, while the corresponding *o*-forms do not (e.g. *he ꞌeku* 'my.a' vs. *hoku* 'my.O'). See also Wilson (1982: 48).

On the other hand, the pervasive presence of the glottal in sg. possessive pronouns (*tā ꞌaku*, *ꞌā ꞌaku*, *mō ꞌoku*, *tō ꞌoku* etc.) suggests that originally the glottal preceded both *a* and *o* (cf. Lynch 1997: 232; Wilson 1982: 50).

<sup>12</sup> Neutralisation of the *o*/*a*-distinction is not uncommon in Polynesian languages. In Niuean (Seiter 1980: 34), the distinction is completely lost. The same is true in a group of Outliers: Nukeria, Takuu, Nukumanu and Luangiua (Wilson 1982: 11; Clark 2000a: 267), while in Rennell, the distinction is lost in third person pronouns (Nico Daams, p.c.).

<sup>13</sup> For the forms of possessive pronouns, see sec. §4.2.1.

<sup>14</sup> A peculiar exception, in which a common noun phrase is an *a*-possessor, is the expression *ꞌa te hau* 'Chilean, from the mainland', as in *va ꞌehau ꞌa te hau* 'Chilean soldiers' (R539-1.616). *Te hau* seems to be used as a name here, meaning something like 'the State'; proper names in Rapa Nui may contain the article *te*.

6.3 The semantics of possessives

(57) *tū* dem *poki* child *era* dist *o* of *rāua* 3pl 'that child of theirs'

The two types of possessive constructions will be referred to as *a*-possession and *o*possession. The choice between the two can often be predicted from the head noun (the possessee): *matuꞌa* is *o*-possessed, *poki* is *a*-possessed. However, many words can be possessed with either *o* or *ꞌa*; Englert (1978: 43) gives the following pair of examples (for more examples, see §6.3.4.1):

(58) *He* ntr *toꞌo* take *tōꞌona* poss.3sg.o *kahu* clothes *mo* for *tata.* wash 'She took her (own) clothes to wash'.

(59) *He* ntr *toꞌo* take *tāꞌana* poss.3sg.a *kahu* clothes *mo* for *tata.* wash

> 'She took her clothes (the clothes that had been given to her as a laundress) to wash.'

The choice for *ꞌa-* or *o*-possession, then, is not an inherent property of the noun; it is determined by the **relation** between the possessor and the possessee, not by the nature of the possessee as such.<sup>15</sup> If many nouns are always *a-*possessed or always *o*-possessed, this is because they always stand in the same relationship to the possessor. For example, when *poki* 'child' is possessed, i.e. 'A is *poki* of B', this usually means that A stands in a child-parent relationship to B, a relationship which is expressed by *a-*possession.

The *o/a* distinction does not only affect possessive pronouns and genitive constituents in the noun phrase (including partitives, see §6.2.2), but benefactives as well: the latter are constructed with either *mā* or *mo* when followed by a singular pronoun or proper noun, depending on the nature of the relationship between the two referents involved (§4.7.8).

### **6.3.3 Possessive relations marked with** *a* **and** *o*

In §6.3.1, a general overview was given of relationships expressed by possessive constructions. The present section provides a detailed discussion of these relationships, categorised by *a*- and *o*-marking.

§6.3.3.1 deals with relationships between people, while §6.3.3.2 and §6.3.3.3 discuss relationships involving non-human possessees. §6.3.3.4 deals with nominalised verbs and their arguments. §6.3.4 addresses the question whether a general characterisation of *ꞌa-* and *o*-possession is possible.

<sup>15</sup> See also Chapin (1978: 151).

#### 6 Possession

#### **6.3.3.1 Human possessees**

When both possessor and possessee are human, the situation is relatively straightforward in the case of kinship relations. These will be discussed in §6.3.3.1.1. Other interhuman relationships are discussd in §6.3.3.1.2.

**6.3.3.1.1 Kinship relations** *ꞌA*-possession is used to express the following kinship relations:


All other kinship relationships are expressed with *o*-possession:

	- 4. further offspring and offspring in general: *tōꞌona hinarere* 'his great-grandchild'; *tōꞌona hakaara* 'his descendants'.

<sup>16</sup> Mulloy & Rapu (1977: 22) quote one example from Métraux' published stories where *taina* is *a*-possessed:

<sup>(</sup>i) *He* ntr *tomo* go\_ashore *Poie* Poie *ki* to *te* art *motu* islet *ananake* together *ko* prom *tāꞌana* poss.3sg.a *ŋā* pl *taina.* sibling 'Poie landed on the island, together with his brothers.' (Mtx-3-01.311)

According to Mulloy & Rapu, this suggests that in the past younger brothers were *a*-possessed, a situation which was changing to *o*-possession in the 1930s, when this story was recorded. However, *tā ꞌana* turns out to be a faulty transcription in the printed text: the text in Métraux' notebook (notebook 4, p. 170) has the regular *tō ꞌona*. Note, however, that Mtx's texts do show some other irregularities in the use of *ꞌa* and *o* possession, without a clear reason: *te matuꞌa ꞌa Ure* 'Ure's father' (Mtx-7-03.108); *ta ꞌu ha ꞌana* 'your brother-in-law' (Mtx-7-30.062); in both cases, *a*-possession is used where one would expect *o*.

<sup>17</sup> As with siblings, Mulloy & Rapu (1977: 22) suggest that a shift has been taking place in the possession class of grandchildren; the text corpus gives no evidence of such a shift, however.

	- 6. all in-law relationships: *tōꞌou hunoŋa* 'your son/daughter-in-law'; *tōꞌou huŋavai* 'your father/mother-in-law'; *tōꞌou taꞌokete* 'your brother/sister-in-law'.
	- 7. the family as such:

However, in the sense of a nuclear family (people living together in one house), family may also be *a*-possessed:

(63) *E* ipfv *noho* stay *era* dist *a* prop *Manutara* Manutara *ananake* together *ko* prom *tāꞌana* poss.3sg.a *huaꞌai.* family 'Manutara lived with his family.' [R309.039]

#### **6.3.3.1.2 Other human relationships**

	- 3. When the *possessor* is higher in status or authority, or in charge of the possessee (e.g. as employer or teacher), *ꞌa* is used.

(66) *He* ntr *uŋa* send *ia* then *e* ag *Ietū* Jesus *e* num *rua* two *o* of *tāꞌana* poss.3sg.a *nuꞌu* people *hāpī.* learn 'Then Jesus sent out two of his disciples.' [Mrk. 14:13]

This also means that *ꞌa* is used for a group of people over which the possessor is in charge:

(67) *Ko* ntr *arma* assemble *ꞌā* cont *a* prop *au* 1sg *i* acc *tāꞌaku* poss.1sg.a *ekipo* group *mai* from *i* at *a* prop *marzo* March *ꞌā.* ident 'From March on, I have put together my group.' [R625.082]

On the other hand, for a group of people to which the possessor belongs, *o* is used.

(68) *He* ntr *aŋa* make *tau* dem *kope* person *era* dist *i* acc *te* art *koro* feast\_house *kumi,* long *ananake* together *ko* prom *tōꞌona* poss.3sg.o *taŋata* person *i* pfv *aŋa* make *ai.* pvp

'That man made a large feast house, together with his people he made it.' [Mtx-4-03.003]

4. Somewhat unexpectedly, when the possessee is a subordinate, *o* tends to be used: *tōꞌoku rarovaꞌe/tāvini* 'my subordinate/servant'.

#### **6.3.3.2 Non-human possessees with** *ꞌa*

With non-human possessees, *ꞌa* is used in the following situations:


This category includes furniture, except furniture supporting the body (see 6c in the next section).

2. The possessee is something produced or caused by the possessor.

6.3 The semantics of possessives

	- 3. The possessee is a dream of the possessor ('to dream' is *moe i te vārua*, lit. 'lie down a spirit').

However, dreams can be *o*-possessed as well:<sup>18</sup>

	- 4. The possessee is land worked by the possessor.
	- 5. The possessee is food. This can be food grown, caught or otherwise obtained by the possessor as in (76) and (77), or food/drink consumed – or destined to be consumed – by the possessor as in (78).

<sup>18</sup> The same variability is seen in Māori, where *moemoeā* 'dream' is *o*-possessed for some speakers and *a*possessed for others (Harlow 2007a: 170).

#### 6 Possession


Horses, however, are *o*-possessed, as they are animals of transport (see (86–87) in the next section).

#### **6.3.3.3 Non-human possessees with** *o*

With non-human possessees, *o* is used in the following situations:


Fruits and flowers of plants can be included in this category, although these may also be *o*-possessed by virtue of being part of a whole (see 2 above).

(81) *ꞌE* and *ꞌi* at *rā* dist *kona* place *he* ntr *tupu* grow *te* art *pua,* kind\_of\_plant *ꞌe* and *he* ntr *ꞌūaꞌa* blossom *tōꞌona* poss.3sg.o *tiare.* flower

'And in that place the *pua* grew and its flowers blossomed' [R532-07.081]


6.3 The semantics of possessives

(83) *…ꞌo* lest *hakameꞌemeꞌe* mock *mai* hither *i* acc *tōꞌoku* poss.1sg.o *veve* poor *e* ag *Makoꞌi.* Mako'i '…so that Mako'i would not mock my poverty.' [R214.050]

This also includes sicknesses: *tōꞌona māuiui/renkē/kokoŋo* 'his sickness/dengue/ cold'.

	- a) clothing and footwear worn by the possessor: *tōꞌoku kahu/kamita/kirivaꞌe/ kete* 'my clothes/shirt/shoes/pocket'.<sup>19</sup> Clothing is *a-*possessed when it does not refer to clothing to be worn, but functions just as a possession or an object to be handled:
	- b) other things covering or adorning the body, such as jewellery, eyeglasses, tattoos and body paint: *tōꞌona karone/hei/tāpeꞌa/hiꞌo* 'her necklace*/* headdress*/*ring*/*glasses'. Watches, however, are *a*-possessed; presumably, they are not classified with jewellery, but with tools and instruments (see 1 in the previous section):

neg poss.1sg.a time

'I don't have a watch.' (Mulloy & Rapu 1977: 17)


<sup>19</sup> *Kete* means 'pocket' in modern Rapa Nui. In the past, *kete* used to mean 'basket' and was *a*-possessed, like any container.

#### 6 Possession

Other furniture is *a-*possessed, like tools and instruments (see 1 in the previous section): *tāꞌaku ꞌamuramaꞌa* 'my table'.

	- 7. The possessee is the country, territory or place to which the possessor belongs.
	- 8. The possessee is property owned by the possessor. This includes:
		- a) land, for example, a plantation or garden:

This means that fields and gardens can be either *a-* or *o-*possessed, depending on whether the focus is on possession (*o*) or labour (*ꞌa*); cf. (75) in the previous section.


6.3 The semantics of possessives


This includes stories, songs, pictures and other work of art with the possessor as theme: *tōꞌoku ꞌaꞌamu* 'the story about me'; *te hohoꞌa o Tiare* 'the picture of Tiare, showing Tiare'.

	- 11. The possessee is a noun referring to time: *tōꞌona mahan poreko* 'his birthday'.
	- 12. The possessor specifies the reference of the possessee, it is a specific instance of the possessee (epexegetical use).

#### 6 Possession

(100) *E* num *hoꞌe* on *ꞌahuru* ten *mā* plus *hoꞌe* one *huru* manner *kē,* different *huru* manner *kē,* different *huru* manner *kē* different *tōꞌona* poss.3sg.o *puka* book *ꞌo* or *tōꞌona* poss.3sg.o *ꞌaꞌamu.* story 'There are eleven different books and different stories.' [R206.019]

#### **6.3.3.4 Possession with nominalised verbs**

The arguments of nominalised verbs are often expressed as a possessor (§8.7). When the possessor is Patient, i.e. undergoes the action, *o*-possession is used:


When the possessee is Agent, i.e. performs the action, the situation is more complicated. Actions as such tend to be *o*-possessed:<sup>20</sup>


'I am now happy because of his chasing me.' [R214.053]

(106) *He* ntr *ꞌui* ask *e* ag *tū* dem *tahutahu* witch *era* dist *i* acc *te* art *tumu* reason *o* of *tōꞌona* poss.3sg.o *tere.* travel 'The witch asked about the reason for his trip.' [R532-07.043]

<sup>20</sup> This is different from the situation in other Polynesian languages, where subjects of transitive verbs (and often intransitive agentive verbs as well) tend to be marked with *a*, while objects and non-agentive subjects are marked with *o* (See e.g. Chung 1973; Clark 1981: 69; Lazard & Peltzer 2000: 197–201; Cablitz 2006: 173–174; Mosel & Hovdhaugen 1992: 540–541; Besnier 2000: 503–505; Elbert & Pukui 1979: 140–142). For Hawaiian, Baker (2012) shows that the choice between *a* and *o* for subjects is pragmatically motivated: *a*-marked subjects are agentive and/or volitional and/or individuated.

6.3 The semantics of possessives

When the noun refers to the product or result of an action rather than the action itself, it is *a*-possessed:


The following pair of examples show the contrast between the action as such as in (109) and the product of an action as in (110):


Verbs expressing verbal utterances ('say', 'tell', 'sing') show the same distinction between the product of an action and the action itself. Utterances made by the possessor – words, stories, songs, et cetera – are *a*-possessed, as in (111–112). On the other hand, when the act of uttering itself is in focus, the possessor is *o*-marked, as in (113–114):


Finally, in the actor-emphatic construction (§8.6.3), Agents are *a*-possessed.

#### 6 Possession

### **6.3.4 General discussion**

#### **6.3.4.1 Summary**

The examples in the previous sections show that the choice between *ꞌa* and *o* depends on the semantic relation between the two referents, not on the actual noun used. A given noun can be *a*- or *o*-possessed, depending on the relation to the possessor. Table 6.2 gives a few examples.


Table 6.2: Some *a*- and *o*-possessed words

The fact that the *o*/*a* distinction has a semantic basis, also means that new words (usually Spanish borrowings) are integrated into the system on the basis of the semantic relation they bear to their possessors. For example, *kōrore* 'colour', *ꞌauto* 'car' and *sobirino* 'nephew' are *o*-possessed, while *koneta* 'trumpet' and *ekipo* 'group' are *a*-possessed.<sup>21</sup>

In fact, apart from lexical changes, the system shows a remarkable stability over time, as far as the sources show. None of the semantic categories described in the previous sections shows shifts in possessive marking between older texts and modern Rapa Nui. (It is only with younger speakers who master the language imperfectly that the *o*/*a* distinction is starting to break down.)

The findings from §6.3.3 can be summarised as follows:


The next section deals with the question whether the *o*/*a* distinction can be explained by a general rule.

<sup>21</sup> See Makihara (2001b: 203) for more examples.

6.3 The semantics of possessives

#### **6.3.4.2 A general rule?**

The *o*/*a* distinction occurs more or less along the same lines in almost all Polynesian languages,<sup>22</sup> and it has been described in various ways.

In general linguistic literature, the distinction between two classes of possession, one of which is more permanent and/or closer to the possessor, is usually labelled alienable/inalienable, and this terminology is followed by Du Feu (1996: 102): *o*-possession is inalienable, *a*-possession is alienable. Pukui & Elbert (1957) use the same terms for Hawaiian. Englert (1978: 42) makes a similar distinction when he states that *o* is used with objects which, in the idea of the speaker, are closer to the possessor. Hohepa (1967) characterises the distinction as one between inherited and acquired possession. According to Capell (1931: 145), "*o* forms indicate a passive relation to the possessor, the *a* forms an active relationship". Biggs (1973: 43) extends this further: *a* is used "when the possessor is active, dominant or superior to that which is possessed"; *o* is used "when the possessor is passive, subordinate or inferior to that which is possessed". Finally, Mulloy & Rapu (1977) propose a distinction between dependence and responsibility.<sup>23</sup>

What, then, is the most appropriate way to characterise the *ꞌa/o* distinction in general terms?

First of all, the distinction between alienable and inalienable is not very accurate in describing which items are *o*- and *a*-possessed. Inalienable possession refers to inherent and/or permanent relationships, such as kinship and part-whole (Dryer 2007b: 185). While it is true that the *o*-possessive indicates inherent and/or permanent possessions like body and soul, body parts and land, its use is much broader, including categories like attitudes and feelings, clothing, jewellery, means of transport and actions undergone. The alienable/inalienable distinction is therefore inadequate as a general characterisation. The same is true for the distinction between inherent and acqired possession.

The distinction between dominant and subordinate makes a number of correct predictions: some possessors that are dominant with respect to their possessees, are *a*marked, while some possessors that are subordinate with respect to their possessees, are *o*-marked. The leader or organiser of a group has a dominant role, while the subjects of a king have a subordinate role. I am dominant with respect to the tools and instruments I handle, the products I make, and the animals and plants I possess.

For other categories, however, this distinction does not work very well. Can a person said to be subordinate with respect to his/her body, voice, feelings and attitudes, or with

<sup>22</sup> There are minor differences between languages. In Tahitian, for example, horses are classified as domesticated animals (*a*-possessed) rather than means of transport (*o*-possessed). Money is *a*-possessed, buildings (except dwellings) are *a*-possessed. Children are *a*-possessed, but young of animals are *o*-possessed (Académie Tahitienne 1986: 86–92). In Māori, grandchildren are *a-*possessed, and so are servants. Food is *a*-possessed, but drinking water is *o*-possessed (Biggs 1973: 44).

<sup>23</sup> This explanation is already suggested – though not accepted – for East Futunan by Capell (1931: 146): "A native explanation of the use of *tiaku* with *tafine*, wife, and *tapakasi*, pig, is that they are 'objects of special care'!"

Other approaches have been suggested. Bennardo (2000a,b) proposes a dichotomy in terms of opposing directionality: for *a*-possession the origin is specified, for *o*-possession the direction/recipient is specified. Finally, Elbert (1969) refrains from a general characterisation, suggesting that the labels "*o*-class" versus "*a*-class" may be the easiest for students.

#### 6 Possession

respect to his/her house, clothing, and vehicle? The subordinate category is inaccurate in certain interhuman relationships as well: spouses are mutually *ꞌa*-marked, yet not mutually dominant; siblings are mutually *o-*marked, yet not mutually subordinate.

Mulloy & Rapu (1977) suggest an alternative: responsibility versus dependence. A possessor who is responsible towards the possessee is expressed with *ꞌa*, a possessor who is dependent versus the possessor is expressed with *o*. From the perspective of the possessee, *ꞌa* is used when it depends on the possessor, *o* is used when it is responsible for the possessor.<sup>24</sup>

This idea enables us, for example, to explain the use of *ꞌa* and *o* with respect to interpersonal relationships. A person is responsible with respect to his or her spouse and children, hence *a*-possession. A person depends on his or her parents and extended family, hence *o*-possession. A child is dependent on its parents, hence *o*-possession. A person is responsible for his/her nuclear family (*ꞌa*), but depends on the wider family as a support system (*o*).

For non-human referents, things which "care for, protect, and shelter the possessor" (Mulloy & Rapu 1977: 23) are *o*-possessed, as the possessor depends on them. On the other hand, possessions which the possessor cares for, shelters and protects, are *a*-possessed.

However, for other categories the responsibility/dependence dichotomy is less satisfactory. In a certain sense, a person is dependent on inherent attributes like body and soul. It is even conceivable that someone is dependent on qualities like size, beauty and poverty, as these attributes define a person. It is a bit of a stretch, however, to qualify attitudes like love, compassion, error and sin under the heading of dependency. The same applies for actions and events undergone, like 'problem, punishment, imprisonment', and even more so for actions performed by the possessor. Further, can a person said to be dependent on his saliva or tears, or a chicken on its eggs? Categories like these are defined by neither dependency nor responsibility.

The dichotomy of active versus passive is more promising as a general explanation. In many cases when *ꞌa* is used, the possessor has an active role towards the possessee. A person is active when performing an act or making an utterance; people are active with respect to the land they work, the instruments they use, the products they make, the animals they care for and the food they eat. They are passive with respect to their spirit, life, age and body parts, with respect to buildings and means of transport (although here passivity is expressed more appropriately as dependence, see above), and with respect to feelings, thoughts, and actions they undergo.

In describing interhuman relationships, the terms "active" and "passive" are somewhat less clear, unless "passive" is explained in terms of dependence or subordinance: a child is "passive" with respect to its parents insofar as it depends on its parents for its needs; a worker is "passive" with respect to his boss, insofar as the latter takes the initiative in telling him what to do. In the same way, "active" in these relationships can be explained in terms of responsibility, being in charge: a king is "active" with respect to his subordinates in the sense that he is responsible of caring for them.

<sup>24</sup> Cf. also Thornton (1998) for an analysis of the *o*/*a* distinction in Māori in cultural terms ("mind set and spirituality", 381), i.e. in terms of *tapu* (sacredness) and *mana* (power).

#### 6.3 The semantics of possessives

However, like the other dichotomies, the active/passive opposition does not explain why *o* possession applies to actions performed. Nor does it explain well why so many interhuman relationships are mutually *o-*possessed. Biggs' conclusion seems justified, that "efforts to generalise in terms of a binary opposition have not met with general acceptance. There are always many examples where the opposition doesn't fit well, if at all." (Biggs 2000) In the next section, a different solution will be proposed.

#### **6.3.4.3** *o* **as unmarked possession**

Clark (1976: 44) suggests that the relationship between *a* and *o* in Polynesian is not symmetrical: "*\*a* […] indicates a relation of control or authority of the adjunct over the head. The relation indicated by *\*o* can best be characterised as covering all relations not included in *a.*" This idea is presented again in Biggs (2000): *a* marks an active or dominant possessor; *o* is the unmarked form, used in all other cases. Wilson (1982: 16) characterises *a-*possession as indicating relationships initiated by the possessor, while *o* is used for everything else.

There are indeed indications that the relation between *a* and *o* in Rapa Nui is not symmetrical. One such indication is the large number of family relationships which are mutually *o*-possessed. Concepts like "dependence" do not explain these well. A child depends on its parent, a person depends on his family. But does an uncle depend on his nephew, or a mother-in-law on her daughter-in-law, to warrant the use of *o*?

Another indication is suggested by those categories of *o*-possession not explained by any of the dichotomies discussed above, e.g. *o*-possessed actions, time words ('your birthday'), distributive constructions ('his day' = 'a certain day'), and epexegetical constructions ('the town of Hanga Roa').

A third indication is the asymmetry displayed within some categories: people under a leader can be either *ꞌa* or *o*-possessed, while on the other hand the leader is always *o*-possessed.

These facts can be explained by stating that *o* is the unmarked possessive marker. *ꞌA* is used to express that the possessor has an active role, which includes being in charge, responsible, or dominant with respect to the possessor; in all other cases, *o* is used. This rule correctly explains why tools and instruments (things to be used) are *a*-possessed, just like animals and plants (things to be cared for), while possessions in general are *o*-possessed.

It also explains why certain categories normally *a-*possessed may in certain cases take *o*-possession: *o-*possession does not imply a passive or dependent possessor, but only refrains from marking the possessor as active or dominant.

Thirdly, this rule explains why *o* is used in constructions where the distinction between active and passive does not play a role, such as distributives, epexegetical possessives and time words. In all these cases, *o* is used as the default marker.

Lastly, this rule goes some way to explaining the use of *a* and *o* possession for actions. A possessor is active with respect the product of his action (e.g. a feast organised, a saying uttered, a teaching performed); on the other hand, it is less clear whether a person can

#### 6 Possession

be said to be active with respect to the action as such; and indeed, here Rapa Nui tends to have *o*-possession.

#### **6.3.4.4 The** *o***/***a* **distinction and the nominal hierarchy**

In Rapa Nui there is one more indication that *o* is the unmarked form: as discussed in §6.3.2, common noun phrases and plural pronouns are *o*-possessors in all contexts, regardless their semantic relationship to the possessee. The marked form *ꞌa* is used only with a subset of nominal constituents: singular pronouns and proper nouns.

This subset coincides with a subset of the "nominal hierarchy". Certain referents are inherently more likely to function as topics of discourse, or to be agents of a verb, than others. Pronouns are more likely agents than common nouns, human referents are more likely agents than inanimates. This has led linguists to propose a nominal hierarchy – a.k.a. "animacy hierarchy" or "topic-worthiness hierarchy" – along the following lines (see Payne 1997: 150; cf. Foley 2007: 413):

(115) 1 <sup>25</sup> > 2 > 3 > proper names > humans > non-human > inanimates

Another distinction cuts partly across the hierarchy above:

#### (116) definite > indefinite

Languages may grammaticalise any part of this hierarchy, for example in case marking.<sup>26</sup> Rapa Nui has grammaticalised this hierarchy with respect to possessive marking: only pronouns and proper names, which are high on the hierarchy, may take the "active" possessive marking with *ꞌa*; elements lower on the hierarchy always get the default marking with *o*.

This leaves the question why only singular possessive pronouns have the option of taking active marking. Why do plural pronouns only get default marking, even though they are higher on the scale than proper names?

This lack of distinction in the plural cannot be explained from the nominal hierarchy as given above, but may have to do with the behaviour of singular and plural in general. Dixon (1994) observes that languages sometimes have more distinctions in the singular than in the plural. Distinctions that exist in the singular, may be neutralised in the plural.

This fact itself may have something to do with the nominal hierarchy. Just like proper names are more topic-worthy than common nouns, and definite nouns more topic-worthy than indefinite nouns, it is conceivable that singular referents are more topic-worthy than plural referents. In all cases a highly individuated referent is more topic-worthy than a less individuated one; highly individuated (singular, definite) referents tend to be topics of discourse.

<sup>25</sup> The numbers refer to first, second, and third person respectively. The complete hierarchy also includes 1st , 2 nd and 3rd person agreement, a category not relevant for Rapa Nui.

<sup>26</sup> In some languages, only constituents high on this hierarchy get accusative case-marking (i.e. are casemarked when used as Patient), while only elements lower on the hierarchy get ergative case-marking (i.e. are case-marked when used as Agent). (See Dixon 1994.)

We may therefore tentatively add another dimension which cuts across the nominal hierarchy:

(117) singular > plural

Under this hypothesis, Rapa Nui makes the *ꞌa*/*o* distinction for a subset of nominal referents which is high on the nominal hierarchy. Items lower on the hierarchy always take the default *o* marking.

### **6.4 Conclusions**

Possessive constructions are widely used. They occur as noun phrase modifiers and as nominal predicates, but may also be used to mark arguments in a verbal clause; the latter happens in the actor-emphatic construction, in clauses introduced by *mo* 'in order to', and occasionally in main clauses.

Possessives are united by the use of a possessive preposition; they are distinguished along three parameters:


Forms with *to* and *ta* are used when the possessor is in determiner position; in older Rapa Nui, they are also found as possessive clause predicates. In all other contexts, Øforms are used.

Possessive constructions express a wide range of semantic relationships, including attributes, parts, verb arguments, and various kinds of associations. They may express prospective possessive relationships, relationships which do not yet hold but are expected to come into being: 'I am looking for my wife to marry'; 'let's search our eggs in the field'.

As in other Polynesian languages, certain relationships are marked with *o*, others with *ꞌa*. Various proposals havs been made in the past to characterise the *o/a* distinction, but the only way to account for the wide range of *o-*marked relationships is to view *o* as default marker; *ꞌa* is only used when the possessor is dominant and/or active in relation to the possessee.

The idea that *o* is the default marker is confirmed by the fact that for plural pronouns and common nouns, *o* is the only marker used, while *ꞌa* is limited to singular pronouns and proper nouns. This can be explained by an expanded version of the nominal hierarchy which has been shown to play a role in various grammatical areas crosslinguistically: only nominal constituents high in this hierarchy exhibit the *o/a* distinction.

## **7 The verb phrase**

### **7.1 The structure of the verb phrase**

In Rapa Nui, the verb phrase consists of a verb, usually preceded by a preverbal marker, and often followed by one or more particles which contribute aspectual, spatial or other nuances.

The structure of the verb phrase is shown in Table 7.1 and Table 7.2. 1


Table 7.1: The verb phrase: preverbal elements

Table 7.2: The verb phrase: postverbal elements


The preverbal constituents 2–5 may occur in different orders, depending on their relative scope. For examples, see (97–100) on p. 342.

<sup>1</sup> Adapted and expanded from Weber (2003b: 26).

#### 7 The verb phrase

In this chapter, the various elements occurring in the verb phrase are discussed. The largest section (§7.2) is devoted to the set of five aspect markers. Aspect markers can largely be described in terms of well-known categories such as perfectivity and imperfectivity; one marker, however (*ka*) is more elusive.

Another major topic is directional marking (§7.5). Two directional markers are used in various ways to orient events with respect to a deictic centre, pointing either towards or away from this deictic centre.

Shorter sections deal with preverbal particles (§7.3), evaluative markers (§7.4) and postverbal demonstratives (§7.6). Finally, §7.7 deals with serial verb constructions, a construction in which two or more verbs occur in a single verb phrase.

As the tables show, the first slot (labelled A/M, aspect/mood) contains not only aspect markers but a variety of other particles as well. Two aspect markers also mark imperative mood; this is discussed in §10.2. The preverbal slot is also home to a group of subordinating particles; these are discussed in §11.5. Two negators occur in the same position; these are discussed in §10.5.

Finally, verb phrase adverbs are discussed with other minor parts of speech in §4.5.1. The particle *hia* 'yet', which occurs in combination with negators, is discussed in §10.5.8.

### **7.2 Aspect marking**

### **7.2.1 Introduction**

As the chart in the previous paragraph shows, the first slot in the verb phrase may be occupied by particles of various nature: aspect markers, subordinators and negators. This means that a verb is either marked for aspect, introduced by a subordinator, or negated by *kai* or *(e) ko*. Combinations of these are impossible. This means, for example, that purpose clauses introduced by *mo* and clauses negated with *kai* are not marked for aspect.<sup>2</sup>

In this section, the use of the aspectual markers is discussed.<sup>3</sup> This discussion will make clear that all markers have indeed an aspectual value and do not mark tense. In other words, they do not specify how the event is located in time, whether it happens before, at, or after the time of utterance. Rather, they are concerned with the internal temporal structure of the event and how the event is temporally related to other events in the context. The aspect markers are listed in Table 7.3.

Certain aspectual functions are expressed by a combination of an aspectual marker and one or two postverbal particles; these particle combinations will be discussed as a whole.

The discussion in this section is largely restricted to main clauses. The use of aspectuals with subordinate clauses (complement clauses, relative clauses and adverbial clauses) is discussed in Chapter 11. As certain subordinate clause types are strongly linked to – and

<sup>2</sup> Neither is aspect marked when the verb is nominalised (§3.2.3).

<sup>3</sup> This section is largely based on the analysis of all clauses in a subcorpus of 29 texts: 15 old texts, containing 2597 clauses; 14 new texts, containing 5834 clauses.

7.2 Aspect marking


Table 7.3: Overview of aspect markers

highly illustrative of – certain aspectuals, reference will be made to Chapter 11 where appropriate.

### **7.2.2 The obligatoriness of aspectuals**

In most contexts, the use of aspectuals is obligatory. Verbs without aspectuals occur in the following situations:


Otherwise, aspectuals are occasionally omitted clause-initially (and especially sentence-initially) in an informal style, if the verb is followed by one or more postverbal particles. As the following examples show, different aspectuals can be omitted. In (1), the perfect aspect marker *ko* is implied; the usual perfect aspect construction is *ko V ꞌā* (§7.2.7). In (2), neutral *he* is implied: the verb is followed by *rō ꞌai*, which points towards the construction *he V rō ꞌai* (§7.2.3.3); the second clause in (2) shows the same construction in full, with aspect marker.


#### 7 The verb phrase

### **7.2.3 Neutral** *he*

#### **7.2.3.1 Introduction**

*He* is the most common aspect marker. It probably developed from the nominal predicate marker *he* (§5.3.4). This development took place only in Rapa Nui – no other Polynesian language has an aspect marker cognate to *he* – so we may tentatively conclude that it took place after the language split off from PEP.<sup>4</sup>

While it may go too far to consider nominal and verbal *he* as one particle synchronically, the two are very similar in function. The nominal predicate marker *he* marks noun phrases as predicates, without attributing any aspectual value to them. Aspect concerns the internal temporal structure of an event; as entities (expressed in a noun phrase) do not have an internal temporal structure, they cannot be marked for any specific aspect. In the same way, the aspectual *he* is the least specific of all aspect markers.<sup>5</sup> Englert (1978: 64) calls it "the most general, most used and least precise tense" (my trl.). Chapin (1978: 153) labels it as a "neutral marker", a term I adopt in this grammar (gloss ntr). The range of use of *he* will be discussed in the next section; the examples will make clear that *he* is used in a wide variety of clauses; these clauses may be punctual, durative or habitual; they may convey events in a narrative, future events or instructions. This confirms the idea that *he* itself expresses none of these functions, but is a neutral marker. The aspectual value of the clause is not expressed by *he* as such, but can be deduced from the nature of the verb and/or the context. In other words, *he* is functionally unmarked.<sup>6</sup>

In many cases, a *he*-marked clause depends on other clauses in the context for its aspectual value. In narrative, a perfective clause may set the scene, after which a series of *he*-marked clauses follow (see (4) below). Another example: *he* may mark a series of instructions, but only when the first of these is explicitly marked as imperative (see (5) below).

*He* is rare in subordinate clauses, which may also be due to its neutral character. Subordinate clauses typically stand in some temporal or aspectual relation to their main clause, whether simultaneous, overlapping, contiguous, anterior or posterior. *He* is not able to supply this temporal link, hence it is not suitable in these contexts.

#### **7.2.3.2 Range of use**

As indicated above, *he* does not express any specific aspect; rather, it depends on the context for its aspectual value. In this section, this will be illustrated through examples of different contexts in which *he* is used.

<sup>4</sup> Interestingly, Cook (1999: 57) gives an example of *he* in Hawaiian preceding a verb.

<sup>5</sup> Another phenomenon linking the predicate marker *he* and the aspectual *he*, is that the negation *ꞌina* is either followed by *he* + noun (never by the article *te*), or by *he* + verb (never by a different aspectual). See section §10.5.1.

<sup>6</sup> Chapin (1978) suggests a different unified account for nominal and verbal *he*: the "noun" after *he* could be a verb, i.e. in *he taŋata koe* 'you are a man', *taŋata* could be analysed as a verb, an analysis also proposed (though in a more cautious wording) by Finney & Alexander (1998: 22). This analysis is syntactically implausible, however, as *he* is followed by a true noun phrase. As the examples in section §5.3.4 show, the noun following *he* may be preceded and followed by noun phrase elements like adjectives and numerals, while verb-phrase particles like *rō*, *atu* and *ꞌai* are excluded.

7.2 Aspect marking

**7.2.3.2.1** In narrative, *he*-marked clauses express the theme line of a story. Strings of *he-*clauses constitute the "back bone" of a story, describing the sequence of narrative events.<sup>7</sup> The following is a typical example:

(3) *He* ntr *eꞌa* go\_out *mai* hither *a* prop *nua* Mum *he* ntr *haka* caus *rivariva* good:red *he* ntr *oti* finish *he* ntr *eꞌa* go\_out *he* ntr *turu* go\_down *ararua* the\_two *ko* prom *Eva* Eva *he* ntr *tuꞌu* arrive *ki* to *Haŋa* Hanga *Piko.* Piko 'Mum went out, she made preparations, she finished, she went out, she went down with Eva, they arrived in Hanga Piko.' [R210.038]

Other aspectuals may interrupt the stream of *he-*clauses to indicate non-theme line elements of the narrative; they serve for example to provide background information, to express events anterior to the theme line, and to mark events which are highlighted in some way. (See the discussion on perfective *i* in §7.2.4.2 below.)<sup>8</sup>

The string of *he-*clauses providing the theme line of the story is usually preceded by one or more clauses which establish a time frame in which the events take place. The following example is the beginning of a story. The story starts with a cohesive clause, a temporal clause providing a time frame for what follows, marked with perfective *i* (§11.6.2.1). After that, the story continues with *he*-marked clauses.

(4) *I* pfv *poreko* born *era* dist *a* prop *Puakiva,* Puakiva *he* ntr *māuiui* sick *a* prop *Kuha,* Kuha *tōꞌona* poss.3sg.o *matuꞌa* parent *vahine.* female 'When Puakiva was born, his mother Kuha got sick.' [R229.001]

The fact that the time reference is established beforehand, confirms the idea that *he* is a neutral aspect marker: *he* has no temporal or aspectual value of its own, but continues within a previously established framework.<sup>9</sup>

In other text types, theme-line clauses are also marked with *he*. For example, in procedural texts the theme line consists of a series of steps which are taken to perform a certain procedure: building a boat, performing a burial, making a traditional cape. In the following example, the speaker describes how to prepare a certain medicine. The first step of the procedure is indicated by the imperative *e*, conveying a general instruction; this is followed by a series of *he*-marked verbs.

<sup>7</sup> The following examples are translated more literally than usual in this grammar, to convey the idea of the concatenation of *he*-clauses.

<sup>8</sup> The interplay of *he* and other aspectuals and their respective functions in narrative have been analysed by Weber (2003b).

<sup>9</sup> Notice that *he* itself does not explicitly indicate either that the events happen sequentially; there is no 'and then' sense implied in *he* as such. This feature is understood in the narrative context (cf. Hooper 1998: 127 on zero-marked narrative events in Tokelauan).

#### 7 The verb phrase

(5) *E* exh *haka* caus *pihaꞌa* boil *i* acc *te* art *vai.* water *Ki* when *oti* finish *he* ntr *toꞌo* take *mai* hither *he* ntr *huri* turn *he* ntr *haka* caus *tano* correct *te* art *matuꞌa puaꞌa matuꞌa puaꞌa ki* to *roto* inside *o* of *te* art *vai* water *pihaꞌa.* boil 'Boil water. When done, take it and pour the right amount of *matuꞌa puaꞌa* (a medicinal plant) into the boiling water.' [R313.160–161]

**7.2.3.2.2** *He*-marked clauses may refer to durative or habitual actions as in (6),<sup>10</sup> or general truths as in (7).


**7.2.3.2.3** *He* is used with stative verbs (including adjectives) to express a state of affairs which holds at the time of reference. This may be the time of speech as in (8), or the time of the narrative as in (9).


**7.2.3.2.4** *He*-marked clauses may express events that are about to happen or foreseen in the (near) future. The time frame is established in the context ('next year' in (10)).

(10) *Matahiti* year *ena* med *he* ntr *hoki* return *a* prop *au* 1sg *ki* to *te* art *hāpī.* learn 'Next year I will return to school.' [R210.003]

To express the future character of the action explicitly, *e V rō* is used (§7.2.5.3). Clauses expressing plans or intentions may also be marked with *ka* (§7.2.6.3).

<sup>10</sup> To mark durativity or habituality explicitly, *e V era* or *e V ꞌā/ꞌana* is used (§7.2.5.4).

7.2 Aspect marking

#### **7.2.3.3** *He* **and postverbal particles**

**7.2.3.3.1 Demonstratives** Unlike the aspectuals*i*, *e* and *ka*, *he* is rarely followed by one of the postverbal demonstratives *nei*, *ena* or *era* (§7.6).<sup>11</sup> Occasional examples are found:

(11) *He* ntr *moe* lie *era* dist *koe* 2sg *e* voc *Hina* Hina *ē* voc *ꞌi* at *te* art *kata.* laugh 'Hina laughed her head off (lit. you, Hina, lied down laughing).' [R313.025]

**7.2.3.3.2** *He V rō ꞌai* More common is the construction *he V rō ꞌai*: a *he*-marked verb followed by the asseverative particle *rō* (§7.4.2) and the sequential particle *ꞌai*. Weber (2003b: 125) shows that *he V rō ꞌai* in narrative texts indicates notable, important events on the theme line: significant developments or culminating points in the story. My analysis of several narrative texts confirms this. *he V rō ꞌai* clauses indicate events which are either climactic in a story, final in a sequence, or both.

In other cases, *he V rō ꞌai* marks an event which is not only final in a series, but which constitutes a climax in the story. In the following example, a sequence of events is concluded with *he V rō ꞌai*: the woman tries to catch her child, which has turned into a fish, but in vain: the child disappears. The last event, the climax of the sequence, is marked with *rō ꞌai*.

(12) *He* ntr *tute* chase *he* ntr *oho* go *e* ag *te* art *viꞌe* woman *nei…* prox *ꞌe* and *he* ntr *ŋaro* disapeaar *rō* emph *atu* away *ꞌai.* subs 'The woman chased the fish…. but it disappeared.' [R338.009]

The use of *rō* in this construction conforms to the general sense of *rō*, asserting the reality of the event (§7.4.2).

*He V rō ꞌai* is also used at points of emotional intensity; in the following example (from the same story as (12)), the mother is grieved because her child has disappeared.

(13) *Te* art *matuꞌa* parent *vahine* female *o* of *te* art *poki* child *nei* prox *he* ntr *taŋi* cry *rō* emph *atu* away *ꞌai.* subs 'The mother of the child cried.' [R338.008]

#### **7.2.3.4 Summary**

The discussion above has shown that *he* does not express one single aspect. It is used in punctual, durative, habitual and stative clauses; the verb may refer to a timeless truth, a narrative event or a future event. This wide range indicates that *he* is a neutral aspect marker, which in itself does not express any aspect. The aspectual value of the clause is contributed by the context, for example a time phrase, a temporal clause or a preceding imperative.

<sup>11</sup> *He V era* occurs relatively often in the stories recorded by Métraux; however, this probably represents the much more common construction *e V era*: Métraux, whose first language was French, sometimes took initial glottals for *h* and vice versa.

#### 7 The verb phrase

*He* is especially common in sequences of clauses expressing successive events; this happens both in narrative and procedural discourse.

### **7.2.4 Perfective** *i*

#### **7.2.4.1 Introduction**

*I* is the perfective marker.<sup>12</sup> The perfective aspect presents an event as a single, unanalysable whole (Comrie 1976: 3; Dixon 2012: 35), without considering its internal structure (e.g. its duration). In other words, the perfective regards the event from the outside, while the imperfective considers its temporal make-up from the inside.

Perfective aspect is naturally correlated with past tense (Comrie 1976: 72), and in fact, *i* usually marks events in the past. *I* has been characterised as a past tense marker in several descriptions of Rapa Nui and other Polynesian languages.<sup>13</sup> In non-narrative contexts *i* is the common aspectual for past events, as the following examples show. As (17) shows, it may also express general facts about the past.


There are cases, however, where *i* conveys a non-past event. For example, in (18) *i* is used with reference to the future:

(18) *I* pfv *oꞌo* enter *era* dist *koe* 2sg *ki* to *roto* inside *i* at *tuꞌu* poss.2sg.o *hare* house *era* dist *e* ipfv *noho* stay *koe.* 2sg 'When you have entered into your house, stay there.' [R310.297]

<sup>12</sup> Perfective *i* is common in Eastern Polynesian languages; non-EP languages have *na*, *ne* or *ni*. Wilson (2012: 314) suggests a development PNP *\*ne* > Central Northern Outliers *\*ni* > PEP \**i*.

<sup>13</sup> See Du Feu (1996: 156) for Rapa Nui, Mutu & Teìkitutoua (2002) for Marquesan, Académie Tahitienne (1986: 172) for Tahitian, Biggs (1973: 34) for Māori. Note that Chapin (1978: 153) labels Rapa Nui *i* as perfective.

7.2 Aspect marking

Conversely, other aspectuals are used besides *i* in clauses referring to past events: narrative *he* (§7.2.3), imperfective *e V ꞌā* (§7.2.5.4). This means that *i* is not a past tense marker; rather, it expresses that an action is temporally closed. This may in turn mean that the event is in the past, or anterior to other events, or finished at a certain point, but neither of these is a necessary condition for the use of *i*.

Comrie (1976: 17–18) stresses that perfective is not the same as punctual. This is true in Rapa Nui as well: while *i* often marks punctual events, it is equally used to mark events that have a certain duration. This is clear in examples like the following, where the perfective is used for events that take place over many years:

(19) *A* prop *Te* Te *Manu* Manu *i* pfv *noho* stay *ai* pvp *ꞌi* at *muri* near *i* at *tū* dem *māmātia* grandmother *era* dist *ōꞌona* poss.3sg.o *ꞌātā* until *ki* to *te* art *nuinui* big:red *iŋa.* nmlz 'Te Manu stayed with his aunt until he had grown up.' [R245.246]

In main clauses, the *i*-marked verb is usually followed by a postverbal demonstrative (PVD). The use of PVDs after *i*-marked verbs will be discussed in more detail in §7.6.5.

#### **7.2.4.2 Neutral** *he* **versus perfective** *i*

As discussed in §7.2.3.2 above, *he* is used to mark the theme line of discourse. This means that the relation between *he* and *i* calls for an explanation. As Timberlake (2007: 293) points out, the perfective is typically the aspect of narrative texts: a perfective event leads to a new state, which is the input for the next event; a string of such events constitutes a narrative. In Rapa Nui, however, sequential events in a narrative are marked with *he*, not *i* (§7.2.3.2 above).

As shown in this section, *i* is used when the event is not sequential to the event in the preceding clause, for example in clauses providing background information. Moreover, *i* is used to highlight events, setting them of from the theme line of *he*-marked clauses. This means that *i* is used in narrative discourse to mark clauses not belonging to the theme line for some reason.

**Anteriority** *I* is used when the event is anterior with respect to the theme line of the story (i.e. where the English equivalent is the pluperfect).

(20) *Māuiui* sick *nei* prox *i* pfv *tuꞌu* arrive *mai* hither *ai* pvp *ki* to *Rapa* Rapa *Nui* Nui *o* of *te* art *nuꞌu* people *empereao* employee *o* of *te* art *Compañía* company *i* pfv *maꞌu* carry *mai.* hither

'This disease had arrived on Rapa Nui, introduced by the employees of the Company.' [R250.084]

*I* with anterior events is especially common in cohesive clauses, temporal clauses preceding a main clause (§11.6.2.1).

#### 7 The verb phrase

**Highlighting** *I* marks events which the speaker wants to highlight in the stream of *he*-clauses. In the following example, Kalia, the protagonist of the story, has been swimming all night to get to Ao Tea Roa to get help for the people of Kapiti. The moment in which she finally arrives and is able to warn the people of Ao Tea Roa, is marked with *i V ai*. As this example shows, the significance of the event may be underlined by the asseverative particle *rō* (§7.4.2).

(21) *Kai* neg.pfv *puhi* blow *rivariva* good:red *ia* then *te* art *haŋu* breath *ꞌi* at *te* art *poto* short *o* of *te* art *aho* breath *i* pfv *ohu* shout *rō* emph *atu* away *ai* pvp *mo* for *haka* caus *ꞌite* know *i* acc *tū* dem *ꞌati* problem *era.* dist 'Short of breath, she shouted to make the trouble known.' [R347.128]

**Intervening events** More specifically, *i* is used when the clause expresses what may be called an intervening event. As Comrie (1976: 3) indicates, the perfective sees the action as an unanalysable whole, without an internal temporal structure. Therefore, the perfective is used in many languages to express punctual events. In Rapa Nui, the perfective is often used with punctual events which take place while something else is happening. The punctual event interrupts another event which has been going on for some time: it intervenes into an existing situation.

This is common after the imperfective *e V nō ꞌā* (§7.2.5.4):


'When he was going up, the wife of Te Manu's father saw him.' [R245.214]

**Background** *I*-marked clauses may express background information. For example, in the introduction of a story, *i*-clauses may serve to set the stage by telling what happened before the beginning of the story, as in (24). *I-*marked clauses may also express restatements or clarifications, as in (25).<sup>14</sup>

(24) *Te* art *ara* way *nei* prox *o* of *te* art *nuꞌu* people *nei,* prox *i* pfv *eꞌa* go\_out *ai* pvp *mai* from *Haŋa* Hanga *Roa* Roa *o* o *Tai* Tai *ꞌi* at *ruŋa* above *o* of *te* art *vaka* boat *nei.* prox

'As to these people's trip, they had left Hanga Roa o Tai by boat.' [R361.004]

<sup>14</sup> Similarly, *i*-marked clauses may express background events in subordinate clauses (§11.6.2.2).

7.2 Aspect marking

(25) *…he* ntr *iri* ascend *he* ntr *oho* go *ki* to *te* art *kona* place *hare* house *era.* dist *I* pfv *tahuti* run *a* prop *Tiare* Tiare *i* pfv *iri* ascend *ai* pvp *ki* to *te* art *kona* place *hare* house *era.* dist

'…she went up to her home. Running, Tiare went up to her home.' [R151.053]

#### **7.2.4.3 Summary**

*I* is the perfective marker: it marks events which are viewed as a whole, without internal temporal structure. The event is usually, but not always, in the past.

In narrative, *i* is used for events which stand out in some way from the thematic backbone of events marked with *he*: *i* marks background events, restatements and conclusions, flashbacks, but also events which are highlighted.

### **7.2.5 Imperfective** *e*

#### **7.2.5.1 Introduction**

*E* is the imperfective marker. It is common throughout Polynesian languages; Pollex (see Greenhill & Clark 2011) glosses it as 'non-past'. According to Comrie (1976: 24), the imperfective makes "explicit reference to the internal temporal structure of a situation, viewing a situation from within" (see also Dixon 2012: 35). Languages may grammaticalise certain subcategories of the imperfective; Comrie divides the imperfective into two subcategories: continuous (an event or situation goes on for some time) and habitual ("a situation which is characteristic of an extended period of time", 27–28). The continuous can be further divided into progressive and nonprogressive: in various European languages, stative verbs may be used in the imperfective (with a continuous interpretation), but not in a progressive form. The progressive is thus a combination of a continuous meaning and non-stativity (35–36).<sup>15</sup>

As it will turn out, the categories mentioned here are relevant in Rapa Nui as well. While *e* as such expresses imperfectivity, finer distinctions are expressed by *e* in combination with certain postverbal particles. Thus the aspectual value of the clause is defined not by *e* alone, but by a combination of *e* and postverbal particles. The following particles contribute to the aspect of the clause: the evaluative markers *rō* and *nō*, the continuity marker *ꞌā/ꞌana*, <sup>16</sup> and the postverbal demonstratives (PVDs) *nei/ena/era*. With *e*, these particles show the following cooccurrence restrictions:

(26) *e e* V V (adverb) (adverb) (*rō/nō*) (*mai/atu*) (*mai/atu*) *nei/ena/era* (*ꞌā/ꞌana*)

<sup>15</sup> Others consider "continuous" and "progressive" as synonymous, see e.g. Dixon (2012: 34).

<sup>16</sup> There is no difference in function between *ꞌā* and *ꞌana*; *ꞌā* is more common (§5.9). In this section, *ꞌā* will be used as a shorthand for *ꞌā/ꞌana*.

#### 7 The verb phrase

In other words, PVDs after *e* do not cooccur with either the evaluative markers *rō* and *nō* or the continuity marker *ꞌā/ꞌana*, <sup>17</sup> but the latter two categories do occur together.<sup>18</sup>

In the following sections, different constructions with *e* will be discussed: bare *e* (i.e. without any postverbal particle) is briefly discussed in §7.2.5.2, *e V rō* in §7.2.5.3. *E V era* and *e V ꞌā* (which largely occur in the same contexts and have similar functions) are treated together in §7.2.5.4. Finally, in §7.2.5.5, the distinction between *e V era* and *e V ꞌā* is explored.

#### **7.2.5.2 Bare** *e*

The aspect marker *e* without any postverbal particle occurs in two contexts only:

	- 2. in the imperfective actor-emphatic construction (§8.6.3):

In all other contexts, the *e*-marked verb is followed by one or more postverbal particles.

#### **7.2.5.3** *e V rō***: future**

The combination of imperfective *e* and the emphatic particle *rō* expresses future events. It is used to express intentions or plans:


<sup>17</sup> This restriction is specific to imperfective *e*; after other aspectuals, postverbal demonstratives do co-occur with *rō*, *nō* and *ꞌā/ꞌana*.

<sup>18</sup> In fact, these two categories co-occur far more often than one would statistically expect: while *e V ꞌā* occurs 35 times and *e V nō*/*rō* occurs 34 times, *e V nō/rō ꞌā* occurs no less than 153 times.

7.2 Aspect marking

#### **7.2.5.4** *E* **with postverbal demonstratives and with** *ꞌā/ꞌana*

As pointed out in §7.2.5.1 above, *e* is used in combination with both PVDs and *ꞌā*. <sup>19</sup> With either of these, the clause has either a habitual or a continuous sense, both of which are subcategories of the imperfective. The question is, whether there is any difference between *e V PVD* and *e V ꞌā*. In this section the use of *e* with these markers is discussed. This discussion will show that there is a great deal of overlap between both constructions, but that there are differences in use as well.

*E V PVD* and *e V ꞌā* occur in main clauses and in temporal subordinate clauses. The former are discussed here, the latter will be discussed in §11.6.2.2 (see (225–228)).

In main clauses, *e V PVD* expresses either a continuous action as in (31–32), or a habitual action as in (33–34):


*E V ꞌā* also expresses either continuous actions as in (35–36) or habitual actions as in (37); the latter is not very common, though.


<sup>19</sup> In this section, *ꞌā* is a shorthand for *ꞌā/ꞌana*.

#### 7 The verb phrase

	- *e V ꞌā* is also used with adjectives, expressing an enduring state:<sup>20</sup>

By contrast, *e V PVD* is rarely used with statives. It never occurs with adjectives of dimension, value of colour (the prototypical adjectives, see §3.5.1.3), only with adjectives from other categories:

(40) *¿He* pred *aha* what *e* ipfv *aŋarahi* difficult *ena* med *mo* for *haka* caus *rehu* forgotten *ōꞌoku* poss.1sg.o *i* acc *a* prop *koe?* 2sg 'Why is it difficult to forget you?' [R452.025–026]

Table 7.4 summarises these findings. Plain x indicates that the category in question is common; (x) indicates uncommon or restricted occurrence.


Table 7.4: Functions of *e V PVD* and *e V ꞌā*

<sup>20</sup> The most frequent stative use of *e V ꞌā* is with the existential verb *ai*: the fossilised expression *e ai rō ꞌā* 'there is' is a very common existential construction (§9.3.1).

7.2 Aspect marking

#### **7.2.5.5 Postverbal demonstratives versus** *ꞌā***; the function of** *ꞌā*

As the discussion in §7.2.5.4 shows, there is a great deal of overlap between *e V PVD* and *e V ꞌā*. Both are used in a habitual and a progressive sense; both are found in main and subordinate clauses. Even so, the two cannot always be used interchangeably. One difference lies in the possibility to express additional meaning elements: as shown in §7.2.5.1, PVDs do not co-occur with the evaluative particles *nō* or *rō* (§7.4); in order to use one of these markers in an imperfective clause, *ꞌā* must be used instead (see (37–39) above).

While *ꞌā* can be used together with *nō* and *rō*, the PVD also has some possibilities of its own: different PVDs indicate different degrees of distance. The default – and by far the most frequent – PVD is *era*, as in (31–34) above; *nei* can be used to indicate proximity to the speaker as in (41), *ena* to indicate proximity to the hearer as in (42).


Apart from these possibilities to express additional meaning elements, there is a more general difference between *e V era* and *e V ꞌā*. This is suggested by two facts:


This raises the question of the function of the marker *ꞌā*. According to Weber (2003b: 52), *ꞌā* is a progressive marker. This would fit many of its occurrences; however, it should be noted that progressive events may also be expressed by *e V PVD*. Moreover, *e V ꞌā* can be used with stative verbs, whereas the progressive (in Comrie's definition, see §7.2.5.1 above) is limited to nonstative verbs.

Another fact which should be taken into consideration, is that *ꞌā* occurs after the perfect marker *ko/ku* as well; in fact, after the perfect marker *ꞌā* is obligatory. Now perfect aspect is incompatible with the progressive; rather, it indicates the continuing relevance of a situation which has come about in the past. As will be discussed in §7.2.7.1 and §7.2.7.4 below, *ko V ꞌā* expresses a state of affairs resulting from an earlier event, rather than the event itself. We may conclude that *ꞌā* marks continuity or stability over

#### 7 The verb phrase

time: *e V ꞌā* expresses that an event or a state continues; *ko V ꞌā* indicates the continuing relevance of a state which has started in the past.<sup>21</sup> Hence the gloss cont(inuity).<sup>22</sup>

Notice that this does not mean that *ꞌā* as such is a marker of continuous aspect. Continuous aspect (expressing events which continue for some time, whether stative or nonstative) is a subcategory of the imperfective, which is expressed by either *e V PVD* or *e V ꞌā*. *ꞌĀ* itself simply emphasises the continuity or stability of a situation, whether in combination with imperfective *e* or perfect *ko*.

This is confirmed by the occasional use of *ꞌā* after the preverbal marker *mai* (§11.5.5): *mai* as such indicates a temporal boundary ('before, until'); in combination with *ꞌana* it expresses the continuation of a state up to a certain point: 'while, as long as'.

The meaning of postverbal *ꞌā* is clearly related to the meaning of postnominal *ꞌā* (§5.9); while postverbal *ꞌā* indicates stability of an event over time, postnominal *ꞌā* underlines the identity of a referent, i.e. stability in reference: 'the same, himself'.

#### **7.2.5.6 Summary**

*E* is the imperfective marker. Its temporal/aspectual value is further defined by certain postverbal particles, as indicated in Table 7.5.


Table 7.5: Functions of imperfective e

In clauses where the verb is non-initial, *e* tends to be used whenever the clause has nonpast reference; this will be briefly discussed in §7.2.8 below.

<sup>21</sup> *Ana* occurs in other Eastern Polynesian languages (which have not retained the Proto-Polynesian glottal plosive) as a post-verbal particle marking a continuing action or state, usually after imperfective *e*, e.g. Hawaiian (Elbert & Pukui 1979: 57–60), Marquesan (Mutu & Teìkitutoua 2002: 67), Mangarevan (Janeau 1908: 32), Māori (Bauer, Parker & Evans 1993: 416–419). Interestingly, in Marquesan the variant *aa* is used as well; given the fact that other languages only have the longer form, it is not unlikely that Rapa Nui *ꞌā* and Marquesan *aa* are independent developments. In Hawaiian, *ana* alternates with postverbal demonstratives as in Rapa Nui. The use of *ꞌana/ꞌā* in the noun phrase (§5.9) is unique to Rapa Nui.

<sup>22</sup> In addition, *ꞌā* is used in negated perfect aspect clauses, marked with preverbal *kai* (see (124–125) on p. 501).

7.2 Aspect marking

### **7.2.6 The contiguity marker** *ka*

#### **7.2.6.1 Introduction:** *ka* **in Polynesian and in Rapa Nui**

*Ka* occurs in most Polynesian languages. It tends to be a somewhat elusive marker. Pawley (1970: 347–348) glosses PPN/PNP *\*kaa* as "anticipatory, future" and PCE *\*kaa* as "inceptive"; Pollex (Greenhill & Clark 2011) has PPN \**ka* as an inceptive marker. In most grammars of Polynesian languages, it is explained as inceptive and/or future and/or imperative; the latter function occurs only in EP languages.<sup>23</sup>

For Rapa Nui, the existing grammars offer little analysis on *ka*. Englert (1978: 63, 72) does not list or discuss *ka* among the "tenses", but only gives examples of its use in the imperative. According to Du Feu (1996: 37) *ka* and *ki* are momentary particles indicating temporal relationships between actions; she gives examples of the use of *ka* in the imperative (38), *ka V rō* in the sense 'until' (52) and *ka* in temporal clauses referring to the future. Chapin (1978: 154) indicates that there are various other uses of *ka* besides the imperative, but that on the basis of his data, it is not possible to reach any satisfactory generalisation regarding these uses.

Weber (2003b: 33), on the contrary, offers a thorough analysis of *ka*. On the basis of a number of newer narrative texts he concludes that *ka* does not give information about the aspectual value of the verb itself, but about its temporal relation to a following or preceding proposition. He postulates that *ka* indicates temporal contiguity between two events, in that the two events are temporally adjacent or overlapping.

My analysis, as outlined below, largely confirms and refines Weber's findings. In many of its uses, *ka* represents a boundary, setting off one event from another; this happens for example when one event represents a temporal limit for another, ongoing event. In other cases *ka* indicatessimultaneity with respect to the event expressed in a preceding or following clause. This simultaneity can be either total or partial (i.e. overlapping). Both situations can be subsumed under the label "contiguity" (cntg), proposed by Weber (2003b).

This section discusses the contiguity marker *ka*; the use of *ka* as imperative marker (which occurs more frequently in discourse) is discussed in §10.2.1. Weber (2003b) treats the contiguity marker and the imperative marker as different particles; in §10.2.1 I will argue that the two are best considered as a single particle.

Another use of *ka* not discussed in the present section, is *ka* preceding numerals (§4.3.2.2). The discussion and examples there show, that *ka* indicates a quantity which has been reached, a use which corresponds neatly to *ka* as a boundary marker.

In the following subsections, different contexts in which *ka* occurs, are discussed in turn. First a number of uses in subordinate clauses are briefly listed (§7.2.6.2), then its use in main clauses is discussed (§7.2.6.3). In §7.2.6.4, some minor uses of *ka* are listed.

<sup>23</sup> PPN *\*ka* reflects a Proto-Oceanic coordinating conjunction *\*ka* 'and then' (Lynch, Ross & Crowley 2002: 85; Lichtenberk 2014), which developed into a marker of sequentiality, future tense, irrealis, imperative and/or inceptive in various (groups of) languages. Evidently, the use of *\*ka* is not narrowed down to a single function in Polynesian.

7 The verb phrase

#### **7.2.6.2** *Ka* **in subordinate clauses**

*Ka* is used in a wide range of subordinate clauses. In this section, these constructions are listed with a single example; they are discussed in more detail in Chapter 11.

*Ka* occurs in complements of perception verbs (§11.3.1.1):

(43) *He* ntr *uꞌi* look *atu,* away *ka* cntg *pū* approach *te* art *manu* bird *taiko. taiko* 'She saw a *taiko* bird come by.' [Ley-9-55.078]

In relative clauses (§11.4.3), *ka* indicates events posterior to the time of reference:

(44) *Te* art *ꞌīŋoa* name *o* of *te* art *kai* food *era* dist *[ka* cntg *maꞌu* carry *mai* hither *era* dist *ki* to *a* prop *koe]* 2sg *he* pred *ioioraŋi. ioioraŋi* 'The name of the food they will bring you is *ioioraŋi*.' [R310.060]

In temporal clauses (§11.6.2.1, §11.6.2.2), *ka* indicates temporal contiguity with the event in the main clause:

(45) *Ka* cntg *hakameꞌemeꞌe* mock *era* dist *he* ntr *riri* angry *a* prop *Taparahi.* Taparahi 'When they mocked, Taparahi would get angry.' [R250.151]

*Ka* marks conditional clauses (§11.6.6):

(46) *Ka* cntg *hāŋai* feed *atu* away *ena* med *ki* to *a* prop *koe,* 2sg *he* ntr *mate* die *koe.* 2sg 'If (the two spirits) feed you, you will die.' [R310.061]

*Ka* occurs after certain temporal conjunctions: *ꞌō ira* 'before' (§11.6.2.4); *ꞌātā*/*ꞌā* 'until', *ꞌahara* 'until' (§11.6.2.5):

(47) *Mai* hither *ki* hort *hāpaꞌo* care\_for *nō* just *tātou* 1pl.incl *i* acc *a* prop *ia* 3sg *ꞌātā* until *ka* cntg *nuinui* big:red *rō.* emph 'Let us take care of him until he is big.' [R211.063]

Without a conjunction, *ka V rō* marks a temporal boundary, 'until' (§11.6.2.5):

(48) *He* ntr *kai* eat *a* prop *Te* Te *Manu* Manu *ka* cntg *mākona* satiated *rō.* emph 'Te Manu ate until he was satiated.' [R245.067]

*Ka V atu* marks a concessive clause (§11.6.7):

(49) *Ka* cntg *rahi* many *atu* away *tāꞌaku* poss.1sg.a *poki,* child *e* ipfv *hāpaꞌo* care\_for *nō* just *e* ag *au* 1sg *ꞌā.* ident 'Even if I have many children, I will care for them myself.' [R229.023]

7.2 Aspect marking

In most of these examples, *ka* expresses temporal contiguity. The event in the *ka*clause is temporally contiguous to the event in the main clause; often it indicates a boundary to the event in the main clause as in (45), (47) and (48); sometimes the event overlaps with or is simultaneous to the main clause event as in (43).

#### **7.2.6.3** *Ka* **in main clauses**

When *ka* occurs in main clauses, the clause often refers to an event posterior to the time of reference, something which happens later than other events in the context. As in subordinate clauses, the verb is often followed by a postverbal demonstrative.

In direct speech, the time of reference is the moment of speech; the *ka*-clause refers to the future, but always the immediate or very near future:


In these cases – different from the subordinate clauses in the previous section – the temporal/aspectual reference of the clause is not determined by its relation to surrounding clauses, but independently anchored in the non-linguistic context. For example, (51) forms a complete speech, so the sentence has no direct linguistic context. The contiguity marker indicates that the event is contiguous to the time of reference, in this case, the moment of speech.

Posterior *ka*-clauses also occur in narrative contexts. These clauses describe events which happen later than the main line of events. As in the examples above, the *ka*-event is posterior to the time of reference (in this case, the main line of the story). An example:

(52) *…he* ntr *oho* go *ararua* the\_two *ki* to *Santiago* Santiago *ki* to *te* art *hare* house *era* dist *o* of *Maꞌatea.* Ma'atea *ꞌI* at *ira* ana *hoꞌi* indeed *ka* cntg *noho* stay *era.* dist 'The two went to Santiago to the house of Ma'atea. There she would stay.' [R210.221]

Sometimes two successive clauses are both marked with *ka*, indicating temporal contiguity between the two events: one event marks the boundary of the other. In this construction, the first clause is a temporal clause providing a time frame for the second. The second clause is the main clause, but this can only be concluded on semantic grounds; the clauses do not differ syntactically, except in their respective order.

#### 7 The verb phrase


'When they arrived, they themselves told about their trouble.' [R361.035]

Finally, *ka* in main clauses is common after certain clause-initial particles, especially deictic particles: *ꞌī* 'here; right now'; *ꞌai* 'there (§4.5.4.1); then'.

#### **7.2.6.4 Other uses of** *ka*

Firstly, *ka* is used in an exclamative construction preceding adjectives (§10.4.1).

Secondly, as discussed in the previous sections, *ka* is commonly used to indicate temporally contiguous events. A natural derivative from this is its use to indicate alternatives. When there are two alternative events or states, either of which can be true, they can be expressed by two *ka-*clauses. An appropriate translation is 'whether … or'.


'The work was done every day, whether (you were) tired or not.' [R539-2.026]

#### **7.2.6.5 Summary**

*Ka* is best characterised as a contiguity marker: it marks events which are temporally contiguous to events in a neighbouring clause. This means that the temporal value of a *ka-*marked clause often depends on a preceding or following clause; not surprising, *ka* often occurs in a subordinating clause, relating it temporally to the main clause.

The *ka*-clause may also be related to an (implied) time of reference; it is usually posterior to this reference time.

7.2 Aspect marking

### **7.2.7 Perfect aspect** *ko V ꞌā*

Perfect aspect is marked by the aspect marker *ku/ko*, in combination with the continuous marker *ꞌana* or *ꞌā* (§7.2.5.5).

First an etymological note. The aspectual particle *ko/ku* reflects PPN *\*kua*, which serves as a perfect aspect marker in almost every Polynesian language (Clark 1976: 30).<sup>24</sup> It has the form *kua* in most languages; apart from Rapa Nui, only a few other languages have dropped the final *-a*. 25

In Rapa Nui both *ku* and *ko* are used as perfect aspect marker. On etymological grounds, *ku* must be the original form, and indeed, in older texts only *ku* is found. Today *ko* is prevalent, while the use of *ku* is limited to certain speakers.

*ꞌĀ* is a reduced form of *ꞌana*; the choice between both variants is free (§5.9). A verb marked with *ku/ko* is always followed by *ꞌana/ꞌā*.

According to Comrie (1976), the perfect aspect relates a state to a preceding situation: the perfect signals that a situation in the past has a continuing relevance in the present.

In Rapa Nui, the perfect *ko V ꞌā* <sup>26</sup> emphasises a current state of affairs. With active verbs, it refers to an event anterior to the time of reference, which has resulted in a current situation. With stative verbs, it refers to the state of affairs itself, which has started at some moment in the past. (In fact, with some verbs it is questionable whether *ko V ꞌā* refers to the anterior event or to a resulting state, an ambiguity which is inherent in the character of the perfect.) The time of reference may be in the present, in the past, or in the future; in other words, *ko V ꞌā* has no temporal value.

In the following sections, different uses of the perfect aspect will be discussed.

#### **7.2.7.1 Anterior events leading to a present situation**

With active verbs, *ko V ꞌā* indicates that the action has taken place and has led to a certain state of affairs which still holds at the time of reference. The time of reference may be the present, in which case the action took place in the past. A few examples:


<sup>24</sup> Massam, Lee & Rolle (2006: 15) mistakenly assume that preverbal *ko* in Rapa Nui is the same particle as the prominence marker *ko*. The historical data show that this cannot be the case.

<sup>25</sup> All of the latter are outliers (e.g. Takuu, Kapingamarangi), except Marquesan (Zewen 1987: 34) and Mangarevan (Janeau 1908: 61), in which the *-a* is dropped before verbs having more than two syllables.

<sup>26</sup> Henceforth, *ko V ꞌā* is used as a shorthand for *ko/ku V ꞌana/ꞌā*.

#### 7 The verb phrase

The time of reference may also be in the past. This happens especially in narrative, where *ko V ꞌā* clauses relate events which have taken place anterior to theme line events. The English equivalent is the pluperfect. The following example shows the alternation between theme-line events (*he*) and anterior events (*ko V ꞌā*).

(60) *He* ntr *eꞌa* go\_out *tau* dem *poki* child *era,* dist *he* ntr *ꞌaꞌaru* grab *mai.* hither *Ku* prf *eꞌa* go\_out *ꞌā* cont *Kaiŋa,* Kainga *ku* prf *kā* kindle *ꞌā* cont *i* acc *te* art *ꞌumu,* earth\_oven *he* ntr *unu* pluck *i* acc *tau* dem *moa* chicken *era…* dist 'The child went out and grabbed (the chickens). Kainga had already gone out and lighted the fire for the earth oven; he plucked those chickens…' [Ley-8-53.004]

The time of reference may be in the future: at a certain point in time something will have happened.

(61) *Ko* prf *eꞌa* go\_out *ꞌā* cont *te* art *ŋā* pl *vārua* spirit *era* dist *ana* irr *tuꞌu* arrive *kōrua.* 2pl 'The spirits will have left when you arrive.' [R310.273]

Sometimes *ko V ꞌā* is used with action verbs without an anterior sense. The event takes place not before, but at the time of reference; for example, it takes place at the same time as events in the immediate context which are marked with *he*. In these cases *ko V ꞌā* emphasises the completed character of the event: the event is done as soon as it is started. An example is the following.

(62) *Hora* hour *hitu* seven *ko* prf *oꞌo* enter *ꞌā* cont *ki* at *rote* inside\_art *hare* house *pure* pray *ki* to *te* art *pure.* prayer 'Seven o'clock they entered into the chapel for prayer.' [R210.140]

The perfect emphasises that at seven o'clock the action of entering was over and done with; in other words, it took place at exactly seven o'clock.

#### **7.2.7.2 Present states**

With stative verbs, *ko V ꞌā* is frequently used to indicate that a state of affairs has been reached. Use of the perfect aspect suggests that some change has taken place, leading to the situation at the time of reference; in other words, the situation has not always been there, but is the result of some unspecified prior process.<sup>27</sup>

Here are a number of examples of *ko V ꞌā* with stative verbs.

(63) *Ko* prf *veꞌaveꞌa* hot:red *ꞌā* cont *ꞌi* at *te* art *rahi* much *o* of *te* art *māuiui.* sick 'She was hot because of her grave illness.' [R229.229]

<sup>27</sup> Cf. Comrie (1976: 57): in many languages, present states are expressed using the perfect, whereas in English, the present is used in such cases: Greek *tethnēkenai*, English 'be dead'.


In all these cases *ko V ꞌā* retains its character as a perfect aspect marker: the present situation is one which has not always existed, but which has come about at some point, often quite recently.

The range of verbs which commonly take *ko V ꞌā* is wide. Roughly speaking, three categories can be distinguished:

In the first place: physical and mental states, including for example pain, sickness, anger, happiness. Also included in this category are *haꞌuru* 'to sleep', *ora* 'to live' (prf 'to recover'), and *mate* 'to die' (prf 'to be dead'), as well as verbs with a more active sense like *kata* 'to laugh', *taŋi* 'to cry', *ꞌekiꞌeki* 'to sob'.


Secondly: verbs of volition.


Thirdly: verbs of perception (esp. *ŋaroꞌa* 'to hear/perceive') and cognition.

(71) *Ko* prf *ŋaroꞌa* perceive *mai* hither *ꞌana* cont *ꞌō* really *e* ag *au* 1sg *te* art *hauꞌa* smell *huru* manner *kē* different *o* of *te* art *kai* food *nei.*

prox

'I smell a strange smell of this food.' [R236.026]

#### 7 The verb phrase

(72) *E* voc *nua,* Mum *ꞌi* at *te* art *hora* time *nei* prox *ko* prf *ꞌite* know *ꞌana* cont *a* prop *au* 1sg *i* acc *te* art *parautiꞌa.* truth 'Mum, now I know the truth.' [R229.495]

Examples (70–72) show, that the use of the perfect aspect with a present sense is not limited to prototypical stative verbs. Verbs like *ꞌite*, *ŋaroꞌa* and *haŋa* are clearly active: they are transitive verbs, the subject of which can be marked with the agent marker *e* (§8.3.2), yet they tend to have the perfect aspect marker.

#### **7.2.7.3** *Ko V era ꞌā***: 'well and truly finished'**

The verb phrase marked by *ko V ꞌā* may contain the demonstrative particle *era*. As discussed in §7.6.4, this particle indicates spatial or temporal distance. When used in a perfect aspect clause, *era* underlines the temporal and conceptual distance between the time of reference and the time at which the event took place: the action is well and truly finished, possibly a considerably time ago.<sup>28</sup> Often 'already' is an appropriate translation.


#### **7.2.7.4 Perfect** *ko V ꞌā* **versus perfective** *i*

As discussed in §7.2.7.1 above, *ko V ꞌā* marks anterior events leading to a present state. Now it is worthwile to compare the use of perfect *ko V ꞌā* and perfective *i*. Both are used to mark events in the (recent) past; to repeat two examples with *i* from §7.2.4.1:


<sup>28</sup> This does not mean that *ko V era ꞌā* indicates a pluperfect, though it can be used in pluperfect sense (§7.2.7.1 above gives examples where *ko V ꞌā* marks pluperfect events).

7.2 Aspect marking

These examples illustrate a typical use of *i*: in many cases, *i*-marked clauses express not just a past event, but an event which has a bearing on the present: the event has led to a state which is relevant right now. For example, in (75), the subject has just arrived, leading to a situation in the present; 'I came' results in 'I am here now'. And in (76), 'you won' means as much as 'OK, I give in, you win', i.e. it describes a current situation, not just something which happened in the past. In other words, *i* is used in situations which seem to be similar to (57–59) in the previous section, where perfect aspect *ko V ꞌā* is used.

Now there is considerable variation between languages in the extent to which the perfect is used (Comrie 1976: 52–56). Examples such as the ones above suggest that in Rapa Nui the perfect aspect is not used in all cases where a past event has resulted in a current state of affairs. A tentative explanation is, that *ko V ꞌā* is used when the emphasis is on the current state resulting from the event, while *i* is used whenever the emphasis is on the event itself.

In this respect it is telling that the *i-*marked verb is often preceded by a subject (as in (75–76)), while *ko V ꞌā* with event/action verbs<sup>29</sup> either has a subject after the verb or no subject at all; only very rarely is *ko V ꞌā* preceded by a subject. As the default constituent order in Rapa Nui is verb—subject, initial subjects are more prominent than subjects following the verb. If *ko V ꞌā* is more state-oriented while *i* is more event-oriented, it is not unexpected that the agent of an *i*-marked verb tends to be more prominent than the agent of a *ko V ꞌā* marked verb.

#### **7.2.7.5 Summary**

*Ko* (var. *ku*) is always accompanied by postverbal *ꞌā* (var. *ꞌana*). *Ko V ꞌā* marks perfect aspect: it indicates a situation holding at the time of reference, which has come about in some way. A comparison with *i*-marked verbs shows, that *ko V ꞌā* is state-oriented, while *i* is event-oriented.

This is confirmed by the fact that *ko V ꞌā* is used with a wide range of verbs which can be characterised as stative.

### **7.2.8 Aspectuals and constituent order**

There is a correlation between the use of aspectuals and constituent order. As a general rule, when the clause contains a preverbal constituent, the range of aspectuals tends to be limited to perfective *i* and imperfective *e*: *i* is used with past reference; *e* (followed by a postverbal demonstrative (PVD) after the verb) is used when the reference is non-past. The other aspectuals (*he*, *ka* and *ko V ꞌā*) are uncommon.<sup>30</sup>

This tendency is very strong with preverbal non-subjects (1–5 below); in some constructions (such as content questions and the actor-emphatic) it is even an absolute rule. It is less strong with preverbal subjects (6 below).

<sup>29</sup> With stative verbs, *ko V ꞌā* does occur with preposed subjects. Using *i* with these verbs would rule out a stative interpretation. *Ko V ꞌā* also occurs with preverbal subjects after the deictic particle *ꞌī* (§4.5.4.1.1).

<sup>30</sup> See Footnote 40 on p. 403 about other phenomena affecting clauses with preverbal constituents.

#### 7 The verb phrase

	- 2. Noun phrases containing a numeral have a strong tendency to be sentence-initial, regardless their semantic relation to the verb; for example, they may be subject as in (78), or adjunct as in (79). After such a preverbal constituent the verb tends to be marked with *i* or *e*.

'One whole day he worked to pay his punishment.' [R250.026]

	- 4. After question words like *ꞌa ꞌai* 'who' and *he aha* 'what, why':

'Why are you crying?' [Mtx-7-12.024]

<sup>31</sup> Temporal phrases, on the other hand, are commonly followed by *he*:

<sup>(</sup>i) *ꞌI* at *te* art *rua* two *ra ꞌā* day *he* ntr *uꞌi* look *atu* away *te* art *hānau* race *momoko…* slender 'The next day, the 'slender race' saw…' [Ley-3-06.028]

<sup>32</sup> *ꞌo ira* is sometimes followed by *he*, but other clause connectors are not.

7.3 Preverbal particles

	- 6. Preverbal subjects show a certain tendency to be followed by *i* or *e*:

However, preverbal subjects followed by *he* are by no means uncommon. For examples, see (73) and (72) on p. 404.

The only preverbal constituent which does not show a correlation with *i* and *e*, is the negator *ꞌina*; as shown in §10.5.1, the verb in a clause negated with *ꞌina* is usually marked with the neutral aspectual *he*.

### **7.3 Preverbal particles**

### **7.3.1** *Rava* **'given to'**

*Rava*<sup>33</sup> always precedes the verb. It indicates either that the action is performed on a regular basis, or that the subject is inclined to perform the action. *Rava* has a variant *vara*; there is little – if any – difference between the two.

*Rava* may occur in a verb phrase which serves as clause predicate:

(86) *¿ꞌIna* neg *ꞌō* really *te* art *hoko* num.pers *toru* three *era* dist *e* ipfv *rava* given\_to *eꞌa* go\_out *era* dist *ananake?* together 'Don't those three always go out together?' [R366.044]

However, this is not very common: usually *rava* + verb occurs after a noun, in a bare relative clause. In these constructions, *rava* + V indicates an action which is not performed at a certain point in time, but which characterises the preceding noun. The expression has therefore a relatively time-stable character. A few examples:<sup>34</sup>

<sup>33</sup> < PPN *\*lawa* 'sufficient, abundant, completed'; cognates in other languages are used as predicate, not as premodifier. Some languages have a postmodifier < PNP *\*lawa*, which has an intensifying sense 'very, completely'.

<sup>34</sup> In (89), the noun is implied: '(the ones) given to sleeping'.

#### 7 The verb phrase

(87) *He* ntr *tuꞌu* arrive *mai* hither *te* art *pahī* ship *rava* given\_to *maꞌu* carry *mai* hither *i* acc *te* art *meꞌe* thing *mo* for *roto* inside *i* at *te* art *hare* hourse *toa.* store 'The ship arrived which used to bring things for the store.' [R250.094]

(88) *Te* art *meꞌe* thing *rava* given\_to *aŋa* do *o* of *tātou* 1pl.incl *ꞌi* at *rā* dist *mahana* day *he* pred *porotē.* parade


### **7.3.2 Degree modifiers**

*ꞌApa* and *ꞌata* are degree modifiers, which precede the verb root.

**7.3.2.1** *ꞌApa ꞌApa* (which is also a noun meaning 'part, portion, piece') indicates a moderate degree: 'somewhat, kind of'.<sup>35</sup> It is often used with stative predicates, but found with actions as well.


In (92), *ꞌapa* semantically quantifies the object: 'we somewhat obtained X' = 'we obtained a few X' (cf. (96–97) for a similar use of *ꞌata*).

(92) *Ko* prf *ꞌapa* part *rovaꞌa* obtain *mai* hither *ꞌā* cont *te* art *meꞌe* thing *pāreherehe* piece:red *matā.* obsidian 'We obtained a few pieces of obsidian.' [R629.030]

<sup>35</sup> *ꞌApa* may be borrowed from Tahitian *ꞌapa* 'half of a fish or animal, cut lengthwise' (Pa'umotu *kapa*). Fischer (2001a: 315) suggests it was borrowed from Tahitian *ꞌafa* 'half' (which was itself borrowed from English).

7.3 Preverbal particles

**7.3.2.2** *ꞌAta ꞌAta* indicates a high degree, either comparative ('more'), superlative ('most') or absolute ('very; thoroughly').<sup>36</sup> It is used in comparative constructions with adjectives (§3.5.1.1); with event verbs it is also used in a comparative sense, comparing the intensity of the event to a previous situation: 'more than before'.


The comparison may also be with respect to another entity as standard of comparison, though this rarely happens. Below is an example; as with adjectives, the standard of comparison is expressed with *ki*:

(95) *¡E* voc *Māria,* Mary *ꞌata* more *haꞌamaitai* bless *koe* 2sg *e* ag *te* art *ꞌAtua* God *ki* to *te* art *taꞌatoꞌa* all *ŋā* pl *viꞌe.* woman 'Mary, you are more blessed by God than all women.' [Luke 1:42]

One could wonder if the verb has been adjectivised here; notice however that the Agent phrase *e te ꞌAtua* is introduced by *e*, which suggests that *haꞌamaitai* retains its status of an agentive verb.

With transitive verbs, *ꞌata* may indicate a multiplication of the object. For example, in (96) *ꞌata* semantically modifies *tāropa*: 'more baskets'.<sup>37</sup>


*ꞌAta* may also be used in a superlative sense:

(98) *Te* art *artículo* article *ꞌata* more *pāpaꞌi* write *o* of *tātou* 1pl.incl *he* pred *'he' he ꞌe* and *he* pred *'te'. te* 'The articles we write most, are *he* and *te*.' [R616.719]

<sup>36</sup> Cognates occur in several EP languages. These are preverbal as in Rapa Nui, but only have an absolute sense: 'carefully, slowly' (Pollex, see Greenhill & Clark 2011; Elbert & Pukui 1979: 74 for Hawaiian *aka*, Bauer, Parker & Evans 1993: 92 for Māori *aata*). Possibly it occurs in SO languages as well: Besnier (2000: 188) mentions a preverbal particle *ata* 'properly, in moderation' in Tuvaluan, though only one example is provided, where it is part of an idiom.

<sup>37</sup> Cf. the use of *tahi* and *rahi* in the verb phrase, sec. §4.4.9 and §4.4.7.2.

#### 7 The verb phrase

**7.3.2.3 Placement** The exact position of *ꞌata* and *ꞌapa* in relation to other preverbal elements varies, depending on their respective scope.

With causative verbs, the degree modifier usually occurs before the causative prefix *haka*, as in (99): 'more [cause to be strong]'. However, it may also occur after *haka*, in which case *haka* has scope over the degree modifier. This is illustrated in (100): 'cause to be [more intelligent]'.


'…Chileans who want to pass themself off as smarter (lit. to cause them to be smarter).' [R428.006]

With the constituent negator *taꞌe*, either the negator or the degree particle may come first. In (101) the negator comes first and has scope over *ꞌata*: 'not [more high]'. In (102), *ꞌapa* has scope over the negation: 'somewhat [not listening]'.


### **7.4 Evaluative markers**

The evaluative markers *nō* and *rō* occur in the same position in the verb phrase; they are mutually exclusive.

### **7.4.1 The limitative marker** *nō*

*Nō* originates in PPN*\*noa*, which occurs as a postverbal marker in a number of languages throughout Polynesia. Rapa Nui is the only language in which the vowel sequence *oa* assimilated to *ō*, apart from Hawaiian (Elbert & Pukui 1979: 100).

7.4 Evaluative markers

*Nō* is a limitative marker; its basic sense is 'nothing else'.<sup>38</sup> The particle has several uses, which can all be related to this basic sense: 'simply, just' (nothing more), 'still' (a lack of change), 'even so, yet' (something happens, despite expectations to the contrary).

In this section, the use of *nō* in the verb phrase is discussed. *Nō* also occurs after other parts of speech, which are discussed elsewhere: nouns (§5.8.2), numerals (§4.3.2.4) and quantifiers (§4.4.10).

*Nō* may indicate that something just happens, without anything more. The implication is that something else or something more could happen, but does not actually happen. The context tells what this 'something else' would be:


*Nō* in this sense 'just' may have the connotation 'without further ado, without thinking, without taking other considerations into account'.

(105) *¿Kai* neg.pfv *haꞌamā* ashamed *koe* 2sg *i* pfv *toꞌo* take *nō* just *koe* 2sg *i* acc *te* art *mauku* grass *mo* for *taꞌo* cook *i* acc *taꞌa* poss.2sg.a *ꞌumu?* earth\_oven 'Weren't you ashamed, that you just took the grass to (as fuel) to cook your

earth oven (without asking, even though the grass was mine)?' [R231.186]

(106) *Te meꞌe nō, ku oho nō ꞌā ki tai hī.*

art thing just prf go just cont to sea to\_fish

'(Nowadays people don't consider the moon and the wind.) On the contrary, they just go out to sea to fish.' [R354.026]

In the previous examples, a contrast is implied between what happens and what could have happened. Sometimes this sense of contrast is more prominent; the clause has a connotation of counterexpectation: 'even so, no matter, still'.<sup>39</sup>

(107) *…e* ipfv *māuiui* sick *nō* just *ꞌana* cont *te* art *ŋā* pl *poki.* child

> '(Nowadays there are all kinds of things to take care of children,) but even so, children get sick.' [R380.138]

<sup>38</sup> Cf. Lazard & Peltzer (2000: 146) on Tahitian *noa*: their basic gloss is 'ne faire que', from which they derive the different uses of *noa*, which largely parallel those of Rapa Nui *nō*.

<sup>39</sup> A contrastive sense of *nō* is also found in expressions like *Te me ꞌe nō* 'however' (see (153) on p. 268) and in the conjunction *nōatu* (§11.6.7).

#### 7 The verb phrase

(108) *Ka* cntg *rahi* many *atu* away *tāꞌaku* poss.1sg.a *poki,* child *e* ipfv *hāpaꞌo* care\_for *nō* just *e* ag *au* 1sg *ꞌā.* ident 'Even if I have many children, I will still take care of them myself.' [R229.023]

*Nō* may be used in a continuous clause, emphasising that the action is still going on. In this sense, it is often used with the imperfective *e*.


'In the morning he went to his work, when Eva was still sleeping.' [R210.025]

An action marked with *nō* is often unremarkable, routine, expected: something is simply going on, nothing significant has happened (yet). Often, the verb phrase expresses a lack of change with respect to a previous situation: the same thing described earlier is still going on. In this sense, *nō* is common in progressive cohesive clauses (see (214) on p. 558):

(111) *E* ipfv *iri* ascend *nō* just *ꞌā* cont *he* ntr *takeꞌa* see *e* ag *Te* Te *Manu* Manu *e* num *tahi* one *hōŋaꞌa* nest *māmari.* eggs '(The two went up and looked for eggs…) While they were still going up, Manu saw a nest with eggs.' [R245.202–203]

As discussed in §5.8.2, *nō* in the noun phrase often serves to limit the reference of a noun phrase. Occasionally, *nō* in the verb phrase has the same effect.<sup>40</sup> In (112), *nō* occurs after the (nominalised) verb *kai*, signalling that the object noun phrase has limited reference.

(112) *Ko* prf *haꞌumani* bored *ꞌana* cont *ꞌi* at *te* art *kai* eat *iŋa* nmlz *nō* just *i* acc *te* art *moa.* chicken 'I'm tired of eating only chicken.' [R229.123]

After certain adjectival predicates, *nō* signals that the object described has only the property in question, implicitly excluding other properties: 'just, altogether'. So while being fundamentally limitative in nature, *nō* in these cases underlines and emphasises the property expressed by the adjective: the object is entirely characterised by this property, to the exclusion of anything else. This use is only found with adjectives expressing

<sup>40</sup> Cf. the use of *tahi* 'all', which occurs in the verb phrase but determines the reference of a noun phrase in the clause (§4.4.9).

7.4 Evaluative markers

a positive evaluation, like *rivariva* 'good', *nene* 'sweet, delicious', *tau* 'pretty'. The adjective is preceded by the aspectual *he*.


Notice that English 'just' can be used in the same way, as the translation of (114) shows.

### **7.4.2 The asseverative marker** *rō*

*Rō* is an asseverative particle. It serves to underline the reality of the event and/or its significance in the course of events. (See also Weber 2003b: 41.) While *nō* underlines the expected, routine nature of the event (for example, because the situation has not changed), *rō* underlines its significance, newsworthiness. In pragmatic terms: while *nō* indicates a low information load, *rō* indicates a high information load. In view of the diversity of its uses, *rō* is glossed emph(atic).<sup>41</sup>

Like *nō*, *rō* is the result of vowel assimilation: it is derived from PNP *\*loa*. Unlike *nō*, *rō* is not used in the noun phrase, but it does occur occasionally in numeral phrases (see (29) on p. 153).

In the verb phrase, *rō* is used in certain well-defined contexts, which are discussed elsewhere in this grammar:


<sup>41</sup> Du Feu (1996: 37) characterises *rō* as a realis particle, glossed as [+REA]; contrasted with *rā* [-REA]. She points out that *rā* is for example used in imperatives, when the speaker has no control over the outcome; *rō*, on the other hand, is for example used in 1st person imperatives (= hortatives) where speaker has greater control over the realisation of the event. While this is only part of the picture, and while *rō* is actually not in a paradigmatic relation with the intensifier *rā* (§4.5.4.4), this correctly underlines the asseverative character of *rō*.

#### 7 The verb phrase

But*rō* is not limited to these constructions. Generally speaking, *rō* marks events which are significant in discourse, for example because they are the culmination of a series of events, or because they change the course of events. This happens in the *he V rō ꞌai* construction mentioned above; it is also found with *i V ai* (§7.6.5). In the following example, Kainga produces a spear point which will play an important role in the events to follow.

(115) *ꞌI* at *rā* dist *pō* night *ꞌā* ident *a* prop *Kaiŋa…* Kainga *i* pfv *aŋa* make *rō* emph *ai* pvp *e* num *tahi* one *matā* obsidian *rivariva.* good:red

'In that night Kainga made a good obsidian spearpoint.' [R304.015]

Events may also be significant by way of contrast:

(116) *A* prop *Pea* Pea *e* ipfv *ko* neg.ipfv *rivariva* good:red *mo* for *hāpaꞌo* care\_for *i* acc *a* prop *Puakiva;* Puakiva *e* ipfv *oho* go *rō* emph *ꞌā* cont *ki* to *te* art *aŋa.* work 'Pea was not able to take care of Puakiva; (rather,) he used to go to work.' [R229.005]

*Rō* may emphasise the reality of a situation: 'really'.<sup>42</sup>


When *rō* emphasises the reality of the clause, there may be a connotation of counterexpectation. In (119) this happens in a question, in (120) as reply to a question.


<sup>42</sup> Examples such as (117) might suggest that *rō* means 'very'. However, *rō* (unlike Tahitian *roa*) is not a common way to express a high degree; rather, this is expressed using *hope ꞌa* 'last' or *riꞌariꞌa* 'terribly'.

7.5 Directionals

In several of the examples above, *rō* occurs in the common construction *e V rō ꞌā*, which expresses an ongoing event or situation (§7.2.5.4). In this construction, the asseverative sense of *rō* is not always clear. Sometimes the clause does convey new, unexpected or even surprising information, as in the following example, where the subject does a somewhat unexpected discovery:

(121) *I* pfv *ꞌara* wake\_up *mai* hither *ai,* pvp *ꞌi* at *rote* inside\_art *piha* room *e* num *tahi* one *e* ipfv *moe* lie *rō* emph *ꞌā…* cont 'When she woke up, she was lying in a room…' [R210.090]

But in other cases the information load of the *e V rō ꞌā* clause does not seem to be particularly high:

(122) *Mo* if *uꞌi* look *atu* away *o* of *te* art *ŋā* pl *poki* child *ki* to *a* prop *Taparahi* Taparahi *e* ipfv *haꞌere* walk *rō* emph *ꞌā* cont *a* by *te* art *ara* road *he* ntr *riꞌariꞌa.* afraid 'When the children saw Taparahi walking by the road, they were afraid.' [R250.190]

Possibly, the sense of *rō* in this construction is weakened, and *e V rō ꞌā* has been developing into a fossilised construction expressing the ongoing duration of a situation.

### **7.4.3 Conclusion**

To give a general characterisation of *rō* and *nō*, one could say that they indicate the cognitive status of the information given in the clause: *nō* indicates that the clause expresses something unchanged, which is often expected or even routine; *rō* indicates that the clause expresses something new and unexpected, which may even be surprising. *Rō* is reminiscent of a "mirative" marker (Payne 1997: 255), though it may not be as strong as elements that have been identified as miratives in other languages.

Even though *rō* and *nō* are in a way opposites, both may involve counterexpectation. That *rō* would express counterexpectation is no surprise, but *nō* may involve a hint of counterexpectation as well: a situation continues to be true or an expected event still happens, despite factors to the contrary. When *nō* involves counterexpectation, this is because of an (unexpected) continuity; *rō* expresses counterexpectation because of a discontinuity.

### **7.5 Directionals**

The directionals *mai* and *atu* indicate direction with respect to a certain deictic centre or locus:

#### 7 The verb phrase


*Mai* and *atu* are the only reflexes in Rapa Nui of a somewhat larger system of directionals in Proto-Polynesian.<sup>43</sup>

The movement indicated by directionals may be of different kinds. Three common types are:


The last type of movement is a metaphorical extension of the idea of movement. Other metaphorical extensions are possible, as will be shown below.

In §7.5.1, the main uses of directionals are discussed, mainly based on three narrative texts (all of which include a considerable amount of direct speech). In §7.5.2, statistics are presented for the use of directionals with certain categories of verbs in the text corpus as a whole. Finally, §7.5.3 raises the question which factors prompt the use of a directional.

### **7.5.1 Use of directionals**

#### **7.5.1.1 In direct speech**

As indicated in the previous section, directionals signal movement with respect to a deictic centre. In direct speech, the deictic centre is usually the speaker. This means that in a conversation, *mai* usually indicates a movement towards the speaker as in (123) below, while *atu* indicates a movement away from the speaker. The latter movement may be towards the addressee as in (124), or away from the speaker in another direction as in (125).

(123) *¿Ko* prom *ai* who *koe* 2sg *e* ipfv *eke* go\_up *mai* hither *ena?* med

'Who are you (who are) coming up?' [R304.084]

(124) *ꞌĪ* imm *au* 1sg *he* ntr *oho* go *atu.* away 'I am coming (towards you) right now.' [R152.010]

<sup>43</sup> Clark (1976: 34) identifies five directionals in PPN: *\*mai* 'toward speaker', *\*atu* 'away from speaker', *\*hake* 'upward', *\*hifo* 'downward', *\*ange* 'along, obliquely'. Most languages preserved at least three of these, Rapa Nui only two. *\*hifo* was retained as *iho*; however, this developed into an adverb meaning 'just then' (§4.5.3.1). Ultimately, *mai* and *atu* stem from a set of directional verbs in POc, which were used as the final verb in a serial verb construction (Ross 2004: 194).

7.5 Directionals

(125) *He* ntr *eꞌa* go\_out *atu* away *a* prop *au* 1sg *ꞌi* at *te* art *hora* time *nei.* now 'I am going away now.' [R245.017]

In the next examples, it is the direct object which moves: towards the speaker in (126), away from the speaker towards the addressee in (127).


In the following examples, there is no physical movement of a participant or object, but a flow of information from the speaker to the addressee. In (128) the speaker is the subject, so the information moves away from the speaker; hence the use of *atu*. In (129) the speaker is addressed by the subject of the clause, so the flow of information is directed towards the speaker; hence the use of *mai*.


The movement may also be more implicit. The following example is spoken by fishermen, who tell what often happens to them: a tuna will come up towards them (i.e. towards the speaker, *mai*), but then it will cut the line. The last verb 'to cut' is not a motion verb, yet the verb is followed by *atu*: the action implies that the tuna swims away from the fishermen, i.e. away from the deictic centre.

(130) *…te* art *kahi* tuna *era* dist *ꞌi* at *raro* below *ꞌā,* ident *e* ipfv *iri* ascend *mai* hither *era,* dist *he* ntr *motu* cut *rō* emph *atu* away *ꞌai* subs *te* art *kahi.* tuna '…the tuna deep below, which when it comes up, the tuna cuts (the line).' [R368.024]

As directionals indicate the direction of movement of a participant, object, or information, a directional may be sufficient to indicate the recipient, addressee or goal of an

#### 7 The verb phrase

event: *kī mai* 'say toward' indicates that something was said to me (or us). Therefore, the recipient, addressee or goal does not need to be stated separately.<sup>44</sup> This is the case in (129) above and in (131) below:

(131) *—¡A* prop *Te* Te *Manu* Manu *hoꞌi* indeed *i* pfv *kī* say *mai* hither *mo* for *turu* go\_down *o* of *māua!* 1du.excl *—¡ꞌĒē,* yes *a* prop *Te* Te *Manu* Manu *hoꞌi* indeed *i* pfv *kī* say *atu!* away '—Te Manu said (to me) that we (=he and I) should go down! —OK, so Te Manu told (you)!' [R245.221]

When the subject is also left implicit, the directional *mai* or *atu* may be the only clue for participant reference. In (132), the subject is not expressed. *Atu* indicates that the request was directed from the speaker ('we') to Meherio, while the help requested went from Meherio to the speaker.<sup>45</sup>

(132) *ꞌĀmui* moreover *i* at *a* prop *Meherio* Meherio *e* ipfv *kī* say *atu* away *era…* dist *mo* for *hāꞌūꞌū* help *mai,* hither *ꞌina* neg *he* ntr *hāꞌūꞌū* help *rō* emph *mai.* hither 'Moreover, when (we) told Meherio to help (us), (she) didn't help (us).' [R315.031]

In the examples so far, *atu* indicates either a movement from speaker to addressee, or a movement away from speaker and hearer in an unspecified direction as in (125).<sup>46</sup> However, *atu* does not always imply a movement away from the speaker: it may indicate a movement from another place or participant towards the addressee. The following examples illustrate this.


'When Hoto Vari comes again (to you), and sticks out his tongue (to you)…' [R304.020]

<sup>44</sup> For this reason, Wittenstein (1978: 4) calls *mai* and *atu* in Rapa Nui "directional pronouns".

<sup>45</sup> The start of the sentence is syntactically unusual. The syntax of *ꞌāmui* (a borrowing from Tahitian, where it is a verb 'to get together, be united') is very flexible in Rapa Nui. In this case, a Tahitian construction seems to be used, in which *i N VP* ('to/at me said') may function as a temporal clause ('when I said'); this construction is not attested otherwise in Rapa Nui.

<sup>46</sup> As mentioned above, according to Clark (1976: 34), *\*atu* in PPN means 'away from the speaker'.

#### 7.5 Directionals

An example like (134) is striking, because it is not at all clear how the location of origin can be considered the deictic centre: the place where Hoto Vari comes from, is not relevant at all in the story; it is not even mentioned. In other words, the use of *atu* seems to be motivated entirely by its destination (the second person), not by a deictic centre. This may thus be an exception to the rule (formulated e.g. by Hooper 2002: 285 for Tokelauan) that the use of directionals always implies the existence of a deictic centre.

These example also show that the sense of *atu* cannot be captured in a single definition: *atu* does not always express movement away from the speaker (see (133–134)), but neither does it always express movement towards the addressee (see (125)). Either one is a sufficient criterion for using *atu*; neither is a necessary criterion.

#### **7.5.1.2 In third-person contexts**

The previous section discussed contexts where a speaker and/or addressee is involved and where movement takes place with respect to the speaker or addressee. As we saw, in these cases *mai* indicates movement towards the speaker, while *atu* indicates movement away from the speaker and/or towards the addressee.

Directionals are also used in third-person contexts, where no speaker or addressee is involved. In such cases, movement does not take place from the perspective of the speaker; rather, the deictic centre is a participant or location in the text. The speaker positions himself (and the hearers) at a certain location or near a certain participant, and events are regarded from the point of view of that location or participant.

There are no fixed rules for determining the deictic centre: it is to a certain extent up to the narrator to choose the perspective from which the text world is regarded.<sup>47</sup> The deictic centre may be fairly constant throughout the story, or it may shift with each scene or even from sentence to sentence. Speakers may have a preference to identify the deictic centre with one central participant, or to vary the point of view. Speakers may also show a preference for *mai* or *atu* with certain verbs or classes of verbs, regardless the context and the subject of the verb.

In other words, there are no hard and fast rules for the use of directionals. However, certain clear tendencies can be observed. In this section some of these tendencies are discussed from individual stories, while §7.5.2 gives statistical data from the whole text corpus. These statistics reveal a number of general tendencies and also show a number of diachronic shifts in the use of directionals.

**7.5.1.2.1 Example 1: a stable deictic centre** The story *Nuahine Rima Roa*, 'The old lady with the long arms' (R368), tells about an old woman with enormous arms, who terrorises the village by stealing food, but who is eventually tricked into defeat by a group of fishermen. In this story, there is one central participant, the old lady; the other participants are hardly mentioned as individuals (they mostly act as a group), let alone mentioned by name. It is not surprising, therefore, that the deictic centre in most of the story is the old

<sup>47</sup> Tchekhoff (1990: 105) likewise stresses optionality and subjectivity for the use of directional particles in Tongan.

#### 7 The verb phrase

lady. Events are regarded from the perspective of wherever the old lady is. Numerous examples could be given, such as the following:


In direct speeches in the story the situation is different: here the deictic centre is the speaker, whether this is the old lady or another participant. But even outside direct speech, not all directionals in this story presuppose the old lady as deictic centre. In the following example, the men come out of the house of the old lady, i.e. they move away from her; yet *mai* is used:

(137) *I* pfv *eꞌa* go\_out *hakaꞌou* again *mai* hither *era* dist *tū* dem *ŋāŋata* men *era* dist *mai* from *te* art *hare* house *era* dist *o* of *tū* dem *rūꞌau* old\_woman *era…* dist

'When those men came out again of the house of that old woman…' [R368.056]

This apparent exception to the rule may have to do with a general tendency of the verb *eꞌa* 'go out' to be followed by *mai* rather than *atu*. As discussed in §7.5.2 below, *eꞌa* commonly takes *mai* while it rarely takes *atu*; a similar tendency is discernible for other motion verbs. This means that the directional after *eꞌa* tends to point to the destination, the place where the subject is going to, as the centre of attention. In this way it provides the reader/hearer with a subtle signal that this location is significant as the location where the next events are going to happen. Notice that in (137) above, *eꞌa mai* occurs in a cohesive clause, which provides a bridge between the previous scene (in the house) and the next one (in the village). *Mai* contributes to paving the way for the change of location and the next scene.

Such examples show that even in a narrative with one protagonist around whom the action revolves, the narrator may use directionals as a device to focus the hearer's attention on locations relevant in the development of the story.

**7.5.1.2.2 Example 2: a shifting deictic centre** The story *He via o te Tūpāhotu* 'The life of the Tupahotu' (R304) tells about wars between two major tribes on the island, the Miru and the Tupahotu. There are various protagonists: Huri a Vai and his father Kainga of the Miru tribe, Hoto Vari and his father Poio of the Tupahotu tribe. These protagonists,

7.5 Directionals

as well as a few other characters, alternate in prominence in different parts of the story, and the deictic centre shifts accordingly.

In the first part of the story, the focus is on Huri a Vai. Not only is he mentioned more than other characters, the directionals point towards him as the deictic centre:

(138) *He* ntr *takeꞌa* see *i* acc *a* prop *Hoto* Hoto *Vari* Vari *ka* cntg *pū* approach *mai.* hither '(Huri a Vai) saw Hoto Vari coming towards him.' [R304.004]

Now what if the movement concerns the protagonist himself, i.e. when Huri a Vai himself moves to a different location? Levinsohn (2007: 142–143) points out that in such cases, languages tend to use one of two strategies: the deictic centre is either a fixed geographical location or it is the next location, i.e. the destination of the movement. As it turns out, in Rapa Nui narrative the second strategy is predominant: when the protagonist moves, *mai* is used to point to the location where the next events are going to happen.<sup>48</sup> The following example illustrates this:

(139) *I* pfv *ahiahi* afternoon *era* dist *he* ntr *hoki* return *mai* hither *a* prop *Huri* Huri *ꞌa* a *Vai* Vai *ki* to *te* art *kona* place *hare* house *era.* dist 'In the afternoon, Huri a Vai returned home.' [R304.009]

This corresponds to a general tendency in Rapa Nui: motion verbs are much more commonly followed by *mai* than by *atu*, as shown in §7.5.2 below.

In the remainder of the story, the deictic centre shifts between various participants and locations. Sometimes one of the major participants is the deictic centre for a while; in the following example, four consecutive verbs are all followed by a directional pointing towards Kainga, one of the protagonists, as deictic centre:

(140) *ꞌĪ* imm *ka* cntg *uꞌi* look *atu* away *ena* med *ko* prom *te* art *ꞌata* shadow *o* of *te* art *taŋata* man *ka* cntg *kohu* shade *mai* hither *ꞌi* at *muꞌa* front *i* at *a* prop *ia.* 3sg *I* pfv *hāhine* near *mai* hither *era* dist *ki* to *muri* near *i* at *a* prop *ia,* 3sg *he* ntr *ꞌui* ask *atu…* away 'Then (Kainga) saw the shadow of a man falling in front of him. When (that man) was close to him, (Kainga) asked…' [R304.095–096]

The deictic centre may also be a minor participant, provided this participant is significant in the scene in question. See (141) in the next section for an example.

#### **7.5.1.3 Directionals with speech verbs**

Directionals in Rapa Nui are commonly used in clauses introducing direct speech ('he said: …'). In such clauses, various strategies are possible:

<sup>48</sup> In the preceding section, the same tendency was observed in the story *Nuahine Rima Roa*; see example (137) and discussion.

#### 7 The verb phrase


All these strategies are used to various degrees in Rapa Nui discourse. Strategy 1 is dominant overall: as the statistics in the next section will show, about 70% of all speech verbs in the corpus do not have a directional. In this section, the other strategies are illustrated from a couple of texts.

In the story *He via* (R304), discussed in the previous section, a mix of strategies is used. In the following short conversation, the directionals all point towards Oho Takatore as deictic centre (strategy 4). Oho Takatore is not a central participant in the story as a whole, but his presence is crucial at this point.

(141) *He* ntr *raŋi* call *atu* away *ia* then *e* ag *ꞌOho* Oho *Takatore…* Takatore *Terā* then *ia* then *ka* cntg *pāhono* answer *mai* hither *e* ag *Poio…* Poio *I* pfv *raŋi* call *mai* hither *era* dist *e* ag *Poio* Poio *pē* like *ira…* ana 'Oho Takatore shouted… Then Poio answered… When Poio had called out like that…' [R304.058-063]

In the following conversation, the deictic centre shifts halfway: in the first two clauses, Kainga is the deictic centre, but then it shifts to Vaha (strategies 2+4).

(142) *He* ntr *ꞌui* ask *atu…* away *Terā,* then *ka* cntg *pāhono* answer *mai* hither *e* ag *Vaha…* Vaha *He* ntr *pāhono* answer *mai* hither *ia* then *e* ag *Kāiŋa…* Kainga '(Kainga) asked… Vaha replied… Kainga then replied…' [R304.096]

These examples show that the speaker has the choice from a variety of strategies. In another story, *Rāꞌau o te rūꞌau ko Mitimiti* 'Medicine of the old woman Mitimiti' (R313), the narrator has a general preference for *atu*, both with speech verbs and other verbs (though not without exceptions). The general pattern in this story is for the first turn in a conversation to be unmarked or marked with *mai*, whereas the following turns are marked with *atu* (strategy 3).

(143) *He* ntr *kī* say *o* of *koro…* Dad *ꞌE* and *he* ntr *kī* say *takoꞌa* also *atu* away *te* art *reꞌo* voice *o* of *nua…* Mum *He* ntr *pāhono* answer *atu* away *ia* then *te* art *reꞌo* voice *o* of *tū* dem *ŋā* pl *repa* young\_man *era…* dist *He* ntr *kī* say *hakaꞌou* again *atu* away *ia* then

7.5 Directionals

*te* art *reꞌo* voice *o* of *koro…* Dad 'Dad said… And Mum's voice also said… The voice of those youngsters replied… Dad's voice said again…' [R313.009–015]

In this text, then, *atu* functions as a sort of continuance marker, marking the next step in a series of speech turns.<sup>49</sup>

#### **7.5.1.4 Lack of movement: more metaphorical uses**

So far, various uses of directionals have been discussed in which some kind of physical or metaphorical movement takes place: movement of a participant or object, or a flow of speech. This section deals with the use of directionals in cases where no movement seems to be involved. In these cases the use of directionals is extended even further than the metaphorical senses discussed so far. Various metaphorical extensions are possible, depending on the verb involved and subject to speaker preference. The examples discussed here do not cover all possibilities, but serve to illustrate the wide range of metaphorical uses of directionals.

Directionals may occur with verbs that do not indicate any movement, nor a transitive action, but rather the absence of movement. In the story *He via* (R304), a directional is used twice with the verb *piko* 'to hide (intr.)', once *mai* and once *atu*:


'(Vaha) came out by that road where Kainga was hiding.' [R304.094]

There is a clear difference between these two examples: in (144), Huri a Vai is hiding from his enemies, he is lying low to avoid being detected. In (145), Kainga is not just hiding away; he is lying in ambush, waiting for Vaha to come by. In other words, the hiding in (144) is self-directed, oriented inwards, while the hiding in (145) is outwardlooking, with the attention away from the person hiding. It is no coincidence that in the first case *mai* is used, indicating orientation towards the subject as deictic centre, while in the second case *atu* is used, pointing away from the subject. In other texts as well, *piko mai* is commonly used when people hide from others, while *piko atu* is used of people lying in ambush, spying on someone else (cf. Fuller 1980: 12). While there is

<sup>49</sup> This use may be influenced by Tahitian, where *atu ra* and *mai ra* are extremely common to mark the next event in a narrative (see e.g. Lazard & Peltzer 2000: 134).

#### 7 The verb phrase

no movement involved, the directionals indicate orientation with respect to the deictic centre.<sup>50</sup>

Directionals are also found with the verb *noho* 'to sit, stay', a verb which would seem to epitomise lack of movement. At least some of these occurrences can be explained as indicating orientation: with *mai* the participant faces inward, is self-oriented; with *atu* the focus is outward. The following two examples illustrate the difference:


'The spirits, when you sit down and look at them, will look away.' [R310.082]

Orientation may also have to do with physical distance from the deictic centre, so that *mai* is similar to 'here' and *atu* is similar to 'away':


#### **7.5.1.5** *Atu* **indicating extent**

Another metaphorical use concerns only *atu*. With stative verbs, *atu* may indicate the extent of a state or characteristic:

(150) *Meꞌe* thing *pararaha* fat *atu* away *te* art *oru* pig *era.* dist 'That pig is very fat.' [Notes]

<sup>50</sup> Hooper (2004: 1751) discusses a similar function of directionals in Tokelauan. She points out (with reference to Jackendoff 1983) that a path (i.e. a directional movement) may play various roles: an object may traverse a path, but it may also be oriented along a path, facing an entity (in this case, the deictic centre).

7.5 Directionals

(151) *E* ipfv *huri* turn *rō* emph *ꞌā* cont *te* art *ꞌāriŋa* face *o* of *Heru* Heru *a* by *ruŋa* above *ꞌe* and *ko* prf *tetea* pl:white *atu* away *ꞌā* cont *te* art *mata.* eye

'Heru's face was turned upwards and his eyes were very white.' [R313.043]

These examples suggest that in some cases *atu* indicates a (light) degree of emphasis. It is not difficult to see how this use could arise: the basic meaning 'away from a deictic centre' may naturally develop into 'away from a point of reference, beyond what is common or expected'.<sup>51</sup>

The sense of an extent is also seen when *atu* is used after the quantifiers *tētahi* 'some/ others', *meꞌe rahi* 'many' and *rauhuru* 'diverse'; in these cases, *atu* emphasises the extent of a quantity (see discussion and examples in §4.4.10.)

Finally, the sense of extent may also explain why *atu* is common – at least for some speakers – in the construction *he V rō ꞌai*, which indicates final or climactic events (§7.2.3.3).<sup>52</sup> *atu* simply makes the construction a little heavier, thereby adding to its prominence.

(152) *He* ntr *mate* die *rō* emph *atu* away *ꞌai* subs *tū* dem *rūꞌau* old\_woman *era* dist *ꞌi* at *te* art *taŋi.* cry 'The old lady burst out in tears (lit. died with crying).' [R313.104]

### **7.5.2 Directionals with motion, speech, and perception verbs**

In the previous sections, the use of directionals was analysed by looking at individual occurrences. One conclusion that could be drawn is that, while the basic meaning of the directionals is clear, the speaker has a certain freedom, both in choosing the deictic centre and in applying directionals in extended uses.

Another method to analyse the use of directionals, is to count the overall use of directionals with different (classes of) verbs in the text corpus. As it turns out, statistical data shed additional light on the use of directionals, revealing a number of general tendencies. These tendencies cannot be discovered by analysing individual texts, but only come to the surface when large numbers of occurrences (and non-occurrences) are taken into account.

This section discusses the use of directionals with three classes of verbs that commonly take directionals: motion verbs, speech verbs and perception verbs. Data are based on the whole corpus of old and new texts.

<sup>51</sup> Hooper (2002: 291) signals a somewhat similar extension of the meaning of *atu* in Tokelauan, where it may signal a point in time beyond the time of reference (e.g. *ananafi* 'yesterday', *ananafi atu* 'the day before yesterday' – the same expression is found in Tahitian).

<sup>52</sup> In the corpus as a whole, *rō atu ꞌai* occurs 186 times, *rō mai ꞌai* only 60 times. Note that directionals are by no means obligatory in this construction: *rō ꞌai* without directional occurs 321 times.

#### 7 The verb phrase

**7.5.2.1 Motion verbs** One class of verbs which often takes a directional, is the class of motion verbs. Table 7.6 gives statistics for the use and non-use of directionals with a number of common motion verbs.<sup>53</sup> Separate figures are given for old texts and newer texts. The most common directional in each corpus is in bold.

Table 7.6: Directionals with motion verbs


As these figures show, *mai* is much more common than *atu* with these verbs. In other words, when the direction of movement is indicated, in most cases the subject moves towards the deictic centre. Put differently, directionals after motion verbs tend to point toward the location where the subject is heading as the deictic centre, subtly signalling that this new location is – or becomes – significant to the action.

This tendency is strong in both older and newer texts. It is strongest for *eꞌa* 'go out', which is followed by *mai* in a total of 415 cases and by *atu* in only 32 cases. One could wonder if *eꞌa mai* is to a degree lexicalised, though we have to keep in mind that in two thirds of all occurrences, *eꞌa* does not have a directional at all.

**7.5.2.2 Speech verbs** As indicated in the previous section, directionals are also commonly used with verbs of speaking. These verbs imply a flow of information from the speaker<sup>54</sup> to the addressee. The use of directionals with speech verbs points to one of the participants as the deictic centre of the speech act: *mai* indicates a movement towards the addressee as deictic centre, *atu* indicates a movement away from the speaker as the deictic centre.

Total occurrences for a number of common speech verbs<sup>55</sup> are given in Table 7.7. This table shows a clear shift over time in the use of directionals. Whereas in old texts *mai* is by far the most common directional and *atu* is rare, in new texts *atu* has become more frequent (though *mai* is by no means uncommon). In other words, in older stories the speaking act is usually considered from the perspective of the addressee, whereas in newer stories it is more commonly seen from the perspective of the speaker (i.e. the subject).

<sup>53</sup> This table includes counts for *oho* 'go', *tuꞌu* 'arrive', *e ꞌa* 'go out', *turu* 'go down', *iri* 'go up', *uru* 'go in', *tomo* 'go ashore'.

<sup>54</sup> "Speaker" is here taken in the sense of "the person uttering the speech referred to by the speech verb", not the speaker/narrator of the text as a whole.

<sup>55</sup> *Kī* 'say', *raŋi* 'call', *pāhono* 'answer' (only in modern Rapa Nui), *ꞌui* 'ask'.

#### 7.5 Directionals


Table 7.7: Directionals with speech verbs

While the relative proportion of *mai* and *atu* has changed, the table also shows that the total use of directionals has not changed much: in both corpora, roughly 30% of the speech verbs under consideration are accompanied by a directional.

**7.5.2.3 Perception verbs** The most common verbs of perception in Rapa Nui are the following: for visual perception, *takeꞌa* or *tikeꞌa* <sup>56</sup> 'to see' and *uꞌi* 'to look, to watch'; for aural perception *ŋaroꞌa* 'to hear, to perceive' and *hakaroŋo* 'to listen'. All these verbs have two arguments: an experiencer (the perceiving entity) and a stimulus (the perceived entity). The experiencer is expressed as subject, the stimulus as direct object or as an oblique marked with *ki*.

The first verb of each pair (*takeꞌa* 'see' and *ŋaroꞌa* 'hear') indicates uncontrolled perception, i.e. the registration of a perceptual stimulus by one of the senses. The other two verbs (*uꞌi* 'look' and *hakaroŋo* 'listen') express controlled perception, i.e. focused attention on the part of the subject (cf. Dixon 2010b: 144). In other words, whereas the subject of *takeꞌa* and *ŋaroꞌa* is merely registrating a visual stimulus, the subject of *uꞌi* and *hakaroŋo* is actively involved in the act of perception.<sup>57</sup>

This difference can be correlated to the direction of movement involved in the act of perception. Concentrating first on visual verbs: *takeꞌa* 'see' indicates that a signal, originating from the stimulus, is perceived by the experiencer: there is a movement from the stimulus (the perceived object) to the experiencer (the subject). *Uꞌi* 'look', on the other hand, indicates that the experiencer directs his/her attention towards the stimulus: there is a movement from the subject to the object.<sup>58</sup>

This has consequences for the use of directionals with these verbs. When a directional is used with *uꞌi*, this directional tends to indicate a movement from the subject (the experiencer) to the object (the stimulus), whereas with *takeꞌa* the directional tends to indicate a movement from the object towards the subject.

Now it is safe to assume that there is a tendency for the subject of the clause to act as the deictic centre, at least in first-person and third-person contexts: as discussed above,

<sup>56</sup> Besides *take ꞌa* and *tike ꞌa*, there are also the less common variants *tikera* and *takera*. All four are synonymous.

<sup>57</sup> Osmond & Pawley (2009) use the terms "sensing" and "attending", respectively. In Rapa Nui, the two pairs of verb also show differences in subject marking (§8.3.2).

<sup>58</sup> See Hooper (2004: 1745) for a similar description of the two possible trajectories.

#### 7 The verb phrase

the deictic centre is usually either the speaker or a central participant in the discourse, both of which tend to be the subject of the clause. This leads us to expect that the controlled perception verb *uꞌi* 'look' will predominantly take the directional *atu*: when the subject is the deictic centre, there is an outgoing movement from the subject/experiencer towards the object/stimulus. On the other hand, the uncontrolled perception verb *takeꞌa* 'see' will predominantly take the directional *mai*: when the subject is the deictic centre, there is a movement from the stimulus towards the subject/experiencer.

This expectation is borne out in newer texts, as shown in Table 7.8: *uꞌi* is followed by *atu* in 339 cases and by *mai* in only 118 cases; by contrast, *takeꞌa* is followed by *mai* in 91 cases and by *atu* in just 4 cases.

In older texts, the difference is not as clear: with *takeꞌa*, only *mai* is used (though only in 18 cases), but with *uꞌi*, both directionals are used with similar frequency. This corresponds to the phenomenon observed above with speech verbs: in older texts there is a general preference for *mai*, while in newer texts *atu* is more common.<sup>59</sup>


Table 7.8: Directionals with verbs of seeing

Verbs of hearing show the same distinction in newer texts, as shown in Table 7.9: the controlled *hakaroŋo* 'to listen' tends to take *atu*, indicating outgoing attention from the subject as deictic centre, while the uncontrolled *ŋaroꞌa* 'perceive' usually takes *mai*, indicating incoming perception towards the subject as deictic centre. Again, in older texts this tendency does not show up, though in the case of *ŋaroꞌa* data are scarce overall.

These tables also show that directionals *as such* are more common with the controlled perception verbs *uꞌi* and *hakaroŋo* than with *takeꞌa* and *ŋaroꞌa*: *uꞌi* takes a directional in about 35% of all occurrences, *hakaroŋo* even over 45%; on the other hand, *takeꞌa* and *ŋaroꞌa* are followed by a directional in less than 15% of all occurrences. This is true in both older and newer texts.

In individual instances, the choice for *mai* or *atu* may be governed by other considerations: with any perception verb, the speaker may choose either the Experiencer or the Stimulus as deictic centre, depending on the dynamics of the discourse. But over the whole of the corpus, there is a clear correlation between verb type (controlled or uncontrolled perception) and the choice of directional.

<sup>59</sup> Notice, however, that for motion verbs there is no such shift: *mai* is predominant both in older and newer material.

<sup>60</sup> Both *hakaroŋo* and *ŋaro ꞌa* are predominantly used for aural perception, though *ŋaro ꞌa* (and occasionally *hakaroŋo*) may be used for perception in general ('to feel, perceive') as well.

7.5 Directionals


Table 7.9: Directionals with verbs of hearing

### **7.5.3 To use or not to use a directional**

In §7.5.1 and subsections, the use of directionals has been discussed as a binary choice: a speaker may use either *mai* or *atu*. However, the statistics in §7.5.2 show that verbs which take directionals only do so in a minority of all cases. For example, only about 30% of all motion verbs in the corpus are followed by a directional. The speaker is thus faced with a ternary choice: *mai*, *atu*, or no directional at all. One more question must therefore be addressed: which factors influence the choice between using a directional and using no directional at all?

Some of the factors which may play a role are the following:

**Directionality** Directionals are used when a movement (physical or metaphorical) is clearly directional and when the speaker wishes to state so. In the following example, Eva first looks into a general direction; then she looks to a more precise location. Only the second verb is followed by a directional.

(153) *He* ntr *uꞌi* look *a* by *ruŋa* above *i* at *te* art *henua…* land *E* ipfv *uꞌi* look *mai* hither *era* dist *a* by *tū* dem *kona* place *kī* say *era* dist *e* ag *nua* Mum *pē* like *nei* prox *ē:* thus *'ꞌI* at *ꞌOhovehi* Ohovehi *mātou* 1pl.excl *ka* cntg *noho* stay *nei'.* prox 'She looked towards the land… She kept looking towards the place about which Mum had said: 'We will stay in Ohovehi'.' [R210.082–083]

**Highlighting** Directionals subtly highlight the deictic centre of the text. The speaker may therefore choose to use directionals to point to the deictic centre, whether this is constant or shifting. For example, the story *Nuahine rima roa* (§7.5.1.2.1) contains numerous occurrences of *mai* which point to the central participant, the old lady.<sup>61</sup>

**Participant reference** As discussed in §7.5.1.1, ex. (131–132), directionals may play a role in participant reference: directionals indicate whether a participant is at the origin

<sup>61</sup> This may explain why *mai* is more frequent overall than *atu*, see the statistics in the previous sections. Hooper (2004: 1742) mentions a 60/40 proportion for *mai* and *atu* in Tokelauan discourse.

#### 7 The verb phrase

or the goal of the movement, so they may be used instead of an overt subject or object. This accounts for many occurrences of directionals, for example with speech verbs in direct discourse, as in (131).

**Distance** Possibly, directionals tend to be used when there is a significant distance between the origin and the goal of movement, e.g. between the speaker and the addressee. I have not found many instances where this is is the only factor involved, but there are examples which can plausibly be explained this way. In (154), *eꞌa* 'go out (of the house)' is not marked by a directional, while *oho* 'go' is; the latter involves movement over a considerable distance, while the former does not.

(154) *Ka* imp *eꞌa* go\_out *koe* 2sg *ka* imp *haka* caus *rivariva* good:red *i* acc *te* art *poki,* child *ka* imp *oho* go *atu* away *kōrua* 2pl *ki* to *Haŋa* Hanga *Piko.* Piko

'Go outside and prepare the child, and go to Hanga Piko.' [R210.036]

This list is not exhaustive, if only because it does not explain all occurrences of *mai* and *atu*. Moreover, many instances can be explained in more than one way. These factors are no more than possible considerations which may play a role; they influence rather than determine the choice for a directional.

### **7.6 Postverbal demonstratives**

### **7.6.1 Introduction**

The postverbal demonstratives (PVDs) *nei*, *ena* and *era* indicate spatial or temporal distance of the event with respect to a place and/or time of reference. The same forms also occur in the noun phrase (§4.6.3). Both in the noun phrase and in the verb phrase they have the following sense:


PVDs cannot be added to just any verb phrase: as the discussion of aspectuals in §7.2 shows, PVDs occur in certain syntactic contexts and convey certain syntactic nuances.


7.6 Postverbal demonstratives

• With the perfect *ko V ꞌā*, *era* is occasionally used to express an action which is well and truly finished (§7.2.7.3).

In addition, PVD's are used in relative clauses (§11.4.5; see also (162) below).

The neutral aspectual *he* is rarely followed by a PVD. PVDs do not occur after the imperative markers *ka* and *e*. Neither are they found after negators *kai* and *e ko*, and after subordinators like *mo* 'in order to' and *ana* 'irrealis'.

The use of PVDs after perfective *i* warrants separate treatment. After *i*, the verb is often followed by a PVD; the list of PVDs after *i* also includes a fourth PVD: *ai*, which is not used after other aspectuals. In fact, *ai* is the default PVD after *i*, except in cohesive clauses. This will be discussed in §7.6.5.

Table 7.10: Frequencies of postverbal demonstratives


In the following subsections, the four PVDs will be discussed in turn. First a statistical note. Table 7.10 shows frequencies for the *era*, *ena* and *nei* in all verb phrases in the text corpus. As this table shows, *era* is far more frequent than *ena* and *nei*: 72% of all PVDs in the text corpus are *era*. This suggests that *era* is the default PVD; it is used whenever a PVD is called for and there is no reason to use *nei* or *ena*. For this reason, the use of *era* will only be discussed as it relates to *nei* and *ena*.

### **7.6.2 Proximal** *nei*

*Nei* marks actions which are either performed by the speaker, take place close to the speaker, or happen at a time close to the time of speech. Any of these is sufficient to warrant the use of *nei*; neither is a necessary condition.

*Nei* often marks an action performed by the speaker, i.e. in the first person, as in (155).

(155) *ꞌO* because\_of *ira* ana *a* prop *au* 1sg *i* pfv *iri* ascend *mai* hither *nei* prox *ki* to *a* prop *koe.* 2sg 'Therefore I have come up to you.' [R229.208]

Alternatively, the event may take place near the speaker as in (156), or is directed towards the location of the speaker as in (157):

(156) *Pē* like *nei* prox *e* ipfv *kī* say *nei* prox *e* ag *te* art *nuꞌu* people *nei:* prox *ko* prf *mate* die *ꞌana* cont *koe.* 2sg 'This is what these people say: you have died.' [R229.316]

#### 7 The verb phrase

(157) *¿ꞌI* at *hē* cq *rā* intens *a* prop *Vaha* Vaha *e* ipfv *taꞌe* conneg *tuꞌu* arrive *mai* hither *nei?* prox 'Where is Vaha, that he doesn't arrive?' [R229.131]

Occasionally *nei* has a temporal rather than a spatial function. In (158), the speaker talks about something habitually taking place in the present.

(158) *Te* art *vaka* boat *o* of *te* art *hora* time *nei,* prox *e* ipfv *hahaꞌo* insert *nei* prox *te* art *aroaro* lining *ꞌi* at *ruŋa* above *o* of *te* art *kavakava* rib *mau* really *ꞌana.* ident 'The boats of nowadays, they put the lining on top of the ribs.' [R200.068]

In narrative contexts, events usually do not take place close to the speaker, nor in the present. Even so, *nei* occurs in narrative as well. By using *nei*, the speaker indicates that the action is spatially close to the locus of discourse, or takes place near the time of reference:

(159) *Mahana* day *nei* prox *i* pfv *iri* ascend *nei* pfv *ki* to *te* art *māmoe* sheep *mo* for *toke* steal *he* ntr *maꞌurima* catch *o* of *tōꞌona* poss.3sg.o *pāpātio.* uncle

'This day when he went to the sheep to steal, his uncle caught him.' [R250.222]

(160) *I* pfv *eꞌa* go\_out *nei* prox *te* art *taŋata* man *nei* prox *ꞌi* at *tū* dem *raꞌā* day *era* dist *he* ntr *oho* go *mai* hither *ē…* on\_and\_on 'When this man had gone out that day, he kept going…' [R310.136]

As these examples show, *nei* in the verb phrase may co-occur with *nei* in the subject or another noun phrase in the clause (cf. also (156) above).

### **7.6.3 Medial** *ena*

*Ena* usually indicates an action performed by the addressee, an event taking place close to the addressee, or an event at a medial distance (i.e. not near the speaker, but not very far either). Either of these factors may trigger the use of *ena*.

Often *ena* marks an action performed by the addressee:

(161) *¿He* pred *aha* what *koe* 2sg *e* ipfv *taŋi* cry *ena?* med 'Why are you crying?' [R229.185]

Sometimes the action takes place near the addressee, as in (162), or at a little distance from both speaker and addressee, as in (163).

7.6 Postverbal demonstratives


Notice that in (163), postverbal *ena* is paralleled by *ena* in the preceding noun phrase. *Ena* may also have a temporal function: it refers to a moment somewhat removed from the present. This may be the near past as in (164), or the near future as in (165):


In narrative, *ena* is especially used after the deictic particle *ꞌī* (§4.5.4.1.1), which signals a shift to the point of view of a participant in the story (often with a verb of perception). The use of *ena* in this construction may be metaphorical, indicating that the reader is conceptually closer to the events in the story than usual, looking as it were through the eyes of the participant.

(166) *ꞌĪ* imm *ka* cntg *uꞌi* look *atu* away *ena* med *ko* prom *te* art *ꞌata* shadow *o* of *te* art *taŋata…* man 'Then he saw the shadow of a man…' [R304.095]

### **7.6.4 Neutral/distal** *era*

*Era* is the default PVD. It is especially common in narrative contexts, where proximity to speaker and hearer does not play a role. *Era* occurs in numerous examples in the discussion of aspectuals in §7.2.

Other PVDs are only used when there is a specific reason to do so. As discussed above, *nei* is used when the action is performed by the speaker, takes place close to the speaker, takes place in the present, or is metaphorically proximate in discourse. Likewise, *ena* can be used when the action is performed by or near to the hearer, takes place at a moderate distance, or at a time somewhat close to the present. This does not mean that

#### 7 The verb phrase

*nei* or *ena* is always used whenever one of these conditions is fulfilled. *Era*, being the default PVD, can be used for an action performed by the speaker as in (167–168), or an action performed by the hearer as in (169–170). In all these cases, however, distance is involved: in (168–170) the event takes place in the past; in (167) the event is hypothetical, and therefore also removed from the here and now.


To summarise: PVDs are used in combination with aspectuals to convey certain aspectual nuances. The default PVD is *era* (except in certain contexts with the perfective marker *i*, where *ai* is more common, see §7.6.5). *Nei* and *ena* may be used to convey proximity and medial distance respectively; distance is usually defined in spatial terms with respect to a participant or locus of discourse, but may also have a temporal sense.

### **7.6.5 Postverbal demonstratives with perfective** *i*

The perfective marker *i* was discussed in §7.2.4. The examples in that section show, that an *i*-marked verb is usually followed by a postverbal demonstrative (PVD). Besides *era*, *nei* and *ena*, *i* (unlike other aspectuals) allows a fourth PVD, *ai*. In fact, *ai* is by far the most common PVD after *i*. Only in cohesive clauses (§11.6.2.1) is the verb usually followed by *era*, while *ai* is rare.

This raises the question what the function of *ai* could be. Now the particle *ai* is common in Polynesian languages. Rapa Nui is different from other languages in that *ai* is not used after all aspectuals; apart from perfective *i*, it is only found after the purpose marker *ki* (§11.5.3). There is also a functional difference. Chapin (1974) shows that in all languages except Rapa Nui, *ai* is anaphoric: it occurs when the verb is preceded by any constituent other than a nominative subject; it serves as a substitute for the preposed constituent. This does not hold in Rapa Nui: in many cases *ai* occurs in verb-initial clauses,

#### 7.6 Postverbal demonstratives

or in clauses where the verb is preceded by a subject. Even so, there is a correlation between the occurrence of preverbal constituents and the use of *ai*. Table 7.11 shows the occurrence of *ai* and other PVDs in *i*-marked clauses (cohesive clauses excepted), differentiated for preverbal constituents: either a core argument (subject or direct object), an oblique constituent (locative or temporal phrase, connective adverb, or question word), or none at all (verb-initial clauses):

Table 7.11: Postverbal demonstratives with *i*-marked verbs


As this table shows, when the verb is preceded by an oblique constituent, it is followed by *ai* in 80% of the cases. By contrast, when the verb is preceded by a core argument, *ai* is relatively rare (21%), while 60% of the cases have no PVD at all. These statistics show a similarity in the use of *ai* between Rapa Nui and other Polynesian languages: *ai* tends to be used after oblique constituents, but not after NP arguments.<sup>62</sup> Still, the situation is much fuzzier than in other languages: *ai* does occur after NP arguments, while after oblique constituents other PVDs occur as well as *ai*.

In verb-initial clauses, *ai* is almost as common as with oblique preverbal constituents (72%), a situation not found in other languages. Possibly the use of *ai* in these clauses can be explained to some extent in terms of inter-clausal (rather than intra-clausal) anaphora. For example, in (171) *ai* could be explained as providing an anaphoric link with the preceding clause.

(171) *¿I mamae rō koe i hiŋa ai?*

> pfv pain emph 2sg pfv fall pvp

'Did you get hurt when you fell down?' [R481.131]

On the other hand, many examples of *ai* cannot be explained in this way.

Turning now to the other PVDs *nei*, *ena* and *era*, these are relatively rare with *i*-marked verbs (except in cohesive clauses, see §11.6.2.1). As Table 7.11shows, out of 219 verbs, only 31 (14%) are followed by one of these. Of these 31 cases, 13 have *nei*, 10 have *ena*, 8 have *era*. These proportions are remarkable, as *era* is much more frequent in general than *nei*

<sup>62</sup> Notice that, different from what Chapin (1974) found in other languages, in Rapa Nui any NP argument, whether subject or object, disfavours the use of *ai*.

In fact, Chapin (1974: 299) found a similar correlation: counting occurrences of *era* and *ai* in Englert's stories (Egt), concludes: "of the 26 cases discovered of verbs in *i* tense with no PVD, all but three or possibly four contain patterns which would lead one on comparative grounds not to expect *ai.* Of the nearly 100 cases of post-verbal *ai,* all but about a dozen appear according to comparative expectations."

#### 7 The verb phrase

and *ena*: as the statistics in §7.6.1 show, *era* accounts for 72% of all occurrences of these three PVDs overall, but in the constructions considered here, *era* represents only 26% of all three PVDs. Even though the sample is small and therefore liable to skewing by a few aberrant examples, the difference is significant.

In other contexts, *era* is the default PVD; *nei* and *ena* are only used when there is a specific reason to use them, to indicate close distance (*nei*) or medial distance (*ena*) (§7.6). By contrast, with *i*-marked verbs, *ai* is the default PVD. *Nei* and *ena* may be used to indicate close and medial distance; *era* may either be a free (but relatively rare) alternative to *ai*, or used only when the speaker wishes to emphasise distance.

### **7.7 Serial verb constructions**

### **7.7.1 Introduction**

Serial verb constructions (SVCs) are constructions in which two or more verbs occur in a single clause, without being so closely linked that they form a verbal compound.<sup>63</sup> Verbs within an SVC have the same specification for tense/aspect/mood and they usually share one or more arguments. They are not separated by a conjunction or by anything marking a clause boundary. The events expressed within an SVC are closely linked: SVCs tend to express a single event, or a set of events considered to be part of a single "macroevent".<sup>64</sup> Certain verb combinations may be lexicalised in a language, but SVCs tend to be productive.

Serialisation is common in Austronesian languages, including Oceanic languages (see Crowley 2002; Senft 2008; Durie 1988), but rare in Polynesian.<sup>65</sup>

### **7.7.2 The syntax of SVCs in Rapa Nui**

Rapa Nui is unusual among Polynesian languages in that SVCs are fairly common.<sup>66</sup> Moreover, it is – to my knowledge – the only Polynesian language in which all verbs in

<sup>63</sup> On SVCs in general see Durie (1988; 1997); Aikhenvald & Dixon (2006); Sebba (1987). On SVCs in Oceanic languages, see Crowley (2002) (+ reviews by Owens (2002) and Bradshaw (2004)); Senft (2008). These studies do not agree on a precise definition of SVCs (it is even uncertain if such a definition is possible, given the crosslinguistic variation in syntax and semantics of SVCs (Crowley 2002: 19)). They differ for example on the question whether SVCs necessarily constitute a single predicate. However, they do agree on the characteristics mentioned here.

<sup>64</sup> The term *macro-event* is discussed by Aikhenvald (2006).

<sup>65</sup> Both in Māori (Harlow 2007a: 150) and Tahitian (Académie Tahitienne (1986: 203)), the only traces of SVCs are motion verbs such as *haere* 'go' modifying another verb: Māori *i tangi haere* 'went weeping'. In Marquesan, this modifying construction also occurs (Cablitz 2006: 205–206). Cablitz also mentions bare complement clauses and clause chaining as examples of serialisation; however, clause chaining constructions are not monoclausal, hence they do not qualify as an SVC as defined above. The only reason to classify clause chaining constructions as SVCs is the absence of an A/M marker on the second verb. Mosel & Hovdhaugen (1992: 397), using the same criterion, identify the same three constructions as SVCs in Samoan. Finally, in Tuvaluan (Besnier 2000: 538) SVCs occur on a limited scale; again, the second verb is not A/M-marked.

<sup>66</sup> SVCs in Rapa Nui are discussed in Weber (2003b: 67-75); Weber uses the term *verb nesting* (anidación de verbos) and especially discusses criteria to distinguish SVCs from clause conjunction.

7.7 Serial verb constructions

an SVC have an aspect/mood (A/M) marker; the A/M markers within an SVC are always identical.<sup>67</sup>

Apart from the A/M marker, nothing can occur between the verbs in an SVC. Postverbal particles – including obligatory particles – only occur after the last verb. Arguments of both verbs are placed after the last verb; preposed arguments occur before the first verb. The structure of a clause with serialisation is thus as follows:

(172) (constituent) [A/M<sup>i</sup> V1 A/M<sup>i</sup> V2 (particles) ]VP (constituents)

Most SVCs have two verbs, but longer series occur. The verbs in an SVC usually share their S/A argument. In fact, the SVC as a whole has a single argument structure, which is determined by the verb with the highest valency: if both verbs are intransitive, the SVC as a whole is intransitive; if one verb is transitive, the SVC is transitive.

Below are a number of examples which illustrate the characteristics of SVCs.


'…lest the disease keeps infecting other people who are still healthy.' [R398.017]

Aspectuals – *ko* in (173), *i* in (174), *he* in (175) – are repeated before each verb. Postverbal *ꞌā* (173) and *era* (174) occur only after the second verb.<sup>68</sup> In (175) – a tripartite SVC – *i tū manu era* is the direct object of the first verb *maꞌu*, yet it occurs after the SVC construction as a whole. (176) shows that subordinators like *ꞌo* are repeated in the same way as aspectuals.

<sup>67</sup> While bare verbs can modify nouns (§5.7.1), they never modify other verbs without a preceding A/M marker. Some Polynesian languages have a V + V construction (see Footnote 65 above), but this does not occur in Rapa Nui.

<sup>68</sup> Both are obligatory, given the construction: perfect *ko* is always accompanied by *ꞌā* (§7.2.7), while *i V era* in (174) marks a cohesive clause (§11.6.2.1).

#### 7 The verb phrase

These examples also show how SVCs can be distinguished from coordinated clauses. As verb arguments are often omitted in discourse, a string of verbal clauses may consist of just *A/M V A/M V…* (see e.g. (3) on p. 317); such a string may at first sight be indistinguishable from an SVC. Diagnostics for SVCs are: the omission of postverbal particles after the first verb, and the placement of the direct object of a verb after the next verb (even when the latter is intransitive). SVCs can also be recognised by semantic criteria, as they often express a single event; this will be discussed in the next section.

In nominalised SVCs, the determiner is repeated. Any preposition preceding the nominalised verb is repeated as well, e.g. *pe* in (178):<sup>69</sup>


There is only one situation in which V2 is unmarked: when the SVC functions as a bare relative clause (§11.4.5), in which case neither verb in the SVC has an A/M marker.

(179) *ꞌI* at *te* art *hora* time *turu* go\_down *oho* go *nei* prox *ōꞌoku* poss.1sg.o *ki* to *Haŋa* Hanga *Roa* Roa *o* o *Tai…* Tai 'When I went down to Hanga Roa o Tai…' [R230.059]

### **7.7.3 Semantics of SVCs**

Most SVCs refer to a single event, which is expressed by one verb (usually the first in the series) and modified in some way by the other verb(s) (categories 1–3 below).<sup>70</sup> Other SVCs express a series of closely connected events which are conceived as one macroevent (category 4).

**7.7.3.1 Aspect** V2 may express an aspectual specification of the event. Only two verbs are used in this way.

*Oho* 'go' is by far the most common V2 in SVCs. It often expresses extended duration, indicating that the action expressed by V1 goes on for a while. As (182) shows, when V1 is an adjective, the SVC expresses an ongoing process.

(180) *Pē* like *rā* dist *nō* just *e* ipfv *kai* eat *e* ipfv *oho* go *era.* dist 'In that way he kept eating.' [R310.225]

<sup>69</sup> (177) is an habitual actor-emphatic construction (§8.6.3).

<sup>70</sup> The same is true crosslinguistically: directional and aspectual SVCs are very common (Aikhenvald 2006, who also mentions all the other categories found in Rapa Nui: manner, synonymy, sequential events)

7.7 Serial verb constructions


*Oti* 'finish' is usually constructed with a complement clause (§11.3.2.2), but it may also function as V2 of an SVC, indicating that an action or process is completely carried out:


**7.7.3.2 Direction** V2 may be a motion verb specifying the direction in which the action expressed by V1 takes place. The motion verb may be *iri* 'go up', *turu* 'go down' or *hoki* 'go back, return'. The idea of movement itself may be expressed by V1 (e.g. *haro* in (185)), but in other cases such as (186), V1 by itself does not express movement.


*Oho* 'go' is mostly used in SVCs to express duration (see 1 above); however, it may also express motion in a certain direction, without specifying the direction itself.<sup>71</sup> In these cases, no extensive duration is implied.

<sup>71</sup> These examples are somewhat similar to category 3, in which the V1 specifies the manner in which V2 is performed. The difference is, that *kau* and *nekeneke* are themselves motion verbs, while the modifying verbs in category 3 are statives.

#### 7 The verb phrase


**7.7.3.3 Manner** One verb in the SVC may be a stative verb indicating the manner in which the action expressed by the other verb is carried out. Usually the stative verb comes first, while the event itself is expressed by V2.


'The fish kept coming up easily, until it got tired of it.' [R361.053]

**7.7.3.4 Other** In other cases, both verbs describe an event. The verbs may be closely related or near-synonyms as in (192), both expressing the same event under different angles; they may also describe different aspects of the same event as in (193) ('ask in writing' or 'write to ask'). Alternatively, they express sequential events considered to be part of the same macro-event, as in (194).


### **7.8 Conclusions**

Verbs are preceded and followed by a range of particles which specify the event for aspect, mood, distance and direction.

Aspect is primarily indicated by a set of five preverbal markers; the use of these markers is obligatory, unless the verb is preceded by a different marker (such as mood and negation) occurring in the same position. The aspectual markers are as follows: neutral *he*, perfective *i*, imperfective *e*, contiguity *ka* and perfect *ko V ꞌā*. The neutral marker *he* is by far the most common one. It marks events which receive their aspectual value from the context in some way; in the absence of other contextual clues, a string of *he-*clauses expresses sequential events in discourse.

The relationship between perfective *i* and perfect *ko V ꞌā* calls for an explanation. Comparison of both markers in similar contexts suggest that *ko V ꞌā* is state-oriented, while *i* is event-oriented. The state-oriented character of *ko V ꞌā* also shows up in its widespread use to mark a state which pertains at a time of reference (usually the present). This happens with typically stative verbs such as 'be hot, big, poor, mad…', but also with verbs of volition and cognition.

Finer aspectual distinctions are indicated by postverbal particles; different classes of particles play a role with different aspectuals:


One pair of postverbal particles operates entirely independently from aspect marking: the directional markers *mai* and *atu*. In direct speech, *mai* indicates movement towards the speaker, while the use of *atu* is varied: movement from the speaker towards the hearer, away from speaker and hearer, or from an unspecified source towards the hearer.

In third-person discourse, the use of *mai* and *atu* marks a deictic centre. The speaker has a high degree of freedom in defining the deictic centre: it may be relatively fixed (often depending on the location of one or more protagonists in the story) or shift rapidly between different locations. Statistics show some general trends, though: with motion verbs, directionals tend to point to the destination of movement as the deictic centre. With perception verbs, there is a difference between controlled perception ('to look, listen') and uncontrolled perception ('to see, hear'): with the former, directionals indicate a movement from the experiencer to the stimulus, i.e. directed attention; with the latter the direction is reversed, i.e. directionals signal the movement from the stimulus towards the experiencer.

Finally, Rapa Nui is the only Polynesian language having a serial verb construction in which the preverbal marker is repeated. Apart from the preverbal marker, nothing may occur between the verbs in this construction. Serial verb constructions form a single predicate with a single argument structure; they often express a single event.

A verbal clause consists of a verb phrase and optional nominal arguments and adjuncts. The number of arguments depends on the verb; different classes of verbs are discussed in §3.4.1.

The verb phrase has been discussed in Chapter 7; the present chapter focuses on the other core constituents of verbal clauses: the arguments of the verb. The chapter is dominated by two main topics: constituent order and argument marking. These two are inextricably linked – the way arguments are marked, depends on their position in the clause – so they will be discussed together; the discussion will focus on the factors determining the marking of subject and object.

Constituent order and argument marking are discussed in sections §8.1–8.7. §8.1 provides a brief introduction and discusses basic and marked constituent orders. §8.2 introduces the topic of case-marking, comparing the situation in Rapa Nui with other Polynesian languages. The next sections deal with S/A marking (§8.3) and O marking (§8.4), respectively. §8.5 discusses passivisation and passive-like constructions. §8.6 discusses a variety of constructions involving non-standard constituent orders and/or noncanonical marking of arguments, e.g. topicalisation and instrumental marking. §8.7 deals with case marking in nominalised clauses.

The last sections deal with miscellaneous constituents, some of which are not restricted to verbal clauses, but which are nevertheless included in this chapter: oblique arguments (§8.8), reflexives and reciprocals (§8.9), comitative constructions (§8.10) and vocatives (§8.11).

Finally, §8.12 discusses causativisation, a process which affects the argument structure of the verb and the expression of arguments.

### **8.1 Introduction; constituent order**

As pointed out above, most of this chapter will be concerned with the order of constituents and the marking of S, A and O arguments.<sup>1</sup> A preliminary question concerns the expression of these arguments as such. The verb phrase is the only obligatory element in the verbal clause: any argument can be omitted if its identity is understood from the context. In discourse, both S/A and O are usually left implicit when they are identical

<sup>1</sup> See Footnote 24 on p. 91 on the terms S, A and O. In this grammar, any clause in which an O argument is either expressed or implied, is counted as transitive (regardless other arguments); a clause without an expressed or implied O is considered intransitive (cf. (85–87) on p. 106). Verbs with a nominalised verb as complement are counted as transitive; verbs with a subordinate clause as complement are counted as intransitive.

to a constituent in the previous clause. An example in which both A and O are implied, is the following:

(1) *He* ntr *moko* rush *ki* to *muri* near *i* at *tū* dem *viꞌe* woman *era* dist *ko* prom *Māhina* Mahina *he* ntr *haꞌi.* embrace 'He rushed toward that woman Mahina and (he) embraced (her).' [R399.191]

Table 8.1 shows how often arguments are expressed or not expressed in a corpus of selected texts.<sup>2</sup> The top part gives figures for intransitive clauses (i.e. clauses in which only one core argument is expressed or implied), the bottom part for transitive clauses (i.e. clauses in which both an A and an O argument are expressed or implied).

This table shows that in 47.8% of all intransitive clauses, the argument is expressed. Of the transitive clauses, only 7.7+15.4=23.1% have an overt A, while 47.1+15.4=62.5% have an overt O. In only 15.4% of all clauses are both arguments expressed, while in 29.8% of the clauses neither argument is expressed.

The default constituent order is VS/VAO. This order is by far the most common one and pragmatically unmarked. Other orders are not uncommon, though. Table 8.2 gives frequencies for all possible constituent orders. Part 1 represents clauses only containing an S/A argument; part 2 represents transitive clauses only containing an O argument; part 3 represents transitive clauses with two overt arguments.

As this table shows, there is a strong preference for verb-initial clauses,<sup>3</sup> but it is not uncommon for S/A to precede the verb (S/AV, AVO, AOV, OAV). It is less common for the object to precede the subject (VOA, OVA, OAV), while clauses in which the object precedes the verb (OV, OVA, AOV, OAV) are rare.<sup>4</sup>

Constituent order can be formulated as a set of three constraints:


<sup>3</sup> These data do not confirm Fischer's suggestion (Fischer 2001a: 323 that SVO is becoming the new unmarked word order (under influence of Spanish). It is true that new texts show a higher proportion of SV(O) clauses than old texts; however, it is also true that OV has become more common in new texts. The former may be under Spanish influence, but these shifts also suggest a move towards a more flexible syntax, in which a greater variety of constructions becomes common.

(i) *¿Mo* for *aha* what [*te* art *ꞌuha* ]<sup>O</sup> chicken [*ꞌā ꞌau* ]<sup>A</sup> poss.2sg.a *i* pfv *tiaŋi* kill *ai?* pvp 'Why did you kill the chicken?' (R250.164)

<sup>2</sup> For the analysis of clause structure and case marking, I used a subcorpus of 15 older texts (pre-1940) and 14 newer texts (post-1970). This corpus contains 7807 verbal clauses (2373 in old texts, 5434 in new texts): 2686 transitive (including three-argument verbs), 4879 intransitive and 242 with zero valency.

<sup>4</sup> The following example, an actor-emphatic construction with preposed object, is an example of OAV order (other orders will be exemplified in detail in the following sections):


Table 8.1: Expression and non-expression of arguments

Table 8.2: Frequencies of constituent orders


Constituent orders which violate only one constraint (like AVO) are more common than orders violating two or three constraints (like OAV). The statistics above also show that constraint 3 is strongest, while 1 is weakest: in clauses with both arguments expressed (413 total), constraint 1 is violated 95x, constraint 2 is violated 48x, constraint 3 is violated 28x.

There are various motivations for non-VAO constituent orders. S/A and O may be preposed as clause topic or because they are thematic (§8.6.1–8.6.2); S/A may be preposed in focus in the actor-emphatic construction (§8.6.3). Preverbal S/A also occurs after various clause-initial elements (§8.6.1.1).

Motivations for the reversal of A and O (i.e. VOA) are also diverse. Some VOA clauses are cases of passivisation (§8.5.1), in other cases the reasons for the marked order are less clear.

### **8.2 Case marking: introduction**

### **8.2.1 Case in Polynesian**

In Polynesian languages, nouns are not inflected for case. As far as case is marked, it is marked by prepositions. The subject of an intransitive clause is usually unmarked, i.e. not preceded by a case-marking preposition. For transitive verbs, three patterns are commonly distinguished (see e.g. Clark 1976: 67):


Certain languages (among which all the Central-Eastern Polynesian languages<sup>5</sup> ) exhibit accusative syntax:<sup>6</sup> the default pattern for all transitive verbs is I, in which A is unmarked like S, while O has an accusative marker. The choice of accusative marker depends on the semantics of the verb: for canonical transitive verbs, it is *i*; for middle verbs (§8.6.4.2), either *i* or *ki* is used. Pattern II is derived by passivisation: the Patient becomes the unmarked case (i.e. the syntactic subject); the Agent becomes an oblique and is marked by agentive *e*; the verb is followed by the passive suffix *-Cia* (where C is a consonant, the identity of which is lexically determined).

Most Tongic and Samoic-Outlier languages exhibit ergative syntax, at least for canonical transitive verbs: the unmarked pattern for these verbs is III, in which O is unmarked (like S) and A is marked with ergative *e*. The suffix *-Cia* may be added, resulting in pattern II; the difference in meaning between II and III is hard to pin down (Clark 1976:

<sup>5</sup> Some linguists have argued that Māori, an EP language, is ergative (see Harlow 2007a: 25, Pucilowsky 2006: 26–36 and refs. there); in this analysis, construction II (which is more common in Māori discourse than I) is considered the normal transitive construction, while the "active" construction I is an antipassive.

<sup>6</sup> On accusative and ergative languages, see e.g. Comrie (1978); Dixon (1994).

8.2 Case marking: introduction

71). Middle verbs in these languages occur in constructions I and II, just as in accusative languages.<sup>7</sup>

### **8.2.2 Case in Rapa Nui**

In a number of respects, Rapa Nui is like other Polynesian languages:

	- 2. O either has the accusative marker *i* or is unmarked.
	- 3. The object of middle verbs is marked with either *i* or *ki* (§8.6.4.2).

Despite these similarities, Rapa Nui seems not to fit either the accusative or the ergative group of languages, as it exhibits a number of differences with respect to both groups:


<sup>7</sup> Whether Proto-Polynesian was an ergative or an accusative language has been debated for decades. Clark (1976) argued that PPN was ergative, a position defended more recently by Kikusawa (2002; 2003) and Otsuka (2011). Hohepa (1969b), Chung (1978) and Ball (2007) argue that PPN was accusative.

As most non-EP languages are ergative and all EP languages apart from Rapa Nui are accusative, an interesting question is whether PEP was ergative or accusative. As Rapa Nui is clearly accusative (see Weber 2003a: 85, as well as the discussion in the following sections), the most natural account is that PEP was accusative as well.

(9) *He* ntr *mātaki* open *mai* hither **Ø** *te* art *ivi* bone *o* of *Ure* Ure *o* o *Hei* Hei *e* ag *te* art *taŋata.* man 'The man unpacked the bones of Ure o Hei.' [Blx-2-01.028]

In other languages, a given verb occurs either in patterns I and II, or in patterns II and III.


This pattern is illustrated in (6) above.


The occurrence of pattern III may give the impression that Rapa Nui is to some degree an ergative language.<sup>8</sup> However, 5, 6 and 7 show that *e* is different from an ergative marker: it occurs with both canonical and middle verbs, it co-occurs with an accusative marker (pattern IV), and it occurs in intransitive clauses. Moreover, as will be shown below, pattern IV is far more common in Rapa Nui discourse than pattern III.<sup>9</sup>

These observations suggest that, rather than looking for accusative or ergative patterns, it is more promising to consider case marking of subjects and objects separately:


Sections §8.3 and §8.4.1 will deal with these questions, respectively.

<sup>8</sup> For example, Otsuka (2011: 296) considers Rapa Nui a transitional language (between the two types), as it exhibits both V S iO and V eS O. See also Mosel (1997: 182).

<sup>9</sup> It is no surprise that the Rapa Nui case system may seem baffling. According to Clark (1973: 575), it is unclear under which conditions case markers in Rapa Nui can be omitted, while Chapin (1978: 168) admits not having found any regularity in the Rapa Nui case system. Alexander (1981a,b) formulates rules for the occurrence of case markers, an approach which yields valuable insights, though it is based on limited (and occasionally erroneous) data. Weber (1988a; 2003a) researches the issue on the basis of more extensive data; her approach, which is informed by discourse analysis, explains many of the patterns found in modern Rapa Nui texts.

8.3 Marking of S/A: the agentive marker*e*

### **8.2.3 Preliminaries to the analysis of case marking**

In order to trace patterns of case marking, I analysed and tabulated the occurrence, order and marking of core arguments in the corpus mentioned in §8.1 (Footnote 2). Now I pointed out in §5.3.2.1 that most prepositions – including agentive *e* and the accusative marker*i* – are obligatorily followed by a determiner. However, prenominal numerals and certain quantifiers preclude the use of determiners, and as a consequence, noun phrases starting with one of these elements cannot be marked by either *e* or *i*. Including these noun phrases in the counts would lead to a skewed picture with a high proportion of unmarked subjects and objects.<sup>10</sup>

Therefore, whenever frequencies of ØS/A and eS/A or frequencies of ØO and iO are compared,<sup>11</sup> noun phrases constructions containing a prenominal numeral or quantifier are disregarded. Also disregarded are other constructions where case marking prepositions are excluded:


All of these are included in the total number of arguments, but disregarded as far as case marking is concerned.

### **8.3 Marking of S/A: the agentive marker** *e*

The default S/A marker is Ø: the S/A argument is unmarked, unless there is a reason for using *e*. The use of *e* depends on syntactic, lexical/semantic, and discourse factors. These will be discussed in turn in the following sections.

### **8.3.1 Syntactic factors**

The use of *e* partly depends on the position of the S/A argument in the clause. Table 8.3 and Table 8.4 show how S/A arguments are marked in clauses with different constituent orders: Table 8.3 gives data for clauses containing only S or A, Table 8.4 gives data for transitive clauses containing both A and O.

<sup>10</sup> This point is also raised by Weber (2003a: 43), who also points out that the acc marker is impossible before complement clauses. As my analysis only considers NP objects, complement clauses are a priori disregarded.

<sup>11</sup> ØS/A = S or A without case marker; eS/A = S or A marked with *e*; ØO = O without case marker; iO = O marked with *i* or (with middle verbs) *ki*.


Table 8.3: Marking of S/A in one-argument clauses

Table 8.4: Marking of S/A in two-argument clauses


From these figures, a number of conclusions can be drawn.

In the first place: as a rule, preverbal S/A is not marked by *e*; it is either unmarked or has different marking. Of the few exceptions to this rule (1x AVO, 5x AV), most are object relative clauses (§11.4.2.2), such as the following:

(12) *Kai* food *nei* prox [*e* ag *au* 1sg *ka* cntg *naꞌa* hide *nei* prox *ꞌi* at *raro* below *i* at *a* prop *koe* ] 2sg *e* exh *maꞌu* carry *hiohio.* strong:red 'This food which I hide below you, hold it tight.' [R310.074]

Secondly: final A in a clause with two expressed arguments (i.e. VOA or OVA) is almost always marked by *e*; the following examples illustrate this:

(13) *He* ntr *tuꞌu* arrive *he* ntr *haka* caus *uru* enter *i* acc *te* art *ꞌuha* chicken *e* ag *Ŋumi* Ngumi *ki* to *roto* inside *ki* to *te* art *hare* house *ki* to *a* prop *Oti.* Oti

'Ngumi arrived and put the chicken in the house for Oti.' [MsE-105b.004]

8.3 Marking of S/A: the agentive marker*e*

(14) *ꞌO* because\_of *ira* ana *au* 1sg *i* pfv *haka* caus *ꞌariki* king *ai* pvp *e* ag *to* art:of *tāua* 1du.incl *matuꞌa.* parent 'Therefore our father made me king.' [Ley-2-06.03]

In the third place: *e*-marked arguments occur in both VS- and VA-clauses (i.e. both in intransitive clauses and in transitive clauses without an expressed O), but in very different proportions. In VA-clauses, 129 subjects are *e*-marked, while 30 are unmarked; in VS-clauses, 185 are *e*-marked, while 1643 are unmarked. That is, 81% of all potentially case-marked A are *e*-marked, against 10% of all potentially case-marked S.<sup>12</sup>

The intransitive examples will be further discussed in §8.3.4 below. Concerning transitive clauses, Weber (2003a) formulates the rule that A is obligatory case-marked when O is not expressed, as in the following examples:


'When this child was born, the old woman raised it with her husband.' [R352.005]

In some cases *e*-marking can be explained as disambiguation, as omission of the case marker would lead to ambiguity: in (15) the *hānau momoko* could also be interpreted as O, were it not for the case marker. But in other cases the sentence is unambiguous: in (16), the verb *hāŋai* 'feed, raise up' is used, which always has the parent as Agent and the child as Patient; even so, *e* is used.

The rule that VA-clauses must have *e* is not without exception: in 30 cases, *e* is omitted. No less than 18 of these occur in older texts.<sup>13</sup> In general, *e* is much less common in older texts than in newer texts (§8.3.5 below); out of 36 VA-clauses in older texts, only 18 are *e*-marked. This suggests that *e*-marking in these clauses was optional in older Rapa

(i) *He* ntr *to ꞌo* take *mai* hither *Kaiŋa* Kainga *matuꞌa,* father *he* ntr *tiŋa ꞌi* kill *i* acc *a* art *Kaiŋa* Kainga *poki.* child 'Father Kainga took (and) killed (his) son Kainga.' (Mtx-3-01.027)

This construction also occurs in new texts, but always with an *e*-marked subject. Possibly *to ꞌo* in this construction was conceived as intransitive in the past.

<sup>12</sup> Weber (2003a: 39) concludes that intransitive subjects marked with *e* are very infrequent. NB In these counts, serial verb constructions consisting of a transitive + intransitive verb have been considered as a single transitive verb phrase. (See Weber 2003a: 39 for examples.)

<sup>13</sup> For example, seven occur in the construction *to ꞌo* 'take' + transitive clause, a sort of clause-chaining construction in which the object of *to ꞌo* is expressed in the next clause. *To ꞌo* seems to indicate an initiative on the part of the subject. An example:

Nui; possibly *e*-marking was mainly used to avoid ambiguity, in cases where the only argument could also be misinterpreted as O.

In modern texts, only 12 VA-clauses have an unmarked A, while 111 are *e*-marked.<sup>14</sup> Weber's rule that A-marking in single-argument transitive clauses is obligatory, thus holds in newer texts with relatively few exceptions.

This rule also implies that A is *e*-marked in relative clauses with object relativisation (§11.4.2.2).

Finally, there is one more syntactic condition on the use of *e*: *e* is obligatory when a subject pronoun is followed by the identity marker *ꞌā* or *ꞌana* (§5.9).

```
(17) He
mātaki
        e
            ia
                mau
                       ꞌā.
```
ntr open ag 3sg really ident

'(His knock was not answered so) he opened [the door] himself.' [R399.189]

This is even true when the subject is preverbal, even though preverbal subjects are normally not *e*-marked:

(18) *E* ag *rāua* 3pl *mau* really *ꞌana* ident *ka* cntg *ꞌaꞌamu* tell *nei* prox *i* acc *te* art *rāua* 3pl *ꞌati.* problem 'They themselves told (about) their misfortune.' [R361.035]

### **8.3.2 Semantic patterns**

As discussed above, in other Polynesian languages the use of agentive *e* is restricted to transitive verbs: either canonical transitives only (in ergative languages), or any transitive verb in the passive (in accusative languages). The figures in the previous section show, that *e* in Rapa Nui is also used in intransitive clauses. Alexander (1981a: 143) suggests that *e* can be used with active intransitive verbs, those involving volition on the part of the subject.<sup>15</sup>

A more refined analysis shows, that there is a correlation between the use of *e* and the type of verb. In Table 8.5, all verbs in the corpus have been assigned to a semantic

<sup>14</sup> Concerning the 12 occurrences of transitive V ØS in newer texts, some may have been conceived as intransitive rather than transitive, i.e. the speaker may not have implied a direct object.

Four examples occur (somewhat unexpectedly) in object relative clauses, such as the following:

<sup>(</sup>i) *He* ntr *mana ꞌu* think *tahi* all *i* acc *te* art *me ꞌe* thing *ta ꞌato ꞌa* all *era* dist [*e* ipfv *aŋa* do *era* dist *a* prop *Kava* Kava *ararua* the\_two *ko* prom *Vaha* ]*.* Vaha 'He thought of all the things that Kava and Vaha did.' (R229.349)

<sup>15</sup> Alexander (1981a: 145) further suggests that Rapa Nui is an "active language", in which intransitive verbs are split along the following lines: Agent subjects can be marked like transitive Agents (i.e. with *e*), while Patient subjects – for example the subject of 'to fall' – can be marked like Patients (i.e. with the acc marker *i*). However, as Weber (2003a: 40) shows, the idea that subjects can be *i*-marked is based on an erroneous interpretation of the data.

8.3 Marking of S/A: the agentive marker*e*

category, and the number of ØS/A and eS/A counted.<sup>16</sup> The verb categories are roughly ordered by agentivity. (Percentages should be read horizontally, e.g.: with prototypical transitive verbs with markable subjects, 67% has zero marking, while 33% has *e*.)


Table 8.5: Verb classes and the use of *e*

This table shows, first of all, that *e* is largely limited to active participants. It is common with prototypical transitive verbs; a prototypical transitive involves a deliberate action performed by a volitional Agent, which affects the patient,<sup>17</sup> e.g. *Kai* 'to eat' and *tiaŋi* 'to kill'. It is also common with cognitive verbs such as *ꞌite* 'to know', perception verbs such as *takeꞌa* 'to see', and speech verbs such as *kī* 'to say'. It occurs occasionally with motion verbs such as *turu* 'to go down' and verbs of affection such as *haŋa* 'to love'. It is hardly – if ever – used with other agentive intransitives such as *piko* 'to hide oneself' and *ruruku* 'to dive', with adjectives/statives such as *ꞌitiꞌiti* '(to be) small', with process verbs (verbs which have a Patient or Theme subject) such as *haꞌuru* 'to sleep' and *hiŋa* 'to fall', with existential verbs such as *ai* 'to be', and with aspectual verbs such as *oti* 'to be finished'.

Even though the use of *e* is clearly correlated with agentivity, it cuts across the transitive/intransitive distinction. Speech verbs are usually intransitive (they may involve an addressee, but usually do not have a nominal object); even so, they commonly take an *e*-marked S. On the other hand, verbs of affection are often transitive, but rarely involve *e*-marking.

Remarkably, the highest proportion of *e*-marking is not found among prototypically transitive verbs. Prototypical transitive verbs do have a relatively high proportion of *e*-

<sup>16</sup> Other subjects, such as possessive subjects and NPs containing a prenominal numeral, have been disregarded. The right-hand column gives the totals of ØS and eS only.

<sup>17</sup> For discussion on prototypical transitivity, see Hopper & Thompson (1980) and more recently Ball (2007); Næss (2007).

marked Agents, but the same is true for cognitive verbs, which do not involve an affected patient. Moreover, there are two verb categories which are not prototypical transitives, yet which show an overwhelming preference for *e*-marking: (1) uncontrolled perception verbs; (2) *rovaꞌa/ravaꞌa* 'to obtain'.<sup>18</sup> For both, around 90% of all A arguments are *e*marked.<sup>19</sup> These will now be discussed in some detail.

The difference between controlled and uncontrolled perception verbs is discussed in §7.5.2.3. Uncontrolled perception verbs indicate the mere registration of a stimulus by the experiencer (*tikeꞌa/takeꞌa* 'to see' and *ŋaroꞌa* 'to hear, perceive'); controlled perception verbs involve deliberate attention on the part of the subject (*uꞌi* 'to look' and *hakaroŋo* 'to listen'). Perception verbs in general are not canonically transitive (as the O is not affected), uncontrolled perception verbs even less so (as the act may be involuntary); even so, about 90% of their A arguments are *e*-marked. An example:

(19) *He* ntr *takeꞌa* see *e* ag *Eva* Eva *tōꞌona* poss.3sg.o *nua* Mum *era* dist *ꞌi* at *tū* dem *kona* place *era.* dist 'Eva saw her mother there.' [R210.086]

By contrast, controlled perception verbs take an unmarked A in almost 90% of all occurrences.

*Rovaꞌa* 'to obtain' (var. *ravaꞌa, varaꞌa, rovā*, redup. *rovarovaꞌa*) also shows a strong preference for *e*-marking.<sup>20</sup> This verb usually involves a deliberate act, but the O is not affected to the same degree as with verbs like 'to eat' and 'to hit'. An example:

(20) *He* ntr *ravaꞌa* obtain *e* ag *rāua* 3pl *i* acc *te* art *vārua* spirit *era* dist *o* of *tū* art *repa* young\_man *era.* dist 'They obtained the spirit of that young man.' [R310.319]

These data make clear that *e* in Rapa Nui does not function as an ergative marker, as it does in Samoic and Tongic languages. It is not restricted to canonical transitives; there is even a tendency for it to be used more frequently with non-canonical transitives, verbs which have an O not affected by the action. *E* is not even restricted to transitives as such: it is used commonly with speech verbs and sometimes with motion verbs.

<sup>18</sup> Interestingly, both *tike ꞌa* (< PPN *\*kite* + \**-a*) and *ŋaro ꞌa* (< PPN *\*roŋo* + \**-na*) are historically passive forms, both of which underwent metathesis. This may well account for the predominance of "passive" syntax with an *e*-marked Experiencer and a Ø-marked Stimulus. The glottal in *tike ꞌa* is secondary, while the glottal in *ŋaro ꞌa* is derived from *n* in Rapa Nui (possibly *-na* > *-ra* > *-ꞌa*; the shift from *r* to glottal is not uncommon in Rapa Nui (§2.5.2; Davletshin 2015)).

<sup>19</sup> Of the few remaining cases, some involve a preverbal subject, which precludes *e*-marking (§8.3.1).

<sup>20</sup> The unusual syntax of *rova ꞌa/rava ꞌa* in Rapa Nui may have to do with its history. It was borrowed from Tahitian *roa ꞌa* and is one of the few borrowings already well established in older texts (§1.4.1). In Tahitian, *roa ꞌa* is a "patientive verb" (Lazard & Peltzer 2000: 241), meaning 'to be obtained, caught'; its Patient is expressed as subject, while the Agent is marked with agentive *i* (§8.6.4.7). The same is true for Hawaiian *loa ꞌa* (Elbert & Pukui 1979: 50). In Rapa Nui, *rava ꞌa/rova ꞌa* became an active and transitive verb, but the frequency of agentive *e*, together with the frequent absence of the acc marker (§8.4.1 below) shows that it retained some of its "patientive", passive-like character, even though its argument structure was fundamentally changed.

8.3 Marking of S/A: the agentive marker*e*

The use of *e* is linked to agentivity, though: it almost exclusively occurs with verbs that involve a volitional agent. (One apparent counterexample is discussed in the following section.) The only exception to this generalisation is, that *e* is far more common with uncontrolled perception verbs than with controlled perception verbs.

However, the notion of agentivity as involving a volitional participant deliberately performing the action, may be too narrow. While Agents are typically animate, *e*-marked constituents sometimes refer to an inanimate entity causing an event; this semantic role can be labelled Force (Payne 1997: 47). This happens especially in passives (see (58) in §8.5.1 below) and pseudopassives (§8.5.2, ex. (65)). The fact that *e* is used with inanimate entities, may indicate a gradual widening of its use, whereby 'agentivity' is defined in a looser way.

### **8.3.3** *E* **with statives?**

Weber (2003a: 36–37) argues on the basis of an example from Englert (1978) that *e* may also be used with stative verbs. The example is as follows:

(21) *E* ipfv *ora* live *rō* emph *e* ag *ia.* 3sg 'He will live.' (Englert 1978: 65)

While this is indeed an *e*-marked S with a non-agentive verb, it seems to be a slender basis to deny the agentivity of *e*. Notice that this is a single isolated example; it occurs without context in Englert's grammar sketch as an example of the future tense marked with *e V rō*. Secondly, the same sentence does in fact occur in a text by Englert (a translated Bible story), but there *e* is absent: *e ora rō ia* (Egt-03.041). This raises the question if (21) is not erroneous, or at least anomalous.

Thirdly, in the corpus I analysed, only one out of 413 *e*-marked arguments involves a stative verb:

(22) *Rohirohi* tired:red *e* ag *tāꞌaku* poss.1sg.a *poki* child *i* pfv *iri* ascend *ai* pvp *i* pfv *here* tie *mai* hither *ai.* pvp

'My son tired himself out when he went up to tie up (the sun).' [R352.099]

*E tāꞌaku poki* is the S of *rohirohi* 'tired'. Now *rohirohi* is normally stative, but in this case it may have an active sense: 'to work hard, to wear oneself out.' (The Spanish translation reads 'Se cansó mucho mi hijo al ir a amarrarlo'.)

We may conclude that the characterisation of *e* as an agentive marker remains valid; apart from a single example from Englert's grammar, all occurrences of *e* involve agentive participants, though – as stated at the end of the previous section – the notion 'agentivity' itself tends to be widened.

### **8.3.4 Pragmatic/discourse factors**

The preceding sections have shown that *e* is more or less obligatory in the following situations:


On the other hand, *e* is not used:


In the remaining situations, *e* is optional, i.e. in the following cases:


In contexts where *e*-marking is optional, the use of *e* is governed by discourse considerations: *e* marks Agents which are highly significant in the context. Usually this means that the participant has a high degree of agentivity.

*E* is used when a new participant is introduced in the Agent role. New participants in a story are usually introduced in a nominal clause or in a non-agentive role. In the following example, however, Kainga – who has not been mentioned before – is introduced as the Agent of the verb *hakaroŋo*, and *e*-marked:

(23) *I* pfv *oti* finish *era* dist *i* acc *te* art *hakaroŋo* listen *e* ag *Kaiŋa,* Kainga *te* art *matuꞌa* parent *tane* male *o* of *Huri* Huri *ꞌa* a *Vai,* Vai *he* ntr *kī…* say

'When Kainga, the father of Huri a Vai, had finished listening, he said…' [R304.011]

In the following example, the turtle (which will play an important role in the story) is introduced as Agent of *oho*:

(24) *He* ntr *oho* go *e* ag *te* art *honu* turtle *ꞌitiꞌiti,* small:red *he* ntr *raŋi* call *a* prop *Uho…* Uho 'A small turtle came by, and Uho shouted…' [Mtx-7-12.007]

*E* is also used when a participant which has been mentioned before, takes the initiative and starts to act.<sup>21</sup>

<sup>21</sup> Levinsohn (2007: 61) uses the term *prominent entities* for entities which have a significant role to play in the subsequent discourse, and which may therefore be highlighted in some way.

8.3 Marking of S/A: the agentive marker*e*

(25) *I* pfv *ꞌōtea* dawn *era* dist *he* ntr *eꞌa* go\_out *mai* hither *a* prop *Kaiŋa* Kainga *ararua* the\_two *ko* prom *Huri* Huri *ꞌa* a *Vai* Vai *ki* to *haho,* outside *he* ntr *kī* say *e* ag *Kaiŋa…* Kainga 'When dawn broke, Kainga went outside with Huri a Vai; then Kainga said…' [R304.017]

More generally, *e* tends to be used in the case of subject shift, when a different participant becomes active. In some dialogues, for example, every turn of conversation is marked with *e*. This explains the large number of *e-*marked S with speech verbs in certain texts:

(26) *He* ntr *kī* say *e* ag *Kuha* Kuha *ki* to *a* prop *Pea…* Pea *He* ntr *ꞌui* ask *e* ag *Pea* Pea *ki* to *a* prop *Kuha…* Kuha *He* ntr *kī* say *e* ag *Kuha* Kuha *ki* to *a* prop *Pea…* Pea 'Kuha said to Pea… Pea asked Kuha… Kuha said to Pea…' [R229.034–038]

Finally, *e* may be used when an Agent is emphatic because it is contrasted with other possible participants. This happens when it is singled out among a group (as in (25) above), when it is followed by *ꞌā/ꞌana* 'identity marker' (see (17–18) above), and in examples like the following:

(27) *E* ipfv *hāpaꞌo* care\_for *rō* emph *e* ag *au* 1sg *i* acc *tāꞌana* poss.3sg.a *poki.* child '*I* will take care of her child.' [R229.081]

These pragmatically motivated uses of *e* confirm that *e* is an agentive marker: when case marking is not determined by the syntax of the clause or the semantics of the verb, *e* is used when the participant is high in agentivity.<sup>22</sup>

### **8.3.5 Diachronic developments in the use of** *e*

Most of the examples in the previous section are from new texts. This is no accident: the use of *e* has significantly increased over time. Table 8.6 shows subject marking in old and new texts.<sup>23</sup> As this table shows, *e* is much more common in new texts than in old texts.<sup>24</sup> In old texts, 40 out of 739 potentially case-marked arguments are *e*-marked (5.4%), in new texts 368 out of 1608 (22.9%).

<sup>22</sup> Pragmatically motivated use of an agentive marker is not unique in Rapa Nui. Duranti (1990; 1994) gives examples from Samoan speeches where ergative *e* is used to emphasise agentivity and responsibility for an action, while other constructions are used to downplay a person's contribution towards an event.

<sup>23</sup> The column "other" includes all types of noun phrases which syntactically do not allow a case marker; see section §8.2.3.

<sup>24</sup> *Pace* Finney & Alexander (1998: 31), who assert that *e* is becoming less frequent under the influence of Tahitian.

In intransitive clauses, the difference is even more remarkable: in old texts, only 9 out of 610 intransitive clauses have *e*-marking (1.5%), in new texts 176 out of 1218 (14.4%).<sup>25</sup>


Table 8.6: Use of *e*: diachronic shifts in one- and two-argument clauses

In VA-clauses, A arguments are almost obligatorily *e*-marked in modern Rapa Nui, while in older texts only half are *e*-marked. Most uses of *e* in older texts can be explained either syntactically (VOA clauses) or lexically (with *rovaꞌa* 'obtain' or passive perception verbs); the pragmatically motivated uses described in the previous section are rare in old texts.

These data suggest that at an earlier stage *e* was only used in transitive clauses, in a limited number of contexts. Texts from the 1930s show the beginning of an extension of its use towards intransitive clauses, a use which is nowadays well established.

### **8.4 Marking of O**

### **8.4.1 Use and non-use of the accusative marker**

The O argument is normally preceded by the accusative marker *i*. With certain verbs, *ki* is used as well (§8.6.4.2).

The accusative marker is used whether A is expressed – postverbal as in (28) or preverbal as in (29) – or implicit as in (30–31):

<sup>25</sup> This partly confirms Finney's assertion (Finney 2001: 409) that *e* (which he labels "ergative") is becoming a marker for all subjects. Notice, however, that *e* is still largely limited to agentive verbs, as shown in sec. §8.3.2.

8.4 Marking of O


These examples also illustrate that the accusative marker is used whether O is definite or indefinite (as in (29)), and whether it is human or non-human (as in (30)). Moreover, disambiguation does not play a role in the use of the accusative marker: in all these examples it is semantically clear that the underlined NP must be O, yet the accusative marker is used.

Under certain conditions the accusative marker is omitted, either obligatorily or optionally.<sup>26</sup> These conditions are as follows:<sup>27</sup>

**Preverbal** Preverbal O is unmarked, just like preverbal S/A. This happens both in OVA clauses (32) and in AOV clauses (33):


'(It was) the tourists (who) took this photo.' [R415.735]

(i) *Ko* prf *māhani* accustomed *ꞌā* cont *a* prop *au* 1sg *e* num *tahi* one *kona* place *…* 'I know a certain place…' [R296.001]

<sup>26</sup> Alexander (1981b: 165) claims that a noun phrase (whether subject or object) is case-marked to bring it into focus. Noticing that the object is marked with *i* more often than not, Alexander suggests that possibly the object is often in focus. I will argue below that, while the presence of the acc marker does not signal focus or salience, its absence sometimes signals non-salience.

<sup>27</sup> As discussed in section §8.2.3, in certain noun phrases the use of a case marker is syntactically impossible. In order to analyse the use and non-use of the acc marker, these noun phrases should be disregarded. Thus, the following example is not counted as a case of an omitted acc marker, as a noun phrase starting with the numeral *e tahi* cannot contain a acc marker at all:

**Imperative** The accusative marker *i* is often omitted in the imperative mood (whether marked with *ka*, exhortative *e* or hortative *ki*):


The marker *ki* (used with middle verbs, see §8.6.4.2) is preserved, though:

(36) *Ka* imp *haŋa* love *ki* to *taꞌa* poss.2sg.a *kenu* husband *ko* prom *Pāpuꞌe.* Papu'e 'Love your husband Papu'e.' [R372.034]

When the subject of an imperative clause is expressed, O is always marked:

(37) *¡Ka* imp *ꞌaꞌaru* grab *mai* hither *koe* 2sg *i* acc *te* art *poki!* child 'Grab the child!' [R210.063]

Even when the subject is not expressed, the accusative marker may be used; this happens especially with pronominal objects:

(38) *¡Ka* imp *maꞌu* carry *i* acc *a* prop *au* 1sg *ki* to *tōꞌoku* poss.1sg.o *kāiŋa!* homeland 'Carry me to my country!' [Ley-9-55.089]

**Nominalised verbs** Certain verbs take a nominalised verb complement, i.e. a verb preceded by a determiner. These complements may or may not have the accusative marker.

Complements of *ꞌite* 'know' and *hāpī* 'learn' usually have the accusative marker:

(39) *¿Ko* prf *ꞌite* know *ꞌā* cont *koe* 2sg *i* acc *te* art *hī?* to\_fish 'Do you know how to fish?' [R245.101]

The complement of *oti* 'finish' may have the accusative marker, but only when the clause has a subject. This subject is the S/A argument of the complement verb, but is raised to the subject position of *oti* (§11.3.2.2). As the following pair of examples shows, after a raised subject the object marker is optional:

8.4 Marking of O


When the subject is not expressed, the complement is not marked.

(42) *He* ntr *oti* finish *te* art *puke* heap\_up *i* acc *te* art *ꞌuhi…* yam

'(When) they finished heaping up (earth mounds for) the yams….' [Mtx-2-01.010]

**Lexical factors** The verbs *rovaꞌa* 'obtain', *takeꞌa* 'see' and *ŋaroꞌa* 'hear, perceive', which usually have an *e*-marked A (§8.3.2), tend to take an unmarked O, especially when A is not expressed:

(43) *Paurō* every *te* art *pō* night *ka* cntg *pere* play *era* dist *he* ntr *rovaꞌa* obtain *te* art *tara.* money

'Every night, when he played, he obtained/won money.' [R250.146]

The accusative marker is sometimes used, but only when A is expressed (see (20) in §8.3.2).<sup>28</sup>

Constructions like (43) can be explained as passives (§8.5 below); this would mean that *rovaꞌa* is constructed passively when the Agent is not expressed – something which is not surprising, given the fact that passives serve to downplay the Agent and to enable the Patient to function as subject.

Apart from the three verbs mentioned above, the presence or absence of the accusative marker is generally unrelated to the way the subject is marked. This is illustrated in Table 8.7, which gives total frequencies for A- and O-marking in VAO clauses.


Table 8.7: Argument marking in VAO clauses

<sup>28</sup> Another peculiarity of *rova ꞌa* is its ability to take an incorporated object (§8.6.4.5).

As this table shows, omission of the accusative marker is relatively rare in VAO clauses (29 out of 249 possible cases, i.e. 11.6%), regardless whether A is *e*-marked or unmarked. As it happens, all but one of these unmarked O belong to categories 3 and 4 above. In other words: apart from the factors discussed so far, accusative marking in VAO clauses is obligatory.

Table 8.8: Object marking in VO clauses


On the other hand, in clauses without an overt A it is more common for the accusative marker to be omitted. Frequencies for VO-clauses are given in Table 8.8. <sup>29</sup> As this table shows, the accusative marker is omitted in 275 out of 1047 possible clauses (26.4%). The percentage is somewhat higher in old texts (32.6%, 126 out of 386) than in new texts (22.7%, 149 out of 655).

Now 138 of these can be explained by the factors above: these objects are nominalised verbs, occur with an imperative, or with one of the verbs in category 4. However, this leaves 137 cases unexplained in VO clauses, i.e. 13.1% of all potentially case-marked objects: 65 in old texts (19.4%), 72 in new texts (11.0%).<sup>30</sup> These will now be considered.

Turning to the 137 unexplained cases of omitted acc markers in VO-clauses: the first observation that can be made, is that almost all of these arguments are non-human. Many of them concern common collocations, verb-object combinations which frequently occur together. The sense of these collocations may or may not be idiomatic, but in all cases the object is highly predictable. Some of these expressions hardly ever occur with an accusative marker. Examples are *hoa (i) te ꞌaka* 'let down (lit. throw) the anchor', *maꞌoa (i) te ꞌumu* 'open the earth oven', *ꞌamo te vaꞌe* 'to lift up the feet = to stride':<sup>31</sup>

<sup>29</sup> OV-clauses are disregarded; as discussed under 1 above, preverbal objects are never marked.

<sup>30</sup> Weber (2003a: 50–51) mentions the possibility that the omission of the acc-marker may be the result of a defective transcription: the transcriber may simply not have heard the particle *i*, especially after words ending in *i*. However, this does not explain why omission of *i* is common in VO-clauses, but rare in VAOclauses (apart from the well-defined contexts described above). The difference is especially telling in older texts. Even though these were transcribed neither by professional linguists nor by native speakers, in VAO clauses only 3 out of 59 "markable" direct objects lack the acc marker, and all of these concern a nominalised verb. We may conclude that the omission of the acc marker cannot be attributed to defective transcription.

<sup>31</sup> Notice that not all common collocations allow omission of the acc marker. For example, *haka te ꞌe i te kōkoma* 'to remove the intestines, to gut' (a common step in food preparation) occurs 13x with acc marker, 1x without.

8.4 Marking of O

(44) *He* ntr *tuꞌu,* arrive *he* ntr *hoa* throw *te* art *ꞌaka* anchor *o* of *te* art *miro.* ship 'They arrived and lowered the anchor of the ship.' [Egt-02.099]

More generally, the accusative marker is frequently omitted when the object is highly predictable. For example, *ao* 'serve food' in (45) is naturally used with food as object, and *haka hū* 'to light, kindle' in (46) has either a fire or an engine as direct object. With both verbs, the accusative marker tends to be omitted:


In other cases it is less clear why the accusative marker is omitted; the only thing that can be said is, that all of these involve a non-human object. Two examples:


We may tentatively conclude that the accusative marker can be left out when the object is non-human and non-salient, <sup>32</sup> in clauses where the subject is not expressed.<sup>33</sup> This tendency may have weakened over time: new texts show fewer examples of unmarked objects than old texts.

Finally, some cases of unmarked objects are best explained as passive constructions. These are discussed in §8.5 below.

<sup>32</sup> "Non-salient" means that the importance of the acc is downplayed. It does not necessarily mean that the object is nonthematic, i.e. does not play a significant role in the larger discourse. Weber (2003a: 50) suggests that in some cases the acc marker may have been omitted because the acc is indeterminate or non-referential. This may explain some cases; however, three of her examples involve the verbs *ŋaro ꞌa* and *take ꞌa*, which allow omission of the acc marker anyway.

<sup>33</sup> Notice that this is somewhat the opposite of the conditions on the use of the agentive marker *e*: *e* is obligatory in transitive clauses when the object is not expressed, and optional when the object is expressed.

### **8.4.2 Conclusion: Rapa Nui is an accusative language**

The preceding sections have shown that Rapa Nui is an accusative language: S and A arguments have identical marking (either Ø or *e*), while O is marked differently (either *i* or Ø). S and A together can be called subject, while O is object.

Case marking of S, A and O is governed by the following rules:

	- a) in VOA and OVA clauses;
	- b) in VA clauses without explicit O;
	- c) with the verbs *rovaꞌa* 'obtain'*, takeꞌa* 'see' and *ŋaroꞌa* 'hear, perceive';
	- d) optionally with any agentive verb (whether transitive or intransitive), to signal a high degree of agentivity.
	- a) in the imperative, if the subject is not expressed;
	- b) usually when the object of *oti* 'to finish' is a nominalised verb;
	- c) often with the verbs *rovaꞌa* 'to obtain'*, take ꞌa* 'to sea' and *ŋaroꞌa* 'to hear';
	- d) in VO clauses, when the object is non-salient.

### **8.5 The passive**

### **8.5.1 Passivisation in Rapa Nui**

In the previous sections, verb arguments have been referred to by their semantic roles, not by their syntactic role. Now in transitive clauses, the A argument (whether *e*-marked or unmarked) is often the subject of the clause, while Patient (whether *i*-marked or unmarked) is object. Alexander (1981a: 136–137) gives arguments to consider the *e*-marked noun phrase as the subject of the clause. For example, when two coordinated clauses have the same subject, one of these may be deleted under Equi-NP-deletion, even when the other is *e*-marked. In the following example, the deletion of the subject in the first clause indicates that *e tōꞌoku pāpā era* is the subject of the second clause.

8.5 The passive

(49) *Ko* prf *oho* go *mai* hither *ꞌā* cont **Ø** *ko* prf *ꞌaꞌaru* grab *ꞌā* cont *i* acc *tōꞌoku* poss.1sg.o *pūꞌoko* head *e* ag *tōꞌoku* poss.1sg.o *pāpā* father *era.* dist

'My father came and touched my head.' (Alexander 1981a: 137; spelling corrected)

In other words, constructions with an *e*-marked Agent are accusative constructions, in which the Agent is subject and the Patient is object.

Other EP languages (such as Tahitian and Māori) have a passive construction, in which not A, but O is syntactically the subject of the clause. These constructions involve a fourfold transformation:<sup>34</sup>


Now Rapa Nui does not have the passive suffix (criterion 3); moreover, the accusative marker is omitted under certain conditions anyway (1), the Agent is *e*-marked under certain conditions (2), and constituent order is relatively free (4). It may thus seem that Rapa Nui cannot have a passive; or if it has a passive, it would be impossible to detect, as all three possible criteria already apply otherwise. It is therefore not surprising that Chapin (1978: 167) denies the existence of a passive in Rapa Nui.

However, Alexander (1981a) and Weber (2003a) show that it is possible to distinguish a passive in Rapa Nui. The passive occurs in clauses such as the following:


In these examples, the Patient is unmarked (criterion 1), the Agent is marked with *e* (2), and the constituent order is VOA (4). More importantly, Weber (2003a: 56–58) argues that in such constructions, the Patient is subject of the clause. This is demonstrated by two phenomena:

Eqi-NP-deletion. In a subordinate clause introduced by *mo* 'in order to', the subject can be omitted if it is coreferential with the subject of the matrix clause. In other words,

<sup>34</sup> See e.g. Weber (2003a: 53); Clark (1976: 67); Harlow (2007a: 171); Harlow (2007b: 90).

if a noun phrase in such a clause is omitted, it must be the subject. The fact that in the following example the Patient is omitted under coreferentiality with the matrix clause subject, shows that the Patient is the subject, while the Agent phrase is an oblique:

(52) *He* ntr *haŋa* want *a* prop *au*<sup>i</sup> 1sg [*mo* for *hoŋi* kiss Ø<sup>i</sup> *e* ag *te* art *poki*]*.* child 'I want to be kissed by the child.' (Weber 2003a: 56, adapted from Alexander 1981a: 134)

This argument may not be as strong as it seems, though, as it is not certain that deletion in these clauses only operates on subjects. Patients (i.e. direct objects) are freely omitted in Rapa Nui, both in main clauses and in *mo*-clauses, without any evidence of passivisation. See for example (53), where the Agent is expressed as a possessor (as is usual in *mo*-clauses, §11.5.1.2), while in (54) the Agent is not expressed at all. In both cases the Patient is left unexpressed under coreferentiality with a constituent of the matrix clause, even though there is no evidence that the Patient is subject of the *mo*-clause.


'He passed by Mahatu, (looking) for palm leaves to break.' [R304.111]

Subject raising. The evidence from subject raising is more compelling. In many Polynesian languages, there is a rule which raises the subject of a subordinate clause to the subject position of the matrix clause. The constructions in which this rule operates, vary per language (see Chung 1978: 132; Weber 2003a: 57); in Rapa Nui, subjects are raised after the aspectual verbs *oti* 'finish' and *haꞌamata* 'begin'.<sup>35</sup>

Now it turns out that when the complement clause contains a transitive verb, the Patient of this verb can be raised to the subject position of the matrix clause. This suggests that the embedded clause is a passive construction with the Patient as subject. This suggestion is reinforced by the fact that the Agent in these constructions, if expressed, is always *e*-marked. Here is an example for each construction:

(55) *Ku* prf *oti* finish *ꞌā* cont *te* art *nua*<sup>i</sup> cape [*te* art *kaui* sew Ø<sup>i</sup> *e* ag *Nune* Nune *rāua* 3pl *ko* prom *te* art *vārua* ]*.* spirit 'Nune had finished sewing the cape (or: the cape had finished being sewn by Nune), together with the spirit.' [Mtx-7-09.051]

<sup>35</sup> See sec. §11.3.2.2 and §11.3.2.1, respectively. Weber mentions *ha ꞌamata*, not *oti*; in addition, she mentions raising of the subject after the negation *ꞌina*; however, this argument depends on the analysis of *ꞌina* as a matrix predicate, an analysis not adopted in this grammar (§10.5.2).

8.5 The passive

(56) *He* ntr *haꞌamata* begin *te* art *hoi*<sup>i</sup> horse [*he* ntr *puꞌapuꞌa* beat:red Ø<sup>i</sup> *e* ag *ꞌOrohe* ]*.* Orohe 'The horse started to be whipped by Orohe.' (Weber 2003a: 58)

We may conclude that Rapa Nui has a true passive construction, in which the unmarked Patient is subject, while the *e*-marked Agent is oblique.<sup>36</sup>

Pragmatically, passives are characterised by the fact that the Patient is the topic of the clause (Keenan & Dryer 2007: 326). The passive construction enables the Patient to function as subject. This is clear in the following sentence, which is part of a story about the arrival of the first airplane on Rapa Nui; the airplane is topical:

(57) *Ko* prf *puru* close *tahi* all *ꞌana* cont *tū* dem *ꞌavione* airplane *era* dist *e* ag *te* art *viꞌe,* woman *e* ag *te* art *taŋata,* man *e* ag *te* art *ŋā* pl *poki* child *hare* house *hāpī.* learn 'The plane was completely surrounded by women, men, and school children.' [R379.012]

One situation in which the Patient tends to be topical, is when the Agent is nonanimate and the Patient is animate. Animate entities tend to be more topical in discourse than non-animate entities; the passive construction may be used to reflect this syntactically. This leads to constructions such as the following, in which a non-animate Agent (more precisely: Force) is *e*-marked:

(58) *A* prop *totoru* red:three *ko* prf *haka* caus *vari* pass *tahi* all *ꞌana* cont *e* ag *te* art *pūai* power *ꞌanaꞌana* splendour *o* of *te* art *ꞌAtua.* God 'The three were completely surrounded by the glorious power of God.' [Luke 9:31]

In all languages that have passives, the Agent of a passive construction can be omitted (Keenan & Dryer 2007: 329). In Rapa Nui, agentless passives can be detected in Patient raising constructions: (59) and (60) are agentless counterparts of (55) and (56), respectively.

(59) *Ki* when *oti* finish *hoꞌi* indeed *te* art *tāua* 1du.incl *kāpē*<sup>i</sup> coffee [*i* acc *te* art *unu* drink Ø<sup>i</sup> ]*…* 'When we have finished our coffee… (lit. when our coffee has finished being drunk).' [R301.043]

<sup>36</sup> Because *e* marks both Agent phrases that are subject and Agent phrases with oblique status, Weber (2003a: 60) distinguish two different particles *e*: a nominative particle, marking subjects (in Weber's view not necessarily agentive in active clauses, see sec. §8.3.3), and an agentive particle, marking oblique Agent noun phrases in passive clauses. In my analysis *e* is treated as a single particle, which always marks Agent noun phrases, whether in subject position or oblique. As the discussion in this section will show, it is not always possible to determine whether a clause is active or passive.

(60) *ꞌAi* there *hoꞌi* indeed *te* art *taŋata*<sup>i</sup> man *e* ipfv *haꞌamata* begin *era* dist [*e* ipfv *tari* transport *era* dist Ø<sup>i</sup> *ki* to *ruŋa* above *i* at *te* art *pahī*]*.* ship

'Then the people started to be transported aboard the ship.' [R210.040]

We would expect agentless passives to occur in simple clauses as well; however, these are harder to detect. An agentless simple passive clause will be a Verb–Patient clause with unmarked Patient, but there are no syntactic criteria to tell whether such a construction is active (Verb–ObjectPatient) or passive (Verb–SubjectPatient): VO-clauses with an unmarked object are not uncommon (§8.4.1).

There are semantic/pragmatic clues, however. A possible indication is, whether or not the clause has an implied Agent. When the Agent is left out in active clauses, this is usually because it is already known; it is coreferential with a noun phrase in a previous clause. In (48) above, here repeated, the identity of the Agent is known, so we may presume that the sentence is active. Moreover, the Agent is topical, therefore likely to be the subject. In other words, this example is an active clause with implied Agent:

(61) *He* ntr *toꞌo* take *mai,* hither *he* ntr *haka pā* fold *te* art *kūpeŋa.* net 'He took the net and (he) folded it.' [Mtx-3-01.171]

With agentless passive clauses, the Agent is not known from the preceding context; the identity of the Agent may simply be irrelevant. In the following example, the identity of the Agent is unknown, so a passive interpretation with the Patient as subject is plausible:

(62) *Hora* hour *iva* nine *ko* prf *tari* transport *ꞌā* cont *te* art *taŋata* man *ki* to *ruŋa* above *ki* to *te* art *pahī.* ship 'At nine o'clock the people are transported aboard the ship.' [R210.037]

Besides this semantic criterion, there are also syntactic clues for passivity. Crosslinguistically, there is a correlation between passive voice and perfect aspect.<sup>37</sup> The perfect aspect focuses on the state resulting from the action, rather than the action itself; similarly, the passive tends to focus on the result of the activity and its effect on the patient. As it happens, quite a few examples of the passive in Rapa Nui are in the perfect aspect, like (50), (51) and (62) above.

Finally, a syntactic indication for passivisation in (62) is the fact that the Patient does not have accusative marking. In general, the accusative marker is obligatory in VOclauses with human Patient.

<sup>37</sup> See e.g. Comrie (1976: 84); Foley (2007: 382); Keenan & Dryer (2007: 340); Dixon (2012: 219). Cf. Milner (1973), who argues that the difference between suffixed and unsuffixed verbs in Samoan (a distinction usually described as active versus passive) has to do with aspect, not voice.

8.5 The passive

In conclusion: there are several clues for passivisation:


Simple VO clauses may be either active or passive. Use of the perfect aspect may be an indication of passivity, but often only the context will tell whether a clause is active or passive. In the first case, the Agent is implied from the context; in the second case, the Agent is unspecified and irrelevant.

### **8.5.2 The pseudopassive**

A few intransitive verbs of motion or position (*uru* and *oꞌo* 'to enter', *eke* 'to mount, embark, climb, fig. to dominate', *noho* 'to sit, stay') exhibit a process very similar to passivisation. These verbs normally take an Agent subject, as well as an optional oblique constituent expressing the target of movement or position:

(63) *He* ntr *eke* go\_up *a* prop *Korikē* Korike *ki* to *ruŋa* above *i* at *te* art *hoi.* horse. 'Korike mounts the horse.' [R616.059]

But there are also examples where the locative constituent becomes the subject and is unmarked or left unexpressed, while the Agent is expressed as an oblique *e*-marked phrase.<sup>38</sup> This construction can be characterised as a pseudopassive: the Agent is expressed as oblique, but unlike the regular passive, it is a locative phrase rather than a Patient which becomes the subject. A few examples:


<sup>38</sup> Hooper (1984b: 40) points out that in Māori, verbs of motion and posture can freely occur in the passive.

Like the regular passive, the pseudopassive tends to be used when the Patient is more topical than the Agent.

### **8.5.3 Two other uses of** *e*

Apart from its use in the passive, *e* also functions as an oblique marker in two other situations.

**8.5.3.1** The verb *ꞌī* 'to be full' has two possible argument structures: the subject either expresses the filled entity (the Container) as in (67), or the filling entity (the Substance) as in (68).<sup>39</sup> When the Substance is subject, the Container may be expressed as a locative phrase (*ꞌi rote vai* in (69)).


Now when the Container is subject, the Substance (whether animate or inanimate) may be expressed with an *e*-marked phrase. In the following example, this happens twice:

(69) *Hai* ins *oho* go *iŋa* nmlz *nei* prox *ko* prf *ꞌī* full *ꞌā* cont *te* art *motu* island *nei* prox *e* ag *te* art *iŋoiŋo.* dirty *ꞌE* and *te* art *vai,* water *ko* prf *ꞌī* full *ꞌā* cont *e* ag *te* art *meꞌe* thing *ꞌiꞌino.* pl:bad

'When this happens, this island will be full of pollution. And the water will be full of bad things.' [R649.119]

(69) can be considered as a kind of passivisation of the construction in (68): the Substance, in (68) expressed as subject, is demoted to an *e*-marked oblique noun phrase, while the Container becomes subject. The difference with regular passivisation is, that the Container subject is not the original direct object: in construction (68), the Container can only be expressed as an oblique, not as direct object. In this respect, (69) is very similar to the pseudopassive construction discussed in the previous section; the difference is, that unlike the pseudopassive examples, the *e*-marked noun phrase does not have an agentive role.

<sup>39</sup> While this alternating argument structure is not uncommon for verbs meaning 'full' in Polynesian (Ross Clark, p.c.), in Rapa Nui it represents an independent development: *ꞌī* was borrowed from Tahitian, where the Container is always subject and the Substance is marked with the multifunctional preposition *i*.

8.6 Non-standard verbal clauses

**8.5.3.2** *E* occasionally marks a noun phrase indicating something potentially harmful. I have found this mainly in the Bible translation with the verbs *hāpaꞌo* 'to take care of' and *uꞌi* 'to watch, look', which can both be used in the sense 'to watch out for, to be on one's guard against'.<sup>40</sup> However, (72) shows that this use of *e* is also found in other contexts.


### **8.6 Non-standard verbal clauses**

Under this heading, clauses are discussed which have a non-standard constituent order, non-canonical marking of arguments, or both. §8.6.1 deals with clauses involving a non-standard constituent order, in which the arguments still have their usual markers (Ø or *e* for the subject, *i* or Ø for the direct object). §8.6.2 discusses topicalisation, in which a preposed subject has a special marker (either *ko* or *he*). §8.6.3 deals with the actor-emphatic construction, which serves to put an Agent in focus. Other constructions involving non-standard marking of arguments are discussed in §8.6.4.

### **8.6.1 Marked constituent orders**

As discussed in §8.1, the default constituent order is VS/VAO, but all other possible orders occur in varying proportions. In this section, different constructions are discussed involving constituent orders other than VS/VAO.

#### **8.6.1.1 Preverbal subjects**

Subjects are often placed before the verb. In certain situations, this is syntactically conditioned.

<sup>40</sup> The complement of these verbs (the negative thing one should watch out for) can also be introduced by *mai* 'from', or as a clause introduced by *ꞌo* 'lest'.

**8.6.1.1.1** Preverbal subjects are common after various clause-initial elements (obligatorily after 1-2, optionally after 3-4):<sup>41</sup>


In fact, there is a general tendency for the subject to be preverbal after any oblique initial constituent, e.g. a prepositional phrase as in (73), or an adverb as in (74):<sup>43</sup>


(i) *ꞌĪ* imm [*te* art *vai*]<sup>i</sup> water [*ko* prf *ha ꞌamata* begin *ꞌana* cont t<sup>i</sup> [*ko* prf *o ꞌo* enter *mai* hither *ꞌā* cont t<sup>i</sup> ] ]*…* 'Immediately the water started to enter (the ship)…' (R210.162)

<sup>41</sup> Subjects can also be raised from the complement of the aspectual verbs *oti* 'finish' and *ha ꞌamata* 'begin' (§11.3.2), but as this places them in the postverbal subject position of the aspectual verbs, this in itself does not result in S V order. However, given the right context, the raised subject can be raised once more to a position before the aspectual verb. In the following example, the original and intermediate position of the subject is indicated by traces t*i*:

<sup>42</sup> It is interesting to note that certain clause-initial elements trigger a number of phenomena that make the clause differ from a standard main clause:

<sup>1.</sup> The subject tends to be preverbal.

<sup>2.</sup> After many of these elements, the aspectual *he* is avoided in favour of *i* or *e* (§7.2.8), a pattern characteristic of subordinate clauses.

<sup>3.</sup> In some cases, the constituent/subordinate negator *ta ꞌe* is used (§10.5.6.6) rather than the main clause negators *ꞌina*, *kai* and *e ko*.

We may conclude that the preposed constituent takes on some characteristics of a predicate, followed by the subject + the rest of the clause as a subordinate clause.

Interestingly, the negator *ꞌina*, for which predicate status has sometimes been argued, is less predicatelike than initial locative and interrogative phrases: while the latter tend to trigger the use of *i* rather than *he*, this is not true for *ꞌina* (§10.5.1).

<sup>43</sup> The same tendency exists in Māori, see Harlow (2007b: 96).

8.6 Non-standard verbal clauses

**8.6.1.1.2** Apart from these syntactically conditioned environments, subjects may be placed before the verb for pragmatic reasons. The frequency with which this happens depends on the speaker, and it is hard to pin down the exact conditions under which this is done (cf. Dryer 2007c: 77). A few generalisations can be made, though.

The preposed subject is often a highlighted topic: preposing the subject signals that the clause is about the entity referred to by the subject.<sup>44</sup> Usually subject shift is involved: the subject is different from the subject of the preceding context. Appropriate paraphrases are 'As for X…' or 'Concerning X…'.


This does not mean that every subject shift is marked by a preposed subject. A subject which is already thematic in the story (or in the current episode of the story) usually occurs in the default postverbal position, even when it is different from the subject of the preceding clause or sentence. In fact, most explicit subjects in discourse – whether pre- or postverbal – involve subject shift, as the subject is usually not expressed when it is identical to the subject of the preceding clause. Subjects are preposed especially when they are not thematic in the wider context, but are the topic of a single sentence or clause. An example:

(77) *Te* art *ŋā* pl *viꞌe* woman *e* ipfv *uruuru* dress:red *rō* emph *ꞌā* cont *i* acc *te* art *kahu* clothes *kākaka,* banana\_leaf *ꞌe* and *te* art *ŋāŋata* men *he* pred *piripō* trouser *tetea* pl:white *he* pred *kamita* shirt *pāreu.* printed\_cloth

'The women wore banana leaf dresses, and the men (wore) white trousers and

coloured shirts.' [R210.132]

Preposed subjects may also mark the start of a discourse or a new episode in the discourse. In the following example the subject *Taparahi* is identical to the subject of the preceding clauses; no subject shift is involved. Even so, the subject is preposed, indicating that the story moves on to a new topic.

(78) *A* prop *Taparahi* Taparahi *e* ipfv *turu* go\_down *era* dist *ki* to *te* art *hāpī,* learn *kona* place *kē* different *e* ipfv *oho* go *era.* dist 'Taparahi, when he went to school, he would go somewhere else.' [R250.033]

<sup>44</sup> Cf. Lambrecht (1994: 131): "A referent is interpreted as the topic of a proposition if in a given situation the proposition is construed as being about this referent".

To summarise: subjects are preposed


#### **8.6.1.2 Preverbal objects**

Just like subjects, direct objects may also be placed before the verb, though this is relatively rare (see Table 8.1 in §8.1). The direct object is preverbal when it is highlighted as topic, often in combination with subject shift with respect to the preceding clause. When the subject is also expressed, the constituent order is usually OVS. As the subject in OVS-clauses is always *e*-marked (§8.3.1) and the preverbal object is unmarked (§8.4.1), these constructions may also be analysed as passives, in which the fronted Patient is actually the subject.

In (79) below, the Patient *au* is topical in the context (the speaker is talking about himself). Example (80) marks the start of a new section in a story, with a shift to a new topic; this topic is the object of the clause, hence it is fronted. (81) is the start of a direct speech, in which the Patient *a koe* is clearly topical.


#### **8.6.1.3 Topic-comment constructions**

In the examples in the previous section, the preposed constituent is subject or object of the clause. Rapa Nui also has a topic-comment construction, in which a topic noun phrase is followed by a complete clause providing information about this topic. The topic NP is left-dislocated: it is not part of the following clause and does not necessarily have a semantic role in relation to the predicate of the clause. The topic may be coreferential to an argument of the verb (as in (82) below, where it is coreferential to the A of *aŋa*), but it is not a verb argument itself; the comment is a complete clause in its own right. Below are a few examples.

8.6 Non-standard verbal clauses


Topic-comment constructions are also found in possessive clauses (§9.3.3).

### **8.6.2 Topicalisation**

As discussed in §8.6.1.1, the subject of a verbal clause may be preposed without special marking. Preposed subjects may also be marked with *ko*, or (occasionally) *he*. These are discussed in the following sections.

#### **8.6.2.1 Topicalisation with** *ko*

Preverbal subjects marked with *ko* are topicalised: they are highlighted as the topic of the sentence or of a longer stretch of discourse. Comparison of preposed subjects with and without *ko* suggests, that topicalisation with *ko* signals that the subject is prominent in some way.<sup>45</sup>

There are various reasons why the topic of the clause may be prominent. Sometimes the participant referred to is contrasted with other participants:

(85) *Ko* prom *nua* Mum *he* ntr *toꞌo* take *i* acc *te* art *huri* banana\_shoot *mo* for *hao…* plant *Ko* prom *Tiare* Tiare *i* pfv *maꞌu* carry *i* acc *te* art *karu* seed *mautini* pumpkin *mo* for *ꞌoka.* plant

'Mum takes the banana shoots to plant… Tiare has brought the pumpkin seeds to plant.' [R184.055–056]

(86) *Ko* prom *tū* dem *hoi* horse *era* dist *i* pfv *eꞌa* go\_up *hakaꞌou* again *ki* to *ruŋa.* above *Ko* prom *tū* dem *poki* child *era* dist *i* pfv *hiŋa* fall *era ki raro he rerehu rō atu ꞌai.*

dist to below pred faint emph away subs

'(The horse and the boy fell.) The horse stood up again. The boy who had fallen down fainted.' [R399.025–026]

<sup>45</sup> Cf. the definition of prominence by Callow (1974: 50): prominence is "any device whatever which gives certain events, participants, or objects more significance than others in the same context".

The noun phrase may also be thematic over a longer stretch of speech: it is the theme of the section that follows. A few examples will illustrate this. In the following sentence, Anisia and Marina have been mentioned earlier in the text, but not very recently. They are now re-introduced as the theme of conversation of a new section: 'As for Anisia and Marina…'

(87) *ꞌĒ,* thus *ko* prom *Anisia* Anisia *ararua* the\_two *ko* prom *Marina* Marina *ꞌi* at *te* art *hare* house *hāpī* learn *era* dist *ꞌā* ident *o* of *rāua* 3pl *e* ipfv *kai* eat *era.* dist

'Anisia and Marina eat in their school.' [R103.191]

*Ko*-marking thus signals a shift to a new theme. This also happens in the following example. The preceding context is about a group of people; the sentence quoted here starts a new section, in which one of the group, Artillero, is the sole participant. To signal the switch to Artillero as theme, the subject is preposed and preceded by *ko*.

(88) *Ko* prom *Artillero* Artillero *i* pfv *hoki* return *i* prom *iri* ascend *ki* to *tōꞌona* poss.3sg.o *kona* place *hare* house *era.* dist 'As for Artillero, he returned to his house.' [R437.055]

Topicalisation tends to happen especially with proper nouns and pronouns; however, it also occurs with common nouns, as (86) above shows. What they all have in common, is that they bring a participant to the front which has been introduced earlier and is known to the hearers.<sup>46</sup> The participant in question is highlighted as the topic of a clause, sentence, or longer stretch of discourse.

*Ko* also serves to mark preverbal direct objects. Just as with subjects, the prominence marker *ko* gives prominence to the preposed constituent, signalling that it is thematic in discourse.

(89) *Ko* prom *te* art *pāherahera* sport *ena* med *e* ipfv *kī* say *nei* prox *ꞌi* at *ꞌaŋarīnā* today.past *he* pred *haka* caus *nini* slide *ꞌi* at *ruŋa* above *i* at *te* art *vave.* wave '(Surfing is an old sport of our ancestors.) Today this sport is called 'sliding on the waves'.' [R645.003]

<sup>46</sup> There are a few interesting exceptions, where topicalisation is used right at the start of a story. The following sentence is an example:

<sup>(</sup>i) *Ko* prom *ꞌOrohe* Orohe *e* num *tahi* one *mahana* day *he* ntr *e ꞌa* go\_out *ꞌi* at *te* art *pō ꞌā* morning *ararua* the\_two *ko* prom *koro.* Dad 'Orohe went out one day in the morning with Dad.' (R154.001)

In such cases, the participant may still have been accessible in the original context of the story; for example, someone may have asked the speaker to tell about such and so. (This possibility was pointed out to me by Stephen Levinsohn.) In fact, R154 is part of a collection or stories, and Orohe has figured in the preceding stories as well – possibly the story should be considered as an episode in an ongoing narrative.

8.6 Non-standard verbal clauses

*Ko* may mark the left-dislocated noun phrase of topic-comment constructions (see §8.6.1.3).

(90) *Ko* prom *te* art *mākini* machine *era,* dist *e* ipfv *haro* pull *mai* hither *e* num *tahi* one *meꞌe* thing *a* by *te* art *muꞌa….* front *he* ntr *roaroa* long:red *mai* hither *mo* for *puē* can *mo* for *toꞌo* take *o* of *te* art *hohoꞌa.* picture '(The captain showed up, carrying a camera.) This device, you had to pull something at the front to take a picture.' [R379.027]

Time adjuncts may also be introduced by *ko*.


These time phrases serve as points of departure (Levinsohn 2007: 39), signalling the shift to another time at the start of a new episode in the story.

Clark (1976: 37) discusses initial *ko*-marked subjects in Proto-Polynesian and analyses these as cleft constructions: the *ko*-marked noun phrase is the predicate of a nominal clause, the subject of which is a relative clause with an empty head:

(93) [ *ko N* ]Pred [ Ø [ *A/M V* …. ]Rel ]<sup>S</sup>

This analysis is suggested by the formal similarities between verbal clauses with topicalisation and nominal clauses with a *ko*-marked predicate (§9.2.2), and enables a unified account of both.<sup>47</sup>

For Rapa Nui however, there is little ground for taking the topicalised constructions under discussion as clefts. Constructions which are unambiguously clefts in Rapa Nui always contain an anchor noun as head of the relative clause (§9.2.6), which suggests that headless relative clauses are not possible in Rapa Nui. Moreover, these clefts have quite a different function from the topicalisation constructions under discussion here: cleft constructions put the initial noun phrase in focus as new information, backgrounding the verb, while topicalisation signals that the initial noun phrase is thematic in discourse.

<sup>47</sup> Bauer (1991) applies the same analysis to Māori, arguing that topicalised *ko*-NPs are clefts in some cases, when the *ko*-NP is in focus and receives sentence stress.

Now it could be argued that Rapa Nui has two types of clefts with different functions:<sup>48</sup> a focus construction with head noun and a topicalisation construction without head noun. However, there are syntactic reasons not to analyse topicalising constructions as relative clauses. In relative clauses, the neutral aspect marker *he* is extremely rare (§11.4.3), while *he* is common in topicalisation constructions like (85) above.

Another characteristic of relative clauses in Rapa Nui is, that the verb tends to be followed by a postverbal demonstrative like *era*, whereas in many topicalisation constructions no PVD is used. And thirdly, while relative clauses often do not have an aspectual (§11.4.5), topicalisation clauses always contain an aspectual.

We may conclude that there are good reasons not to consider topicalisation constructions in Rapa Nui as clefts. Besides, even if such an analysis were adopted, it would not account for all occurrences of *ko* in verbal clauses; §8.6.4.5 discusses cases where the *ko*-marked noun phrase is even more clearly part of the main clause.

Finally, one more topicalisation construction deserves attention: occasionally a *ko*marked topicalised subject is followed by a nominalised verb:

(94) *ꞌO* because\_of *ira* ana *ko* prom *tū* dem *ŋā* pl *taina* sibling *rikiriki* little *nō* just *era* dist *ōꞌona* poss.3sg.o *te* art *oho* go *ki* to *te* art *hāpī.* school

'Therefore only her little brothers went to school.' [R441.011]

(95) *Te* art *taŋata* person *e* ipfv *mate* die *tahi* all *rō* emph *ꞌā;* cont *ko* prom *koe* 2sg *nō,* just *te* art *ora* live *te* art *oho.* go 'All people die, but you just keep on living.' [R445.016–017]

These constructions mostly have a habitual or continuous sense. They are very similar to the nominalised actor-emphatic construction (§8.6.3), which likewise has a preposed subject followed by a nominalised verb with habitual sense. They are also similar to constructions in which *ko* is followed by a nominalised verb (see (23–24) in §3.2.3.1).

#### **8.6.2.2 Topicalisation with** *he*

Occasionally, a preverbal subject is introduced by the nominal predicate marker *he*. This is somewhat surprising, as *he* normally introduces non-referential noun phrases and is limited to nominal predicates and other non-argument NPs (§5.3.4.1).

The sentence may state a general fact about a category as a whole as in (96), or refer to a specific entity or group as in (97):

<sup>48</sup> Clark (1976: 38) points out that in most Polynesian languages, clauses with a topicalised *ko*-marked NP are ambiguous: the *ko*-NP can be either predicate ('it was John who was chopping the yam') or topic ('as for John, he was chopping the yam'). Regardless the analysis of topicalised constructions, it is clear that the two are syntactically distinguished in Rapa Nui.

8.6 Non-standard verbal clauses

(96) *He* pred *nuꞌu* people *paꞌari* adult *ꞌina* neg *e* ipfv *ko* neg.ipfv *aŋa* do *tahaŋa* aimlessly *nō* just *i* at *a* prop *koe* 2sg *i* acc *te* art *aŋa* work *i* pfv *manaꞌu.* think

'Grown-ups don't simply do<sup>49</sup> the work they think of (i.e. without preparation).' [R363.145]

(97) *…ꞌe* and *he* pred *mataroa* sailor *repahoa* friend *o* of *koro* Dad *ko* prf *maꞌu* carry *mai* hither *ꞌā* cont *ka* cntg *rahi* many *atu* away *te* art *pahu* can *peti.* peach

'(When she entered there were many people,) and some sailors, friends of Dad, brought many cans of peaches.' [R210.125]

In both examples above, the *he-*marked noun phrase refers to an entity not mentioned before and therefore not yet known to the hearer. In other cases, the subject refers to someone or something whose existence the hearer may be presumed to know or infer from the context. The following examples illustrate this:


What all these examples have in common, is that the *he*-marked subject is singled out from among other entities; in other words, the *he-*construction serves as topicalisation. This is particularly clear in (98), where the subject *he taŋata* is contrasted with other actors in the discourse, and in (100), where the two babies introduced in the preceding context are mentioned individually.

<sup>49</sup> The phrase *i a koe* 'to/regarding you' seems to be s a second person of personal involvement (§4.2.4.2), involving the addressee in the discourse in some way.

In fact, clauses with *he*-marked subjects are very similar to clauses with topicalised *ko*-marked subjects (§8.6.2.1). *Ko*-marked topicalisations refer to individuated entities (a single referent or a clearly defined group) which are accessible to the hearer; in other words, the exact referent of the *ko*-marked subject has been introduced in the preceding context. By contrast, the *he*-marked subjects in the examples above are not accessible as individuated entities. Even though the hearer can infer their existence from the context, they have not been mentioned as such. In (98), the context tells about a group of people – men, women and children – who go on an outing; prior to the sentence quoted here, the men have not been mentioned separately. Similarly, in (100), the preceding sentences tell about the birth of two boys; the hearer can infer the existence of an oldest and a youngest boy, but it is only in the sentence quoted here that each boy is singled out. In both cases, the referent of the *he*-marked noun phrase is not accessible as such, as it has only been introduced as undefined part of a larger group. By contrast, subjects topicalised with *ko*-marking are always accessible as individual referents; for example in (85) above, both mother and Tiare play a role in the preceding context.

It is not very surprising that topicalisation with *ko* often involves a pronoun or proper noun, noun types typically associated with accessible referents.

The difference between *ko* and *he* in marking topicalised subjects is reminiscent of the use of *ko* and *he* with nominal predicates (§9.2.1): in both cases, *ko* marks an accessible, individuated entity, while in other cases *he* is used.

### **8.6.3 The actor-emphatic construction**

Many Polynesian languages have a construction commonly called the actor-emphatic (AE).<sup>50</sup> This construction is used when the Agent is in focus and shows the following characteristics:


<sup>50</sup> See Harlow (2001: 196–197); Harlow (2007a: 175–176); Elbert & Pukui (1979: 147–148); Lazard & Peltzer (2000: 62–63). More extensive treatments are found in Clark (1976: 111–123) for PPN; Waite (1990) and Bauer (2004) for Māori; Potsdam & Polinsky (2012) for Tahitian. The construction is mainly found in EP languages, but Wilson (2012: 315) gives an example from Luangiua and suggests that the AE construction is an innovation in the Northern Outlier + EP subgroup he proposes.

8.6 Non-standard verbal clauses

• The construction tends to be limited to transitive verbs, or to (transitive or intransitive) agentive verbs.

The actor-emphatic construction also occurs in Rapa Nui.<sup>51</sup> As in other languages, it serves to put the Agent in focus and to background the action; the action is often presupposed. Different from other languages, there are three varieties, which are discussed in turn below.

**Perfective** In the perfective AE, the Agent is a possessive pronoun or noun phrase. For pronouns, this is a Ø-possessive, i.e. a pronoun without determiner (§4.2.2.2). As pointed out in §6.3.2, singular pronouns show a distinction between *a*- and *o*-possessives: *ꞌāꞌaku* versus *ōꞌoku*. The same distinction is made with proper nouns: preposition *ꞌa* versus *o*. <sup>52</sup> In the AE construction, the *a*-possessive is used with singular pronouns and proper nouns. For plural pronouns and with common nouns, no *a*-forms exist, so the default *o* is used.

The verb in this construction is always marked with perfective *i*; the construction refers to actions prior to the time of reference, usually in the past. Two examples:


As these examples show, the Patient either follows or precedes the verb. When it follows the verb as in (101), it is *i*-marked; when it precedes the verb as in (102), it is unmarked. This corresponds to the general pattern of object marking in Rapa Nui (§8.4.1), so there are good reasons to consider the Patient as direct object in either position.

It is remarkable that the object of an AE construction is often preverbal, while preverbal objects in general are rare (see Table 8.2 in §8.1). This may have to do with the

(i) *A/ꞌA* prop/of.a *Kuha* Kuha *i* pfv *kī* say *mai* hither *ki* to *a* prop *au* 1sg *mo* for *iri* go\_up *mai* hither *ki* to *nei.* prox 'Kuha told me to come up here.' (R229.105)

<sup>51</sup> Actor-emphatic constructions are also common in questions; these are discussed in section §10.3.2.1 and §10.3.2.2.

<sup>52</sup> As initial glottals are not contrastive, perfective AE's with proper nouns (marked with *ꞌa*) are difficult to distinguish from a preposed topical subject construction with perfective aspect, in which the noun is marked with the proper article *a*. The following example is syntactically ambiguous:

In the context it is more likely that the subject is in focus ('It was Kuha who said…') than that it is topical ('As for Kuha, she said…'), so that an AE reading is plausible. On the other hand, topicalised subjects sometimes occur at the start of a direct speech without further apparent reason, which may be the case here.

pragmatic status of Agent and Patient. The AE construction is used when the Agent is in focus: the Agent is presented as new information, an appropriate paraphrase is 'It was X who…' or 'X was the one who…'. The rest of the sentence, including the Patient, is known information. The Patient will often be topical, and this may be the reason it is preverbal: there is a tendency in Rapa Nui to place topical constituents early in the clause (§8.5.1 on passives; §8.6.1.1 on preposed subjects and objects; cf. Potsdam & Polinsky (2012: 68) for a similar observation about Tahitian). Another example is the following:

(103) *O* of *te* art *rūhia* tourist *ia* then *te* art *hohoꞌa* image *nei* prox *i* pfv *toꞌo.* take '(It was) the tourists (who) took this photo.' [R415.735]

**Imperfective** In the imperfective AE, the verb has the ipfv marker *e* and the Agent is expressed as a benefactive pronoun or noun phrase (§4.2.3; §4.7.8). Again, *a*-forms are used when available: *māꞌa*- in singular pronouns, *mā* in front of proper nouns. Plural pronouns and common nouns, which lack *a*-forms, are marked with the default *mo*.

The imperfective AE refers to an action posterior to the time of reference. Usually this is the future, but as (106) shows, this is not necessarily so:


As in the perfective AE, postverbal objects have the acc marker, while preverbal objects are unmarked.

**Nominal** The third AE construction has a possessive Agent as in construction 1, but the verb is nominalised. This construction refers to habitual actions, regardless the time of action:

(107) *ꞌĀꞌana* poss.3sg.a *te* art *haka* caus *tere* run *i* acc *te* art *henua.* land

'He was the one who governed the country.' [R370.005]

8.6 Non-standard verbal clauses

(108) *O* of *te* art *ika* fish *nei* prox *te* art *kai* eat *i* acc *te* art *oꞌioꞌi* larva *o* of *te* art *naonao.* mosquito 'This fish eats the larvae of the mosquito.' [R535.110]

The nominal AE construction is almost identical to a proprietary clause (§9.4.2), which equally consists of a possessive constituent (the predicate) + a noun phrase (the subject):

(109) *ꞌĀꞌana* poss.3sg.a *hoꞌi* indeed *te* art *viꞌe* woman *era.* dist 'The woman is his.' [R416.1156]

The only difference is that the subject of the AE construction is a nominalised verb, which may have an object. As in other AE constructions, this object may be preverbal, in which case the acc marker is omitted:

(110) *ꞌĀꞌana* poss.3sg.a *a* prop *au* 1sg *te* art *hāpaꞌo* care\_for *mai* hither *te* art *hāŋai* feed *mai.* hither 'She was the one who took care of me and fed me.' [R310.480]

All examples of AE constructions so far involve a transitive verb. The AE also occurs with intransitive verbs, but only with agentive verbs, i.e. verbs with an Agent argument:


AE constructions are negated using the constituent negator *taꞌe*, which precedes the Agent:


Finally: AE constructions are also used in 'who'-questions (§10.3.2.1).

To summarise: Rapa Nui has two actor-emphatic constructions which differ in aspect. The perfective AE uses the Ø-possessive, while in other languages a *n*-possessive

is used.<sup>53</sup> As in other languages, the imperfective AE has a *ma*/*mo*-marked Agent. A third construction, not found in other languages, uses a nominalised verb and expresses habitual actions.

As in other languages, the object may either precede or follow the verb. No special rules are needed to account for object marking in AE constructions.

### **8.6.4 Other non-canonical arguments**

In the next subsections, constructions are discussed in which the S/A argument is marked differently from *e* or Ø, or in which O is marked differently from *i* or Ø. The last subsection (§8.6.4.7) discusses agentive phrases marked with *i*; these are not syntactic subjects or objects, but are discussed under this heading because agentive *i* is similar in function to the agentive subject marker *e*.

#### **8.6.4.1 Possessive S/A arguments**

Sometimes the S/A argument is expressed as a possessive, using the preposition *o* or a possessive pronoun of the *o*-class. In two contexts this is the normal marking: with nominalised verbs (§8.6.4.7), and in subordinate clauses introduced by *mo* (§11.5.1.2). However, possessive S/A arguments are found in main clauses as well. They may occur when the following two conditions are met:


Two examples:


The use of possessives to express arguments in main clauses is largely speaker-dependent: this construction is frequent in some texts, but absent in others. Though the precise conditions are not clear, possessive marking appears to be a device to demote a nonsalient Agent. The fact that this is only found with the neutral marker *he*, which is also the nominal predicate marker, suggests that these clauses have been nominalised: (115)

<sup>53</sup> On the relation between the Rapa Nui Ø-possessive and the *n*-possessive in other languages, see Footnote 9 on p. 290.

8.6 Non-standard verbal clauses

could tentatively be paraphrased as '(There was) the saying of that old woman…'. In that case, the verb is nominalised and *he* is the nominal predicate marker.<sup>54</sup>

#### **8.6.4.2 Middle verbs:** *ki***-marked objects**

Certain transitive verbs take *ki* as object marker rather than *i*. This happens with verbs of perception, knowledge, emotion, speech, as well as a few others. In Polynesian linguistics, these verbs are distinguished as a separate subclass, which has been labelled *middle verbs* or "experiential verbs".

Generally speaking, *ki*-marked objects are not affected by the action. Rather, they are the Goal of the event: the focus of attention, the person or thing at which a feeling is directed, the content of knowledge or the addressee of a speech.<sup>55</sup>

Some verbs always take a *ki*-marked object, while other verbs allow both *i* and *ki*. With some verbs there is a clear difference in function between *i* and *ki*-marked objects; in other cases the difference is less clear. In this section, different semantic classes of verbs taking *ki*-complements are discussed.

**Perception verbs** The controlled perception verbs<sup>56</sup> *uꞌi* 'look' and *hakaroŋo* 'listen' take either *i* or *ki*, though *i* is more common.<sup>57</sup>

Generally speaking, *ki* tends to be used with more intensive or purposeful actions. *Uꞌi i* and *uꞌi ki* both mean 'to look at, to watch', but *uꞌi ki* may indicate a more focused attention as in (118), or is used in the sense 'to look for, to search', as in (119):


'She looked to the houses where the smoke of the earth oven rose (in order to snatch the food as soon as it was cooked).' [R368.004]

(119) *He* ntr *uꞌi* look *a* by *roto* inside *i* at *te* art *vai* water *ki* to *tū* dem *ika* fish *era,* dist *ꞌina* neg *kai* neg.pfv *takeꞌa.* see

'He looked for that fish in the water, but did not see it.' [R301.232]

<sup>54</sup> The directional *atu* in (116) may suggest that the phrase is still a verb phrase, but notice that *atu* occasionally occurs with nominalised verbs (§3.2.3.3).

<sup>55</sup> Notice that *ki* also expresses the Goal of motion, as well as the Recipient of an act of giving.

<sup>56</sup> The object of uncontrolled perception verbs either takes *i* or zero marking (§8.4.1; see also §7.5.2 on the difference between active and passive perception).

<sup>57</sup> *i* may have become more popular over time, as the following rough count suggests: in old texts, *uꞌi (mai/atu) (ena/era)* is followed 51x by *i*, 28x by *ki* (proportion *i/ki* roughly 2:1); in new texts, it is followed 152x by *i*, 34x by *ki* (proportion *i/ki* roughly 5:1).

*Hakaroŋo i* means 'to hear' or 'to listen'. *Hakaroŋo ki* likewise means 'to listen', but is also used in a more intensive sense: 'to pay attention' or 'to obey', as in (121).


**Emotive verbs** With verbs expressing emotion (feeling, attitude), object marking depends on the verb.

The object of *haŋa* 'love, like, want' is always marked with *ki*, never with *i*:

(122) *Ko* prf *haŋa* love *ꞌā* cont *a* prop *au* 1sg *ki* to *tōꞌoku* poss.1sg.o *koro.* Dad 'I love my Dad.' [R210.115]

With *ꞌaroha*, there is a clear semantic difference: *ꞌaroha i* means 'to pity', while *ꞌaroha ki* means 'to greet'.

*Riri* 'to be angry' can be followed by *i* or *ki*, without a clear difference in meaning:


Other emotive verbs taking a *ki*-marked object are e.g. *koromaki* 'to miss, long for', *manava mate* 'to be in love with'.

**Cognitive verbs** *Manaꞌu* 'to think' and *māhani* 'to be accustomed to, acquainted with' take either *ki* or *i*, without a clear difference in meaning. There may be a tendency for *ki* to be used with human objects and *i* with non-human objects, but see the following examples, which both have a non-human object:

(125) *E* ipfv *manaꞌu* think *nō* just *ꞌā* cont *a* prop *Te* Te *Manu* Manu *i* acc *tou* dem *meꞌe* thing *taꞌatoꞌa* all *era.* dist 'Te Manu thought about all those things.' [R245.011]

8.6 Non-standard verbal clauses

(126) *He* ntr *manaꞌu* think *ki* to *te* art *hora* time *era* dist *ōꞌona* poss.3sg.o *e* ipfv *noho* stay *era* dist *ꞌi* at *muri* near *i* at *tōꞌona* poss.3sg.o *nua* Mum *era.* dist

'He thought about the time when he lived with his mother.' [R245.003]

*ꞌIte* 'to know', on the other hand, always takes an *i*-marked object.

**Speech verbs** Two types of speech verbs should be distinguished.

'say'-type verbs include *kī* 'say', *ꞌaꞌamu* 'tell'*, raŋi* 'call'*, pāhono* 'answer', *pure* 'pray'*, hāꞌaki* 'inform, make known'. These verbs are often followed by direct speech. Alternatively, they may have a direct object expressing the content of speech:<sup>58</sup>

(127) *He* ntr *kī* say *ki* to *a* prop *Kava* Kava *i* acc *tū* dem *vānaŋa* word *kī* say *era* dist *e* ag *Pea.* Pea 'She said to Kava the words said by Pea.' [R229.075]

'talk'-type verbs include *hakameꞌemeꞌe* 'to mock', *ture* 'to scold' and *haꞌahanahana* 'to praise'. These verbs are usually not followed by direct speech<sup>59</sup> and do not take a direct object expressing the content of speech. The addressee may be expressed with a *ki*-marked noun phrase, but with some verbs *i* can be used as well. The latter would not be possible with a 'say'-type verb. Compare the two following examples:


**Other verbs** Various other verbs take either *ki* or *i*.

With *hāꞌūꞌū* 'help', the person helped is usually expressed with *ki* (though *i* is found as well), while *i* marks the activity. In (130), both are used together:

(i) *He* ntr *ture* scold *e* ag *nua* mother *ki* to *a* prop *Taparahi* Taparahi *he* ntr *kī…* say 'Mother scolded Taparahi and said…' (R250.018)

<sup>58</sup> As the example shows, the addressee of these verbs may be expressed by a *ki*-marked noun phrase, but this noun phrase is not the direct object.

<sup>59</sup> The following example shows that the speaker felt a second speech verb was needed to introduce the direct speech giving the content of the scolding:

(130) *Ko* prom *Tiare* Tiare *i* pfv *oho* go *i* pfv *hāꞌūꞌū* help *ki* to *a* prop *nua* Mum *i* acc *tāꞌana* poss.3sg.a *aŋa.* work 'Tiare went and helped Mum with her work.' [R334.125]

*Kimi* 'search' is marked with either *ki* or *i*, though *i* is more common. *Tiaki* 'wait' is usually marked with *ki*; *i* is used when the verb has the sense 'to guard':

(131) *Ko* prom *au* 1sg *mo* for *noho* stay *mo* for *tiaki* guard *i* acc *te* art *tātou* 1pl.incl *hare.* house 'I will stay (home) to guard our house.' [R399.130]

*Ki* **+ NP is direct object** While *ki* often marks an oblique constituent (e.g. a Recipient, or the Goal of motion), there are several indications that the *ki*-marked object of middle verbs is the direct object of the clause.

Firstly, the *ki*-marked constituent can be relativised in the same way as object noun phrases (§11.4.2.2): the constituent is not expressed in the relative clause and the subject is *e*-marked:

(132) *…Tahiti,* Tahiti *henua*<sup>i</sup> land [*haŋa* love Ø<sup>i</sup> *e* ag *te* art *taꞌatoꞌa* ]*.* all '…Tahiti, the island loved by all.' [R303.019]

By contrast, other constituents marked with *ki* use a different relativising strategy (§11.4.2.3).

Secondly, *ki*-marked constituents can be passivised. In the following example the Goal of *haŋa* is not expressed, but the fact that (a) it is topical in the context, and (b) the Agent is *e*-marked, suggests that it is the implicit subject of the clause.

(133) *E* ipfv *haŋa* love *rahi* much *rō* emph *ꞌā* cont *e* ag *te* art *ꞌAtua* God *ꞌe* and *e* ag *te* art *taŋata* man *taꞌatoꞌa.* all

'(Jesus grew up…) He was loved much by God and by all the people.' [Luke 2:52]

Thirdly, in causative constructions, the causee (the S/A of the root verb) is expressed as direct object (DO). However, when the root verb is transitive, its Patient is often expressed as DO, in which case the DO position is not available for the causee; in that case the causee is expressed as an oblique, introduced by *ki* (§8.12.3). In the following example, the verb *manaꞌu* (which may take a *ki*-marked object, see (126) above) is causativised. The causee *ki a koe* 'you' is expressed with *ki*, not *i*, which suggests that the DO position is already occupied by the noun phrase *ki tū vānaŋa..*.

	- *a* prop *au.* 1sg

'(We came) to remind you of the words you said to me.' [R229.207]

8.6 Non-standard verbal clauses

In other words, the presence of the oblique phrase *ki a koe* indicates that the other *ki*-marked noun phrase occupies the DO position.

#### **8.6.4.3 Patient marked as instrument**

Sometimes a Patient or Theme is not expressed as direct object, but as an instrument phrase, marked with the instrumental preposition *hai*. This preposition (§4.7.10) is usually not followed by a determiner (§5.3.2.2) and marks a non-specific entity.<sup>60</sup>

This construction only occurs when the Patient is non-specific. The entity may be unknown (e.g. something which is being sought, bought or asked for) as in (136), but this is not necessarily so. Two pairs of examples: (135) and (137) show the usual construction with the Patient as DO, while in (136) and (138) the same argument is marked with *hai*. 61


Argument expression by means of *hai* is especially common with verbs that involve both a Theme (or Patient) and a Goal (or Recipient or Beneficiary), like *vaꞌai* 'to give' and *hoa* 'to throw'. The Theme of *vaꞌai* is usually expressed as DO, while the Goal is marked with *ki*, as in (139). In (140) however, the Goal is expressed as DO, while the Patient 'food'

<sup>60</sup> The *hai*-marked Patient in (136) could be considered a "demoted object", which would imply that the construction in (136) is derived from the one in (135). While this may seem plausible in some cases, in other cases it is not at all clear that the construction with Patient as DO is more basic than the instrumental construction (see the discussion on (142–146) below). Goldberg (1995) argues that it is unnecessary and often unwarranted to assume a transformational relationship between two constructions with alternative argument expression.

<sup>61</sup> Notice that non-specific objects can also be constructed with a acc marker + determiner; see examples (34–36) in section §5.3.3.

is marked with *hai*. The motivation for this may be pragmatic: the Goal is more topical in discourse, hence expressed as a core argument.<sup>62</sup>


Thus with three-argument verbs we encounter the following two patterns:

(141) (i) verb *i* + Theme *ki* + Goal (ii) verb *i* + Goal *hai* + Theme

With the verbs discussed so far, (i) is the rule, while the *hai*-Theme construction in (ii) only occurs occasionally. However, with a number of verbs, the *hai-*Theme construction is very common; all of these have two arguments, apart from the Agent.

**Throwing** The Patient of *tau* 'throw' may be expressed as DO, as in (142); alternatively, the Goal is expressed as DO as in (143), and the Patient is marked with *hai*.


**Covering/filling** Verbs referring to covering or filling potentially have a Container argument (the object filled or covered) and a Substance argument (the stuff filling or covering the object). Either one can be expressed as a core argument. For intransitive verbs such as *ꞌī* 'to be full', this can result in a passive-like construction, illustrated in (69) in §8.5.3, repeated here:

(144) *Hai* ins *oho* go *iŋa* nmlz *nei* prox *ko* prf *ꞌī* full *ꞌā* cont *te* art *motu* island *nei* prox *e* ag *te* art *iŋoiŋo.* dirty *ꞌE* and *te* art *vai,* water *ko* prf *ꞌī* full *ꞌā* cont *e* ag *te* art *meꞌe* thing *ꞌiꞌino.* pl:bad

'When this happens, this island will be full of pollution. And the water will be full of bad things.' [R649.119]

<sup>62</sup> In the terminology of Haspelmath 2005 (quoted in Reesink 2013), the DO-Theme construction is "indirective", while the *hai*-Theme construction is "secundative". Reesink finds that in a sample of 72 Papuan languages, a large majority has a secundative construction as the only option. In languages that allow a choice between both constructions, the choice may be determined by a variety of pragmatic factors.

8.6 Non-standard verbal clauses

For transitive verbs, the following examples illustrate the two options. In (145), the Container is direct object, while the Substance is marked as instrument; in (146), the Substance is direct object, while the container is marked with a locative preposition.


What unites the *hai*-Theme constructions, is that in most cases the Theme has a somewhat instrumental sense: it refers to an object used to perform the act (e.g. a stone thrown at somebody), or a means to reach a certain goal (e.g. food to alleviate hunger). This also means that the entity is usually non-human. However, the following example shows that it may be human as well:

(147) *He* ntr *kimi* search *mai* hither *hai* ins *nuꞌu* people *mo* for *oho* go *hai* ins *ika* fish *mo* to *ruku* dive *mai.* hither 'He looked for people to go looking for fish, to dive.' [R309.100]

#### **8.6.4.4 Variable argument assignment**

The previous section showed that Patients may be expressed as instrument, allowing other arguments to be expressed as direct object. This is in fact part of a wider phenomenon: with many verbs, arguments can be expressed in different ways, depending on which arguments are relevant or topical in the context. An exhaustive treatment is beyond the scope of this grammar; a few examples show the types of variation involved.

First: *amo* 'to clean, to wipe'. The direct object may express either the object cleaned as in (148), or the substance removed as in (149).


Another example is *oŋe* 'to lack, to be in need, to suffer shortage'. The subject may express either the person(s) in need as in (150), or the substance which is lacking as in (151).


#### **8.6.4.5** *Ko* **with non-topicalised arguments**

As discussed in §8.6.2.1, *ko* in verbal clauses marks topicalised subjects, preverbal subjects which are highlighted as topics. Occasionally, *ko* is used to mark postverbal arguments. In these cases, the argument is marked as prominent for a certain reason; different motivations can be distinguished.

*Poreko* **'to be born'** The subject of *poreko* is sometimes *ko-*marked. The noun phrase introduced by *ko* introduces a new participant ('new' in an absolute sense!); *ko* may indicate that the referent will be thematic in the text that follows.


**Naming verbs** *Ko* is used in the complement of the naming verbs *kī* 'to call' and *nape* 'to name, convey a name to'.<sup>63</sup> The complement of these verbs can be analysed as an identifying clause with the *ko*-marked noun phrase as predicate. Its subject can be implicit as in (154), or expressed as *ꞌīŋoa* (which is case-marked as direct object of *nape* or *kī*) as in (155).<sup>64</sup>

(154) *He* ntr *poreko* born *te* art *poki* child *he* ntr *nape* name *ko* prom *Tikitiki* Tikitiki *ꞌa* a *Ataraŋa.* Ataranga 'The child was born, they called it Tikitiki a Ataranga.' [R532-02.005]

<sup>63</sup> The same happens in Tongan and Samoan (Clark 1976: 45); in Tongan, *hoko* 'become' and *ui* 'call' both take a *ko*-marked complement.

<sup>64</sup> These constructions are sometimes characterised as "small clauses", a pair of constituents which are in a subject-predicate relation, but which may not be a single constituent syntactically (see Bowers 2001). Bauer (1991: 12) also analyses the same constructions in Māori as embedded equative (=identifying) clauses. See sec. §9.2.2 and §9.2.1 on identifying and classifying clauses, respectively.

8.6 Non-standard verbal clauses

(155) *He* ntr *nape* to\_name *i* acc *te* art *ꞌīŋoa* name *ko* prom *Māhina* Mahina *Tea.* Tea 'They called her (lit. her name) Mahina Tea.' [R399.003]

With common nouns, the noun phrase is *he*-marked, and the complement can be analysed as a classifying clause; again, its subject is case-marked as direct object of the main verb.

(156) *He* ntr *nape* to\_name *i* acc *te* art *rāua* 3pl *ꞌīŋoa* name *he* pred *hānau* race *momoko.* slender 'They called them 'slender race'.' [R370.008]

Under this analysis, the noun phrase marked with *ko* or *he* as such is not a complement of the verb, but rather the predicate of a complement clause.

**Perception verbs** As a prominence marker, *ko* signals information that the speaker wishes to highlight in some way, for example because it is thematic. This may explain why *ko* can be used to mark the complement of perception verbs like *takeꞌa* 'see', *uꞌi* 'look' and *ŋaroꞌa* 'perceive'. Normally, perception verbs take a direct object as in (157):<sup>65</sup>

(157) *He* ntr *uꞌi* look *a* prop *Makemake* Makemake *i* acc *te* art *kona* place *rivariva,* good:red *kona* place *Kauhaŋa.* Kauhanga 'Makemake saw a good place, the place Kauhanga.' [Mtx-1-01.026]

But the perceived object may also be marked with *ko*, which highlights the significance of the object for the participant. The perceived object may be surprising and unexpected. This use of *ko* can be characterised as participant-oriented.


'In the early morning I heard an insect chirping.' [R109.005]

If the complement involves an event or action, this is expressed as a clause following the *ko-*marked noun phrase, for example *e oho nō mai ꞌā* 'it was coming' in (158) (§11.3.1.2).

<sup>65</sup> With passive perception verbs, the object may be unmarked – see (19) in sec. §8.3.2. With *uꞌi*, the object may be marked with *ki* – see (118) in sec. §8.6.4.2.

#### **8.6.4.6 Object incorporation with** *rovaꞌa*

Object incorporation is rare in Rapa Nui; it mainly occurs with modifying verbs in noun phrases (§5.7.2.3). On the clausal level, object incorporation does not occur, but there is one exception: the object of *rovaꞌa/ravaꞌa* 'obtain'. As shown in §8.4.1 above, the object of *rovaꞌa* is usually expressed as a regular noun phrase, though the accusative marker *i* is often omitted. However, when the object is something edible, it tends to be incorporated into the verb.<sup>66</sup> The incorporated object shows the following characteristics:


In the following example, the object noun *meꞌe* is incorporated into the verb phrase, but the relative clause *mo kai*, which modifies *meꞌe*, is left stranded at the end of the clause.

(162) *Māuruuru* thank *ki* to *tū* dem *taŋata* man *era* dist *i* pfv *rovaꞌa* obtain [*meꞌe* ] thing *ai* pvp *rāua* 3pl [*mo* for *kai*]*.* eat 'Thanks to that man they had obtained something to eat.' [R349.021]

#### **8.6.4.7 Agentive use of** *i*

As discussed in §8.3 above, the Agent is marked with *e* under certain conditions. Rapa Nui also has a second Agent marker: the locative preposition *i* (§4.7.2.2) is used to mark agents which are not an argument of the verb. Usually, *i*-marked Agents occur with intransitive verbs which have a non-agentive S, such as *motu* 'break (intr.)' and *ꞌōꞌotu* 'be cooked'.<sup>67</sup> Agentive *i* is especially common with a few patientive verbs: *mate* 'die', *rehu* 'be forgotten' and *ŋaro* 'to be lost; to be forgotten'; however, it may occur with any intransitive verb. A few examples:

<sup>66</sup> Object incorporation with *rova ꞌa* only occurs in modern Rapa Nui; in older texts *rova ꞌa* always takes a full NP object.

<sup>67</sup> Hooper (1984b) discusses the same construction in Māori, where agentive *i* is used with a closed class of verbs. Hooper characterises these verbs as inherently passive. In Māori, these verbs share certain syntactic characteristics: they do not occur in the imperative and cannot function as NP head or modifier. (See also

8.6 Non-standard verbal clauses


(166) is remarkable because *tuꞌu* is an active verb; its subject is an Agent. Even so, an agentive *i*-phrase is added, expressing an external (higher order) Agent which causes the event to happen. A construction like this is semantically similar to causativisation of an agentive verb (§8.12.2).

In the examples so far, the Agent noun phrase is animate. This is to be expected, as prototypical agentivity implies that the action is done volitionally, something which is only possible with an animate agent. However, the event can also be caused by an animal, object, force or event. In the latter case, Rapa Nui orthography uses the preposition *ꞌi* (with glottal); note however that *i* and *ꞌi* are merely different spellings of the same preposition (§2.2.5; §4.7.2). Below are two examples; more examples are given in §4.7.3.

	-

Pucilowsky 2006: 33.) Different from Māori, Rapa Nui allows agentive *i* with any intransitive verb, even active verbs, though it mostly occurs with verbs having a Patient subject.

Alexander (1981a: 135) claims that agentive *i* in Rapa Nui only occurs with inanimate subjects, while *e* is only used with animate subjects. Neither claim is correct: see the examples in this section and in §8.5.2. Finney & Alexander (1998: 21) characterise the agentive *i* construction as a "lexical passive": a passive without a corresponding active form. The subject expresses a Patient, while the Agent is expressed as an oblique, just as in passive constructions. Notice, however, that the Agent is not part of the case frame of the verb; the inherent meaning of the verb is not such that the action is carried out by the Agent on the Patient; rather, the Patient undergoes the event, without an Agent being in view.

### **8.7 Case marking in nominalised clauses**

With nominalised verbs and verbal nouns (§3.2.3), case marking is governed by the same rules as with verbs in general, with one important difference: arguments that would be unmarked in a verbal clause, are expressed as possessives when the verb is nominalised. Agents may be either *a*- or *o*-possessed; Patients are *o*-possessed (§6.3.3.4).

This results in the following situation:

1. The Agent is usually possessive, as in (169–170).

	- 2. The Agent is *e*-marked in the situations listed in §8.4.2, for example in VOA clauses as in (171), and in transitive clauses with implied O as in (172):

'There were no men left, they had all been handed over by Vaha (lit. the giving by Vaha had finished) to (the people in) the canoe.' [Mtx-3-01.122]

	- 4. Sometimes the Patient is expressed as possessive; this may indicate passivisation, i.e. the Patient has become subject.

It never happens that both subject and object are expressed as possessive.

8.8 Obliques

### **8.8 Obliques**

### **8.8.1 Indirect object?**

Apart from the subject and the direct object, verbs may have various other arguments, expressing roles such as Goal (of movement or action), Beneficiary and Recipient. This includes semantic roles traditionally labelled as "indirect object", e.g. the Recipient of *vaꞌai* 'give'. In Rapa Nui, these constituents are syntactically not different from any other oblique role; they do not share the characteristics of the core grammatical relations, subject and direct object, but behave like other obliques, as the following evidence shows.

	- 2. Patients can be expressed as subject in passivisation (§8.5.1), but arguments such as recipients and beneficiaries cannot; the following construction is impossible:
	- 3. In relative clauses, relativised subjects and direct objects can be omitted; other constituents (including recipients) need to be expressed (§11.4.2).
	- 4. This also has consequences for content questions with the interrogative pronoun *ai* 'who'. Interrogative subjects and objects are constructed as clefts, nominal clauses containing a relative clause (§10.3.2.1–10.3.2.2). As a consequence, an interrogative direct object is not marked with the accusative marker *i*, but as a nominal predicate. By contrast, interrogative obliques are always preceded by the appropriate preposition; this includes recipients, as in (179):
		- (178) *¿ꞌI* at *muri* near *i* at *a* prop *ai* who *a* prop *Eva* Eva *ka* cntg *noho* stay *era* dist *ꞌi* at *a* prop *Tire?* Chile 'With whom will Eva stay in Chile?' [R615.660]

(179) *¿Ki* to *a* prop *ai* who *ki* to *a* prop *ai* who *i* pfv *kī* say *ai* pvp *mo* for *hāꞌūꞌū* help *mai* hither *ꞌi* at *tōꞌona* poss.3sg.o *ꞌati?* problem 'Whom and whom did he ask to help him in his problem?' [R615.145]

We may conclude that all constituents other than subject and direct object are obliques, none of which has a special status as "indirect object".

### **8.8.2 Marking of obliques**

Any preposition may serve to mark an oblique constituent. Examples are given in the sections discussing these prepositions (subsections of §4.7). Two prepositions which are particularly common with obliques, are the directional preposition *ki* 'to' (§4.7.4) and, to a lesser extent, *mo* 'for'. Because of their wide range of uses, both will be discussed in some detail here.

#### **8.8.2.1** *Ki* **'to'**

*Ki* marks the object of middle verbs (§8.6.4.2 above). In addition, it marks semantic roles such as Recipient, Addressee and Goal:


*Ki* marks the oblique argument of a diverse group of verbs, including for example *nonoꞌi* 'ask for', *moe* 'sleep with', *tauꞌa* 'to fight against', *māhani* 'get to know' and *koromaki* 'to miss, long for':


8.8 Obliques

Finally, *ki* expresses the causee in causative constructions based on a transitive verb (§8.12.3).

#### **8.8.2.2** *Mo* **'for'**

Some verbs take a Goal complement marked with the benefactive preposition *mo*. The complement of *riri* 'to be angry' may be marked with *i* or *ki* (see (123–124) on p. 418), but also by *mo*:

(185) *ꞌIna* neg *koe* 2sg *ko* neg.ipfv *riri* angry *mo* for *tuꞌu* poss.2sg.o *māmā* mother *era.* dist 'Don't be angry with your mother.' [R103.071]

With other verbs both *ki* and *mo* are possible. With verbs of giving, benefactive *mo*/*mā* may be used instead of the more usual *ki*. The choice between *mo* and *mā* depends on the semantic relationship between the Recipient and the given object (§6.3.3).


As a benefactive preposition (expressing intended possession), *mo* emphasises possession of the object by the Recipient which results from the act of giving. By contrast, a *ki*-marked Recipient is not necessarily the possessor of the given object. In the following example, Tiare is not the (ultimate) possessor of the given object. In this case, benefactive *mo* may not be appropriate.

(188) *He* ntr *toꞌo* take *mai* hither *he* ntr *vaꞌai* give *i* acc *tou* dem *mareni* watermelon *era* dist *ki* to *a* prop *Tiare…* Tiare 'She took the watermelon and gave it to Tiare (and told her to give it to Merahi).' [R309.072]

*Mo* may mark the object of a feeling (like *ki*), or the topic of an utterance ('about'):


'Tahonga cried because of the words Hotu said about his fish.' [R301.295]

Nowadays, 'about' tends to be expressed by *o ruŋa* (see (153) on p. 128).

#### **8.8.2.3 Order of constituents**

As the examples above show, oblique constituents usually come after the subject. They may also be preposed as in (175–176) above. If there is also a direct object, the oblique constituent usually comes last as in (180), though the reverse order also occurs:

(191) *He* ntr *tuhi* point\_out *e* ag *Vaha* Vaha [*ki* to *a* prop *Kava* ]OBL Kava [*i* acc *te* art *kona* place *mo* for *titi* stack *o* of *te* art *hare.*]DO house

'Vaha pointed out to Kava a place to build the house.' [R229.110]

This happens when the oblique is more topical than the direct object, or when the oblique is lighter (i.e. shorter and structurally simpler) than the direct object. In (191) the oblique is short, while the direct object is a complex noun phrase.

### **8.9 Reflexive and reciprocal**

Reflexivity occurs when two constituents in a clause are coreferential, in most cases subject and object. Reciprocality involves two participants which are mutually involved in an action. Rapa Nui does not have specific pronouns or other forms to express these categories; various strategies are used, which are discussed below.

### **8.9.1 Reflexive**

Reflexivity can be expressed by a personal pronoun in the appropriate person and with the appropriate preposition:<sup>68</sup>


Reflexive reference may be made more explicit by the identity particle *ꞌā* or *ꞌana* (§5.9):

(194) *Ko* prf *riꞌariꞌa* afraid *ꞌana* cont *ꞌi* at *tū* dem *māuiui* sick *era* dist *ōꞌona* poss.3sg.o *e* ipfv *maꞌu* carry *era* dist *ꞌi* at *roto* inside *i* at *a* prop *ia* 3sg *ꞌā.* ident 'She was afraid of the sickness she carried inside herself' [R301.091]

<sup>68</sup> According to Anderson & Keenan (1985: 265), there "appears to be a universal constraint against using ordinary pronouns and noun phrases for referring to the same individual twice within a single clause". This constraint does not operate in Rapa Nui.

8.10 Comitative constructions: 'with'

(195) *He* ntr *noho* sit *ꞌi* at *ruŋa* above *i* at *te* art *māꞌea* stone *e* num *tahi,* one *he* ntr *kī* say *ki* to *a* prop *ia* 3sg *ꞌā…* ident 'He sat down on a stone and said to himself…' [R229.365]

A reflexive action, which the subject performs on itself, can also be expressed by a causative verb with unexpressed object (§8.12.2).

### **8.9.2 Reciprocal**

Like reflexivity, reciprocality may be expressed by a pronoun + identity marker *ꞌā/ꞌana* as in (196), or left implicit as in (197):


### **8.10 Comitative constructions: 'with'**

### **8.10.1 Introduction**

A comitative relationship ('X with Y') is expressed by the following construction:

(198) NP<sup>1</sup> comitative marker *ko* NP<sup>2</sup>

An example:

(199) *He* ntr *hoki* return *Makemake*<sup>1</sup> Makemake *rāua*COM 3pl *ko* prom *Haua*2*.* Haua 'Makemake and Haua returned.' [Mtx-1-01.029]

The comitative marker can be realised in several ways:


1 is specified for number and person, 2 for number. Both 1 and 2 are used in an inclusory sense: they denote the total set of referents of NP<sup>1</sup> and NP2. In other words, when both NP<sup>1</sup> and NP<sup>2</sup> are singular, the comitative marker is dual if a dual form is available (i.e. in the 1st person) and plural otherwise; when either NP<sup>1</sup> or NP<sup>2</sup> is dual or plural, the comitative marker is plural. This is illustrated in (199) above, where the plural *rāua* connects two singular noun phrases.

All four constructions are used commonly in both older and newer texts. They will be discussed in turn in sections §8.10.2–8.10.5. §8.10.6–8.10.7 discuss issues concerning pronouns, especially the inclusory pronoun construction, which is a truncated variant of comitative constructions.

The particle *ko* in these constructions is best considered as the prominence marking *ko*, rather than a separate lexeme meaning 'with'. As §4.7.12 shows, *ko* has a wide variety of uses; in §4.7.12.4 I suggest that *ko* is a default preposition, marking any noun phrases without a thematic role assigned by a verb or preposition. The second noun phrase in a comitative construction is exactly that: it is not governed by a verb or preposition, and therefore gets the default preposition *ko*. 69

The noun phrase as a whole is plural, even when NP<sup>1</sup> is singular. This is shown by the fact that a plural verb can be used with a singular NP<sup>1</sup> (whether the latter is explicit or implied). In the example below, the implied NP<sup>1</sup> is singular 'he', yet the verb is plural.

(200) *He* ntr *nonoho* pl:sit Ø<sup>1</sup> *ararua* the\_two *ko* prom [*tū* dem *repa* young\_man *era* dist *ꞌāꞌana* ]2*.* poss.3sg.a 'He sat down together with his son.' [R310.020]

### **8.10.2 Pronouns as comitative markers**

When the comitative marker is a pronoun, this pronoun is inclusory: it refers to the total group of NP<sup>1</sup> and NP2. In other words, dual/plural *rāua* is used regardless whether NP<sup>1</sup> and NP<sup>2</sup> are singular or plural. An example:

(201) *He* ntr *mataku* fear *Rapu* Rapu *ꞌi* at *te* art *hatutiri* thunder *rāua* 3pl *ko* prom *te* art *ꞌuira.* lightning 'Rapu feared the thunder and the lightning.' [Fel-40-028]

<sup>69</sup> Additional confirmation that the sense 'with' is not expressed by *ko* as such but by the construction as a whole, comes from the following example from the Bible translation. Here *ararua* 'the two' (§8.10.3 below) is used in an inclusory sense in a relative clause; in this construction, *ko* is not used: due to the gapping strategy, there is no constituent to be marked with *ko*.

<sup>(</sup>i) *…mo* for *haka* caus *ūtuꞌa* punish *i* acc *te* art *kope* person [*ararua* the\_two *i* pfv *ture* quarrel *ai*]*.* pvp 'to condemn the person with whom she had a conflict (lit. …the person the two quarrelled)' (Luke 18:3)

8.10 Comitative constructions: 'with'

In practice, *rāua ko* is mostly used to connect pairs of single participants, just like *ararua ko* (§8.10.3 below); larger sets are usually expressed by *ananake ko* (§8.10.3).

First and second person pronouns can also be used as comitative markers, and they are inclusory as well. First person pronouns make a distinction between dual and plural. Dual pronouns are used when NP<sup>1</sup> and NP<sup>2</sup> are both singular as in (202); when NP<sup>1</sup> and/or NP<sup>2</sup> is plural as in (203), the pronoun is plural.


These constructions are similar to the inclusory pronoun construction, discussed in §8.10.7 below.

### **8.10.3** *Ararua* **and** *ananake* **'together' as comitative markers**

*Ararua* is a definite numeral meaning 'the two' (§4.3.4). *Ananake* means 'together' in modern Rapa Nui; in older texts has the more general sense 'all' (§4.4.4). Both words indicate a collectivity or group: *ararua* refers to a group of two, *ananake* to a group larger than two.

Just like pronouns, *ararua and ananake* as comitative markers are used in an inclusory way. *Ararua* is used when the total set denoted by NP<sup>1</sup> + NP<sup>2</sup> is two; it is thus similar in use to *rāua ko*, which usually connects two singular noun phrases. Though both constructions are common, the use of *ararua ko* has increased over time, while the use of *rāua ko* has decreased. An example:

(204) *He* ntr *eeke* pl:go\_up *a* prop *Rāvī* Ravi *ararua* the\_two *ko* prom *Hotu* Hotu *ki* to *ruŋa* above *i* at *te* art *hoi.* horse 'Ravi mounts the horse with Hotu.' [R616.736]

*Ananake* is used for sets larger than two, in which NP<sup>1</sup> and/or NP<sup>2</sup> is plural. In (205) NP<sup>1</sup> 'the people' is plural, while NP<sup>2</sup> is singular. In (206), NP<sup>1</sup> is singular while NP<sup>2</sup> is plural.

(205) *He* ntr *takataka* gather:red *he* ntr *oho* go *te* art *taŋata* man *ꞌi* at *rote* inside\_art *hare* house *pure* prayer *ki* to *te* art *hāpī* learn *ananake* together *ko* prom *te* art *matuꞌa.* parent

'The people gathered in the church to learn together with the priest.' [R231.288]

(206) *He* ntr *noho* stay *a* prop *Uho* Uho *ananake* together *ko* prom *te* art *matuꞌa,* parent *ko* prom *te* art *taina.* sibling 'Uho lived together with her parents and siblings.' [Mtx-7-12.055]

In the examples so far, the comitative construction is a constituent in a clause, usually the subject. The comitative construction can also be a (nonverbal) clause by itself, meaning 'A is with B.' In that case, its construction is as follows:

(207) *ararua*/*ananake* NP<sup>1</sup> *ko* NP<sup>2</sup>

Here are two examples. In these constructions, both the comitative marker (*ararua*/*ananake*) and the NP<sup>1</sup> pronoun are inclusory, denoting the total set of referents of NP<sup>1</sup> and NP2: plural in (208), dual in (209).


### **8.10.4** *Koia ko* **'with'**

*Koia* is originally the third person singular pronoun *ia* preceded by the prominence marker *ko*. However, when used as a connector it has lost the character of a pronoun and is written as one word.

*Koia ko* marks attendant circumstances. As such it may introduce both noun phrases and circumstantial clauses: the following noun phrase or clause indicates an action, state of mind, person or object which in some way accompanies the main clause or a participant in the main clause.

#### **8.10.4.1** *Koia ko* **+ noun**

Like *rāua ko* and *ananake/ararua ko*, *koia ko* may serve as a comitative marker connecting two noun phrases which are closely associated:

(210) *Te* art *huaꞌai* family *nei* prox *a* prop *Paio* Paio *koia* com *ko* prom *tāꞌana* poss.3sg.a *viꞌe* woman *ko* prom *Uka* Uka *ꞌa* a *Nei* Nei *Ariro.* Ariro

'This family (consisted of) Paio with his wife Uka a Nei Ariro.' [R439.003]

8.10 Comitative constructions: 'with'

However, *koia ko* + noun phrase usually indicates a looser connection to the preceding context than other comitative markers. The following example, in which *rāua ko* and *koia ko* are both used, is illustrative. The first set of referents, 'he' (implied) and 'his two brothers', is connected by *rāua ko*. Another entity, 'the warriors' is added to this first set.

(211) *He* ntr *hoki* return *mai* hither *rāua* 3pl *ko* prom *tōꞌona* poss.3sg.o *ŋā* pl *taina* sibling *e* num *rua,* two *koia* com *ko* prom *te* art *nuꞌu* people *pāoa.* warrior 'He returned together with his two brothers, and with the warriors.' [Fel-64.107]

Often *koia ko* + NP is not directly connected to a preceding noun phrase, but introduces an additional item to the clause as a whole. In the following example, *rēkaro* 'present' is not linked to a specific noun phrase, but adds an additional circumstance to the clause: 'they gathered around Eva with (=while carrying) gifts.'

(212) *Ko* prf *takataka* gather:red *tahi* all *ꞌana* cont *te* art *rapa* Rapa *nui* Nui *ꞌi* at *muri* near *koia* com *ko* prom *te* art *rēkaro* present *e* ipfv *vaꞌai* give *era* dist *ki* to *a* prop *Eva.* Eva 'All the Rapa Nui people gathered next to (her) with presents they gave to Eva.' [R210.218]

Notice that there are a few other expressions in which *ko* is not followed by a determiner (§5.5.2).

#### **8.10.4.2** *Koia ko* **+ verb**

*Koia ko* followed by a verb phrase or adjective indicates a secondary circumstance under which the main clause takes place (§11.6.8). This circumstance can be expressed by anything ranging from a single word to a full clause:


in\_fact

'She cried bitterly (lit. she died from crying), while asking how that accident had happened.' [R437.101]

In modern Rapa Nui, the verb is sometimes is preceded by the article *te*; this does not seem to make much difference in meaning.

(215) *He* ntr *kī* say *atu* away *a* prop *nua* Mum *koia* com *ko* prom *te* art *taŋi…* cry 'Mum said crying…' [R237.014]

### **8.10.5** *Ko* **without comitative marker**

Occasionally *ko* on its own, without comitative marker, is used to connect two noun phrases. As with *koia ko*, the determiner before the noun may be left out.

(216) *ꞌI* at *te* art *pō* night *varu* eight *tōꞌou* poss.2sg.o *miro* tree *ko* prom *aka* root *ko* prom *raurau.* branch:red 'On the eighth day your tree (will arrive), with roots and branches.' [Mtx-7-18.003]

This construction is used both in older and newer texts, but is not common in either corpus.

*Ko* on its own is somewhat more common in comitative constructions containing three or more elements: the first two nouns are linked by one of the comitative markers; after that, only *ko* is used without repeating the comitative marker. An example:

(217) *He* ntr *noho* stay *Rano* Rano *rāua* 3pl *ko* prom *tāꞌana* poss.3sg.a *poki,* child *ko* prom *te* art *viꞌe.* woman 'Rano lived with his child and his wife.' [Mtx-7-18.001]

This construction reminds of the use of *ko* in lists: in a list of items every item may be preceded by *ko*, regardless its syntactic function in the clause (§4.7.12.1).

#### **8.10.6 Pronouns as NP<sup>1</sup> : inclusory and exclusory use**

When NP<sup>1</sup> in a comitative construction is a pronoun, it may either include or exclude the referent(s) of NP2; in other words, it may be either inclusory or exclusory.<sup>70</sup> In (218), the dual pronoun *māua* is inclusory: its reference includes NP<sup>2</sup> *Peŋipeŋi*: 'me and Pengipengi'.

(218) *E* num *tahi* one *mahana* day *māua* 1du.excl *ararua* the\_two *ko* prom *Peŋipeŋi* Pengipengi *e* ipfv *noho* stay *era* dist *ꞌi* at *Tuꞌu* Tu'u *Tapu.* Tapu

'One day, Pengipengi and I were staying in Tu'u Tapu.' [R133.001]

<sup>70</sup> Other examples are found in the previous sections. Inclusory use of a pronoun as NP<sup>1</sup> is found in (208–209) above, exclusory use in (202–203). Notice that in the last two examples, an exclusory pronoun as NP1 is followed by an inclusory pronoun as comitative marker.

8.10 Comitative constructions: 'with'

In (219), NP<sup>1</sup> is exclusory: the pronoun is singular and does not include NP<sup>2</sup> 'a person of your house'.

(219) *¿ꞌI* at *ꞌaŋahē* when.past *te* art *mahana* day *hopeꞌa* last *i* pfv *noho* sit *i* pfv *vānaŋa* talk *rivariva* good:red *koe* 2sg *ararua* the\_two *ko* prom *te* art *kope* person *e* num *tahi* one *o* of *tuꞌu* poss.2sg.o *hare?* house

'When was the last day you sat down and talked well with a person of your house?' [R209.032]

### **8.10.7 The inclusory pronoun construction**

The inclusory<sup>71</sup> pronoun construction consists of a dual or plural pronoun denoting a set of referents, followed by a noun phrase denoting a subset of these referents.<sup>72</sup> The referents indicated by the noun phrase are included in the set indicated by the pronoun. Here is an example:

(220) *Te* art *parautiꞌa,* truth *e* ipfv *haŋa* love *rō* emph *ꞌā* cont *a* prop *au* 1sg *ki* to *a* prop *kōrua* 2pl *ko* prom *koro.* Dad 'The truth is, I love you and Dad very much.' [R229.498]

This sentence is spoken by a child to her mother. The dual/plural pronoun *kōrua* refers to mother and father (even though the father is not directly addressed); the father, indicated by the noun *koro*, is a subset of this group. The noun *koro* is marked with the default preposition *ko*. <sup>73</sup> The phrase as a whole could be paraphrased 'you, including father'.

Two more examples:


<sup>71</sup> I use the term *inclusory* following Lichtenberk (2000) for pronouns which denote the complete set of referents of the noun phrase; by analogy, *exclusory* means that a pronoun denotes a subset of referents, excluding another subset. Dixon (2010b: 207) uses the term *pronoun elaboration*: the noun phrase elaborates on the reference of the pronoun.

<sup>72</sup> Anderson & Keenan (1985: 267), consider the pronoun in such cases to be semantically singular. Schwartz (1988) takes the same approach. However, it seems more straightforward to take the pronoun as the head of the construction with plural reference.

<sup>73</sup> Schwartz (1988: 241) points out that connectors used in inclusory pronoun constructions are generally not the same connectors used in coordination, but rather elements also used to indicate accompaniment. This is also true in Rapa Nui, where *ko* without any further marker can be used in comitative constructions (§8.10.5).

The inclusory construction is a concise or truncated variety of the comitative construction: where other comitative constructions have a NP<sup>1</sup> + comitative marker, in the inclusory pronoun construction there is only a pronoun. The latter can be analysed in several ways:


### **8.11 The vocative**

Vocative phrases occur in various positions in the sentence: initial as in (223), after the first phrase or clause as in (224), and final as in (225).


<sup>74</sup> The reasons for this deletion could be pragmatic: a repeated pronoun can be perceived as redundant or stylistically awkward.

8.12 The causative

As (224) and (225) show, the end of a vocative phrase is often marked by *ē*. <sup>75</sup> This particle is always used when the vocative occurs in the middle or at the end of a clause, but rarely with initial vocatives.

These examples also illustrate that names and proper nouns like *koro* 'father' in the vocative are not preceded by the proper article *a*.

When common nouns occur in the vocative, they occur with or without article *te*. In modern Rapa Nui the following tendencies can be observed regarding the use of the article:

	- 2. Nouns preceded by the plural marker *ŋā* do not have the article.
	- 3. In other situations, the article tends to be used. This includes familial terms never used as personal nouns, as well as other common nouns.

### **8.12 The causative**

### **8.12.1 Introduction**

The preverbal particle *haka* (< PPN *\*faka*) expresses causation. When *haka* is placed before a verb, the valency of the verb is increased by one: an Agent-subject is added expressing an entity which causes the event to happen; the S/A of the root verb is demoted to direct object.

*Haka* is used with intransitive verbs, transitive verbs, adjectives and nouns; examples will be given in the following sections. Furthermore, it is used with copula verbs (see

<sup>75</sup> This particle is glossed voc, but it may be compared to two other particles *ē* which occur before a pause: *ē* in the expression *pē nei ē* 'thus', which occurs before an indirect speech (§4.6.5.1), and *ē* indicating an ongoing event (found e.g. in (160) on p. 364).

(248) below)<sup>76</sup> and with adverbs (see (260)). There are no examples in the corpus where it is used with locationals.

*Haka* is fully productive, though in certain cases the meaning of *haka* + root is lexicalised (§8.12.5 below).

Regarding the morphological status of *haka*: in many analyses of Polynesian languages, *haka* is considered a prefix and written together with the root. Semantic motivations may play a role here: *haka* + root forms a derived verb with its own argument structure, often with a lexicalised (i.e. non-predictable) meaning. However, phonologically and morphologically *haka* is no different from preverbal particles. There is no phonological integration with the root: *haka* does not affect the root phonologically any more than particles do. Moreover, *haka* may be separated from the root by preverbal particles, both degree modifiers (see (261) below) and the constituent negator *taꞌe*:

(229) *te* art *nuꞌu* people *haka* caus *taꞌe* conneg *au* pleased *ꞌo* because\_of *tōꞌona* poss.3sg.o *rē* win *i* acc *tū* dem *ꞌāꞌati* contest *era* dist 'the people who were not pleased because of her winning the contest' [R347.036]

### **8.12.2 Causativisation of intransitive predicates**

In many cases *haka* expresses direct causation, as illustrated by the following pair of examples. In (230) the Theme argument 'the fire' is the subject; in (231), an (implicit) Agent has been added, expressing the persons causing the fire to burn; the Theme argument is now expressed as direct object.


This example involves an inanimate argument and a non-agentive verb. With agentive verbs, the Agent of the root (the causee) is often actively involved in the event, despite the presence of additional Agent (the causer). In (233) the children are just as much the Agent of going as in (232), even though another Agent has been added.

(232) *ꞌI* at *te* art *ahiahi* afternoon *he* ntr *oho* go *te* art *ŋā* pl *poki* child *ki* to *Mataveri.* Mataveri 'In the afternoon, the children went to Mataveri.' [R159.015]

<sup>76</sup> According to Dixon (2012: 251), this is crosslinguistically unusual.

8.12 The causative

(233) *He* ntr *haka* caus *oho* go *i* acc *te* art *ŋā* pl *poki* child *ki* to *haho* outside *mo* for *kokori.* pl:play 'She made the children go outside to play.' [R236.013]

The degree to which the causee is actively involved in the event, may vary. The following example can mean either that the subject lifts Poreone up and puts him on the horse, or that he helps him to mount the horse. (The latter is more likely, as Poreone happens to be about ten years old.)

(234) *He* ntr *haka* caus *eke* go\_up *ki* to *ruŋa* above *i* at *te* art *hoi* horse *i* acc *tū* dem *poki* child *era* dist *ko* prom *Poreone.* Poreone 'He made the child Poreone mount the horse.' [R105.095]

Non-agentive verbs – which have a Theme as subject – become agentive by causativisation, as the following pair of examples shows. In (235), the subject *koe* is Theme, while *i* marks the Agent (§8.6.4.7 on non-argument Agents). In (236), the subject *koe* is Agent, while *i* marks the direct object.


When adjectives are causativised, the property expressed by the adjective is brought about by an Agent; the person or thing having the property is expressed as direct object, *i te pista* in (237), *te tātou mahana* in (238).


*Haka* may also express indirect causation (cf. Dixon 2012: 274). In this case, the Agent does not actively bring about the event, but facilitates the event in some way. This may involve letting a natural process run its course as in (239), giving permission as in (240), or waiting for something to happen as in (241):


### **8.12.3 Causativisation of transitive verbs**

All the examples so far involve intransitive predicates; with these predicates, the original subject (the causee) is demoted to direct object (DO), while the causative Agent occupies the subject position. Now when the root is a transitive verb, the DO position is already occupied.<sup>77</sup> In that case, the original direct object remains the direct object; the causee is expressed as oblique, marked with the preposition *ki*. This happens whether the DO is expressed as in (242–243),<sup>78</sup> or implied as in (244).


The same happens when the root is a middle verb which takes a direct object introduced by *ki*; see the discussion about (134) in §8.6.4.2.

When no Patient is implied, the verb is intransitive and the causee is expressed as DO. The verb *ꞌaꞌamu* 'to tell' is often transitive (with a story as DO), but in the following example it is used intransitively, which means that the DO position is available for the causee 'her grandmother':

<sup>77</sup> Languages employ various strategies in this case; Dixon (2012: 256 263) lists five.

<sup>78</sup> In (242) the object is preverbal, therefore it does not have the acc marker (§8.4.1).

8.12 The causative

(245) *He* pred *haka* caus *ꞌaꞌamu* tell *te* art *aŋa* do *i* acc *tōꞌona* poss.3sg.o *māmārūꞌau* grandmother *era* dist *ko* prom *Kena.* Kena 'She always made her grandmother Kena tell (stories).' [R380.015]

The DO position is also available for the causee when the Patient is realised as instrument (§8.6.4.3). This is what happens in the following phrase: the Patient *moni* 'money' is realised not as DO but as instrument; the DO position is available for the causee *te taŋata* 'the people'.

(246) *te* art *nuꞌu* people *haka* caus *ꞌauhau* pay *i* acc *te* art *taŋata* man *hai* ins *moni* money *mo* for *te* the *rōmano* Roman 'those who made the people pay money for the Romans (= the tax collectors)' [Luke 7:34]

### **8.12.4 Reflexive and implicit causatives**

Causatives of intransitive predicates may be used reflexively: the causee is a pronoun which is coreferential with the Agent. The causative verb *haka paka* (lit. 'cause to be conspicuous') means 'to honour, praise'; in (247) it is used reflexively in the sense 'to praise oneself, to brag', and the object *i a ia* is coreferential to the implied subject.

(247) *Ku* prf *mate* die *atu* away *ꞌā* cont *a* prop *au* 1sg *ꞌi* at *te* art *kata* laugh *ꞌi* at *tū* dem *haka* caus *paka* conspicuous *era* dist *i* acc *a* prop *ia.* 3sg 'I laughed my head off from his bragging.' [R230.172]

Another example is the following causative of the copula verb *riro*:

(248) *He* ntr *haka* caus *riro* become *i* acc *a* prop *ia* 3sg *pa* like *he* pred *pīkea.* crab 'She made herself like a crab.' [R310.050]

More commonly, the reflexive object is not expressed. This means that the clause is on the surface no different in argument structure from a clause with the corresponding simple verb: the subject/causer is coreferential with the implied object, which is identical to the subject of the original verb.


This is illustrated in the following two pairs of examples, first with *riro* 'become', then with *takataka* 'gather':


These 'implicit reflexives' are part of a larger phenomenon: in many cases, causatives do not add a new argument to the verb, so the argument structure of the root is not modified. What addition of *haka* does in such cases, is adding a semantic element, usually an element of agentivity, activity or intensity. For example, while *ꞌui* means 'to ask', *haka ꞌui* is used in the sense 'to ask persistently and/or repeatedly, to inquire'. Both verbs have the same argument structure, but the causative verb is more intensive.

(254) *He* pred *haka* caus *ꞌui* ask *mai* hither *te* art *aŋa* do *e* ag *te* art *ŋā* pl *poki* child *repahoa* friend *ōꞌoku* poss.1sg.o *pē* like *nei* prox *ē…* thus

'My friends kept asking me as follows…' [R380.042]

*ꞌAvaꞌava* means 'to be at a distance' or 'to move away, to withdraw'; *haka ꞌavaꞌava* also has the latter sense, but underlines that the act of withdrawing is volitional. Compare the following pair of examples:


The same phenomenon can be observed with adjectives. While *haka* + adjective may be a true causative, expressing that the property is brought about by an external Agent (see (237–238) above), it may also express that the subject reaches a state or acquires a property through intentional action. A few examples:

8.12 The causative



*Haka* + adjective or adverb may also indicate that the subject acts in a way characterised by the root. As (261) shows, this may involve simulating a certain characteristic.<sup>79</sup>


### **8.12.5 Lexicalised causatives**

A number of *haka* forms have a meaning which cannot quite be predicted from the meaning of the root. Some examples:


Another example is found in (259) above: *haka honu* 'bodysurfing' is a lexicalised causative from *honu* 'turtle'. The same sentence also contains the verb *hakatiu* 'to watch'; even though this is formally a causative, there is no word *tiu* in the Rapa Nui lexicon. There are a few more *haka* forms for which the root as such is not a Rapa Nui word. This includes two very common words: *hakaꞌou* 'again', *hakarē* 'to leave (abandon; permit)'.

<sup>79</sup> Cf. Moyle (2011: 14) about "similative use" of the causative prefix in Takuu.

### **8.12.6 The causative prefix with nouns**

When the root of the *haka* construction is a noun, the causative expresses an action which is in some way characterised by the noun. The noun may be the product of the action: *haka N* = 'to cause the object to be N, to make something into N', or more generally 'to make/create N':


Other relationships to the noun are possible, whether conventional or creative. In (265) the noun refers to something used in the action, or something characterising the direct object as a result of the action. In (266) the noun is used in a figurative way (cf. English 'cheeky').


### **8.12.7 Lexical causatives**

The term *lexical causative* refers to a situation where there are two lexemes, unrelated in form, one of which is semantically the causative of the other. (Dixon 2012: 248.)

There are very few lexical causatives in Rapa Nui; the following two are possible candidates.

**8.12.7.1** *Hāŋai* 'to feed' can be considered a causative of *kai* 'to eat'. Apart from the obvious semantic relationship between the two, there are two reasons to assume a causative relationship between the two:


8.13 Conclusions

(267) *I* pfv *tuꞌu* arrive *era* dist *he* ntr *hāŋai* feed *i* acc *a* prop *Ure* Ure *ka* cntg *oti* finish *rō.* emph 'When she arrived, she fed Ure completely.' [R310.291]

But when the object of eating is implied, the causee is marked with *ki*. 80

(268) *He* ntr *toꞌo* take *mai* hither *tū* dem *kai* food *era,* dist *he* ntr *hāŋai* feed *ki* to *tū* dem *ŋā* pl *matuꞌa* parent *era* dist *o* of *Tiare.* Tiare 'They took the food and fed (it) to Tiare's parents.' [R238.009]

Not all instances of *hāŋai* can be considered as lexical causatives, though: the verb is also used in the sense 'to raise/tend (animals); to raise/rear (children)'.

**8.12.7.2** Another possible lexical causative is *tiŋaꞌi* (var. *tiaŋi*) 'to kill', causative of *mate* 'to die':

(269) *He* ntr *tiŋaꞌi* kill *i* acc *te* art *taŋata,* man *i* acc *te* art *viꞌe,* woman *i* acc *te* art *poki.* child 'They killed men, women and children.' [Mtx-3-01.250]

Apart from the sense 'to kill', *tīŋaꞌi* also has a different (though obviously related) sense: 'to strike, hit'. Note also that the sense 'to kill' is also expressed occasionally by the morphological causative *haka mate*.

**8.12.7.3** A number of causative verbs were borrowed as a whole from Tahitian. These are clearly recognizable as borrowings, as they start with *haꞌa-* (< Tah. *faꞌa*-) rather than *haka*. Most of these are isolated lexical items, the root of which does not occur on its own in Rapa Nui: *haꞌatura* 'to obey, respect' (Tah. *tura* 'respect.N'); *haꞌatiꞌa* 'to permit' (Tah. *tiꞌa* 'to stand'). For most of these words, it is not at all obvious that *haꞌa-* has a causative sense in Rapa Nui.

For a few *haꞌa-* forms, however, the root as such was also borrowed into Rapa Nui: *ꞌī* 'to be full', *haꞌaꞌī* 'to fill'. As *haꞌa-* is not a productive prefix in Rapa Nui, *haꞌaꞌī* can be considered as a lexical causative of *ꞌī*, rather than a form derived through prefixation of *haꞌa-*.

### **8.13 Conclusions**

This chapter has explored the expression of core constituents of verbal clauses.

Rapa Nui patterns with other Polynesian languages in that the S/A argument is marked with *e* or unmarked, while the O argument is marked with *i* or unmarked. However, the resulting case marking patterns are different from those in other languages. At first sight Rapa Nui may seem to have ergative traits, but a close analysis shows that the language is unambiguously accusative. The case marking patterns which seem to deviate from regular accusativity can be explained by the following features:

<sup>80</sup> There are no examples in the corpus where the causee and the object of eating are both expressed.


Agent marking is determined by an interplay of heterogenous factors: syntactic (preverbal subjects are always unmarked), lexico-semantic (some verbs show a strong preference for *e-*marked Agents) and pragmatic (Agents which start to act, tend to be *e*marked). The same is true for object marking: the object marker is omitted under certain conditions, which may be syntactic (OV clauses), lexico-semantic (with certain verbs) or pragmatic (non-salient objects).

Rapa Nui has a passive construction, in which the Patient is expressed as subject while the Agent is an optional oblique (but without morphological changes in the verb). In fact, passivisation in Rapa Nui is part of a wider phenomenon: several (groups of) verbs exhibit variation in argument assignment. For example, the verb *ꞌī* 'to be full' has two argument structures, with the Container and the Substance as subject, respectively. Variable argument structure can also be observed with transfer verbs like 'to feed' and 'to throw': with these verbs, either the Patient or the Goal/Recipient is expressed as direct object; the other argument is expressed as an oblique. When the Patient is oblique, it is marked as an instrument ('he threw the enemy with a spear').

Another argument-related operation is the addition of an external Agent to intransitive verbs; this Agent is marked with the preposition *i*.

Rapa Nui has a number of different comitative constructions ('A with B'). In most of these, a comitative marker is used, followed by the prominence *ko*; this marker is often a plural pronoun ('Makemake they *ko* Haua') or collective marker ('the people together *ko* the priest'). These comitative markers are used in an inclusory way: their number corresponds the total set of referents of both noun phrases. Similar are constructions with comitative sense – but without a comitative marker – in which the first noun phrase is an inclusory pronoun: 'we *ko* the child', meaning 'the child and I'.

The final topic of this chapter is causativisation. Causativisation is very common in Rapa Nui; moreover, it is very versatile:


## **9 Nonverbal and copular clauses**

### **9.1 Introduction**

This chapter deals with clauses which do not have a lexical verb as predicate. These clauses contain either no verb, an existential verb, or a copula verb.

The following types can be distinguished and will be discussed in turn:


### **9.2 NP NP clauses**

When a nominal clause consists of two noun phrases, one of them is the subject; for the other noun phrase, there are two possibilities: it may either be referential or nonreferential. When the noun phrase is non-referential, it is a true predicate, which gives new information about the subject, expressing that the subject belongs to a certain class. When the non-subject noun phrase is a referential noun phrase, the clause establishes a relation of identity between the two noun phrases, expressing that both are descriptions of the same referent. In this grammar, these two constructions are labelled classifying and identifying clauses, respectively.<sup>1</sup>

In Rapa Nui, these two types of clauses are distinguished by the use of the predicate marker *he* in classifying clauses and the preposition *ko* in identifying clauses.

A third type of NP NP clauses, attributive clauses, is characterised by the absence of any prenominal marker and the presence of an adjective in the predicate NP.

<sup>1</sup> Various terms are used in the literature. Dryer (2007a: 233) distinguishes between "equational clauses" and "true nominal predicate clauses". The distinction is fundamental in some Polynesian languages; terms used in Polynesian linguistics include: classifying and equative predicates (Bauer, Parker & Evans 1993: 78), predicational and identificational NPs (Chung, Mason & Milroy 1995: 430), predicate nominals and equatives (De Lacy 1999), class-inclusion and equational sentences (Cook 1999: 45).

9 Nonverbal and copular clauses

### **9.2.1 Classifying clauses**

In classifying clauses, a nominal predicate provides information about the subject by expressing that the subject belongs to a certain class of entities. The predicate is introduced with *he*, which indicates non-referentiality (§5.3.4).

The unmarked order in these clauses is Subject–Predicate.


The predicate may also come first. This happens only when the subject is well-established, i.e. topical in discourse; it tends to be expressed by a pronoun as in (4), or a generic noun phrase as in (5). In this construction, the predicate is prominent. In (5), for example, the predicate conveys unexpected, surprising information.


In (6), Tangaroa (who has transformed himself into a seal, and is mistaken for a seal by the people) wants to emphasise that he is the king, not a real seal as the people think. The predicate *he ꞌariki* is counterexpectative and occurs before the subject.<sup>2</sup>

(6) *He* ntr *raŋi* call *mai* hither *te* art *reꞌo* voice *o* of *te* art *pakia:* seal *'He* pred *ꞌariki* king *au* 1sg *ko* prom *Taŋaroa'.* Tangaroa 'The voice of the seal cried: I am king Tangaroa.' [Mtx-1-05.008]

Just as in verbal clauses, the subject of classifying clauses may be left out:

<sup>2</sup> Notice that *ko Taŋaroa*, which is an apposition to the predicate, is not fronted but remains in its post-subject position; see sec. §9.2.5 for more examples of split predicates.

9.2 NP NP clauses


### **9.2.2 Identifying clauses**

Identifying clauses serve to identify the referent of one noun phrase with the referent of the other noun phrase in the clause. Both NPs are preceded by a *t*-determiner (§5.3.2) such as the article *te*, indicating that they are referential. In all identifying clauses, one noun phrase is preceded by the prominence marker *ko* (§4.7.12).

A few examples:


'The first people of Rapa Nui were king Hotu Matu'a with his family.' [R350.015]

Notice that the *ko*-marked NP, in the case of a common noun, is always followed by a postnominal demonstrative *nei*, *ena* or *era*; the combination of the article *te* with one of these demonstratives indicates definiteness (§4.6.3.1).

As both noun phrases are referential and definite, and both refer to the same entity, it is not always clear which NP is subject and which is predicate. Constituent order cannot be used as the sole criterion, as both subject and predicate of a nominal clause may come first.<sup>3</sup> It is even questionable whether the term *predicate* is appropriate at all in identifying clauses (see Anderson 2004: 440): as both noun phrases are referential expressions, they are fundamentally different from predicates, which designate properties or events rather than referring to entities.

<sup>3</sup> See examples (1–6) in classifying clauses; the same is true in other types of nominal clauses, e.g. locative clauses (§9.4.1).

#### 9 Nonverbal and copular clauses

Even so, the terms *subject* and *predicate* may be used in identifying clauses in a loose way, in the sense that the subject is the entity to be identified, and the predicate is the identifying expression. In some cases it is clear which NP is the subject, as this NP functions as discourse topic. In other cases, however, it is difficult to identify subject and predicate – unless we adopt a simple syntactic definition. As indicated above, in every identifying clause one noun phrase is marked with *ko*, while the other is an unmarked NP. Taking the *ko*-marked NP as predicate provides a simple criterion. Moreover, this analysis coincides with the intuitive assignment of subject and predicate in those cases where the distinction is clear: in examples like (11), it is clear that the unmarked NP is subject, while the *ko*-marked NP serves to identify this subject.

In the examples so far, the identifying clause consists of two common noun phrases. When the clause contains a pronoun or proper noun, the use of *ko* is described by the following two rules:


This is illustrated in the following examples. Common NP + proper noun:

(12) *Te* art *kona* place *hope ꞌa* last *o* of *te* art *nehenehe* beautiful *ko* prom *ꞌAnakena.* Anakena 'The most beautiful place (of the island) is Anakena.' [R350.013]

Pronoun + common NP:

(13) *Pero* but *ko* prom *au* 1sg *te* art *suerekao* governor *o* of *te* art *hora* time *nei.* prox 'But I am the governor now (or: the governor now is me).' [R201.007]

Pronoun + proper noun:

(14) *A* prop *au* 1sg *ko* prom *Omoaŋa.* Omoanga 'I am Omoanga.' [R314.101]

These patterns make sense if we assume that *ko* always marks the predicate. Proper names are inherently highly identifiable (their reference is always unique and unambiguous in a given context), so it is not surprising that they serve as an identifying expression (predicate) rather than as a referent to be identified (subject). The same is true for pronouns. Between proper nouns and pronouns, the former are identifiable to a higher

<sup>4</sup> I have not found any exceptions to this rule in the text corpus, though there are a few exceptions in the New Testament translation.

9.2 NP NP clauses

degree: within a given context, a proper noun has unambiguous unique reference; for a pronoun, more contextual clues may be needed to establish its reference. This can be represented in a *hierarchy of identifiability*:

(15) proper nouns > pronouns > common nouns

The idea that *ko* marks the predicate is also confirmed by the fact that an identifying clause may consist of a *ko*-phrase only; this follows from the general rule in Rapa Nui that the predicate is obligatory, while the subject can be omitted:


In (14) above, the pronoun is not marked with *ko* when the other constituent is a proper noun. There are also a few cases in the corpus where a pronoun and a proper noun are both *ko*-marked. Two examples are provided below:


If the pronoun is taken as the subject, these clauses are counterexamples to the claim that only the predicate is marked with *ko*. However, a different analysis is also possible: the pronoun can be analysed as the predicate (with implicit subject), with the proper noun added as apposition, 'It's me, Totimo'. In both examples above this analysis is plausible. In (18), for example, the situation is as follows: there is a blind girl, Mahina Tea, who knows a boy called Totimo. Totimo walks up to her, embraces her and utters the clause quoted here. An analysis as predicate + apposition is appropriate here.<sup>5</sup>

In other cases this analysis is less plausible, as in the following exchange:

<sup>5</sup> This analysis is reinforced by the fact that in some cases the two constituents are separated by a comma:

<sup>(</sup>i) *Ko* prom *au,* 1sg *ko* prom *Hotu* Hotu *ꞌIti* Iti *te* te *Mata ꞌiti* Mata'iti *ꞌa* of.a *Hotu* Hotu *Matuꞌa.* Matu'a 'It's me, Hotu Iti te Mata'iti, son of Hotu Matu'a.' (Ley-2-08.025)

9 Nonverbal and copular clauses

(20) *—¿Ko* prom *ai* who *koe?* 2sg *… —¡Ko* prom *au* 1sg *nei* prox *ko* prom *Vaha* Vaha *ko* prom *toꞌo* take *i* acc *a* prop *Huri* huri *ꞌa* a *Vai!* Vai *… —¡ꞌE* and *ko* prom *au* 1sg *nei* prox *ko* prom *Kaiŋa* Kainga *ko* prom *toꞌo* take *i* acc *a* prop *Vaha!* Vaha '—Who are you? —I am Vaha, who takes (=kills) Huri a Vai! —And I am Kainga, who takes Vaha!' [R304.97-101]

Especially in the last clause, an appositional analysis doesn't appear to be appropriate. Possibly, these constructions can be analysed as topic + comment constructions (§8.6.1.3): '(As for) me, I'm Kainga.'<sup>6</sup>

### **9.2.3 Comparing classifying and identifying clauses**

In the examples of classifying clauses in §9.2.1 above, the predicate NP clearly indicates that the subject belongs to a certain class of entities; the subject is part of a category described by the predicate.

In some cases however, the class of entities described by the predicate has only one member, i.e. this class coincides with the referent of the subject. This is illustrated in the following examples:


These clauses are very similar in sense to identifying clauses, which express that two noun phrases have identical reference (§9.2.2). In fact, in most examples above, the predicate is translated with a definite noun phrase in English, which is characteristic of an identifying clause. Some examples of identifying clauses are very similar to the classifying clauses above:

<sup>6</sup> De Lacy (1999: 47) discusses cases in Māori where both constituents are *ko*-marked; these are different in that both constituents are a (long) common noun phrase. This enables De Lacy to analyse these as clefts, i.e. biclausal constructions.

9.2 NP NP clauses

(24) *He* ntr *ꞌite* know *ia* then *Tuꞌu* Tu'u *Koihu* Koihu *ꞌi* at *tū* dem *hora* time *era* dist *tū* dem *ŋā* pl *nuꞌu* people *era* dist *ko* prom *tū* dem *ŋā* pl *tahutahu* witch *era.* dist

'At that moment Tu'u Koihu knew that those people (whom he saw) were those witches.' [R233.023]

	- fibre

'(in a description of fishing techniques:) The lines they fished with, were those lines braided with three strands.' [R539-1.218]

These examples show that the choice between the two constructions in Rapa Nui is not determined by the criterion of uniqueness, that is, whether or not the predicate defines a single unique entity.<sup>7</sup> Rather, classifying constructions serve to *describe* the subject by giving new information about it, while identifying clauses serve to *identify* a referent with an entity already known to the hearer. The referent of the identifying noun phrase must be accessible to the hearer, otherwise a classifying construction with *he* is used.

To give an example, in the context preceding (21) above, there has been no mention of the son of the sun and the high God, so the hearer does not necessarily know that there is such a person as the child of the sun, or that the people in the story had a high God at all. Therefore, this person is not accessible to the hearer. By contrast, in (24), 'those witches' refers to witches who have been mentioned earlier in the story; the identifying clause enables the hearer to identify this known entity with the subject 'those people'. Similarly, in (25) the speaker refers to a type of fishing line which he assumes to be known by the hearer (even though it has not been mentioned in the text itself).

The referent of a noun phrase in an identifying clause must not only be unique and accessible, it also needs to be a specific, bounded entity. In the following two examples, the predicate noun phrase could be considered as unique and accessible; nevertheless, it is marked with *he*, i.e. the construction is a classifying clause. In (26), the noun phrase refers to priests in general, not to any specific priest. Likewise, in (27), the reference is to adults in general.<sup>8</sup>

(26) *He* pred *iviꞌatua* priest *nō* just *te* art *kope* person *era* dist *e* ipfv *puā* touch *era* dist *ꞌi* at *te* art *taꞌu* year *era* dist *i* at *a* prop *ia* 3sg *te ꞌao.*

art reign

'The priest was the only person (lit. only the priest was the person) who would touch him (te bird man) in the year in which he reigned.' [R641.008]

<sup>7</sup> Lyons (1999) mentions uniqueness as one of the necessary conditions for definiteness. Uniqueness is defined as: "there is only one entity satisfying the description used, relative to the context."

<sup>8</sup> These examples are cleft constructions, which are discussed in more detail in sec. §9.2.6 below.

#### 9 Nonverbal and copular clauses

(27) *Te* art *ŋā* pl *poki* child *ko* prf *ꞌite* know *ꞌana* cont *o* of *ruŋa* above *i* at *te* art *meꞌe* thing *taꞌatoꞌa* all *o* of *te* art *naonao* mosquito *ꞌe* and *he* pred *paꞌari* adult *te* art *meꞌe* thing *i* pfv *taꞌe* conneg *ꞌite.* know

'The children know everything about the mosquitoes, and the adults are the ones who don't know.' [R535.159]

We may conclude that a nominal predicate designating an accessible, individuated, bounded entity is marked with *ko*; in all other cases, a classifying construction with *he* is used. A similar difference between *ko* and *he* can be observed with topicalisation in verbal clauses (§8.6.2).

As the examples above show, classifying predicates always consist of a common noun phrase. Proper nouns and pronouns never serve as a classifying predicate; in fact, they are never preceded by *he*. This is to be expected, as proper nouns and pronouns by definition qualify as identifying predicates: they have unique reference, are accessible in the context, and refer to a specific, bounded entity. On the other hand, they do not designate a class of entities, hence are not suitable as classifying predicates.

### **9.2.4 Constituent order in identifying clauses**

The order of constituents in an identifying clause depends to some extent on the type of noun phrases involved. When both constituents are common noun phrases, the predicate usually occurs after the subject, as illustrated in (24–25) and (11) above. The predicate may come first when it conveys significant and possibly surprising information as in (28), or when it is a discourse topic established in the preceding context as in (29).<sup>9</sup>

(28) *Taꞌe* conneg *he* pred *ꞌatua* god *tau* dem *meꞌe* thing *era,* dist *he* pred *taŋata;* person *ko* prom *te* art *ŋā* pl *io* youngster *era* dist *ꞌĀmai* Amai *tau* dem *meꞌe* thing *era.* dist

'These beings are not gods, they are men; these beings are the Amai guys.' [Mtx-7-37.029]

(29) *Ko* prom *te* art *meꞌe* thing *nei* prox *te* art *meꞌe* thing *uꞌi* see *rahi* much *o* of *te* art *muꞌa* front *ꞌā.* ident 'This (= the difficulties mentioned in the previous clause) was something often seen in the past.' [R107.009]

When the identifying clause contains a pronoun (whether subject or predicate), this is always in an initial position, as illustrated in (13–14) above.

When the clause consists of a proper noun and a common noun phrase, they may occur in either order, as the following examples show. Putting the predicate before the subject

<sup>9</sup> Cf. Levinsohn (2007): prominence may involve both new information (focal prominence) and established information (thematic prominence).

9.2 NP NP clauses

gives it more prominence. In (32), for example, the predicate *ko Korikē* is contrasted with other persons. In (33), Anakena is singled out between other places on the island.


med

'Anakena was the only place where the people who lived there did not get sick.' [R231.098]

### **9.2.5 Split predicates**

When a clause has a pronominal subject and the predicate comes first, certain postnominal modifiers of the predicate are placed after the subject. In (34), *ōꞌou* is a postnominal possessive modifying the predicate *repahoa*; it is separated from the nucleus by the subject *au*.

(34) *He* pred *repahoa* friend *nō* just *au* 1sg *ōꞌou.* poss.2sg.o 'I am just your friend.' [R308.032]

This predicate split is obligatory; clauses like the following do not occur:

```
(35) *He repahoa nō ōꞌou au.
```
As discussed in §4.6.6, this process also takes place when the subject is a demonstrative pronoun; this is illustrated in (36–37) below. The stranded element is often a possessive as in (34); it may also be a numeral as in (36), or a relative clause as in (37). While in (37) the relative clause as a whole is separated from the head noun *famiria*, in (38) the relative clause itself is split up: the verb phrase (*aŋa mau ꞌā*) moves along with the head noun, while the direct object is stranded after the subject (§11.4.5 on raising of relative clause verbs).

9 Nonverbal and copular clauses


Split constituents also occur around the particle *ia* 'then' (§4.5.4.1), which occurs after the first constituent of the clause; postnuclear elements occur after *ia*:


Clark (1976: 119–120) analyses this process as extraposition of the second constituent of the predicate over the subject. Alternatively, the split can be described as movement of the predicate with stranding of the postnominal modifier; fronting of a constituent is a common process (both crosslinguistically and in Rapa Nui), while it is difficult to see why a modifier would be moved to the right.

### **9.2.6 Clefts**

A cleft construction consists of two noun phrases, one of which is a simple noun phrase, while the other contains a relative clause, often without head noun (Payne 1997: 278). Clefts are formally identifying clauses – their main constituents are two coreferential NPs – but they express an event or action; the latter is relegated to the relative clause. The effect of a cleft construction is to put the simple NP in focus.

In Rapa Nui cleft constructions, the simple NP comes first and is marked with *ko*, as is expected with predicates of identifying clauses (§9.2.2). The second noun phrase contains an anchor noun functioning as head of the relative clause; this is either a repetition of the noun in focus, or a generic noun like *meꞌe* 'thing'. The cleft construction is thus similar to the English construction 'Mary was the one who won',<sup>10</sup> though a noun is used instead

<sup>10</sup> Cleft constructions of the type 'X was the one who…' are often called pseudo-clefts (Payne 1997: 279; Bauer 1991: 9 for Māori). On the question whether Rapa Nui also has "real" clefts, i.e. without anchor noun, see sec. §8.6.2.1.

9.2 NP NP clauses

of 'one' and there is no copula verb. As in all relative clauses, the verb is usually marked with *i*, *e* or unmarked (§11.4.3).

A few examples:


'These things (which have just been listed) are the things that people need to go diving nowadays.' [R360.002]


The effect of relegating the verb to a relative clause is that the initial noun phrase is in focus, while the event or action is backgrounded. Clefts are used when the event or action as such is presupposed; it has already been mentioned as in (42), or can be inferred from the context: in (41), the context of a musical contest presupposes that there is a winner, while the important new information is the identity of the winner. The act of winning is therefore backgrounded, while the noun phrase referring to the winner is put in focus.

The examples so far represent the most common construction, in which clefts are constructed as identifying clauses with a *ko*-marked predicate. Clefts may also be classifying clauses, with a *he*-marked predicate. As discussed in §9.2.1, identifying clauses are used when the predicate refers to a unique individual which is accessible to the hearer; in other cases, classifying clauses are used. This happens for example when the noun phrase is generic:

(45) *Te* art *ŋā* pl *poki* child *ko* prf *ꞌite* know *ꞌana* cont *o* of *ruŋa* above *i* at *te* art *meꞌe* thing *taꞌatoꞌa* all *o* of *te* art *naonao* mosquito *ꞌe* and *he* pred *paꞌari* adult [*te* art *meꞌe* thing *i* pfv *taꞌe* conneg *ꞌite* ]*.* know

'The children know everything about the mosquitoes, and the adults are the ones who don't know.' [R535.159]

#### 9 Nonverbal and copular clauses

Classifying cleft constructions are especially common with the verb *haŋa* 'want' and other expressions of volition/desire (§3.2.3.1.1 on the nominal tendency of volition verbs). With these verbs, the noun phrase does not contain a full relative clause, but a bare modifying verb, such as *haŋa* in (46); if the subject of this verb is expressed, it is a possessive pronoun (*tāꞌaku* in (46)) or a genitive phrase (§11.4.4):


'For him, the thing (he) liked most was climbing his horse (and going around the island).' [R439.008]

Clefts also occur in questions, when a verb argument is questioned: identifying clefts with *ko ai* 'who' (§10.3.2.1), classifying clefts with *he aha* 'what' (§10.3.2.2).

As discussed in §8.6.3, the actor-emphatic (AE) construction also serves to put a noun phrase in focus. It is not entirely clear which conditions determine the choice between an AE construction and a cleft. However, AE's are only used to put agentive subjects in focus; in order to put non-agentive subjects in focus as in (42) or non-subjects as in (46), only clefts can be used.

### **9.2.7 Attributive clauses**

In an attributive clause, an inherent – and usually permanent – property is attributed to the subject.<sup>12</sup> This property is in most cases expressed as an adjective. Now an adjective as such cannot serve as a nominal predicate in Rapa Nui, and therefore an anchor noun is needed to fit the adjective into the syntactic structure. This anchor noun is either identical to the subject noun or a generic noun like *meꞌe* 'thing'.<sup>13</sup>

The predicate may be marked with *he* as in (48), in which case the clause is a classifying clause (§9.2.1). This is rare, though; usually the predicate is a bare noun phrase, lacking any determiner.

Below are some examples, with the anchor noun emphasised.

With repetition of the subject noun:

(48) *Te* art *ꞌati* problem *ena* med *o* of *te* art *kahu* clothes *ꞌi* at *rā* dist *noho* stay *iŋa* nmlz *he* pred *ꞌati* problem *nuinui* big:red *e* num *tahi.*

one

'The problem of clothing at the time was a big one.' [R380.093]

<sup>11</sup> *Manava mate* is an idiom expressing love or endearment.

<sup>12</sup> Non-permanent properties are expressed as verbal predicates, see sec. §3.5.1.5.

<sup>13</sup> In related languages, cognates of *me ꞌe* also serve as anchor noun for adjectival or verbal predicates; see e.g. Lazard & Peltzer (2000: 38) on Tahitian.

9.2 NP NP clauses

(49) *Taŋata* person *ꞌuriꞌuri* black:red *te* art *taŋata* person *nei* prox *ꞌe* and *taŋata* person *rakerake.* bad:red 'This man is dark and ugly.' [R372.133]

With a generic noun:


These examples show that, as in other nominal clauses, either the subject may come first as in (48) and (50), or the predicate as in (49) and (51).

In the examples above, the property is an adjective. It may also be another type of noun modifier: a verbal clause as in (52–53), or a modifying noun as in (54).


As (52–53) show, the modifying verb may be preceded by preverbal particles, including the negator *taꞌe*.

As in other clause types, the subject of attributive clauses may be omitted:


Finally, Rapa Nui has a somewhat peculiar construction consisting of a bare noun phrase headed by *meꞌe* or another generic noun, followed by a *he*-marked NP. This construction is not very common, but entirely grammatical. It is especially used to express general truths.

9 Nonverbal and copular clauses


This construction is unusual in that both noun phrases seem to be marked as a nominal predicate. However, a more plausible analysis is also possible: the construction may be a subjectless attributive clause, in which the predicate *meꞌe X* is followed by an apposition introduced by *he*. (57) could be paraphrased as 'It's (a) mortal (thing), man is.' This appositional analysis is suggested by the use of *he* (see §5.12.1 for the use of *he* in appositions), and by the fact that the *he*-marked NP always occurs after the *meꞌe* phrase.

### **9.3 Existential clauses**

Existential clauses state the existence of a person or thing. In Rapa Nui, they are either constructed as a verbless clause or with the existential verb *ai*. 14

### **9.3.1 Verbless and verbal existential clauses**

Verbless existential clauses contain only one core consituent, which is introduced by *he*; the use of *he* shows that this constituent is predicate rather than subject.<sup>15</sup> This means that existential clauses conform to the general rule that the predicate is the only obligatory constituent.

(60) *He* pred *taŋata* man *ko* prom *Eŋo.* Engo 'There was a man (called) Engo.' [Mtx-7-28.001]

<sup>14</sup> In this respect, Rapa Nui shows characteristics of both EP languages (where existential clauses are verbless, with a *he*-marked Existee as in Rapa Nui), and non-EP languages (where existential clauses are constructed with the verb *ai*/*iai* (Clark 1976; 1997: 101.

<sup>15</sup> According to Dryer (2007a: 241), it is in many languages unclear whether the theme of an existential clause should be considered a subject. In many languages, it is clear that the theme is not subject, e.g. in European languages like Dutch ('Er is een hond in de tuin' = there is a dog in the garden, rather than 'Een hond is in de tuin') and French ('Il y a un chien dans le jardin' = there is a dog in the garden).

9.3 Existential clauses

(61) *He* ntr *repa* young\_man *e* num *rua* two *te* art *ꞌīŋoa* name *ko* prom *Makita* Makita *ko* prom *Rokeꞌaua.* Roke'aua 'There were two young men, named Makita and Roke'aua.' [R243.001]

The noun phrase may contain a prenominal numeral; as discussed in §5.3.5, prenominal numerals are in determiner position, hence they replace the predicate marker *he*:

(62) *E* num *tahi* one *poki* child *te* art *ꞌīŋoa* name *ko* prom *Eva* Eva *ka* cntg *hoꞌe* one *ꞌahuru* ten *matahiti.* year 'There was a child called Eva, ten years old.' [R210.001]

Existential clauses can also be expressed with the verb *ai* 'to exist', with the Theme or Existee as subject of the clause. This construction is rare in older texts, but in modern Rapa Nui it is more common than the verbless construction.

Usually *ai* has continuous aspect marking *e V ꞌā/ꞌana* (§7.2.5.4), while the verb phrase also has the emphatic particle *rō*. *E ai rō ꞌā/ꞌana* is such a common combination that it almost seems to be a frozen expression.


However, *ai* is used with other aspectuals as well, for example neutral *he* (65) and exhortative *e* (66):

(65) *ꞌI* at *tōꞌona* poss.3sg.o *mahana* day *he* ntr *ai* exist *mai* hither *te* art *aŋa* work *he* pred *ꞌāua* fence *titi,* build *ꞌo* or *he* pred *rau* leaf *kato…* pick

'On some days there was work: building fences or picking leaves…' [R380.084]

(66) *Mo* if *oho* go *e* num *tahi* one *taŋata* person *ki* to *tai,* sea *e* exh *ai* exist *te* art *meꞌe* thing *taꞌatoꞌa* all *o* of *te* art *hī.* to\_fish 'If someone goes to the sea, he needs all the fishing gear (lit. there should be all the things of fishing).' [R354.002]

9 Nonverbal and copular clauses

### **9.3.2 Existential-locative clauses**

Many existential clauses do not just state the existence of something, but rather its existence in a certain place: 'There is water here'. These clauses can be labelled 'existentiallocative'.<sup>16</sup>

Just like plain existential clauses, existential-locative clauses may be either verbless as in (67–68) or verbal as in (69–70). In older texts, they are always verbless.


As the examples above show, the locative adjunct in these constructions is often introduced by *to* (in older Rapa Nui) or *o* (in modern Rapa Nui).<sup>17</sup> The possessive preposition *o*, when used in a locative construction, often indicates that a referent belongs to a certain place, i.e. comes from that place or is located there permanently. It may, however, also indicate the location of a referent at a given moment, and therefore be similar in sense to *ꞌi* (see (44–46 in §6.3.1).

### **9.3.3 Possessive clauses**

Possessive clauses establish a relationship of possession between two entities:<sup>18</sup> 'John has a book' expresses that John is the possessor of a book. In Rapa Nui, this relation is expressed by an existential clause,<sup>19</sup> in which the possessee noun phrase is modified by a possessor; the construction can be paraphrased as 'John's book exists' or 'There is John's book.'

<sup>16</sup> These are different from locative clauses, which predicate the location of a certain referent: 'The water is here.' Rapa Nui, like many other languages, employs different constructions for these two clause types. See Dryer (2007a: 241) for general discussion.

<sup>17</sup> *To* is a contraction of the article *te* + the genitive preposition *o* (§6.2).

<sup>18</sup> Possessive clauses ('John has a book') are different from proprietary clauses ('The book is John's', §9.4.2). See Clark (1969).

<sup>19</sup> This is common in many languages, see Dryer (2007a: 244).

9.3 Existential clauses

In modern Rapa Nui, possessive clauses are constructed as verbal existential clauses, in which the existential verb *ai* takes the possessee as subject. (71) is literally 'His house in Hanga Roa existed', (72) is 'Two their children existed'.


As these examples show, the possessor is expressed in the subject noun phrase: it is either a possessive pronoun as in (71–72), or a possessive noun phrase as in (73). (For more details, see §6.2.1 on possessives in the noun phrase, §6.3.1 on the semantic range of possessive constructions, and §6.3.2 on the choice between *o* and *ꞌa*.)

The clause may be preceded by a noun phrase coreferential to the possessor; this happens especially when the possessor is a full noun phrase. This noun phrase is leftdislocated and is syntactically not a constituent of the clause that follows; the clause as a whole is a topic-comment construction (§8.6.1.3). (74) can be translated literally as 'All the tribes, their leaders existed.'


In these topic-comment constructions, the possessor is often not expressed again in the subject NP. (76) is literally: 'We, money exists'; (77) is 'This woman, there were two daughters.'


#### 9 Nonverbal and copular clauses

In older texts, possessive clauses may also be constructed as a verbless existential clause. Instead of the verb *ai* with its subject, these have a *he*-marked nominal predicate. The possessor is expressed as *to* + NP or a *t-*possessive pronoun.


In modern Rapa Nui, verbless possessive clauses only occur in the following circumstances:

When the predicate noun phrase contains a numeral:

(80) *E* num *tahi* one *ōꞌoku* poss.1sg.o *hoa* friend *repa* friend *ko* prom *Hoahine* Hoahine *te* art *ꞌīŋoa.* name 'I have a friend whose name is Hoahine.' [R213.014]

When the clause is negated, using *ꞌina* (§10.5.1):

(81) *ꞌIna* neg *paꞌi* in\_fact *o* of *māua* 1du.excl *kona* place *mo* for *noho.* stay 'For we do not have a place to live.' [R229.210]

As these examples show, in these cases the possessor is a Ø-possessive pronoun within the predicate noun phrase. These clauses are different from the old constructions illustrated in (78–79), where the possessor is a separate constituent.<sup>20</sup>

### **9.3.4 Conclusion**

Whether an existential clause is verbless or verbal, depends on the type of clause: simple existential, existential-locative, or possessive. However, there is a general development over time in which verbless constructions are replaced by verbal ones. This is summarised in Table 9.1:

<sup>20</sup> If the possessives in (78–79) were part of the predicate noun phrase, the possessor would be marked with the preposition *o* in (78), and a Ø-possessive pronoun in (79).

9.4 Prepositional predicates


Table 9.1: Types of existential clauses

### **9.4 Prepositional predicates**

Various types of prepositional phrases may serve as predicate of a nonverbal clause.

### **9.4.1 Locative clauses**

Locative clauses consist of a subject noun phrase and a prepositional phrase with locative sense as predicate. Either phrase may come first. The locative phrase is often introduced by *ꞌi*, marking stationary location, possibly followed by a locational as in (82). Other prepositions may also be used, as (84) shows.

	- '—Where are you (coming) from? —From the seaside.' [R245.084]

### **9.4.2 Proprietary clauses**

Proprietary clauses (also known as "genitive predicates", Dryer 2007a: 248) consist of a subject noun phrase and a predicate expressing a possessor. In Rapa Nui, the latter is either a noun phrase marked with genitive *o* or *ꞌa*, or a Ø-possessive pronoun. (§6.3.1 on the semantic range of possessive constructions, §6.3.2 on the choice between *o* and *ꞌa*.)

#### 9 Nonverbal and copular clauses

(85) *Te* art *hare* house *nei,* prox *taꞌe* conneg *ōꞌoku;* poss.1sg.o *o* of *tāꞌaku* poss.1sg.a *māꞌaŋa* adopted\_child *ena* med *ko* prom *Puakiva.* Puakiva

'This house is not mine; it belongs to my adopted child Puakiva.' [R229.268]


As these examples show, the predicate may come after the subject as in (85–86), or before the subject as in (87–88).

Occasionally, proprietary clauses are constructed with the locative preposition *i*, which may have a possessive sense (§4.7.2.2). *I* in proprietary clauses tends to indicate possession in an abstract sense, e.g. possession of qualities or attributes; however, as (90) shows, it is also used with concrete entities.


The proprietary clause construction also serves to form nominalised actor-emphatic clauses (§8.6.3).

### **9.4.3 Other prepositional predicates**

Any prepositional phrase may serve as the predicate of a nominal clause. This results in clauses that could be labelled "benefactive" (91), "instrumental" (92) or "comparative" (93); however, these labels should not obscure the fact that these clauses simply follow the general pattern of a NP PP clause.

9.5 Numerical clauses


As with all types of nominal clauses, the constituent order is not fixed, though the subject tends to come first, as (91–93) show.

### **9.5 Numerical clauses**

In numerical clauses,<sup>21</sup> the predicate is a numeral phrase, consisting of a numeral with preceding particle (§4.3.2). The numeral predicate comes first; it is followed by the subject noun phrase.

(94) [*E* num *tahi*] one [*te* art *rāua* 3pl *poki* child *vahine* female *nehenehe* ]*.* beautiful

'They had one beautiful daughter (lit. one [was] their beautiful daughter).' [R338.001]

In this example, the numeral phrase *e tahi* is predicated of the subject *te rāua poki vahine nehenehe*. *E tahi* is not part of the noun phrase that follows, as is indicated by the determiner introducing that noun phrase; numerals within a noun phrase are never followed by a determiner (§5.4.1).

In the following example, the numeral is followed by a *t*-possessive pronoun, which occupies the determiner position in the noun phrase (§6.2.1); again, this indicates that the numeral is not part of the subject NP, but a separate constituent.

(95) *He* ntr *tuꞌu* arrive *mai…* hither *e* num *tahi* one *paiheŋa,* dog *e* num *rua* two *tōꞌona* poss.3sg.o *pūꞌoko.* head 'One day a dog came, which had two heads (lit. two its heads).' [R435.003]

The following sentence, which is superficially almost identical to (94), has a fundamentally different structure.

<sup>21</sup> See Clark (1969: 108) on this term.

9 Nonverbal and copular clauses

(96) *E* num *tahi* one *rāua* 3pl *poki* child *vahine* female *nehenehe.* beautiful

'They had one beautiful daughter (lit. one their beautiful daughter).'

This is an existential clause, which consists of a single NP containing the numeral *e tahi*; the absence of a determiner after *tahi* indicates that the numeral is part of the noun phrase. This is confirmed by the fact that the noun phrase as a whole can be used as constituent of a larger clause, for example as subject of an existential verb:

(97) *E* ipfv *ai* exist *rō* emph *ꞌana* cont [*e* num *tahi* one *rāua* 3pl *poki* child *vahine* ]*.* female 'They had one daughter (lit. there was one their daughter).' [R338.001 revised]

Numerical clauses are not very common. It is more common for a numeral to be embedded within a noun phrase, as in (96) above. This is also illustrated in (61–63) in §9.3.1.

### **9.6 Copula verbs**

Copula verbs serve to link a nominal subject to a nominal or otherwise non-verbal predicate. While copula verbs may have all the morphosyntactic trappings of a verb, they are semantically empty (Payne 1997: 115) or nearly empty. Copula verbs are unusual in Polynesian languages; the only example I am aware of concerns the contact-induced development of verbs 'have' and 'be' in Mele-Fila and Emae in Vanuatu (Clark 1986: 337; Clark 1994: 119), though there is a possible example in Hawaiian (see Footnote 24 on p. 475).<sup>22</sup> In Rapa Nui, the existential verb *ai* is used as a copula verb in some constructions. This use is absent in older texts; possibly it is developing under influence of Spanish, where copular clauses have *ser* or *estar* 'to be'. Another recent introduction is *riro* 'become', which equally functions as a copula verb. In the following sections, these verbs will be discussed in turn.

### **9.6.1** *Ai* **'to exist' as a copula verb**

*Ai* usually functions as an existential verb 'to be, exist' (§9.3.1). Existential constructions with *ai* can be analysed as intransitive verbal clause with the Existee as subject. However, *ai* is also used in a construction involving both a subject and a nonverbal predicate. This construction is uncommon, but it does occur. Examples in the text corpus are scarce; more examples are found in the Bible translation, probably due to the higher frequency of subordinate clause constructions in Biblical texts.

<sup>22</sup> Harlow (2007a: 154) mentions *ai* as a copula verb in older Māori; however, as this verb only takes a single argument, it seems to be an existential verb like Rapa Nui *ai* in existential clauses, rather than a copula. (The example *Kia ai he moenga…* is translated 'Let there be a bed…') As Dixon (2010b: 160) points out, "a defining feature for a copula verb is that it *must* be able to occur in a construction with two core arguments."

9.6 Copula verbs

At first sight, the following two examples involve a copula verb construction. The verb *ai* (preceded by the subordinators *mo* 'if' and *ana* 'irrealis', respectively) is followed by two noun phrases: a subject and a *he*-marked noun phrase. In both cases, *ai* appears to be a copula verb in a classifying clause.


Jew

'God does not consider whether you are a Jew or not a Jew.' [Colossians, introduction]

However, on a closer look, *ai* may not be a copula verb here. As it turns out, *ai* in subordinate clauses can be followed by a complete verbal clause; the latter is no different in structure from a main clause. Below are two examples, again introduced by *mo* and *ana*:


In (100–101) it is clear that *ai* is not the predicate of the clause between brackets. Rather, *ai* is an (existential) verb followed by a complete (independent) clause.<sup>23</sup> The same analysis is possible for (98) above; in that case *koe he Kiritō* is a complete (nominal) clause, in which *ai* does not play a role. The same is true for (99). If this analysis is correct, *ai* in (98–99) is not a copula verb. A compelling reason to adopt this analysis of (98) is, that the subject of a verb marked with *mo* is normally expressed as a possessive (§11.5.1.2). The fact that the subject in (98) is nominative *koe*, makes it an unlikely candidate for the subject position of the *mo-*clause.

In other cases, however, the analysis above is implausible. First, the subject after *mo ai* may be expressed as a possessive, strongly suggesting that it is indeed the subject of the *mo*-clause, hence an argument of *ai*. This suggests that *ai* in (102) is bivalent (hence copular), taking two arguments just like the transitive verb *ꞌui* in (103).

<sup>23</sup> See further §11.5.1.1 (*mo*) §11.5.2.2 (*ana*) on the use of *ai* with subordinating markers.

9 Nonverbal and copular clauses


Second, a copular analysis of *ai* is plausible when it occurs in a main clause. Although (104) below could be interpreted as existential *ai*, this is not very plausible, as there are no unambiguous examples of *ai* in main clauses followed by an independent clause expressing the Existee. A monovalent analysis is even less likely when the two noun phrases occur on either side of the verb, as in (105).


We may conclude that *ai* is occasionally used as a copula verb. Using *ai* enables a speaker to embed nominal clauses into constructions which only allow verbal clauses, for example subordinate clauses as in (102), and exhortations as in (104).

While all examples so far concern classifying clauses, other types of verbless clauses may have the copula as well. Here is an example of a locative clause. Again, the subject is possessive, as the verb *ai* is nominalised.

(106) *He* pred *koa* happy *tōꞌona* poss.3sg.o *matuꞌa* parent *ꞌo* because\_of *te* art *ai* exist *hakaꞌou* again *mai* hither [*ōꞌona* ] poss.3sg.o [*ꞌi* at *nei*]*.* prox

'Her parents were happy because she was here again.' [R441.018]

### **9.6.2** *Riro* **'to become'**

*Riro* 'to become' expresses the transformation of an entity into something else. It was borrowed from Tahitian relatively recently: *riro* is not found in older texts, the oldest occurrences are in the stories collected in the early 1970s by Felbermayer (Felbermayer 1971; 1973; 1978).

*Riro* occurs in a few stories in which a person turns into an animal. In older versions of these stories, the process of transformation is implicit and the new identity is expressed by a non-verbal clause; in new versions, *riro* is used. The following examples are from

9.6 Copula verbs

two versions of the same story, which tells about a child turning into a fish. In the old version in (107), no verb is used to describe the transformation; the new version in (108) employs the verb *riro*.

	- dist

'A wave came in and took the child. The child (became) a fish.' [Mtx-7-10.019]

(108) *He* ntr *riro* become *rō* emph *atu* away *ꞌai* subs *tū* dem *poki* child *era* dist *he* pred *ika.* fish 'The child became a fish.' [R338.006]

As (108) shows, the verb *riro* has two arguments: the subject *tū poki era* and a *he*marked noun phrase expressing the class to which the subject belongs after the transformation. Apart from the verb, the clause has the same structure as the verbless classifying clause in (107). This shows that *riro* is a true copula verb, linking two noun phrases with an identity relation. Two more examples of the same construction:


While the form and meaning of*riro* were borrowed from Tahitian, its status as a copula verb is unique to Rapa Nui.<sup>24</sup> In Tahitian, the resulting entity after *riro* is marked with the preposition *ꞌei*: 25

(111) *ꞌUa* prf *riro* become *tō* art:of *ꞌoe* 2sg *tuahine* sister *ꞌei* to *pōtiꞌi* girl *purotu.* pretty (Tahitian) 'Your sister has become a beautiful girl.' (Académie Tahitienne 1986: 272)

<sup>24</sup> There is one possible exception: for Hawaiian, Cook (1999: 63) gives an example from an old text (1918) where *he* (which is a nominal predicate marker, as in Rapa Nui) marks the resulting entity after the verb *lilo*, an argument normally marked with *i* (related to Tahitian *ꞌei* in (111)?). Apparently, this construction, which corresponds exactly to the Rapa Nui construction *riro he*, is unknown nowadays.

<sup>25</sup> Tahitian *ꞌei* has various uses, all of which have to do with a state not yet realised; see Académie Tahitienne (1986: 364–365).

9 Nonverbal and copular clauses

### **9.7 Conclusions**

This chapter has dealt with various types of clauses, all of which do not have a lexical verb as predicate. Many of these are verbless; others have either the existential verb *ai* or – occasionally – a copula verb.

Regarding clauses with a noun phrase predicate, two types can be distinguished. Classifying clauses contain a true predicate providing information about the subject by including it in a certain class; identifying clauses express an identity relation between two referents. In classifying clauses the predicate has the predicate marker *he*; in identifying clauses, it has the prominence marker *ko*. The identifying construction is only used if the predicate is already known to the hearer as an individual entity.

Rapa Nui has a cleft construction, which consists of an identifying or classifying predicate followed by a subject noun phrase containing a relative clause. Unlike other Polynesian languages, Rapa Nui requires the relative clause to contain a head noun, resulting in the construction sometimes called "pseudo-cleft".

Like clefts, attributive clauses (those with an adjectival predicate expressing an inherent property) need a head noun in the predicate; in other words, rather than 'This tomato [is] yellow', Rapa Nui has 'This tomato [is] a yellow tomato'. This makes attributive clauses very similar in structure to classifying clauses, but while the predicate marker is obligatory in classifying clauses, in attributive clauses it is usually omitted.

Existential clauses may be verbless (with the Existee as nominal predicate) or verbal (using the verb *ai*, with the Existee as subject). They may be expanded with a possessor to form possessive clauses; these are usually constructed with a verb: 'His house existed' = 'He had a house'. Possession may also be expressed in a topic-comment construction: 'As for him, there was a house.'

In recent years, Rapa Nui has seen the emergence of two copula verbs: *ai* 'to be' and *riro* 'to become'. This development becomes clear by comparing old and new versions of stories in which a person transforms into an animal: in old versions the transformation is expressed in a nominal clause, in new versions *riro* is used. In copula constructions, the nominal predicate is marked with *he*, just as in nonverbal clauses. *Riro* was borrowed from Tahitian, but only in Rapa Nui did it develop into a copula verb.

## **10 Mood and negation**

### **10.1 Introduction**

Mood concerns the pragmatic status of a sentence, the speech act performed by uttering the sentence: a sentence can either be a statement (declarative mood), command (imperative mood) or question (interrogative mood) (Dixon 2010a: 95; Payne 1997: 294). A fourth (minor) speech act is the exclamative, in which the speaker gives an affective response to a fact presumed to be known by the hearer (König & Siemund 2007: 316).

This chapter deals with mood; sections §10.2–10.4 discuss imperative, interrogative and exclamative constructions, respectively. Furthermore, this chapter discusses negation (§10.5).

### **10.2 Imperative mood**

### **10.2.1 The imperative**

Imperatives are expressed by two preverbal markers, which also have an aspectual value: the contiguity marker *ka* (§7.2.6) and the imperfective marker *e* (§7.2.5). *Ka* is used for actions which are to be performed immediately; *ka* with imperative function is glossed imp(erative). *E* is used for actions which are to be performed in the future or which are to be performed repeatedly or habitually, as well as for general instructions; *e* with imperative function is glossed exh(ortative). *Ka* and *e* can be characterised as marking direct and indirect injunctions, respectively. A few examples of both markers:


#### 10 Mood and negation


As these examples show, the subject can be either omitted (1) or expressed (2–5). If expressed, it is a 2nd person pronoun placed after the verb. Unlike other subject pronouns, it is not preceded by the proper article *a* (§5.13.2.1).

In a series of commands, only the first imperative tends to have an expressed subject:

(6) *Ka* imp *ꞌara* wake\_up *mai* hither *koe,* 2sg *ka* imp *kai* eat *tāꞌau* poss.2sg.a *o* of *te* art *kai.* food 'Wake up, eat some food (lit. your [part] of the food).' [R310.104]

As discussed in §8.4.1, the direct object has the accusative marker *i* when the subject is expressed (as in 5); when the subject is not expressed, the accusative marker is omitted.

There are clear functional similarities between the imperative use of the markers *ka* and *e* and their aspectual uses.


While *ka* and *e* can be followed by any postverbal particle (depending on the function of the clause), in imperatives the range of postverbal particles with both aspectuals is limited. As the following example shows, the verb can be followed by evaluatives (*nō* and *rō*) and directionals (*mai* and *atu*); postverbal demonstratives and the continuity marker *ꞌā/ꞌana* are excluded.

(7) *Ka* imp *haka* caus *noho* stay *nō* just *atu* away *koe* 2sg *i* acc *a* prop *au* 1sg *ꞌi* at *nei.* prox 'Let me just stay here.' [R229.013]

<sup>1</sup> *Ka* is used as an imperative marker in various EP languages. In most descriptions, all uses of *ka* are subsumed under a single particle. Imperative or subjunctive *e* is found for example in Māori (Waite 1990: 403; Bauer, Parker & Evans 1993: 30), Tahitian (Lazard & Peltzer 2000: 28) and Hawaiian (Elbert & Pukui 1979: 61). Weber (2003b) describes imperative *ka* and aspectual *ka* as distinctive particles; he also distinguishes exhortative *e* from imperfective *e*.

10.2 Imperative mood

The imperative can be used with any verb. It is rarely used with adjectives, but this may have pragmatic rather than syntactic reasons: there are simply not many situations in which it is appropriate to order someone to have a certain property. For an example of an imperative-marked adjective, see (91) on p. 108.

As the examples above show, the imperative has a wide range of pragmatic usages, including commands, requests, invitations and permissions. It is used between persons of equal or of different status; it is not inappropriate to address a higher-status person with an imperative. In the Bible translation, the imperative is commonly used in prayer; in the following example from the corpus, a chief is addressed in the imperative:

(8) *E* voc *te* art *ꞌariki* king *ē,* voc *e* voc *Tuꞌu* Tu'u *Kōihu* Koihu *ē,* voc *ka* imp *vaꞌai* give *mai* hither *koe* 2sg *i* acc *to* art:of *mātou* 1pl.excl *mōai.* statue

'O king Tu'u Koihu, give us a statue (lit. our statue).' [Mtx-4-01.048]

Very occasionally, the imperative marker is omitted; this happens especially before the causative marker *haka*, possibly for euphonic reasons, to prevent the sequence *ka haka*.

(9) *Haka* caus *rito* ready *koe,* 2sg *e* voc *nua* Mum *ē,* voc *mo* for *kā* kindle *i* acc *to* art:of *tātou* 1pl.incl *ꞌumu* earth\_oven *āpō.* tomorrow

'Get ready, Mum, to light our earth oven tomorrow.' [R352.041]

### **10.2.2 Third-person injunctions (jussives)**

*Ka* or *e* are also used to express instructions or advice to be carried out by a third-person Agent. This happens for example in procedural texts, which describe how something is done or should be done.

As the following examples show, the subject may occur either before or after the verb, as in declarative clauses.


'For underwater fishing, you need (lit. there should be) the following things: a harpoon, glasses, fins…' [R360.001]

#### 10 Mood and negation

### **10.2.3 First-person injunctions (hortatives)**

First-person injunctions (hortatives)<sup>2</sup> are marked with *ki*, the marker also used in certain purpose clauses (§11.5.3). As with imperatives, the subject is optional; if expressed, it is a pronoun which occurs after the verb and which is not preceded by the proper article *a*.


As (14) shows, *ki* is also used to mark proposals in question form.

The hortative may be introduced by *matu* 'come on, let's do it', an interjection which also occurs in isolation. It can also be introduced by the directional *mai*. Note that this is an atypical use of the directional, which normally occurs postverbally (§7.5).


### **10.3 Interrogatives**

### **10.3.1 Polar questions**

Polar questions (also known as yes/no questions) usually do not have a special marker, though the particle *hoki* may be used (see below); nor do they differ from statements in word order. The only difference between polar questions and statements is intonational: whereas in statements the final phrase of the sentence is normally pronounced in a low tone, polar questions have a high rise on the final stressed syllable (§2.4.2; cf. Du Feu 1995: 27). Here are a few examples of polar questions:

<sup>2</sup> On the term 'hortative', see König & Siemund (2007: 305, 313)) and Andrews (2007b: 207).

10.3 Interrogatives


As these examples show, various particles can be added after the first constituent:


When a constituent within the clause is questioned, it is in focus position: it is fronted as in (20).

Polar questions may be marked with *hoki* (glossed pq = polar question), which is placed at the start of the sentence. *Hoki* is less common in modern Rapa Nui than in older texts, but it does occur. It is used especially when the speaker expects a certain answer to the question, whether affirmative as in (21–22) or negative as in (23–24). For example, in (22) the context makes clear that the speaker assumes that the hearer has indeed heard the dream; on the other hand, in (23), the speaker does not believe that the hearer has ever seen a devil.


#### 10 Mood and negation

(24) *¿Hoki* pq *e* ipfv *ketu* raise *rō* emph *koe* 2sg *i* acc *te* art *hare* house *o* of *te* art *taŋata* man *ki* to *raro?* below '(one wind to another:) Could you destroy someone's house (lit. raise down a house of a man)⁈' [R314.121]

When a question contains a negation, it depends on the underlying presupposition which answering strategy ('yes' or 'no') is appropriate. In the following examples, the person asking the question presupposes that the underlying proposition is true; in (25) for example, the speaker expects that the person pointed out is indeed Vivika. The positive reply 'yes' confirms this expectation. In (26), the asker expects the addressee to want to have him as father; negative response 'no' refutes this expectation.


On the other hand, when the speaker presupposes that the underlying proposition is not true, this negative expectation can be confirmed with a positive answer:

(27) *—¿ꞌIna* neg *he* pred *pepe?* chair *—ꞌĒē.* yes *E* ipfv *nohonoho* sit:red *nō* just *ꞌā* cont *ꞌi* at *raro.* below '—There were no chairs? —Indeed. They sat on the floor.' [R413.635]

### **10.3.2 Content questions**

Content questions are formed with one of the following question words: *ai* 'who', *aha* 'what', *hē* 'where, when, which', or *hia* 'how many, how much'. These are always the nucleus of the first constituent of the clause. Each question word belongs to a different word class, as can be seen from the elements preceding them. For example, *ai* is a pronoun, while *aha* is best categorised as a common noun. In the following sections, these question words will be discussed in turn.

10.3 Interrogatives

#### **10.3.2.1** *Ai/ꞌai* **'who'**

The question word 'who' has two forms: *ai* and *ꞌai*. <sup>3</sup> *Ai* occurs after prepositions and after the proper article *a*, while *ꞌai* occurs in possessive and benefactive forms. Syntactically, *ai/ꞌai* is a pronoun: like personal pronouns, it is preceded by the proper article *a* after the prepositions *ꞌi/i* and *ki* (*ki a ai*), it follows immediately after other prepositions (*ko ai*), and is never preceded by a determiner.

*Ai* is always in focus. In nominal clauses, this means that *ai* is preposed and receives the main clause stress. It is marked with *ko*, just like all pronouns used as identifying predicate (§9.2.2). Two examples:


In a verbal clause, when *ai* is a core argument (S, A or O), it is not only preposed and stressed, but the clause takes a focus construction. Just as in declarative clauses, two constructions are possible: the actor-emphatic or a cleft.

When *ai* is Agent, an actor-emphatic construction can be used (§8.6.3). In this construction, the Agent is marked as possessive (if the clause is perfective) or benefactive (if the clause is imperfective); this means that the interrogative is *ꞌa ꞌai* or *mā ꞌai*, respectively. The object is often placed before the verb and tends to be unmarked.


<sup>3</sup> Reflexes of *ai* occurs in most or all Polynesian languages. In Tongic the form is *hai*, which suggests that the PPN form was *\*hai*. In some EP languages (Tahitian, Māori, Hawaiian), the form is *vai/wai*. No other language has a form *ꞌai* except Rarotongan, where the glottal is the regular reflex of PEP *\*h*, *\*f* or *\*s* (§2.5.2).

#### 10 Mood and negation

When *ai* is any core argument (regardless of its semantic role), a cleft construction can be used (§9.2.6).<sup>4</sup> In this construction, *ko ai* is a nominal predicate, followed by a subject containing a relative clause. The subject noun is usually the generic *meꞌe*, though other nouns are also used. A few examples:

S/A questioned:


O questioned:

(35) *¿Ko* prom *ai* who *te* art *meꞌe* thing *ena* med *e* ag *kōrua* 2pl *ka* cntg *haka* caus *tere* run *ena?* med '(If everybody wants to govern the island,) whom will you govern?' [R647.370]

Only very occasionally is *ko ai* immediately followed by a verb; this happens especially in older texts. It is impossible to tell whether this is a simple clause, or a cleft with headless relative clause (a construction not attested otherwise, §8.6.2.1).

(36) *¿Ko* prom *ai* who *i* pfv *mate?* die 'Who died?' [MsE-046.009]

When a possessor is questioned, the form *ꞌai* is used, preceded by *o* or *ꞌa*: like all singular pronouns, *ꞌai* is subject to the *o*/*a* distinction (§6.3.2). The clause is a proprietary clause (§9.4.2) with fronted predicate. Two examples:


<sup>4</sup> Potsdam & Polinsky (2011) distinguish three questioning strategies in Polynesian: displacement (= preposing the Wh-constituent), clefts, and pseudo-clefts (=clefts in which the relative clause has a head noun; in Rapa Nui, this is the only cleft strategy possible, see sec. §9.2.6 and §8.6.2.1). They tentatively analyse Rapa Nui as using the displacement strategy, but admit that data are scarce. One example is given of a construction as in (36), as well as a number of oblique examples (which indeed have a displacement structure), and one example of *he aha* in the sense of 'why' (which is also an oblique with displacement). However, in Rapa Nui texts, pseudo-clefts abound in questions, both with *ai* 'who' (such as in (33–35)) and with *aha* 'what' (such as (44) in the next section).

10.3 Interrogatives

When *ai* questions an obliqe constituent, this constituent is simply fronted.<sup>5</sup>


*Ai* asks about persons, while *aha* 'what' asks about things. To ask about names, *ai* is used. This applies even when the name asked for is the name of an inanimate entity:

(41) *¿Ko* prom *ai* who *te* art *ꞌīŋoa* name *o* of *rā* dist *kona?* place 'What is the name of that place?' [R124.014]

#### **10.3.2.2** *Aha* **'what, why'**

Unlike *ai* 'who', *aha* 'what'<sup>6</sup> is a common noun: it is preceded by the predicate marker *he* or the article *te*, never by the proper article *a*. Apart from this, it is also used as noun modifier. Like *ai*, *aha* is in focus: it always occurs initially and is stressed.

As a nominal predicate, *aha* is preceded by *he*; the construction is a simple classifying clause (§9.2.1).


When questioning arguments in a verbal clause, a cleft construction is used: *aha* is marked as nominal predicate, while the subject noun phrase consists of an anchor noun (usually *meꞌe*) followed by a relative clause.

Sometimes the S/A argument is questioned; as *aha* questions non-human entites, this is not very common:

<sup>5</sup> As (39–40) show, if the clause also has a subject, it is usually preverbal. This is usual after most preverbal constituents (§8.6.1.1).

<sup>6</sup> Cognates of *aha* (< PPN*\*hafa*, going back to PAN) occur throughout Polynesian languages, but especially in Eastern Polynesian (outside EP e.g. in Kapingamarangi and Nukuoro). Most Tongic and Samoic languages have a reflex of PPN *\*haa* instead (Pollex, see Greenhill & Clark 2011).

10 Mood and negation

(44) *¿He* pred *aha* who *te* art *meꞌe* thing *i* pfv *topa* happen *ki* to *a* prop *koro?* Dad 'What happened to Dad?' [R615.594]

More commonly, the O argument is questioned. As in all object relative clauses (§11.4.2), the subject is either marked with *e* as in (45), or the possessive-relative construction is used (§11.4.4) as in (46).


When *aha* has an obliqe role, it is simply preposed as constituent of the verbal clause. As with *ai* 'who', the subject is usually preverbal. After prepositions, *aha* is preceded by the article *te* (like all common nouns), with the exception of the instrumental preposition *hai* (which is never followed by a determiner) and benefactive *mo*. *Mo aha* is used to ask about the purpose of an event.


*He aha* is also used in the sense 'why'. In this case, it is an oblique, which is part of a simple verbal clause with preverbal subject: *he aha S V*. The structure of the sentence is thus different from *he aha* as subject or object, which have a cleft structure *he aha* [NP Rel]; compare (51) with (45–46) above:

(51) *¿He* pred *aha* what *koe* 2sg *e* ipfv *taŋi* cry *ena?* med 'Why are you crying?' [Ley-9-55.064]

10.3 Interrogatives

*Aha* is used as an adjective 'what, which', especially after time nouns. The noun phrase containing *aha* is clause-initial:


As these examples show, the noun is not preceded by a determiner. Cf. the use of *hē* as an adjective (§10.3.2.3).

#### **10.3.2.3** *Hē* **'where, when, how, which'**

*Hē* is used to ask about places, times and situations.<sup>7</sup> Because of its wide range of functions, it is glossed 'cq' (content question). Syntactically, it is a locational (§3.6): it is immediately preceded by prepositions, without any determiner. Like all question words, it is in focus and always occurs as the first constituent of the clause.

**Location** When preceded by a locative preposition (*ꞌi* 'at' *ki* 'to', *mai* 'from'*, a* 'by, towards') or *o* 'of', *hē* has a locative sense 'where'. As the examples show, *hē* may be the predicate of a verbless clause as in (54–55), or an oblique in a verbal clause as in (56–58). In verbal clauses, the subject is usually preverbal.


<sup>7</sup> According to Pollex (Greenhill & Clark 2011), *hē* is a reflex of PPN *\*fē* 'where', which occurs in a number of Samoic and Tongic languages. However, it is more plausible that *hē* reflects PNP *\*fea* 'where', which is widespread both in Samoic and EP languages (e.g. Tahitian, Hawaiian, Marquesan *hea*, Māori *whea*, Rarotongan, Mangarevan *ꞌea*). Cf. §2.5.2 on monophthongisation of particles.

10 Mood and negation

(58) *¿Ki* to *hē* cq *kōrua* 2pl *ko* prom *te* art *poki* child *i* pfv *iri* ascend *mai* hither *ena?* med 'Where did you and the child go up to?' [R229.205]

In nominal clauses, *hē* is also used without a preceding preposition. Its sense is similar to *ꞌi hē* 'where', but it is only used to ask about things that are situationally close; often, the addressee is directly involved. Compare (59–60) with (54) and (56) above:


**Situation** *Pē hē* 'like what, how' asks about a situation; it is the interrogative counterpart of *pē ira* 'like that' (§4.6.5.2). It occurs for example in the common greeting *Pē hē koe* 'how are you'. As with other prepositions, in a verbal clause the subject is usually preverbal.


**Time** To ask about time, *hē* is preceded by *ꞌaŋa-* (past) or *a* (future). *ꞌAŋahē* is written as one word; *a hē* is written as two words and is homophonic to *a hē* 'by what place' (see (57) above). The particles *ꞌaŋa-* and *a* also occur with other roots (§3.6.4). As (64) shows, *ꞌaŋahe* is preceded by locative prepositions.


10.3 Interrogatives

**Adjectival use** Finally, *hē* is used as an adjective 'which'. As the examples below show, the questioned noun is preceded by the appropriate preposition marking its semantic role, but does not have a determiner. For example, the questioned NP in (65) is *o huaꞌai hē*, not \**o te huaꞌai hē*, even though the preposition *o* must normally be followed by a determiner (§5.3.2.1).

(65) *¿O* of *huaꞌai* family *hē* cq *te* art *rūꞌau* old\_woman *era* dist *ꞌai?* deic

'From which family is that woman over there?' [R413.305]

(66) *—¿Ko* prom *poki* child *hē* cq *rā* dist *poki* child *hiko* snatch *era* dist *i* acc *taꞌa* poss.2sg.a *meꞌe?* thing *—Poki* child *tane.* male '—Which child [was the child who] snatched your things? —A boy.' [R172.012–014]

There is no sharp difference in meaning between *hē* and *aha* used as adjective (see (52–53) above), except that the latter only occurs with time nouns, while *hē* occurs with any type of noun. Possibly *hē* implies a choice from a closed range, though (66) above appears to be a counterexample.

#### **10.3.2.4** *Hia* **'how much, how many'**

*Hia* 'how much, how many' (< PPN *\*fiha,* with reflexes throughout Polynesia) is a numeral: it is always preceded by one of the numeral particles *e*, *ka* and *hoko* (§4.3.2). *Hia* may occur in a noun phrase as in (67–68), or as a separate constituent as in (69–70). In either case, it is placed at the start of the sentence.


10 Mood and negation

### **10.3.3 Dependent questions**

Dependent questions, i.e. questions in subordinate clauses, occur mainly after speech verbs and cognitive verbs.

**Polar questions** Dependent polar questions are optionally introduced by *hoki* as in (71). In (72), *hoki* is not used, but here the question has a tag *ꞌo ꞌina*.


Alternatively, the question is marked with the irrealis marker *ana* (§11.5.2.2):

(73) *ꞌO* because\_of *ira* ana *a* prop *au* 1sg *i* pfv *ꞌui* ask *atu* away *ena* med [*ana* irr *haŋa* want *koe* 2sg *mo* for *turu* go\_down *mai* hither *ki* to *nei…* ] prox

'Therefore I asked you if you wanted to come here…' [R315.269]

**Content questions** Dependent content questions are marked with one of the question words discussed in the previous sections. Just as in main clause questions, the questioned constituent is placed at the start of the clause.


10.4 Exclamatives

### **10.4 Exclamatives**

There are three constructions in Rapa Nui specifically used for exclamations. They are marked with the aspectual *ka*, the preposition *ko* and the deictic particle *ꞌai*, respectively. These constructions will be discussed in turn in the next subsections.

### **10.4.1** *Ka* **in exclamations**

With certain adjectives the continguity marker *ka* (§7.2.6) is used in an emphatic sense, often in exclamations. In this construction, the quality expressed by the adjective is emphasised. This construction is only used with a limited number of adjectives, all of which express a positive evaluation: *riva* 'good', *reka* 'pleasant', *tau* 'beautiful, handsome', in older texts also *maꞌitaki* 'clean; beautiful'. A few examples:


This construction is similar in function to *ꞌai te* preceding an adjective (§10.4.3 below); in fact, in (79) above the two constructions are used side by side. The choice between the two is lexically determined: while *ka* is only used with adjectives denoting a positive evaluation, *ꞌai te* is used with adjectives of size.

The origin of this use of *ka* may lay in the tendency of *ka* to denote an extent, a use which is for example seen in the construction *ka V rō* 'until' (§11.6.2.5) and in the use of *ka* with numerals (§4.3.2.2).

### **10.4.2** *Ko* **in exclamations**

In modern Rapa Nui, *ko te* X is used in exclamations to convey a strong emotion about something.<sup>8</sup> This usage does not occur in older texts. Sometimes it involves a noun as in (80), but more commonly, exclamative *ko te* is followed by an adjective as in (81). The speaker expresses his or her emotion about the quality expressed, implying that the quality is true to a high degree: 'How beautiful!'.

<sup>8</sup> Moyse-Faurie (2011: 149) points out, that predicate (i.e. *ko*-marked) noun phrases in Polynesian languages often have an exclamative function.

10 Mood and negation


The person or thing possessing the quality in question is marked with the preposition *i* 'corresponding to' (§4.7.2):


A similar construction is *ko te aha* 'what', followed by a noun phrase:


### **10.4.3** *ꞌAi* **in exclamations**

Adjectives of size, such as *nuinui* 'big' and *kumi* 'big, long', occur in a nominal construction in which they are preceded by the deictic particle *ꞌai* (§4.5.4.1.2).


10.5 Negation

### **10.5 Negation**

Rapa Nui has three clausal negators:

*ꞌina* neutral (discussed in §10.5.1–10.5.2) *kai* perfective (§10.5.3) *(e) ko* imperfective (§10.5.4–10.5.5)

The neutral character of *ꞌina* is shown by the fact that it occurs in a variety of contexts, is always followed by the neutral aspectual *he*, and can be combined in a single clause with one of the other negators.

While *ꞌina* is a phrase head, *(e) ko* and *kai* are preverbal particles which occur in the same position as – and thus replace – the aspectual marker (§7.1). This means that there are fewer aspectual distinctions in negative clauses than in positive ones (cf. Dixon 2012: 129).

Apart from the three clausal negators, Rapa Nui has a constituent negator *taꞌe* (§10.5.6) and an existential/noun negator *kore* (§10.5.7).

The verb phrase particle *hia/ia* 'not yet', which occurs in combination with different negators, is discussed in §10.5.8.

### **10.5.1 The neutral negator** *ꞌina*

*ꞌIna* is the most neutral negator; of all the negators, it has the widest range of use.

#### **10.5.1.1 Verbal clauses**

*ꞌina* is a common negator in verbal clauses, as the following examples show:


#### 10 Mood and negation

(92) *He* ntr *haꞌamata* begin *he* ntr *riri,* angry *ꞌe* and *ꞌina* neg *he* ntr *hakaroŋo* listen *ki* to *tū* dem *vānaŋa* word *era* dist *o* of *tū* dem *hoa* friend *era* dist *ōꞌona.* poss.3sg.o

'He began to get angry, and did not listen to the words of his friend.' [R237.152]

These examples illustrate a number of characteristics of *ꞌina*:


For the sake of comparison: the unmarked positive counterpart of (88) would be:

(93) *He* ntr *uꞌi* watch *rō* emph *mai* hither *a* prop *Heru* Heru *hai* ins *mata.* eye 'Heru watched (her) with his eyes.' [R313.165]

Only occasionally is the subject in postverbal position. Usually a postverbal subject is marked with the agentive marker *e*. In general, preverbal subjects are not *e*-marked (§8.3.1), which could be the reason why the *e*-marked subject is placed after the verb.

(94) *ꞌIna* neg *he* ntr *aŋiaŋi* certain:red *e* ag *tū* dem *ŋā* pl *ꞌakuꞌaku* spirit *era* dist *e* ipfv *aha* what *ꞌā* cont *te* art *ꞌariki.* king 'Those spirits did not know what the king was doing.' [R532-06.018]

In (88–90) above, the subject is a proper noun or pronoun. When the subject is a common noun and preverbal, it is usually not preceded by the article *te*, but by the predicate marker *he*. This happens despite the fact that it refers to a definite entity, while *he* normally marks nonreferential noun phrases (§5.3.4.1).


10.5 Negation

(97) *¿ꞌIna* neg *ꞌō* really *he* pred *mata* tribe *o* of *Hotu* Hotu *ꞌIti* Iti *he* pred *taŋitaŋi* cry:red *ki* to *te* art *Tūpāhotu?* Tupahotu 'The tribe of Hotu Iti doesn't mourn for the Tupahotu, does it?' [R304.070]

*ꞌIna* may be followed by the article or another *t*-deteminer, but this happens only occasionally:

(98) *Te* art *ꞌati* problem *nō* just *ꞌina* neg *te* art *ŋā* pl *poki* child *he* ntr *haŋa* want *mo* for *ꞌite.* know

'The problem is that the children don't want to know.' [R647.094]

In (95–97) above, the construction *ꞌina he N VP* is a verbal clause in which *he N* is the preverbal subject. However, the same sequence of elements may also be an existential clause, in which the verb phrase is part of a relative clause (see (108–110) below on the negation of existential clauses).


Constructions like (99–100) are quite distinct from (95–97) above. Firstly, the noun phrase does not refer to a specific entity, but predicates the existence of the category as a whole: 'there is not…' In the second place, the verb is marked in ways typical of relative clauses. While the verb in (95–97) has the neutral marker *he*, verbs in relative clauses are typically marked with the aspectuals *i* or *e* or with zero marking, but not by *he* (§11.4.3; §11.4.5).<sup>9</sup>

A third difference between verbal *ꞌina* clauses and existential constructions is, that in the latter the noun phrase after *ꞌina* is not always the S/A argument of the verb. This is illustrated in (101–102), where the noun phrase following *ꞌina* is the Patient. As (102) shows, the Agent may be expressed as a possessive, a construction common in relative clauses (§11.4.4).

(101) *ꞌIna* neg *he* pred *meꞌe* thing *i* pfv *rovā* obtain *o* of *tū* art *pō* night *era.* dist

> 'They did not catch anything (lit. there was no thing obtained) that night.' [R359.005]

<sup>9</sup> That these two constructions are distinct is confirmed by the fact that *i*, *e* and zero marking never occur after *ꞌina* + proper noun or pronoun; they are limited to constructions with a common noun, which are open to an existential analysis.

#### 10 Mood and negation

(102) *¿ꞌIna* neg *ꞌō* really *he* pred *ꞌaꞌamu* story *ꞌāꞌau* poss.2sg.a *i* pfv *maꞌu* carry *mai* hither *mai* from *Haŋa* Hanga *Roa?* Roa 'Haven't you brought any news (lit. are there no stories you brought) from Hanga Roa?' [R380.039]

Constructions as in (99–102) are relatively unusual. More commonly, the noun phrase in negative existential constructions is preceded by the numeral *e tahi* 'one'. *ꞌIna e tahi N* has become the usual way to express 'not one, no one, nobody':


All examples so far involve *ꞌina* as sole negator in the clause. However, more often than not, *ꞌina* as verbal clause negator co-occurs with a second clausal negator, either perfective *kai* or imperfective *(e) ko*. Table 10.1 gives the number of occurrences of *ꞌina* in verbal clauses in the text corpus with and without a second negator.


Table 10.1: Frequencies of single and double negators

A few examples of double negation:


10.5 Negation

(107) *ꞌIna* neg *e* ipfv *ko* neg.ipfv *kai* eat *i* acc *te* art *kahi* tuna *o* of *tōꞌona* poss.3sg.o *vaka.* boat '(The fisherman) would not eat the tuna (caught with) his boat.' [Ley-5-27.013]

*ꞌIna* … *e ko* and *ꞌina* … *kai* are multiple markings of a single negation. The effect of multiple marking may be a slight reinforcement or emphasis; notice however that multiple marking is so common, that it cannot be a highly marked form.<sup>10</sup> As the examples illustrate, the subject is usually preverbal, just like constructions where *ꞌina* is the only negator in the clause.

In one situation the use of the double negation is almost exceptionless: the imperative. This is discussed in §10.5.5.

#### **10.5.1.2 Nonverbal clauses**

Several types of nonverbal clauses are negated by *ꞌina*.

Existential clauses (§9.3) are negated by placing *ꞌina* in front of the nominal predicate as in (108). The same is true for subtypes of existential clauses: existential-locative clauses as in (109), possessive clauses as in (110).


Notice that positive existential clauses are nowadays usually constructed with the existential verb *ai* (§9.3.1); negative clauses, however, are constructed without a verb, as these examples show.

Locative clauses (§9.4.1) can be negated with *ꞌina* in front of the subject. As in verbal clauses, the subject has the predicate marker *he*, even when it has definite reference (see (95–97) above).

(111) *ꞌIna* neg *he* pred *māmā* mother *o* of *nā* med *poki* child *o* of *nei.* prox

'The mother of that boy is not here.' [R403.051]

Alternatively, the locative phrase is negated by the constituent negator *taꞌe* (see (144) on p. 505).

<sup>10</sup> See Dixon (2012: 91) on multiple marking. According to Payne (1985: 224), there is a strong crosslinguistic tendency for negatives to be reinforced by other elements in the clause.

10 Mood and negation

#### **10.5.1.3 Independent polarity item**

Besides negating verbal and nominal clauses, *ꞌina* also functions as independent polarity item 'no':


### **10.5.2 Status and origin of** *ꞌina*

In many Polynesian languages, some negators are verbs, or at least have important characteristics in common with verbs: they occur in the position of the predicate and they are preceded and/or followed by VP elements such as aspectuals (see Payne 1985: 209–211; Broschart 1999 on Tongan). The rest of the sentence may be constructed as a subordinate clause, as evidenced by the constituent order (subject raising) and by the fact that the choice of aspectuals on the main verb is limited in the same way as in other subordinate clauses. The latter happens for example in Tahitian (Lazard & Peltzer 1999; Lazard & Peltzer 2000: 49) and Māori (Hohepa 1969a; Bauer, Parker & Evans 1993: 139–141).

The question is whether Rapa Nui *ꞌina* can be analysed as a matrix verb followed by a subordinate clause.<sup>11</sup> Weber (2003a: 57) assumes a biclausal structure, when she analyses subject placement in *ꞌina* constructions by a raising rule, in which the subject is moved to the subject position of the higher clause. Stenson (1981: 159–160) gives several arguments to treat *ꞌina* as a matrix verb: it may be separated from the negated verb by the subject (while the otherwise common VSO order is marginal in *ꞌina*-clauses); it may co-occur with the negators *kai* and *e ko*, and unlike the latter, it co-occurs with an aspect marker. It should be noted, however, that the last two points only show that *ꞌina* has a different status from *kai* and *e ko*, without demonstrating its verbal character. After all, the aspect marker does not occur in front of *ꞌina* itself, but in front of the following verb.

Another possible indication for the verbal character of *ꞌina* is, that it can be followed by a wide range of verb phrase particles: certain adverbs (*mau* 'really', *takoꞌa* 'also'), the emphatic marker *rō*, the directional *atu*, postverbal demonstratives and the identity marker *ꞌā*. This is illustrated in (90) above and in the following example:

<sup>11</sup> Note that *ꞌina* is not related to verb-like negators in other Polynesian languages (but see the discussion on Mangarevan *inau* below). The latter either do not have a cognate in Rapa Nui or a cognate with a different status; for example, the negative verb *ꞌikai* in Tongan is related to the negative particle *kai* in Rapa Nui.

10.5 Negation

(114) *ꞌIna* neg *rō* emph *atu* away *he* ntr *noho* stay *i* at *a* prop *au.* 1sg 'I couldn't keep (my fishing line) steady (lit. It didn't stay at all to me).' [R230.162]

Despite these arguments, there are good reasons not to analyse *ꞌina* as a verb followed by a subordinate clause.


The main argument for analysing *ꞌina* as a matrix verb in a biclausal construction, is that it attracts the subject: after *ꞌina*, the subject is usually preverbal. In this respect, *ꞌina* constructions are similar to constructions with auxiliary verbs such as *haꞌamata* 'begin' (§11.3.2.1), and it may be tempting to analyse both along the same lines. However, auxiliary verbs in Rapa Nui are not the only elements that trigger preverbal subject placement. Subjects tend to be preverbal after a wide range of initial elements, including adjuncts and deictic particles (§8.6.1.1; cf. Footnote 42 on p. 404).

We may conclude that *ꞌina* is not a verb and that *ꞌina* constructions are monoclausal. Even so, it should be noted that *ꞌina* is significantly different from other negators: *ꞌina* is a phrase nucleus, while other negators are prenuclear particles. *ꞌIna* forms a constituent on its own, which may contain various postnuclear particles. This is confirmed by the fact that second-position particles (which are placed after the first constituent) occur immediately after *ꞌina*. Here is an example with *paꞌi* (§4.5.4.2):

(115) *ꞌIna,* neg *paꞌi,* in\_fact *a* prop *mātou* 1pl.excl *kai* neg.pfv *māuiui* sick *ꞌi* at *te* art *rōviro.* smallpox 'In fact, we were not sick with smallpox.' [R539-1.680]

#### 10 Mood and negation

The fact that *ꞌina* is consistently initial, conforms to a general crosslinguistic tendency for negative particles to come first (Miestamo 2007: 560 and refs. there). It may also be explained by the possible origin of *ꞌina*. Unlike other negators in Rapa Nui, *ꞌina* is not widely found in other Polynesian languages. The only plausible cognate I have found is Mangareven *inau*. <sup>12</sup> The latter is used both as independent negator 'no' and as verb 'to deny a proposition; to refuse' (Tregear 2009: 24; Rensch 1991: 83).

If *ꞌina* is indeed related to Mangarevan *inau*, this suggests that it originated as an independent polarity item.<sup>13</sup> This would confirm Clark's suggestion (Clark 1976: 104) that *ꞌina* started out as reinforcement of another negator ('no, we will not go') and developed into a clausal negator, a cross-linguistically common process which is known as Jespersen's Cycle (Miestamo 2007: 566). This analysis would provide a historical explanation for the fact that *ꞌina* is always clause-initial, and the fact that it is often accompanied by another negator.

### **10.5.3 The perfective negator** *kai*

*Kai* negates clauses in the perfective aspect.<sup>14</sup> It precedes the verb and occurs in the same position as aspectuals. As discussed in §10.5.1 above, it is often combined with *ꞌina*, in which case the subject usually precedes the verb.

*Kai* is used to negate events in narrative as in (116–117), and any past events as in (118– 120).<sup>15</sup> If these clauses were positive, the former would be marked with *he*, the latter with perfective *i* or – if the speaker wishes to emphasise their present relevance – perfect *ko V ꞌā*.


<sup>12</sup> *Inau* may in turn be related to *kinau*, found in some languages in West-Polynesia in the sense 'to persist against something' (Pollex, Greenhill & Clark 2011). In East Futunan and East Uvean, this verb has 'to deny' as one of its senses.

<sup>13</sup> The verbal use in Mangarevan may be a secondary development, one which is not unexpected given the great freedom of cross-categorial use in Polynesian languages.

<sup>14</sup> The negator *kai* occurs in a few other Polynesian languages (Māori, Pukapuka, Tikopian) but only as a negative imperative marker and/or in the sense 'lest' (Pollex, see Greenhill & Clark 2011). More widespread are reflexes of PPN *\* ꞌikai*, which has various negative senses in all branches of Polynesian.

<sup>15</sup> The latter point is illustrated somewhat more extensively, to show that *kai* does indeed negate past tense clauses, the positive counterpart of which would have perfective *i*. In this respect my analysis is different from Englert (1978: 79), who claims that *i*-clauses are negated by *ta ꞌe* (an analysis followed by Chapin 1978: 158 and Stenson 1981: 157). In fact, *ta ꞌe* is not the default negator of *i*, but is used to negate certain constructions with *i* and *e* (§10.5.6 below).

10.5 Negation


'Mum, you didn't tell me how to cook that food.' [R236.091]

*Kai* is also used to negate stative verbs. In positive clauses, these verbs are commonly marked with perfect aspect *ko V ꞌā* (§7.2.7.2).


In (118), (121) and (123), *kai* co-occurs with the neutral negator *ꞌina*. There is little or no semantic or pragmatic difference between clauses with and without *ꞌina*, though he examples with *ꞌina* may be slightly more emphatic than constructions with *kai* alone.

Just like any verb phrase, a verb phrase marked with *kai* may contain various kinds of postverbal particles, such as directionals (*mai* in (118) and (120) above). When the clause has perfect aspect, the continuity marker *ꞌā/ꞌana* may be added. This marker is obligatory with the perfect marker *ko* and indicates continuity of a state (§7.2.5.5); in combination with *kai* it indicates that the negative state still continues, i.e. that a positive action has not yet taken place, or that a positive state has not yet been reached.

(124) *E* ipfv *ꞌitiꞌiti* small:red *nō* just *ꞌā* cont *a* prop *koe;* 2sg *kai* neg.pfv *ꞌite* know *ꞌana* cont *e* num *tahi* one *meꞌe* thing *o* of *te* art *via* life *taŋata.* person

'You are (still) little; you don't know anything about human life (yet).' [R210.052]

10 Mood and negation

(125) *¿Kai* neg.pfv *ꞌara* wake\_up *ꞌana* cont *ꞌō* really *a* prop *nua* Mum *era* dist *ko* prom *Kava,* Kava *e* ipfv *taꞌe* conneg *tuꞌu* arrive *mai* hither *nei?* prox

'Hasn't mother Kava not woken up (yet), that she doesn't come?' [R229.359]

### **10.5.4 The imperfective negator** *(e) ko*

*(E) ko*<sup>16</sup> is the imperfective negator. Like *kai*, it replaces the aspectual in front of the verb. The first element *e* (tentatively glossed as imperfective) is almost always included, except in the imperative. Like *kai*, *e ko* it can be reinforced with *ꞌina*, which triggers preverbal verb placement; compare (127) and (128) below.

*E ko* has the same range of use as imperfective *e*. Is is used in sentences expressing a future event or intention:


It also negates habitual actions and general facts.

(129) *ꞌIna* neg *a* prop *ꞌOrohe* Orohe *e* ipfv *ko* neg.ipfv *hoa* abandon *i* acc *tōꞌona* poss.3sg.o *taina* sibling *ꞌitiꞌiti.* small:red '(When they walk to school,) Orohe does not leave his little sister alone.' [R166.005]

<sup>16</sup> The origin of *ko* is unclear. Pollex (Greenhill & Clark 2011) mentions a negative imperative form \**kaua* in Fijian and Polynesian, which could have assimilated > *\*kō* > *ko*. The semantic correspondence is tempting, but the evidence for *\*kaua* is not very strong; more common is *ꞌaua*, which occurs throughout Polynesian and which could be at the root of Rapa Nui *ꞌo* 'lest' (§11.5.4). Alternatively, *ko* could be a shortening of *kore*, which is the main negator in verbal clauses in Central-Eastern Polynesian languages (Clark 1976: 100). This would explain the fact that *e* is a fixed part of the negation in most contexts: in CE languages, *kore* fused with preceding aspectuals (esp. *ka* and *e*). NB *kore* itself also occurs in Rapa Nui as a lexical negator (§10.5.7).

10.5 Negation

(130) *Mo* if *taꞌe* conneg *eꞌa* go\_out *o* of *te* art *nuꞌu* people *hī* to\_fish *ika,* fish *e* ipfv *ko* neg.ipfv *ai* exist *te* art *ika* fish *mo* for *kai.* eat

'If the fishermen don't go out, there is no fish to eat.' [R334.261]

Finally, *e ko* negates stative verbs. This includes auxiliaries like *puē*, as in (133).

(131) *E* ipfv *ko* neg.ipfv *rivariva* good:red *te* art *kāiŋa,* homeland *e* ipfv *ko* neg.ipfv *nahonaho* comfortable *te* art *noho* stay *oŋa.* nmlz

'The land wasn't good, life was not comfortable (up until now).' [R368.103]


### **10.5.5 Negation of the imperative**

Negative commands are marked by the imperfective negator *(e) ko*, usually in combination with *ꞌina*. While *e* is obligatory in other uses of the imperfective negator, in imperatives it is often left out, as in (134) and (136) below. However, when *ꞌina* is not included, as in (137), *e* is obligatory.

As with other uses of *ꞌina*, the subject – if expressed at all – tends to be placed before the verb.

The following examples show, that *(ꞌina) (e) ko* negates both immediate commands (marked with *ka* when positive, §10.2.1) and non-immediate commands (marked with *e* when positive).


'You two take care of Puakiva. Don't beat him.' [R229.420]

10 Mood and negation


First and third person injunctions are negated in the same way. Notice that in (139) below, the subject remains in postverbal position.


### **10.5.6 The constituent negator** *taꞌe*

anymore.' [R310.160]

*Taꞌe* has a wide range of uses, all of which can be characterised as constituent negation: *taꞌe* is used whenever something other than a main clause is negated, i.e. a subordinate clause or a constituent of a clause.<sup>17</sup> Besides, *taꞌe* is used to negate the predicate of certain types of nonverbal clauses.

**10.5.6.1** *Taꞌe* negates noun phrases which are the predicate of a nonverbal clause. This may be a classifying clause with a *he*-marked predicate (§9.2.1) as in (140),<sup>18</sup> or an

<sup>18</sup> There is a difference between:

(i) *Ta ꞌe he taŋata* 'It is not a man' (classifying)

(ii) *ꞌIna he taŋata* 'There is no man' (existential, (103) in §10.5.1)

<sup>17</sup> Cognates of *ta ꞌe* are widespread; they occur in most Samoic-Outlier languages, as well as in Tongan and a number of CE languages (Māori, Marquesan, Mangarevan). The glottal only occurs in those languages that preserved the PPN glottal, such as Tongan and Rapa Nui. The initial vowel was assimilated to *e* in all languages except Tongan and Rapa Nui, and in most Samoic-Outlier the initial consonant changed to *s-* (or a reflex of *\*s-*) or *l-*. As a result, the current form is *see*, *hee* or *lee* in most SO languages, and *tee* in CE languages. Clark (1976: 85–87) argues for *\*ta ꞌe* as the PPN form. This had probably assimilated to *\*te ꞌe* in PNP (see also Hamp 1977); the question remains whether Rapa Nui *ta ꞌe* should be explained as subsequent dissimilation, or whether *\*ta ꞌe* survived alongside *\*te ꞌe* in PNP (Clark 1976: 87).

In SO languages, reflexes of *\*ta ꞌe* are the unmarked negator. In Mangarevan as well, *tē* seems to be a main clause negator (Janeau 1908: 78; examples are found in Pupu-takao (1908), e.g. Mark 4:40 *Tē kereto ana noti ra kotou?* 'Do you still not believe?'). In Marquesan, on the other hand, *tē* is a preverbal modifier (Mutu & Teìkitutoua 2002: 52.

10.5 Negation

identifying clause with a *ko*-marked predicate (§9.2.2) as in (141).


*Taꞌe* does not negate nouns as such: nouns are negated with *kore* (§10.5.7).

**10.5.6.2** *Taꞌe* negates other phrases: prepositional phrases serving as arguments in a verbal clause as in (142–143), prepositional predicates as in (144), possessive predicates as in (145):


**10.5.6.3** *Taꞌe* negates nominalised verbs:

(146) *Kai* neg.pfv *puē* can *takoꞌa* also *a* prop *ia* 3sg *mo* for *hāpī* learn *ꞌo* because\_of *te* art *taꞌe* conneg *rava* sufficient *o* of *te* art *moni.* money

'He could not study as well (like his brother), because there was not enough money (lit. because of the not sufficient of the money).' [R231.006]

(147) *¿Ko* prf *takeꞌa* see *ꞌā* cont *e* ag *koe* 2sg *tuꞌu* poss.2sg.o *taꞌe* conneg *hakaroŋo* listen *ena?* med 'Do you see how disobedient you were (lit. your not listening)?' [R481.117]

#### 10 Mood and negation

**10.5.6.4** *Taꞌe* negatessubconstituents, such as adjectives (148) and quantifiers (149) in the noun phrase.


**10.5.6.5** *Taꞌe* also occurs in the verb phrase. It negates subordinate clauses introduced by a subordinating marker. These markers are in the same position as aspectuals (§11.5); *taꞌe* occurs between the marker and the verb. Below are examples with *mo* 'to, in order to' and *ana* 'irrealis':


Subordinate clauses without subordinating marker are also negated by *taꞌe*. In these cases, *taꞌe* co-occurs with an aspect marker, usually *i* or *e*. As in the examples above, *ta ꞌe* occurs between the marker and the verb. Below are examples of relative clauses (152– 153) (the second without aspectual), a temporal clause (154), and the conjunction *ꞌāhani* (155).

(152) *Te* art *vānaŋa* word *rapa* Rapa *nui* Nui *taꞌe* conneg *he* pred *meꞌe* thing [*e* ipfv *taꞌe* conneg *haŋa* like *rō* emph *ꞌā* cont *e* ag *au* ]*.*

1sg

'The Rapa Nui language is not something I don't like.' [R648.251]


10.5 Negation

(155) *ꞌĀhani* if\_only *ꞌō* really *tōꞌoku* poss.1sg.o *nua* Mum *era* dist *i* pfv *taꞌe* conneg *mate,* die *ꞌī* imm *au* 1sg *ꞌi* at *muri* near *i* in *a* prop *ia* 3sg *ꞌi* in *te* art *hora* time *nei.* prox 'If my mother had not died, I would be near her at this time.' [R245.007]

**10.5.6.6** Interestingly, *taꞌe* also occurs in the verb phrase in main clauses, mainly with the aspect markers *i* and *e*. This happens when the verb phrase is preceded by an oblique constituent. As suggested in Footnote 42 on p. 404, this preposed constituent acts somewhat like a subordinating predicate.

(156) [*Hai* ins *ꞌarero* ]*,* tongue *paꞌi,* in\_fact *e* ipfv *taꞌe* conneg *ŋaro* lost *ena* med *te* art *haka* caus *tere* run *iŋa* nmlz *o* of *te* art *motu* island *nei.* prox

'By means of the language, the culture of this island will not be lost.' [R647.155]


However, in such cases, main clause negators are also used. This is illustrated in (118) above and in the following example:

(160) *ꞌO* because\_of *ira,* ana *ꞌina* neg *e* ipfv *ko* neg.ipfv *ŋaro* lost *te* art *kaikai.* string\_figure 'Therefore, the (art of making) string figures will not be lost.' [R648.133]

**10.5.6.7.** Finally, *taꞌe* is used in combination with the other negators to express double negation; *taꞌe* and the other negator cancel each other out, resulting in a strong affirmation. The other negator may be *kai* or *e ko*; as (162) shows, it may be reinforced by *ꞌina*.

(161) *Kai* neg.pfv *taꞌe* conneg *haka* caus *ꞌite* know *ko* prom *ai* who *a* prop *ia* 3sg *hai* ins *meꞌe* thing *rivariva* good:red *aŋa.* do '(God) did not fail to make known who he is, by the good things he did.' (Acts 14:17]

10 Mood and negation

(162) *…ꞌina* neg *e* ipfv *ko* neg.ipfv *taꞌe* conneg *ravaꞌa* obtain *te* art *ika.* fish '(If the mother does not eat the fish caught by her firstborn son,) he will not fail to catch fish.' [Ley-5-27.008]

### **10.5.7 The negator** *kore*

*Kore*<sup>19</sup> is a verb, meaning 'to lack, be absent, be gone':


Besides, *kore* is used to negate nouns, indicating that the entity expressed by the noun does not exist in the given context; as a noun negator it immediately follows the noun in adjective position. When the noun is a modifier as in (165), *kore* can be translated as 'without N'; in other cases as in (166–167), it can be translated as 'lack of N':


<sup>19</sup> *Kore* is common in EP languages; in all languages but Rapa Nui, it is either an existential negation ('there is not'), or negates certain types of verbal clauses. In Rapa Nui, existential clauses are negated with *ꞌina* (§10.5.1). *Kore* has the more specific sense 'to be lacking'. It does not occur in non-EP languages; outside Polynesian, Clark (1976: 98) mentions a verb *ore* in Sa'a (Solomon Islands) 'to remain behind' and Lau (Fiji) 'to fail, lack'. This may suggest that *kore* originated as a verb meaning 'to lack' and developed into something more like a negator in PCE (Clark 1976: 101–102).

10.5 Negation

### **10.5.8** *Hia***/***ia* **'not yet'**

*Hia* (etymology unknown) is used after negated verbs; the sense of the negator + *hia* is 'not yet'. *Hia* occurs immediately after the verb, before directionals:

(168) *¡Kai* neg.pfv *topa* descend *hia* yet *atu* away *ꞌō* really *tāꞌaku* poss.1sg.a *vānaŋa* word *koe* 2sg *i* pfv *pāhono* answer *rō* emph *mai* hither *ai!* pvp

'I hadn't finished speaking yet when you answered!' [R314.139]

*Hia* may occur with any negator: *kai* as in (168) above, *e ko* (169) or *taꞌe* (170).

(169) *E* ipfv *ko* neg.ipfv *ꞌoꞌoa* crow *hia* yet *te* art *moa* chicken *ka* cntg *kī* say *ena* med *e* ag *koe* 2sg *e* num *toru* three *kī* say *iŋa* nmlz *kai* neg.pfv *ꞌite* know *koe* 2sg *ko* prom *ai* who *a* prop *au.* 1sg

'Before the rooster crows, you will say three times that you don't know who I am.' [Jhn. 13:38]

(170) *He* ntr *ꞌaꞌamu,* tell *mata* while *taꞌe* conneg *ꞌite* know *hia* yet *pē* like *nei* prox *ē:* thus *he* pred *tahutahu.* witch 'She told it, without knowing yet that (the other person) was a witch.' [R532-07.044]

As (168–169) show, *hia* is often used when an action or event is interrupted by another event. In these cases, the function of the negator + *hia* is similar to a temporal marker 'before'.

Sometimes the variant *ia* is found. This should not be confused with the sentential particle *ia* 'then' (§4.5.4.1): while the latter occurs after the verb phrase, *ia* 'yet' occurs before other postverbal particles:

(171) *Kai* neg.pfv *tomo* go\_ashore *ia* yet *mai* hither *ꞌā* cont *ꞌi* at *te* art *ahiahi* afternoon *i* pfv *ꞌite* know *tahi* all *rō* emph *ai* pvp *te* art *ꞌuta* inland *i* acc *tū* dem *parau* word *ꞌāpī* new *era.* dist

'They had not arrived yet in the afternoon when all people ashore knew the news.' [R345.015]

#### 10 Mood and negation

### **10.6 Conclusions**

In this chapter, non-declarative moods have been discussed, as well as negation.

Two aspect markers serve to mark imperatives: the contiguity marker *ka* is used for direct commands, imperfective *e* for indirect commands. Though imperatives usually occur in the second person (often with explicit subject), they may occur in the third person as well. For first-person injunctions (e.g. exhortations), the purpose marker *ki* is used.

Polar questions usually do not have any special marking. Sometimes they are marked with the question marker *hoki*; in addition, the particles *ꞌō* and *hō* may be used to add a note of counterexpectation or doubt, respectively.

Content questions are marked by four question words, each of which belongs to a different word class:


Rapa Nui has three main clause negators: neutral *ꞌina*, perfective *kai* and imperfective *(e) ko*. *ꞌIna* is a phrase head; it may seem to have some properties of a predicate (e.g. triggering subject raising), but the same is true for a number of other clause-initial elements, such as deictic particles, while *ꞌina* lacks crucial features of a predicate.

The other two negators are preverbal markers; they are often combined with *ꞌina*. All units other than main clauses are negated by *taꞌe*: noun phrases, nominalised verbs, subconstituents and subordinate clauses. *Taꞌe* is also used to negate certain types of main clauses: those which have an *e*- or *i*-marked verb, preceded by an initial oblique constituent. This suggests that these clauses have some features of subordinate clauses: the initial oblique functions as a kind of matrix predicate (see Footnote 42 on p. 404).

## **11 Combining clauses**

### **11.1 Introduction**

Clauses can be combined in several ways. Two or more main clauses can be linked by juxtaposition or by using a coordinating conjunction (§11.2). Alternatively, one clause may contain another as subordinate clause: various categories of verbs take a clausal complement (§11.3); nouns may be modified by a relative clause (§11.4); adverbial clauses serve as an adjunct in a main clause (§11.6).

In Rapa Nui, different strategies are used to combine clauses, depending on the type of clause. Some constructions have a conjunction, others have a preverbal subordinating marker, others yet are unmarked.<sup>1</sup> Conjunctions only occur in certain types of adverbial clauses and will be discussed in the appropriate subsections of §11.6. Preverbal markers cut through the distinction between types of subordinate clauses, therefore they are discussed separately in §11.5.

### **11.2 Coordination**

### **11.2.1 Asyndetic and syndetic coordination**

Old Rapa Nui did not have a coordinating conjunction. Both phrases and clauses were linked by simple juxtaposition (i.e. asyndetic coordination, see Haspelmath 2007: 7). (1) shows juxtaposed clauses, while (2) contains a string of juxtaposed noun phrases.

(1) *He* ntr *oho* go *a* by *te* art *ara,* road *he* ntr *tikeꞌa* see *te* art *kohe; kohe he* ntr *rei* step *hai* ins *vaꞌe,* foot *he* ntr *hati* break *te* art *kohe… kohe*

'He went along the road, he saw a *kohe* plant; he stepped on it, the *kohe* broke…' [Ley-2-01.018]

(2) *He* ntr *māmate* pl:die *te* art *taŋata,* man *te* art *viꞌe,* woman *te* art *poki,* child *te* art *korohuꞌa.* old\_man 'Men, women, children, old people died.' [Ley-2-01.010]

<sup>1</sup> The latter can be recognised as subordinate clauses by the use of the negator *ta ꞌe* (see (211) on p. 557), and by the use of aspectual marker (only *i*, *e* or *ka*; the verb is always followed by a postverbal demonstrative).

#### 11 Combining clauses

Sometimes the adverbs *takoꞌa* and *hoki* 'also' (§4.5.3.2–4.5.3.3) are used to link clauses or phrases. In (3), two clauses with similar information about different participants are linked with *takoꞌa*. In (4), the last item in a list of noun phrases is marked with *hoki*. The latter happens only in older texts.


'To the side are the parents, the fathers, the mothers, the pretty girls, also the handsome boys.' [Ley-5-24.013]

In modern Rapa Nui, the conjunction *ꞌe* 'and' (probably a Tahitian borrowing<sup>2</sup> ) is used to link clauses and phrases; it occurs in clause- and phrase-initial position.

In old texts *ꞌe* is found a few times in Mtx, but not in other corpora (Ley and MsE); this suggests that *ꞌe* was emerging in the 1930s. In newer texts, it occurs over 3,000 times; this can be (partially) explained by changing speaking and writing styles under the influence of Spanish and other languages.

Even though *ꞌe* is very common nowadays, the most common strategy for linking clauses is still juxtaposition. Juxtaposition is especially used to link clauses referring to successive events in discourse. For example, in narrative, sequences such as the following are common:

(5) *He* ntr *tahuti* run *a* prop *Eva* Eva *ki* to *haho,* outside *he* ntr *oŋa* look *ki* to *te* art *vaka,* boat *he* ntr *takeꞌa* see *tōꞌona* poss.3sg.o *koro.* Dad 'Eva ran outside, stared at the boat, saw her Dad.' [R210.095]

In other situations, the conjunction *ꞌe* tends to be used. *ꞌE* is common in the following situations (the list is not exhaustive, and neither are these categories mutually exclusive):


good also

<sup>&#</sup>x27;Just go, eat and sleep.' [R304.013]

<sup>2</sup> Concerning the origin of *ꞌe* in Tahitian: although it is phonologically identical to French 'et', the fact that *ꞌe* is already common in the 1838 Tahitian Bible translation (Bibilia 1996) indicates that it predates French influence. A similar conjunction (spelled *e*, *ꞌe* or *ē*) occurs in Pa'umotu, Rarotongan and Mangarevan, but not in Marquesan and Māori.

	- 2. To link a pair of clauses not referring to successive events; these clauses are often parallel in some way and may involve a contrast between two items:
	- 3. To link subordinate clauses:

'You will return to finish your studies and to direct the mission work.' [R231.244]


#### 11 Combining clauses

When two clauses are both under the scope of a single initial constituent, they are usually juxtaposed without conjunction and without repetition of the initial constituent. Examples are *ꞌo ira* 'therefore' in (13) and the interrogative phrase in (14). As (14) also shows, verb phrase particles – both the aspectual and the negator *taꞌe* – are repeated in the second clause.


'Why didn't you come and tell me…?' [R313.106]

When two noun phrases are coordinated in modern Rapa Nui, they are usually linked with *ꞌe*. When the list is longer than two as in (16), *ꞌe* occurs only before the last item; the other items are juxtaposed:


'The rock piles, the rock gardens and the chicken houses gradually fell apart.' [R621.018]

When noun phrases marked with prepositions are coordinated, the preposition is repeated, including the accusative marker *i*; the last item may be preceded by *ꞌe* as in (17–18), but juxtaposition is also common, as in (19–20):


11.2 Coordination


In modern Rapa Nui, Spanish *pero* 'but' is often used as adversative conjunction:

(21) *He* ntr *maꞌu* carry *mai* hither *he* ntr *tunu* cook *i* acc *te* art *māmoe* sheep *pero* but *kai* neg.pfv *mākona* satiated *tū* dem *nuꞌu* people *era.* dist

'He carried the sheep and cooked it, but the people were not satiated.' [R183.033]

Despite its frequent use in everyday speech, *pero* is perceived as an intrusion, as witnessed by the fact that it is little used in the written texts in the corpus. In the Bible translation, it is not used at all. As (8) above shows, *ꞌe* is also used in situations where other languages would have an adversative conjunction.

### **11.2.2 Disjunction**

In old texts, disjunction is expressed by juxtaposition:

(22) *He* ntr *tia* sew *i* acc *te* art *nua* cape *hai* with *ivi* bone *manu,* bird *ivi* bone *moa,* chicken *ivi* bone *taŋata.* man '(The women of old) sewed capes with (needles made of) bird bones, chicken bones (or) human bones.' [Ley-5-04.013]

In modern Rapa Nui, disjunction is expressed by *ꞌo* 'or',<sup>3</sup> a conjunction borrowed from Spanish *o*. *ꞌO* may connect clauses as in (23–24) or phrases as in (25):

(23) *Te* ntr *ŋā* pl *kai* food *ꞌāpī* new *raꞌe* first *era* dist *ana* irr *momore,* red:cut *ana* irr *paꞌo,* chop *ꞌo* or *ana* irr *keri,* dig *e* ipfv *maꞌu* carry *to* art:of *te* art *hare* house *pure* pray *ꞌi* at *raꞌe.* first

'The first new food which would be picked, cut or dug up, had to be taken to the church first.' [R539-3.150]

(24) *…he* ntr *oho* go *ꞌo* or *he* ntr *hāpī* learn *ꞌo* or *he* ntr *eꞌa* go\_out *he* ntr *haꞌere* walk *ꞌo* or *he* ntr *oho* go *ki* to *kampō.* countryside '(When his work was finished,) he would go or study or go out for a walk or go to the countryside.' [R302.051]

<sup>3</sup> This particle should not be confused with preverbal *ꞌo* 'lest' (§11.5.4), or with the preposition *ꞌo* 'because of' (§4.7.3).

#### 11 Combining clauses

(25) *He* ntr *ꞌaiua* help *i* acc *te* art *aŋa* work *ki* to *a* prop *nua* Mum *ꞌo* or *ki* to *a* prop *koro.* Dad 'They help Mum or Dad with the work.' [R157.001]

Unlike *ꞌe* 'and', *ꞌo* may also connect nouns; in that case, the parts on either side of *ꞌo* are not complete noun phrases. In the following examples, *ꞌo* is directly followed by the second noun; prenominal elements, such as determiners and the plural marker *ŋā* in (26), precede the first noun, while the postnominal demonstrative *era* follows the second noun:


'Also bring a bag or box to put your sawfish in.' [R364.031]

### **11.3 Clausal arguments**

This section deals with verbs which take a clausal argument, i.e. an argument containing a predicate. This includes a number of different types of verbs: perception verbs such as *tikeꞌa* 'to see'; aspectual verbs such as *haꞌamata* 'to begin'; cognitive verbs such as *ꞌite* 'to know'; speech verbs such as *kī* 'to say', attitude verbs such as *haŋa* 'to want'; modal verbs such as *puē* 'can'. These verbs occur in a variety of multiclausal constructions:

	- 2. Nominalised complement clauses, in which the verb is introduced by the article *te*; it may be preceded by the acc marker *i*, as in the following example:

'Therefore, they didn't catch the people on top (of the islet) quickly.' [R304.048]

11.3 Clausal arguments

	- 4. Independent clauses:

It depends on the matrix verb which type of construction is used. Only types 1 and 2 involve a proper complement, that is, a constituent which is syntactically an argument of the matrix verb. For lack of a better term, constructions of types 3 and 4 will sometimes be referred to as "complement" or "complement clause" in the following sections, but one should bear in mind that this does not imply that they are syntactically a complement of the verb.

Types 3 and 4 are quite similar; in fact, 3 is a subset of 4, with the following two restrictions:


Despite their similarities, types 3 and 4 should be distinguished, as they occur with different verbs.

In addition to these four types of constructions, the same matrix verbs may also have a involve monoclausal constructions: nominal arguments and serial verbs. An example of a serial verb construction with complementation function is the following:

(32) *ꞌO* because\_of *ira* ana *i* pfv *hōrou* quick *i* pfv *oho* go *mai* hither *era* dist *ꞌi* at *tū* dem *mahana* day *era.* dist 'Therefore he went quickly that day.' [R105.108]

In the following subsections, the different categories of verbs mentioned above will be discussed in turn. In §11.3.7, the use of these different constructions will be summarised.

### **11.3.1 Perception verbs**

Perception verbs like *uꞌi* 'to see, watch', *hakaroŋo* 'to listen' and *ŋaroꞌa* 'to hear' can be followed by a nominal complement (§8.6.4.2), or by a clause which is syntactically independent of the perception verb (type 4 above). The latter will be discussed in the following subsections.

#### 11 Combining clauses

#### **11.3.1.1 Use of aspectuals**

When a perception verb is followed by a clause describing the perceived event, the range of aspect markers in this clause is limited: *ka*, *ko V ꞌā* and *e* are used, while *i* and *he* do not occur. The absence of perfective *i* is not surprising: events which are over and done with are usually not the object of perception. The absence of neutral *he* is not unexpected either: *he* is not able to provide the necessary temporal/aspectual link between the two clauses.

**Contiguity marker** *ka* When the clause expresses an activity or event which is perceived while it is happening, it is often marked with the contiguity marker *ka*. *Ka* (§7.2.6) expresses simultaneity between the event of perception and the event which is perceived: both take place at the same time.


**Perfect aspect** *ko V ꞌā* When the clause expresses a state of affairs which is perceived, it is marked with the perfect aspect *ko V ꞌā* (§7.2.7). This state of affairs may be the result of an event which has taken place before; what is seen is not the event itself but a situation from which the event can be inferred.

The *ko*-marked complement is often a stative verb or a temporal noun like *pō* 'night'; the perfect aspect expresses that this state has come about in some way, without specifying how. In (35) it is night because it has *become* night, and the ship is far from Rapa Nui because it has been *moving* further and further away.

(35) *He* ntr *uꞌi* look *atu* away *ko* prf *pō* night *ꞌā,* cont *ꞌe* and *ko* prf *roaroa* distant:red *ꞌana* cont *te* art *pahī* ship *mai* from *Rapa* Rapa *Nui.* Nui

'She saw that it was night, and that the ship was far from Rapa Nui.' [R210.116]

**Imperfective** *e* The third aspectual used after verbs of perception is imperfective *e*, usually followed by the continuity marker *ꞌā/ꞌana*. While *ko V ꞌā* indicates a state which has come about, *e V ꞌā* underlines the continuous nature of a situation, without implying the process by which it has come about (§7.2.5.4 on *e V ꞌā*).

(36) *ꞌĪ* imm *ka* cntg *uꞌi* look *atu* away *ena* med *ko* prom *te* art *repa* young\_man *ꞌi* at *roto* inside *e* ipfv *moe* lie *rō* emph *ꞌā.* cont 'Right then she saw a young man inside, lying down.' [R310.045]

11.3 Clausal arguments

(37) *He* ntr *uꞌi* look *atu* away *e* ipfv *huri* turn *rō* emph *ꞌā* cont *te* art *ꞌāriŋa* face *o* of *Heru* Heru *a* by *ruŋa.* above 'They saw that Heru was lying face up.' [R313.043]

#### **11.3.1.2 NP + clause**

Often a perception verb is followed first by an object NP expressing the person or thing which is perceived, then a clause specifying what happens to this referent (cf. English 'he saw someone coming'). The object NP in this construction may be marked in several ways: with the accusative marker *i* as in (38–39), but also with the topic marker *ko* as in (40–41) (§8.6.4.5). The verb in the complement clause is often marked with *ka*.


How should these constructions be analysed? At first sight, the complement clause in (38–41) can be considered as a relative clause to the object. One argument against this is the function of the aspect marker: whereas relative clauses marked with *ka* usually express an event posterior to that in the surrounding clause(s) (§11.4.3), in these examples the *ka*-marked clauses express an event simultaneous to the perception event of the matrix clause. Moreover, as (39) shows, the clause may follow a proper noun, even though proper nouns normally do not take relative clauses.

A second possibility would be to regard the object NP and the complement clause as two complements of the perception verb. This would mean that perception verbs, which normally take one complement, take two complements in this construction. Such an analysis would only be plausible if these arguments fulfilled different semantic roles. However, the noun phrase and the clause do not express different semantic roles connected to the action; neither do they express two instances of the same semantic role (\*'I saw him *and* coming'); rather, they are two aspects of a single semantic role: the nominal

#### 11 Combining clauses

complement refers to the perceived referent, while the clause expresses an action which is not only performed by that entity, but also part of the same perceived situation.

Therefore it seems more plausible to consider the nominal complement and the complement clause as a single constituent. The fact that the noun phrase can be marked with *ko* (which is the default case marker in the absence of other markers) is an argument for this analysis. Constructions (38–39) suggest that this noun phrase can be raised to the object position of the matrix verb.

### **11.3.2 Aspectual and manner verbs**

#### **11.3.2.1** *Haꞌamata* **'begin'**

*Haꞌamata* 'begin' is usually followed by a clause expressing the event which begins. In most cases this clause is juxtaposed, with the same verb marking as *haꞌamata*. Thus, both verbs may be marked with neutral *he* as in (42), perfective *i* as in (43), or perfect *ko V ꞌā* as in (44):


This identical marking is not limited to aspect markers. In (45), both verbs are marked with the negator *kai*. In (46), *haꞌamata* is the verb of a bare relative clause (§11.4.5), which is characterised by the absence of an aspect marker; the complement verb *tuꞌu* is likewise unmarked.


As (42–45) show, the S/A of the second verb is often placed in the subject position of the matrix clause. However, it may also be placed after the complement verb:

11.3 Clausal arguments

(47) *He* ntr *haꞌamata* begin *he* ntr *taŋi* cry *a* prop *Puakiva* Puakiva *ki* to *a* prop *Vaha.* Vaha 'Puakiva began to cry for Vaha.' [R229.149]

A second construction is that in which the complement is expressed as a nominalised verb (i.e. preceded by the determiner *te*). This complement may have the accusative marker *i* as in (48), but usually this marker is omitted, as in (49):


Despite the nominalised character of the complement, it still has verbal characteristics: its Patient (*i te māꞌea era* in (48)) is marked with *i*.

#### **11.3.2.2** *Oti* **'finish'**

The verb *oti* has several senses: 'to be finished, done, over' (e.g. a story), 'to run out', 'to be the only one'. One common use is 'to finish doing something', where *oti* is followed by a complement clause.

The complement verb is nominalised, i.e. marked with the article *te*. Sometimes it is preceded by the accusative marker *i*, in other cases *i* is omitted.<sup>4</sup> As the examples show, the subject of the second verb may be placed in the subject position of *oti* as in (50) and (52), or follow the complement verb as in (51) and (53).


burn

'…therefore it wasn't long before the house was completely burned.' [R250.120]

<sup>4</sup> In Māori, the complement verb never has the accusative marker; Hooper (1984a) argues that the complement verb is the subject of *oti*.

#### 11 Combining clauses

(53) *I* pfv *oti* finish *era* dist *te* art *kī* say *au,* 1sg *he* ntr *turu* go\_down *ko* prom *au* 1sg *ko* prom *te* art *viꞌe.* woman 'When I had finished saying this, I went down (to the coast) with my wife.' [Egt-02.066]

When the complement verb is transitive, the Patient may be raised to the subject position of *oti*, showing that the complement clause is passivised:

(54) *Ki* when *oti* finish *te* art *kōrua* 2pl *parau* document *te* art *tuhaꞌa* distribute *ꞌi* at *te* art *pōꞌā…* morning 'When your certificates have been handed out in the morning…' [R315.368]

*Oti* as a matrix verb with a complement may also be expressed in a serial verb construction. For examples, see (183–184) in §7.7.3.

#### **11.3.2.3** *Hōrou* **'hurry'**

*Hōrou* 'to hurry, (be) quick' is used as an adjective or adverb, but more commonly it is a main verb taking a clausal argument. This argument can be expressed in a variety of ways:


'The fish would not come quickly to be speared with a harpoon that had not been tied properly.' [R360.019]

11.3 Clausal arguments

As (58) shows, the subject of the second verb may be raised to the subject position of *hōrou* (in this case, the Patient is raised, showing that the complement clause is passivised).

#### **11.3.2.4** *Oho* **'go, about to'**

*Oho* 'go' usually refers to physical movement; in this sense, it is the most unmarked motion verb. *Oho* is also used as an aspectual verb, indicating that an event is about to happen (possibly under influence of Spanish *ir*, cf. Fischer 2007: 392). In this sense, *oho* is followed by a complement clause introduced by *mo*.


### **11.3.3 Cognitive verbs**

Cognitive verbs include *ꞌite* 'to know',<sup>5</sup> *aŋiaŋi* 'to know, be certain', *manaꞌu* 'to think' and the obsolete *maꞌa* 'know'. They may take a nominal object, which – depending on the verb – is marked with *i* or *ki* (§8.6.4.2).

The content of knowledge or thought may also be an event. This is expressed by an independent clause, which can be nominal as in (61) or verbal as in (62–63). As (64) shows, the clause may also be a dependent question. In each example, the bracketed part could function as a clause by itself.


'Ataranga knew for sure that his wife would not return.' [R532-01.019]

(63) *He* ntr *manaꞌu* think *rō* emph *ꞌai* subs *te* art *taŋata* person *o* of *nei* prox [*ko* prf *māmate* pl:die *ꞌā* cont *a* prop *koā* coll *Taparahi*]*.* Taparahi

'The people here thought that Taparahi and the others had died.' [R250.243]

<sup>5</sup> For *ꞌite* expressing possibility or ability, see §11.3.6 below.

11 Combining clauses

(64) *Ko* prf *ꞌite* know *ꞌana* cont *hoꞌi* indeed *kōrua* 2pl [*ꞌi* at *hē* cq *a* prop *ia* ]*.* 3sg 'For you know where she is.' [R229.277]

The content clause may be introduced by the phrase *pē nei ē* 'like this' (§4.6.5.1).

(65) *Ko* prf *ꞌite* know *rivariva* good:red *ꞌā* cont *e* ag *koe* 2sg *pē* like *nei* prox *ē:* thus *ko* prf *haŋa* want *ꞌā* cont *a* prop *ia* 3sg *mo* for *oho* go *mo* for *hāpī.* study

'You know very well that she wants to go and study.' [R210.066]

### **11.3.4 Speech verbs**

As discussed in §8.6.4.2, there are two types of speech verbs in Rapa Nui, 'say'-type and 'talk'-type verbs. Only the former, which include e.g. *kī* 'say' and *ꞌaꞌamu* 'tell', can be followed by a clause (or longer discourse) expressing the content of speech. This can be a direct speech, which usually follows without a specific marker:

(66) *He* ntr *kī:* say *'¡Ka* imp *moe* lie *ki* to *raro!'* below 'He said: "Lie down!"' [Ley-5-28a.003]

When the speech verb is followed by an indirect speech, it is often introduced by *pē nei (ē)* 'like this' (§4.6.5.1):

(67) *Kai* neg.pfv *kī* say *atu* away *e* ag *te* art *nuꞌu* people *hāpaꞌo* care\_for *i* acc *a* prop *koe* 2sg *pē* like *nei* prox *ē:* thus *a* prop *koe* 2sg *he* ntr *poki* child *ꞌa* of.a *Hakahonu.* Hakahonu 'The people who took care of you haven't told you that you are the child of Hakahonu.' [R427.016]

*Kī* 'say' may also be followed by a complement clause introduced by the purpose marker *mo* (§11.5.1); usually with a different subject, in the sense 'tell/ask someone to…', occasionally with the same subject, in the sense 'to tell one's intention'. The identity of the subject can only be known from the context.


11.3 Clausal arguments

### **11.3.5 Attitude verbs**

Under this heading a varied group of verbs is included which involve emotion, mental state, volition and desire. These include *haŋa* 'to want', *pohe* 'to desire', *riꞌariꞌa* 'to fear', *haꞌamā* 'to be ashamed', *manaꞌu* 'to consider, intend, decide' (for *manaꞌu* as cognitive verb, see §11.3.3).

These verbs may take a nominal complement introduced by *i* or *ki* (§8.6.4.2). They may also take a clausal complement introduced by *mo* (§11.5.1):


As these examples show, the complement clause usually has the same subject as the matrix clause and is unexpressed. A different subject is possible, though; this subject is expressed in the same way as in all *mo*-clauses (§11.5.1.2): usually as possessive, but sometimes with the agent marker *e*:


'I want you to wash me every day.' [R313.178]

Negative complements can be introduced by *ꞌo* 'lest' (§11.5.4), which expresses an adverse effect to be avoided.

(75) *ꞌAi* then *a* prop *Vai* Vai *Ora* Ora *ka* cntg *riꞌariꞌa* afraid *nō* just [*ꞌo* lest *māuiui* sick *rō* emph *ꞌi* at *te* art *rari*]*.* wet 'Then Vai Ora was afraid (her child) would get ill from being wet.' [R301.151]

#### 11 Combining clauses

(76) *Ana* irr *haŋa* want *koe* 2sg [*ꞌo* lest *manaꞌu* think *rahi* much *koe* 2sg *ki* to *te* art *poki*]*,* child *mo* for *tāua* 1du.incl *ꞌana* ident *e* ipfv *hāpaꞌo* care\_for *i* acc *a* prop *rāua* 3pl *ko* prom *Kava.* Kava 'If you don't want to worry (lit. if you want lest you think much) about the boy, we will care for him and Kava.' [R229.028]

### **11.3.6 Modal verbs**

Various verbs can be used to express modal concepts such as ability, possibility and obligation. These verbs are followed by a complement clause, which is in most cases introduced by *mo*. Most of these verbs are also used in other constructions, e.g. with a nominal complement. If the subject is expressed, it occurs in the main clause (except with *tiene que*, see below).

*Riva* and *rivariva* 'good', followed by *mo V,* express ability, possibility or permission:


'—Can't the children enter the canoe building site? —They can.' [R363.137–138]

When *ꞌite* 'to know' is followed by *i te V* (i.e. a nominalised verb marked as direct object), it often expresses ability, often a particular skill. Alternatively, it may express a habit or inclination, as in (80).

(79) *Ko* prf *ꞌite* know *ꞌā* cont *i* acc *te* art *pāpaꞌi,* write *i* acc *te* art *taiꞌo,* read *i* acc *te* art *vānaŋa* speak *i* acc *tētahi* other *ꞌarero…* tongue

'He could write, read, speak other languages…' [R539-1.052]

(80) *ꞌIna* neg *a* prop *au* 1sg *kai* neg.pfv *ꞌite* know *i* acc *te* art *kai* eat *i* acc *te* art *ꞌate.* liver 'I don't eat liver, I'm not used to eating liver.' [R245.238]

*Rovaꞌa/ravaꞌa* 'to obtain', followed by *mo V*, is used in the sense 'to be able, to succeed':

(81) *Kai* neg.pfv *ravaꞌa* obtain *e* ag *roto* inside *mo* for *haka* caus *raꞌu* hook *mai* hither *i* acc *te* art *kūpeŋa.* net 'Those inside (the net) did not succeed to hook the net.' [R304.128]

11.3 Clausal arguments

Possibility is often expressed by *puē*. This word is borrowed from Spanish *puede*, the third person sg. present tense of *poder* 'can, be able',<sup>6</sup> but is used in all persons and numbers. It is followed by *mo V*.


'Now I can teach the people well how to pray.' [R231.195]

*Tiene que*, which expresses both obligation ('have to') and necessity ('must'), is borrowed from Spanish *tiene*, the third person sg. present from *tener*. Just like *puē*, it is used for all persons and numbers. The complementiser *que* was borrowed along with the verb;<sup>7</sup> *que* is followed by a clausal complement, as in Spanish.<sup>8</sup>

The subject usually comes after the main verb as in (86); in this respect *tiene que* is different from other modal verbs, where the subject follows the modal verb immediately. However, (87) shows that the subject can be raised to the subject position of *tiene*.


<sup>6</sup> It is not uncommon for Spanish words to be borrowed in the 3rd person sg. present (Makihara 2001b: 197). The weak pronunciation of intervocalic *d* in Chilean Spanish facilitates its elision (§2.5.3.1); the resulting VV sequence coalesces into a single long vowel.

<sup>7</sup> In this respect *tiene* is less integrated into the language than *puē*, which takes the Rapa Nui complementiser *mo*. *Puē* is much more common in the text corpus (176x *puē*, 20x *tiene*). The difference in complementiser can also be explained from Spanish itself: the auxiliary *poder* (3sg. *puede*) is followed by a bare verb, a construction which would be highly unusual in Rapa Nui, hence the insertion of *mo*.

<sup>8</sup> See Makihara (2001b: 207–210) for more examples and discussion.

#### 11 Combining clauses

### **11.3.7 Summary**

As stated in the introduction to this section, while certain verbs are followed by a complement clause marked with a subordinating marker, other verbs are followed by a juxtaposed clause which is interpreted as semantic complement; yet others are followed by an independent clause. Table 11.1 summarises the use of these strategies for different types of verbs.

### **11.4 Relative clauses**

### **11.4.1 Introduction**

Relative clauses modify the head noun in a noun phrase. In Rapa Nui, as in most languages, the head noun is external to the relative clause itself; it is a constituent of a higher clause. As this noun has a semantic role both in the higher clause and in the matrix clause, Dixon (2010b: 317) uses the term *common argument* (CA).

In Rapa Nui, relative clauses are not marked by special markers or relative pronouns. They have the following syntactic features:


All these features will be discussed and illustrated below. First, a number of preliminary remarks.

• Relative clauses always modify an overt head noun. Headless relative clauses do not occur in Rapa Nui.

11.4 Relative clauses


#### Table 11.1: Complementation strategies

#### 11 Combining clauses


### **11.4.2 Relativised constituents**

In many languages, there are restrictions on the types of constituents that can be relativised. Keenan & Comrie (1977; 1979) account for this by proposing a "noun phrase accessibility hierarchy":

(88) subject > direct object > indirect object > oblique > possessor

All languages allow subject relativisation; not all languages allow relativisation of other constituents. A language may have one or more relativisation strategies; according to Keenan & Comrie, a given strategy will always apply to a continuous segment of this hierarchy.

This principle holds in many languages, though exceptions have turned up; among Polynesian languages, the hierarchy does not hold in Māori (see Harlow 2007a).

In this section, relativisation of different constituents in Rapa Nui will be discussed and illustrated. At the end of the section, the issue of the noun phrase hierarchy will be revisited.

**11.4.2.1 Subject** Subject relativisation is common. The subject is not expressed in the relative clause.


11.4 Relative clauses

**11.4.2.2 Object** When the object is relativised, it is not expressed in the relative clause. In object relative clauses, the subject is often *e-*marked. This conforms to a general pattern: *e*-marking of the subject is the rule in transitive clauses without an expressed object (§8.3.1).


Interestingly, the *e*-marked subject may precede the verb if it is pronominal, even though preverbal subjects in general are not *e*-marked (§8.3.1), and even though preverbal constituents in relative clauses are rare.

(94) *He* ntr *vaꞌai* give *tahi* all *e* ag *ꞌOho* Oho *Takatore* Takatore *i* acc *tū* dem *ŋā* pl *meꞌe* thing *taꞌatoꞌa* all *era* dist [*e* ag *ia* 3sg *i* pfv *maꞌu* carry *era* ]*.* dist

'Oho Takatore gave (him) all the things he had brought.' [R304.115]

Pronominal subjects are not always *e*-marked; in the following example, the subject pronoun is marked with the proper article *a*:

(95) *Ka* imp *hakaroŋo* listen *rivariva* good:red *mai* hither *tāꞌaku* poss.1sg.a *vānaŋa* word *nei* prox [*a* prop *au* 1sg *ka* cntg *kī* say *atu* away *nei* prox *ki* to *a* prop *koe* ]*.* 2sg

'Listen well to my words I am going to say to you.' [R229.243]

**11.4.2.3 Oblique** When oblique arguments<sup>9</sup> are relativised, the common argument is expressed pronominally in the relative clause.<sup>10</sup> As examples in native texts are scarce, two example from the Bible translation are given.

(96) *…nuꞌu* people [*ki* to *a* prop *rāua* 3pl *a* prop *au* 1sg *i* pfv *vaꞌai* give *ai* pvp *i* acc *te* art *māramarama* intelligent *mo* for *te* art *rāua* 3pl *aŋa* ] work

'people to whom I have given intelligence for their task' [Exo. 28:3]

<sup>9</sup> This includes "indirect objects" (§8.8.1).

<sup>10</sup> Silva-Corvalán (1978: 1) gives an example of an oblique argument relativised with gapping, but such a construction does not occur anywhere in my the text corpus.

#### 11 Combining clauses

(97) *A* prop *au* 1sg *he* pred *ꞌAtua,* God *kope* person [*ki* to *a* prop *ia* 3sg *e* ipfv *haꞌamuri* worship *ena* med *e* ag *te* art *kōrua* 2pl *tupuna* ]*.* ancestor 'I am God, the one whom your ancestors worshipped.' [Mat. 22:32]

**11.4.2.4 Adjunct** Adjuncts are relativised without being expressed in the relative clause. These usually express place as in (98) or time as in (99), but other adjuncts are possible as in (100).


'Do you know what the reason was (why) I didn't answer?' [R363.109]

In two situations a relativised locative constituent is represented by the pro-form *ira* (§4.6.5.2):

1. When the relative clause is a locative clause, i.e. the relativised phrase is predicate:

	- 2. When a preposition is needed to specify the nature of the locative relationship, for example, when a movement is involved from (*mai*) the referent:

As these examples show, the *ira* constituent is in clause-initial position in the relative clause.

11.4 Relative clauses

**11.4.2.5 Possessor** Relative clauses with possessor relativisation are rare, but they do occur. The possessor is expressed pronominally in the relative clause, in the same position where it would be in a main clause. In the following example, *te rāua* is coreferential to the head noun *nuꞌu*.

(103) *…tētahi* other *atu* away *nuꞌu* people *tuꞌu* arrive *atu,* away [*haru* grab *takoꞌa* also *i* acc *te* art *rāua* 3pl *henua* land *e* ag *te* art *fiko* ]*.* government

'…other people who had arrived, whose land the government had also grabbed.' [R649.055]

**11.4.2.6 Identifying predicates** Predicates of identifying clauses (§9.2.2) may also be relativised. In this case, the predicate is expressed in the relative clause as a pronoun preceded by *ko*. 11


'We'll ask the people who are the ones who know about this work.' [R535.193]

**11.4.2.7 Existential clauses** To relativise existential clauses, the verb *ai* 'to exist' is used. As discussed in §9.3, there are two subtypes of existential clauses: existentiallocative ('there is a house in the field', see §9.3.2) and possessive ('there is his house' = 'he has a house', see §9.3.3). An example of a relativised existential-locative is the following:

(106) *Kona* place [*ai* exist *o* of *te* art *miro* tree *o* of *rā* dist *hora* ] time *ko* prom *te* art *hare* house *pure.* prayer 'The place where there were trees at the time was the church.' [R539-1.524]

In this example the location (*kona*) is relativised, while the Existee (the entity that exists in a given place) is expressed in the relative clause, marked with the possessive preposition *o*.

The Existee can also be relativised, with the location expressed in the relative clause:

(107) *…he* pred *aha* what *te* art *meꞌe* thing [*i* pfv *ai* exist *ꞌi* at *Rapa* Rapa *Nui*] Nui

'(they want to know) what are the things that exist on Rapa Nui' [R470.006]

<sup>11</sup> See §9.2.2 for arguments to consider the *ko*-marked pronoun as predicate.

#### 11 Combining clauses

In possessive clauses, the possessor can be relativised as in (108); in this case, the possessee is expressed in the relative clause. The possessee can also be relativised as in (109–110), in which case the possessor is expressed in the relative clause.


'What was the problem that the woman in this story had (lit. that existed of this woman)?' [R616.603]

(110) *He* pred *haŋu* strength *pūai* strong [*taꞌe* conneg *ai* exist *i* at *te* art *taꞌatoꞌa* all *taŋata* ]*.* person '(*Mana*) was a strong force that not everyone had.' [R634.002]

As these examples show, possessees in the relative clause are marked with genitive *o* as in (108); possessors are marked either with *o* as in (109) or the general-purpose preposition *i* as in (110).

To summarise: there are two relativising strategies in Rapa Nui: one involving a gap (non-expressed constituent), one involving a resumptive element. It depends on the role of the relativised constituent which strategy is used; this is shown in Table 11.2. 12


Table 11.2: Relativisation strategies

We can now look once again at the noun phrase hierarchy given in (88) on p. 530 above. As Table 11.2 shows, whether or not the situation in Rapa Nui conforms to Keenan & Comrie's generalisation that every relativising strategy involves a continuous segment of the hierarchy, depends on how the syntactic categories of Rapa Nui are mapped to

<sup>12</sup> Existential clauses are not included separately in this table. When the Existee/possessee is relativised, it is the subject of the clause; when the possessor or location is relativised, it can be considered as an adjunct.

11.4 Relative clauses

this hierarchy. If oblique arguments (a category including arguments such as Recipients) are taken as a rough equivalent of their category of "indirect object", the gapping strategy in Rapa Nui does not apply to a continuous segment of the hierarchy: it applies to subjects, direct objects and adjuncts (with the latter, the pronoun strategy also occurs, but marginally), but not to "indirect objects".

### **11.4.3 Aspect marking in relative clauses**

The most common aspect markers in relative clauses are perfective *i* and imperfective *e*. *Ka* and *ko* are not unusual either, but *he* is rare. All of these will be briefly discussed in turn.

Perfective *i* is the most general aspectual in relative clauses. It may mark events performed at the same time as the events in the main clause as in (111), or completed prior to the events in the main clause as in (112); it may also mark states as in (113).

The verb may be followed by a postverbal demonstrative (including *ai*), but this is optional.

(111) *ꞌI* at *tū* dem *hora* time *era* dist [*Eva* Eva *i* pfv *ŋaroꞌa* perceive *era* dist *i* acc *tū* dem *vānaŋa* word *era* dist *ꞌa* of.a *koro* ]*,* Dad *he* ntr *hakaroŋo* feel *atu…* away

'At the moment Eva heard those words Dad (spoke), she felt…' [R210.075]


'He was the only person in the house who wasn't sick.' [R250.091]

Imperfective *e* in relative clauses often refers to events which are going on at the time of reference, as in (114). Alternatively, it may indicate events which happen repeatedly or habitually, as in (115). The verb is usually followed by a postverbal demonstrative (§7.2.5.4).

(114) *…ꞌi* at *te* art *reka* entertaining *o* of *te* art *rāua* 3pl *ara* way [*e* ipfv *oho* go *era* ]*.* dist '(Eva stopped crying,) because of the enjoyment of the trip they were making.' [R210.137]

#### 11 Combining clauses

(115) *Te* art *aŋa* work *ꞌa* of.a *Puakiva* Puakiva [*e* ipfv *ꞌavai* give *era* dist *e* ag *Pipi*]*,* Pipi *he* pred *apaapa* gather *hukahuka…* firewood:red 'The work Puakiva got assigned by Pipi, was gathering firewood…' [R229.396]

Perfect *ko/ku V ꞌā* indicates a state which has come about in some way: with event verbs as in (116), the state is the result of the event described by the verb; with statives as in (117), the situation has resulted from some unspecified process.


When the contiguity marker *ka* is used in a relative clause, the clause expresses an event posterior to the events in the context. In direct speech this means the clause refers to the future, as in (118); in narrative texts the *ka*-marked relative clause is posterior with respect to the time of the main action, as in (119). The verb is always followed by a postverbal demonstrative.


Neutral *he* is rarely used in relative clauses. In the few examples I found, its function seems to be similar to *ka*:

(120) *ꞌI* at *te* art *mahana* day *era* dist [*he* ntr *oho* ]*,* go *ko* prf *ꞌara* wake\_up *ꞌā* cont *a* prop *Eva* Eva *ꞌi* at *te* art *hora* time *ono* six *o* of *te* art *pōꞌā.* morning 'On the day she was going to leave, Eva woke up at six in the morning.' [R210.028]

11.4 Relative clauses

Finally, relative clauses may be marked with the purpose marker *mo* (§11.5.1), in which case they express an event destined to happen:


### **11.4.4 Possessive-relative constructions**

In possessive-relative constructions, the head noun is preceded or followed by a possessor, which is coreferential to the subject of the relative clause; the latter is not expressed in the relative clause itself. (These constructions occur in Rapa Nui as well as in various other Polynesian languages.) Possessive-relative constructions only occur when a constituent other than the subject is relativised; they are found with both object and adjunct relativisation. An example is the following:

(123) *¿He* ntr *aha* what *te* art *kōrua* 2pl *meꞌe* thing [*i* pfv *aŋa* do *ꞌi* at *ꞌApina* ]*?* Apina 'What did you do (lit. what [is] your thing did) in Apina?' [R301.197]

Syntactically, *te kōrua* is a possessive pronoun modifying *meꞌe* 'thing'; it is coreferential to the implied subject of the relative clause.

When the possessor is pronominal, it may either precede the noun as in (124–125), or follow it as in (126) (§6.2.1):

(124) *…mo* for *haka* caus *oho* go *ki* to *tāꞌana* poss.3sg.a *kona* place *era* dist [*i* pfv *pohe* ]*.* desire

'…to make (the horse) go to the place he wanted (it to go).' [R345.087]


<sup>13</sup> That this is a relative clause, not just a modifying verb, is shown by the verb phrase particle *mai*.

#### 11 Combining clauses

When the possessor is a full noun phrase, it must occur after the noun:


Possessive-relative constructions occur in other Polynesian languages as well. There has been some discussion on the question whether the possessor is raised from the subject position of the relative clause (e.g. Harlow 2000: 367; Harlow 2007a: 185), or whether it is a genuine noun phrase possessor which happens to be coreferential to the relative clause subject (Clark 1976: 116). In Rapa Nui the second option is more plausible. First, the possessor can be in the same positions as in any other noun phrase, which suggests that it is no different from other possessors in the noun phrase. Second, as the examples above show, the form of the possessive construction varies between *a-* and *o-*possession: *a*-possession in (124), (126) and (128), *o*-possession in (125) and (127). As *a-* and *o-*possession express different semantic relationships between possessor and possessee (§6.3.2), this suggests that there is a direct relation between the possessor and the head noun, even though the primary function of the possessor seems to be the expression of the relative clause subject.<sup>15</sup> And indeed, in most of these cases the choice between *ꞌa* and *o* is governed by the same principles guiding this choice in possessive constructions in general. In (128), where the relation between possessor and head noun is one between husband and wife, *ꞌa* is used (§6.3.3.1.1). In (126), the use of *ꞌa* is possibly motivated by the active relationship of the possessee to the head noun 'work' (§6.3.3.2 item 2). In (127), *o* is used with a temporal noun, again conforming to a general pattern (§6.3.3.3 item 11). In fact, given the wide range of relationships expressed by possessive constructions in Rapa Nui, all possessive-relatives seem to exhibit some kind of possessive relationship also attested in simple possessive constructions.

If this analysis is correct, the possessor is not the result of raising, but is a normal noun phrase possessor which happens to be coreferential to the relative clause subject. Under coreferentiality, the latter is left unexpressed.

This analysis is confirmed by the fact that there are also possessive-relative constructions where the possessor is not the subject of the relative clause, but an oblique/embedded constituent as in (129):

<sup>14</sup> Examples such as (128) are potentially ambiguous. As discussed above, in object relative clauses the subject is sometimes preverbal and preceded by the proper article *a* (see (95) in §11.4.2 above). Now the proper article *a* is homophonous to the possessive preposition *ꞌa*, and both may be followed by proper nouns; therefore, in examples such as (128), the subject could also be analysed as a nominative subject marked with the proper article *a.* However, an analysis as genitive (i.e. *ꞌa* rather than *a*) is more plausible, as only pronouns occur unambiguously as preverbal subjects in the relative clause; noun phrase subjects in relative clauses are always postverbal (see e.g. (92–93) above).

<sup>15</sup> Herd, Macdonald & Massam (2011) make a similar observation for other Polynesian languages. They propose a structure where there is a relation between the possessor and the relative construction as a whole. This involves a control relation (not raising) between possessor and relative clause subject.

11.4 Relative clauses

(129) *He* pred *pura* only *mata* eye *te* art *kōrua* 2pl *meꞌe* thing [*takeꞌa* see *mai*]*.* hither 'Your eyes are the only thing that can be seen (lit. mere eyes are your thing seen).' [R245.217]

### **11.4.5 Bare relative clauses; verb raising**

Bare relative clauses are relative clauses in which the verb is not preceded by an aspectual. In Rapa Nui in general, the aspectual is obligatory, except in a few well-defined contexts (§7.2.2), one of which is when the verb is adjectival (i.e. functions as noun modifier). Even so, in §5.7.2.3 I argued that bare relatives are different from adjectival modifiers: unlike the latter, they are truly verbal in that they indicate an event taking place at a specific time; moreover, they can be followed by verb phrase particles and verb arguments.

Here are a number of examples of bare relative clauses.


now.' [R364.001]

These examples show that bare relative clauses are not limited to one single aspect. In most cases they express a one-time event which has been completed as in (130), i.e. the clause has perfective aspect; however, they may also be habitual as in (131), durative as in (132), or stative as in (133).

As these examples also show, the verb tends to come straight after the head noun. Only in (130) are noun and verb separated by the adjective *varavara*. Other elements occasionally occurring between noun and verb are quantifiers as in (134) and postnominal demonstratives as in (135):

11 Combining clauses


Even though noun and verb can be separated by these noun phrase elements, there is a strong tendency to place the verb adjacent to the noun. Often the verb is raised to a position straight after the noun, before other noun phrase elements. In (136), the verb *hatu* is raised to a position before the quantifier *taꞌatoꞌa*, while the subject of the relative clause is stranded after *taꞌatoꞌa*. (The status of *era* is discussed below.)

(136) *He* ntr *oho* go *tū* dem *poki* child *era* dist *pē* like *tū* dem *meꞌe* thing [*hatu* ] advise *taꞌatoꞌa* all *era* dist [*e* ag *tū* dem *rūꞌau* old\_woman *era* ]*.* dist 'The boy went (and did) like all the things advised by the old woman.' [R310.105]

Similarly, in (137), the verb *tuꞌu* is raised over the postnominal possessor *ꞌāꞌana*. Notice that even though the relative clause only consists of a verb, it is still a true relative clause, not an "adjectival" verb: *tuꞌu* refers to a specific event, it is not a time-stable property of the child (§5.7.2.3).

(137) *He* ntr *vānaŋa* talk *ararua* the\_two *ko* prom *tū* dem *poki* child [*tuꞌu* ] arrive *era* dist *ꞌāꞌana.* poss.3sg.a 'She spoke with her child who had arrived.' [R532-01.007]

In (138) the verb *hiŋa* is raised both over the particle *ꞌā* and the possessor *o te poki*. <sup>16</sup> The same happens in (139), where the possessor 'of the morning' modifies the head noun, while the next phrase 'to school' is the part of the relative clause left stranded.


'…from the morning time, when he went down to school, until then' [R245.009]

<sup>16</sup> *ꞌĀ* occurs both in the noun phrase (expressing identity) and in the verb phrase (expressing continuity); here it is a noun phrase particle, modifying the noun: 'the very same day'.

11.5 Subordinating markers

Examples (136–139) all involve a demonstrative *era*. Now this demonstrative (as well as *nei* and *ena*) is common both in the noun phrase and in the verb phrase, so *a priori* it may be either a postnominal particle over which the verb has been raised, or a verb phrase particle belonging to the relative clause. The position of *era* in the examples suggest that the former is the case, as indicated by the brackets. *Era* occurs after the quantifier in (136), but before the possessor in (137) and before the particle *ꞌā* in (138–139); in other words, *era* occurs in its usual noun phrase position (see the chart in §5.1). If *era* were a verb phrase particle, it would be unclear why it is raised with the verb in (137–139), but left stranded in (136).

Another reason to consider *era* as postnominal rather than postverbal, is that it cooccurs with the demonstrative *tū*, which is always accompanied by a postnominal demonstrative (§4.6.2.1). When *tū* co-occurs with *era* after the verb, this suggests that the verb has been raised.<sup>17</sup> This is illustrated in (136) above; the same analysis can be extended to examples such as the following:


In other words, even though *kī era e nua* in (141) seems to be a relative clause, the presence of *tū* suggests that *era* is not part of the relative clause, but is a noun phrase particle which has been leapfrogged over by the verb.

Examples such as (140–141) are quite common. In fact, the tendency to leave out the aspectual and (if needed) to raise the verb is strongest with definite/anaphoric noun phrases like the ones illustrated here. Leaving out the aspectual has the effect of downplaying the action/event character of the relative clause: what the relative clause denotes is not so much an event but rather a fact; this fact is part of the referential description in the noun phrase.

### **11.5 Subordinating markers**

The preverbal markers *mo*, *ana*, *ki*, *ꞌo* and *mai* are used to mark certain types of clauses. As these markers occur in the same position as aspectuals (§7.1), they do not co-occur with the latter, which means that a clause containing one of these particles is not marked for aspect.

<sup>17</sup> Relative clause verbs may have a postverbal demonstrative, even when the head noun also has a demonstrative; see *nuꞌu era* [*oho era* ] in (135). Nevertheless, raised verbs never have a demonstrative of their own: two consecutive demonstratives never occur (*\*nuꞌu* [*oho* ] *era* [*era* ]). This can be accounted for by a rule deleting one of two consecutive demonstratives.

#### 11 Combining clauses

In subordinate clauses, these markers are always clause-initial; no constituents are placed before the verb phrase. *Ana*, *ki* and – somewhat marginally – *mo* also occur in main clauses. As their functions in main and subordinate clauses are clearly similar, all their uses will be discussed together in the following sections, with two exceptions:


### **11.5.1 The purpose/conditional marker** *mo*

*Mo* is by far the most common subordinating marker.<sup>18</sup> It is used to mark complements of cognitive verbs (§11.3.3), speech verbs (§11.3.4), attitude verbs (§11.3.5) and modal verbs (§11.3.6). In addition, it marks both purpose clauses and conditional clauses; these will be discussed in §11.5.1.1. §11.5.1.2 discusses the expression of arguments in *mo-*clauses. Occasionally *mo* occurs in main clauses; this is discussed in §11.5.1.3.

#### **11.5.1.1** *Mo* **in adverbial clauses**

*Mo* marks purpose clauses. As the examples below show, the *mo*-clause usually follows the main clause.


*Mo* also marks conditional clauses.

(145) *Mo* if *mate* die *tāꞌue,* by\_chance *ꞌo* or *mo* if *ŋaro,* lost *he* ntr *rahi* much *tuꞌu* poss.2sg.o *māuiui* sick *ꞌi* at *te* art *ꞌaroha.* pity 'If (the bird) dies accidentally, or if it gets lost, you would suffer much from feeling sorry.' [R213.027]

<sup>18</sup> Preverbal *mo* probably developed from (or is an extended use of) the benefactive preposition (§4.7.8). To my knowledge, Rapa Nui is the only language in which *mo* developed into a preverbal marker. The fact that the subject is often expressed as a possessor (§11.5.1.2) may be a trace of the prepositional character of *mo*.

11.5 Subordinating markers

(146) *Meꞌe* thing *ihu* nose *piꞌipiꞌi;* crushed:red *mo* if *vānaŋa* talk *mai,* hither *meꞌe* thing *reꞌo* voice *huru* manner *kē.* different 'They are snub-nosed; if they talk, they have a strange voice.' [R310.252]

Conditional clauses usually precede the main clause as in these examples, though this is not a rigid rule.

As *mo* is a preverbal marker, it is always immediately followed by a verb. This means that *mo* constructions would be impossible with nominal clauses, or in contexts where a different preverbal marker is called for (e.g. the negation *kai*). In such cases, the existential verb *ai* can be employed as an auxiliary verb. *Ai* in turn is followed by a clause which is structured as a main clause (this construction is further discussed in §9.6.1).


#### **11.5.1.2 Arguments in the** *mo***-clause**

**The S/A argument** The S or A argument of the *mo*-clause is often coreferential to the subject of the main clause, in which case it is usually not expressed. See e.g. (143) above.

When the S/A argument is expressed, it is either as a possessive (with preposition *o* or a possessive pronoun of the *o*-class) as in (149–151), or with the agent marker *e* as in (152–155). The latter occurs more or less in the same contexts as in main clauses (§8.3): in transitive VA-clauses without explicit object as in (152); in VOA-clauses as in (153); with verbs like *ŋaroꞌa* as in (154); when it is contrasted with other referents as in (155).


#### 11 Combining clauses


The fact that the S/A argument is often expressed as a possessive, does not mean that the *mo*-clause is nominal. Apart from the possessive constituent, the clause is wholly verbal: the verb is not preceded by a determiner, it may be followed by VP particles such as *mai* in (155), and as the same example also shows, the object may have the accusative

**The O argument** The O argument of a *mo-*clause is either expressed as a direct object – preceded by the accusative marker *i* – or as a possessive. (155) above and (156) below show *i*-marked direct objects; in (157–158), the O is expressed as a possessive.


marker *i*.

11.5 Subordinating markers

(158) *…he* ntr *vahivahi* divide:red *mo* for *tatau* to\_milk *o* of *te* art *puaꞌa,* cow *mo* for *hāŋai* feed *o* of *te* art *oru* pig *ꞌe* and *mo* for *puru* close

*o te hoi.*

of art horse

'…he divided (the piece of land) to milk cows, to raise pigs and to enclose horses.' [R250.047]

I have not noticed any difference between the two constructions. There may be a distinction in prominence, with less significant objects marked as possessive. However this may be, object marking in *mo-*clauses is significantly different from object marking in main clauses: contexts where the object is possessive are not the same contexts where the object would be zero-marked in main clauses (§8.4.1).

#### **11.5.1.3** *Mo* **in main clauses**

Occasionally preverbal *mo* is used in main clauses. In these clauses, the subject is always expressed; the constituent order is almost always SV/AVO. When the subject is a pronoun or proper noun, it is marked with *ko*. This structure reminds of clauses with *ko*-marked topicalised subjects (§8.6.2.1).

The general sense is that of a subject being destined in some way to perform the action described by the verb. Depending on the context, the clause may express a plan or intention as in (159), an instruction as in (160), or permission as in (161).


With a negation, *mo*-clauses may express a prohibition or dissuasion. Several negative constructions occur. The constituent negator *taꞌe* can be used to negate the subject as in (162) or the predicate as in (163). A construction with the clause negator *ꞌina* is also possible, as in (164).

(162) *Taꞌe* conneg *māua* 1du.excl *mo* for *moto* fight *hakaꞌou.* again 'We should not fight any more.' [R211.014]

#### 11 Combining clauses


More work is needed to find out the exact function of *mo* in main clauses, and the syntactic constraints that apply in this construction.

### **11.5.2 The irrealis marker** *ana*

*Ana* is an irrealis marker.<sup>19</sup> The irrealis mode, as defined by Payne (1997: 244), does not assert that the event has happened or will happen. Neither does it assert that the event did *not* happen or will not happen: the irrealis refrains from any claim about the truth of the proposition expressed by the clause.

*Ana* is mostly used to mark events which may or may not happen, for example intentions, possibilities and obligations; this will be amply illustrated in the following subsections.

In some cases the event has actually happened; this is not inconsistent with the irrealis as defined above. In the following example, the speaker refers back to a question her interlocutor has just asked:

(165) *¿Mo* for *aha* what *ꞌana* ident *koe* 2sg *ana* irr *ꞌui* ask *rō* emph *mai?* hither 'Why would you ask this?' [R315.028]

Even though the asking is a real event, the speaker refers to it as something unrealised, perhaps conceived as a more general truth ('why would anybody ask something like this?'), or as something which is inherently improbable.

*Ana* occurs in the same structural position as aspect markers; *ana* and aspect markers are mutually exclusive. Clauses marked by *ana* are therefore not differentiated for aspect (but see (185) below).

As (165) shows, *ana* can be followed by evaluative markers (*rō*) and directionals (*mai*). It cannot be followed by postverbal demonstratives or the VP-final particles *ꞌā* and *ꞌai*.

The following subsections will deal with uses of *ana* in main clauses (§11.5.2.1) and subordinate clauses (§11.5.2.2), respectively.

<sup>19</sup> This particle does not occur in any other language, with the exception of Māori *ana* 'if and when' (Biggs 1973: 130), which corresponds to the use of Rapa Nui *ana* in conditional/temporal clauses.

11.5 Subordinating markers

#### **11.5.2.1** *Ana* **in main clauses**

**11.5.2.1.1** *Ana* is used to expressintentions. While the outcome of the intended event is inherently uncertain, the intention itself may be quite firm: (166) occurs in a context where two parents have just agreed to call their baby Tahonga; in the quoted sentence, this decision is confirmed.


**11.5.2.1.2** *Ana* may express potential events, events which may or may not happen.

(168) *A* by *ꞌuta* inland *hō* dub *a* prop *Vaha* Vaha *ana* irr *oho* go *rō.* emph 'Vaha might go by the inland way.' [Mtx-3-01.142]

Whether the event will happen or not, may depend on a condition which is stated explicitly. Thus, *ana* may occur in the apodosis, the clause expressing the consequence of a conditional or temporal clause.

(169) *Ki* when *hāhine* close *nō* just *tāua* 1du.incl *mo* for *tuꞌu* arrive *ana* irr *maꞌu* carry *iho* just\_then *e* ag *au* 1sg *te* art *kai.* food 'When we are close to arrival, then I will take the food.' [R215.026]

Even without a conditional clause construction, the occurrence of the event marked by *ana* may be contingent on another event: it is the result of, or at least follows upon, an event expressed in an earlier clause: 'X, only then Y'. In this case – as in (94) above – the verb is usually followed by *iho* 'just then'.


'We will go up to the crater or to Ro'iho, where we (will/may) find water.' [R487.035]

<sup>20</sup> *He me ꞌe ꞌo* is a now obsolete construction expressing prohibitions.

#### 11 Combining clauses

As a marker of potentiality, *ana* is also used in content questions. The question may be a real one to which an answer is expected as in (172), or a rhetorical one as in (173):


**11.5.2.1.3** *Ana* also has a deontic use: it is used to express instructions, obligations or norms, as well as permission.


In the second person, deontic *ana* is similar in function to imperative *ka* and exhortative *e*. While the latter two are only used with clause-initial verbs, *ana* is especially used when the verb phrase is non-initial. In (177), initial *e* alternates with non-initial *ana*:


'Worship Jehovah your God, and serve only him.' [Mat. 4:10]

**11.5.2.1.4** *Ana* may also mark clauses which express a general practice, something which **is** normally/usually done in a given situation. This use is found especially in procedural contexts, where the speaker describes how certain things are normally done or should be done. In Rapa Nui, procedures are generally expressed by strings of *he*clauses, with occasional imperatives (see (5) on p. 317). But *ana* may be used as well, especially when the verb is non-initial.

11.5 Subordinating markers


Examples like (178) could be considered as deontic, prescribing how something should be done. However, (179) shows that *ana* is used even when the procedure is not an instruction to the present-day hearer, but a description of how something was done in the past. Such contexts can be considered irrealis, as they do not describe events which happened at a specific occasion.<sup>21</sup>

#### **11.5.2.2** *Ana* **in subordinate clauses**

**11.5.2.2.1** In subordinate clauses, *ana* is used to express a condition: the event may or may not happen, but only if it happens will the event in the main clause take place. The conditional clause tends to precede the main clause.


As these examples show, the apodosis is usually marked with an aspectual, i.e. in the realis mood. Alternatively, the apodosis may also be marked with *ana* (cf. (169) above). This can lead to a situation in which both the conditional clause and the apodosis are marked with *ana*:

(182) *Ana* irr *haŋa* want *mo* for *hakarere* leave *nō* just *ꞌi* at *Orohie,* Orohie *ꞌi* at *rā* dist *ꞌana* ident *ana* irr *hakarere…* leave 'If they want to leave (the statue) in Orohie, there they leave it…' [Ley-4-06.015]

<sup>21</sup> Payne (1997: 245) points out that habitual aspect is less realis than perfective aspect.

#### 11 Combining clauses

In other cases, the question is not *whether* the event in the subordinate clause happens, but *when*: the event is expected to happen or has already happened, and the same is true for the main clause event dependent on it. However, *ana* signals that the clause is still irrealis in some way. It may indicate an event which takes or took place habitually (see the discussion about (179) above), or an event which is expected (with more or less certainty) to take place in the future. *Ana* is not used with events which have taken place at a definite moment in the past.


**11.5.2.2.2** *Ana* also occurs in dependent polar qestions ('whether'):

(185) *ꞌĪ* imm *ꞌō* really *a* prop *au* 1sg *he* ntr *oho* go *he* ntr *uꞌi* look *ana* irr *ai* exist *ko* prf *ꞌara* wake\_up *ꞌana.* cont 'I'm going straightaway and look whether she has woken up.' [R229.366]

While *ana* is usually followed by the main verb of the clause, sometimes it is followed by the existential verb *ai* 'exist' (just like *mo*, see (147) on p. 543); the rest of the clause follows as a complement to this verb. This allows the speaker to use *ana* with a nonverbal clause as in (186), or to express aspect in addition to irrealis, as in (185) above, where the main verb is marked with perfect aspect *ko*.

(186) *He* ntr *ꞌui* ask *e* ag *Aio* Aio *ki* to *tū* dem *korohuꞌa* old\_man *era* dist *ana* irr *ai* exist [*pē* like *ira* ana *mau* really *te* art *parautiꞌa* ]*.* truth 'Aio asked the old man if those things were true (lit. if it was: like that [was] the truth).' [R532-14.016]

### **11.5.3 The purpose/temporal marker** *ki*

The preverbal marker *ki* is used in subordinate clauses expressing time ('when') and purpose ('in order to, so that'). In main clauses it marks hortatives, i.e. first-person injunctions. In this section, its use in subordinate clauses is discussed; hortatives are discussed in §10.2.3.

Even though *ki* is homophonous to the preposition *ki*, the two are probably etymologically distinct. The verbal marker *ki* is probably derived from PPN *\*kia*, which occurs in

11.5 Subordinating markers

many languages with an optative and/or purposive sense.<sup>22</sup> If this is correct, the preposition and the verbal marker *ki* were distinct lexemes in the protolanguage. However, because of the goal-oriented character of preverbal *ki*, it is glossed 'to', just like the preposition.

**11.5.3.1** For purpose clauses, the default marker is *mo* (§11.5.1.1). *Ki* is used especially in the following circumstances:

In the first place, after an imperative or hortative.


When the *ki*-clause has a first person plural subject as in (188), the clause may have hortative overtones: 'so we (can) go' > 'let's go'.

Secondly, when *mo* would be potentially ambiguous. In (189), the main verb *pohe* is followed by a complement clause marked with *mo*. If the next clause were also marked with *mo*, it could be read as a second complement of *pohe*; to ensure a reading as purpose clause, *ki* is used. The same happens in (190): while the *mo*-clause expresses the purpose of the preceding main clause, the *ki*-clause after that expresses the ultimate purpose, the higher-order goal of the preceding clauses as a whole.

(189) *ꞌĪ* imm *e* ipfv *pohe* desire *atu* away *ena* med *mo* for *ꞌata* more *noho* stay *mai* hither *ki* to *ꞌata* more *keukeu* labour:red *ai* pvp *tētahi* other *aŋa.*

work

'I would like him to stay here a bit more, in order to get other projects done.' [R204.005]

(190) *O* of *te* art *hānau* race *ꞌeꞌepe* corpulent *i* pfv *keri* dig *ai* pvp *i* acc *te* art *rua…* hole *mo* for *pae* finished *o* of *te* art *hānau* race *momoko,* slender *ki* to *noho* stay *e* ag *hānau* race *ꞌeꞌepe* corpulent *nō.* just 'The 'corpulent race' dug a hole… to exterminate the 'slender race', so the

'corpulent race' would be the only ones (left).' [Ley-3-06.019]

<sup>22</sup> *Kia* was shortened to *ki* in various languages. Clark (1976: 30) mentions Kapingamarangi, Nukumanu, Sikaiana and Luangiua. Hawaiian *i* (Elbert & Pukui 1979: 61) seems to represent the same particle. As the particle is *kia*/*ꞌia* in most CE languages, the shortening to *ki* in Rapa Nui must have been an independent development which took place after Rapa Nui broke off from PEP (§2.5.2 on the monophthongisation of particles). This process may have taken place relatively recently: there are a few occurrences of *kia* in older texts, mostly in fossilised phrases such as *ka oho, kia tika* 'go straight' (Mtx-2-03.018; Mtx-6-07.014); see discussion in Fischer (1994: 429). Nowadays *kia* survives in *kiahio* 'keep courage, be strong' (cf. *hiohio* 'strong').

#### 11 Combining clauses

Thirdly, to express a result not intended by the main-clause subject. This is illustrated in the following two examples. The *ki*-clause does not express a purpose which the main-clause subject had in mind; rather, it is a purpose external to the intentions of the subject.


As these examples show, the subject of the *ki*-clause is expressed in the same way as in main clauses: either unmarked as in (187–188) or with the agent marker *e* as in (190). In this respect, *ki*-clauses are different from *mo-*clauses, which usually have a possessive subject.

A peculiarity of *ki*-clauses with purpose sense, is that the verb is often followed by *ai*, the postverbal demonstrative which otherwise only occurs after *i* (§7.6.5). This is illustrated in (189) and (191) above.

**11.5.3.2** *Ki* also marks temporal clauses.<sup>23</sup> As the examples below show, these occur in various contexts: with past reference, with future reference, or habitual. *Ki*-clauses usually occur before the main clause, but as (196) shows, they may also be placed after the main clause.


<sup>23</sup> The double function of reflexes of PPN *\*kia* as both optative/purposive and temporal markers is also found with Māori *kia* (Bauer, Parker & Evans 1993: 62, 459) and Tahitian *ꞌia* (Lazard & Peltzer 2000: 138–139); unlike Rapa Nui, in these languages the particle is not used in temporal clauses referring to the past. In Rapa Nui, the purposive sense of *\*kia* has to a large degree been taken over by *mo*, as discussed above.

11.5 Subordinating markers

(195) *Ki* when *oho* go *ararua* the\_two *e* exh *maꞌu* hold *te* art *rima.* hand 'When the two of you walk together, hold hands.' [R166.004]

(196) *He* ntr *aŋa* work *tātou* 1pl.incl *he* ntr *haka* caus *hōrou* quick *mo* for *turu* go\_down *o* of *tātou* 1pl.incl *ki* to *tai* sea *ki* when *tuꞌu* arrive *mai* hither *a* prop *nua.* Mum

'We will work quickly, so we can go to the sea when Mum comes.' [R229.456]

*Ki*-marked clauses may indicate a goal or temporal boundary: 'until'. This occurs for example after the verb *tiaki* 'wait'.<sup>24</sup>


'The people stayed there until Mahina Tea's eyes had healed well.' [R399.235]

The preposition *ki* has the same use, see (272) on p. 214. This shows that the two particles *ki*, though etymologically distinct, are closely related.

In fact, there is not an absolute distinction between the senses 'when' and 'until'. Whether *ki* is translated as one or the other, mainly depends on whether it is connected to the preceding clause ('X until Y') or to the following clause ('when Y, then Z'). When connected to both, the *ki*-clause marks a boundary point or "hinge" between two events:

(199) *ꞌI* at *roto* inside *e* ipfv *hāpaꞌo* care\_for *era* dist *ki* to/when *takataka* gather:red *tahi* all *te* art *tarake* corn *he* ntr *toꞌo* take *mai* hither *he huhu.*

ntr strip

'Inside they stored (the corn) until all the corn was gathered, (then) they would take it and strip it.' [R250.068]

(200) *He* ntr *uru* enter *ki* to *raro* below *i* at *te* art *roꞌi* bed *he* ntr *piko,* hide *ki* to/when *roa* long *te* art *hora* time *he* ntr *eꞌa* go\_out *he* ntr *tere* run *mai.* hither

'He would go under the bed and hide, when/until a long time (had passed), then he would come out and run away.' [R250.185]

<sup>24</sup> In other contexts, 'until' is more commonly expressed by *ka V rō*, and/or using *ꞌātā* (§11.6.2.5).

#### 11 Combining clauses

### **11.5.4** *ꞌO* **'lest'**

The preverbal marker *ꞌo* <sup>25</sup> indicates a consequence which is to be avoided. It can be translated as 'lest' or 'so that … not'.

*ꞌO*-marked clauses usually occur after the main clause and are always verb-initial. The subject is expressed in the same way as in main clauses: unmarked as in (201), or with the agent marker *e* as in (202).

(201) *He* ntr *oho* go *a* prop *Eva* Eva *he* ntr *piko* hide *ꞌo* lest *kī* say *rō* emph *a* prop *koro* Dad *mo* for *taꞌe* conneg *oho* go *ki* to *hiva.* mainland 'Eva went and hid lest Dad would tell her not to go to the mainland.' [R210.026]

(202) *He* ntr *tētere* pl:run *he* ntr *pipiko* pl:hide *tahi* all *ꞌo* lest *varaꞌa* catch *rō* emph *e* ag *te* art *Miru* Miru *i* acc *a* prop *rāua* 3pl *mo* for *tiaŋi.* kill

'All of them fled and hid, lest the Miru would catch them to kill them.' [R304.039]


'Kainga waited for Vaha, so Vaha wouldn't climb on the islet.' [Mtx-3-01.124]

In modern Rapa Nui, a verb marked with *ꞌo* is usually followed by the asseverative particle *rō* (§7.4.2), as illustrated in (201–202) above.

Occasionally *ꞌo* is found in complement clauses expressing a negative complement: *riꞌariꞌa ꞌo* 'to fear lest', *haŋa ꞌo* 'to want that not…' (see (75–76) in §11.3.5).

<sup>25</sup> The origin of *ꞌo* is unclear. It may be a reflex of PPN *\* ꞌaua* 'negative imperative', which occurs throughout Polynesia (Tongic, Samoic-Outlier and EP). Cf. also Footnote 16 on p. 502 on the origin of the negator *(e) ko*.

Another possible cognate is Tahitian *ꞌo*, which introduces clauses after "des verbes exprimant la crainte, la méfiance, et parfois l'eventualité" (verbs expressing fear, mistrust, and sometimes contingency), and which is followed by a nominalised verb (Académie Tahitienne 1986: 197). However, given the fact that Rapa Nui *ꞌo* occurs in old texts already, it is relatively unlikely that it is a borrowing from Tahitian.

11.5 Subordinating markers

### **11.5.5** *Mai* **'before; while'**

*Mai*, which is common as a preposition 'from' (§4.7.5) and as a directional 'movement towards deictic centre' (§7.5), also occurs occasionally as a preverbal marker. It indicates an event prior to the event in the main clause: 'before'.

(205) *He* ntr *tunu* cook *atu* away *au* 1sg *i* acc *to* art:of *tāua* 1du.incl *kai* food *mai* from *pō.* night 'I will cook our food, before it gets dark.' [R229.140]

*Mai* is often reinforced by the constituent negator *taꞌe*, which in this construction does not invert the polarity of the clause.

(206) *¡Ka* imp *hōrou* hurry *mai,* hither *mai* from *taꞌe* conneg *taŋi* cry *te* art *oe!* bell 'Hurry up, before the bell strikes!' [R334.077]

As these examples show, the event in the *mai*-clause indicates the end point of a time frame, which limits the time available to accomplish the action in the main clause. Event A should be done before (*mai*) event B happens.<sup>26</sup>

The event in the *mai*-clause may also be something which is to be avoided altogether: A should be done before B happens, so that B will not happen at all.


Occasionally, the *mai*-clause marks not the boundary of a time frame, but the time frame as such during which the action in the main clause is to be performed: 'while, as long as'. In this case, the verb is followed by the continuity marker *ꞌā/ꞌana* (§7.2.5.5):

(209) *¿ꞌO* because\_of *te* art *aha* what *koe* 2sg *i* pfv *taꞌe* conneg *hāꞌaki* inform *mai* hither *ai* pvp *mai* from *noho* stay *ꞌana* cont *ꞌi* at *Hiva,* Hiva *ꞌi* at *te* art *kāiŋa?* homeland 'Why didn't you tell me when we still lived in Hiva, in the homeland?' [Ley-2-07.028]

<sup>26</sup> Interestingly, in Hawaiian *mai* marks events to be avoided; it marks both negative imperatives and events (always unpleasant ones) which almost happen, but not quite: *Mai hā ꞌule ke keike* 'The child almost fell' (Elbert & Pukui 1979: 61–63). This is somewhat similar to temporal *mai* in Rapa Nui, though the latter is limited to subordinate clauses.

#### 11 Combining clauses

(210) *ꞌO* because\_of *ira* ana *ka* imp *hāꞌere* walk:pl *ꞌi* at *roto* inside *i* at *te* art *māꞌeha,* light *mai* from *ai* exist *atu* away *ꞌana* cont *te* art *mōrī.* light

'Therefore walk in the light, while there is still light.' [John 12:35]

### **11.5.6 Summary**

In the preceding sections, five preverbal markers have been discussed which introduce subordinate clauses; two of these also introduce certain types of main clauses. Table 11.3 summarises the different functions of these markers.


Table 11.3: Functions of preverbal markers

The sections above also show, that case marking in subordinate clauses follows the same rules as in main clauses: the S/A argument is marked with Ø or *e*, the O argument with *i* or Ø, depending on the factors described in §8.3–8.4. The only exception is *mo*, where both arguments are often marked as possessive.

11.6 Adverbial clauses

### **11.6 Adverbial clauses**

### **11.6.1 Adverbial clause strategies**

Adverbial clauses provide an adverbial modification of the main clause. They can be constructed in various ways:


Type 3 clauses are subordinate, even though they lack a conjunction or subordinating marker; this is indicated by the fact that they are negated with the constituent negator *taꞌe* (§10.5.6.5), not by a main clause negator. Here is an example of a negated temporal clause. Cf. also (258) on p. 567 (a reason clause marked with *he*).

(211) [*I* pfv *taꞌe* conneg *kore* lack *era* dist *tuꞌu* poss.2sg.o *tokerau* wind *era* ] dist *he* ntr *manaꞌu* think *mo* for *haka* caus *tītika* straight *i* acc *te* art *vaka* boat *ki* to *Tahiti.* Tahiti 'When the wind did not die down, they decided to steer the boat to Tahiti.' [R303.064]

In the following subsections, adverbial clauses are discussed, grouped by function: time (§11.6.2), purpose (§11.6.3), reason/result (§11.6.4), condition (§11.6.6), concession (§11.6.7) and circumstance (§11.6.8). This is followed by an overview (§11.6.9) summarising the different strategies used.

### **11.6.2 Time**

A temporal clause is a subordinate clause which provides a temporal framework for the event in the main clause. Rapa Nui has a variety of temporal clause constructions. Some of these involve a conjunction or a nominal construction; in others, the temporal relation is expressed by an aspectual marker.

#### 11 Combining clauses

#### **11.6.2.1 Cohesive clauses**

In Rapa Nui discourse – especially in narrative – it is common to find an unmarked subordinate clause at the beginning of a sentence, which provides a temporal framework for the main clause. Weber (2003b: 116) labels these *cohesive*: they connect the events to the preceding context and provide a setting for the events that follow. Two examples:


Cohesive clauses are characterised by the following features:


As the examples above show, cohesive clauses marked with perfective *i* express an event anterior to the event in the main clause (§7.2.4.2), which provides the setting for the event in the main clause.

Cohesive clauses marked with imperfective *e* indicate events simultaneous to the event in the main clause. They may be continuous as in (214) or habitual as in (215):


<sup>27</sup> In a representative corpus containing 304 *i*-marked cohesive clauses, 281 (92.4%) have *era*; *ai* occurs in 13 clauses (4.3%), while the remaining clauses have *nei* (7x), *ena* (1x) or no PVD at all (2x).

11.6 Adverbial clauses

The contiguity marker *ka* in cohesive clauses expresses temporal contiguity: the event in the subordinate clause marks the starting point of the event in the main clause.


Perfect aspect *ko V ꞌā* in cohesive clauses (as in main clauses) expresses a state resulting from a process. In cohesive clauses, *ko V ꞌā* only occurs with stative verbs.


Concerning the function of cohesive clauses in discourse: in many cases the preposed clause expresses an event which is predictable from the situation or from the preceding events. The event is just to be expected, and therefore it is backgrounded to a subordinate clause. In the following example, the person in question is on his way to Hanga Oteo. Puna Marengo is a place that lies on the way to Hanga Oteo, so it is only natural that he passes it on the way.

(220) *He* ntr *eꞌa* go\_out *he* ntr *oho* go *ki* to *Haŋa* Hanga *ꞌŌteo.* Oteo *I* pfv *haka* caus *noi* incline *atu* away *era* dist *ꞌi* at *ruŋa* above *o* of *te* art *nihinihi* curve:red *era* dist *o* of *Puna* Puna *Māreŋo,* Marengo *he* ntr *uꞌi* look *atu* away *ko* prom *te* art *ꞌau…* smoke 'He went out to Hanga Oteo. When he had come down the slope of Puna Marengo, he saw smoke…' [R313.091]

The preposed clause is not always closely connected to the preceding context, however. It may also have a transitional function, marking the start of a new scene or episode in the story. Such transitional clauses may express a lapse of time between the previous and the next event, or indicate the point in time at which the next events take place:

(221) *I* pfv *hinihini* delay:red *era* dist *he* ntr *oho* go *mai* hither *he* ntr *haꞌi* embrace *i* acc *tū* dem *poki* child *era* dist *ꞌāꞌana.* poss.3sg.a 'After that, he went to embrace his child.' [R210.068]

<sup>28</sup> *ꞌŌꞌotu* is a stative verb meaning 'to be cooked, done', not an active verb 'to cook'.

#### 11 Combining clauses

(222) *I* pfv *tuꞌu* arrive *nei* prox *ki* to *te* art *mahana* day *e* num *tahi* one *he* ntr *eꞌa* go\_out *hakaꞌou* again *te* art *taŋata* man *nei…* prox

'When a certain day came, this man went out again…' [R310.025]

#### **11.6.2.2 Other unmarked temporal clauses**

Apart from cohesive clauses, there are other temporal clauses without a conjunction or subordinator. The only way in which these clauses are marked, is by an aspectual which is different from the aspectual in the main clause. They may be marked with *i*, *e* or *ka*.

**Perfective** *i* In §7.2.4.2 on perfective *i*, it was shown that *i*-marked clauses may express a restatement, conclusion or clarification of the preceding clause. Subordinate *i*-marked clauses are somewhat similar in function; they express an event which is simultaneous to the event expressed in the preceding clause.


**Imperfective** *e* Temporal clauses may also be marked with imperfective *e*. These clauses express a continuous event simultaneous to the one in the main clause. As discussed in §7.2.5.4, *e-*marked verbs in main clauses are followed either by a postverbal demonstrative (PVD) or the continuity marker *ꞌā/ꞌana*. The same is true in temporal clauses: the verb is either followed by a PVD as in (225–226), or by *ꞌā/ꞌana* as in (227– 228).


11.6 Adverbial clauses


Though all these clauses are similar in function, there is a difference between clauses marked with *e V PVD* and the ones marked with *e V ꞌā*. The constructions with a PVD can be characterised as true temporal clauses, indicating an event which takes place at the same time as the main event. The clauses with *ꞌā* are more like circumstantial or manner clauses, further defining the nature of the event in the main clause or the manner in which it takes place. They have less the character of an independent event and can often be translated with a participle.

There are two indications for the more participial character of the *ꞌā* constructions:


**Contiguity marker** *ka* Subordinate clauses marked with the contiguity marker *ka* indicate an event which is simultaneous with the event expressed in the main clause:


'When we go down, father Vaha will come here.' [R229.187]

As these examples show, the subordinate clause may precede or follow the main clause. As in (230), the verb is often followed by a postverbal demonstrative.

#### 11 Combining clauses

#### **11.6.2.3 Development of** *hora* **'time' into a pseudo-conjunction**

Temporal adjuncts can be expressed by a temporal noun preceded by a preposition; the most general temporal noun is *hora* 'time'. The adjunct can be further specified by a modifier, e.g. a genitive as in (231) or a relative clause as in (232):


go to art place work

'At the time when her husband was still alive, she was the one who would go to work…' [R349.005]

Now as discussed in §5.3.2.3, the article can be omitted before clause-initial nouns followed by a demonstrative like *era*. At the same time, the preposition *ꞌi* can be omitted as well. This results in constructions like the following:


In the constructions above, *hora ena/era* resembles a temporal conjunction; semantic bleaching is taking place, where *hora ena/era* comes to mean little more than 'when'. Notice however, that the construction is syntactically still a nominal phrase with relative clause: as (234) shows, the aspectual can be omitted, something which is only possible in relative clauses (§11.4.5). (Also, the verb *takeꞌa* has been raised from the relative clause.)

#### **11.6.2.4 Anteriority: 'before'**

Rapa Nui has a variety of devices to express that the event in the subordinate clause takes place prior to the event in the main clause. One of these is preverbal *mai*, discussed in §11.5.5. The following strategies are also used:

11.6 Adverbial clauses

*ꞌI raꞌe Raꞌe* is a locational meaning 'first' (§3.6.4.1). *ꞌI raꞌe ki*, followed by a nominalised verb, means 'before':

(235) *Paurō* every *te* art *mahana* day *e* ipfv *ꞌara* wake\_up *era* dist *ꞌi* at *te* art *pō* night *era* dist *ꞌā,* ident *ꞌi* at *raꞌe* first *ki* to *te* art *eꞌa* go\_out *o* of *te* art *raꞌā.* sun

'Every day he woke up early in the morning, before the sun came up.' [R448.003]

*Ante Ante* (< Sp. *antes*) is used as an adverb meaning 'before, earlier, previously'. It is also used as a conjunction, followed by *ki* + nominalised verb:

(236) *Pero* but *ante* before *ki* to *te* art *uru,* enter *he* ntr *oho* go *tahi* all *te* art *ŋā* pl *poki* child *he* ntr *fira* line *raꞌe.* first 'But before going in, the children first go and stand in line.' [R151.012]

*ꞌŌ ira ꞌō ira* 'before'<sup>29</sup> consists of the otherwise unknown particle *ꞌō*, followed by the pro-form *ira* (§4.6.5.2). It is always followed by a *ka*-marked verb. As (237) shows, the subject after *ꞌō ira* is usually preverbal.

(237) *Te* art *rāua* 3pl *henua* land *raꞌe* first *i* pfv *noho* stay *ko* prom *Perú,* Peru *ꞌō\_ira* before *te* art *Inca* Inca *ka* cntg *tuꞌu.* arrive 'The first land where they lived was Peru, before the Incas arrived.' [R376.011]

*Hia* The postverbal marker *hia*, combined with a negation, means 'not yet'; in a multiclause construction it indicates that an event has not happened before another occurs (§10.5.8).

#### **11.6.2.5 Temporal limit: 'until'**

'Until' is often expressed by the aspectual *ka* (§7.2.6) in combination with the emphatic marker *rō* (§7.4.2). This is in line with the function of *ka* as a contiguity marker: the event or state expressed in the *ka-*clause marks the temporal boundary of another event, often indicating the natural or expected outcome of an action performed to completion. These *ka-*clauses usually occur sentence-finally.

(238) *He* ntr *kai* eat *a* prop *Te* Te *Manu* Manu *ka* cntg *mākona* satiated *rō.* emph 'Te Manu ate until he was satiated.' [R245.067]

<sup>29</sup> Not to be confused with *ꞌo ira* 'therefore' (§11.6.4).

#### 11 Combining clauses

(239) *I* pfv *noho* stay *ai* pvp *a* prop *Te* Te *Manu* Manu *ꞌi* at *muri* near *tū* dem *pāpārūꞌau* grandfather *era* dist *ka* cntg *rovaꞌa* obtain *rō* emph *hoꞌe* one *ꞌahuru* ten *tūmaꞌa* more\_or\_less *matahiti.* year 'Te Manu stayed with his grandfather until he was about ten years old.' [R245.159]

In the examples above, the subject of the main clause reaches a certain state or end point; for example, in (239), Te Manu reaches a state of satiation after having eaten. The stative verb in the *ka V rō* clause may also specify the action of the main clause, which is performed – or is to be performed – to a certain extent or in a certain way. (Cf. the use of *ka* before numerals to mark an extent, §4.3.2.2).


A second way to express 'until' is by means of *ꞌātā* (< Sp. *hasta*). *ꞌĀtā* is used in nominal constructions before the preposition *ki* (see (273) on p. 214), but also in verbal constructions, followed by *ka V rō*. As (243) shows, *ꞌātā* may be shortened to *ꞌā*:


In the third place: less commonly, the conjunction *ꞌahara* is used, followed by *ka*:

(244) *He* ntr *noho* stay *rō* emph *ꞌai* subs *tāua* 1du.incl *ꞌahara* until *ka* cntg *haka* caus *hoki* return *rō* emph *koe* 2sg *i* acc *a* prop *au* 1sg *ki* to *muꞌa* before *ki* to *tōꞌoku* poss.1sg.o *nuꞌu.* people 'We will stay, until you make me return to my people.' [Fel-1978.115]

Finally, 'until' may be expressed by the subordinator *ki*, especially after verbs like *tiaki* 'wait' (§11.5.3).

11.6 Adverbial clauses

### **11.6.3 Purpose: bare purpose clauses**

Purpose clauses are often marked with preverbal *mo* (§11.5.1.1) or *ki* (§11.5.3). Purpose may also be expressed by a bare verb, i.e. a verb without aspect marker.<sup>30</sup> This verb is always initial in the clause. Bare purpose clauses are found especially after motion verbs. A few examples:


More commonly, the purpose of an action is expressed by a noun phrase introduced by the preposition *ki*, followed by a bare verb. Here are a few examples:


'We all went down to watch a movie at school.' [R410.010]

In these examples, the main verb is a motion verb; the *ki*-marked noun phrase is the Goal of movement. This noun phrase is followed by a bare verb, of which the preceding noun is the Patient.<sup>31</sup>

The noun in this construction is not an incorporated object of the following verb: it is the head of a regular noun phrase, marked with the article *te* and preceded by a preposition. A somewhat more plausible analysis would be to consider the verb as incorporated

<sup>30</sup> Clauses with a bare verb cannot be analysed as juxtaposed main clauses, as main clause verbs always have an aspect marker, except occasionally when the verb is followed by certain postverbal particles (§7.2.2).

<sup>31</sup> Clark (1983b: 424) points out that the same construction occurs in Marquesan and Mangarevan. Different from what Clark suggests, in Rapa Nui this construction is not limited to generic objects, as (248) shows.

#### 11 Combining clauses

into the noun; however, the directional *mai* in (248) shows that the verb is the head of a true verb phrase. It is best to analyse these constructions simply as a combination of a noun phrase and a bare purpose clause, rather than assuming that the noun phrase + verb are a single constituent. An additional reason to do so is that this construction is not an isolated phenomenon, but an instance (admittedly, the most common instance) of a group of constructions in which a locative noun phrase and a purpose clause occur together. Related constructions include:

	- a Source noun phrase (with preposition *mai* 'from') followed by a bare verb:
	- a *ki*-marked Goal noun phrase followed by a *mo*-marked purpose clause:
	- a *ki*-marked Goal noun phrase, with the associated action left implicit:

(253) *—¿Ki* to *hē* cq *a* prop *kuā* coll *ꞌOrohe* Orohe *i* pfv *iri* ascend *ai* pfv *ꞌi* at *ruŋa* above *i* at *te* art *vaka?* boat *—Ki* to *te* art *rāua* 3pl *ika* fish *ꞌi* at *ruŋa* above *i* at *te* art *toka.* rock '—Where did Orohe and the others go by boat? —To their fish (i.e. to catch fish) on the rocks.' [R154.038]


11.6 Adverbial clauses

These examples suggest that *ki te N V* in (247–249) should not be analysed as a special construction involving a single NP+V constituent. Rather, it is a combination of two constituents, a nominal Goal phrase followed by a bare purpose clause, either of which may also occur on its own.

### **11.6.4 Reason**

Reason clauses can be constructed in several ways. In the first place, reason is often expressed by nominalised clauses marked with the prepositions *ꞌi* and *ꞌo* (§4.7.3).

Secondly, in modern Rapa Nui, the phrase *ꞌi te meꞌe (era)* (lit. 'in the thing' or 'because of the thing') is used as a conjunction introducing a reason clause. As the examples show, the reason clause either precedes or follows the main clause.


'OK, because I was little, my father didn't allow me to mount a horse.' [R101.004]

Thirdly, the reason clause may also be a subordinate clause marked with the aspectual *he*. That this is a subordinate clause, is shown by the fact that it is negated with the constituent negator *taꞌe* (§10.5.6); main clauses would have a different negator.


'These people fled because their family did not allow them to marry.' [R303.144]

In these constructions, *he* can also be considered as a nominal predicate marker followed by a nominalised verb (hence the double gloss in the examples above). One reason to do so, is that other nominal constructions are also used to express reasons: in (259) a nominalised verb preceded by a possessive pronoun, in (260) a subordinate existential construction (an existential main clause would be *ꞌIna he meꞌe mo kai*):

#### 11 Combining clauses


### **11.6.5 Result**

Results may be marked by the adverbial connector *ꞌo ira* 'because of that; therefore' (the reason preposition *ꞌo* followed by the pro-form *ira*). As (262) shows, it is possible to mark both the reason clause (in this case, a nominal construction) and the result clause.


As these examples show, the subject tends to be placed straight after *ꞌo ira*. This conforms to a general preference for preverbal subjects after initial oblique constituents (§8.6.1.1).

### **11.6.6 Condition**

Conditional clauses can be marked by one of the subordinators *mo* (§11.5.1.1) and *ana* (§11.5.2.2).

Condition is not always marked, however: clauses with a conditional sense may also occur without special marking. The verb is marked with one of the aspectuals *i*, *e* or *ka* and followed by a postverbal demonstrative. Two examples:

(263) *ꞌE* and *i* pfv *haŋa* want *era* dist *koe* 2sg *mo* for *rere* fly *ki* to *taꞌa* poss.2sg.a *kona* place *i* pfv *manaꞌu,* think *he* ntr *rere* fly *rō* emph *ꞌai* subs *koe….* 2sg 'And if you want to fly to the place you think of, you (can) fly…' [R378.006]

11.6 Adverbial clauses

(264) *Ka* cntg *hāŋai* feed *atu* away *ena* med *ki* to *a* prop *koe,* 2sg *he* ntr *mate* die *koe.* 2sg 'If (the two spirits) feed you, you will die.' [R310.061]

The contiguity marker *ka* is relatively common in clauses expressing a condition. It seems natural that a marker which indicates temporal contiguity (simultaneous or sequential events) also marks logical contiguity, i.e. contingency of one event on another. To mark irreal conditions, the conjunction *ꞌāhani* (var. *ꞌani, ꞌahari*) is used.<sup>32</sup>

(265) *ꞌĀhani* if\_only *ꞌō* really *au* 1sg *he* pred *ꞌono,* rich *ko* prf *hoꞌo* buy *mai* hither *ꞌā* cont *au* 1sg *i* acc *te* art *hare* house *e* num *tahi…*

```
one
```
'If I were rich, I would buy a house…' [R399.182]

(266) *ꞌĀhani* if\_only *ꞌō* really *tōꞌoku* poss.1sg.o *nua* Mum *era* dist *i* pfv *taꞌe* conneg *mate,* die *ꞌī* imm *au* 1sg *ꞌi* at *muri* near *i* at *a* prop *ia* 3sg *ꞌi* at *te* art *hora* time *nei.* prox

'If my mother hadn't died, I would be with her now.' [R245.007]

As these examples show, the subject after *ꞌāhani* is usually preverbal (§8.6.1.1).

### **11.6.7 Concession**

The aspectual marker *ka*, in combination with the directional *atu*, can be used in a concessive sense, indicating a circumstance which might be expected to prevent – but actually does not prevent – the event in the main clause.<sup>33</sup>

(267) *Ka* cntg *rahi* many *atu* away *tāꞌaku* poss.1sg.a *poki,* child *e* ipfv *hāpaꞌo* care\_for *nō* just *e* ag *au* 1sg *ꞌā.* ident 'Even if I have many children, I will care for them myself.' [R229.023]

As discussed in §7.2.6, *ka* expresses temporal contiguity; the concessive sense follows in a way from this basic sense. By explicitly juxtaposing two events or situations which are temporally contiguous or simultaneous, the contrast between the two is highlighted.<sup>34</sup>

<sup>32</sup> ꞌĀhani < Tah. *ꞌahani*, a var. of *ꞌahari/ꞌahiri/ꞌahini*, which is likewise a conjunction introducing an irreal condition clause.

<sup>33</sup> This does not mean that all *ka V atu* constructions have a concessive sense, see e.g. example (264) above.

<sup>34</sup> The same use can be observed for constructions expressing simultaneity in other languages. English 'while' can be used in the sense 'even though' ('While he had a good job, he did not earn enough to support his expensive tastes.'). The French *gérondif*, preceded by 'tout en', has a concessive sense ('La police a des soupçons tout en ignorant l'identité du coupable' = 'The police has suspicions, but does not know the identity of the culprit.').

#### 11 Combining clauses

The *ka V atu* construction with concessive sense is especially common with the existential verb *ai*, in the expressions *ka ai atu* 'even' and *ka ai atu pē ira/nei* 'even though; even so':


A second way to express concession is by means of the preposition *nōatu*, <sup>35</sup> followed by a nominalised verb:

(270) *Nōatu* no\_matter *te* art *paŋahaꞌa,* heavy *te* art *mahana* day *te* art *mahana* day *e* ipfv *hāpī* teach *ena* med *ꞌi* at *ira.* ana 'Even though it's heavy, they teach there day after day.' [R537.023]

Finally, concession is expressed by the adverbial expression *te meꞌe nō* 'however, even so', which functions as a coordinating conjunction (§5.8.2.4):

(271) *ꞌApa* half *te* art *toe* remain *a* prop *au* 1sg *he* ntr *mate;* die *te* art *meꞌe* thing *nō,* just *ꞌī* imm *a* prop *au* 1sg *e* ipfv *ora* live *nō* just *ꞌā.* cont

'I almost died; even so, I am alive.' [R437.050]

### **11.6.8 Circumstance**

Circumstantial clauses may be expressed by *koia ko* 'with' preceding the verb (§8.10.4.2):

(272) *He* ntr *hoki* return *mai* hither *a* prop *Kāiŋa* Kainga *koia* com *ko* prom *taŋi.* cry 'Kainga returned crying.' [R243.173]

Alternatively, *mā* 'and, with' may be used, followed by a nominalised verb.<sup>36</sup> As (274) shows, *mā te* may be assimilated to *mata*.

<sup>35</sup> *Nōatu* < *nō* 'just' + *atu* 'away', but probably borrowed as a whole from Tahitian *noātu* (Académie Tahitienne 1986: 310).

<sup>36</sup> *Mā* has a limited distribution in Rapa Nui: it is only used in the construction under discussion and in numerals. Both uses are also found in (and were probably borrowed from) Tahitian (see Footnote 9 on p. 148).

11.6 Adverbial clauses


When circumstances are states rather than events, they tend to be expressed in a clause in the perfect aspect (*ko V ꞌā*), without a special marker.


I have not found this construction in older texts, so it may be a modern development. Perfect aspect clauses expressing circumstances are especially common in the construction *ko V ꞌā e V era*. In this construction, the second clause is marked with *e V era* and expresses an action, while the preceding *ko V ꞌā* clause expresses a quality (e.g. a feeling or attitude) possessed by the subject performing the action. Even though *e V era* in general expresses durative actions, in this construction it is not necessarily durative.


Notice that *e V era* is obligatory when the circumstantial *ko V ꞌā* clause comes first; when the circumstantial clause follows the main clause, the main clause may be *he*marked, as in (275–276) above.

### **11.6.9 Summary**

Events which modify the event in the main clause, can be expressed in several ways. Certain interclausal relationships are expressed using a subordinating marker or conjunction. In other cases no special marker is used; even so, the modifying clause is subordinate, as is shown by the fact that these clauses are negated by the subordinate negator *taꞌe* rather than a main clause negator. The various strategies are summarised in Table 11.4.

#### 11 Combining clauses


#### Table 11.4: Overview of adverbial clauses

### **11.7 Conclusions**

This chapter has explored the ways in which clauses are combined. A common way to combine clauses is simple juxtaposition. In fact, older Rapa Nui did not have any coordinating conjunction. In modern Rapa Nui *ꞌe* 'and' is used, but juxtaposition is still the default strategy for coordinating clauses. Juxtaposition is not only used to express sequential events, but also to express semantic complements of the verbs *haꞌamata* 'begin' and *hōrou* 'hurry'.

Rapa Nui has various strategies to express the argument of a matrix verb. Only some of these involve a proper complement clause, i.e. a clause which is syntactically dependent on the main verb; they may involve the subordinating marker *mo* 'for, in order to', or a nominalised complement. Other verbs are followed by a juxtaposed clause or an independent clause.

11.7 Conclusions

The subordinator *mo* marks both complement clauses and adverbial clauses; interestingly, it marks both purpose and condition. The marker *ana* has an even wider range of functions, all of which can be characterised as irrealis: an *ana-*marked clause refrains from claiming the truth of the proposition expressed. *Ana*-marked clauses express intentions, potential events and obligations, but also general truths. In subordinate clauses, *ana* marks conditional clauses and dependent questions.

Relative clauses in Rapa Nui are not marked by a conjunction or preverbal marker, but they have various distinctive properties: they are invariably verb-initial and the choice of aspectuals is limited. A peculiar feature is, that the aspect marker may be left out (in most other clause types, unmarked verbs are rare or nonexistent). In these "bare relative clauses", the verb is often raised to a position immediately after the head noun, before any postnominal markers.

A wide range of constituents can be relativised; most of these are not expressed in the relative clause, others are expressed as a pronoun. The distribution of these two constructions does not entirely conform to the noun phrase hierarchy proposed by Keenan & Comrie (1977): while subjects, objects and adjuncts are left unexpressed, oblique arguments (which are higher in the hierarchy than adjuncts) are expressed pronominally, just like constituents low in the hierarchy like possessors.

There is a tendency to express the entity which is subject of the relative clause as a possessor before or after the head noun: 'your thing [did yesterday]' = 'the thing you did yesterday'. Syntactically there is nothing special about these constructions: the possessor is no different from other possessors in the noun phrase; the relative clause is no different from other relative clauses, apart from the fact that the subject is not expressed.

## **Appendix A: Interlinear texts**

Below are three glossed and interlinearised texts, all of which are part of the PLRN text corpus (§1.6.2). The first text is a children's story, written during a writer's workshop in 1984. The second text is a dramatic retelling of a traditional story by Luis Avaka Paoa ('Papa Kiko'), a renowned storyteller; a very short version of the same story was published by Blixen (1974). Number three is a description of a fishing trip, composed as part of a schoolbook containing stories about traditional activities on Rapa Nui.

### *Te tātane taŋata* **– The devilman (R215)**

by Virginia Haoa Cardinali (Haoa Cardinali 1984; Weber, Weber & team 1990a: Vol. 3:118– 119)

**01** *Ko* prf *ahiahi* evening *pō* night *ꞌā.* cont **02** *Te* art *ŋā* pl *poki* child *nei* prox *e* num *rua:* two *e* num *tahi* one *ko* prom *Kihi* Kihi *te* art *ꞌīŋoa,* name *e* num *rima* five *ōꞌona* poss.3sg.o *matahiti,* year *poki* child *teatea,* white *ritorito,* clear *he* ntr *tau* pretty *nō;* just *te* art *rua* two *poki* child *ko* prom *ꞌAtera* Atera *te* art *ꞌīŋoa,* name *e* num *hitu* seven *matahiti,* year *meꞌe* thing *rakerake* bad:red *a* prop *vērā.* poor\_thing **03** *Mahana* day *tāpati* Sunday *ꞌi* at *te* art *pō,* night *he* ntr *haka* caus *rivariva* good:red *e* ag *te* art *rāua* 3pl *māmā* mother *mo* for *maꞌu* carry *i* acc *te* art *kai* food *ki* to *te* art *rāua* 3pl *koro.* Dad **04** *Te* art *aŋa* work *iŋa* nmlz *ꞌi* at *te* art *kona* place *motore* engine *mo* for *haka* caus *pura* shine *i* acc *te* art *mōrī* light *paurō* every *te* art *mahana* day *ꞌi* at *te* art *pō.* night **05** *Ko* prf *haꞌaꞌī* fill *ꞌā* cont *tū* dem *kai* food *era* dist *e* ag *nua* Mum *ꞌi* at *roto* inside *i* at *te* art *pani* pan *e* num *tahi,* one *ꞌai* subs *ka* cntg *viri* wrap *rō* emph *hai* with *pānio* towel *teatea.* white

**01** It was evening. **02** There were two children: one was called Kihi, she was five years old, a fair child, light-skinned, just pretty; the other child was called Atera, seven years old, the poor one was ugly. **03** On Sunday night, their mother made preparations to take food to their father. **04** He worked at the electrical power plant every day at night. **05** Mother had put the food in a pan and wrapped it in a white towel.

**06** *He* ntr *hahari* comb *i* acc *te* art *pūꞌoko* head *o* of *tū* dem *ŋā* pl *poki* child *era,* dist *he* ntr *haka* caus *uru* dress *i* acc *te* art *paratoa,* jacket *ꞌai* subs *ka* cntg *vaꞌai* give *rō* emph *tū* dem *pūꞌahu* bundle *kai* food *era* dist *ki* to *te* art *poki* child *ꞌatariki* firstborn *era.* dist **07** *He* ntr *kī* say

*ia* then *e* ag *nua:* Mum **08** *—Ka* imp *maꞌu* carry *hiohio* strong *te* art *kai* food *ena* med *mā* for.a *koro.* Dad **09** *E* exh *maꞌu* carry *hiohio* strong *ꞌi* at *roto* inside *i* at *tuꞌu* poss.2sg.o *rima;* hand *taꞌe* conneg *mo* for *haka* caus *pakō* loose *tāꞌue.* by\_chance **10** *E* num *tahi* one *rima* hand *ena* med *ōꞌou* poss.2sg.o *ka* cntg *maꞌu* carry *ena* med *i* acc *te* art *kai,* food *e* num *tahi* one *i* acc *te* art *rima* hand *o* of *tuꞌu* poss.2sg.o *taina;* sibling *ararua* the\_two *nō* just *ꞌina* neg *ko* neg.ipfv *haka* caus *pakoꞌo* loose *ki* to *te* art *hora* time *hopeꞌa.* last

**06** She combed the children's hair, put on their jacket and gave the bundle of food to the oldest one. **07** Then mother said, "Hold the food for Dad firmly. **09** Hold it tight in your hands; you must not let go of it. **10** With one hand carry the food, with the other hold on to you sister's hand; don't ever let go of each other."

**11** *He* ntr *eꞌa* go\_out *ia* then *tū* dem *ŋā* art *poki* child *era* dist *a* by *te* art *vāeŋa* middle *o* of *te* art *ara* road *he* ntr *haꞌere* walk *he* ntr *iri.* ascend **12** *ꞌIna* neg *e* num *tahi* one *vānaŋa* word *rere* fly *ararua;* the\_two *ko* prf *momou* pl:quiet *ꞌā* cont *ꞌi* at *te* art *riꞌariꞌa* fear *ꞌi* at *te* art *pōhāhā.* dark **13** *ꞌE* and *ꞌi* at *te* art *riꞌariꞌa* fear *ꞌi* at *tū* dem *pōhāhā* dark *era,* dist *he* ntr *paꞌahia* sweat *te* art *rima* hand *ararua.* the\_two **14** *He* ntr *haꞌamata* begin *te* art *paŋahaꞌa* heavy *o* of *te* art *rāua* 3pl *vaꞌe,* foot *pē* like *ira* ana *ꞌā* ident *tū* dem *kai* food *era* dist *mā* for.a *koro.* Dad

**11** The children went out by the middle of the road and walked up. **12** The two didn't say anything; they were silent, because they were afraid in the dark. **13** And because they were afraid in the dark, both had sweaty hands. **14** Their feet started to feel heavy, and the food for Dad felt heavy as well.

**15** *Ko* prf *hinihini* delay:red *ꞌā* cont *te* art *hora* time *e* ipfv *iri* ascend *era,* dist *he* ntr *ꞌui* ask *ia* then *e* ag *ꞌAtera* Atera *ki* to *a* prop *Kihi:* Kihi **16** *—¿E* ipfv *ko* neg.ipfv *haŋa* want *ꞌō* really *rō* emph *koe* 2sg *mo* for *maꞌu* carry *i* acc *te* art *kai* food *nei* prox *mā* for.a *koro?* Dad **17** *Ka* imp *maꞌu* carry *ꞌitiꞌiti* little *koe* 2sg *mo* for *haka* caus *ora* rest *ꞌitiꞌiti* little *o* of *tōꞌoku* poss.1sg.o *rima.* hand **18** *Terā* then *ka* cntg *kī* say *e* ag *Kihi:* Kihi **19** *––¡Ko* prf *aha* what *ꞌā* cont *i* at *a* prop *koe!* 2sg **20** *¿Ko* prf *haŋa* want *ꞌana* cont *ꞌō* really *pēaha* perhaps *koe* 2sg *mo* for *pakoꞌo* loose *tōꞌoku* poss.1sg.o *rima?* hand **21** *He* ntr *momou* pl:quiet *hakaꞌou* again *ararua,* the\_two *ꞌai* subs *ka* cntg *iri* ascend *nō.* just

**15** When they had walked for a while, Atera asked Kihi, **16** "Don't you want to carry the food for Dad? **17** Carry it for a little while, so my hand can rest a little." **18** Kihi

said **19** "What are you thinking! **20** Do you really want to let go of my hand?" **21** The two were silent again, while they kept going up.

**22** *I* pfv *roaroa* long *hakaꞌou* again *era* dist *te* art *hora,* time *he* ntr *kī* say *hakaꞌou* again *e* ag *ꞌAtera:* Atera **23** *—¡Mo* if *taꞌe* conneg *haŋa* want *ōꞌou* poss.2sg.o *mo* for *maꞌu* carry *i* acc *te* art *kai* food *nei* prox *mā* for.a *koro,* Dad *he* ntr *haka* caus *pakō* loose *e* ag *au* 1sg *tuꞌu* poss.2sg.o *rima* hand *nei!* prox **24** *He* ntr *oho* go *ia* then *a* prop *Kihi* Kihi *mo* for *taŋi,* cry *ꞌai* subs *ka* cntg *kī* say *rō:* emph **25** *—¡E* voc *te* art *taina* sibling *riva* good *ōꞌoku* poss.1sg.o *ē,* voc *ꞌina* neg *koe* 2sg *ko* neg.ipfv *haka* caus *pakō* loose *i* acc *tōꞌoku* art *rima* hand *ꞌo* lest *riꞌariꞌa* fear *rō* emph *au* 1sg *ꞌi* at *te* art *tātane!* devil **26** *Ki* when *hāhine* near *nō* just *tāua* 1du.incl *mo* for *tuꞌu* arrive *ana* irr *maꞌu* carry *iho* just\_then *e* ag *au* 1sg *te* art *kai.* food

**22** After a long time, Atera said again, **23** "If you don't want to carry the food for Dad, I will let go of your hand!" **24** Kihi was about to cry and said, **25** "My dear sister, don't let go of my hand, or else I will be afraid of the devil! **26** When we are almost there, then I will carry the food."

**27** *He* ntr *kī* say *ia* then *e* ag *ꞌAtera:* Atera **28** *—¡Ko* prf *aha* what *ꞌā* cont *koe* 2sg *i* pfv *vānaŋa* speak *mai* hither *ai* pvp *i* acc *te* art *vānaŋa* word *o* of *te* art *tātane!* devil **29** *¿Hoki* pq *ko* prf *tikeꞌa* see *ꞌā* cont *e* ag *koe* 2sg *te* art *tātane* devil *raꞌe?* first **30** *He* ntr *pāhono* answer *mai* hither *ia* then *e* ag *Kihi:* Kihi **31** *—Te* art *parautiꞌa,* truth *kai* neg.pfv *tikeꞌa* see *ꞌā* cont *e* ag *au* 1sg *te* art *tātane* devil *raꞌe.* first **32** *O* of *tētahi* other *ŋā* pl *poki* child *ꞌō* really *tāꞌau* poss.2sg.a *i* pfv *ꞌaꞌamu* tell *mai* hither *pē* like *nei* prox *ē:* thus *e* ipfv *ai* exist *rō* emph *ꞌā* cont *te* art *tātane* devil *ꞌe* and *ꞌi* at *te* art *pō* night *e* ipfv *eꞌa* go\_out *e* ipfv *haꞌere* walk *nei.* prox

**27** Atera said, **28** "What are you talking about the devil! **29** Have you ever seen a devil?" **30** Kihi replied, **31** "The truth is, I have never seen a devil. **32** Other children have told that devils exist and that they go out and walk around at night."

**33** *Ka* cntg *topa* happen *tū* dem *vānaŋa* word *era* dist *ꞌa* of.a *Kihi,* Kihi *ꞌī* imm *rāua* 3pl *ka* cntg *uꞌi* look *atu* away *ena* med *e* ipfv *noho* sit *nō* just *mai* hither *ꞌā* cont *te* art *tātane* devil *e* num *tahi* one *ꞌi* at *ruŋa* above *i* at *te* art *ꞌāua.* fence **34** *A* by *ruŋa* above *i* pfv *peꞌe* jump\_up *ai* pvp *ararua* the\_two *ko* prf *kikikikiu* red:shriek *atu* away *ꞌana* cont *ia.* then **35** *E* ipfv *tuꞌu* arrive *nō* just *ꞌā* cont *te* art *reꞌo* voice *ki* to *te* art *rāua* 3pl *hare* house *era* dist *ꞌe* and *pē* like *ira* ana *ꞌā* ident *ki* to *te* art *kona* place *aŋa* work *era* dist *o* of *koro.* Dad **36** *He* ntr *momoko* pl:rush *mai* hither *te* art *taŋata* man *taꞌatoꞌa* all *mai* from *roto* inside *i* at *te* art *rāua* 3pl *hare.* house **37** *Pē* like *ira* ana *ꞌā* ident *a* prop *nua* Mum *ararua* the\_two *ko* prom *koro* Dad *ko* prf *tahuti* run *mai* hither *ꞌā* cont *i* pfv *ŋaroꞌa* perceive *era* dist *te* art *reꞌo* voice *o* of *tū* dem *ŋā* pl *rikiriki* small.pl *era* dist *o* of *rāua.* 3pl **38** *I* pfv *tuꞌu* arrive *era* dist *ki* to *tū* dem *kona* place *era* dist *o* of *tū* dem *ŋā* pl *poki* child *era* dist *ko* prom *te* art *kikikikiu* red:shriek *haŋa* nmlz *ꞌā,* ident *e* ipfv *puru* close *rō* emph *ꞌā* cont *te* art *ꞌāriŋa* face *ararua* the\_two *hai* ins *tū* dem *paratoa* jacket *era* dist *o* of *rāua.* 3pl

**33** Just when Kihi had said that, they saw a devil that was sitting on a fence. **35** They both jumped up and yelled. **35** The sound could be heard at their house and also at the place where Dad worked. **36** All the people rushed out from their houses. **37** Mum and Dad came running when they heard the voice of their little ones. **38** When they came to the place where their children were, the two were still yelling, covering their face with their jackets.

**39** *He* ntr *haka* caus *hāhine* near *atu* away *ia* then *a* prop *nua* Mum *ararua* the\_two *ko* prom *koro,* Dad *ꞌai* subs *ka* cntg *kī* say *rō:* emph **40** *—Ka* imp *momou,* pl:quiet *e* voc *koā* coll *vovo* dear\_girl *ē.* voc **41** *Taꞌe* conneg *he* pred *tātane* devil *te* art *meꞌe* thing *era* dist *o* of *ruŋa* above *i* at *te* art *ꞌāua.* fence **42** *He* pred *mautini* pumpkin *piro* rotten *ko* prf *kakaro* hollow\_out *ꞌā* cont *te* art *roto* inside *e* ag *Hao* Hao *Kūmā.* Kuma **43** *ꞌAi* subs *ka* cntg *aŋa* make *rō* emph *te* art *mata,* eye *te* art *haha,* mouth *ꞌai* subs *ka* cntg *puꞌa* cover *rō* emph *hai* with *paratoa* jacket *ꞌe* and *he* ntr *toꞌo* take *mai* hither *he* ntr *haka* caus *eke* go\_up *ki* to *ruŋa* above *o* of *te* art *ꞌāua* fence *mo* for *haka* caus *riꞌariꞌa* fear *o* of *te* art *huaꞌai haꞌere pō.*

family walk night

**39** Mum and Dad came near and said, **40** "Be quiet, dear girls. **41** That thing on the fence is not a devil. **42** It's a rotten pumpkin which Hao Kuma has hollowed out. **43** He made eyes and a mouth, then he covered it with a jacket and put it on top of the fence to scare the people who walk by at night."

### *Tikitiki ꞌa ꞌAtaraŋa* **– Tikitiki a Ataranga (R352)**

by Luis Avaka Paoa (Weber, Weber & team 1990a: Vol. 4:93–98)

**01** *Hora* time *nei* prox *he* ntr *ꞌaꞌamu* tell *atu* away *au* 1sg *i* acc *te* art *ꞌaꞌamu* story *era* dist *o* of *Tikitiki* Tikitiki *ꞌa* a *ꞌAtaraŋa.* Ataranga **02** *A* prop *Tikitiki* Tikitiki *ꞌa* a *ꞌAtaraŋa,* Ataranga *kai* neg.pfv *aŋiaŋi* certain *mai* hither *e* ag *au* 1sg *he* pred *aha* what *tōꞌona* poss.3sg.o *mata* tribe *he* pred *aha* what *tōꞌona* poss.3sg.o *hakaara,* descendance *he* pred *Tūpāhotu* Tupahotu *ꞌo* of *he* pred *Miru.* Miru **03** *ꞌIna* neg *kai* neg.pfv *aŋiaŋi* certain *rivariva* good:red *mai* hither *e* ag *au.* 1sg **04** *Te* art *meꞌe* thing *ōꞌona* poss.3sg.o *i* pfv *aŋiaŋi* certain *mai* hither *e* ag *au,* 1sg *he* pred *nua* Mum *ōꞌona* poss.3sg.o *te* art *ꞌīŋoa* name *ko* prom *Nuahine* Nuahine *ꞌa* a *Raŋi* Rangi *Kotekote.* Kotekote **05** *I* pfv *poreko* born *era* dist *te* art *poki* child *nei,* prox *he* ntr *hāŋai* raise *e* ag *te* art *rūꞌau* old\_woman *nei* prox *ararua* the\_two *ko* prom *tāꞌana* poss.3sg.a *kenu.* husband **06** *He* ntr *nuinui,* big:red *he* ntr *mate* die *te* art *koro.* Dad **07** *I* pfv *mate* die *era* dist *te* art *koro* Dad *o* of *te* art *poki* child *nei,* prox *he* ntr *toe* remain *he* pred *rūꞌau* old\_woman *nō* just *nei,* prox *he* ntr *noho* stay *he* ntr *hāŋai* raise *ē…* on\_and\_on **08** *he* ntr *nuinui.* big:red

**01** Now I'm going to tell the story of Tikitiki a Ataranga. **02** Concerning Tiki a Ataranga, I don't know what his tribe was or his descendance, whether Tupahotu of Miru. **03** I don't know it well. **04** The thing about him I do know, is that his mother's name was Nuahine a Rangi Kotekote. **05** When this child was born, this old woman brought it up with her husband. **06** When he was bigger, his father died. **07** When the child's father had died, just the old woman was left; she continued to raise the child **08** until he was grown up.

**09** *He* ntr *moe* lie\_down *tou* dem *poki* child *era* dist *ko* prom *Tikitiki* Tikitiki *ꞌa* a *ꞌAtaraŋa* Ataranga *ki* to *tāꞌana* poss.3sg.a *viꞌe* woman *Tūpāhotu.* Tupahotu **10** *He* ntr *poreko* born *e* num *toru* three *poki:* child *e* num *rua* two *poki* child *tane,* male *e* num *tahi* one *poki* child *vahine.* female **11** *He* ntr *hāŋai* raise *i* acc *tū* dem *ŋā* pl *poki* child *era* dist *he* ntr *nunui.* pl:big

**09** The boy Tikitiki a Ataranga married a Tupahotu wife. **10** Three children were born: two boys and one girl. **11** They raised the children and they grew up.

**12** *I* pfv *hāŋai* raise *era* dist *i* pfv *nunui* pl:big *era,* dist *te* art *aŋa* work *o* of *tū* dem *rūꞌau* old\_woman *māmā* mother *era* dist *he* pred *kā* kindle *i* acc *te* art *ꞌumu* earth\_oven *paurō* every *te* art *mahana.* day **13** *E* ipfv *kā* kindle *era* dist *i* acc *tou* dem *ꞌumu* earth\_oven *era* dist *paurō* every *te* art *mahana,* day *ꞌina* neg *he* ntr *ꞌōꞌotu* cooked *te* art *ꞌumu* earth\_oven *e* ipfv *pō* night *rō* emph *era.* dist **14** *E* ipfv *pō* night *rō* emph *ꞌā* cont *e* ipfv *ꞌōꞌotu* cooked *era* dist *te* art *ꞌumu* earth\_oven *e* ipfv *maꞌoa* open\_earth\_oven *era,* dist *ꞌai* subs *ka* cntg *kakai* pl:eat *ananake* together *ko* prom *te* art *ŋā* pl *poki.* child **15** *He* ntr *uꞌi* look *paꞌi* in\_fact *i* acc *te* art *meꞌe* thing *era* dist *ꞌina* neg *ꞌō* really *he* ntr *kai* eat *ꞌi* at *te* art *ꞌōtea* daytime *ꞌā.* ident **16** *E* ipfv *pō* night *rō* emph *ꞌana* cont *ꞌō* really *e* ipfv *kai* eat *era,* dist *e* ipfv *mōkirokiro* nightfall *rō* emph *ꞌā.* cont **17** *He* ntr *ꞌaroha* pity *paꞌi* in\_fact *he* ntr *kī:* say **18** *—¡Kai* neg.pfv *riva* good *ꞌō* really *te* art *kai* eat *nei* prox *o* of *tātou,* 1pl.incl *e* voc *repa* young\_man *ē!* voc **19** *ꞌI* in *te* art *ꞌao* day *nui* big *ꞌō* really *e* ipfv *kai* eat *nei* prox *te* art *ŋā* pl *poki,* child *e* ipfv *haꞌuru* sleep *rō* emph *ꞌō* really *te* art *ŋā* pl *poki.* child

**12** When they had raised them and they had grown up, what the old mother did was cooking food in the earth oven every day. **13** When she cooked food in the oven every day, the food was not cooked before it got dark. **14** When it was dark, the food would be done and she would open the earth oven; then she would eat with the children. **15** So she saw that they did not eat in the daylight. **16** It was night when they ate, it was dark. **17** So she took pity and said, **18** "The way we eat is not right, my son! **19** The children eat at midnight, when they are asleep."

**20** *He* ntr *oti* finish *ꞌana* cont *he* ntr *kī* say *e* ag *tou* dem *poki* child *era* dist *o* of *tou* dem *rūꞌau* old\_woman *era* dist *ko* prom *Tikitiki* Tikitiki *ꞌa* a *ꞌAtaraŋa:* Ataranga **21** *—¿Maꞌa* know *ꞌā* cont *rō* emph *koe,* 2sg *e* voc *nua* Mum *ē?* voc **22** *He* ntr *kī* say *tou* dem *rūꞌau* old\_woman *era:* dist **23** *—ꞌE,* and *¿he* pred *aha?* what **24** *—A* prop *au* 1sg *he* ntr *eꞌa* go\_out *a* prop *au* 1sg *he* ntr *oho.* go **25** *He* ntr *oho* go *au* 1sg *ki* to *ŋā* pl *hare* house *he* pl *noꞌinoꞌi* request:red *hai* ins *kona* place *mahute,* mulberry *hauhau* kind\_of\_tree *mo* for *hiro* braid *o* of *te* art *taura* rope *mo* for *tāea* catch\_with\_lasso *e* ag *au* 1sg *i* acc *te* art *raꞌā* sun *mo* for *here.* tie **26** *¿ꞌE* and *ku* prf *tano* correct *ꞌā,* cont *e* voc *nua* Mum *ē?* voc **27** *—e* ipfv *kī* say *era* dist *ki* to *tōꞌona* poss.3sg.o *nua* Mum *nei.* prox **28** *He* ntr *kī:* say **29** *—ꞌĒē,* yes *ku* prf *tano* correct *ꞌā.* cont

**20** Then the son of the old woman, Tikitiki a Ataranga, said, **21** "You know what, Mum?" **22** The old woman said, **23** "Well, what?" **24** "I will go out. **25** I will go to the houses and ask for mulberry and hauhau fibers to braid a rope to catch the sun with a lasso. **26** Is that okay, Mum?" **27** he said to his mother. **28** She said, **29** "Yes, that's okay."

**30** *I* pfv *eꞌa* go\_out *era* dist *ꞌi* at *te* art *ꞌao* dawn *era* dist *ꞌā* ident *o* of *te* art *rua* two *raꞌā,* day *he* ntr *oho* go *he* ntr *tuꞌu* arrive *he* ntr *paꞌo* chop *mai* hither *i* acc *tāꞌana* poss.3sg.a *mahute* mulberry *i* acc *tāꞌana* poss.3sg.a *hauhau,* kind\_of\_tree *he* ntr *hahaꞌo* insert *ki* to *roto* inside *te* art *vai* water *he* ntr *haka* caus *pā* double *mo* for *haka* caus *marere* fall\_apart *mo* for *haka* caus *piro* rotten *haka* caus *kōpiro* ferment *ꞌi* at *roto* inside *i* at *te* art *vai* water *mo* for *te* art *taura* rope *mo* for *hiri.* braid **31** *He* ntr *oho* go *ki* to *te* art *hare* house *o* of *tōꞌona* poss.3sg.o *ŋaruhoa* friend *e* ipfv *noho* stay *era* dist *ananake.* together **32** *ꞌE* and *i* pfv *tupuꞌaki* near *ꞌā* cont *ꞌi* at *tōꞌona* poss.3sg.o *hare,* house *he* ntr *oho* go *he* ntr *tuꞌu* arrive *he* ntr *noꞌinoꞌi* request:red *hai* ins *mahute* mulberry *hai* ins *hauhau.* kind\_of\_tree **33** *He* ntr *rovaꞌa* obtain *mai.* hither *34 He* ntr *maꞌu* carry *he* ntr *oho* go *mai* hither *he* ntr *tuꞌu,* arrive *he* ntr *puꞌa* cover *he* ntr *haka* caus *kōpiro* ferment *mo* for *te* art *taura* rope *mo* for *hiri.* braid

**30** When he had gone out at dawn the next day, he went and chopped mulberry and hauhau trees; he put the fibres in the water and folded them so they would fall apart and rot and ferment in the water, to braid a rope from. **31** He went to the houses of his friends who lived together with him. **32** And when he was near their houses, he went and asked for mulberry and hauhau. **33** He got them. **34** He carried them away and covered them (with water) to ferment to braid a rope from.

**35** *I* pfv *oti* finish *era* dist *tū* dem *meꞌe* thing *era,* dist *he* ntr *toꞌo* take *koroꞌiti* slowly *mai* hither *tū* dem *taura* rope *era* dist *i* acc *tou* dem *hau* cord *era,* dist *he* ntr *hiro* braid *i* acc *te* art *taura.* rope **36** *He* ntr *hiro* braid *ka* cntg *hiro* braid *era* dist *ē…* on\_and\_on **37** *ka* cntg *kumi* long *ka* cntg *kumi.* long **38** *He* ntr *toꞌo* take *mai* hither *i* acc *tū* dem *taura* rope *era* dist *he* ntr *aŋa* make *ꞌā* until *ka* cntg *oti* finish *rō.* emph **39** *I* pfv *oti* finish *era* dist *tū* dem *taura* rope *era,* dist *he* ntr *oho* go *mai* hither *he* ntr *kī* say *ki* to *tū* dem *rūꞌau* old\_woman *māmā* mother *era:* dist **40** *—Āpō* tomorrow *ꞌi* at *te* art *ꞌao* dawn *era* dist *ꞌā,* ident *e* ipfv *oŋa* appear *iho* just\_then *nō* just *mai* hither *ꞌā* cont *te* art *raꞌā,* sun *he* ntr *rere* jump *au* 1sg *he* ntr *oho* go *he* ntr *tuꞌu* arrive *he* ntr *tāea* catch\_with\_lasso *i* acc *te* art *ŋao* neck *o* of *te* art *raꞌā,* sun *he* ntr *here* tie *a* by *ruŋa* above *i* at *te* art *māꞌea.* stone **41** *Haka* caus *rito* ready *koe,* 2sg *e* voc *nua* Mum *ē,* voc *mo* for *kā* kindle *i* acc *to* art:of *tātou* 1pl.incl *ꞌumu* earth\_oven *āpō.* tomorrow **42** *He* ntr *kī* say *tū* dem *rūꞌau* old\_woman *era:* dist **43** *—Ku* prf *tano* correct *ꞌā.* cont

#### Appendix A: Interlinear texts

**35** When that was done, he took the fibres bit by bit and braided a rope. **36** He braided and braided **37** until the rope was very long. **38** He took the rope and kept working until it was finished. **39** When the rope was finished, he went and said to his old mother, **40** "Tomorrow early in the morning, just when the sun appears, I will jump and catch the neck of the sun with a lasso and tie it to a rock. **41** Get ready, Mum, to cook our earth oven tomorrow." **42** The old woman said, **43** "That's okay."

**44** *Ka* cntg *moe* lie\_down *ka* cntg *ꞌara,* wake\_up *ka* cntg *moe* lie\_down *ka* cntg *ꞌara* wake\_up *ē….* on\_and\_on **45** *ꞌI* at *te* art *ꞌao* dawn *era* dist *ꞌā* ident *i* pfv *ꞌara* wake\_up *rō* emph *ai,* pvp *he* ntr *tiaki* wait *he* ntr *haka* caus *tau* hang *he* ntr *noho* stay *he* ntr *uꞌi* look *ki* to *te* art *raꞌā.* sun **46** *I* pfv *oho* go *mai* hither *era* dist *mo* for *hāhine* near *mai* hither *mo* for *eꞌa* go\_out *mai* hither *o* of *te* art *raꞌā,* sun *he* ntr *rere* jump *he* ntr *oho.* go **47** *Ka* cntg *rere* jump *atu* away *ka* cntg *oho* go *atu,* away *ka* cntg *oŋa* appear *mai* hither *te* art *raꞌā,* sun *he* ntr *haka* caus *eke* go\_up *hai* ins *taura* rope *ki* to *te* art *ŋao,* neck *he* ntr *here* tie *hiohio* strong *taꞌa* poss.2sg.a *ika.* victim **48** *He* ntr *totoi* drag *he* ntr *oho* go *mai* hither *i* acc *te* art *taura,* rope *he* ntr *tuꞌu* arrive *mai* hither *he* ntr *here* tie *ki* to *ruŋa* above *ki* to *te* art *puku* boulder *māꞌea* stone *ena,* med *ꞌā* until *ka* cntg *ŋōŋō* firm *rō* emph *ka* cntg *ŋōŋō* firm *rō,* emph *ꞌā* until *ka* cntg *harara* stiff *rō.* emph **49** *He* ntr *here,* tie *ꞌā* until *ka* cntg *hiohio* strong *rō,* emph *ꞌina* neg *kai* neg.pfv *hōrou* quickly *hakaꞌou* again *mai* hither *tū* dem *raꞌā* sun *era.* dist **50** *He* ntr *iri* ascend *mai* hither *tū* dem *raꞌā* sun *era* dist *koroꞌiti* slowly *nō* just *koroꞌiti* slowly *nō* just *i* pfv *iri* ascend *mai* hither *ai.* pvp

**44** He slept and woke up, he slept and woke up… **45** Early in the morning he woke up and stayed on the lookout, keeping an eye on the sun. **46** When the sun was close to rising, he jumped up. **47** Just when he jumped, the sun appeared; he put the rope to the neck and tied his victim firmly. **48** He dragged the rope and tied it to a stone boulder, so it was very firmly tied and the rope was taut. **49** He tied it firmly and the sun did not go quickly any more. **50** The sun came up; slowly, slowly it came up.

**51** *He* ntr *eꞌa* go\_out *mai* hither *tou* dem *rūꞌau* old\_woman *era* dist *ꞌi* at *tou* dem *hora* time *ꞌā* ident *e* quickly *ki* to *tāꞌana* poss.3sg.a *ꞌumu,* earth\_oven *he* ntr *amo* wipe *ka* cntg *maꞌitaki* clean *rō.* emph **52** *He* ntr *penapena* arrange\_firewood *i* acc *te* art *ꞌumu* earth\_oven *pae,* hewn\_stones *he* ntr *tutu.* set\_fire **53** *He* ntr *oho* go *ki* to *tāꞌana* poss.3sg.a *kūmā* sweet\_potato *he* ntr *keri* dig *mai* hither *he* ntr *tata,* wash *ki* to *tāꞌana* poss.3sg.a *moa* chicken *he* ntr *unu.* pluck **54** *Pōpōrā* quickly *e* ipfv *keꞌokeꞌo* hurry *mai* hither *era* dist

*te* art *rima* hand *o* of *tou* dem *rūꞌau* old\_woman *era.* dist **55** *He* ntr *toꞌo* take *mai* hither *he* ntr *kā* kindle *i* acc *tou* dem *ꞌumu* earth\_oven *era* dist *he* ntr *ꞌōꞌotu.* cooked **56** *He* ntr *taꞌo* cook *i* acc *te* art *kūmā,* sweet\_potato *i* acc *te* art *taro,* taro *i* acc *te* art *ꞌuhi,* yam *i* acc *te* art *moa.* chicken **57** *Ko* prf *turu* go\_down *ꞌā* cont *tū* dem *taŋata* man *era* dist *ki* to *tai* sea *ko* prf *hī* to\_fish *mai* hither *ꞌā* cont *i* acc *te* art *ika,* fish *e* ipfv *here* tie *ꞌā* cont *i* acc *te* art *kōreha.* eel **58** *Tanu* bury *te* art *ꞌumu* earth\_oven *raꞌe* first *nei* prox *o* of *te* art *rūꞌau* old\_woman *nei,* prox *ko* prf *tata* approach *mai* hither *ꞌā* cont *tū* dem *taŋata* man *era* dist *mo* for *tou* dem *ika* fish *era.* dist

**51** At that time the old women went quickly to her earth oven and wiped it clean. **52** She arranged the firewood over the earth oven made of hewn stones, and kindled it. **53** She went and dug up sweet potatoes and washed them, and plucked chickens. **54** The hands of the old woman moved quickly. **55** She lighted the earth oven and got the food cooked. **56** She cooked sweet potatoes, taro, yam and chicken. **57** The man had gone down to sea to catch fish with hook and line and to catch eels with a trap. **58** The old woman covered the first earth oven when the man had come back with his fish.

**59** *Puhi* blow *hakaꞌou* again *ꞌā* cont *ka* cntg *rua* two *ꞌumu.* earth\_oven **60** *Ko* prf *ꞌōꞌotu* cooked *hakaꞌou* again *mai* hither *ꞌā* cont *te* art *taꞌo,* cook *penapena* arrange\_firewood *hakaꞌou* again *mai* hither *ꞌā* cont *ka* cntg *toru* three *ꞌumu.* earth\_oven **61** *He* ntr *eꞌa* go\_out *mai* hither *tou* dem *rūꞌau* old\_woman *era* dist *ki* to *tou* dem *ꞌumu* earth\_oven *raꞌe* first *era,* dist *he* ntr *mātaki* open *mai.* hither **62** *He* ntr *raŋi* call *mai* hither *ki* to *te* art *ŋā* pl *poki:* child **63** *—¡Ka* imp *oho* go *mai!* hither **64** *He* ntr *oho* go *atu* away *tou* dem *viꞌe* woman *hunoŋa* child\_in\_law *era,* dist *tou* dem *ŋā* pl *poki* child *era* dist *he* ntr *takataka,* gather:red *he* ntr *maꞌoa* open\_earth\_oven *tou* dem *ꞌumu* earth\_oven *era.* dist

**59** She lighted another earth oven, the second one. **60** When the food was cooked again, she arranged firewood again for the third earth oven. **61** The old woman went to the first earth oven and opened it. **62** She cried to the children, **63** "Come!" **64** Her daughter-in-law and the children gathered and she opened the earth oven.

**65** *He* ntr *hoa* throw *te* art *kai* food *ki* to *ruŋa,* above *he* ntr *kī:* say **66** *—Pē* like *nei* prox *hoki* indeed *taꞌa* poss.2sg.a *meꞌe.* thing **67** *ꞌĪ* imm *tātou* 1pl.incl *ka* cntg *rovā* obtain *kai.* food **68** *Ka* cntg *kai* eat *iho* just\_now *tātou* 1pl.incl *ꞌi* at *te* art *hora* time *māꞌeha* light *ꞌi* at *te* art *ꞌōtea.* daytime **69** *Kai* neg.pfv *riva* good *hoki* indeed *taꞌa* poss.2sg.a *meꞌe* thing *era* dist *tātou* 1pl.incl *e* ipfv *kakai* pl:eat *era* dist *ꞌi* at *te* art *pō* night *ꞌi* at *te* art *ꞌao* dusk *nui.* big **70** *ꞌIna* neg *he* pred *māꞌeha* light *mo* for *uꞌi* see *iŋa* nmlz *i* acc *te* art *kai.* food **71** *E* ipfv *hāꞌuru* pl:sleep *rō* emph *era* dist *te* art *ŋā* pl *poki,* child *ꞌina* neg *he* ntr *kakai.* pl:eat **72** *¡ꞌĀ,* ah *meꞌe* thing *ꞌaroha!* pity **73** *ꞌO* because\_of *ira* ana *ꞌī* imm *tātou* 1pl.incl *ka* cntg *toꞌo* take *rō* emph *mai* hither *ka* cntg *kai* eat *rō* emph *atu.* away **74** *He* ntr *kai* eat *mai* hither *i* acc *tou* dem *ꞌumu* earth\_oven *era,* dist *he* ntr *oti.* finish **75** *He* ntr *maꞌoa* open\_earth\_oven *hakaꞌou* again *mai* hither *ka* cntg *rua* two *ꞌumu* earth\_oven *ꞌi* at *ruŋa* above *ꞌā,* ident *he* ntr *kai* eat *hakaꞌou* again *mai.* hither **76** *He* ntr *taꞌo* cook *hakaꞌou* again *mai* hither *i* acc *te* art *ꞌumu.* earth\_oven **77** *I* pfv *taꞌo* cook *i* pfv *oho* go *nō* just *i* pfv *taꞌo* cook *i* pfv *oho* go *nō.* just

**65** She took the food out and said, **66** "Here, look at this. **67** Now we have something to eat. **68** We will eat in the daytime, when it is light. **69** It was not good what we did, eating in the dark at midnight. **70** There was no light to see the food. **71** The children were asleep and did not eat. **72** Ah, what a pity! **73** So we will go ahead now and eat." **74** They ate the food from the earth oven and finished. **75** Then they opened the second earth oven and ate again. **76** They cooked another earth oven. **77** They just kept cooking and cooking.

**78** *I* pfv *ahiahi* evening *era,* dist *he* ntr *tikeꞌa* see *e* ag *te* art *taŋata* man *ko* prf *here* tie *ꞌā* cont *i* acc *te* art *raꞌā.* sun **79** *He* ntr *oho* go *mai* hither *he* ntr *tuꞌu* arrive *mai* hither *he* ntr *haka* caus *takataka* gather:red *ꞌi* at *tū* dem *hare* house *era* dist *o* of *tū* dem *taŋata* man *era.* dist **80** *He* ntr *taū* to\_fight *te* art *tauꞌa,* fight *he* ntr *rakerake* bad:red *te* art *tauꞌa* fight *he* ntr *noho* stay *he* ntr *kakai.* quarrel **81** *Kakai* quarrel *i* acc *te* art *tauꞌa* fight *nei,* prox *ka* cntg *rakerake* bad:red *rō* emph *ꞌi* at *te* art *hare* house *era* dist *o* of *tū* dem *taŋata* man *era.* dist **82** *He* ntr *kī:* say **83** *—¿He* pred *aha* what *koe* 2sg *i* pfv *toꞌo* take *mai* hither *ai* pvp *i* pfv *here* tie *ai* pvp *i* acc *te* art *raꞌā?* sun

**78** In the evening, the people saw that the sun had been tied. **79** They came and gathered at the house of that man. **80** A fight broke out, there was an ugly quarrel. **81** There was a bad quarrel at the house of that man. **82** They said, **83** "Why did you take the sun and tie it up?"

**84** *He* ntr *kī* say *e* ag *tū* dem *taŋata* man *era:* dist **85** *—Ka* imp *koa* happy *hoꞌi* indeed *kōrua.* 2pl **86** *¿Kai* neg.pfv *koa* happy *ꞌō* really *kōrua* 2pl *ꞌi* at *te* art *meꞌe* thing *rivariva?* good **87** *Ku* prf *toꞌo* take *mai* hither *ꞌā* cont *a* prop *au* 1sg *i* acc *te* art *raꞌā* sun *ku* pfv *here* tie

*ꞌā* cont *mo* for *haka* caus *māꞌeha* light *mo* for *tātou* 1pl.incl *mo* for *kakai.* pl:eat **88** *ꞌO* because\_of *ira* ana *hoꞌi* indeed *au* 1sg *i* pfv *here* tie *ena.* med **89** *Nā* med *ka* imp *uꞌi* look *rā* intens *kōrua,* 2pl *ka* cntg *hia* how\_many *ꞌumu* earth\_oven *nei* prox *ꞌāꞌaku* poss.1sg.a *e* ipfv *kā* kindle *atu* away *ena.* med **90** *He* ntr *kā* kindle *te* art *ꞌumu* earth\_oven *nei* prox *he* ntr *maꞌoa* open\_earth\_oven *he* ntr *kai.* eat **91** *He* ntr *kā* kindle *hakaꞌou* again *he* ntr *maꞌoa* open\_earth\_oven *he* ntr *kai.* eat **92** *ꞌAi* subs *e* ipfv *kā* kindle *hakaꞌou* again *mai* hither *era.* dist

**84** The man said, **85** "Be happy! **86** Aren't you happy with a good thing? **87** I have taken the sun and tied it up to give us light to eat. **88** That's why I tied it up. **89** There, look how many earth ovens I've been cooking. **90** We lighted this earth oven, opened it and ate. **91** We lighted another, opened it and ate. **92** Then we lighted another."

**93** *He* ntr *ture* scold *ka* cntg *rakerake* bad:red *rō* emph *te* art *tauꞌa.* fight **94** *He* ntr *rere* jump *mai* hither *e* ag *te* art *taŋata* man *he* ntr *horehore* cut:red *i* acc *tou* dem *taura* rope *era.* dist **95** *He* ntr *momotu* break\_completely *he* ntr *haka* caus *pokoꞌo.* loose **96** *I* pfv *haka* caus *pokoꞌo* loose *atu* away *era* dist *i* acc *te* art *raꞌā* sun *era,* dist *he* ntr *nini* spin *rō* emph *ꞌai* subs *tū* dem *raꞌā* sun *era* dist *he* ntr *oho* go *he* ntr *ŋaro.* disappear **97** *He* ntr *raŋi* call *e* ag *te* art *rūꞌau* old\_woman *nei:* prox **98** *—E* ipfv *paꞌo* chop *rō* emph *kōrua* 2pl *i* acc *te* art *raꞌā* sun *here* tie *e* ag *tāꞌaku* poss.1sg.a *poki.* child **99** *Rohirohi* tired *e* ag *tāꞌaku* poss.1sg.a *poki* child *i* pfv *iri* ascend *ai* pvp *i* pfv *here* tie *mai* hither *ai* pvp *mo* for *haka* caus *māꞌeha* light *o* of *tātou.* 1pl.incl **100** *ꞌE* and *ꞌī* imm *au* 1sg *ka* cntg *raŋi* call *rō* emph *atu:* away *e* ipfv *toꞌo* take *rō* emph *au* 1sg *i* acc *taꞌa* poss.2sg. *meꞌe* thing *hukahuka* firewood *ena* med *mai* from *ruŋa* above *i* at *te* art *henua.* land **101** *ꞌĪ* imm *ka* cntg *toꞌo* take *rō.* emph

**93** They made reproaches and quarreled bitterly. **94** The people jumped up and cut the rope. **95** It snapped and the sun was loose. **96** When the sun was loose, it turned around and disappeared. **97** The old woman cried out, **98** "You're chopping the rope with which my son tied the sun. **99** My sun wore himself out to go up and tie the sun to give us light. **100** Now I'll tell you: I will take your firewood from the earth. **101** I'll take it straightaway."

**102** *He* ntr *rere* jump *mai* hither *te* art *rūꞌau* old\_woman *nei* prox *ki* to *tāꞌana* poss.3sg.a *ꞌumu* earth\_oven *pae,* hewn\_stones *ki* to *tāꞌana* poss.3sg.a *hukahuka,* firewood *ki* to *tāꞌana* poss.3sg.a *kai,* food *ki* to *tāꞌana* poss.3sg.a *meꞌe* thing *taꞌatoꞌa.* all **103** *He* ntr *ꞌapa* carry *tahi* all *ko* prom *te* art *ŋā* pl *poki,* child *ko* prom *te* art *hare,* house *ko* prom *te* art *meꞌe* thing *taꞌatoꞌa.* all **104** *He* ntr *rere* fly *he* ntr *oho* go *rō* emph *atu* away *ꞌai,* subs *ko* prom *te* art *iri* ascend *iŋa* nmlz *ꞌā* ident *ko* prom *te* art *oho* go *oŋa* nmlz *ꞌā,* ident *ē…* on\_and\_on **105** *he ŋaro ki roto ki te māhina, he oti rō ꞌai.* **106** *Ki roto i te*

ntr disappear to inside to art moon ntr finish emph subs to inside at art *māhina* moon *i* pfv *ŋaro* disappear *ai* pvp *te* art *rūꞌau* old\_woman *nei.* prox **107** *Te* art *ꞌīŋoa* name *o* of *tou* dem *rūꞌau* old\_woman *era* dist *ko* prom *Nuahine* Nuahine *ꞌa* a *Raŋi* Rangi *Kotekote.* Kotekote **108** *He* ntr *ŋaro* disappear *rō* emph *atu* away *ꞌai* subs *ki* to *roto* inside *ki* to *te* art *māhina.* moon

**102** The old woman jumped to her stone earth oven, to her firewood, to her food, to all her things. **103** She carried all the children, the house, everything. **104** She flew away, going up, going away… **105** and disappeared into the moon, and that was it. **106** Inside the moon the old woman disappeared. **107** The name of the old woman was Nuahine a Rangi Kotekote. **108** She disappeared into the moon.

**109** *He* ntr *kī* say *te* art *taŋata:* man **110** *—¡ꞌĀ,* ah *ko* prf *tere* run *ꞌā* cont *te* art *rūꞌau!* old\_woman **111** *I* pfv *ꞌōtea* daylight *mai* hither *era* dist *ki* to *te* art *popohaŋa* dawn *he* ntr *uꞌi* look *te* art *taŋata* man *ku* prf *oŋe* shortage *ꞌā,* cont *ꞌina* neg *he* pred *hukahuka.* firewood **112** *Ku* prf *oŋe* shortage *ꞌā* cont *te* art *kai,* food *ko* prf *pakapaka* dry:red *ꞌā* cont *ku* prf *mei* wither *ꞌā.* cont **113** *He* ntr *kī:* say **114** *—¡ꞌĀ!* ah **115** *O* of *tou* dem *rūꞌau* old\_woman *era* dist *ꞌā* ident *te* art *kai,* food *e* ipfv *topa* happen *era* dist *te* art *mau,* abundance *te* art *hukahuka* firewood *e* ipfv *ai* exist *era.* dist **116** *Ku* prf *kore* lack *ꞌā* cont *te* art *hukahuka* firewood *mo* for *tunu* cook *o* of *te* art *kai.* food **117** *Ku* prf *pakapaka* dry:red *ꞌā* cont *te* art *henua.* land **118** *Ku* prf *oŋe* shortage *ꞌā* cont *tātou.* 1pl.incl **119** *He* ntr *oti* finish *mau* really *ꞌā.* cont **120** *¿Ka* cntg *aha* what *rā* intens *ia?* then **121** *Ku* prf *tere* run *ꞌā* cont *te* art *rūꞌau* old\_woman *nei.* prox

**109** The people said, **110** "Ah, the old woman has run off!" **111** At dawn the next day the people saw that there was shortage, there was no firewood. **112** The food was scarce, it was dry and had withered. **113** They said, **114** "Ah! **115** It was because of the woman that there was food, that there was abundance, that there was firewood. **116** Now there is no firewood to cook food. **117** The land is dry. **118** We are in need. **119** That's how it is. **120** What can we do? **121** The old woman has run off."

**122** *Kai* neg.pfv *ꞌite* know *atu* away *hoꞌi* indeed *e* ag *te* art *taŋata* man *pē* like *nei* prox *ē:* thus *tou* dem *rūꞌau* old\_woman *era* dist *ꞌō* really *e* ipfv *ai* exist *rō* emph *ꞌā* cont *tōꞌona* poss.3sg.o *mana;* power *rūꞌau* old\_woman *mana.* power **123** *Pē* like *ira* ana *ꞌā* ident *tū* dem *taŋata* man *era.* dist **124** *ꞌO* because\_of *ira* ana *hoꞌi* indeed *i* pfv *iri* ascend *era* dist *i* pfv *here* tie *era* dist *i* acc *te* art *raꞌā* sun *era.* dist **125** *E* ipfv *ai* exist *rō* emph *ꞌā* cont *to* art:of *rāua* 3pl *mana.* power

**122** The people did not know that the old woman had mana (supernatural power); it was a woman with power. **123** The same was true for the man. **124** That's why he went up and tied the son. **125** They had supernatural power.

**126** *He* ntr *oti* finish *mau* really *ꞌā.* cont **127** *Ku* prf *ŋaro* disappear *ku* prf *oti* finish *ꞌā* cont *te* art *rūꞌau* old\_woman *nei* prox *ki* to *roto* inside *ki* to *te* art *māhina.* moon **128** *ꞌE* and *ku* prf *noho* stay *ꞌā* cont *te* art *taŋata* man *o* of *te* art *henua* land *he* ntr *taŋi* cry *rō* emph *ꞌai* subs *mo* for *tou* dem *rūꞌau* old\_woman *era* dist *ꞌi* at *tū* dem *meꞌe* thing *era* dist *ko* prf *ŋaro* disappear *ꞌā.* cont **129** *He* ntr *oti* finish *mau* really *rō* emph *ꞌai* subs *te* art *ꞌaꞌamu* story *nei* prox *o* of *Tikitiki* Tikitiki *ꞌa* a *ꞌAtaraŋa.* Ataranga

**126** That's all. **127** The old woman had disappeared into the moon. **128** And the people of the land stayed and cried for the woman, because she had disappeared. **129** The story of Tikitiki a Ataranga is finished.

Appendix A: Interlinear texts

### *He oho iŋa o te nuꞌu hoko rua rama* **– The trip of two people who went torch fishing (R357)**

(Weber, Weber & team 1990a: Vol. 5:22–23)

**01** *Ko* prom *te* art *ꞌāvaꞌe* month *era* dist *o* of *te* art *evinio* Lent *ꞌe* and *ko* prom *te* art *mahana* day *maha* four *ia,* then *he* ntr *eꞌa* go\_out *te* art *nuꞌu* people *hoko* num.pers *rua,* two *he* ntr *puꞌa* cover *i* acc *te* art *hoi,* horse *he* ntr *aŋa* make *i* acc *te* art *rama,* torch *he* ntr *haka* caus *rito* ready *i* acc *te* art *rāua* 3pl *meꞌe,* thing *he* ntr *eke* go\_up *te* art *hoi,* horse *he* ntr *oho* go *ararua* the\_two *a* by *te* art *pāꞌeŋa* side *ena* med *o* of *Vaihū.* Vaihu **02** *I* pfv *tuꞌu* arrive *era* dist *ki* to *ꞌAkahaŋa,* Akahanga *he* ntr *noho* stay *he* ntr *tiaki* wait *ki* to *te* art *tai* sea *hahati* break *mo* for *rama.*

fish\_with\_torch

**01** In the month of Lent on a Thursday, two people went out; they saddled their horses, made torches and prepared their things; they mounted their horses and went together to the side of Vaihu. **02** When they arrived at Akahanga, they stayed and waited for the rising tide (suitable) for torch fishing.

**03** *Ko* prom *rote* inside\_art *ꞌahuru* ten *mā* plus *hoꞌe* one *o* of *te* art *pō* night *ia,* then *he* ntr *eꞌa* go\_out *ararua,* the\_two *he* ntr *tutu* set\_fire *i* acc *te* art *rama,* torch *he* ntr *topa* descend *he* ntr *rama* fish\_with\_torch *ꞌi* at *rote* inside\_the *haŋa* bay *era* dist *o* of *ꞌAkahaŋa.* Akahanga **04** *E* ipfv *rama* fish\_with\_torch *nō* just *ꞌā,* cont *i* pfv *hoa* throw *rō* emph *atu* away *ai* then *te* art *ꞌua* rain *ki* to *te* art *kāveꞌuveꞌu* ache *o* of *te* art *tuaꞌivi.* back **05** *Kai* neg.pfv *rovaꞌa* obtain *hia* yet *e* num *tahi* one *meꞌe.* thing **06** *He* ntr *eꞌa* go\_out *ararua,* the\_two *he* ntr *tētere* pl:run *ki* to *rote* inside\_the *ꞌana,* cave *he* ntr *nonoho* pl:stay *koia* with *ko* prom *mamae* pain *ꞌi* at *tū* dem *meꞌe* thing *taꞌe* conneg *rovaꞌa* obtain *era.* dist

**03** Around eleven o'clock at night the two went out, lighted their torches, went down and went torch fishing in the bay of Akahanga. **04** While they were fishing, suddenly the rain came down, lashing their backs. **05** They didn't catch anything. **06** The two left and fled into a cave, sad because they hadn't caught anything.

**07** *I* pfv *nonoho* pl:stay *era,* dist *he* ntr *papaŋahaꞌa* pl:heavy *ꞌi* at *te* art *haꞌuru.* sleep **08** *E* ipfv *haꞌuru* sleep *nō* just *ꞌā,* cont *he* ntr *tuꞌu* arrive *atu* away *hoko* num.pers *rua* two *hakaꞌou* again *nuꞌu* people *mai* from *te* art *puhi* fish\_at\_night *iŋa* nmlz *mo* for *te* art *evinio.* Lent **09** *He* ntr *hipa* pass\_by *atu* away *he* ntr *haka* caus *ꞌara* wake\_up *i* acc *tū* dem *ŋā* pl *haꞌuru* sleep *era.* dist **10** *I* pfv *ꞌaꞌara* pl:wake\_up *mai* hither *era,* dist *he* ntr *eke* go\_up *te* art *hoi,* horse *he* ntr *oho* go *mai* hither *mai* from *ꞌAkahaŋa,* Akahanga *he* ntr *tuꞌu* arrive *ki* to *Motu* Motu *o* o *Pope.* Pope **11** *He* ntr *topa* descend *hakaꞌou* again *ararua* the\_two *ꞌi* at *tū* dem *kona* place *era* dist *he* ntr *tutu* set\_fire *te* art *rama.* torch

**07** While they stayed there, they fell asleep. **08** While they were sleeping, two other people arrived, who had been fishing at night for Lent. **09** They came by and woke the sleepers up. **10** Once awake, they mounted their horses, left Akahanga and went to Motu o Pope. **11** There the two descended again and lighted their torches.

**12** *I* pfv *māꞌeha* light *atu* away *era* dist *tū* dem *rama* torch *era,* dist *ꞌī* imm *ka* cntg *uꞌi* look *atu* away *ena* med *ko* prom *te* art *ꞌura* lobster *ku* prf *noho* stay *ꞌā* cont *ꞌi* at *tū* dem *kona* place *era.* dist **13** *He* ntr *topa* descend *ararua* the\_two *he* ntr *ꞌaruꞌaru* grab:red *mai* hither *i* acc *tū* dem *ŋā* pl *ꞌura* lobster *era* dist *mai* from *tū* dem *kona* place *era.* dist **14** *He* ntr *rovaꞌa* obtain *e* num *hoꞌe* one *ꞌahuru* ten *mā* plus *piti* two *ꞌura.* lobster **15** *Ku* prf *koa* happy *ꞌā* cont *ko* prom *ruŋa* above *i* at *te* art *hoi,* horse *he* ntr *ꞌehu* twilight *he* ntr *oho* go *rō* emph *mai* hither *ꞌai* subs *ki* to *Haŋa* Hanga *Roa* Roa *ꞌi* at *tū* dem *hora* time *era.* dist

**12** When their torches gave light, they saw lobsters sitting in that place. **13** The two went down and grabbed the lobsters from there. **14** They got twelve lobsters. **15** Happily they climbed their horses, and in the early morning light they went to Hanga Roa.

**16** *Tū* dem *oho* go *era* dist *hoꞌi* indeed *i* pfv *oho* go *ai* pvp *e* ipfv *oho* go *ꞌā* cont *mo* for *rovaꞌa* obtain *ꞌura* lobster *mo* for *eꞌa* go\_out *ki* to *ruŋa* above *ki* to *te* art *vaka* boat *ki* to *te* art *nanue* kind\_of\_fish *hī* to\_fish *ꞌi* at *te* art *motu* islet *mo* for *kai* eat *o* for *te* art *evinio.* Lent **17** *ꞌI* at *te* art *meꞌe* thing *era,* dist *te* art *taŋata* person *o* of *nei* prox *ꞌi* at *rā* dist *hora* time *ꞌina* neg *he* ntr *kai* eat *i* acc *te* art *kiko* meat *ꞌi* at *te* art *taꞌatoꞌa* all *mahana* day *pae* five *o* of *te* art *evinio.* Lent

**16** This trip happened when they went to catch lobsters (as bait), to go out by boat to fish for nanue at the islets, to eat during Lent. **17** Because at that time the people here did not eat meat on Fridays during Lent.

## **Appendix B: The text corpus**

Below is a list of texts in the corpus. The first column gives the reference as given in the example sentences in this grammar (*x* represents any digit). The second column gives a basic characterisation of the text type. The third column provides a short description; for published texts, a bibliographic reference is given. The final column gives the number of words in thousands.

The corpus is described in §1.6.2.

### **1. Older texts**


### **2. Texts from the 1970s**


Appendix B: The text corpus

### **3.** *Programa Lengua Rapa Nui* **corpus**



#### Appendix B: The text corpus


### **4. Bible translation**




(Baessler-Archiv, Beiträge zur Völkerkunde, Neue Folge, Beiheft 9). Berlin: Dietrich Reimer.



Campbell, Ramon. 1970. *La herencia musical de Rapanui*. Santiago: Editorial Andrés Bello.


Churchward, Clerk Maxwell. 1985 [1953]. *Tongan grammar*. Nuku'alofa: Vava'u Press.



Diamond, Jared. 2005. *Collapse: How societies choose to fail or succeed*. New York: Viking.


Du Feu, Veronica M. 1995. Asking questions in Rapa Nui. *Rongorongo Studies* 5(1). 27–30. Du Feu, Veronica M. 1996. *Rapanui* (Descriptive Grammars). London: Routledge.


Fedorova, Irina K. 1988. *Mify i legendy ostrova Paskhi*. St. Petersburg: Nauka.




Hotus Chavez, Alberto. 2008. *Diccionario práctico Rapanui – Castellano*.



Lichtenberk, Frantisek. 2000. Inclusory prononimals. *Oceanic Linguistics* 39(1). 1–32.




*memory of William T. Mulloy* (Oxbow Monographs in Archaeology 32), 169–171. Oxford: Oxbow Books.


Zewen, P. François. 1987. *Introduction à la langue des îles Marquises. Le parler de Nukuhiva. Hamana ha'avivini 'i te 'eo 'enana: Kohu'otetua*. Tahiti: Editions Haere Po no Tahiti.

Aikhenvald, Alexandra Y., 108, 368, 370 Alexander, James D., 14, 22, 125, 126, 316, 380, 384, 389, 391, 396– 398, 427 Anderson, Atholl, 3 Anderson, John M., 274, 453 Anderson, Stephen R., 128, 194, 432, 439 Anderson, Victoria, 42, 242 Andrews, Avery D., 272, 480 Atán, Karlo Huke, 21, 594 Ayres, Gabriella S., 21 Ayres, William S., 3, 21 Bahn, Paul, 2, 4 Bailey, Nicholas, 184 Baker, Christopher M. (Kaliko), 304 Bakker, Dik, 22 Ball, Douglas, 379, 385 Barthel, Thomas S., 24, 25, 591 Bauer, Winifred, 44, 51, 91, 96, 106, 121, 133, 152, 194, 226, 241, 278, 290, 328, 341, 409, 412, 424, 451, 460, 478, 498, 552 Bennardo, Giovanni, 307 Bergmann, Hans-Georg, 4, 10, 34, 140, 146, 195 Besnier, Niko, 44, 79, 82, 121, 241, 242, 304, 341, 368 Bhat, D. N. S., 108, 116, 255 Bierbach, Annette, 21 Biggs, Bruce, 23, 75, 77, 79, 107, 121, 123, 134, 249, 294, 307, 309, 320, 546 Blixen, Olaf, 19, 22, 24, 34, 56, 575, 591 Blust, Robert K., 5, 61, 72, 110 Boersema, Jan J., 4

Bork, Hans-Rudolf, 3, 4 Bowden, John, 121 Bowers, John, 424 Bradshaw, Joel, 368 Broschart, Jürgen, 90, 498 Buse, J. E., 75, 77, 107, 193, 250 Cablitz, Gabriele H., 123, 129, 179, 195, 241, 278, 304, 368 Caillot, A.-C., 2 Cain, Horst, 21 Calderón Haoa, Ivonne, 19 Callow, Kathleen, 407 Campbell, Ramon, 25 Capell, Arthur, 294, 307 Cardinali, Samuel, 592 Carroll, Vern, 220 Cea Egaña, Alfredo, 21 Chapin, Paul G., 19, 21–23, 201, 209, 233, 295, 316, 320, 329, 366, 367, 380, 397, 500 Chávez Ika, Ana, 592 Chung, Sandra, 106, 107, 241, 242, 304, 379, 398, 451 Churchill, William, 21, 33 Churchward, Clerk Maxwell, 121, 140 Clark, Ross, 3, 5, 7, 8, 11, 36, 42, 54, 86, 88, 89, 121–123, 125, 129, 133, 134, 150, 154, 193, 194, 198, 206, 209, 216, 217, 220, 226, 237, 242, 273, 276–278, 290, 294, 304, 309, 323, 329, 333, 341, 348, 350, 378–380, 397, 409, 410, 412, 424, 460, 464, 466, 471, 472, 485, 487, 500, 502, 504, 508, 538, 551, 565

Comrie, Bernard, 91, 320–323, 333, 334, 337, 378, 400, 530, 573 Conte Oliveros, Jesús, 19, 21 Cook, James, 4 Cook, Kenneth William, 241, 242, 244, 290, 316, 451, 475 Cooke, George H., 21, 33 Corney, Bolton Glanvill, 21 Crockford, Susan J., 3 Croft, William, 81, 83, 108, 111 Crowley, Terry, 5, 329, 368 Cuadros Hucke, Melinka, 24, 594 Cuny, Sara L. Collins, 2 Daams, Nico, 294n12 Davletshin, Albert, 24n29, 25, 31, 34, 45, 52, 61, 386 De Lacy, Paul, 44, 51, 451, 456 Di Castri, Francesco, 16 Diamond, Jared, 4 Dixon, R. M. W., 23, 108, 112, 113, 125, 152, 193, 231, 274, 276, 310, 320, 323, 359, 368, 378, 400, 439, 442–444, 448, 472, 477, 493, 497, 528 Dooley, Robert A., 188 Dryer, Matthew S., 23, 105, 141, 232, 239, 258, 307, 399, 400, 405, 451, 464, 466, 469 Du Feu, Veronica M., 3, 19, 21, 22, 53, 121, 190, 231, 237, 249, 251, 307, 320, 329, 345, 480 Duranti, Alessandro, 389 Durie, Mark, 368 Dutrou-Bornier, Jean-Baptiste, 4 Elbert, Samuel H., 5, 9, 77, 106, 107, 121, 123, 179, 241, 249, 276, 304, 307, 328, 341, 342, 386, 412, 478, 551, 555 Emory, Kenneth P., 2–5, 11, 25 Englert, Sebastian, 19, 21, 24, 34, 37, 42, 109, 119, 120, 131, 146, 152, 195,

201, 242, 251, 295, 307, 316, 329, 387, 500, 591 Evans, Te Kareongawai, 44, 51, 91, 96, 133, 152, 194, 241, 278, 290, 328, 341, 451, 478, 498, 552 Fedorova, Irina K., 21, 145, 150 Felbermayer, Fritz, 24, 474, 591 Feldman, Harry, 51 Fennig, Charles D., 5 Ferdon Jr., Edwin N., 2 Finney, Joseph C., 14, 22, 41, 66, 68, 125, 126, 316, 389, 390, 427 Fischer, Steven Roger, 1–4, 7, 9, 15–18, 21, 24, 25, 28, 36, 53, 57, 125, 150, 185, 201, 259, 276, 282, 340, 376, 523, 551 Flenley, John, 2, 4 Foerster, Rolf, 21 Foley, William A., 310, 400 Frontier, Arturo, 24, 591 Fuentes, J., 19, 21, 34, 37 Fuller, Judith, 22, 355 Geiseler, Kapitänleutnant, 21, 33 Geraghty, Paul, 10, 57 Gleisner, Christine, 20, 25 Goldberg, Adele E., 421 Gómez Macker, Luis, 22 González, Don Felipe, 4 Gordon, Lynn, 22 Gorrie, Colin, 240 Green, Roger C., 2, 3, 5, 7, 133, 194, 206, 290 Greenhill, Simon J., 8, 42, 123, 133, 134, 154, 194, 206, 209, 216, 217, 220, 323, 329, 341, 485, 487, 500, 502 Guerra Eissmann, Ana María, 21, 28, 34, 52 Gunckel, Hugo, 21 Hajek, John, 108 Haji-Abdolhosseini, Mohammad, 69

Hamp, Eric P., 242, 504 Haoa Cardinali, María Virginia, 22, 575, 592 Haoa Rapahango, Ana Betty, 21 Harlow, Ray, 31, 37, 51, 57, 106, 110, 121, 290, 299, 368, 378, 397, 404, 412, 472, 530, 538 Haspelmath, Martin, 422, 511 Hayes, Bruce, 38 Herd, Jonathon, 538 Hernández Sallés, Arturo, 21 Hey Chávez, Elena, 592 Hey Icka, Alicia, 592 Hey Icka, Sonia, 592 Heyerdahl, Thor, 2 Hohepa, Patrick W., 107, 110, 307, 379, 498 Hooper, Robin, 106, 107, 196, 317, 351, 356, 357, 359, 361, 401, 426, 521 Hopper, Paul J., 385 Horley, Paul, 24 Hotus Chavez, Alberto, 21 Hotus Tuki, Marta Raquel, 21, 592 Hovdhaugen, Even, 38, 44, 75, 77, 79, 96, 107, 121, 152, 242, 304, 368 Howard, Irwin, 7 Hucke Atán, Blanca, 592 Hucke Atán, Leticia, 592 Hunt, Terry L., 3, 4 Hyslop, Catriona, 108 Ibañez, P., 21 Inkelas, Sharon, 61 Jackendoff, Ray, 356 Janeau, V.-F., 328, 333, 504 Jaussen, Mgr. Tepano, 4 Johnston, Marla Speas, 22, 69 Kager, René, 38, 42 Kahn, Jennifer G., 2, 3 Keenan, Edward L., 128, 194, 399, 400, 432, 439, 530, 573 Kellner, Alexandra, 240

Kieviet, Antje, 2, 57 Kieviet, Paulus, 2, 57, 246 Kikusawa, Ritsuko, 140, 379 Kirch, Patrick V., 2, 3 Knoche, Walter, 16 Knorozov, J.V, 4 König, Ekkerhard, 477, 480 Krupa, Viktor, 5 La Pérouse, Count, 4 Labbé, Lilian López, 24 Lambrecht, Knud, 405 Langdon, Robert, 2, 4, 7, 10, 11, 131, 201 Lazard, Gilbert, 77, 79, 152, 175, 176, 187, 189, 190, 246, 276, 290, 304, 343, 355, 386, 412, 462, 478, 498, 552 Lee, Josephine, 226, 333 Lehmann, Christian, 20 Lemaître, Yves, 147 Lemaréchal, Alain, 240 Levinsohn, Stephen H., 182, 184, 188, 226, 353, 388, 409, 458 Levinson, Stephen C., 129 Lewis, M. Paul, 5 Lichtenberk, Frantisek, 281, 293, 329, 439 Liller, William, 21 Lipo, Carl P., 3 Lynch, John, 5, 36, 150, 281, 294, 329 Lyons, Christopher, 238, 457 Macdonald, Catherine, 538 Makihara, Miki, 1, 17, 18, 22, 57, 60, 88, 306, 527 Mangor, Inger Spaabaek, 150 Marck, Jeff, 2, 3, 5, 7–10, 30, 31, 34, 281 Martínez, E., 21, 33 Martinsson-Wallin, Helene, 3 Mason, Te Haumihiata, 241, 242, 451 Massam, Diane, 69, 226, 240, 333, 538 Matras, Yaron, 57 McAdams, Gwen, 22 McCall, Grant, 4

McCarthy, John J., 61 McEnery, Tony, 22 Mellén Blanco, Francisco, 150 Métraux, Alfred, 16, 19, 24, 33, 281, 296n16, 319n11, 591 Meyerhoff, Miriam, 64, 65 Miestamo, Matti, 500 Mieth, Andreas, 3, 4 Milner, G. B., 400 Milroy, Wharehuia, 241, 242, 451 Montt, Sara, 20, 25 Mosel, Ulrike, 21, 22, 44, 57, 75, 77, 79, 96, 107, 121, 152, 242, 304, 368, 380 Moyle, Richard, 9, 447 Moyse-Faurie, Claire, 77, 82, 88, 491 Mulloy, Emily Ross, 22, 131, 142, 294, 296, 301, 303, 307, 308 Mulrooney, Mara A., 4 Munro, Pamela, 21 Mutu, Margaret, 320, 328, 504 Næss, Åshild, 197, 385 Oda, Kenji, 69 Osmond, Meredith, 359 Otsuka, Yuko, 42, 242, 379, 380 Owens, Melanie, 368 Pakarati Tuki, Felipe, 24, 592, 594 Pakomio Ika, Toribio, 592 Palmer, Bill, 130 Parker, William, 44, 51, 91, 96, 133, 152, 194, 241, 278, 290, 328, 341, 451, 478, 498, 552 Paté Tuki, María, 24, 592 Pauly, Stephanie, 21, 594 Pawley, Andrew, 2, 5, 7, 9, 13, 106, 140, 194, 201, 329, 359 Payne, John R., 497, 498 Payne, Thomas, 41, 81, 89, 271, 310, 347, 387, 460, 472, 477, 546, 549 Pearce, Elizabeth, 226

Peltzer, Louise, 77, 79, 152, 175, 176, 187, 189, 190, 246, 276, 290, 304, 343, 355, 386, 412, 462, 478, 498, 552 Philippi, Rodulfo A., 21, 33 Pinochet Carte, Eduardo, 21 Polinsky, Maria, 241, 242, 412, 414, 484 Potsdam, Eric, 412, 414, 484 Prince, Alan S., 61 Pucilowsky, Anna, 378, 427 Pukui, Mary Kawena, 77, 106, 107, 121, 123, 179, 241, 276, 304, 307, 328, 341, 342, 386, 412, 478, 551, 555 Ramirez, J. M., 21 Randall, John E., 21 Rapu, Sergio Alejo, 22, 142, 294, 296, 301, 303, 307, 308 Rauch, M., 21 Reesink, Ger, 422 Rehg, Kenneth, 38 Rensch, Karl H., 500 Reynolds, Bill, 64, 65 Rigo, Bernard, 194, 240 Roggeveen, Jacob, 4 Rolle, Nicholas, 226, 333 Ross, A. S. C., 21, 150 Ross, Malcolm, 5, 329, 348 Roussel, Hippolyte, 19, 21, 25, 33, 39, 42 Rubino, Carl, 21 Rutter, Lawrence Kenji, 7, 31 Sakel, Jeanette, 57 Salas Palomo, Rosa, 22 Salas, Adalberto, 19, 21, 34, 37 Schuhmacher, W. Wilfried, 2, 21, 150, 251 Schwartz, Linda, 439 Sebba, Mark, 368 Seiter, William J., 107, 294 Senft, Gunter, 368 Siemund, Peter, 477, 480 Silva-Corvalán, Carmen, 22, 531 Simons, Gary F., 5

Smith, Jan, 22 Soulik, Tobias, 220 Spriggs, Matthew, 3, 150 Stefan, Vincent H., 2 Stenson, Nancy, 22, 498, 500 Stimson, Frank, 123, 152 Stokes, John F. G., 2 Stolz, Thomas, 22 Tchekhoff, Claude, 77, 351 Teìkitutoua, Ben, 320, 328, 504 Tent, Jan, 57 Tepano Haoa, Clementina, 592 Tepano Haoa, Flora, 592 Tepano Kaituoe, Uka, 25 Tepano Pont, Marcelo (Pai Hiti Uira Rano), 592 Tetahiotupa, Edgar, 128 Thompson, Sandra A., 385 Thomson, William J., 21, 33, 53, 54 Thornton, Agathe, 308 Timberlake, Alan, 321 Topping, Donald M., 150 Tregear, Edgard, 123, 500 Tryon, Darrell, 2, 4, 7, 10, 11, 201, 250 Tsukamoto, Akihisa, 51 Tuki Hey, Benita, 592 Tuki Pakarati, Hilaria, 592 Tuki Paté, María Dolores, 592 Tuki Tepano, Eusebio, 592 Tuki Tepano, Felipe W., 592 Van den Berg, René, 38 Vernaudon, Jacques, 194, 240, 256 Vonen, Arnfinn Muruvik, 77 Waite, Jeffrey, 79, 412, 478 Walworth, Mary, 3, 7, 11, 57, 194 Ward, Jack H., 22, 34 Weber, Nancy, 1, 18–22, 24, 25, 28, 34, 37, 379–381, 383, 384, 387, 394, 395, 397–399, 498, 575, 579, 588, 593, 594

Weber, Robert, 1, 1n2, 18–21, 21n24, 22, 24, 25, 28, 34, 37, 60n41, 190, 239, 241, 265, 313, 317, 319, 327, 329, 345, 368, 478, 558, 575, 579, 588, 593, 594 Weisler, Marshall I., 2, 3 White, Ralph G., 150 Wilkins, David P., 129 Wilmshurst, Janet M., 2, 3 Wilson, Andrew, 22 Wilson, William H., 3, 7, 9, 13, 30, 36, 140, 143, 217, 283, 290, 294, 309, 320, 412 Wittenstein, Andreas, 22, 350 Wurm, Stephen A., 1

Zec, Draga, 61 Zewen, P. François, 249, 333

## **Language index**

Anejom, 150n13

Bislama, 150n13

Chamorro, 150n13

Dutch, 2, 4, 293, 464n15


Hawaiian, 7, 9, 13, 23, 31n4, 34, 38n12, 55n36, 77n4, 106n29, 123n48, 179n36, 179n37, 194n53, 195n57, 219n79, 219n80, 241, 242n12, 244n13, 249n17, 276n35, 290n9, 304n20, 307, 316n4, 328n21, 341n36, 342, 386n20, 472, 475n24, 478n1, 483n3, 487n7, 551n22, 555n26, 592

Hebrew, 184n42

Kapingamarangi, 7, 333n25, 485n6, 551n22

Lau, 508n19 Luangiua, 7, 294n12, 412n50, 551n22 Mangarevan, 7, 7n14, 9, 13, 21, 31n4, 56n38, 123n48, 150n13, 182, 183n41, 328n21, 333n25, 412, 487n7, 498n11, 500, 500n13, 504n17, 512n2, 565n31 Marquesan, 7, 9, 54n34, 123n48, 129n55, 133n62, 179n36, 195n57, 219n80, 240, 249n17, 320n13, 328n21, 333n25, 368n65, 487n7, 504n17, 512n2, 565n31 Muna, 38n12 Māori, 7, 10, 14, 23, 31n4, 37n9, 44n25, 51n32, 57n40, 64n48, 65n49, 75n1, 77n4, 79n8, 86n17, 91n23, 96n25, 106n29, 107n33, 110n35, 123n48, 125n50, 133n62, 133n63, 134n64, 152n16, 194n53, 219n79, 219n80, 220n81, 226, 241, 241n9, 242n12, 249n17, 290n9, 299n18, 307n22, 308n24, 320n13, 328n21, 341n36, 368n65, 378n5, 397, 401n38, 404n43, 409n47, 412, 412n50, 424n64, 426, 427n67, 456n6, 460n10, 472n22, 478n1, 483n3, 487n7, 498, 499, 500n14, 504n17, 512n2, 521n4, 530, 546n19, 552n23, 594

Niuafo'ou, 51n32 Niuean, 69n55, 226, 226n86, 240n8, 294n12

#### Language index

Nukeria, 52n33, 294n12 Nukumanu, 294n12, 551n22 Nukuoro, 7, 8, 220n81, 485n6 Pa'umotu, 9, 10, 56, 123n48, 133n62, 152n16, 180, 182n41, 194n53, 340n35, 512n2 Penrhyn, 30n3 Pileni, 197n59 Pukapuka, 500n14 Rapa, 2n4, 3n7, 10n17, 13, 31n4, 57n40, 194n54 Rarotongan, 13, 31n4, 75n1, 77n4, 180, 193n52, 194n53, 250n17, 483n3, 487n7, 512n2 Rennell, 3n7, 7n13, 33, 34, 36n8, 217n75, 276n36, 294n12 Samoan, 9, 23, 38n16, 57n39, 75n1, 77n4, 96n25, 96n26, 133n63, 134n64, 152n16, 237n4, 242n11, 368n65, 389n22, 400n37, 424n63 Sa'a, 508n19 Sikaiana, 551n22 Spanish, 4, 15–18, 20, 21, 21n24, 22, 24, 25n30, 28n1, 44n24, 57– 59, 73, 88, 104, 125n52, 129, 132, 149, 150n11, 150n13, 172, 172n30, 173, 175, 176, 180, 183, 185n44, 187, 198, 201, 218, 219, 251n19, 268, 306, 376n3, 387, 472, 512, 515, 523, 527, 527n6, 527n7 Tahitian, xiii, 4, 9, 15–18, 21, 23, 30, 31n4, 34, 34n6, 36n8, 37n11, 37n9, 41, 53, 55, 57, 57n40, 60, 72, 77n4, 84n14, 96n25, 106n29, 119, 123n48, 128n54, 133n62, 134n64, 148n9, 150, 150n10, 150n13, 151, 152n14, 152n16, 158, 160, 162, 169, 170n29, 173n31, 175, 176,

179n36, 179n37, 180, 182n41, 186, 187n45, 189n46, 190n47, 193n52, 194n53, 227, 240n8, 246n14, 249n17, 251n19, 253, 256n21, 257n23, 260, 265, 276n35, 276n36, 282n3, 290n9, 307n22, 320n13, 340n35, 343n38, 346n42, 350n45, 355n49, 357n51, 368n65, 386n20, 389n24, 397, 402n39, 412, 412n50, 414, 449, 462n13, 474, 475, 475n24, 475n25, 476, 478n1, 483n3, 487n7, 498, 499, 512, 512n2, 552n23, 554n25, 570n35, 570n36 Takuu, 7, 9n16, 294n12, 333n25, 447n79

Tikopian, 500n14 Tokelauan, 196n58, 317n9, 351, 356n50, 357n51, 361n61


ꞌā (identity), 91, 98, 135, 195, 230, 268– 270, 275, 384, 389, 432, 433 ꞌa (possessive prep.), *see* Possession > o/a distinction ꞌa (possessive prep.), 35, 143, 281, 282 ꞌā (postverbal), 30, 56, 313, 327, 333–337, 478, 501, 555, *see* e (imperfective) > e V ꞌā, *see* ko V ꞌā a (preposition), 35, 128, 216–217 a (proper article), 36, 52, 75, 101–105, 122, 126, 136, 141, 156, 157, 207, 208, 215, 222, 231, 232, 234n1, 237, 245, 276n34, 274–279, 413n52, 441, 478, 480, 483, 485, 531, 538n14 a- 'future', 134, 488 aꞌa 'your', 143 ꞌāꞌaku 'my', 143 ꞌāꞌana 'his/her', 143 ꞌāꞌau 'your', 143 Accessibility, 202, 412, 457, 458 Actor-emphatic construction, 14, 80, 90, 107, 218, 287, 290, 290n9, 305, 324, 370n69, 376n4, 381, 403, 412–416, 462, 470, 483 Adjective, 108–120 in noun phrase, 264–266 used nominally, 116 Adverb, 97, 111, 112n38, 112n39, 113, 114, 117, 134, 135, 139, 154, 159, 170, 173, 177, 178, 178n33, 178n34, 178n35, 179, 181, 265, 266, 275, 323, 348n43, 367, 404, 447, 522, 563 in noun phrase, 266 ꞌafa 'half', 158

Agentivity, 387, 388, 446 aha 'what', 8, 235, 462, 485–487, 492 ꞌāhani 'if only', 404, 569 ꞌahara 'until', 330, 564 ꞌahuru 'ten', 147, 148, 150 ai (deictic), 230, 270–271 ꞌai (deictic), 185–186, 270, 332, 404, 492 ai (postverbal), 14, 313, 366–368 ꞌai (postverbal), 319, *see* he (aspect marker) > he V rō ꞌai ꞌai (preposition), 186, 221–222 ai 'to exist', 15, 326n20, 464, 465, 467, 533, 543, 550 as copula verb, 472–474 ai 'who', 429, 462, 483–485 ꞌana (identity), 91, *see* ꞌā (identity) ꞌana (postverbal), 8, 14, *see* ꞌā (postverbal) ana 'irrealis', 14, 546–550 anake 'all', 163 ananake 'together', 157, 157n21, 159, 163– 164, 435 in comitatives, 435–436 Anaphora, 194, 194n53, 195, 195n57, 196, 197, 199, 200, 202, 205, 206, 206n67, 206n68, 207, 248, 366, 367, 541 Animacy, 140, 206, 209n70, 387, 399, 402, 427, 427n67 ꞌanīrā 'later today', 53, 134 ante 'before', 93, 563 Anteriority, 317, 321, 333, 334 ꞌaŋa- 'recent past', 133, 488 ꞌaŋahē 'when (past)', 133 ꞌaŋahuru 'ten', 55, 147, 148, 152 ꞌaŋanīrā 'earlier today', 133

ꞌaŋapō 'last night', 133 ꞌaŋarīnā 'earlier today', 133 ꞌaŋataiahi 'yesterday', 133 ꞌapa 'part', 158, 265, 313, 340, 342 āpō 'tomorrow', 134 Apposition, 171, 223, 226, 227, 229, 240, 243, 273n33, 271–274, 452n2, 455, 464, 530 ararua 'the two', 156, 157, 157n21, 435 in comitatives, 435–436 ꞌarīnā 'later today', 53, 134 Aspect, 430, *see* e, *see* i (perfective), *see* ko V ꞌā continuous, 323, 325, 327, 344, 370, 558 durative, 292, 316, 318, 539, 571 habitual, 89, 91, 313, 316, 318, 323, 325–327, 362, 370n69, 414, 416, 478, 494, 502, 539, 549n21, 552, 558 imperfective, 323, *see* e (imperfective) perfect, 108, 184, 327, 333–336, 400, 401, 501, 518, 550, 571, *see* ko V ꞌā Aspect marker, 34, 76, 77n5, 79, 91, 109, 110, 113–115, 117, 119, 153, 169, 314–339, 410, 416, 493, 498, 506, 513, 517, 519, 520, 522, 535, 546, 557, 565, 569 Asseverative, *see* rō ꞌata 'more', 118, 119, 265, 313, 342 ꞌātā 'until', 93, 214, 330, 564 Attenuative, 72 atu 'away', 174, 569, *see* Directional aꞌu 'your', 143 au '1sg', 140 Benefactive preposition, 217–218 Borrowing, 4, 27, 28, 30, 33n5, 38n13, 39, 40, 42n23, 44, 44n24, 51, 52, 55, 57, 57n40, 60, 88, 129n57, 150n13, 386n20, 449 calque, 259n26

Case marking, 378–432 in nominalised clauses, 428 object, 379, 390–396 subject, 428 Cataphora, 200 Causative, 83, 84n14, 107, 111, 182, 183n41, 313, 342, 420, 431, 433, 441–449, 479 Central-Eastern Polynesian, 5, 5n12, 7n14, 10n17, 7–13, 30, 30n3, 31, 33, 53, 133n62, 206n67, 290n9, 329, 378, 502n16, 504n17, 508n19, 551n22 Clause adverbial, 557–571 attributive, 171, 451, 462–464 causal, 92, *see* Clause > reason circumstantial, 561, 570–571 classifying, 242, 424n64, 425, 451– 453, 453n3, 457, 461, 462, 473– 475, 485 cohesive, 317, 321, 327, 344, 363, 366, 367, 558n27, 558–560 complement, 516–528 concessive, 569–570 conditional, 190n47, 330, 542, 543, 547, 549, 568–569 equative, equational, 451n1 equative/equational, *see* Clause > identifying existential, 96, 242, 249, 451, 464n14, 464n15, 464–469, 472, 472n22, 495, 497, 508n19, 533 existential-locative, 466, 533 identifying, 222, 267, 451, 453–462, 533 locative, 469, 474, 532 nominal, 80, 93, 96, 115, 169, 181, 224, 236, 262, 388, 409, 429, 453n3, 451–476, 483, 488, 498, 543 nominalised, 516, *see* Verb > nominalised

non-verbal, *see* Clause > nominal numerical, 153, 283n4, 471–472 possessive, 142, 288, 290, 290n9, 407, 466–468, 534 possessive-relative, 486, 495, 537– 539 proprietary, 211, 287, 290n9, 415, 466n18, 469–470, 484 purpose, 93, 314, 542, 551, 565 bare, 565–567 reason, 92, 567–568 relative, 101, 113, 114, 114n41, 168, 169, 198, 199, 202, 206, 207, 229, 263, 270–272, 272n32, 293, 314, 330, 382, 384, 384n14, 409, 410, 420, 425, 426, 429, 434n69, 459–462, 484, 484n4, 485, 486, 495, 506, 511, 519, 537n13, 538n14, 538n15, 528– 541, 562 bare, 14, 114, 263, 315, 339, 363, 370, 520, 539–541 headless, 256, 409, 528 result, 552, 568 small clause, 424n64 temporal, 93, 210, 317, 321, 327, 329–331, 480, 506, 546n19, 547, 552, 557–564 verbal, 375–450 verbless, *see* Clause > nominal Cleft, 223, 226, 409, 409n47, 457n8, 460– 462, 484, 484n4, 485, 530 Code switching, 57 Collocation, 394 Comitative, 157n21, 163, 223, 226, 375, 439n73, 433–440 Comparative, 114, 118, 118n44, 214, 341, 367n62, 470 Complement, 93, *see* Clause > complement, *see* Object Compound, 113, 147, 229, 256, 257n22, 257n23, 259n25, 258–264, 315, 368

Conjunction, 15, 36, 190n47, 268, 329n23, 330, 343n39, 368, 368n66, 404, 506, 511, 512, 512n2, 514, 557, 558, 560, 562–564, 567, 569 Consonant cluster, 59 Consonant inventory, 28 Constituent order, 375–378, 397, 452, 453, 458, 459, *see* Object > preverbal, *see* Subject > preverbal Contiguity, temporal, 329, 331, 478, 569 Cooccurrence restrictions, 43–45 Coordination, 370, 396, 511–516 Counterexpectation, 343, 346, 481 Definiteness, 193, 198, 198n61, 199, 237– 239, 239n7, 453, 457n7, *see* Accessibility Deictic centre, 347–362 Deictic particle, 184–186 Deixis, 46, 135, 196, 199, 202, 205, 207, *see* Locational Demonstrative, 78, 98, 122, 135, 139, 160, 179, 194n53, 197n59, 192–208, 219n79, 230, 237, 324n17, 453, 539, 541n17 determiner, 13, 15, 194, 201–204 postnominal, 98, 129n56, 146, 194– 202, 204, 208, 236, 239, 248, 270, 271, 275, 453, 541 postverbal, 98, 115, 319, 323–328, 331, 367n62, 362–368, 410, 478, 498, 535, 536, 541n17, 546, 552, 558, 558n27, 560, 561 Deontic mode, 548, 549 Determiner, 232–246 demonstrative, *see* Demonstrative > determiner Devoicing, 51, 51n32, 52 Diphthong, 37, 37n11, 37n9, 38n12, 42, 60 Directional, 14, 46, 79, 82, 98, 114, 174, 179, 348n43, 350n44, 351n47, 356n50, 357n52, 347–362,

370n70, 478, 480, 498, 501, 509, 546, 555, 566 Dislocation left, 236, 406, 409, 467 Distal, *see* era Distributive, 69, 70, 145, 152n16, 235, 303, 309 Dual, 140, 140n2 e (agent marker), 14, 15, 107, 383n13, 379–390, 396–403, 494, 525, 531, 543 e (exhortative), 477–479 e (imperfective), 8, 35, 115, 323–328, 362, 502, 507, 535, 558, 560 e V nō ꞌana, 115, 322 e V PVD, 318n10, 319n11, 324–328, 560, 571 e V rō, 318, 324, 345 e V ꞌā, 321, 324, 324n18, 326n20, 325–328, 347, 465, 518, 560, 561 e (numeral particle), 153 e ko (negator), 496, 497, 499, 502–504, 507 ꞌe 'and', 35, 512–515 Eastern Polynesian, 2n6, 3, 3n7, 5, 5n12, 10n17, 10n17, 7–11, 14, 21, 23, 30, 30n3, 31, 33, 42, 54, 55n36, 56, 60, 123n48, 133n62, 136n65, 140, 140n2, 143n6, 180, 182n41, 194n53, 201n65, 206n67, 216n74, 220n81, 237n5, 242, 242n12, 249, 290n9, 320n12, 329, 341n36, 378n5, 379n7, 397, 412n50, 464n14, 483n3, 485n6, 487n7, 508n19, 551n22, 554n25 Elision, 28, 33n5, 34, 51, 52, 55, 56, 58–60, 527n6 ena (medial distance), 135, *see* Demonstrative postnominal, 201 postverbal, 313, 364–365

Epenthesis, 59 Equative, 219 era (distal), *see* Demonstrative postnominal, 196–199, 541 postverbal, 78n6, 313, 365–366 Ergativity, 10, 106n29, 310n26, 378, 378n5, 378n6, 379, 379n7, 380, 384, 386, 389n22, 390n25 Evaluative marker, 342, *see* nō, *see* rō Exclamative, 15, 115–117, 191, 211, 223, 226, 332, 477, 491n8, 491–492 Exclusive, 140 Exhortative, 52, 324, 392, 465, 478n1, 548 Focus, 184, 185, 226, 391n26, 403, 409, 412–414, 460–462, 481, 483, 485 Foot, 38, 41, 42, 45, 59, 61–63, 66, 67 degenerate, 38, 38n15, 41, 42, 42n22, 45, 52, 59, 62, 63, 65n49, 66, 66n50 Fraction, 158 Fricative, 28, 28n1, 29, 58, 59 Full word, 34, 75 Future, 318, 324, 331, 334, 478, 536, 552 Genitive, 282, *see* Possession Geographical names, 101, 102, 122, 130 Glottal plosive, 9, 19, 28–30, 33n5, 34n6, 36n8, 36n8, 33–37, 45, 45n28, 54, 55, 55n35, 55n36, 60, 106n29, 143, 150, 152n14, 180, 183n41, 209, 210, 217, 217n75, 217n76, 222, 270, 276n36, 294n11, 319n11, 328n21, 386n18, 413n52, 427, 483n3, 504n17 hā 'four', 147 haꞌa- (causative), 449 haꞌamata 'to begin', 404n41, 520, 521 haho 'outside', 122

hai (instrumental prep.), 13, 166, 208, 220–221, 234, 235, 247, 277n37, 421–423 haka (causative), 83, 441, *see* Causative haka roŋo 'to listen', 359, 361, 417, 418 case marking, 386 hakaꞌou 'again', 155, 175, 177, 182–183, 266 hānere 'hundred', 148, 149 haŋa (nominaliser), 83, 87, 96, 97 hau 'to exceed', 118 he (aspect marker), 14, 115, 316–321, 404n42, 410, 416, 493, 494, 499, 528, 536 he V rō ꞌai, 186, 315, 319, 345, 346, 357 hē (content question marker), 56, 235, 264, 487–489 he (nominal predicate marker), 14, 88, 96, 98, 155, 219, 226, 240, 241, 241n10, 242–245, 272, 274, 316, 452, 457, 458, 462, 464, 468, 473, 475, 485, 494, 497, 567 in topicalisation, 410–412 Heavy shift, 271 hia 'how many', 147, 489 hia 'not yet', 55, 313, 509, 563 hitu 'seven', 147 hō (dubitative), 192, 481 hoꞌe 'one', 147, 149, 150 hoꞌi 'in fact', 186, 188 hoki (polar questions), 481–482 hoki 'also', 181–182, 512 hoko (numeral particle), 154 hopeꞌa 'last', 119 hora 'summer', 105 hora 'time', 105, 134 hōrou 'to hurry', 522, 523 Hortative, 392, 480, 542 hū (demonstrative), 195 i (accusative marker), 79, 209, 224, 279, 378–381, 390–395, 397, 417–

421, 426, 428, 429, 478, 514, 519, 521, 521n4, 522, 544 ꞌī (deictic), 184–185, 332, 365, 404 i (perfective), 8, 320–323, 363, 507, 535, 558, 560 i (preposition), 124–126, 210–211, 277, 470 agent marker, 108, 211, 426n67, 426–427 ꞌi 'in, at', 35, 92, 209–211, 277, 292, 469 ia '3sg', 140 ia 'not yet', 55, 509 ꞌia 'not yet', *see* hia 'not yet' ia 'then', 189–190, 460 Identifiability, 219, 238, 238n6, 243, 273, 454, 455 iho 'just now', 177, 179 Imperative, 47n29, 52, 56, 107, 190, 191, 277, 314–317, 329, 329n23, 345n41, 392, 394, 396, 426n67, 478n1, 477–480, 497, 500n14, 502, 502n16, 503, 504, 542, 548, 551 Imperfective aspect, *see* e (imperfective) ꞌina (negator), 75, 243, 283, 316n5, 404, 468, 493–501, 503, 507, 545 Inclusive, 140 Inclusory pronoun, 438, *see* Pronoun > inclusory Incorporation, *see* Object incorporation Indirect object, 429–430 Interrogative, 49, 550, *see* Question Intonation, 35, 47n29, 47–51, 480 iŋa (nominaliser), 83, 87, 96, 97 ira (anaphor), 9, 206, 568 iri 'to go up', 131 Irrealis, 89, 89n21, 93, 217n76, 217n77, 329n23, 490, 506, 546–550, *see* ana ꞌite 'to know', 419, 526 Iterative, 61n43, 69, 69n54, 70–72 iva 'nine', 147 Jespersen's Cycle, 500

ka (aspect marker), 115, 329–332, 362, 491, 518, 519, 536, 559, 561, 563, 569 ka V rō, 329, 330, 345, 491, 553n24, 564 ka (imperative marker), 477–479 ka (numeral particle), 153, 155, 329 kā 'each', 159, 172, 173 kai (negator), 496, 497, 499–502, 507 kampō 'countryside', 129, 130, 132 kē 'different', 159, 172, 177 ki (preposition), 118, 119, 213–214, 277, 417–421 ki (preverbal), 56, 550–553, *see* Hortative Kinship term, 103, 231, 274, 281n2, 296– 297, 441 ko (locative prep.), 224 ko (prominence marker), 91, 222–227, 245, 273, 483, 519, 520 in comitatives, 434, 438 in exclamations, 491 in identifying clauses, 453–462 in topicalisation, 407–410 non-topicalised arguments, 424– 425 ko V ꞌā (perfect aspect), 54, 108, 115, 184, 315, 327, 328, 333n26, 336n28, 333–337, 337n29, 363, 500, 501, 518, 520, 536, 559, 571 koā (collective), 231 koe '2sg', 140, 144, 145 koia ko 'with', 93, 236, 436–438, 570 kōnui 'far', 129, 132 koraꞌiti, *see* koroꞌiti kore (negator), 8, 508 koroꞌiti 'slowly, softly', 54, 177 kōrua '2du/pl', 140 koā (collective), *see* kuā ku, 56, *see* ko V ꞌā (perfect aspect) kuā (collective), 13, 15, 104, 229, 231–232, 274

Lengthening, 52, *see* Vowel > lengthen-

ing Lexical word, 31, *see* Full word Limitative, *see* nō Loanword, *see* Borrowing Locational, 75, 76, 95, 101–103, 121n45, 121n46, 123n48, 123n48, 129n56, 129n57, 129n58, 130n60, 121–136, 164, 178n34, 193, 194, 202, 206n67, 206n69, 204–207, 209–211, 213, 215, 217, 221, 224, 229, 247, 288, 442, 469, 487, 563 mā (benefactive prep.), 55, 217–218 mā 'with', 148, 149, 570 maha 'four', 147, 150 mai 'from', 214–215 mai 'hither', 98, *see* Directional mai 'while, before', 555–556 mātou '1pl.excl.', 140 matu 'come on', 480 mau (plural marker), 253 mau 'really', 135, 155, 177, 230, 266, 275 māua '1du.excl.', 140 Medial distance, *see* ena meꞌe rahi 'many', 159, 160, 167–169, 174, 176, 240 meꞌe 'thing', 165, 200, 207, 255, 256, 460, 462–464, 484, 485, *see* meꞌe rahi ꞌi te meꞌe 'because', 567 te meꞌe nō 'however', 570 Metathesis, 13, 53 Metrical structure, 13, 38–43, 59–61, 66 mo (benefactive prep.), 55, 144, 217–218 marking obliques, 431 mo (preverbal), 14, 93, 287, 523–526, 537, 542–546, 551 Months, names of, 104 Mood, 477–492 Mora, 38, 38n15, 39, 41, 42, 45, 45n28, 46, 61, 61n43, 62, 63, 75, 156 muꞌa 'front', 122, 128, 215 muri 'near', 122, 126

na (demonstrative) nā (demonstrative) deictic, 186 nā (demonstrative) deictic, 186 deictic locational, 204 demonstrative determiner, 204 demonstrative pronoun, 207, 208 Nasal, 28, 29, 44, 56, 58, 59 Negation, 14, 116, 241, 256, 292, 313, 314, 404n42, 498n11, 493–509 nei (proximal) deictic locational, 204 demonstrative determiner, 203 demonstrative pronoun, 207, 208 postnominal, 199–200 postverbal, 313, 363–364 nī (demonstrative determiner), 203–204 nō 'just', 56, 98, 114, 135, 155, 174, 230, 275, 313, 327, 347, 478 in noun phrase, 266–268 nōatu 'no matter', 570 Nominal hierarchy, 310–311 Nominalisation, 84n14, 83–99, 117, 178 lexical, 80, 83, 87, 88 syntactic, 80, 83, 87–99 Noun, 101–105 as modifier, 113, 256–266 common, 75, 76, 101, 102, 104, 122, 157, 217, 221, 222, 224, 224n83, 226, 229, 234n1, 237, 243, 271, 274, 278, 294, 310, 408, 413, 414, 425, 441, 453–455 countable, 101 deverbal, 71 non-countable, 101 proper, 75, 76, 101–105, 122, 141, 217, 220, 222, 224, 227, 243, 273, 274, 278, 294, 310, 408, 413, 414, 441, 454, 455, 458, 519 verbal, 88, 89, 92–99, 223, 226, 428, *see* Nominalisation Noun phrase, 229–280

headless, 14, 161, 162, 247n15, 254, 255, 282, 282n3, 285, 286, 288 proper, 274–279 Numeral, 15, 147–158, 471, 472 definite, 13, 105, 156, 157, 158n21, 163 in noun phrase, 245–249 ordinal, 148, 155, 156, 264 Numeral, cardinal, 147 nuꞌu 'people', 253 ŋā (plural marker), 14, 99, 104, 229, 232, 236, 237, 249–253, 261, 441 ŋaroꞌa 'to perceive', 53, 359, 361 case marking, 386, 393 nō 'just', 342–345 ꞌō (asseverative), 155, 191–192, 481 o (possessive prep.), 35, 124–126, 209, 281, 282, 466 ꞌō ira 'before', 330, 563 ꞌo 'because of', 35, 92, 212–213, 568 ꞌo 'lest', 56, 345, 525, 554 ꞌo 'or', 515–516 Object, 375–381, 390–396 demotion, 421–423 incorporation, 262–263, 393n28, 426, 565 nominalised verb, 94 possessive, 428 preverbal, 367, 376n4, 406, 413 Oblique, 246, 429–432, 444 oho 'to go', 370, 371, 523 ono 'six', 147 ōꞌoku 'my', 143 ōꞌona 'his/her', 143 ōꞌou 'your', 143 Orthography, 19–20, 28, 31, 34, 34n6, 34n7, 36, 60, 258n24, 427, 591 oti 'to finish', 371, 404n41, 521, 522 oꞌu 'your', 143 pa 'like', 219, *see* pē pae 'five', 147, 150 paꞌi 'in fact', 186–187

Particle postnuclear, 31, 75, 139, 499 prenuclear, 31, 34, 36, 41, 75, 499 Partitive, 142, 254, 285–286 Passive, 107, 375, 378, 393, 396–403, 420, 428, 429, 522, 523, *see* Pseudopassive Past, 320, 334, 500, 552 paurō 'every', 159, 160, 162, 164, 176 pē 'like', 120, 205, 208, 215n73, 219–220, 234, 240, 243, 244, 247, 488 pē nei, 205, 524 pe 'towards', 128, 215–217 pēaha 'perhaps', 8 Perfective aspect, 30, 500, *see* i (perfective) pero 'but', 515 Personal article, *see* a (proper article) Phoneme inventory, 27 piti 'two', 147, 150 Plosive, 28, 29, 44, 58, *see* Glottal plosive Pluperfect, 336n28 Plural adjective, 111 noun, 249–254 verb, 68, 70 Polynesian languages, 5–7 Possession, 9, 81–83, 141, 202, 211, 217, 270, 281n1, 282n3, 294n14, 281–311, 431, 466–468, 468n20, 469, 484, 528, 530, 533, 534, 534n12, 537, 538, 538n15, 540, 541, *see* ta (possessive prep.), *see* to (possessive prep.) o/a distinction, 13, 141, 143, 144, 217, 218, 294–311, 484, 538 Posteriority, 215, 330, 331, 414, 519, 536 Preposition, 208–227 Pro-form, 75, *see* ira (anaphor), *see* Pronoun Progressive, 89, 323, 323n15, 327, 344, 362 Prominence, 226, 407, 407n45, 459, *see*

ko (prominence marker) Pronoun, 13 benefactive, 75, 140, 144, 217, 281, 281n1 inclusory, 434, 434n69, 435, 439n71, 439n73, 438–440 personal, 103, 139, 140, 140n2, 141, 143, 206, 206n68, 207, 277, 278, 283, 432, 454, 483 as comitative marker, 434 possessive, 13, 98, 139–141, 141n4, 142, 144, 146, 196, 232, 233, 237, 276, 279, 281, 281n1, 281n2, 282–285, 287–288, 290, 290n9, 294, 294n11, 294n13, 295, 303, 310, 412, 468n20 t-class, 141–144, 217n76, 229, 244, 254, 274, 282, 283n4, 284, 285, 288, 468, 471 Ø-class, 9, 141, 143, 229, 230, 282–284, 287, 290, 290n9, 413, 415, 416n53, 468, 468n20, 469 Proper article, 103, *see* a (proper article) Proper noun, 274, *see* Noun > proper Prosodic phrase, 29, 34–36, 45, 46, 51, 52 Prosodic word, 38 Proto Central-Eastern Polynesian, *see* Central-Eastern Polynesian Proto Eastern Polynesian, 2, *see* Eastern Polynesian Proto-Polynesian, 2n6, 8–10, 13, 21, 30, 30n3, 31, 33, 34, 36n8, 42, 53–55, 55n35, 55n36, 56, 73, 123n48, 130, 133n62, 133n63, 134, 140n2, 148n9, 154n17, 179, 180, 193n52, 194n53, 209, 213, 217, 217n75, 219n79, 220n81, 226, 237n4, 242, 260, 276, 276n36, 277n37, 281n2, 328n21, 329, 329n23, 333, 339n33, 342, 348, 348n43, 350n46, 379n7, 386n18, 409, 412n50, 441, 483n3, 485n6, 487n7, 489,

500n14, 504n17, 550, 554n25 Prototype, 80, 81, 81n10, 82, 83, 105, 108n34, 109–111, 113, 114, 118, 182n40, 229, 326, 336, 385 Proximal, 364, *see* nei Pseudopassive, 401, 402 puē 'can', 503, 527 pura 'mere', 159, 173 Quantifier, 15, 75n2, 82, 99, 101, 136, 139, 159n22, 157–176, 182, 230, 236, 240, 245, 250, 254, 264, 266, 283, 343, 357, 381, 506, 539– 541 Question, 241, 292, 338, 404, 480–490 content, 190, 192, 227, 429, 482–490 dependent, 490 polar, 49, 480–482, 490, 550 rā (demonstrative) deictic locational, 204, 205 demonstrative determiner, 202– 203 demonstrative pronoun, 207, 208 rā (intensifier), 190–191 raꞌe 'first', 135, 155, 563 rahi 'much/many', 159, 169–172, 264, *see* meꞌe rahi Raising, *see* Subject raising, *see* Verb raising raro 'below', 122, 127 rau 'hundred', 164 rāua '3du/pl', 140 rauhuru 'diverse', 159, 160, 164–165, 174, 240 rava 'usually', 14, 53, 313, 339–340 ravaꞌa 'to obtain', 53, 54, *see* rovaꞌa Realis, 89, 98, 217n77, 345n41, 412, 549, 549n21 Reciprocal, 70, 432–433 Reduplication, 27, 37–39, 39n18, 42, 44, 51, 61n42, 61n43, 61n44, 66n50, 68n53, 69n54, 69n55, 60–73, 110, 110n36, 156, 163

Referentiality, 26, 226, 233, 235, 236, 239, 240, 240n8, 241n9, 242, 253, 274, 279, 398, 410, 451– 453, 497, 538 Reflexive, 179, 269, 432–433, 445, 446 rima 'five', 147 riro 'become', 474–475 rō (emphatic marker), 56, 114, 155, 313, 345–347, 465, 478, 546, 554 roto 'inside', 122 rovaꞌa 'to obtain', 54, 386, 526 case marking, 386, 393 object incorporation, 426 rua 'two; other', 147, 156 ruŋa 'above', 122, 127, 128 rō (emphatic marker), *see* he (aspect marker) > he V rō ꞌai Salience, 395, 395n32, 396, 416 Serial verb, 14, 314, 348n43, 368–372, 383n12, 522 Simultaneity, 184, 329, 330, 518, 519, 559–561 Spanish influence, 15, 17–18, 57, 88, 149, 172, 172n30, 527 Specific reference, 236–239, 239n7, 242, 249, 272, 391, 410, 457 non-specific reference, 220, 238, 242, 248, 421, 421n61 Split predicate, 459–460 Stranding, 426, 459, 460 Stress, 27, 31, 32, 37, 37n11, 38, 38n15, 43, 45–47, 49, 51, 52, 59, 60, 65n49, 204, 409n47, 483 Subject, 375–381 demotion, 416 possessive, 416, 428, 525, 543, 544 preverbal, 337, 337n29, 339, 367, 382, 384, 388, 404n42, 403– 406, 410, 424, 494, 499, 531, 538n14, 545, 568, 569, *see* Topicalisation raising, 392, 398, 399, 404n41, 498, 520, 521, 523, 538

Subject, passim, *see* Case marking > subject demotion Superlative, 119, 120, 341 Syllable, 27, 32, 37–38, 38n13, 38n15, 39n18, 41, 41n19, 42, 44, 45, 45n28, 46–50, 52, 53, 56, 59, 61, 61n45, 63, 65, 65n49, 67, 72, 96n26, 163, 193n52, 217, 268, 333n25, 480 taꞌa 'your', 142, 146 tāꞌaku 'my', 142 tāꞌana 'his/her', 142 taꞌatoꞌa 'all', 159–162, 164, 176, 240 tāꞌau 'your', 142 taꞌe (negator), 98, 171, 313, 342, 404n42, 415, 463, 497, 504–508, 545, 555, 567 tahaŋa 'simply', 177 tahatai 'seashore', 129, 132 tahi 'all', 159, 173, 174, 176, 177 tahi 'one', 147, 239, 248–249 Tahitian influence, 4, 15–16, 30, 34, 34n6, 37n11, 53, 60, 84n14, 150, 158, 160, 162, 173n31, 175, 176, 186n45, 253, 257n23, 260, 340n35, 386n20, 402n39, 449, 474, 475, 512, 554n25 tai 'seaside', 129–132 takeꞌa 'to see', 360 case marking, 386, 393 takoꞌa 'also', 53, 177, 180–181, 266, 512 tātou '1pl.incl.', 140 tau (demonstrative), 194, *see* tū taꞌu 'your', 142, 146 tāua '1du.incl.', 140 tāꞌue 'by chance', 177 taꞌutini 'thousand', 55, 149, 150 te (article), 88, 98, 142, 143, 166, 196, 198, 199, 232–241, 244, 276, 278, 441 tētahi 'some', 159, 160, 165–167, 174, 176 Thematicity, 226, 405, 408, 409, 416, 424, 425, 434, 458n9

tiene que 'must', 527 tikeꞌa 'to see', 53, 54, 359, *see* takeꞌa to (possessive prep.), 142, 466, 468 tomo 'to go ashore', 131 tōꞌoku 'my', 142 tōꞌona 'his/her', 142 tōꞌou 'your', 142 Topic, topicality, 226, 399, 400, 402, 405, 406, 413n52, 414, 420, 423, 432, 452, 454 Topic-comment construction, 406–407, 467 Topicalisation, 103, 223, 226, 242n12, 244, 409n47, 410n48, 407–412, 424, 545 toru 'three', 147 tou (demonstrative), 194, *see* tū toꞌu 'your', 142 tū (demonstrative determiner), 56, 194– 199, 202, 203, 284 tuꞌa 'back', 122, 127, 128 tūmaꞌa 'more or less', 148 tupuꞌaki 'near', 122 turu 'to go down', 131 tuꞌu 'your', 142 uꞌi 'to look', 359, 360, 417 case marking, 386 ꞌuta 'inland', 129–132 uꞌu 'your', 143 vāeŋa 'middle', 122 vara 'usually', *see* rava varaꞌa 'to obtain', 53, *see* rovaꞌa varu, 152 vaꞌu 'eight', 147 Verb, 105–108 active, 107–109, 115, 333, 360, 427, 427n67, 559n28 as modifier, 113 aspectual verb, 89, 94, 385, 520, 521 cognitive, 94, 106, 335, 418, 523, 524 copula, 15, 242, 441, 472–475 denominal, 71, 99

emotion, 418 intransitive, 68, 105, 106, 106n31, 107, 108, 304n20, 369, 370, 375n1, 376, 376n2, 377, 378, 380, 382, 383, 383n12, 383n13, 384, 384n14, 384n15, 385, 390, 396, 401, 413, 415, 422, 426, 427n67, 441, 442, 444, 445, 472 middle, 106, 106n29, 109, 214, 378– 380, 392, 417–421, 430 modal, 15, 526, 527, 542 motion, 348, 349, 352, 353, 357, 358, 360n59, 361, 368n65, 371, 371n71, 385, 386, 388, 523, 565 naming, 227, 424 nominalised, 80, 83–88, 91–95, 97– 99, 210, 212, 291, 295, 304, 305, 370, 375n1, 392, 394, 394n30, 396, 414–416, 428, 505, 521, 522, 526, 554n25, 563, 567, 570 patientive, 106, 108, 109, 386n20, 426 perception, 95, 106, 330, 335, 357, 359, 385–388, 390, 417, 417n56, 425, 425n65, 517–520 case marking, 386 raising, 540, 541 speech, 84, 95, 348, 353, 354, 357– 360, 362, 385, 386, 389, 419, 490, 524, 542 stative, 107, 107n33, 108, 109, 110n35, 118, 318, 323, 327, 333, 334, 336, 337n29, 356, 387, 501, 503, 559 transitive, 68, 70, 106, 106n29, 106n31, 107, 304n20, 336, 341, 355, 369, 375n1, 376, 376n2, 377, 378, 378n5, 379–383, 383n12, 383n13, 384, 384n14, 384n15, 385, 386, 386n20, 390, 395n33, 396, 398, 413, 415, 417, 420, 423, 428, 431, 441, 444, 473, 543

volition, 335, 462, 525, 526 weather, 82, 85 Verb phrase, 313–373 Vocative, 145, 229, 232, 277, 440–441 Vowel lengthening, 42n22, 51, 52, 59, 61, 61n43, 64n48, 65, 217, 443 Vowel inventory, 31 Vowel length, 31, 33, 38, 42n21, 45n28, 46, 60, 66, 96n26 vāeŋa 'middle', 127, 128

with, 433–440

# Did you like this book?

This book was brought to you for free

Please help us in providing free access to linguistic research worldwide. Visit http://www.langsci-press.org/donate to provide financial support or register as a community proofreader or typesetter at http://www.langsci-press.org/register.

## A grammar of Rapa Nui

This book is a comprehensive description of the grammar of Rapa Nui, the Polynesian language spoken on Easter Island. After an introductory chapter, the grammar deals with phonology, word classes, the noun phrase, possession, the verb phrase, verbal and nonverbal clauses, mood and negation, and clause combinations.

The phonology of Rapa Nui reveals certain issues of typological interest, such as the existence of strict conditions on the phonological shape of words, wordfinal devoicing, and reduplication patterns motivated by metrical constraints.

For Polynesian languages, the distinction between nouns and verbs in the lexicon has often been denied; in this grammar it is argued that this distinction is needed for Rapa Nui. Rapa Nui has sometimes been characterised as an ergative language; this grammar shows that it is unambiguously accusative. Subject and object marking depend on an interplay of syntactic, semantic and pragmatic factors.

Other distinctive features of the language include the existence of a 'neutral' aspect marker, a serial verb construction, the emergence of copula verbs, a possessiverelative construction, and a tendency to maximise the use of the nominal domain. Rapa Nui's relationship to the other Polynesian languages is a recurring theme in this grammar; the relationship to Tahitian (which has profoundly influenced Rapa Nui) especially deserves attention.

The grammar is supplemented with a number of interlinear texts, two maps and a number of indices.