# A grammar of Palula

Henrik Liljegren

Studies in Diversity Linguistics 8

### 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 Palula

Henrik Liljegren

Henrik Liljegren. 2016. *A grammar of Palula* (Studies in Diversity Linguistics 8). Berlin: Language Science Press.

This title can be downloaded at: http://langsci-press.org/catalog/book/82 © 2016, Henrik Liljegren Published under the Creative Commons Attribution 4.0 Licence (CC BY 4.0): http://creativecommons.org/licenses/by/4.0/ ISBN: 978-3-946234-31-9 (Digital) 978-3-946234-32-6 (Hardcover) 978-3-946234-33-3 (Softcover) 978-1523743-72-8 (Softcover US) ISSN: 2363-5568 DOI:10.17169/langsci.b82.85

Cover and concept of design: Ulrike Harbort Typesetting: Birgit Jänen, Henrik Liljegren, Sebastian Nordhoff Illustration: Henrik Liljegren, Sebastian Nordhoff, Ljuba Veselinova Proofreading: Mario Bisiada, Jonathan Brindle, Christian Döhler, Tom Gardner, Eitan Grossman, Andreas Hölzl, Michelle Natolo, Alexis Palmer, Mathias Schenner, Alec Shaw, Aviva Shimelman, Charlotte van Tongeren Fonts: Linux Libertine, Arimo, DejaVu Sans Mono, ScheherazadeRegOT\_Jazm 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.

To the Palula people









## **Acknowledgments**

This grammar of Palula is essentially a revised version of my doctoral thesis, successfully defended at Stockholm University in June 2008, at the time titled *Towards a grammatical description of Palula*. Since then, the text has been subject to a series of necessary corrections, structural as well as content-related changes and a few but substantial additions, and it is my sincere hope that the end result is an improved product as far as user-friendliness, comprehensiveness and accuracy are concerned.

A large number of people and circumstances have played a vital role in the completion of this work.

First of all, none of this would ever have been possible had it not been for the many open hearts and homes of the Palula community that my family and I have experienced since the first contacts were established in the summer of 1998. Thank you all for sharing your lives, your knowledge and your stories with us. Although hospitality is something self-evident and a matter of honour in Chitral, I would nevertheless like to thank those who so generously have provided shelter, sustenance and good company. Especially (but not exclusively) my thanks go to Muhammad Atiqullah (Biori), Hanifullah (Drosh), Hayatuddin (Kalkatak), Khitabuddin (Ashret), (late) Sahib Shah (Biori), (late) Said Habib (Ashret) and Shaukat Ali (Kalkatak). Thanks also to the local leadership in the Palula-speaking localities and to the Palula Community Welfare Organization (formerly known as Anjuman-e-taraqqi-e-Palula) for your cooperation and encouragement.

For his help in getting to know Chitral, and the Palula community in particular, I must mention my friend Fakhruddin Akhunzada, who, more than anyone else, helped me establish many of those contacts and connections that turned out to be the longest lasting and most fruitful in my work and continues to provide me with local knowledge, advice and insights. My main Palula language consultants and collaborators are duly acknowledged in the introductory chapter of this work, and informants and recorded speakers are listed in the references at the end, the sole reason why many of these key people are not mentioned here.

Many thanks, to my main PhD supervisor, Östen Dahl, who, over an extended period of time, read and reread all the different parts and versions of my text

### Acknowledgments

up to the time of my defence, constantly commenting and challenging my analysis, while also suggesting improvements, pointing out crucial typological correlations and linguistic references, and widening my own awareness on a range of linguistic topics. I am also deeply indebted to my second supervisor, Joan Baart, who, besides carefully reading and insightfully commenting on my text, served more than anyone else as an enormously inspiring and competent mentor in the investigation and analysis of a largely undocumented language, especially during our overlapping time periods in Pakistan.

I am very grateful to Ruth Laila Schmidt, who encouraged my Palula research from the time we first met, and subsequently read many of my drafts, often offering interesting comparisons with other Shina varieties and ancient forms of Indo-Aryan. Without any formal agreement, she provided qualified mentorship and stirred an interest in the diachronic aspects of my Palula studies.

Thanks go to a number of other people who at different (both pre- and postdissertation) stages took the time to read drafts of the work, either in its entirety or parts of it, and offered crucial input on structure, style and content: Lamont Antieau, Kimberly Caton, Henk Courtz, Marie Crandall, Peter Hook, Andreas Jäger, Amy Kennemur, Maria Koptjevskaja-Tamm, Johanna Liljegren, Eva Lindström, Stephen Marlett and (late) Carla Radloff. Also a special thanks to Ljuba Veselinova, who helped me produce one of the maps included in this work, and to Emil Perder for the many interesting and rewarding conversations we have had about the area and its languages at the centre of our mutual interest. Correspondence and conversations at different points in time with a number of other people engaged in linguistic fieldwork in the region have also been inspirational and contributed to spurring a further interest in areal-linguistic perspectives on my Palula research. Here I would like to mention Elena Bashir, Jan Heegård, Tariq Rahman, Khawaja Rehman, Ronald Trail, Matthias Weinreich and Claus Peter Zoller in particular.

My chief language consultant, Naseem Haider, is acknowledged in the introductory chapter, but I want to mention him here too. Apart from carrying out many interviews and recordings, he made a first transcription of most of the Palula texts that my analysis is based on, which we later worked on together and gradually refined and discussed at great length and in extensive detail. Without his ceaseless patience in providing answers and explanations, and in listening to and helping to thresh out my own hypotheses, the annotation would still have been full of question marks, and my resulting analysis would have been at a much less advanced stage than it is now. Thank you, Naseem, for the privilege of working with you!

Moving to Pakistan and settling in Peshawar would have been much harder and not as enjoyable and fascinating as it turned out to be, if it had not been for all the many new friends and people who gave me and my family a warm welcome and continued to make us feel at home during those years. At the Forum for Language Initiatives (formely Frontier Language Institute) in Peshawar, a special thanks go to Wayne and Valerie Lunsford (and their sons, Sean and Jordan), without whose unceasing friendship and support we would never have made it. Thanks also to Fakhruddin Akhunzada (again) and Muhammad Zaman Sagar, who faithfully included us in their lives and in the lives of their families and communities. I am also indebted to all the other FLI staff members and all of those language activists and local scholars in FLI's network, not specifically named here, who have enriched my life and many times made sense of the seemingly (and actually) contradictory cultures and sentiments that Pakistan comprises.

Furthermore, I am extremely grateful to the staff at Language Science Press. A special thanks to Martin Haspelmath for encouragement and for offering me the opportunity to publish my work at LSP. I also want to thank Birgit Jänen for the tremendous work with LaTeX conversion and typesetting, and Sebastian Nordhoff for managing the publication process and offering quick answers and relevant solutions to various problems. I also want to thank the four anonymous reviewers who offered many valuable and insightful comments and contributed suggestions that greatly improved the final version of the work. Thanks also to the proofreaders who were engaged at the final stages of the production process.

During my PhD periods spent in Stockholm, the Department of Linguistics at Stockholm University, Elisabeth & Herman Rhodin's Foundation and P.A. Siljeström's Foundation contributed by providing me part-time employment and doctoral student scholarships. My field work in Pakistan was conducted while holding a post as a language development consultant (from 2003 under the auspices of FLI) financially supported by Sida (Swedish International Development Cooperation Agency), through its frame organisation PMU, 1998—2000, 2003— 2006, and 2008—2010. For the period when most of the revisions were implemented, I was supported by the Swedish Research Council as part of the ongoing research project *Language contact and relatedness in the Hindukush region* (421-2014-631).

Finally, my deepest love and appreciation to Maarit, Johanna and Jonathan, my own family, for your love, participation and support throughout all of our adventures.

Henrik Liljegren *Stockholm, February 2016*

### **General abbreviations**


### **Grammatical abbreviations**

Abbreviations are listed in upper-case characters. They are used also in various formats: all capitals, small capitals and lower case, for different purposes.





### **Abbreviations of example sources**

### **Palula data references (Ashret dialect)**



### **Palula data references (Biori dialect)**


## **1 Introduction**

Palula [phl] is an Indo-Aryan language belonging to a group of speech varieties subsumed under the heading Shina. It is spoken by approximately 10,000 people in the Chitral Valley in northern Pakistan's mountain region. This study is the first attempt at a systematic description of the grammar of a language that, until recently, has been unwritten and largely undocumented. It is based on first hand data, collected and analysed in close collaboration with Palula-speaking language consultants (specially acknowledged in §1.6), mainly during the period 1998–2008. The two main dialects, both represented in this work, correspond to the primary distribution of the speakers into the two side valleys Ashret and Biori, with the geographical coordinates 35° 26′ 6″ N 71° 44′ 41″ E and 35° 28′ 24″ N 71° 48′ 2″ E, respectively.<sup>1</sup>

### **1.1 Language name**

*Palula* and *Phalura* are the names most commonly used in linguistic and other literature with reference to this speech variety, the former almost exclusively used in more recent publications (Cacopardo & Cacopardo 2001; Bashir 2003; Heegård Petersen 2006; Schmidt & Kohistani 2008; Perder 2013; Baart 2014). In the earliest reference in print to this particular ethnolinguistic group (Biddulph 1986 [1880]: 64), the ethnonym *Dangariké* is used, but only a few years later, a British officer named Gurdon, stationed in Chitral between 1895 and 1902, mentions that people in some villages in southern Chitral speak a language called *Palola*, or *Dangarikwar*<sup>2</sup> (Morgenstierne 1941).

Despite mainly using the name *Palula* in his earlier references (e.g., 1932: 54– 59), the Norwegian linguist Georg Morgenstierne who conducted field research in the region in the 1920s (Morgenstierne 1992) was the one who introduced the form *Phalûṛa* (*Phalura* without diacritics), which is how the language was referred to in what was to remain the sole source of scholarly knowledge about

<sup>1</sup> The coordinates provided are those corresponding to the most densely populated section of each of the two main settlements, Atshareet-xaás and Bhiúuṛi.

<sup>2</sup> The derivational suffix *-war* denotes 'language' in Khowar, the *lingua franca* of Chitral.

### 1 Introduction

this language for decades (Morgenstierne 1941) and set a standard followed by others (Buddruss 1967; Èdel'man 1983; Masica 1991; Decker 1992b). It seems likely, however, that this form was misconstrued, as there is no other primary source supporting it and no recollection of it in the present-day community of Palula speakers. The name *Palula* was brought to the fore again by Richard Strand (1997/2015), based on his own field studies (of the Ashret dialect) in the 1980s and was transcribed as *palôlâ'*, although he himself primarily refers to the language as *açharêtâ'* (appr. 'the speech of Ashret').

*Palula* corresponds to the self-designation *paalúula* 'Palula people' (sg. *paalúulu* 'a male person of the Palula') and *paaluulaá* 'the language of the Palula', both represented orthographically as "Palula". There is, however, no complete internal consensus on what name to use for the language, one reason being that many speakers prefer to identify their language as well as themselves with a geographical location (a village or a valley), so that, for instance, the people of the Ashret Valley more readily refer to their tongue as *atshareetaá* (in my transcription corresponding to Strand's açharêtâ'), as observed already by Morgenstierne (1941), or the even more generic *asíi čoolaá* 'our speech'. While the name Palula is indeed recognised by people with some historical awareness and a certain educational level<sup>3</sup> in both of the main geographical locations, it is also accepted at large by the speakers in the Biori Valley as a reference to their own speech as well as that of Ashret, whereas the common man in the Ashret Valley tends to regard Palula as designating the (slightly different) speech variety of the Biori Valley, often implying an additional "tribal"<sup>4</sup> and genealogical distinction.

There are no suggestions as to the origin of the name Palula in the older literature, apart from Morgenstierne's (1941: 53) comment that the name of the language (in its assumedly misapprehended form) is formally identical to the plural of 'grain'. Recently however, the Italian anthropologist Alberto Cacopardo (2001: 91), has suggested a historical link between the fifth- to eighth-century Paṭola, or Palola, dynasty with its heart-land in the upper Indus Valley and the ancestors of the Palula speakers, who according to local oral history, migrated from the very same area. The similarity between the names is, as Schmidt & Kohistani (2008: 3) point out, too striking to be merely coincidental.

<sup>3</sup> This was reflected in the choice of a name for the local language society formed in 2003 by representatives from all the major locations: *Anjuman-e-Taraqqi-e-Palula*, approximately meaning 'The Society for the Promotion of Palula'.

<sup>4</sup> While "tribal" in today's Western context is a marked term in comparison to "ethnic", *tribal* in the Pakistani context is on the contrary the politically correct choice as opposed to ethnic, the latter being avoided as carrying a connotation of separatism; compare for instance with the fully accepted designations *Federally Administered Tribal Areas* and *Home and Tribal Affairs Department*.

### 1.2 The general setting

In popular use by most outsiders – mainly speakers of Khowar (the linguistic majority of Chitral District) – is the ethnonym *Dangarik*, alternatively *Dangeri* or *Dangarikwar*, for the language itself. That use, however, is frowned upon by many Palula speakers, particularly in the Ashret Valley. Following Biddulph (1986 [1880]: 113), the term Dangariké was in the past applied to Shina-speaking people in general, including the Chitrali Dangariké (i.e., the inhabitants of Ashret and Biori), carrying the connotation "cow-people". The same IA source lexeme *ḍanɡara-* (Turner 1966: 5526, 5524) has given rise to derogatory meanings such as 'defective, bad, unpleasant' (see Cacopardo & Cacopardo 2001: 81 for a discussion on this). In any case, the term has often been, and is still, interpreted as derogatory by some (Decker 1992b: 69; 1996: 160, and own observations), and should therefore be avoided. A positive explanation of the same word has, on the other hand, been offered to me by the local historian Muhammad Atiqullah from Biori as merely pointing to the geographical origin of the speech community in or near to Tangir in the Indus Valley east of Chitral. It is of course not entirely unlikely that it is exactly the use of wordplay itself that has given rise to the interpretational ambiguity, i.e., a designation intended to refer to geography being similar to a derogatory word and is subsequently applied to a community already socially stigmatised or markedly different.

Finally, some of the more educated speakers of Palula would occasionally identify their language as a dialect of Shina (Decker 1992b: 82 and my own observations), although there is no evidence of any regular modern-day interaction between Palula speakers in Chitral and speakers of any other Shina variety.

### **1.2 The general setting**

### **1.2.1 Where and by whom the language is spoken**

The geographical home of Palula is the southern part of Chitral District, in Pakistan's Khyber Pakhtunkhwa Province, displayed in Map 1.1. All of the Palulaspeaking localities are situated within a stretch of only 40 kilometres on the eastern side of the Kunar River (also called Chitral River), which crosses into neighbouring Afghanistan a few kilometres down-river from the area where the language is spoken. The two main settlements of Palula speakers are the two side valleys Ashret and Biori. The main road, leading from lowland Pakistan to Chitral through the 10,000 ft. Lowari Pass, goes right through the Ashret Valley.

Although formally regarded as one large village, Ashret (Atshareét) is more accurately described as a rather long stretch of more or less interconnected smaller

### 1 Introduction

villages or hamlets (coming from the Lowari Pass going downstream towards the main valley): Buzeeghaá, Bharaaḍám, Patoodhám, Lookúṛi, Kaṇeeghaá, Šaṛadeéš, Kooḍghaá, Ghróom (or Atshareet-xaás 'Ashret proper'), Feerimaá and Looṭanghaá. Apart from Buzeeghaá, which has a majority of Pashto speakers, the population of all of these hamlets is solidly Palula-speaking. The main settlement is situated at an altitude of 1,500 metres above sea level.

Four kilometres south of the bazaar town Drosh, the Biori Valley meets the main valley along the Kunar River. There are three distinct villages in this narrow side valley, which together with the Ashret Valley is to be considered the heartland of the Palula-speaking community, all of them solidly Palula-speaking: Mingál (or Lower Biori), Dhamareét (or Middle Biori), and Bhiúuṛi (or Upper Biori/Biori proper), the latter being the largest of the three, situated at an altitude of 1,800 metres.

Apart from those two main settlements, there are a few other non-adjacent villages where Palula is also spoken, or at least has been spoken in the recent past. The northernmost of these villages is Purigal (Púri in Palula), which is situated about twenty kilometres north of Drosh, about seven kilometres into the Shishi Koh Valley. Once an exclusively Palula-speaking village, the language is now competing, more than likely on the losing side, with Khowar. The next location with a substantial number of Palula speakers – though the number is markedly in decline – is Kalkatak (Galaṭáak), a village situated in the fertile main valley, six kilometres by the main road south of Drosh, near the mouth of the Biori Valley. In this village, Palula competes primarily with Khowar, and to a considerably lesser extent with Pashto, and the usage of Palula is dramatically decreasing in the younger generation. This may in some ways be the repetition of the scenario in this village about a century ago, when Palula obviously won ground over the then dwindling Kalasha language (Decker 1996: 165; Cacopardo & Cacopardo 2001: 95). The small village Serdur (Sawdár), situated right at the mouth of Biori Valley, where the majority is reported to speak Palula, is for official purposes treated as a part of Kalkatak, but it is on the local level mostly considered a separate unit (Fakhruddin Akhunzada, pc). On the arid mountainside a couple of kilometres east of Drosh is a small and remote village called Ghos (Ghoós), which apparently was Palula-speaking in the recent past (Decker 1992b: 75, 84), but, according to an informal survey carried out in 2005 by my local language consultants, it has switched entirely to Khowar. Another village that deserves to be mentioned is Badrugal (Baaḍurghaá), which is located in the main valley between Ashret and Kalkatak, and according to local sources is inhabited by a substantial number of people able to communicate in Palula. The first language of

Map 1.1: The Palula-speaking area in southern Chitral

### 1 Introduction

Photo 1.1: Cluster of houses in Mingal, Biori Valley, 2002 (Dietmar Polster)

most (male) villagers is given as Shekhani or Shekhwar, a Nuristani language, but many have reportedly acquired Palula in frequent contact (especially through intermarriage) with the nearby Palula settlements.

Outside of what is in a strict sense part of the Palula-speaking area in Chitral, one other place that should be mentioned is a village called Gumandand (Gumaaṇḍáṇḍ) in Dir Kohistan (in the neighbouring Upper Dir District to the east). Previously, Morgenstierne (1941: 9) mentioned the tribal connection between this village and the Palula of Chitral, and a number of people in Ashret have similarly pointed out that they indeed have relatives there, as a result of a past migration from Ashret. However, I have yet to come across any solid evidence for the same or a similar language being spoken there. It is also unclear whether there is a linguistic or tribal connection between this village and another village in Dir Kohistan, namely Kalkot, where a closely-related speech variety has been recorded (see §1.3.1 for a discussion on the relationship between the Palula of Chitral and the speech of Kalkot).

As far as the total number of Palula speakers is concerned, we can only provide a rough estimate at best (shown in Table 1.1), due partly to the lack of any reliable census directly mapped to language use, and partly to the changeable and

1.2 The general setting


Table 1.1: Estimated number of Palula speakers in each location

multilingual situation of the region. Based on a combination of population data provided by a local health network in Ashret, the results of surveys carried out by the local ethnographer and sociologist Fakhruddin Akhunzada in Biori and Kalkatak, interviews with local respondents in Purigal, and an assumption that there are a couple hundred speakers of Palula in Badrugal (out of a total population of 740 in 1987 according to Decker 1992b: 143), we conclude that there are approximately 10,000 speakers of the language in Chitral, a figure that should be treated with a certain amount of healthy scepticism.<sup>5</sup> Provided there has been no dramatic increase in any of the other speech communities in the district, Palula is still the second largest language community in Chitral, as it was according to a survey carried out in the 1990s (Decker 1992b: 11).

### **1.2.2 The socioeconomic environment**

All of the locations with any higher concentration of Palula speakers are villages with a rudimentary infrastructure. The community is mainly agricultural, often combined with animal husbandry, and its inhabitants also receive income from timber harvesting, in the form of royalties on the cedar forest and through the sale of firewood. The main subsistence crops cultivated are wheat and maize, but also a variety of fruit and vegetables is grown. In most of the villages there is

<sup>5</sup> The seemingly exact figure 8,600 that (Decker 1992b: 74–76) arrived at (later cited by Ethnologue) was in fact also the result of quite a rough estimate, based on a combination of respondent opinion and a 1987 census report.

### 1 Introduction

Photo 1.2: Police check post in the Ashret bazaar, 2006 (Henrik Liljegren)

an ample supply of water for irrigation. A portion of the population in the Ashret Valley, as well as in the Biori Valley, practice transhumance, in the spring taking their herds of sheep and goats to the high pastures situated at the extreme ends of these valleys and staying there throughout the summer months. This practice, and interrelated activities such as the production of a large variety of dairy products, was a central part of community life and traditions but has given way to today's mainly agricultural society. Whereas irrigated land in and adjacent to the villages and nearby winter pastures are owned by individual families, the distant summer pastures are communal.

As the educational level is steadily increasing and the demand for an income source besides agriculture is deemed necessary, a growing number of people are being employed by the government or carry out business within the private sector, either in their home village, e.g., as school teachers, or in the bazaar town Drosh or in Chitral town, the administrative centre of the district. There is also a local bazaar in Ashret proper, occupying a smaller number of shopkeepers and

1.2 The general setting

craftsmen from the valley. In pursuit of other employment opportunities, a number of Palula families have migrated to urban areas, settling more or less permanently in larger cities such as Peshawar, Rawalpindi and Lahore.

### **1.2.3 Local history and cultural identity**

Palula has usually been described or seen as a single-language community (Morgenstierne 1941: 7; Decker 1992b: 67; 1996: 160; Masica 1991: 21; Strand 2001: 253, 258) as well as a single-ethnic community (Cacopardo & Cacopardo 2001: 79–143; Akhunzada & Liljegren 2009: 5). Although the former is not very surprising, considering the relatively minor dialectal differences, the latter is a more complex issue. From the outsider perspective, and more specifically from the perspective of the Khowar-speaking majority of Chitral District, the people and the speech of the Ashret and Biori Valleys are indistinguishable, and they refer to the people as a whole as the Dangarik and their speech – dramatically different from

Photo 1.3: Wedding ceremony in Ashret, 2005 (Henrik Liljegren)

### 1 Introduction

their own language – as Dangarikwar. Internally, however, the picture is less clear. Indeed, from the perspective of the "southerners" in Ashret, the speech of the "northerners" in Biori is rather similar and largely comprehensible, and vice versa, but from both sides the idea is also common that the other variety is a speech form that is somehow debased or has deteriorated from its pure or original form. Although speakers of the two varieties have interacted and to some extent have also intermarried for a long time, it is obvious that the community in Ashret do not consider the community in Biori to be related to them, the main reason being that they have no genealogy in common.

As is the case in many other communities in the region, there are preserved oral traditions concerning a particular place of origin in these communities as well as genealogies memorised from generation to generation. The most extensive and consistent traditions of this kind is found in Ashret,<sup>6</sup> where the entire population claims descent from a common ancestor.

According to local history, the ancestor of the people of Ashret was Choke (C̣hoók), son of Machoke (Machoók), ̣ <sup>7</sup> who migrated to the present location from Chilas in the Indus Valley some fifteen to sixteen generations ago, a scenario convincingly corroborated by recent research into local history and culture carried out by the anthropologist Alberto Cacopardo (Cacopardo & Cacopardo 2001: 84–93). One particular source states how Choke and two of his brothers arrived in Chitral and reached Drosh but subsequently separated. One brother went to Kalas, a village in the Shishi Koh Valley north of Drosh, another continued to Sau in the Kunar Valley (in present-day Afghanistan), where he settled, and Choke himself settled in Ashret (Cacopardo & Cacopardo 2001: 84). This would support a common origin of the Palula speakers of Ashret and the inhabitants of Sau, the language of the latter closely related to Palula.<sup>8</sup> The inhabitants of present-day Kalas, on the other hand, are all Khowar speakers, but according to respondents in Puri (the only remaining Palula-speaking village in Shishi Koh Valley), the people of Kalas formerly spoke Palula. An independent local tradition among the Bozhokey in the Laspur Valley (about 200 km northeast of the Palula-speaking

<sup>6</sup> These traditions, including the important historical manuscript Tarikh-i-Ashret, written by the local poet and historian Mirza Guldali Shah in Persian verse, are commented on at great length and in great detail by Cacopardo & Cacopardo (2001: 79–143), and a translation of it into English is included in its entirety as an appendix to their work.

<sup>7</sup> In some versions of this tradition, Choke and Machoke were brothers (Cacopardo & Cacopardo 2001: 85, and my own notes).

<sup>8</sup> This was confirmed by respondents from Sau interviewed in an Afghan refugee camp in Dir in 2000 by Ajmal Nuristani and myself who stated that "the people of Ashret are our brothers".

### 1.2 The general setting

area) also speaks of a migration from Chilas some twelve to fifteen generations ago. Inayatullah Faizi, himself a Bozhokey, has documented this tradition, according to which the two brothers Choke and Machoke left Chilas after a power struggle with their elder brother, and after having parted during their exile, Machoke arrived in Laspur where his elder son, Laphur, subsequently settled. The descendants in the Laspur Valley have since been linguistically assimilated by their Khowar-speaking neighbours. Also, some Ashreti sources claim that the Chilasi immigrants indeed came to Ashret from the north, perhaps via Laspur, rather than through the Lowari Pass and Dir Kohistan, which would agree with this Laspuri tradition (Cacopardo & Cacopardo 2001: 85, 125–126).

What then about the Palula speakers in Biori? They are not mentioned in any of the migration traditions from Ashret (or those agreeing with it), whereas it has been explicitly pointed out to me by Ashretis that the people of Biori are descendants neither of Choke nor Machoke but are Kohistanis from Dir who later adopted the language of Ashret (compare Strand 2001: 255; Saeed 2001: 296). The first part of the statement can very well be true, as the main Biori genealogies lack any convincing links to the genealogies of Ashret, but I suspect the second part to be an overinterpretation on the part of the Ashreti informants. Although the ethnic composition of the Biori Valley population is much more complex (including considerable sections with Kalasha and Nuristani origin), and with much less consensus around its origin than what is the case in Ashret, there is indeed a local tradition that connects a major section of the population (especially in the uppermost village Bhiúuṛi) with Dir Kohistan, in particular with a village called Biyar (Bhiáaṛ). Although somewhat speculative, it is possible that a genitive or perhaps an adjectival form along with a regular sound change from /aː/ to /uː/ would render the form Bhiúuṛi, i.e., 'from Biyar'. Also Cacopardo & Cacopardo (2001: 111–108) draws the conclusion that the Palula<sup>9</sup> of Biori most likely came to the valley from Dir Kohistan, somewhat later than the Palula of Ashret, possibly to escape conversion to Islam, which was common at the time in Dir Kohistan. While present-day Biyar is a Kohistani (Gawri-speaking) village, and there are no obvious traces of any previous language spoken in the village, it is not wholly unlikely that the population speaking a language closely related to Palula in the relatively close village Kalkot (both being situated along the Panjkora River) is a remnant of a once more widely spoken Shina variety in this part of Dir Kohistan. Muhammad Atiqullah, a local historian and also my main lan-

<sup>9</sup> Keeping in mind that the ancestors of a good portion of the ethnically-mixed population most likely were speakers of a variety of Kalasha, and that there may also have been other linguistic enclaves before Palula became the sole surviving language of the valley.

### 1 Introduction

guage consultant in Biori, further claims that the people who came from Biyar were originally from Tangir, one of the Indus side valleys west of Chilas (hence the name Tangiri/Dangarik as mentioned earlier).<sup>10</sup> In Puri (also part of the same general dialect area as Biori), a local elder told me that their village was founded by two brothers, Dúuši and Kaṇúuši, who came via Dogdarra, a valley in Dir Kohistan, from a village in the Tangir Valley called Dangeri Phuruṛi, possibly corresponding to a very real place in central Tangir that is rendered Phurori on some maps. As a possible explanation for the exclusiveness on the part of the Ashretis vis-à-vis the Palula speakers of Biori, Inayatullah Faizi (pc) has suggested that those ancestors of the Biori Palula who came from Dir Kohistan may very well have been Shina speakers, but being Yeshkun rather than Shin<sup>11</sup> would immediately have placed them in a non-kin category, regardless of the their linguistic relatedness.<sup>12</sup>

What this gives us are (at least) two possible migration routes from the Indus Valley to Chitral. One would have originated in the Chilas area, in the main Indus Valley, taking the way over the Shandur Pass to Laspur and then continuing south through Chitral to the Ashret Valley, branching out quite early on to Sau in the Kunar Valley. The other one would have originated in the Tangir Valley,<sup>13</sup> taking the route over Swat and Dir Kohistan, leaving a trace in Kalkoti speech, and ending up in the Biori Valley. Whether or not there is any linguistic evidence supporting this hypothesis is something we will have reason to return to briefly below (§1.3.1).

The arrival of the Palula in Ashret must, if the local genealogical evidence and other historical traditions are taken into account, be dated to some time before the mid-17th century (Cacopardo & Cacopardo 2001: 88), and the migration from Dir Kohistan to Biori somewhat later (2001: 118). As for the religion of the Palula speakers, they were clearly still unconverted to Islam when they first entered

<sup>10</sup> Interestingly, while Ashreti people usually dislike the Chitrali designation Dangarik or Dangarikwar, the people of Biori do not seem to mind and sometimes even use the term referring to their own community.

<sup>11</sup> According to Jettmar (2002: 17), the population in the traditional Shina territory was organised into four castes – Shins, Yeshkuns, Kamins and Doms – with the Shins considering themselves ritually cleaner than the other three, possibly modelled on Hindu communities in adjacent areas.

<sup>12</sup> This idea, however, is disputed by Alberto Cacopardo (pc), who holds that the awareness of such a caste identity in relatively recent times is out of question.

<sup>13</sup> As rightly pointed out by Alberto Cacopardo (pc), there is a narrow, modern-day use of Tangir that refers to a particular side valley to the west of Chilas in the Indus Valley, but there is also a broader, past use of Tangir that refers to the larger region around Chilas, including many of the major areas where varieties of Shina are spoken.

### 1.3 The linguistic setting

Chitral, and it was probably not until the latter half of the 19th century that they embraced Islam (2001: 83), and even then only gradually and probably earlier in Ashret than in Biori. What kind of religion was practiced before the Muslim conversion remains uncertain, but elements in it are shared with the non-Muslim Kalasha as well as what is known about other pre-Islamic religions in the region. Finally, I will quote Alberto Cacopardo's brief but interesting summary of his own findings about the great-grandfathers of today's Palula:

They were goat-herders, whose women in the summer watered the fields, while the men went up on their own to the high pastures. Cows were impure to them, like the women at the time of seclusion. They had something like a priestly lineage, and some of them held feasts of merit. They lived in densely forested valleys haunted by murderous bandits and by the fear of monstrous ghosts. They were mostly dressed in goatskins and travelled only on foot, but they remembered ancestors who rode on horses and fought with bows and arrows against princes who lived in forts. They had rites in which they drank wine and men and women danced together at night. They had temples and shrines, sacred stones and wooden idols, and they said they heard the voices of the fairies (Cacopardo & Cacopardo 2001: 143).

### **1.3 The linguistic setting**

### **1.3.1 Genealogical affiliation**

Palula belongs to a group of Indo-Aryan (IA) languages spoken in the Hindukush region that are often referred to as "Dardic" languages. Totally 27 named varieties are sorted under this grouping (Lewis, Simons & Fennig 2015). It has been and is still disputed to what extent this primarily geographically defined grouping has any real classificatory validity.<sup>14</sup> On the one hand, Strand (2001: 251) suggests that the term should be discarded altogether, holding that there is no justification whatsoever for any such grouping (in addition to the term itself having a problematic history of use), and prefers to make a finer classification of these languages into smaller genealogical groups directly under the IA heading, a classification we shall return to shortly.

Others have continued using the term, for various reasons, in a classificatory sense. Bashir (2003: 822) concludes that there are enough similarities across

<sup>14</sup> For an overview of the terms Dard, Dardic and Dardistan and their different uses, see Mock 1997.

### 1 Introduction

Map 1.2: Languages in the Hindukush region

these languges, in terms of shared retentions, areal features and innovations affecting at least subsets of the languages to justify the umbrella term "Dardic" even though she expresses doubt about the possibility of tracing them all to a single node in a Stammbaum model. Zoller (2005: 10–11) identifies the Dardic languages as the modern successors of the Middle Indo-Aryan (MIA) language Gandhari (also Gandhari Prakrit), but along with Bashir, Zoller concludes that the family tree model alone will not explain all the historical developments.

Masica (1991: 460) questions the value of trying to sort out a complete and accurate New Indo-Aryan (NIA) historical taxonomy and contends that it would be far more interesting to focus on and recognise a number of overlapping genealogical zones, some of them more strongly defined than others. Following that position, I will in the remainder of this work refer to these languages as Hindukush Indo-Aryan languages (HKIA), without any claim of taxonomical correctness but recognising the shared historical developments pointed out by Bashir and Zoller.

### 1.3 The linguistic setting

Some words should now be said about the lower-level grouping of these HKIA languages, which have as their home an extremely mountainous region that stretches from north-eastern Afghanistan over northern Pakistan all the way to Kashmir (see Map 1.2). Strand recognises in his classification (2001: 258, which represents an updated and corrected version of Strand 1973, which in its turn is based on the work of Morgenstierne) eight groups (some of them represented by one language alone):


<sup>15</sup> An entirely accurate classification is still lacking. Bashir (2003: 824), for instance, groups Dameli with Gawarbati in a Kunar group.

### 1 Introduction

Bateri [btv], all spoken in the Indus Valley, and Torwali [trw] in upper Swat.


```
Chilasi
```

```
Kohistani Shina [plk]
    Eastern dialects
                 Astori
                 Drasi
    Dispersed dialects
                 Palula [phl]
                 Ushojo [ush] (?)
                 Kalkoti [xka] (?)
Gilgiti [scl]
```
Brokskat [bkk] (a Tibetanised offshoot)

In an ambitious survey of Shina dialects, Radloff (1992) obtained and compared word lists from 27 different locations (including only what she, implicitly, regarded as representatives of Shina proper, and thereby excluding obvious "outliers", such as Palula at the Western extreme and Brokskat at the Eastern extreme), and came to the conclusion that Shina, on the one hand, can be divided into four clusters based on geographical proximity (Northern, Eastern, Diamer and Kohistan), and on the other hand, can be sampled into two major centres, one around Gilgit and the other around Chilas. A continuum, with two end points, Gupis and Kolai, as described by Radloff (1992: 132), corresponds quite well to the full extension of Shina-speaking locations along the Gilgit/Indus River system,

### 1.3 The linguistic setting

an analysis which in many ways also justifies the labels [scl] and [plk] chosen to pinpoint two markedly different poles with a great number of intermediate forms, particularly realised in the varieties that Strand labels "Eastern dialects".

Strand further identifies the dispersed and archaic varieties Palula and Sawi [sdg] (or Sauji) as of particular interest (the reason he does not include Sawi in the list is perhaps that he sees Palula and Sawi as basically one and the same variety), and that two other speech enclaves, Ushojo and Kalkoti, probably should be included as well (his question marks indicating the need for more and better data in order to draw reliable conclusions). Yet another distinct variety that would be a highly likely candidate for inclusion under "dispersed dialects" is the language of Kundal Shahi [shd], a speech variety of a small community residing in the Neelum Valley in Pakistan-administered Kashmir (Rehman & Baart 2005).

It is primarily in a phonological comparison between cognates that Palula appears to be conservative vis-à-vis the other Shina varieties (Table 1.2). In a comparative study between some major varieties (Palula not included), Schmidt (2004a: 36) shows that three features are common to all of them: a) the preservation of the OIA three sibilants *s, š, ṣ*, as in HKIA languages in general, b) the development of retracted (retroflex) fricatives *c̣*and *ẓ* from OIA clusters *tr, kṣ, dr, bhr,* etc., and c) the development of contrastive pitch accent. While Palula shares the first and the third of those with the other varieties, it has to a larger extent preserved consonant clusters.


Table 1.2: Comparison of cognates in five Shina varieties (based on Schmidt 2004a: 37, with Palula (A variety) data added)

Something should be said about two of the other so-called "dispersed dialects", Sawi/Sauji and Kalkoti, and their relationship to Palula.

### 1 Introduction

*Sauji* (the form used by my respondents when referring to their language in contrast to Sawi, which is the way it has been referred to in most literature) is the speech variety of Sau, a village situated in Afghanistan, on the east bank of the Kunar River, about 20 kilometers south of the Pakistani border town Arandu in southern Chitral. It is uncertain to what degree and by how many people Sauji is spoken today in this village, but according to Decker's (1992) informants, approximately 8,000–12,000 lived in the village prior to the long period of war and unrest in Afghanistan.<sup>16</sup> That the variety spoken in Sau is closely related to Palula was pointed out previously by Morgenstierne in the first half of the last century (Morgenstierne 1941: 7), and was further confirmed by the more extensive study of the language undertaken by Buddruss in the 1950s:

Dagegen ist die nahe Verwandtschaft des Sawi mit dem Phal. bereits durch einen Blick in Grammatik und Vokabular evident und wird überdies durch die Angabe meines Gewährsmannes bestätigt, daß er die Sprache der Leute von Ashret verstehen könne. Dennoch sind die beiden Sprachen keineswegs identisch mit einander. (The close relationship of Sawi with Phal. is, however, already evident by a glance at its grammar and vocabulary, and is also confirmed by my informant's own statement, i.e., that he could understand the language of the people of Ashret. Yet, the two languages are by no means identical to each other. *HL's translation.*) (Buddruss 1967: 11)

That there is a historical connection between Sau and Ashret according to local tradition was already mentioned (see §1.2.3), but no major interaction or contact between the two communities seems to have taken place in the recent past, and for a long time, the population of Sau has been included and been seen as a part of the all-surrounding Gawar community, sharing their identity in all aspects save the language, something rather remarkable from a regional perspective (Cacopardo & Cacopardo 2001: 232).

*Kalkoti* (or Khalkoti), on the other hand, has remained largely undocumented. It is spoken by approximately 6,000 people, apparently confined to a village called Kalkot, situated in the upper Panjkora Valley in Dir Kohistan. Since no systematic survey has been carried out in Dir Kohistan, however, it is too early to exclude the existence of other locations where this variety or something similar is spoken. Most other villages in the main valley, from Rajkot (Patrak) and

<sup>16</sup> A good portion of a refugee camp in Timergara in Lower Dir that I visited along with Ajmal Nuristani in 2000 was made up of Sauji speakers from Sau. However, in the last few years many are said to have returned to their home village in Afghanistan.

### 1.3 The linguistic setting

upstream, are Gawri-speaking,<sup>17</sup> but that the speech of this village<sup>18</sup> could be something rather different was first hinted at in a sociolinguistic survey carried out by Rensch (1992: 7) and his colleagues: "The linguistic variety spoken in the village of Kalkot in Dir Kohistan seems to be quite distinct from that spoken in the surrounding villages of Dir Kohistan and in Kalam, although it is obviously related". Richard Strand, as was pointed out above, tentatively classified Kalkoti as a Shina variety closely related to Palula, based on the word list presented in the already mentioned survey report (Rensch 1992: 159–176), particularly pointing to the strikingly similar forms of the personal pronouns. Perhaps there is, as was mentioned above (§1.2.3), a historical connection between the Palula speakers of Biori, who tradition says came from Dir Kohistan, and the Kalkotis.

My own study (Liljegren 2013), based on data collected by Naseem Haider and Muhammad Zaman Sagar in Kalkot in early 2006, confirms the hypothesis that Palula and Kalkot are closely related and also that Kalkoti, although heavily influenced by Kohistani (Gawri), is essentially a Shina variety. The assumption is that certain types of words (see Table 1.3) are much less likely to be borrowed, such as personal pronouns, lower numerals, kinship terms, and basic verbs (Trask 1996: 23).

Based on this, it seems that the two main Palula dialects (A and B), Sauji and Kalkoti, although separated for perhaps several centuries, form some sort of cluster of their own (Liljegren 2009).

So far nothing has been said about Ushojo, one of the other "enclaves" mentioned by Strand. Not much was known about this language or its speakers prior to the sociolinguistic survey referred to above (Decker 1992c). Ushojo is spoken by an estimated 2,000 people in the Bishigram Valley in Swat, and the ancestors of the present-day speakers are said to have migrated from Kolai, a Shinaspeaking area in Indus Kohistan (Decker 1992c: 69), and the word list and other data<sup>19</sup> suggest that it is at its core another dispersed Shina variety (compare Strand 2001: 255), although Zoller (2005: 9), without further justifying it, refers to it as "a Kohistani language with Shina elements". In any case, the preliminary lexical comparison made by Sandra Decker (1992c: 70) did not show any greater

<sup>17</sup> I use Gawri to refer to the Kohistani varieties that have been variously called Bashkarik, Kalam (Kalami) Kohistani or Dir Kohistani. Gawri seems to be the preferred name and one that can be accepted by speakers from locations both in Swat and Dir Kohistan.

<sup>18</sup> In fact, about 70 per cent of the Kalkot population are speakers of Kalkoti, whereas the remaining 30 per cent are speakers of a Gawri variety (Muhammad Zaman Sagar, pc).

<sup>19</sup> Such as the "core" verbs *han-/as-* 'be' and *th-* 'do' (Decker 1992c: 71–72, 199–203), which are typical features of Shina.

### 1 Introduction


Table 1.3: Lexical comparison between Palula (A variety), Kalkoti and Gawri

similarities with Palula or Kalkoti than with any other Shina varieties, rather the other way around, and there is nothing really that would directly connect the Ushojo community and its migration routes with the Palula-Sauji-Kalkoti cluster. The same probably holds true for the Kundal Shahi variety, which might be an outlier of the Eastern cluster of Shina (Rehman & Baart 2005: 9).

Outside this particular cluster, there are indications that it is the Shina subvarieties spoken in the Tangir and Darel Valleys, in today's Diamer District, that may be most similar to Palula (Radloff 1992: 142–143).

### **1.3.2 Areal affinities**

There are of course other factors beside genealogical relatedness responsible for similarities between languages, one being contact. When looking at the Hindukush-Karakorum region where Palula is spoken, one finds a highly interesting and relatively little researched region, areally as well as linguistically, which lies at the crossroads between different geographical as well as cultural zones.

### 1.3 The linguistic setting

From the larger perspective, the region is part of (although at the fringe of) the Indian subcontinent, and since there are many more languages beside the Indo-Aryan in the area, many with even longer histories, the question of a linguistic area or a *Sprachbund* in South Asia becomes relevant. Besides the IA languages, there are other Indo-European languages, mainly Iranian languages in the west, as well as English (although it has a relatively short history in the region). In addition to Indo-European languages, there are representatives of Tibeto-Burman in the far north and east, Dravidian in the south (with the 'remnant' Brahui in the north-west), Austro-Asiatic in the northeast, the isolate Burushaski in the northwest, and even Turkic languages in the northwest, although the latter are found outside of what is normally seen as part of the subcontinent. A number of traits (such as retroflex consonants and dative subjects) have been suggested as areal, shared across languages and language families in South Asia, but not without lengthy discussions about what really should be considered areal (Emeneau 1965; 1980; Masica 1976; 1991; 2001; Ebert 2006).

According to Masica (2001), areal traits or features are not only of one kind, which is why he suggests a finer differentiation between 1) *essentially areal features*, which would be those features really defining the language area (as, for example, the above mentioned ones); 2) *macroareal features*, for features found not only in South Asia, but in most parts of mainland Asia (such as SOV word order); 3) *marginal* or *linking features*, for features "spilling over" from adjacent areas but not necessarily affecting the entire South Asia (such as the "essentially" Southeast Asian use of numeral classifiers in the northeast, or ergativity linking together South Asia with areas of Western Asia); and 4) *subareal features*, for features defining a smaller area within South Asia. Others (Dahl 2001; Koptjevskaja-Tamm & Wälchli 2001) have been skeptical towards the notion of linguistic areas as such, questioning whether such areas have a reality of their own rather than merely being convenient ways of summarising certain phenomena.

Regardless of the particular view one takes on areality in the end, there is reason to further inspect the languages and traits found in the northwestern corner of the subcontinent (i.e., the Hindukush region) "where conflicting areal patterns meet and interact, and many peculiar languages ('Dardic', Burushaski,<sup>20</sup> the Pamir group of Eastern Iranian), at once archaic and innovating, find their home" (Masica 2001: 225). To this collection of languages should be added the Tibeto-Burman language Balti, which is spoken in the eastern part of the region adjacent to the main Shina belt, and the Nuristani languages, mainly found in the part of Afghanistan bordering Chitral and, to a lesser extent, on the Pakistani

<sup>20</sup> A language isolate, spoken in a few valleys in the extreme north of Pakistan.

### 1 Introduction

side of the border. The latter were initially lumped together with the "Dardic" languages but are now considered a branch of Indo-Iranian, beside IA and Iranian (Degener 2002; Strand 2001).

A few other scholars of South Asian languages, in addition to those mentioned, have identified features or a certain convergence of features that seem to be of particular relevance to this region (or some similarly defined region). Some of the more promising are those suggested by Bashir (2003), including grammaticalisation of evidentiality, multi-differentiating deictic systems, predominantly leftbranching structures, contrasts between dental, retroflex and palatal sibilants and affricates (compare Tikkanen 2008), and the development of tonal or accentual systems, the latter feature further elaborated upon by Baart (2014). An interesting feature, only affecting a few (mostly adjacent but not closely-related languages) in this subregion, is the development of retroflex vowels, described in Mørch (1997) and Heegård & Mørch (2004). The loss of gender distinctions in the geographically-peripheral Indo-Aryan language Khowar (and also in closelyrelated Kalasha) was pointed out already by Emeneau (1965) as resulting from Persian influence (via the Iranian Pamir languages). More research may reveal whether Emeneau's conclusion is correct, and also how this phenomenon is related to an assumed grammaticalisation of animacy distinctions, present not only in Khowar and Kalasha (Bashir 1988: 401), but also in largely undocumented and not yet fully classified Dameli (Perder 2013: 4–6). Possible substratal influence, related to the isolate Burushaski or other now extinct languages, is also to a large extent *terra incognita* (Tikkanen 1988).

Without any claim of completeness, there are also a number of other relevant works relating to areality in the region, or in a similarly defined region, that the interested reader is advised to consult (Bashir 1988; 1996b; Èdel'man 1980; 1983; Fussman 1972; Skalmowski 1974; Tikkanen 1999; Toporov 1970).

### **1.3.3 "Next-door" linguistic neighbours**

Surveying the languages of the region in the 1920s, Morgenstierne stated that "Lower Chitral is one of the most polyglott [*sic*] regions of Asia" (1941: 7), and if something characterises the immediate surroundings of the Palula area, it is multilingualism and an ample opportunity for cross-language interaction. Decker (1992b: 10–23), in his and his team's sociolinguistic survey, identified as many as twelve languages native to the 200,000+ population of Chitral District at the time, and another four non-native languages that played some role in the lives of people in the district. Some of the nearest linguistic neighbours of the Palula – historically or presently – are the following:

### 1.3 The linguistic setting

**Khowar [khw].** Although not spoken in any area immediately adjacent to Palula until quite recently, Khowar is now the overall dominant one in the district. It is spoken as the first language of approximately 82 per cent of the population (Decker 1992b: 11), and functions very much as a *lingua franca*, but in competition with Pashto south of Drosh Bazaar. Khowar is an IA language classified as belonging to the Chitral group (see §1.3.1). The historical heartland of Khowar and the Kho people is the northern part of the Chitral District, north of Chitral town, from where the language apparently expanded southward to incorporate a number of other ethno-linguistically distinct groups that previously used their own languages (Morgenstierne 1932: 46–47). Today, the total number of speakers of the language is 200,000–300,000 (Decker 1992b: 31–32). As previously mentioned (§1.2.1), Khowar competes with Palula in some of the Palula-speaking locations.

**Kalasha [kls]** is the language spoken by and most intimately associated with the only surviving non-Muslim population in the entire region. The language is today the first language of 3,000 to 6,000 individuals (Trail & Cooper 1999: xi; Heegård Petersen 2006: 8) in a few famous side valleys – Birir, Bumboret and Rumbur – to the west of the Kunar River, near the Afghan border, and it also includes a slightly different variety used in Urtsun Valley (where the speakers are Muslim). Due to the unique features of the traditional Kalasha religion and culture, it has received a great deal of attention from anthropologists and folklorists throughout the years and is therefore probably one of the best documented communities of the region. The Kalasha language was once spoken in a much larger area – corresponding to the local political power held by the Kalasha in earlier times (Siiger 1956: 33) – possibly the only language truly indigenous to southern Chitral (Strand 2001). The language shares a number of features with Khowar, linguistically its nearest relative in the Chitral group. Although not spoken in any location adjacent to Palula today, it was probably the closest linguistic neighbour of Palula for centuries, sometimes spoken side by side in the very same location (particularly in Kalkatak and the Biori Valley).

**Dameli [dml]** is spoken by about 5,000 people (Decker 1992b: 118) in eleven small villages in the Damel Valley, the next populated side valley south of the Palula-speaking Ashret Valley. Classification has proved to be a complicated issue with this language, as it it is similar in most respects to other IA languages in the region, while it is also strongly influenced by neighbouring Nuristani languages (Strand 2001: 254). The dominant second language in the Damel Valley is Pashto and all men in the valley are reported to be able to communicate in that language, whereas most women in the valley seem to be monolingual in Dameli (Decker 1992b). Its position as the primary language for in-group communication

### 1 Introduction

is largely unthreatened (Perder 2013: 7). Although there is some intermarriage between Dameli and Palula, the contact between the two groups may have been more frequent in the past, as indicated by shared vocabulary (Morgenstierne 1932: 59–60).

**Shekhani [xvi]**, or Kamviri, is a northern Nuristani language.<sup>21</sup> From its old heartland around Kamdesh in southern Bashgal Valley in Afghanistan, the language had already found its way into Chitral by the end of the 19th century. There are approximately 1,500 to 2,000 speakers of this language (Decker 1992b) in the villages Langorbat and Badrugal, the latter situated along the main road between the mouths of the Biori and Ashret Valleys, while about 4,000 speakers of the language remain in the old heartland in Afghanistan. The most widespread second language of Shekhani speakers is Pashto, a language preferred even in interaction with Khowar speakers, but (as mentioned above) in Badrugal, Palula seems to have a rather strong position as a second (or possibly third) language, apparently resulting from intermarriage with Palula-speaking families in nearby Palula locations.

**Pashto [pbu]**. In spite of its current dominant position as a trade language and the unrivalled *lingua franca* of the entire province,<sup>22</sup> Pashto has a fairly short history in Chitral, spoken only by a few settlers less than a century ago (Morgenstierne 1932: 67). Many of those families who now count Chitral as their home district started moving into Chitral in the 1930s. Although no more than 3,000 (Decker 1992b) individuals in Chitral had Pashto as their first language in the beginning of the 1990s (a figure that more than likely has increased a great deal), the Pashtuns (or Pathans as they are often referred to) is a very influential group and, according to one report, control 85 per cent of all trade in the district. While earlier immigrating Pashtuns learnt to speak Khowar in order to carry out their trade, it is now common for speakers of other languages to learn Pashto to have access to trade in Chitral. The largest concentration of Pashto speakers is probably in and around Drosh and along the Kunar River southwards from there. There are also Pashtun settlements in the upper parts of the Ashret Valley.

<sup>21</sup> Although Kamviri has been given its own ISO 639-1 code, it is according to Strand (1973: 298) and Degener (2002: 104) a mere sub-variety of Kati [bsh], or Kamkata-viri as Strand (2001) more recently refers to it, the largest and most widespread of the languages classified as Nuristani.

<sup>22</sup> Pashto is an Iranian language with a total of 25–50 million speakers on both sides of the Pakistan-Afghanistan border (David 2013: 7–8). A fourfold dialect division based mainly on phonological features, places the Pashto of this region in a Northeastern group (Elfenbein 1997: 739–740), roughly corresponding to Northern Pashto [pbu] in the Ethnologue (Lewis, Simons & Fennig 2015) classification.

1.3 The linguistic setting

### **1.3.4 Patterns of language use**

Palula is almost exclusively used among people who speak Palula as their first language. Within the Biori and Ashret Valleys, it is in most cases the only language in communicative use, and there are very few native speakers of other languages residing in those locations. However, as soon as there is a non-Palula speaker present, even in one of the main Palula villages, the language switches to one of wider communication, vz. Khowar with most other-tongue speakers from within Chitral, or Pashto, which is used with some non-Khowar speakers from within Chitral, as well as with Pashtuns from places beyond the Lowari Pass.<sup>23</sup> Decker (1992b) draws the conclusion that Palula speakers are probably less proficient in these other languages in domains requiring little or no outgroup interaction.

The only indication that Palula may be learnt by speakers of other languages is in the, now historical, case of Kalasha speakers in Kalkatak shifting from Kalasha to the use of Palula (Decker 1992a: 55, 79; Akhunzada & Liljegren 2009: 5), and the Shekhani speakers of Badrugal of whom reportedly a fair number use Palula as a second language (Decker 1992a: 59).

The only location where a substantial number of people might be termed semispeakers of Palula is Kalkatak. Many children of one or two Palula-speaking parents have some proficiency in Palula, but due to lack of practice and the predominance of Khowar in many daily life situations, as well as a conscious language shift even in originally Palula-speaking families, their Palula skills are imperfect (according to other speakers in Biori and Ashret). The only situation where they have to use Palula is when speaking with elderly monolingual relatives or in interaction with people from other Palula-speaking locations.

Women with other first languages marrying into families in Biori and Ashret also learn or are expected to learn and use Palula with their in-laws as well as with their own children, which is generally the normal pattern in the region. That pattern, however, is often reversed in the case of the in-laws' language being a low-prestige language. Therefore, in a village like Kalkatak, and to some extent in Puri, where Palula identity is weak, a strong reason for choosing a Khowarspeaking wife is, contrary to this overall pattern, to ensure that the children-to-be are brought up with a good command of Khowar. A secondary effect is that also the husband and the in-laws benefit from this, by being offered an opportunity to shift from low-prestige Palula to high-prestige Khowar.

<sup>23</sup> Communication with other outsiders is still rather infrequent, but Urdu would be the natural choice of language to the extent the Palula speaker is able to communicate in it (which constituted about 50 per cent of Decker (1992)'s Palula respondents).

### 1 Introduction

Photo 1.4: The author (left) interviewing group of men in Puri, 2004 (Emil Perder)

Most, if not all, children of Palula-speaking parents in Ashret and Biori learn Palula as their first language, but there is a tendency to pick up Khowar or Pashto as second languages at a very early age, although children can be said to remain virtually monolingual until they start school. Only a few Palula-speaking parents (according to Decker 1992b) report that their children do not speak Palula at all, mostly in Palula locations outside Biori and Ashret, i.e., locations with an already weak or vanishing Palula identity.

There is widespread multilingualism in all of the Palula villages, and the ability to speak and understand other languages has possibly increased over time. There is, generally speaking, a strong pressure to learn other languages, regardless of location, for educational and business purposes. As soon as children start school, they need to learn Urdu in order to gain anything from the teaching. And as soon as they leave their home village or valley to go to the bazaar or to meet people from non-Palula villages, they need either Khowar or Pashto to make themselves understood. In contrast, there is, as far as can be observed, not much pressure

### 1.3 The linguistic setting

to reject their own language. The only exception to this rule might be Kalkatak, where Palula is viewed as being lower in prestige and less useful than Khowar.

The exact pattern of multilingualism varies from location to location. Most of those interviewed by Decker (1992b) said they speak at least three languages besides Palula. Overall, Khowar is the most common second language among Palula speakers. It is most frequently used in the bazaar town Drosh as well as with non-Palula neighbours (in the locations outside Biori and Ashret) and in most contacts with local officials. Only in Ashret is Pashto the most common second language. Particularly in Kalkatak and in the Biori Valley, many people are proficient in Khowar as well as in Pashto. Only a few people are purely monolingual: mainly older people, women to a larger extent than men, and small children (Decker 1992b).

The prescribed language of most formal education is the national language Urdu, which in practical terms means that teachers in the lower grades, in as far as they themselves are locals, teach and give explanations in Palula. Instruction in Palula is supposed to be gradually replaced by "Urdu only" instruction as students move into the high school level, but to what degree that is realised practically is rather uncertain. It may be foreseen that an increasing number of people in the near future will be exposed to Urdu as a language of wider communication, as a result of more and more people getting access to radio, TV and other media, and there will almost certainly be a growing emphasis on Urdu skills for obtaining qualified jobs.

English plays a role similar to that of Urdu, as it is the medium (at least technically) of instruction in the private schools of the district, including private schools in Ashret and Biori. Exposure to English as a language of natural interaction is, however, restricted to ephemeral contact with foreigners coming to Chitral in the summer months. As the official language of all civil services, knowledge of English is considered very attractive and is very important for those applying for education outside their home district, a position within the local or regional administration or a job within the tourist sector. Both Urdu and English function as markers of prestige. Educated speakers tend to mix in Urdu lexical material relatively freely into their Palula speech, and to a more limited degree they also use English, almost exclusively via Urdu.

Arabic in its literary form has a status similar to that of Urdu or English, but its use is strictly limited to the topic of religion. There are probably very few if any first language speakers of Arabic in the district, though a number of religious scholars have or claim to have a command of the language. A special area of loans has to do with religion, and naturally we find a large number of Arabic or Perso-Arabic loans here, but generally it is difficult to discern for certain whether they

### 1 Introduction

are borrowed from Arabic or Persian directly or have come into the language via Pashto, Urdu or even Khowar.

A historically deeper level of loans in the lexicon of Palula suggests more frequent contacts with speakers of other languages, particularly Dameli and Kalasha, as was mentioned above. Persian, as the official language in the former kingdom of Chitral, once played a role similar to that played by Urdu today.

### **1.4 Internal variation**

There is no significant dialect variation between the different Palula locations, but there is reason to speak of two dialect areas, each representing one of the two major concentrations of Palula speakers: Ashret and Biori. Although it is not entirely clear how the speech varieties of Ghos (now extinct or nearly so) and Puri relate to these two main dialect areas, the data available for the Palula of Puri agrees more closely with the Biori variety than it does with the Ashret variety. We shall therefore regard the dialect of Puri, together with that of Kalkatak, as a subvariety of the Biori variety. There is no dialect continuum between the Biori and Ashret varieties (as the population is confined to these two relatively fertile side valleys), and the geographical proximity within each valley is high enough to exclude any significant in-dialect variations.

Individual examples representing the speech of Ashret and Biori will be indicated by A and B, respectively, in front of the references given throughout this work, and if nothing is explicitly indicated it is the speech of Ashret that is intended.

Although individual differences will be pointed out as different features are discussed, a few salient divergent features should be mentioned briefly. They have mainly to do with morphophonology and the lexicon, and to a much lesser extent with phonology proper, morphology or syntax.

Due to differences in the historical development of accented vowel sounds, there are some regular correspondences (as shown in Table 1.4) between A and B cognates. That some vowel raising and lengthening have taken place in A regardless of syllable structure, whereas the same processes have been conditioned by accent position as well as syllable structure in B, has resulted in some paradigms, particularly the nominal, displaying morphophonemic alternations in B (sg: *sáan*, pl: *súuna; krám, kráama; chéetr, c ̣ híitra ̣* ) that are not found in A. But conversely, and for the same reason, some other alternations have arisen in A (*sáar* 'lake', *sarí* 'lakes'; *dáar* 'door', *dará* 'doors') that are *not* found in B (*sár, sarí; dár, dará*). See §3.5.1 for further examples.

As far as the inventory of phonological segments and suprasegmentals are

### 1.4 Internal variation


Table 1.4: Reconstructed vowel development from proto-forms to A and B forms, respectively

concerned the two dialects are basically identical, whereas there are some phonological processes that seem to be confined to one of the two dialects. One of them is /l/-velarisation (see §3.1.2), exclusively found in B, possibly resulting from Kalasha substratal influence. Some entirely non-predictable differences in pronunciation are also found, as when a word is pronounced with an *h*-segment in one of the dialects and not in the other: B *ɡhaḍé-* vs. A *ɡaḍé-* 'take off'. In other cases, such differences are limited to a subset of a paradigm. In B, some inflectional forms of 'come' contain an *h*-segment (*yhéendi* 'she comes'), others lack it (*yéeli* 'she came', *yíi* 'she will come'); in A, however, all the forms of that verb include an *h*-element (*yhéendi*, *yhéeli*, *yhíi*).

Lexically there are some differences, but these primarily have to do with separate sources of influence. While loans from Pashto have been prevalent in A, it seems that Khowar has been the more common donor language as far as B is concerned. Sometimes a "native" word has been replaced in one of the dialects, whereas it has been preserved in the other: B *niwešé-* (from Khowar) vs. A *čooṇṭá-* 'write'; B *neeṛíi* vs. A *zeelái* (from Pashto) 'root'. In other cases, an identical, or corresponding, lexical form differs in usage or scope. The word *ɡaaḍbáabu*, literally "big father", can only refer to one's father's older brother in A, whereas in B, *ɡaaḍubáabu* is the word also used for one's grandfather, paternal and maternal alike (correspoding to A *dóodu* and *móomu*, respectively). The general word for 'clothes' in B is *dóbal*, a word that in A has the markedly negative meaning 'rags, tattered clothing', while *paaṇṭí* is the word typically used for 'clothes' in the latter variety.

Syntactically, the Perfect formation with the perfective finite verb and *hin-* 'is',

### 1 Introduction

the most common construction used in A, seems to be missing altogether in B. Instead a construction with the nonfinite converb and *hin-* is used for expressing the same category (see §10.1.7).

### **1.5 Previous research**

Georg Morgenstierne was the first scholar to collect linguistic data from the language. In 1929 he visited Chitral and collected, apart from information and data on a number of other languages, Palula language material. Some preliminary notes on Palula are included in a summary report, but most of his Palula material and analysis can be found in a research article titled *Notes on Phalūṛa: An unknown Dardic language of Chitral* (1941). The latter is mainly based on the Ashret dialect, "en meget eiendommelig, og hittil ganske ukjent indisk dialekt" (a most peculiar, and so far, fairly unknown, Indic dialect), as he describes it in his personal journal (Morgenstierne 1992: 38). Briefly (in about 5o pages), Morgenstierne (1941) outlines its assumed relationship to other Indo-Aryan languages, describes summarily its phonology and morphology. The article also includes a translation of a standard text (the beginning part of The Prodigal Son) and a word list.

The next scholar to approach the subject, or something closely related, was Georg Buddruss, who studied the language spoken in Sau. During the winter of 1955–56, he sat with a native informant from the village, and about a decade later published his results in *Die Sprache von Sau in Ostafghanistan: Beiträge zur Kenntnis des dardischen Phalura* (Buddruss 1967), including an outline of the phonology and morphology of this variety from a historical-comparative perspective and some comparisons with Morgenstierne (1941)'s Palula material. Syntax is given only a few pages, but 11 shorter texts with German translation and a word list are generously included.

A sociolinguistic survey of northern Pakistan was carried out by the Summer Institute of Linguistics (SIL) in 1989–90. In the fifth volume of the report, *Languages of Chitral* (Decker 1992b), a chapter on Palula written by Kendall Decker discusses the sociolinguistic environment of the language, including its geographical extension, and summarises its history as viewed by the community. It also describes the social and economic environment and presents some factors having to do with language use. Attached to this study is a 210-item word list with words from Ashret, Biori, Purigal and Sau respectively, as well as some partly interlinearised texts. The material discussed later by Decker (1992a; 1996) overlaps to a large extent with that of the survey (Decker 1992b).

Richard Strand has continuously posted on the Internet (1997/2015) results

### 1.6 Current study

from a short stretch of fieldwork carried out on the Ashret dialect, including some historical-genealogical material from Ashret, a phonological statement and a semantically-structured lexicon (incorporating Morgenstierne (1941)'s word list). This project is all part of a larger attempt at documenting the languages spoken in the Hindukush region, with special references to Nuristan and the Nuristani languages.

Elena Bashir, focusing her scholarly work on Kalasha and Khowar, includes a section on Palula in *The Indo-Aryan languages*, a recently published standard work (2003). Although mainly based on Morgenstierne (1941), it is supplemented by Bashir's own field notes from the B variety. Some original B data of hers is also included and discussed in a paper on quotatives and complementisers in the region (1996).

Another work touching on the subject, although not linguistic per se, is the ethnohistorical work carried out by Alberto and Augusto Cacopardo (2001). In *Gates of Peristan – history, religion and society in the Hindu Kush*, one entire chapter by Alberto Cacopardo is devoted to Palula as a group, including a review and discussion of a wealth of historical sources and the local tradition, combined with a presentation of carefully recorded first hand observations and interviews. A previously unpublished essay from 1987, written by the former major Ahmad Saeed (from Ashret), is included as an appendix. In it, Saeed draws up a historical-genealogical background of the community and the area where the language is spoken today, drawing from its rich oral tradition.

Summarising all the previous research carried out, we conclude that there is still no description of Palula that has taken larger amounts of data into account or one that reflects the structures of the language at more than a rudimentary level. Outside Palula specific studies, a number of scholars have been and continue to be actively involved in research in the region, but I will not make an attempt to paint that picture at the present time.

### **1.6 Current study**

The aim of the current work is to provide an accurate description of Palula phonology, morphology, and syntax, based on first hand data and generously illustrated with examples from natural use. Some discourse features appearing in the language will also be mentioned, but no section as a whole will be dedicated to this topic. The present work includes two interlinearised sample texts, one in the Ashret dialect, and the other in the Biori dialect. For further information on the lexicon of the language, the interested reader may consult *Palula Vocabulary* (Liljegren & Haider 2011), a limited Palula-English dictionary produced primarily

### 1 Introduction

with the scholarly community in mind, and for more text samples, *Palula Texts* (Liljegren & Haider 2015b) may be consulted, a collection of annotated texts representing various genres. The project has also generated a few other topic-specific publications (Liljegren 2009; Liljegren & Haider 2009; Liljegren 2010; Liljegren & Haider 2015a).

Photo 1.5: The author (right) recording a narrative with Muhammad Zahir Shah in Puri, 2004 (Emil Perder)

The main bulk of the material on which this grammatical description is based was collected during several periods of field work in Pakistan from 1998 to 2006. I resided along with my family in Peshawar during two longer periods, the first being 1998–2000, as a student of Pashto at the Area Study Centre of Peshawar University, and then 2003–2006 while involved in the establishment and development of the Forum for Language Initiatives (FLI) in Peshawar.<sup>24</sup> After the completion of my dissertation in 2008, I served for another two-year period, 2008–2010,

<sup>24</sup> The organisation was initially established as the Frontier Language Institute. In 2009, it relocated to Islamabad.

as a consultant with FLI while also following up on my Palula research and being engaged in areal-linguistic research. My time spent in Chitral extends from a few days to periods of two months at a time, variously staying in Drosh, Kalkatak, Biori and Ashret.

My main philosophy has been to diminish the gap between myself and my Palula respondents/consultants and their community as much as possible, thus avoiding unnecessary filtering. Therefore, I have gradually acquired not only a passive understanding of the language, but also as much as possible I have used it in interaction with Palula speakers. However, I am in no position to claim that my research has been carried out entirely monolingually. English and Pashto have also functioned as important communication languages. A more passive and rudimentary understanding of Urdu has also been of some help.

Although all this started out very much as a one-man enterprise, it developed gradually into more of a collaborative undertaking, and throughout the project a number of people from the Palula community have been involved at varying levels of activity, independence and expertise:

**Naseem Haider**, a schoolteacher from Ashret, came to be my chief language consultant. He worked full-time together with me in Peshawar, from mid-2003 to 2006, and even after my return to Sweden, he continued assisting me from a distance. After being trained through FLI in basic language documentation, he carried out a large number of interviews and recordings with people in his community, transcribed massive amounts of text, worked with me on translation into English, filled in a number of questionnaires, participated in and gave valuable input to the on-going analysis, and worked on a Palula lexical database. Presently, he leads a mother-tongue based educational project in Ashret, while also serving with FLI as a consultant to multiple language communities in the area of literacy.

**Haji Muhammad Atiqullah**, a school principal from Dhamaret, Biori, became my main Biori language consultant in 1998. Although never under any formal agreement, he was instrumental in systematically introducing me to his language during my early visits to Chitral and helped me go through lots of language material and sort out a number of phonological and grammatical issues. He went through the same training at FLI as Naseem Haider.

**Ikram ul-Haq**, a schoolteacher from Ashret, assisted me voluntarily in my research during the period 1998–2000. He was introduced to transcription and basic recording methods and made a number of important interviews and text recordings, which he also transcribed and provided with free translations.

**Sher Haider Khan**, a schoolteacher from Ashret and older brother of Naseem

### 1 Introduction

Photo 1.6: Main language consultants Naseem Haider (left) and Muhammad Atiqullah (right), 2006 (Henrik Liljegren)

Haider, assisted me voluntarily on different occasions, filling in questionnaires, providing natural examples and explaining various aspects of his language. He participated in parts of the FLI training programme.

A few other people spent considerable time with me, contributing in important ways to my research, understanding and ability to speak Palula: Saeed Ahmad, Kalkatak; Atahullah, Biori; Sardar Hayat, Ashret; (late) Said Habib, Ashret; Sher Habib, Ashret; and Munir Ahmad, Ashret.

My material on phonology is primarily based on separate B and A word lists<sup>25</sup> that I began compiling and recording with speakers in 1999. Those have been supplemented later with other lists and recordings.

The morphological and syntactic parts of my study are primarily based on text

<sup>25</sup> This included lexemes (primarily nouns and verbs) recorded in differently inflected forms as well as recordings of words in isolation and in frames.

material, in all 62 texts from 37 speakers/writers, of different length and analysed with varying accuracy and detail. Most of the texts are recorded and transcribed oral narratives, but also other textual genres are represented. A few texts in the material were written and in some cases only later recorded when read out. However, even those written texts represent an oral rather than a literary style, as there is no literary tradition. A tentative orthography was introduced as late as 2004, and the written texts were produced during an experimental phase in its development. The textual material was supplemented with elicitation of full paradigms, local proverbs, expressions and sentences, questionnaires and notes of language use in the community.

For details on textual genres, types of data and individual speakers and writers or informants, the list in *Abbreviations* at the beginning of this work should be consulted. The references given after each example utterance throughout the work can be found in that list: for example, A:SHY028 refers to a particular narrative (abbreviated SHY) in the A variety (hence A:) written by Sher Haider, occurring as entry 28 in my interlinear text database.<sup>26</sup>

Although only marginally referred to and used, a few smaller field studies carried out parallel with my Palula research also need to be mentioned:

As already hinted at earlier, some **Sauji** data was collected in 2000, by Ajmal Nuristani<sup>27</sup> and myself, in Timar Camp in Lower Dir, where a substantial number of people from Sau resided at the time. Later, Ajmal Nuristani, at my request, recorded and transcribed a few other texts and word lists, both in Pakistan and in Sau (Afgha

AlsoMuhammad Khanntioned briefly above, a survey trip to Kalkot in Dir Kohistan was carried out under Joan Baart's (SIL) and my auspices by Naseem Haider and another FLI fellow, Muhammad Zaman Sagar (from Kalam, Swat), in 2006. Apart from obtaining sociolinguistic information, they recorded a few **Kalkoti** texts, a word list with two different speakers, a questionnaire focusing on verbs and another focusing on pronouns.

As for **Dameli**, data was chiefly collected in the form of a questionnaire (dealing with pronouns and verb agreement) filled out in the summer of 2005 by Hayat Muhammad Khan of Aspar (Damel Valley), who participated in the FLI training program. The data was subsequently discussed with and commented upon by

<sup>26</sup> A few texts are not entered as interlinear texts and therefore references to individual strings in them are not numbered.

<sup>27</sup> Ajmal Nuristani's father is Nuristani and his mother's relatives come from Sau, but he is primarily a Pashto speaker who grew up as a refugee in Chitral.

Emil Perder (Stockholm University).<sup>28</sup>

Also in 2005, I carried out some interviews (based on the questionnaire also used for Dameli) in Kalkatak with Muhammad Salaam and Faiz Muhammad, two **Gawarbati** speakers, both in their thirties and originally from Nari in Afghanistan and since the early 1980s settled in the refugee section of the village Kalkatak.<sup>29</sup>

### **1.7 Palula as a written language**

Until recently, Palula was unwritten and largely undocumented, a condition shared with most smaller (and even some larger) language communities in the region. Before the commencement of the current research, only a handful of local poets saw the need for writing Palula, making use of Urdu writing when composing poetry to be read aloud by themselves. There had been no systematic or collective attempts at creating a consistent and practical orthographic representation of Palula. The community is, after all, relatively small, with only a limited number of highly educated people, and uses a language entirely deprived of any outside recognition. Although there are primary schools throughout the Palulaspeaking area, only Urdu (and more recently English) is the recognised medium of instruction and formal literacy.

In 2003, representatives from all the major settlements came together and formed *Anjuman-e-taraqqi-e-Palula*, <sup>30</sup> a society for the promotion of Palula, with the purpose of facilitating the development of Palula as a vehicle for literary and educational efforts. At that time, an orthography proposal was put together by Naseem Haider and myself, with input from a few local scholars and teachers. It was endorsed by the society, which agreed that a Perso-Arabic script, conforming closely with the way it is applied to Urdu, should be used as a basis, with the addition of symbols representing a few consonant sounds not present in Urdu.<sup>31</sup>

<sup>28</sup> Emil Perder began field work on Dameli in 2003, through FLI, mainly with the aforementioned Hayat Muhammad Khan and his uncle Asmatullah Khan Dameli, also from Aspar, as co-researchers and informants.

<sup>29</sup> I am indebted to my friend and FLI colleague Fakhruddin Akhunzada of Kalkatak for arranging these interviews.

<sup>30</sup> The organization was registered 2014 under the new name *Palula Community Welfare Organization*.

<sup>31</sup> Words which are clearly identical to words in the Perso-Arabic stock are normally written in the traditional way, resulting in a fair degree of orthographic overrepresentation. The phoneme /s/ is, for example, represented by<س<in most native Palula vocabulary, but could be written as <س<,> ث <or <ص <if occurring in a loanword, depending on the particular spelling

### 1.7 Palula as a written language

The alphabet adopted is presented in Table 1.5, displayed along with sound correspondences in the same broad transcription ("Palula Common Transcription", or PCT) that is used throughout this work. Starting with a group of Urdu-literate people, the basic spelling principles were further discussed and applied in a writers' workshop in 2004. Following some fine-tuning of the orthography, the two first-ever Palula booklets were printed in 2006, one of them an alphabet book (Haider 2006a) <sup>32</sup> and the other a collection of short stories (Haider 2006b). In 2012, Forum for Language Initiatives and Anjuman-e-taraqqi-e-Palula published a collection of about 300 Palula proverbs and sayings, along with Urdu translations (Haider 2012b).

Through the Forum for Language Initiatives, the concept of multilingual education was introduced to the community in 2005. This eventually led to the establishment of a mother-tongue pre-school in Ashret proper in 2008, and somewhat later a second school in Kooḍghaá. More than 30 native Palula speakers, both men and women, took part in producing the curricula from scratch, representing a set of culturally relevant themes. In preparation for the first school year, a package was produced consisting of a pre-reader, a pre-writer, a primer, a collection of reading stories, a collection of listening stories, and a compilation of songs and rhymes, all in Palula. In 2010, a first batch of pupils completed two years of education in Palula, with gradual introduction to Urdu and English (Rehman & Sagar 2015).

The most visible modifications made to the existing Urdu alphabet in order to write Palula have to do with symbols representing a number of retroflex sounds: <ڇ <for *c*,*̣* <ڙ <for *ẓ*, <ݜ <for *ṣ*, and <ݨ <for *ṇ*. To write *ts*, the symbol<څ< has been borrowed from the Pashto alphabet. Aspiration as well as *h* occurring in clusters with voiced consonants (see Table 1.6) are represented by <ھ<.

Another modification relates to the representation of the ten vowels. Because of the many crucial contrasts between long and short vowels, particularly in word-final position, the developers of the Palula orthography introduced a language-particular use of a diacritic <ۡ> to mark a short vowel, as displayed in Table 1.7. Mainly due to challenges in the area of font development, Palula materials are exclusively produced in naskh style, not in nasta'liq.

in the donor language.

<sup>32</sup> A second edition was published in 2012 (Haider 2012a).

### 1 Introduction


Table 1.5: The Palula alphabet with corresponding transcription (PCT)

Table 1.6: Examples of Palula representation of "aspiration"



1.8 Remarks on transcription and glossing

### **1.8 Remarks on transcription and glossing**

The trancription system used for Palula in the main bulk of this work (always occurring with italics, for example, *paaluulaá*) largely corresponds with that in general use in South Asian linguistics and by most indologists (what Masica refers to as "Standard Orientalist", 1991: xv, and some others refer to as an "indological" system, Radloff 1999: 9). Since this is also the basic system (with very slight variations) used by other contemporary scholars of Shina varieties (Buddruss 1987; 1993; 1996; Degener 2008; Schmidt & Kohistani 2001; 2008; Schmidt 2000; 2001; 2003; 2004a,b; Radloff 1992; 1999; Radloff w. Shakil 1998) <sup>33</sup>, including the conventions for representing accent (for details, see §3.4.3), I have seen no reason to abandon it in favour of any other standard, such as a consistent use of IPA symbols, since any such decision would make inter-variety comparisons more troublesome and less straight-forward. I will, as mentioned (§1.7), refer to this broad transcription, taking only phonological contrasts into account, as *Palula common transcription* (PCT). The PCT symbols that differ from IPA notation are listed in Table 1.8.

However, to make the work accessible to the general typologist or readers not familiar with this particular system, IPA transcription (following International Phonetic Association, 1999) has been used (along with PCT within parentheses) for phonemic and phonetic transcriptions in §3.1–§3.4 (and occasionally elsewhere in the grammar), the former consistently surrounded by forward slashes, for example /paːluːlǎː/, and the latter by square brackets, for example [paːɫuːɫǎː].

When citing other sources or scholars on individual IA languages, I have in most cases kept the transcription used by them, without any attempts at making them conform to the particular conventions used in Palula examples.

I should hasten to add that PCT is not to be considered a community orthography. For that purpose the Arabic -based script (as described in §1.7) is used and promoted in the community.

Each glossed example sentence in this work consists (with a few exceptions) of three lines or analytical strings: 1) A transcribed utterance, 2) morpheme glossing, and 3) a free translation. The first line is a phonemic representation of the utterance, following PTC (as described above), with word breaks and morpheme breaks. It is, as far as has been possible, a regularised (within each of the two main dialects) surface form, i.e., a form not necessarily reflecting underlying morphology nor individual sub-phonemic variation. Some deviations from these

<sup>33</sup> Also the transcriptions in Hook (1990a; 1996) and Hook & Zia (2005) agree to a large extent with the principles applied by the other scholars.

### 1 Introduction


Table 1.8: Symbols used in Palula common transcription (PCT) deviating from IPA notation

principles have been applied when there is considerable (intra-dialectal) variation in the output form, such as in the representation of the present tense suffix *-áan* when following on an *a*-ending stem. In most cases, however, morpheme breaks (e.g., *phed-í* 'arrive-cv') have only been applied (i.e., signalled explicitly with a hyphen) when they line up with the surface form, whereas sequences of morphemic elements obscured by the surface form (e.g., *seé* 'sleep.cv') are only separated with a period in the glossing while represented as a single element in the transcription itself. On the second line, each morpheme is shown with a gloss, either in the form of a one-word English equivalent of a lexical morpheme, or in the form of an abbreviation of a grammatical category. The third line gives a free translation into English of the utterance as a whole; it strives at capturing the meaning in idiomatic English.

Slightly different principles have been applied to the two sample texts included. Instead of the three analytical strings of the glossed examples, there are four separate strings: 1) A transcribed utterance, 2) morpheme breakdown, 3) morpheme glossing, and 4) a free translation. The first line, again reflecting a regularised surface form, is only given with word breaks, no word-internal morpheme breaks. The second line has the transcribed Palula words broken down into morphemes. These are, to the extent it has been possible, displayed in an underlying morphemic form, which in some cases differ slightly from the output form of the first line. The third line, that of morpheme glossing, lines up with the morpheme breakdown of the second line. Frequently occurring markers have in this version been supplied with more specific glossing. The same principles as above have been applied to the free translation.

## **2 Typological overview**

This chapter is a brief overview covering the most central features of Palula. For more in-depth coverage of each topic, and for information on those not explicitly covered here, the later chapters will need to be consulted.

### **2.1 Phonology**

With its 32–37 members, the Palula consonant inventory (Table 2.1) is moderately large to large (Maddieson 2013a). There are five basic places of articulation (labial, dental, retroflex, palatal and velar), with a voicing contrast in the plosive and fricative sets, and an aspiration contrast in the plosive and affricate sets.


Table 2.1: Palula consonants

Palula has ten phonemic vowels, comprising five basic qualities, each having a long and a short counterpart. This inventory (Table 2.2) forms a symmetrical and typologically common system. Vowel nasalisation is a marginal, possibly emerging, feature in the language, but not so far fully contrastive.

The language has a complex syllable structure (Maddieson 2013b), permitting three consonants in the onset position and two in the coda position (although a limited number of consonant combinations are permitted before or after the

### 2 Typological overview

Table 2.2: Palula vowels

vowel nucleus, as shown in Table 2.3). There is a tendency to drop the final consonant in word final clusters.

Main stress falls on the final or the penultimate syllable of the lexical root. One of the vocalic moras of the stressed syllable receives pitch accent, phonetically realised as: a) high level or high falling on a short vowel [ˊ], represented in this work as *á* (in polysyllabic words, elsewhere no marking); b) low rising on a long vowel [ˇ], represented as *aá*; or c) high falling on a long vowel [ˆ], represented as *áa*. Pitch accent is contrastive, as illustrated in (1).


Table 2.3: Word boundary syllable clusters (vd = voiced; vl = voiceless)


2.2 Morphology

### **2.2 Morphology**

Palula morphology is suffixing, and formatives are almost exclusively concatenative (Bickel & Nichols 2013b), with a moderately high degreee of synthesis (Bickel & Nichols 2013c).

Nouns are inflected for number (singular, plural) and case (nominative, oblique, genitive). In most of the declensional classes, nominative plural and oblique singular are cumulated into a single formative (Bickel & Nichols 2013a). The genitive (at least in the plural) can be analysed as suffixed to the oblique rather than to the nominative stem. The noun exemplified in Table 2.4 is *ṣinɡ* 'horn'.

Table 2.4: Inflection of nouns


There are three main functions of the oblique case of nouns: a) as the transitive subject in the perfective (i.e., as an ergative case marker); b) as the form to which postpositions are added; and c) as a locative. A number of other case-like functions (such as recipients) and more peripheral arguments appear as postpositional phrases.

Palula displays core-case asymmetry (Iggesen 2013), within the category of nouns as well as for NPs in general (more on pronouns below). While most nouns (those belonging to the two major *a*- and *i*-declensions) make a nominative-oblique distinction, one declensional class (the *m*-declension) in particular does not make this distinction at all, whereas some of the pronouns make an even more fine-tuned nominative-accusative-oblique distinction, as seen in Table 2.5.

Table 2.5: Core case distinctions


### 2 Typological overview


Table 2.6: Pronominal case distinctions

Palula has a fairly typical Indo-European two-gender system, which is primarily sex-based (Corbett 2013a). A noun is either masculine or feminine, a property established through morphological agreement. The basis for gender assignment is semantic as well as formal (Corbett 2013b).

The pronoun system proper (i.e., 1st and 2nd person) is interesting in that it makes more distinctions in the plural than in the singular (Table 2.6), as there are dedicated ergative case forms only in the plural.<sup>1</sup>

The demonstratives, which are used as third-person pronouns, essentially make the same case distinctions as the plural personal pronouns (although there are other uses of the oblique apart from its ergative function). Additionally, they display gender distinctions (in the nominative singular) as well as a three-way deictic contrast (Table 2.7).


Table 2.7: Demonstrative distinctions

<sup>1</sup> The forms *ma* and *tu* are glossed throughout this work as nominative and *míi* and *thíi* as genitive, regardless of their functions in the clause.

2.3 Syntax

Table 2.8: Inflection of adjectives


Adjectives are inflected for agreement in gender (masculine, feminine), number (singular, plural) and case (nominative, non-nominative). The adjective in Table 2.8 is *paṇáaru* 'white'.

Finite verbs are inflected for tense(-aspect), mood (in a limited sense) and agreement in a) gender/number, *or* b) person (the type of agreement expressed depending on tense-aspect, see §2.3 below). There are also some non-finite forms. For the sake of a more economical presentation that takes verbs of different inflectional classes into account, all verbs are analysed as having a perfective and a nonperfective stem. The verb in Table 2.9 is *til-* 'walk'.

### **2.3 Syntax**

Four of the most frequently occurring TMA-categories (Future, Present, Simple Past (=perfective) and Imperative) make use of inflectional morphology only (as displayed in Table 2.9).<sup>2</sup> Another three basic TMA-categories (Past Imperfective, Perfect and Pluperfect) are expressed periphrastically, by adding auxiliaries to inflected verb forms (as shown in Table 2.10).

There is relatively little synchronically productive derivational morphology in the language, but a productive process for deriving verbs from other categories from within the language as well as from entirely novel or non-native elements is the use of verbalisers such as *the-* 'do' and *bhe-* 'become'. Some examples are shown in (2).


<sup>2</sup> Language-specific, and functionally defined, verbal categories are capitalised to set them apart from grammatical terms applied cross-linguistically. See Chapter 10 for details.

### 2 Typological overview


### Table 2.9: Inflection of verbs

Table 2.10: Periphrastically formed TMA-categories


Word order in Palula is typically head-final (Table 2.11). This is seen in the word order in noun phrases (determiner–noun, adjective–noun, numeral–noun, genitive–noun), adjective phrases (adjunct–adjective) and in adpositional phrases (noun phrase–postposition). As far as entire clauses are concerned, the word order (or rather constituent order) is more flexible, but the most frequent and pragmatically unmarked order is intransitive subject–verb, transitive subject– verb, and direct object–verb.

2.3 Syntax


Table 2.11: Word order features

As far as alignment is concerned, Palula displays an intricate split system. In the perfective categories (Simple Past, Perfect and Pluperfect), the pattern is essentially ergative, as seen in example (3), with a non-nominatively marked agentsubject and verbal agreement with the feminine direct object. In the non-perfective categories (Future, Present and Past Imperfective), in contrast, it is essentially accusative, which can be observed from the nominatively marked agentsubject in (4), which is also the NP that the transitive verb agrees with in gender and number.

(3) *íṇc-ạ* bear[msg]-obl **A** *čhéeli* she.goat[fsg] **O** *khéel-i* eat.pfv-f **V** 'The bear ate the goat.' (A:PAS056)

(4) *iṇc̣ áaṇc-ạ kha-áan-u*

bear[msg] **A** raspberry[m]-pl **O** eat-prs-msg **V**

'The bear is eating raspberries.' (A:KAT145)

Agreement is part of all finite verb forms, but the particular agreement features realised are related to tense-aspect. In Future and Past Imperfective, the verb agrees with its target in person (and number), as in (5), whereas in Present and the categories based on the perfective, the verb agrees in gender and number, as can be seen in (6).

### 2 Typological overview


(5) *so* 3msg.nom **A** *múṛee* dead.person.pl **O** *ǰand-óo* make.alive-3sg **V** *de* pst

'He was resurrecting the dead.' (A:ABO034)

(6) *táapaṛ-a* hill-obl *túuri* below *íṇc-ạ* bear[msg]-obl **A** *čhéeli* she.goat[f] **O** *ɡhašíl-i* catch.pfv-f **V** *hín-i* be.prs-f

'Below the hill, the bear has captured the goat.' (A:PAS054)

Alignment in the realm of verbal alignment is summarized in Table 2.12. Several NP splits further complicate the picture. Apart from the singling out of the transitive subject (A) in the perfective (*asím* in (7)), we also have pronominal forms particular to the direct object (O) (*asaám* in (8)), both of them different from the form used as the subject (S) of an intransitive clause (*be* in (9)).


2.3 Syntax


Table 2.13: Alignment: Case marking

Alignment in the realm of case marking is summarized in Table 2.13. This is a somewhat simplified overview in that some minor noun declensions (in which some case distinctions are upheld only in the plural) are not included.

Sentences lacking an overt copula are allowed, and for predicate nominals in the Present tense, as the one shown in (10), they are typical.

(10) *míi* 1sg.gen *báabu* father *áak* idef *zamindaár* farmer *míiš* man 'My father is a farmer.' (A:OUR002)

Although it is possible to conjoin clauses with a conjunctive suffix (also used for conjoining noun phrases), other strategies are preferred, such as juxtaposition for symmetrical clauses or the overwhelmingly favoured Converb construction, exemplified in (11), which is used for a great variety of same-subject clause combinations.

(11) *tíi* 3sg.obl *ba* top [*bhun* down *wha-í* come.down-cv *ba*] top [*so* def.msg.nom *mhaás* meat *muṭ-í* tree-gen *bhun* down *wheel-í* take.down-cv *ba*] top [*teeṇíi* refl *ɡhooṣṭ-á* house-obl *the* to *ɡhin-í*] take-cv *ɡáu* go.pfv.msg 'He came down [having come down], took down the meat from the tree

[having taken down the meat from the tree], and brought it to his house.' (B:SHB762)

In complex constructions, the unmarked order is a complement clause followed by (or embedded in) the main clause, as in (12), and similarly an adverbial clause followed by (or, again, embedded in) the main clause, as in (13).

### 2 Typological overview


However, a post-posed construction with the complementiser *ki* is also commonly used (14), especially for utterance complements.

(14) *ɡhueeṇíi-am* Pashtun-pl.obl *maníit-u* say.pfv-msg *ki* comp *ni* 3pl.prox.nom *bíiḍ-a* much-mpl *zinaawúr* wild *xálaka* people *hín-a* be.prs-mpl 'The Pashtuns said, "These are very wild people."' (A:CHA008)

Polar interrogatives are formed with a clitical sentence-final question particle *ee* (B *aa*), as in (15), whereas an indefinite-interrogative pronoun (or other proform), such as *kasée* (B) 'whose' in (16), is used in content interrogatives.


Negation is formed with a separate and invariable negative particle *na*, preceding the predicate (17).

(17) *muṣṭúk-a* of.past-mpl *xálak-a* people-pl *dhii-á* daughter-obl *díi* from *na* neg *khooǰ-óon* ask-3pl *de* pst 'People in the old days were not asking their daughter [who she wanted to marry].' (A:MAR018)

## **3 Phonology**

### **3.1 Consonants**

### **3.1.1 Consonant inventory**

The consonant inventory is rather symmetrical, with the dental and retroflex places of articulation displaying the most well-developed system of manner contrasts. The ancient (OIA) contrast between the three sibilants /s ʂ ɕ/ is preserved (Cardona & Luraghi 2009: 375), with the present voicing contrast probably not evolving until quite recently, partly through lenition of voiced affricates, partly through foreign loans.

While the plosive and fricative sets show a contrast in voicing except for in


Table 3.1: Inventory of consonants (IPA). Marginal or doubtful phonemes within parentheses

### 3 Phonology

the (marginal) labial and glottal places of articulation, voiced counterparts are missing in the small affricate set. Mirroring that is a general allophonic variation (see below) between voiced fricative and affricate pronunciations. The voiced palatal fricative could equally well be treated as an affricate, as that is the more common allophone (especially in the A dialect), but to provide more symmetry to the system, I have chosen to include it among the voiced fricatives,<sup>1</sup> while in the common transcription it is represented as *ǰ*.

The post-velar (or uvular) place of articulation is represented by a voiceless post-velar or uvular plosive /q/ alone. This marginal phoneme is only pronounced distinctly post-velar by some educated speakers – and even then rather inconsistently – when occurring in loanwords of mainly Perso-Arabic origin. In many speakers' pronunciation, however, it normally tends to approximate a velar fricative pronunciation [x], thus not contrasting with the phoneme /x/. The fricatives /z, x, ɣ/ are rather frequent in present-day Palula, and many of the words probably have a long history in the language, although they to a large extent are found in vocabulary borrowed from languages in the immediate region, and to a much lesser extent are found in inherited vocabulary affected by phonological processes. A labio-dental [f] is sometimes heard in more recent loans, primarily from Urdu and English, but with many speakers it alternates freely with or is entirely replaced by the native voiceless aspirated bilabial plosive /pʰ/, hence /f/ is considered a marginal phoneme.

The voiced retroflex fricative /ʐ/ is also a marginal phoneme, but it is included for comparative reasons; an even more rarely occurring voiced retroflex affricate sound [ɖʐ] is tentatively analysed as an allophone of the same phoneme.

There is insufficient proof to regard a velar nasal [ŋ] as a phoneme independent from /n/, as it only occurs before /k/ and /ɡ/, or as a variant pronunciation of /nɡ/: [ŋɡ]*∼*[ŋ].

Although initial findings identified several voiced aspirated consonants, later observations favoured a cluster analysis, e.g., /b/ followed by /h/ rather than a phoneme /bʰ/. However, it should be noted that voiceless aspirated sounds do share some characteristics with (frequently occurring) clusters of voiced consonants and /h/, as will be further discussed in §3.4.1.

<sup>1</sup> As pointed out by Zoller (2005: 34), this particular asymmetry within the affricate/fricative sets is a feature shared by a number of languages of northern Pakistan, due to a "lenition process which is more advanced in case of the voiced phonemes than in case of the voiceless phonemes".

3.1 Consonants

### **3.1.2 Distribution and variation**

Examples of the distribution of consonants are shown in Table 3.2.

The retroflex consonants are in some descriptions called "retracted" (Schmidt & Kohistani 2008: 16) or "cerebrals" (Morgenstierne 1941); it has been questioned whether these consonants in HKIA languages are retroflex in the same sense or to the same extent as in the main NIA languages or in Dravidian languages.

I am presently in no position to determine the exact nature of retroflexion in Palula, but I prefer, nevertheless, to retain the term, as the most prominent feature in the pronunciation of these consonants is the articulation with the tip of the tongue against a place at the rear end of the alveolar ridge and usually with the tongue slightly curled back. The dental consonants on the other hand are indeed dental, often articulated against the lower as well as the upper teeth. Generally, the area of contact between the tongue and the place of articulation is larger than in the case of the retroflex consonants.

The palatal consonants can also be described as alveolo-palatal, with the blade of the tongue against the area covering the rear part of the alveolar and the front part of the palate, and with the tip of the tongue behind the lower teeth.

### **Plosives**

The set of plosives includes: /p/: [p], /pʰ/: [pʰ]*∼*[f], /b/: [b], /t/: [t], /tʰ/: [tʰ], /d/: ̪ [d], /ʈ/: [ʈ] ̪ *∼*[ṯ], /ʈʰ/: [ʈʰ], /ɖ/: [ɖ]*∼*[ḏ], /k/: [k], /kʰ/: [kʰ], /ɡ/: [ɡ], (/q/: [q]*∼*[x]).

With respect to frequency, the voiceless plosives can be considered the unmarked subset, occurring almost twice as often as their voiced counterparts. The voiced plosives do not commonly occur word-finally, and when they do they tend to be devoiced, as in /ɕid/ *(šid)* 'coldness': [ɕid̪ ̥ ]. Voiceless aspirated plosives occur in the majority of cases word-initially, only seldom word-medially, and never (as far as has been determined) in word-final position.

Intervocalically, the voiced plosives are often slightly fricativised. They frequently occur in clusters with /h/ (see §3.4.1). Some of the voiceless aspirated plosives show lenition. An example is /pʰ/, which has an alternating pronunciation [f]*∼*[ɸ]*∼*[pʰ], as in /pʰeːrimaː/ *(pheerimaá)* 'Ferima (place name)'.

The phonemic status of [q] was already commented on above (see §3.1.1).

<sup>2</sup> In Palula common transcription these would be: *píili, šópu, ṭip, phoó, aḍaphaár, bíiḍi, ǰabál, ḍáab, fásil, xafá, muaáf, míiša, hiimaál, bráam, wíiwai, heewaánd, ɡhaáu, reetí, béeriṣ, anɡóor, léedi, balíi, čhaál, téeti, phútu, baát, thúuṇi, suthaán, déedi, léedi, šid, néeṛi, ǰáanu, sóon, seetí, búusi, deés, zeerí, baazóor, anɡreéz, tsiipí, bútsu, uts, batsaár, šéemi, húuši, díiš, čéeri, kučúru, baalbáč, čhéeli, pičhaá, ǰéeli, béeǰi, ráaǰ, yíiṛi, lháaya, babaái, ṭaaká, beeṭí, báaṭ, ṭhónɡi, buṭheé, ḍaaká, ɡéeḍi, haáḍ, déeṇi, bheéṇ, déeṛi, kiroóṛ, ṣéeti, khaṣíi, báaṣ, ẓamí, ẓaṇẓá, ríiẓ, ciinkí, téec ̣ i,̣ dhraác, c̣ hiír, ac ̣ híi, buc ̣ h, katí, bakáara, ḍóok, khur, nikhaí, ɡaḍí, síɡal, phaáɡ, xatí, maaxaám, ̣ mux, qisá, alaaqá, ašáq, ɣeerí, kaaɣaáz, baáɣ, harí, kuhíi*.


### 3 Phonology

Table 3.2: The

distribution

 of

consonants:

 medial,

 and final.

2 The

occurrences

 within

 are

parentheses

word-initial,



3.1 Consonants

### 3 Phonology

### **Affricates**

The set of affricates includes: /ʦ/: [ts], /ʦʰ/: [tsʰ]*∼*[s], /ʈʂ/: [ʈʂ], /ʈʂʰ/: [ʈʂʰ]*∼*[ʂ], /ʨ/: [tɕ].

Affricates occur at three places of articulation, dental, retroflex and palatal, but with respect to frequency the dentals are quite limited as compared to the other two. The explanation of the missing voicing contrast is partly explainable (as already commented on above) by the overlap or neutralisation between the affricate and fricative sets.

There is also a less consistent neutralisation of the contrast between aspirated dental (1) and retroflex voiceless (2) affricates and their fricative counterparts (but as far as I have been able to observe, never between aspirated voiceless palatal affricates and fricatives), apparently limited to certain lexical items.


### **Fricatives**

The set of fricatives includes: (/f/: [f]*∼*[pʰ]), /s/: [s], /z/: [z], /ʂ/: [ʂ]*∼*[ʈʂʰ], /ʐ/: [ʐ]*∼*[ɖʐ], /ɕ/: [ɕ], /ʑ/: [dʑ]*∼*[ʑ], /x/: [x], /ɣ/: [ɣ], /h/: [ɦ]*∼*[h].

As already pointed out in connection with the affricates, there is a close link between the affricate set and the fricative set, with some overlaps and neutralisations taking place between them. The voiced palatal fricative is alternately realised as [ʑ] and [dʑ] (more often with an affricate pronunciation in A) and the voiced retroflex as [ʐ] and [ɖʐ], whereas /z/ seems to occur consistently as [z] and never with an affricate pronunciation.

The marginal phoneme /f/ is often realised as [pʰ], thus neutralised with /pʰ/, as in /faːjda/ *(faaidá)* 'benefit': [pʰaːjdá]*∼*[faːjdá].

The voiced retroflex fricative is extremely rare, occurring only in a few words. [ɖʐ] is most likely an allophone of it, as in /ʐʰaɳʐiːr/ *(ẓhaṇẓíir)* 'chain': [ɖʐʰaɳɖʐ îːɾ].

There is a strong affinity between /h/ and historical aspiration (§3.4.1), especially when occurring in clusters of voiced consonants and /h/, in which case it is mostly realised as [ɦ]. Historical occurrences of word medial /h/ through

3.1 Consonants

movement to syllable onsets have most likely been reinterpreted as voiced aspiration. In the present language, a single /h/ only rarely occurs intervocalically, and even then often with an interpretational ambivalence: *(rhayíi)* 'footprints': /rhajiː/ or /rahiː/.

### **Nasals**

The set of nasals includes: /m/: [m], /n/: [n̪]*∼*[ɲ]*∼*[ŋ], /ɳ/: [ɳ].

Phonetically there are at least five places of articulation attested for nasals: labial, dental, retroflex, palatal and velar. The palatal nasal, however, is analysed as deriving from a sequence of /n/ + a palatal consonant, as it never occurs in any other environment. The same analysis may be applied to the velar nasal, where the sequence /n/ + a velar stop usually is the likely source.

The case is a little more complicated with the retroflex nasal, /ɳ/. Although it is clear in some cases that retroflexion is the result of assimilation with an adjacent retroflex consonant, this cannot always be concluded. Whereas a retroflex nasal normally does not occur word initially (although the word /ɳiɳeː/ *(ṇiṇeé)* 'popcorn' can be cited as an isolated counterexample), it contrasts intervocalically with dental /n/, compare /deːni/ and /deːɳi/ in (3), and word-finally, /kan/ and /kaɳ/ in (4).


The labial and the dental nasals are very frequent in the language, as these two segments are part of some of the most productive inflectional forms in the language.

### **Flaps**

The set of flaps includes: /r/: [ɾ], /ɽ/: [ɽ].

While /r/ commonly occurs word-initially, intervocalically, and word-finally, the occurrence of /ɽ/ is more restricted. In B it does not occur word-initially at all, whereas in A it occurs in free variation with /l/ in weak forms of a series of

3 Phonology

demonstratives (5), for example *(lo)* or *(ṛo)* 'he, that', related to strong forms of the same series with an intervocalic /ɽ/, *(eeṛó)* 'he, that', but otherwise not.

(5) *lo∼ṛo* /lo/*∼*/ɽo/ 'he, that'

### **Lateral approximant**

There is one lateral approximant: /l/: [l](*∼*[ɫ] B).

Preceded by a back vowel /a aː o oː u uː/, /l/ is velarised, but only markedly so in the B variety: compare non-velarised *khéeli* and velarised *khúulu* in (6).

(6) *khéeli – khúulu* /kʰeːli/ [kʰêːli] – /kʰuːlu/ [kʰûːɫu] 'ate fsg' – 'ate msg' (B)

### **Approximants**

The set of approximants includes: /w/: [β̞]*∼*[ʋ], /y/: [j].

In the speech of my main A consultant, the front-most approximant /w/ is usually pronounced bilabially [β̞], but with many speakers this phoneme seems to alternate between a bilabial and something close to a labiodental [ʋ] pronunciation.

The two approximants are sometimes challenging to interpret, whether they should be regarded as consonants or vowels, and they are in various ways susceptible to articulatory fluctuation or variation, especially when occurring intervocalically, an issue that will be further discussed in connection with vowels (see §3.2.4).

### **3.2 Vowels**

### **3.2.1 Vowel inventory**

For the vowels, there are five contrasting places of articulation, as can be seen in Table 3.3: a) close front, b) close back, c) open front, d) open back rounded, and e) open back unrounded. Together with phonemic length contrasts, there is a tenvowel system. A convincing and consistent contrast (as the one shown by Radloff (1999: 19) for Gilgiti Shina) between oral and nasalised vowels has not been found.


Instead, nasalisation seems to be a marginal suprasegmental feature of a limited number of lexemes. Apart from those, nasalisation is a non-contrastive phonetic property of vowels occurring adjacent to a nasal consonant.


Table 3.3: Inventory of vowels (IPA)

### **3.2.2 Distribution and variation**

Table 3.4 exemplifies target articulations of the vowels, all of which take on more centralised qualities in natural and connected speech. Generally, the short vowels /i/, /a/, and /u/ tend to be pronounced as less peripheral than their long counterparts. The short /i/ is not necessarily more open than the long /iː/, but it has a rather more central pronunciation; the short /u/, on the other hand, is both more open and slightly more central than the long /uː/; the short /a/ is also slightly less open and more fronted than the long /aː/.

Phonetically, there is a significant difference between short and long vowels. The duration of a long vowel like /aː/ as compared to its short counterpart /a/ is not just slightly longer, but usually of at least twice the duration.

Environment as well as accent (see §3.4.3) further influences the exact pronunciation of each of the ten vowels. Under certain conditions, some neutralisations take place (see §3.2.3).

As pointed out already by Morgenstierne (1932: 58), the most important – if not all – phonological dialect differences between A and B concern the vowels rather than the consonants.

<sup>2</sup> In Palula common transcription, the words are: *ɡir, ɡiír, ṭíki, tíiṇi, preṣ, keéṇ, ṭéka, teeká, šak, káaṇ, ṭáka, ṭaaká, sum, kúuṇ, thúki, thúuṇi, khoṇḍ, kóoṇ, tróki, ṭooká*.

### 3 Phonology

Table 3.4: Vowel contrasts exemplified (see §3.4.3 for details on pitch accent)<sup>3</sup>


### **3.2.3 Vowel neutralisation**

While there is a consistent contrast between all the five vowel qualities as well as a contrast in length, when the vowels are accented (see §3.4.3 for details on accent), these contrasts are fewer and less convincing when the vowels are unaccented. The two main dialects also show some differences in this regard.

Whereas B maintains a word-final /a/ vs. /e/ contrast – as is clearly evidenced in the morphological contrast between the general oblique inflection *-a*, as in /ˈdiːɕa/ 'in the village', and the genitive singular *-e*, as in /ˈdiːɕe/ 'of the village', of many masculine nouns – there is no evidence of contrast between these two unaccented short vowels in A (where the unaccented genitive ending instead is /iː/). In contrast, a non-variable masculine ending [u] in B, corresponds to two different (but grammatically identical) masculine endings [o] and [u] in A. Curiously, the realisations of these two variants are in complementary distribution, although there is no obvious phonological motivation behind it. When preceded by /aː/ (in the previous syllable), the allophone is [u], while it is [o] when preceded by any other vowel, as can be seen in (7).

(7) *paṇáaru –* [paɳâːɾu] – *tóoru –* [tôːɾo] – *bhíiru –* [bɦîːɾo] – *bhúuru –* [bɦûːɾo] – *léku* [léko] 'white msg' – 'star [m]' – 'he-goat [m]' – 'deaf msg' – 'small msg'

As a consequence of these observations, all instances of unaccented word-final [u] and [o] are consistently transcribed as *u* in the Palula common transcription, whether A or B, and only word-final unaccented *a* occurs in A examples, not *e*.

3.2 Vowels

### **3.2.4 The status of diphthongs**

A complex issue still needing more careful study concerns the interpretation and representation of ambiguous vowel sequences. However, for the time being there is no strong evidence for stipulating any phonemic diphthongs with a status comparable to that of the ten vowels already introduced.

The sequences of vowels in lexical stems all consist of at least one close vowel (also interpretable as an approximant), such as [ai], [aːi], [ui], [oːi], [oi], [eːi], [ueː], [uaː], [iaː], [ioː], [aːu], [au]. Probably most, if not all, combinations of a short close vowel and another long or short vowel are possible. Some examples are given in (8).


Taking a number of factors into account, such as mother-tongue speakers' counting ("knocking") syllables, the apparent absence of sequences not including any of the two close vowels [i] and [u], and the evidence for approximants occurring word-initially as well as intervocalically (and therefore if not being interpreted consonantally leaving a gap word-finally), would favour an approximant interpretation, which would render the following phonemic output: /brhaːˈʑaj/, /baˈbaːj/, /ʑaˈbuj/, /bhoːj/, /ˈlhojlu/, /jeːj/, /kakˈweːki/, /ˈswaːl/, /taːˈpjaːɽ/, /pʰjoːɽ/, /ɡhaːw/, /maɳˈɖaw/.

However, in a number of words with a vowel + [i] sequence, the final [i] can be considered a feminine gender suffix (in some cases derived by that suffix, at least diachronically, from a masculine stem), and in the morphological behaviour of monomorphemic stems, such as those exemplified above, it is an advantage to show that there is an underlying vowel /i/ or /u/ (rather than a consonant) involved. Therefore, I have chosen to represent them as *bhraaǰái*, *yéei*, etc., in

### 3 Phonology

Palula common transcription, to signal precisely the connection between a stem and its derivations or inflected forms.<sup>4</sup>

The latter representation makes even more sense for sequences in polymorphemic words, such as *dhióomii* 'of the daughters' from *dhií* + *-óom* (obl.pl) + *-ii* (gen), although the surface phonemic representation would be /djhoːmiː/, the latter taking de facto syllabification into account at the expense of morphemic transparency. This holds for inflected forms of verbs as well: A purely phonemic representation such as /swâːnu/ 'is sleeping msg' obscures the fact that we have the verbal stem *só-* 'sleep' inflected for present tense with *-áan*, and therefore a Palula common transcription *suáanu* has been chosen for it.

When, on the other hand, there is a need to show that there indeed is a syllable break between two successive vowels, whether the word is mono- or polymorphemic, an approximant, *y* or *w*, is inserted: *bhooyóomii* /bhoːˈjoːmiː/ 'of the daughters-in-law', and *bharíiwa* /bhaˈriːwa/ 'husbands'.

### **3.3 Phonotactics**

### **3.3.1 Syllable structure**

A typical syllable in Palula is an open syllable consisting of a consonant and a vowel. This is the most common type when the syllable is unaccented. Long, as well as short vowels (9) could constitute the nucleus of such a syllable: CV or CVV. There are monosyllabic words (/be/, /wiː/) which conform to this basic CV pattern, but most words are polysyllabic, consisting of two or more CV (or CVV) syllables (such as /ɡuː.li/ and /ku.ɳaː.koː.miː/).


The closed-syllable pattern, CV(V)C, is also a very common syllable, see examples in (10), and the most common one in accented syllables. This type occurs in monosyllabic as well as in polysyllabic words. Commonly, however, a word is made up of a combination of open and closed syllables.

<sup>4</sup> Although not attempted here, an alternative analysis of [ái] would be to consider it an allophone of first-mora accented /ée/.

3.3 Phonotactics


Even onsetless syllables, V(C) or VV(C), occur in Palula (11), though less frequently. That means that both the onset and the coda is optional, i.e., a vowel nucleus can occur alone or at least word-initially. Whether this is also possible word-medially or word-finally is an interpretational issue, but in any case, there are no single phonological words consisting of only a vowel nucleus.<sup>5</sup>


The minimal word can therefore be defined as consisting of a vowel nucleus plus either an onset or a coda consonant. There seems also to be further constraints on words belonging to the major open classes as opposed to words from closed classes when it comes to minimal words. Nouns, adjectives and verbs (except for imperative forms and a few participle forms) must consist of at least a short vowel plus a coda, or an onset plus a long vowel. Pronouns, on the other hand, may very well consist of only a short vowel with an onset: /ma/ 'I', /be/ 'we', etc.

### **3.3.2 Consonant clusters**

The preservation of a number of clusters, especially some that occur word finally, sets Palula off as more conservative than most other Shina varieties. In addition, a set of changes, at least partly related to, on the one hand, vowel metathesis and re-syllabification, and on the other hand, laryngeal metathesis and the subsequent reinterpretation of what was earlier voiced aspirates (see §3.4.1), have produced a few new, primarily word-initial, clusters.

There is a maximum onset of three consonants in Palula words, as can be seen in (12). These are clusters of voiced consonants only, whose third member al-

<sup>5</sup> An alternative analysis not attempted in this work would be to regard a glottal stop as a consonant phoneme preceding all vowels that are here considered word-initial, thus doing away with onsetless syllables altogether.

### 3 Phonology

ways is /h/ (phonetically realised as [ɦ]). One type, whose middle member is /r/ preceded by a plosive, go back to old (or secondarily formed) voiced aspirates followed by /r/. The aspirates have been reinterpreted as plosive + /h/ clusters, and a subsequent realignment has taken place, whereby the more sonorant /h/ has changed to the position closest to the syllable nucleus: /brhoː/ 'brother' < \*/bhraː/ < *bhrā́tr̥-*. Another type, whose middle member is one of the two approximants /j/ or /w/ (if we go with the analysis presented above, §3.2.4) preceded by a plosive or a nasal, have arisen through de-syllabification of a short unaccented closed vowel, in some cases subsequent to vowel metathesis, such as is the case (in A) with /ɡwheːɳiː/ 'Pashtun' (< /ɡhweːɳiː/ < /ɡhueːɳiː/ < /uɡheːɳiː/, the latter which is still the form heard in the conservative B dialect). That other clusters similar to the last-mentioned type seem to be in the process of evolving is evidenced by co-existing forms: /ukʰaːndu/*∼*/kʰwaːndu/ *(ukháandu)* 'is coming/going up msg', perhaps pointing to a preference for Cw and Cj clusters vis-à-vis word initial V-syllables.


Initial two-consonant clusters, see (13), share many of the features already mentioned for three-consonant onsets. The second member of such a cluster is either /r/ (most of them of considerable age), an approximant (with a recent history, derived along the same lines as was presented above for /ɡwheːɳiː/), or /h/ (which is either historical voiced aspiration reinterpreted as a cluster, or a new initial cluster arisen through laryngeal metathesis). Usually, but not exclusively, /r/ is preceded by a plosive (in the majority of cases a voiceless one). Voiceless aspirated plosives in clusters with /r/ are rare indeed, the verb /pʰrajaːnu/ 'send' is the only example found so far in the data with a following /r/. Approximant may be preceded by plosive, fricatives or nasals. Nearly any voiced consonant may precede /h/ in initial clusters (the only exceptions in the data being the "new"phoneme /ɣ/, and the distributionally limited consonants /ɳ/ and /ɽ/).

### 3.3 Phonotactics


Two-consonant clusters in coda position, see (14), seem to be subject to a much higher degree of variability, although the position also seems to be slightly more permissive than the onset. The more frequently occurring type observed at word boundaries consists of nasal + plosive/affricate/fricative, the other types being more marginal in occurrence.


The final affricate or fricative is always articulated, even if sometimes only weakly, whereas the nasal (homorganic with the affricate or fricative) is sometimes – more with some speakers than others and depending on word emphasis – phonetically absent but leaves a trace of nasalisation on the preceding vowel. Even the nasal + plosive sequences are subject to much variability. With some speakers and dialects, one of the phonemes in the sequence is altogether absent, sometimes the nasal (then leaving the preceding vowel nasalised, for example in

### 3 Phonology

/naːnɡ/ (náanɡ) 'finger or toe-nail' B: [n̪ã̂ːɡ]), sometimes the stop (for example in /ɕaːnɡ/ (šáanɡ) 'branch' B: [ɕâːŋ]), whereas in the corresponding inflected forms the stop would never be omitted: [n̪ûːŋɡa] (núunɡa), [ɕûːŋɡa] (šúunɡa). This especially pertains to the /n + d/ sequences in singular nouns, where it seems to be more rule than exception that the final stop is dropped, especially in B, whereas these are clearly articulated when occurring non-finally in the inflected forms: /dan/ 'tooth', but /daːnda/ 'teeth'; /pan/ 'path', but /paːnda/ 'paths'.

In the final /tr/-cluster, the /r/ is present as a segment in the speech of all my informants, but its articulation is not exactly identical to its non-final allophones (as in the inflected forms of the same lexical items). There is a strong tendency for it to be pronounced with very little energy, almost always being devoiced and sometimes also followed by an optional very short schwa-like sound: [putɾ̥(ə)] in /putr/ 'son'.

As for the realisation of the final /st/ and /ʂʈ/-clusters, there are differences between different speakers, and possibly between different dialects as well. My B informants tend to articulate both members of the cluster, even in final position, though the plosive is somewhat softened, whereas my A informants seem to prefer to omit the plosive altogether in final position, e.g [ɡɦǒːʂ] 'house', [n̪aːs] *(náas)* 'nose'. However, in all speech varieties both the fricative and the plosive are clearly present when occurring medially, i.e., in the corresponding inflected forms: [ɡɦoːʂʈá], [nastí].

A special case is the final cluster /ndr/ in /jaːndr/, see (15). This is the only threeconsonant cluster at a word boundary discovered so far, but its exact phonetic realisation is not entirely easy to define in terms of segments. With some speakers, the /n/ is clearly articulated, whereas the /dr/ part is only faintly present, and in other pronunciations the final /r/ gets a schwa-like sound attached to it, in practice making /dr/ the onset of an additional syllable. As with the above-mentioned clusters occurring at the end of singular nouns, the same cluster stretching over a syllable boundary in an inflected form of the same noun is clearly and unambiguously articulated: /jaːn.dra/ 'mills'.

### (15) /-ndr/ /jaːndr/ *(yáandr)* 'mill'

Clusters occurring in syllable onsets and syllable codas intervocalically, see (16), are subject to the same restrictions as the clusters at word boundaries described above, but a few of them, particularly those containing /h/, are extremely rare in that position.

3.4 Suprasegmentals


Apart from those, a number of other consonant clusters not permitted at wordboundaries are found word-internally. However, in all of those cases the clusters are analyzable as occurring across a syllable boundary (and not seldom across a morpheme-boundary as well). Only a few examples are shown in (17).


### **3.4 Suprasegmentals**

### **3.4.1 Aspiration and breathiness**

In the present description, seven consonants, all voiceless, with aspiration as a secondary articulation have been postulated: /pʰ/, /tʰ/, /ʈʰ/, /kʰ/, /ʦʰ/, /ʈʂʰ/, and /ʨʰ/. The two affricates /ʦʰ/ and /ʈʂʰ/ are treated with a higher degree of tentativeness, as their contrastiveness vis-à-vis their non-aspirated counterparts is far less convincing than is the case with the other five aspirated segments, and would deserve further, more detailed, investigation.<sup>6</sup> There are, however, certain characteristics (synchronic as well as diachronic) that the voiceless aspirated sounds share with clusters of voiced consonants and /h/ (as described in 3.3.2), suggesting an alternative (or perhaps complementary), unitary treatment of aspiration as a feature of a word (or more correctly the lexical stem), as is already reflected in the consistent use of *h* in the Palula common transcription.<sup>7</sup>

<sup>6</sup> At least as far as /ʈʂ/ is concerned, it seems, it is almost by "default" more or less clearly aspirated. A similar hesitation has been expressed on the /ʈʂ/-/ʈʂʰ/ contrast in Khowar (Endresen & Kristiansen 1981: 239).

<sup>7</sup> It is also possible that a more careful analysis will result in all instances of /h/ (here considered a consonant in its own right when occurring alone in syllable onset) being treated in the same way.

### 3 Phonology

This feature, realised as [ʰ] with a voiceless consonant and [ɦ] (or [ʱ]) with a voiced consonant, occurs only once in a word<sup>8</sup> (in a majority of the cases wordintially). Some minimal pairs in (18) illustrate the contrastiveness of this feature.


The vowel immediately following /h/ occurring in clusters with voiced consonants is normally phonetically realised with (at least partially) breathy voice: [bʱo̤ːla].

Most voiced consonants can occur in clusters with /h/, from plosives to approximants (as described in 3.3.2). This generous occurrence of aspiration (in the wider sense) is not a feature of most other languages in the immediate region, possibly with the exception of Indus Kohistani, where OIA aspiration, like in Palula, has been preserved and where aspiration is concomitant with most of its consonants (Hallberg & Hallberg 1999: 19–25).

Whereas the phonetic realisation of the aspiration with the voiceless consonants is more or less equal to a secondary pronunciation of the voiceless segment, the "breathiness" affecting the pronunciation of the following vowel is somewhat mobile within the syllable, and for some words even beyond the realm of that syllable. Especially in B, there is a fluctuation in some words, as seen in (19) and (20), between a realisation as a single intervocalic /h/-segment and the occurrence of /h/ as part of a word-initial cluster, as described above, the intervocalic /h/ probably representing an older pattern.<sup>9</sup>

<sup>8</sup> The process of dissimilation of aspirates in two successive syllables is known as *Grassman's Law* within Indo-European historical linguistics (Szemerényi 1996: 19, 56; Lehmann 1992: 153– 154, 162–163) and has been applied to the development of OIA as well as Greek. A synchronic process or rule restricting the occurrence of aspirated sounds to one per word has also been stipulated for other NIA languages (see for example Losey (2002: 32) for Gojri, and Shackle (1976: 34–35) for Siraiki). This process, however, is not confined to Indo-European languages: compare Tibeto-Burman Manipuri (Bhat & Ningomba 1997: 13–14).

<sup>9</sup> Possibly this is preserved to a larger extent in the conservative variety spoken in Puri (mainly agreeing with Biori Palula), where the following nominative-oblique alternation was recorded: /brhuː/ *(bhruú)* 'brother' vs. /brahu/ 'brothers'. The unstable character of the phoneme /h/ and voiced aspiration in Kalasha is also commented on by Mørch & Heegaard (1997: 50).

3.4 Suprasegmentals


In A, too, there are words, as in example (21), for which the location of aspriation is alternating (between speakers and possibly even with one and the same speaker).

(21) /ɡhaɖeːró/*∼*/ɡaɖheːró/ *(ɡhaḍeeró)* 'elder'

The aspiration feature (whether synchronically a cluster with /h/ or a voiceless aspirated consonant) has multiple diachronic sources: One is the OIA aspiration, thus preserved in Palula to an extent not evidenced in the major Shina varieties,<sup>10</sup> such as in /ɡhuːɽu/ *(ɡhúuṛu)* 'horse' < OIA *ɡhōṭa* (Turner 1966: 4516). Another is the above-mentioned intervocalic /h/ advanced to a more word-initial position. Finally, an old point of aspiration can be advanced, such as in /ɡhoːʂʈ/ *(ɡhoóṣṭ)* 'house' < OIA *ɡōṣṭhá* (Turner 1966: 4336). Other words may have followed other routes, possibly further reinforced by the rising pitch of a second mora accent (see §3.4.3 below). In any case, not all occurrences of aspiration, even when concomitant with plosives, are justified or explained solely by etymology (as pointed out by Morgenstierne 1932: 57).

Further study is needed to determine to what extent aspiration is preserved in two aspirated words that are compounded. There is an indication that the primary stressed part of the compound keeps its aspiration, while the other point of aspiration is entirely or partly deaspirated (compare with the comment in footnote 8 on *Grassman's Law*): /dhut/ *(dhut)* 'mouth' + /ɡhaːnu/ *(ɡháanu)* 'large' > /dutaɡhaːnu/ *(dutaɡháanu)* 'talkative'.

The interaction between accent and aspiration is another topic for further research. As breathiness (or a voiced cluster with /h/) quite often precedes a second-mora accented long vowel, the feature may have influenced or reinforced

<sup>10</sup> On the contrary, Schmidt & Kohistani (2008: 30) state that the voiced aspirates in modern Kohistani Shina have come into the language through borrowing, while they have been lost in OIA voiced aspirate cognates.

### 3 Phonology

the rising pitch of that accent. However, it should be pointed out that breathiness also occurs before unaccented, as in /ʑhamatroː/ *(ǰhamatroó)* 'son-in-law', as well as first-mora accented vowels, as in /oːɖhoːl/ *(ooḍhóol)* 'flood'; hence the two suprasegmental features are, in essence, independent.<sup>11</sup>

### **3.4.2 Nasalisation**

As mentioned earlier, obligatory nasalisation seems to be a marginal suprasegmental feature associated with an extremely limited number of lexemes, among which figure those in (22). It is unclear whether in all those lexemes there is a historical loss of a nasal segment. In any case, there are too few instances to postulate any individual nasalised vowel phonemes, or even less a whole set of nasalised vowels. Neither are there in the data any examples of a lexical distinction being made solely by contrasting an oral vowel with a corresponding nasalised vowel.


### **3.4.3 Pitch accent**

A phonological word in Palula may carry one, and only one, accent. Phonetically the accent is primarily realised as relatively higher pitch, accompanied to some extent by higher amplitude. Generally speaking, in a single word, accent is associated with high pitch, and the corresponding lack of accent is associated with low (or default) pitch.

The accent-bearing unit is the vocalic mora, which means that accent can be associated with a short vowel (as in (23)), or with the first mora of a long vowel (as in (24)), or with the second mora of a long vowel (as in (25)).

<sup>11</sup> This contrasts with the situation in Gawri and Torwali (both languages belonging to the Kohistani group of HKIA languages), where aspiration is only contrastive with voiceless consonants, whereas breathiness is a feature only optionally concomitant with low tone (Baart 1999b: 92; Lunsford 2001: 36–37).

3.4 Suprasegmentals



This means that the pitch accent<sup>12</sup> (henceforth only "accent") has one of the following phonetic manifestations:


A word, as referred to here, is either a bare stem, such as /děːs/ *(deés)* 'day', /paːɽú/ *(paaṛú)* 'magician', or a stem with one or more suffixes added to it, such as /deːs-ôːm-iː/ *(dees-óom-ii)* 'of the days', /ʑhôːn-um/ *(ǰhóon-um)* 'I will know'. Even though some combinations of syllables and accents are more common than others and there are restrictions on accent placement (see below), the location of the accent within a given word is not entirely predictable. Therefore, accent in Palula must be defined as lexical.

Difference in accent placement is in a few cases, as in (26), the only phonemic contrast between two lexical items.

<sup>12</sup> Pitch accent here is of the kind also observed in e.g., Lithuanian (Szemerényi 1996: 73–82), a mora accent which is "free within limits".

### 3 Phonology


Although this kind of minimal pair is not particularly common in the language, the system definitely allows them to occur.

### **The position of accent on stems**

Accent falls either on the final or the penultimate syllable in a non-verbal lexical stem, as seen in (27) and (28); whereas in verbal stems, as in (29), the accent is always on the final syllable. In terms of vocalic moras, only one of the three last moras, of any lexical stem in Palula, may be the locus of the pitch accent.


### **Accent properties of suffixes**

There are two types of suffixes: those that carry their own accent, which will be referred to as accent-bearing suffixes, and those that are accent-neutral. When a suffix of the first type is added to a stem, the accent of the stem is eliminated, and the word accent falls on the suffix. When a suffix of the second type is added to a stem, the stem accent may be retained. However, under certain conditions,

### 3.5 Morphophonology

even in the latter case, the accent (defined by the lexical stem) shifts from the stem to the suffix, a matter I will return to in §3.5.1.


Table 3.5: Accent-bearing suffixes

The most productive accent-bearing suffixes (mainly verbal) are presented in Table 3.5. Some of them also have other allomorphs, such as *-áan*: *-éen*, *-áand*, and *-éend*.

Some of these suffixes may be added cumulatively, in which case the accent is carried by the last accent-bearing suffix.

As with the accent-bearing suffixes, only the most productive accent-neutral suffixes are presented in Table 3.6, excluding possible allomorphs.

Some quantitative and qualitative morphophonemic alternations relating to the position of the accent in the word will be dealt with in §3.5.1.

### **3.5 Morphophonology**

### **3.5.1 Morphophonemic alternations relating to accent**

A number of segmental modifications (primarily in the nominal paradigm) are related to accent, or more precisely to the position of the accent within the word;

### 3 Phonology


Table 3.6: Accent-neutral suffixes

whether they are described as synchronically productive processes or the result of a diachronic process, the latter also offering some explanations to the more regular dialectal variation observed.

Table 3.7 illustrates the main types of alternations: *ḍheér–ḍheerí* has an accent alternating between the stem and an accent-bearing suffix without any concomitant segmental modification (see §3.4.3 and Table 3.5), *kuṇaák–kuṇaaká* has an accent shifting from the stem to an accent-neutral suffix without any concomitant segmental modification (see §3.4.3 and Table 3.6), *ṣiṣ–ṣiṣóom* has an accent shifting from the stem to an affix-neutral suffix with accompanying suffix modification (the unaccented allomorph being *-am*), and *haál–halá* has an accent shifting from the stem to an affix-neutral suffix with accompanying stem vowel modification. These four and some of their more salient subtypes will be further discussed and illustrated in the rest of this subsection.

### **Accent alternation (accent-bearing suffix) without modification**

A stem inflected with one of the accent-bearing suffixes presented in Table 3.5 will without exception carry its accent on the accent-bearing unit of the suffix. Some further examples are given in Table 3.8.

### 3.5 Morphophonology


Table 3.7: Accent-related alternations in the nominal paradigm

*a* In this section, only the Palula common transcription is being used (as in the rest of this work) without any accompanying IPA transcription.



### **Accent alternation (accent-neutral suffix) without modification**

In some cases, the accent shifts to the suffix in spite of it being an accent-neutral suffix. While the previously mentioned process is fully predictable from the type of suffix itself, the reason for the current process happening is instead to be found in the lexical stem itself.

For the most part (as far as the nominal paradigm is concerned), this shift takes place when the lexical stem accent is on the last mora, as in Table 3.9.

While this is true for all polysyllabic stems, there are quite many monosyllabic stems for which accent shift remains non-predictable from a purely synchronic perspective. On the one hand, there are those final-mora accented stems that do not produce an accent shift (Table 3.10), and on the other hand, there are those non-final mora accented stems that contrary to expectation do undergo an accent shift (the latter will be discussed further below).<sup>13</sup>

<sup>13</sup> An alternative, and possibly more economic, way of describing accent patterns in Palula, would be to regard final-mora accent as the default accent on any phonological word (whether inflected or not), and any from that deviating placement as specified in the lexicon.

### 3 Phonology

### **Accent alternation with suffix modification**

In some cases, an accented suffix vowel is, qualitatively or quantitatively, modified as compared with an unaccented allophone. This can be seen in Table 3.11. This particularly concerns the accent-neutral plural oblique suffix *-am* and the genitive suffix *-e* (in B only).

The alternation in the paradigms to some extent reflects general vowel shifts in the language *(a > aa, aa > oo/uu, ee > ii)* conditioned by accent, and, in the B dialect, also by syllable structure, to which I will have reason to return to in the discussion below on stem modification.

In the A dialect, *-am* regularly has the form *-óom* when accented.

In the B dialect, *-am* regularly takes the form *-áam* in closed syllables, and  *úum* in open syllables (as, for example, when followed by a genitive suffix), as illustrated in Table 3.12. The unaccented genitive suffix *-e* in B corresponds regularly to an accented form *-í*.

### **Accent alternation with stem modification**

In the nominal paradigm of the A dialect, some vowel modifications affecting the nominative have been blocked by accent shift in the inflected forms, resulting in alternations between *aa* and *a*, which is obvious with the nouns in Table 3.13. The lengthening of the accented vowels has produced a second-mora accent in polysyllabic stems and vowels preceded by aspiration (in the wider sense, see 3.4.1).

Along the same lines, there are regular alternations (Table 3.14) between accented stems with *ii, oo* (B *uu*) and *aa* vs. unaccented stems with *ee, aa* and *a*.

### **Other alternations**

Some other alternations having to do with the interaction between stem vowels and suffix vowels will be discussed at length in the chapters on noun and verb morphology, see Chapter 5 and Chapter 9, respectively.

### **3.5.2 Morphophonemic alternations relating to syllable structure**

Morphophonemic alternations relating to syllable structure, briefly touched upon above (see Table 3.12), is exclusively a feature of the B dialect. The alternations are between closed syllables with *ee, aa*, and *a* and open syllables with *ii, uu*, and *aa*. It is most clearly displayed in the nominal paradigm, as shown in Table 3.15.


Table 3.9: Accent shift from final-moraic accented stems to accent-neutral suffix

Table 3.10: Stems with final-mora accent not displaying accent shift


Table 3.11: Accent shift with suffix modification (A dialect)


Table 3.12: Accent shift with suffix modification (B dialect)



Table 3.13: Alternations between *a* and *aa* (A dialect)

Table 3.14: Alternations in the verbal paradigm: *a–áa*, *aa–óo* and *ee–íi*


Table 3.15: Alternations between *a–áa*, *áa–úu* and *ée–íi*, respectively (B dialect)


3.5 Morphophonology

### **3.5.3 Umlaut**

There are numerous examples, in the nominal (Table 3.16) and verbal paradigms (Table 3.17) of anticipatory fronting ("umlaut") of *aa* to *ee* when preceding an *i* in the following syllable. The anticipation normally does not occur if the *a* is short. The nouns 'glass' and 'book' in Table 3.16 show that this process has been productively extended even to relatively recent loans.

Table 3.16: Alternations in the nominal paradigm between *aá* and umlaut-*ee*


Table 3.17: Alternations in the verbal paradigm between *áa* and umlaut-*ee*


In addition to that, umlaut is applied to: a) verbal suffixes (or the final part of the perfective stem) in anticipation of a following adjectival feminine agreement suffix -*i* (Table 3.18), b) the adjectival stem in anticipation of a feminine agreement suffix (Table 3.19), and c) to derivations of various kinds in which the derivational suffix contains *i* (Table 3.20).


Table 3.18: Umlaut in verbal suffixes anticipating feminine agreement suffixes

Table 3.19: Umlaut in adjectival stems anticipating feminine agreement suffixes


Table 3.20: Umlaut in derivations


## **4 Parts of speech and the lexical profile**

### **4.1 Part-of-speech categories**

The primary criteria for the part-of-speech classification applied in this work (as well as in Liljegren & Haider 2011) are grammatical, although there is an obvious semantic core to each of the classes thus established (Givón 2001a: 49-54; Dixon 2010: 47-54, 102-106; Croft 2003: 183-188). These criteria include languageparticular distribution, functional range and morphological behaviour (Schachter & Shopen 2007: 1-3). Palula has four open classes and another nine clearly defined closed classes, but it should be stressed that this division is by no means to be seen as an entirely discrete one. Even within the open classes (particularly among the adverbs), there are closed subclasses, and even some of the closed classes are indeed open to occasional additions, through loans or derivation:


In addition, there is a small number of other words or word-like elements that has been given a somewhat more tentative classification as something other than one of the above-mentioned categories. Some of them constitute a very small category; others are special for other reasons. Below follows a brief summary of the main characteristic features of each category.

### 4 Parts of speech and the lexical profile

The four open part-of-speech classes are treated in depth, each in a separate chapter (nouns in Chapter 5, verbs in Chapter 9, adjectives in Chapter 7, and adverbs in Chapter 8), as is also the closed class of pronouns (Chapter 6). The remaining closed classes are either treated alongside functionally or form-wise closely related categories, or as part of the discussion of higher-level structures. Determiners, in particular demonstrative determiners, are dealt with along with pronominal demonstratives (§6.2), as there is an obvious diachronic relationship as well as plenty of paradigm-sharing between the two categories. Quantifiers, in particular numerals, are due to their shared function as noun modifiers, treated along with adjectives (§7.4). Postpositions are due to their functional overlap with adverbs, especially in the spatial-temporal realm, treated in the same chapter (§8.2). Auxiliaries are due to their role in expressing TMA distinctions mentioned and further exemplified in the chapters dealing with verbs (Chapter 9) and verbal categories (Chapter 10). Examples of the use of the size-wise limited set of mood markers appear e.g., in the treatment of imperative sentences (§10.2.1), hearsay (§10.2.4) and interrogative constructions (§15.2). Conjunctions and discourse markers are exemplified throughout Chapter 13 and Chapter 14. Most of the rest of the strictly limited part-of-speech categories are apart from the examples and brief characterisation given below, not given any further treatment in this grammar.

### **4.2 Nouns**

Examples: *báaṭ* (m) 'stone', *achíi ̣* (f) 'eye', *ṭíinčuk* (m) 'scorpion', *kúṛi* (f) 'woman', *biaabaán* (f) 'wilderness', *rúuṣ* (f) 'anger'.

Semantically characterised by relative temporal stability of its referent, a Palula noun typically, and unsurprisingly, denotes concepts such as things, places, animals and people, but a great deal of abstract entities are also encoded as nouns. They may be further specified for animacy or humanness. A noun primarily functions as the head of an argument (subject, object, etc.), but can also be the predicate, in the latter case often without an accompanying overt copula. The most important subclassification of nouns is the one between masculine and feminine gender nouns. Gender assignment is almost exclusively inherent and part of the lexical specification. Only for a smaller group of nouns is gender assigned contextually only.

The typical noun is inflected for number (singular vs. plural) and case (nominative vs. oblique vs. genitive), although the realisation of and formal expression of each category are subject to declensional differences. There are three main

declensions, two minor ones, and a smaller number of nouns that display idiosyncratic inflectional behaviour.

### **4.2.1 Proper nouns**

Examples: *atshareét* (m) 'Ashret (the name of a valley and its main settlement)', *deeúli* (f) 'Dir (the name of a district and a former principality)', *machoók ̣* (m) 'Machoke (the name of a tribal ancestor)'.

A distinction can be made between common nouns and proper nouns. The latter are used to refer to specific persons or places. They are normally not pluralised, and only rarely occur with any preceding modifiers. In most cases, however, they can be identified as either masculine or feminine, and as belonging to one of the above mentioned declensions.

### **4.3 Verbs**

Examples: *utráp-* (itr) 'to run', *mar-* (itr) 'to die', *ɡhin-* (tr) 'to take, buy', *phaalé-* (tr) 'to break, tear', *bhe-* (copula, itr) 'to become; to come into existence'.<sup>1</sup>

References to less-stable experiences or transitory states rather than to a particular entity, are typically, and again, unsurprisingly, encoded as verbs in Palula. The characteristic function of a Palula verb is as a predicate, with the most important subclassification being one between transitive and intransitive verbs. This is a strict distinction, and almost without exception, a particular verb stem is either intransitive or transitive and cannot (without further derivation, see below) be ambivalent or polyvalent. In addition to those two main classes, there is a subclass of copulative verbs, some of which overlap functionally with intransitive verbs on the one hand, and with auxiliaries on the other.

Verbs are primarily inflected for tense, aspect and argument agreement, and in addition to that, a few tense-mood-aspect (TMA) categories (such as Perfect and Past Imperfective) are expressed periphrastically by means of auxiliaries. Two different kinds of agreement are part of the paradigm, person agreement and gender/number agreement. The former is confined to the non-tense marked categories Future and the Past Imperfective, and the latter with Present and the perfective-based categories (Simple Past, Perfect and Pluperfect). Apart from finite inflectional categories, there are a number of important non-finite forms.

As far as inflectional morphology is concerned, there are two main morphological verb classes, here referred to as L-verbs (an open, productive and large class)

<sup>1</sup> The citation forms with a final hyphen are underlying imperfective verb stems.

### 4 Parts of speech and the lexical profile

and T-verbs (a closed, non-productive, small and rather heterogeneous class). Additionally, there are a few verbs with stems that to a varying degree are suppletive. Within the class of L-verbs there are predictable variations in the inflectional paradigms due to accent position and the quality of stem vowels. Many Tverbs form their perfectives with a plosive segment (in the clear cases a *t*-suffix), but often this has been assimilated with preceding stem segments, and it makes sense to identify a perfective stem as distinctly different from an imperfective stem.

### **4.3.1 Secondary stems**

Examples: *pašíǰ-* (passive) 'to be seen' (from *paš-* 'to see'), *thawá-* (causative) 'to have someone do' (from *the-* 'to do').

There is a fairly productive valence changing morphology by which "new" stems can be derived: a secondary one-argument verb stem can be derived morphologically from a corresponding primary transitive verb, and in the reverse, many two-argument verb stems are (at least in a historical sense) derived morphologically from corresponding primary one-argument verb stems. Similarly, a secondary three-argument verb can be derived morphologically from a corresponding primary transitive verb stem.

### **4.3.2 Conjunct verbs**

Examples: *ǰhaní the-* (non-incorporating) 'to marry', *káaṇ the-* (incorporating) 'to listen, give heed to', *milaáu bhe-* (incorporating) 'to meet', *póo de-* (non-incorporating) 'to step on'.

Conjunct verbs are frequently occurring complex predicate constructions that, albeit phonologically existing as a combination of two words, function as lexical units. Usually they consist of a simplex verb preceded by a noun or an adjective; words, or rather lexical elements, that cannot easily be identified as belonging to either of these part-of-speech categories may also occur in this position (see §4.15). The verb in such a construction comes from a small set of verb stems (mostly *bhe-* 'become', *the-* 'do', *de-* 'give; fall'), and it is the non-verb element that contributes the main semantic content to the complex.

There are two main types of conjunct verbs: incorporating and non-incorporating. In the non-incorporating conjunct, the non-verb element functions as the direct object, whereas in the incorporating conjunct, the non-verb element is never treated as an argument of the clause.

### **4.4 Adjectives**

Examples: *piṇḍúuru* 'round', *ṣíiṛu* 'blind', *ḍanɡ* 'hard', *purá* 'full, complete'.

The typical adjective functions as an attributive modifier of a noun, or as a predicate, in the latter case often without an accompanying overt copula. The only formally substantiated subclassification is one based on agreement properties. On the one hand, there are those adjectives that inflectionally indicate gender, number and case of the nouns they modify or, when they function as predicates, the nouns that are their subjects. On the other hand, there are those adjectives that are invariable in form. Adjectives pertaining to dimensions, age and human propensity show a strong tendency to be substantivised.

The great majority of inflecting adjectives occur in three forms ending in *–u* (msg.nom), *-a* (mpl.nom/m.obl) and *–i* (f), respectively, the latter with an additional umlaut for those stems that have an accented *á* or *áa* in its underlying form. There is also a marginal feminine plural in *-im*, largely limited to predicative use.

### **4.5 Adverbs**

The fourth, and only remaining, open class is adverbs. It is of only moderate size compared to the other three open classes, and some of its rather disparate subclasses are closed rather than open. Adverbs function as modifiers of constituents other than nouns on various grammatical levels. A fair number of adverbs are also part of the cross-cutting category pro-forms, in this case belonging to the subcategory pro-adverbs. At least five subclasses of adverbs can be identified: spatial adverbs, temporal adverbs, manner adverbs, degree adverbs, and sentence adverbs. It should be noted that many adverbial meanings are expressed by words primarily belonging to other categories, or by entire phrases (often postpositional or noun phrases).

### **4.5.1 Spatial adverbs**

Examples: *bhun* 'down, down below', *aǰá* 'up, up there', *nhiáaṛa* 'near, nearby'.

Spatial adverbs usually modify verbs or verb phrases, and as such, specify the direction of a movement or the location of an event expressed by a verb. Some of these adverbs are closely related to, yet in most cases clearly distinct from, nouns. While these do not pluralise, nor are they assigned to a gender category, they do occur with case inflections reminiscent of those found in the noun paradigm.

### **4.5.2 Temporal adverbs**

Examples: *típa* 'now, nowadays', *heeṇṣúka* 'this year', *dhoóṛ* 'yesterday'.

There is a certain degree of overlap between temporal and spatial adverbs, and like the spatial adverbs, a number of temporal adverbs are related to nouns. In some cases it is not altogether obvious whether a particular word is primarily a noun or primarily an adverb. It has been decided here to categorise such a word as a noun when assignment to one or the other gender can be established beyond doubt.

### **4.5.3 Manner adverbs**

Examples: *ɡúči* 'freely, for nothing', *táru* 'quickly', *bhraáš* 'slowly'.

Although manner mainly is expressed by non-finite verb forms (especially converbs and copredicative participles), there is a small class of non-derived manner adverbs whose main function is to modify verbs or verb phrases.

### **4.5.4 Degree adverbs**

Examples: *phaṣ* (with 'white': *phaṣ paṇáaru* 'white as a sheet'), *bak* (with 'bright': *bak práal* 'shining bright'), *tap* (with 'dark': *tap chiṇ ̣* 'pitch dark').

The function of degree adverbs is mostly to modify adjectives or other adverbs. Apart from a few quantifiers that besides their noun-quantifying role also function as adjectival and adverbial degree modifiers, there is a set of what is referred to here as co-lexicalised intensifiers (some examples given above). These are highly specialised (or idiomatic) elements having a degree-modifying or intensifying function when preceding a certain adjective or adverb (with which they are co-lexicalised).

### **4.5.5 Sentence adverbs**

Examples: *ɡóo* 'maybe', *inšaalaáh* 'God willing', *rištaá* 'really, in truth'. Sentence adverbs is a small subclass that modify entire utterances, i.e., they specify the speaker's attitude toward the event.

### **4.6 Pronouns**

Pronouns are in fact a subset of a cross-cutting category of pro-forms, including words that belong to a variety of part of speech categories as well as some that

### 4.6 Pronouns

correspond to larger constituents. There are two major kinds of these, functionally-semantically defined, pro-forms: demonstrative pro-forms and indefiniteinterrogative pro-forms, the former mostly recognised by an initial *ee*-element, the latter by an initial *k*- or *ɡ*-element. In Table 4.1, a few examples of such proforms can be seen.


Table 4.1: Cross-cutting pro-forms

However, the class of pronouns has a central position with its many members, especially of the demonstrative kind, and is therefore deserving of being treated as a part of speech in its own right. Pronouns substitute for a noun or an entire noun phrase. A number of subclasses can be identified.

### **4.6.1 Personal pronouns**

Personal pronouns are words that refer to the speaker or the person spoken to. They occur in singular and plural, respectively, with two case forms available in the singular and four in the plural. Third person, i.e., words that refer to contextually identifiable referents other than speakers or hearers, is expressed by forms belonging in the demonstrative subcategory.

### **4.6.2 Demonstrative pronouns**

Demonstrative pronouns use a basic three-way distance/visibility differentiation extensively: a proximal category for referents close at hand, a distal category for referents further removed from the speaker, and a remote category for referents out of sight. Within each subset, there is a further differentiation in number, case, and gender, the latter restricted to the singular nominative. It is also possible to differentiate between strong and weak forms, where strong forms with an initial *ee* (*eesó* corresponding to *so*) tend to be used for deictic or anaphoric functions in order to keep track of less accessible discourse referents, whereas the weak forms are the default choice with easily accessible discourse referents. For the proximal and distal sets, additional forms with an initial *a* (*anú* corresponding to *nu*) are available, seemingly in free variation with the "bare" forms.

### **4.6.3 Indefinite-interrogative pronouns**

Another subset of pronouns directly corresponding to the demonstratives, does double duty as indefinite and interrogative pronouns. While the same case distinctions are made as with demonstratives, there is no differentiation in gender or number. Different indefinite pronouns (*ɡubáa, ɡa*) must be used when referring to an inanimate referent as opposed to those indefinite-interrogative pronouns that refer to animate, in particular human, referents. In addition to this particular closed set, there is a small number of other words that can be used as, and therefore labelled as, indefinite pronouns. These often have more specialised functions, some of which in fact primarily belong in the class of quantifiers.

### **4.6.4 Reflexive pronouns**

There is one frequently used pronoun, *teeṇíi* 'self's, own', identified as reflexive, i.e., a pronoun that is co-referential with another nominal in the clause. It occurs almost exclusively in this form. Usually, but not exclusively, it is the possessor in a possessive construction, and its referent is identical to the clause subject.

### **4.6.5 Reciprocal pronouns**

There is a single reciprocal pronoun, *akaadúi* 'one another, each other'. It is used, although rarely, in a few other case forms. It is, like the reflexive pronoun, coreferential with another nominal, but is restricted to mutual actions.

### **4.7 Determiners**

Examples: *anú* 'this', *áa* 'a', *dúi* 'another', *daašúma* 'the tenth'.

While attributes (of nouns) are expressed by adjectives, and quantity by quantifiers, determiners establish the reference of a particular noun (and in some cases of a pronoun). Almost all determiner words have dual membership (or are polysemous) and can occur pronominally as well as adnominally. Although clearly derived from quantifiers, words that are normally described as ordinal numerals are for functional reasons included here among the determiners. The special subset of demonstrative determiners displays agreement in gender, number and case (a nominative masculine singular agreement form vs. a non-nominative/plural/feminine agreement form), using different forms for proximal, distal and remote referents.

As with the closely related demonstrative pronoun set, a further differentiation is made between strong and weak forms, where the strong forms are used along with less accessible discourse referents, whereas the weak forms occur when the referents are easily accessible. A further, and probably still ongoing, grammaticalisation of this distinction is the use of the weak forms of the remote set (*so* and *se*, respectively) to indicate definiteness or identifiability, often systematically contrasting with the indefinite use of *áa* 'a, an'.

In addition to the basic three-way differentiation, the demonstrative determiners can be compounded with preceding spatial adverbs to derive more specialised determiners that indicate finer degrees of distance or vertical-horizontal position in relation to the speaker, e.g., *bhun* 'down there' + *aṛó* 'that' > *bhunaṛó* 'that down there'.

### **4.8 Quantifiers**

Examples: *tróo* 'three', *bíiḍu* 'many, much', *buṭheé* 'all', *khéli* 'quite some'.

Quantifiers are modifiers in much the same sense as adjectives, but while adjectives are descriptive, i.e., denoting qualities and attributes, quantifiers are limiting, thus indicating quantity or scope of the nouns they modify. While most quantifiers show no agreement features, there are a few that agree in gender and case with the nouns that they modify, much like adjectives. Unlike adjectives, quantifiers do not agree in number (naturally, as they are inherently either plural or singular) with the modified noun. The quantifiers that indicate exact quantities can combine either with other such quantifiers or with non-exact quantifiers to form compound quantifier expressions; a few of them can be pluralised when

being modified by another quantifier. As with some subsets of adjectives, quantifiers have a strong tendency to be substantivised.

### **4.9 Postpositions**

Examples: *the* 'to, for', *sangí* 'with, at', *maǰí* 'among, in, inside'.

Postpositions are markers of syntactic-semantic roles or spatial-temporal relations that are held by the nouns or pronouns they follow. These markers also form phrasal constituents with the nouns or pronouns about which they convey some information. Under certain circumstances, some postpositions form a single phonological word with the nominal form to which they are postposed.

With most postpositions, the preceding noun occurs in the oblique case. Apart from single word postpositions, there are also a number of complex postpositions, consisting either of a sequence of postpositions or a postposition followed by an adverb. In both cases, the phrase functions just like any single word postposition. Some spatial and temporal adverbs can also function as postpositions.

### **4.10 Auxiliaries**

Examples: *de* Past tense marker, *bhóo* 'can, to be able to', *ṣáatu* 'began'.

Auxiliaries is a small set of verb-related words which, in addition to verbal morphology, express certain TMA distinctions. Although some of them can take verbal inflections, they always occur in a clause along with a (main) verb. Some of the auxiliaries can be combined with each other.

### **4.11 Mood markers**

Examples: *ee* Polar question marker, *maní* Hearsay marker, *neé* Request marker.

Mood markers is another size-wise very limited set of words that, one way or another, specify the relationship between an utterance as a whole and the speaker and/or hearer. A mood marker mostly occurs in utterance-final position, sometimes cliticised to the immediately preceding element.

### **4.12 Conjunctions**

Examples: *ee* 'and', *yaá* 'or', *heentá* 'if', *ki* 'that'.

The function of conjunctions (some of them clitics) is to connect or signal the relationship between constituents on various levels. Primarily they indicate what kind of relationship exists between two adjacent clauses, or between a dependent unit and a larger unit that the former is a part of. With a few exceptions, the conjunctions can be characterised as postpositional, since the conjunction forms a structural unit with the conjunct it follows.

### **4.13 Discourse markers**

Examples: *ba* Switch-topic marker, *bi* Separation marker, *ta* Contrast marker, *eé* Amplification marker.

Discourse markers are words (or clitics) that specify the discourse role of a particular (preceding) unit vis-à-vis adjacent units. The units that are being indicated thus are primarily phrasal in nature (mostly noun phrases), but not exclusively so. A secondary effect of some discourse markers is that they indicate how larger units (such as clauses) are interrelated, especially when used in pairs, or when the same marker is used repeatedly in two adjacent clauses, thus partly overlapping with the function of the conjunction category.

### **4.14 Interjections**

Examples: *óo* 'yes', *ohoó* 'wow!', *čo* 'go ahead!', *ée* 'hey!'.

Although the category of interjections, at least theoretically, may be an open class, there are relatively few examples included in this work. These words can in themselves be used as entire utterances, and there is in most cases no clear syntactic connection with any other co-occurring words.

### **4.15 Other words or word-like elements**

A single-word word category, at least as far as this vocabulary is concerned, consists of the high frequency negator word *na*. Belonging in the cross-cutting category pro-forms, but not really fitting into any of the aforementioned classes, is the indefinite-interrogative *keé* 'why', substituting for an entire clause. Another minor category is labelled honorific; such lexical units are titles or title-like elements prefaced to, or cliticised after, names of certain highly respected people. Closely related to that are some ritualistic expressions, such as *aleehisalaám* 'on

whom be peace', which is a phrase that has to be added when mentioning one of the prophets according to Islamic beliefs.

As already mentioned above, some conjunct verbs consist of a simplex verb preceded by a lexical element that only occurs as part of that particular complex. Such elements have been classified as host elements. Two other processes involving what may be termed "semi-words" are echo formation (usually by repeating a word and substituting the onset with *m*, as in: *ɡúuli* 'bread' + echo > *ɡúuli múuli* 'bread and other eatables'; *nirkízi* 'henna' + echo > *nirkízi mirkízi* 'henna and stuff') and reduplication (*áak* 'one' + red > *aakáak* 'one each'; *teeṇíi* 'their, etc. (refl)' + red > *teeṇteeṇíi* 'each their, etc.').

## **5 Nouns**

### **5.1 The noun and its properties**

Distributionally (i.e., syntactically) the noun functions as the head of a noun phrase argument. Within the noun phrase, the head is placed finally, as in 'man' in example (1), and preceding modifiers agree with the head in gender, number and case (for further details, see Chapter 11).

(1) *aní* prox *dúu* two *dhríɡ-a* tall-mpl *míiš-a* man-pl 'these two tall men' (A:ADJ048)

On the clause level, such a phrase, where a noun is the head, can serve as a subject, as in (2), or a direct object, as in (3).

(2) *se* def *kúṛi* woman *búd-i* understand.pfv-f *ki* compl *anú* prox.msg.nom *míi* 1sg.gen *bharíiw* husband *na* neg

'The woman understood that this wasn't her husband.' (A:WOM646)

(3) *teeṇíi* refl *čúti-m* paw-pl *de* give.cv *baṭ* stone *uch-áan-u ̣* pick.up-prs-msg 'It [the leopard] picked up a stone with his paws.' (B:SHB749)

It can also be the head of a postpositionally marked noun phrase, as in (4), a noun phrase expressing location, as in (5), or a nominal predicate, as in (6).


5 Nouns

(6) *lhoók-u* small-msg *díiš* village *de* be.pst 'It was a small village.' (A:JAN003)

Phonologically, the most frequent Palula noun stem (i.e., the nominative singular form) in my data consists of two syllables, comprising about half of all nouns in my database, 30 per cent are monosyllabic, and 16 per cent are three syllables. Four-syllable nouns are rare in my database, and possibly most if not all of these are either derived from three-syllable words or are compounds of two noun stems. There is no evidence of noun stems exceeding four syllables. The minimal Palula noun stem consists of a rhyme where the nucleus is built up by a short vowel and a coda consonant: *uṭ* 'camel' (B), *uts* 'well'. Though simple vowel words do exist in Palula, there are no such nouns.<sup>1</sup> A noun stem with a consonant onset must have at least a long vowel nucleus, or a short vowel and a consonant coda: *bíi* 'seed', *kuḍ* 'wall'. There are no nouns consisting of only consonant onset and a short vowel, though there are such words belonging to other parts of speech.

Morphologically, the prototypical Palula noun is inflected for number (§5.4) and case (§5.5), which is intimately related with declensional membership (§5.6) as well as inherent grammatical gender (§5.3).

### **5.2 Noun morphology**

### **5.2.1 Inflectional morphology**

Palula has two grammatical genders: masculine and feminine. Gender is an inherent, lexically defined, property of the noun and is partly predictable on semantic and formal grounds (see §5.3). Nouns are inflected to show number (§5.4) and case (§5.5). An example paradigm is displayed in Table 5.1, representing one of a handful of Palula noun declensions (§5.6).

As far as case inflection is concerned, the primary distinction is between nominative (§5.5.1) and oblique (§5.5.2) case, but other inflectionally expressed cases, such as genitive (§5.5.3), will also be discussed. Gender, number and case control agreement within the noun phrase (§11.3), and therefore have relevance also for adjectives, demonstratives and numerals. Gender and number have further relevance for clause-level agreement patterns (§12.1–§12.2). Definiteness and deixis

<sup>1</sup> A potential counter-example is *(óo∼)* 'mouth', although the status of the nasalised vowel as a single segment is phonologically questionable.

5.2 Noun morphology


Table 5.1: Inflection of nouns

are central components of the noun phrase but are not part of the inflectional system of the noun and will be discussed in §6.2.

As in many other NIA languages, the Palula nominal paradigm is built up from a combination of inherited synthetic elements, "new" agglutinative elements and analytical elements (Masica 1991: 212). The declensional classes (§5.6) suggested in this chapter are mainly based on the formation of the plural, but case forms as well as gender play a role. On functional grounds, this chapter includes slightly more than what would typically be included under the heading of morphology proper.

### **5.2.2 Derivational morphology**

There are only a few regular processes by which nouns are derived morphologically from other parts of speech. The clearest examples are nominalisations of verbs, although it should be kept in mind that those are usually occurring in specialised grammatical constructions, such as the Verbal Noun (described in §10.3.3), occurring as the non-finite predicate in a number of dependent clauses, and the Agentive Verbal Noun (see §10.3.4), the latter which is for instance used in forming an inceptive construction.

The derivation of deadjectival nouns is idiosyncratic, and semi-regular at best, with a few pairs pointing to what might earlier have been productive morphological processes: *šidal-aár* 'coldness' < *šidáalu* 'cold'; *taapi-aál* 'warmth' < *táatu* 'warm, hot'.

Certain adjectives, especially dimensional adjectives modifying humans, and numerals can also be used as heads of noun phrases. Apart from the application of case forms normally reserved for nouns, there is no morphology *per se* involved in such derivation of nouns from adjectives or numerals (see §7.3.2 and §7.4.2).

### 5 Nouns

### **5.3 Gender**

A feature associated with, or assigned to, each noun is its grammatical gender. The language has two grammatical genders; each noun is either *masculine*, see examples in (7), or *feminine*, see examples in (8), following the general tradition in describing IA languages.



In my database, which comprises about 1,300 nouns,<sup>2</sup> masculine nouns were slightly more numerous than feminine nouns with 58 per cent masculine as compared to 42 per cent feminine.

There are no traces here of the three-gender system of OIA, still present in some western NIA languages (Masica 1991: 220–221). As in many of the twogender systems found in NIA, it is mainly the old masculine and neuter that have merged. Although gender has otherwise been fairly stable as far as inherited vocabulary is concerned, there has been a rather radical restructuring of the old declensional system, partly as a result of segmental loss, a phenomenon we will have reason to discuss further in §5.6.

While gender is an inherent and classificatory property of the noun lexeme itself, it is essentially established through morphological agreement with adjectives/demonstratives (9) and verbs (10), for which it is a variable property.


<sup>2</sup> This figure refers to the number of nouns in the A dialect. There is an additional database with about 400 B dialect nouns, many of them overlapping with the former.

### 5.3 Gender

### **5.3.1 Gender assignment**

There are plenty of semantic and formal clues to gender assignment in Palula. Looking again at the examples in (7) and (8), we do not have a problem making intelligent guesses as to the gender of the words for 'boy' and 'older sister', respectively, if we assume that there is a connection between biological gender or sex and this particular grammatical two-gender differentiation between feminine nouns and masculine nouns (Dahl 2000: 102). Those nouns, as is the case for other nouns denoting humans and higher animates, are indeed assigned gender according to meaning – masculine gender if male, and feminine if female, a phenomenon we refer to as semantic gender assignment (Corbett 1991: 7–32). For the word 'horse, stallion', we can similarly draw the conclusion that it is a masculine noun (especially if we are told that another word is used to refer to a mare). The other three words, referring to either lower animates, inanimate objects or phenomena, have also – some way or another – been assigned to one of the two genders: *hiimaál* and *phúti* are feminine nouns and *ɡhoóṣṭ* is a masculine noun. We cannot, however, see any obvious semantic reason for this to be so, and we will therefore have to look at other assigning criteria.

Male–female word pairs with a common lexical "root" are frequent in the Palula lexicon, especially as kinship terms, and the derivation of a feminine counterpart from a masculine (except in a few cases when it may be the other way around) could be described as a rather productive state of affairs, even synchronically. Some examples are given in Table 5.2.

Beside the obvious correlation between biological and grammatical gender, we have a somewhat related but less regular connection between gender and rela-


Table 5.2: Male/female pairs vis-à-vis gender

### 5 Nouns


Table 5.3: Masculine/feminine lexical pairs

tive size/power. Alternatively there is some kind of complementarity implied. This is mainly detectable in lexical pairs of the kind displayed in Table 5.3, which although similar to the male–female pairs in Table 5.2, are related to other scales than pure masculinity versus femininity and also with a more approximate similarity in kind. While this derivative process (in its essence very much like diminutive formation) may have been productive in the past, it is uncertain how productive this extended gender differentiation is among today's Palula speakers.

For most higher animates, the masculine noun is the "default" gender, and the feminine, when used, is a marked form (compare with §5.3.3). Practically, that means that either only a masculine form is available or that the masculine member in a masculine–feminine pair is used when no specification is needed. However, in a few cases, the feminine is the default, even when there is a distinct masculine form available, such as *luumái* 'fox' vs. *luumóo* 'male fox', and the above exemplified (Table 5.2) pair *púši* 'cat' vs. *púšu* 'tom-cat' (compare with Dahl 2000: 103–104). A different relation holds between masculine *čhaál* 'goat kid' and feminine *čhéeli* 'goat', where the feminine noun is not the female counterpart to the masculine but instead the generic term for goat, whereas the masculine is used to refer to the young kid regardless of sex.

Now we will turn to formal properties that may stand in relation to gender association. First, a comparison of a large number of nouns reveals that particular phonological properties really seem to be related to either of the two genders. A word-final *u* is for instance a formal feature of a major group of masculine nouns, as are *á* and *oó* for smaller groups of nouns that are all masculine (11).


5.3 Gender


Table 5.4: Toponyms and gender assignment

A word-final unaccented *i* is in the same way a feature of a very large group of feminine nouns, and *ái* for a smaller group of them (12).


This kind of correlation between phonological properties and gender, not uncommon in IA (Masica 1991: 219) and referred to as "overt gender"<sup>3</sup> (see also Corbett 1991: 44, 62), is of course significant, but whether we should talk about "gender marking" is not equally clear. We can hypothesise that *-u* is a nominative case marker, available for some nouns like *ɡhúuṛu*, for which segmental material is added to what we may view as a root *ɡhúuṛ-* when inflecting for case and number. For a large number of feminine nouns with a final *i* (including *ɡhúuṛi* 'mare'), we may in addition to that describe them as derived from such a masculine root with the help of a derivational affix *-i* denoting something female (or something smaller of approximately the same kind). In a diachronic perspective both seem in fact to go back to OIA derivational suffixes with – at least partly – diminutive senses: a masculine *-aka-* and a feminine *-ikā-* (Masica 1991: 222; Morgenstierne 1941: 15; Buddruss 1967: 29).

Also proper nouns referring to inanimates, such as toponyms (Table 5.4), have gender assigned to them. Gender assignment here seems to be primarily morphologically determined, but since phonology is interrelated with morphological behaviour there is no surprise that those toponyms that end with *i* or *ai* are

<sup>3</sup> Alternative terms offered by Masica (1991: 219) are *thematic, strong thematic, extended, augmented* or *enlarged*.

### 5 Nouns

feminine, whereas most of the rest are masculine. In some cases the etymology is overt enough to connect the last part of a toponym with a proper noun with a particular and known gender (such as *-deéš* in *šaṛadeéš*, which is related to the masculine noun *díiš* 'village').

There are also, as we shall see in §5.6, obvious connections between noun morphology and gender, or what is referred to as morphological gender assignment (Corbett 1991: 34–50), which is really a matter of co-variation rather than one feature necessarily being primary and the other secondary. Some declensions and subdeclensions consist entirely or to a large extent of feminine or masculine nouns, whereas for other declensions the gender distribution is quite even. Gender can, for instance, be predicted to some degree from the plural formation or the case inflections applied to a particular noun.

We can thus say that an interplay of semantic, phonological and morphological properties make a noun masculine or feminine, or at least tells us whether a particular noun is more likely to be one or the other.

### **5.3.2 Gender stability and consistency**

An issue that deserves some attention is the relative stability and variability of gender in Palula. First, across the two main dialects, there is almost no variation at all,<sup>4</sup> at least as far as words with a longer history in the language are concerned. This almost complete correspondence is on the other hand not surprising, considering that the noun paradigms and the declensions are near-identical in those two varieties. Comparing with something less closely related, the Shina variety of Gilgit, we observe (see Table 5.5) more divergence between that variety and Palula as far as gender assignment is concerned. If we limit our comparison to inanimate and lower animate cognate nouns, and exclude higher animates, which are all assigned gender according to rather transparent semantic criteria, as well as what we with confidence can regard as recent loans, we come up with a 79 per cent correspondence between gender in Palula and gender in Gilgiti. This is still quite a high figure, considering that Palula and Gilgiti speakers most likely have not had any shared development or even any non-trivial contact for at least a few centuries (see §1.2.3).

<sup>4</sup> In my present database I have only eight nouns that differ in gender assignment between A and B.

### 5.4 Number



### **5.3.3 Gender markedness**

Morphologically and otherwise, the feminine tends to be the marked gender in Palula. Many nouns with an ending *i* are derived from a masculine noun (see Table 5.2), while the opposite seems to be extremely rare (*phóopu* from *phéepi* is a possible example). Also, the most common complement-taking verbs agree, by default, with the masculine singular (see §14.5.1) regardless of the content of the complement clause. There is similarly an indication that in other cases of agreement without an accessible controller, the masculine singular form of the target is chosen. Perhaps the obligatory number distinction in agreement with masculine heads vis-à-vis the optionality as far as feminine heads are concerned (see §11.3) is related to this as well. These observations suggest that the "default" in Palula is the masculine. Interestingly, this stands in direct contrast to the feminine as the default gender in Gilgiti Shina (Hook & Zia 2005: 176). This, however, is a topic for further studies.

### **5.4 Number**

Number is one of the basic, and possibly the most straightforward, of the inflectional categories related to nouns in Palula. There are two number categories, singular and plural, with no traces of the OIA dual category. There are a number of different plural formations in the language: *-a, -i, -m, -ee* and *-aan*, as seen in Table 5.6, some of them with additional stem changes. As these inflections (along with oblique case inflections) form the basis of the declensional distinctions, the various forms will be discussed in more detail in §5.6.

Some groups of nouns (§5.4.1–§5.4.3) are special in that they normally do not pluralise or display a regular contrast between a singular and a plural form.

### 5 Nouns


### Table 5.6: Examples of plural formation

### **5.4.1 Non-count nouns**

Non-count or mass nouns normally do not have plural forms, nor are they modified by numerals. Many of them refer to what is perceived as a substance. Some examples are provided in Table 5.7.

Table 5.7: A selection of non-count nouns


Such nouns are usually quantified by adjectives with the meaning 'much', 'some', 'a little', a measuring unit or a noun referring to some sort of container or a quantifiable part of the whole, as in (13) and (14).


### 5.4 Number

Only exceptionally, when speaking of more than one variety of a substance, for instance, may a mass noun be pluralised. Some mass nouns may also occur in a plural form, as 'blood' (16), as well as in a singular form (15), while still being regarded as a substance, sometimes with the connotation of large volumes or weights somehow divided up into several separate chunks or volumes. It would, however, still be ungrammatical to quantify such a noun directly with a numeral, whether in its singular or its plural form.


Also, some abstract nouns (examples provided in Table 5.8) are treated as noncount nouns that do not normally pluralise.

Table 5.8: Examples of abstract non-count nouns


However, since non-count nouns also occur in oblique forms (such as *inseefí* from *insaáf* 'justice'), and the singular oblique in most Palula declensions is formally identical to a nominative plural, there is usually no need to come up with a novel plural form, since the step towards pluralisation is never far away, so to speak.

Other nouns have a unique reference, at least in one or more senses of the word, and therefore normally do not occur pluralised, but must be considered as special cases of non-count nouns: *súuri* 'sun', *ɡhuaaṇaá* 'Pashto (language)' and *xudái* 'God'.

### **5.4.2 Collective nouns**

While the nouns dealt with above as non-count nouns, often referring to substances or abstract concepts, are largely limited to the singular, a few nouns (17)

### 5 Nouns

have a collective meaning and occur in a plural form only, or almost exclusively so.


### **5.4.3 Proper names**

A structural property typical of proper names is that they do not normally pluralise, as they are seen as having a unique reference, tied to one person or to one place. There is, however, no such restriction on case inflection. As proper names can take on any function in the clause, they also inflect like other nouns for case, as can be seen in (18).

(18) *fazelnuur-á* Fazal.Noor-**obl** *díi* from *ba* top *panǰ* five *putr-á* son-pl *bhíl-a* become.pfv-mpl 'Fazal Noor had five sons.' (B:ATI079)

Complex names consisting of several elements are common and are treated as lexical units and often as single phonological words (see §11.1.3).

### **5.5 Case**

Noun phrases are marked for case, either with noun suffixes, or – as far as pronouns are concerned – through distinct forms and also by cross-referencing within the noun phrase between the noun head and its dependents. There is also cross-referencing within the clause between the verb and one of its arguments (see Chapter 12).

Even though the following presentation focuses on case inflection rather than grammatical relations, it is, as Masica (1991: 230–231) phrases it, "easily the most problematic nominal category in NIA". Case in Palula can, as in many other NIA languages, be described as accumulative inflectional layers with case-like functions, and the comparison between languages is complicated by the fact that a function in a given layer in one language is managed in a different layer in another language. The actual case forms (mainly the oblique case) in Palula differ between the declensions as do the actual occurrences of certain case distinctions.

The most basic formal distinction is that between the nominative (or direct) case and the oblique case, but even that is not realised in all declensions, but is,

5.5 Case

for instance, totally missing in the *m*-declension (see §5.6.3). Form syncretism (Matthews 1991: 27) between nominative plural and oblique singular is found in most declensions, reminiscent of the paradigm of "overtly" masculine nouns in Urdu-Hindi (Schmidt 1999: 1), or those of some typically masculine and typically feminine vowel-ending nouns in Pashto (Robson & Tegey 2009: 726-728).

The only other inflectionally distinct case for nouns is the genitive, although it must be considered more peripheral than the basic distinction between nominative and oblique. The evidence for ergative, instrumental and vocative cases, respectively, will also be discussed below, whereas a rather large number of caselike functions expressed with postpositions will be discussed in §8.2.

### **5.5.1 Nominative case**

The nominative is the form of the noun used as the citation form, the subject of intransitive verbs, as in (19), and as the direct object, as in (20), of most transitive verbs. In the non-perfective categories, it is also the case with which subjects of transitive verbs occur, as in (21).


For all nouns ending in a consonant, the nominative is identical to the noun stem, whereas for nouns ending in gender-typical vowels, it could be argued that these vowels are in fact nominative case-markers. The latter interpretation may be especially relevant for those nouns ending in *u* and *i*, for which there are distinct vocative forms (see §5.5.4) without gender-typical "nominative" endings.

### **5.5.2 Oblique case**

For those nouns that take an oblique case suffix, this is the form the noun occurs in when followed by a postposition, as in (22), when serving as the agent of

### 5 Nouns

perfective transitive verbs, as in (23), and when a noun is used as an adverbial of time or place, as in (24).


This is also the form used for the causee in causative constructions, as in the case of 'son' in (25).

(25) *ma* 1sg.nom *teeṇíi* refl *putr-á* son-obl *čéi* tea *pila-áan-u* give.to.drink-prs-msg 'I make my son drink tea.' (A:DHE6693)

As mentioned above, only in some of the declensions (see §5.6) is there a nominative vs. oblique differentiation. The singular oblique is formed by a suffix *-a, -í*, etc. added to the noun stem, alternatively by an ending *eé, a*, etc., replacing the ending vowel of the nominative. In most declensions, the oblique singular is formally identical to the nominative plural. The oblique plural is formed by a suffix *-am, -óom, -íim*, or *-eém* added to the noun stem, except for nouns forming a plural with *-aán*, for which the plural oblique suffix *-óom* (or *-úm*) may be attached subsequent to the plural suffix.

### **5.5.3 Genitive case**

The genitive is used for the noun that heads a possessive construction, see (26)– (27), and for an ablative function, either along with an additional postposition, as in (29), or alone, as in (28).

(26) *khanɡar-íi-e* sword-obl-gen *záxum* wound *lab* quickly *saáz* whole *bh-áan-u* become-prs-msg 'The wound of a sword heals quickly.' (B:PRB018)

5.5 Case


The genitive is also used for the object in constructions with the verb *ǰe-* 'hit', see (30). The background of this special case may possibly be explained (as Baart 1999a: 43, has done for Kohistani Gawri) with an implicit object with the nominal meaning 'hit'.

(30) *ma* 1sg.nom *paalaá* leaf *ɡhin-í* take-cv *phút-am-e* fly-obl.pl-gen *ǰ-áan-u* hit-prs-msg 'I'm driving away flies with leaves.' (B:DHN4851)

The genitive case inflection is less variable than the oblique. In A, the genitive is formed with an invariable *-ii* (accented or unaccented, depending on declensional membership), whereas in B it has one accented form *-í* and one unaccented *-e* (also depending on declensional membership). It can be argued that the genitive belongs in a layer outside of or based on the oblique. This makes most sense for the genitive plural, which is formed by attaching *-ii* to the oblique plural form of the noun: *-am + -ii > -amii, -íim + -ii > -íimii, -óom + -ii > -óomii*. The genitive singular on the other hand is usually the noun stem followed by the genitive suffix, save for the nouns of the *i*-declension in B, where the (strengthened) oblique singular suffix mediates between the stem and the genitive suffix: *ḍheer-íi-e* 'belly-obl-gen'. It is, therefore, possible that an old genitive plural has been replaced by this new "peripheral" genitive, formed analogically from the singular genitive.

### **5.5.4 Other cases or case-like categories**

A few other case categories should also be mentioned, although they are rather marginal or seldom used.

### 5 Nouns

One of the expressions of an *instrumental* function, or possibly case, is (as far as I have evidence) formally identical to the oblique plural and occurs in the form of 'gun' in example (31), regardless of number reference.

(31) *se* def *míiš-a* man-obl *ba* top *huṇḍíi* from.above *thíi* from *se* def *bhalá-ii* spirit-gen *toobak-íim* gun-ins *ǰít-i* hit.pfv-f 'The man shot from above with his gun at the spirit.' (A:WOM671)

This example of case syncretism is in itself a hint that the origin of the ergative marking is in the instrumental. However, instrumentality is also, and more frequently, expressed with a converb (see §14.4.1), such as *ɡhiní* from *ɡhin-* 'take' (as in example (30) above).

Unique *vocative* forms (whether we regard it as a case or not) of nouns only occur with some kinship terms, as in (32)–(33). These forms are shorter than the corresponding nominative forms, appearing without their "gender signaling" *u*or *i*-endings.

(32) *phéep* father's.sister.voc *séer-i=ee* fine-f=q

> 'How are you auntie [politely addressing any middle-aged woman]?' (A:HLE3088) [*phéepi* in nom.sg]

(33) *dóod* father's.father.voc *sóor-u=ee* fine-msg=q

> 'How are you grandpa [politely addressing any aged man]?' (A:HLE3090) [*dóodu* in nom.sg]

A great number of other case-like meanings are expressed with the help of postpositions, usually preceded by the noun in its oblique case form (see §8.2 for a further discussion). Postpositions tend to receive stress like other free morphemes, but in a few cases, where the postposition is de-accented, we may see the early grammaticalisation of a postposition into a case suffix. This is especially true of the postposition *wée* 'in, into', as can be seen in (34), which in some speech styles is phonologically fused with the singular oblique suffix, possibly developing into a more specialised locative case suffix.

(34) *so* he *meedóon-ee* field-"locative" *(=meedóon-a* [=field-obl *wée)* in] *nikháat-u* appear.pfv-msg 'He appeared in the field.' (A:ROP003)

5.6 Declensions

### **5.6 Declensions**

The formal realisations of the categories – gender, number and case – can be grouped into inflectional classes, which I will refer to as noun declensions. There are three main declensions (to which between 80 and 90 per cent of all nouns belong) and a few minor ones. These are primarily based on the various plural formations (hence the declension labels, "*a*-declension", etc., given here) and to a lesser degree on oblique case forms. Masica (1991: 219) points out that gender in NIA "often entails declensional differences". That is true to some extent also with Palula, but there is far from any one-to-one mapping between gender and declensional affinity. A general overview of the various declensions is displayed in 5.9, their characteristic number and case formations, and the gender categories that are represented within each of them.


Table 5.9: Noun declensions, an overview

Below I present the declensions, one by one; there is a short characterisation of each (including phonological forms and any correlation to semantic content), its relative frequency,<sup>5</sup> a number of examples, important subgroups, and some suggestions as to the historical development as far as it is traceable.

### **5.6.1** *a-***declension**

Nouns belonging to the *a*-declension form their plural and oblique case with an *a*-suffix, as is evident from Table 5.10. This is the largest declensional class as far as my data goes with 50 per cent of my nouns belonging to this particular

<sup>5</sup> Based on a database with about 1,700 nouns.

### 5 Nouns

declension. A clear majority (79 per cent) of them are masculine. A great number of toponyms are also included.


Table 5.10: *a*-declension nouns

Quite a number of nouns in this declension can be traced back to the large OIA declension of masculine and neuter nouns ending in *a*. The OIA neuter nouns have largely fused with the masculine. In the nominative singular there is a regular loss of the final OIA segment *a(s)*, as in a number of other NIA languages (Masica 1991: 222), while the Palula plural inflection *-a* may reflect one or more of the OIA dual or plural forms that include a long *ā: kā́māu, kā́mās, kā́mān, kā́mānām* (of *kā́ma* 'love', see Whitney 1960 [1889]: 330). I would like to suggest the diachronic developments (partly based on Turner 1966) in (35).

(35) *anɡóor* 'fire' (m) < \**anɡáar* < \**ánɡaar* < *ánɡāra-* 'glowing charcoal' (m, n) *chíitr ̣* 'field' (m) < \**chéetr ̣* < *kṣḗtra-* 'land' (n) *bíi* 'seed' (f) < \**bīya* < *bīǰa-* 'seed, semen' (n) *ṣínɡ* 'horn' (m) < \**ṣínɡa* < *šr̥nɡa-* 'horn' (n)

Also a few nouns of the OIA declension with stems ending in a consonant have ended up here, particularly neuters with a final *n* in their stems. Some examples are shown in (36).

(36) \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   
 \   

However, this is far from a totally uniform declension in Palula. Nouns with an alternating accent (as explained in §3.5.1) form their plural oblique with *-óom*

### 5.6 Declensions

instead of *-am*, exemplified in Table 5.11, due to an historical tensing and raising of the accented vowel (*ám > áam > óom*).


Table 5.11: *a*-declension nouns with accent shift

With a rather high level of confidence we can trace this Palula accent shift to an originally stressed (or accented) root final *-á* (found among OIA nouns with a stem ending in *a*), which was deleted through apocope some time between MIA (Pischel 2011 [1900]: 247–248) and the emergence of a Palula proto-language (in proto-Shina or at an even earlier stage), leaving the last remaining vocalic mora of the noun in the nominative singular with the accent, while preserving it on the original segment in the inflected noun forms, as in the OIA *devā́s, devā́n, devā́nām*, the nominative, accusative and genitive plural forms of *devá* 'god' respectively (Whitney 1960 [1889]: 330). The proposed diachronic developments are shown in (37).

(37) *deés* 'day' (m) < *\*deesá* < *divasá-* 'heaven, day' (m) *ɡhoóṣṭ* 'house' (m) < *\*ɡhooṣṭá* < *\*ɡooṣṭhá* < *ɡōṣṭhá-* 'cow-house, meeting-place' (m) *pútr* 'son' (m) < *putrá-* 'son' (m) *ṣíṣ* 'head' (m) < *\*ṣiṣá* < *\*šiiṣá* < *šīrṣá-* 'head, skull' (n)

These words all go back to the OIA declension with nouns ending in *a* being mostly masculine, but with a few Palula masculine nouns in this group originating in OIA neuters. That final-accented nouns in OIA always end up as accentshifting nouns in Palula is, however, not an infallible rule. There are indeed counterexamples, like *díiš* 'village' (m) < *deešá-* 'point, region, part, province, country' (m), where it is likely that a stress-shift took place prior to the apocope process (an assumption that is further supported by the regular strengthening of firstmora accented *ée* into *íi* in the Palula proto-language, while the quality of second-mora accented *eé* in most cases has been preserved): *díiš < déeš < déeša < deešá-*.

### 5 Nouns

Among *a*-inflecting nouns in Palula, I have at least one example of a word that can be traced back to the rather limited OIA declension of stems with a final syllabic *r̥: dhií* 'daughter' < *duhitr̥-* 'daughter'.

A sub-irregularity in the *a*-declension is a group of nouns (Table 5.12) ending with an accented vowel *ú* or *í* that usually have a shared singular form (nominative and oblique), while the two plural cases are formally distinct.


Table 5.12: *a*-declension nouns with ending *ú* or *í*

Because of the historical development mentioned above, by which accented *á* was strengthened, some nouns (examples in Table 5.13) now show an alternation (in A) between a long (accented) vowel *áa* or *aá* in the singular and a short (unaccented) vowel *a* in the inflected forms. All disyllabic nouns and monosyllabic nouns with aspiration (including a preceding *h*) have developed a second-mora accent *aá* on the lengthened vowel.

Table 5.13: *a*-declension nouns with length alternation


There are also many highly frequent nouns (examples in Table 5.14) that in the nominative singular end with an unaccented *u*, as already pointed out in connection with gender assignment, where this ending *u* does not appear in the plural or in any of the case inflected forms. This could in fact qualify as a very important subdeclension.

### 5.6 Declensions


Table 5.14: *a*-declension nouns with ending unaccented *u*

Morgenstierne (1941: 15), as well as Buddruss (1967: 29), suggests the OIA derivation *-aka-* (also belonging to the large a-ending declension in OIA) as the origin of these typically masculine endings. According to Masica (1991: 222), this *-aka-* (in the nominative singular *-akas*) has been subject to weakening via *-akō > -aɡō > -ahu > -au*, ending up as *-ō* or *-ā* in many NIA languages. That may be so, but I also hold it for very likely that this rather major group of masculine nouns has been further expanded through analogy with the *aka*-derived nouns, some of them possibly with an adjectival origin. In any case, for a few Palula nouns with an ending *u* there is indeed evidence of OIA cognates (or established reconstructions) with *-aka-* (or *-uka*) formation, see (38), although it will be necessary to posit a stress or accent-shift (due to reasons I am not able to formulate now) to have taken place for those words that were stressed on the final syllable in OIA.

(38) *bóolu* 'hair' (m) < *\*báalo* < *\*báalau* < *vālaka-* 'tail of horse or elephant' (m) *tóoru* 'star' (m) < *\*táaro* < *\*táarau* < *tāraká-* 'belonging to the stars' (m)

Earlier, a derivational suffix forming adjectives from nouns, *-aka-* (and a group of similar endings) included already in OIA diminutive formations as well as a number of less easily definable noun-to-noun derivations. In the older sources, one basic form occurs as well as a derived form with seemingly identical semantic content (Whitney 1960 [1889]: 1222). The above exemplified *vālaka-* 'tail of horse or elephant' is for instance derived from *vā́la-*, glossed very similarly as 'hair of tail, tail, hair' by Turner (1966: 12056). Masica (1991: 222) describes it as a diminutive suffix, which later became a meaningless "extension". There is a corresponding (Whitney 1960 [1889]: 1181, 1222) feminine formation made with *-ikā* that will be discussed below. However, as pointed out already, we will certainly find nouns in this declensional subgroup with an altogether different origin. One

### 5 Nouns

such example is *híṛu* 'heart' (m) < *hr̥daya-* 'heart', where the final *-aya-* may have gone through a weakening process, similar to that of *-aka-*, also ending up with a final *u* vowel.

In a few nouns (Table 5.15) with an accented final *ó* in the nominative singular, this *ó* also disappears in the inflected forms, and the inflectional endings are *á* and *óom*, respectively.



Another relatively large group of nouns (Table 5.16), ending in *ái*, may be considered a subdeclension and may eventually develop into a declension of their own or even fuse with the *ee*-declension (see below), one of the minor declensions described below.<sup>6</sup>

Table 5.16: *a*-declension nouns with ending *ái*


The actual pronunciation of the inflected forms seems to be rather variable, with a preserved ending *a* (in line with the typical *a*-declined nouns) in some of the variant forms. These nouns are almost exclusively feminine.

<sup>6</sup> Note, however, that the ending with the feminines is a first-mora accented *ée*, whereas the ending of the masculines is a second-mora accented *eé*.

### 5.6 Declensions

### **5.6.2** *i***-declension**

Nouns belonging to the *i*-declension form their plural and oblique case with an *i*-suffix, as exemplified in Table 5.17. This is the second-most frequent noun declension with 20 per cent of the nouns in my database belonging to this declension. Most of them, about 70 per cent, are feminine. These nouns regularly form their plural and oblique forms with an accented suffix.

Table 5.17: *i*-declension nouns


Like the *a*-declension this is far from a uniform class, and in some cases it is not entirely clear whether a noun should be included or rather be considered part of a separate declension. I have, for instance, chosen to regard a group of masculine nouns ending in *oó* or *á* as a separate minor declension (see §5.6.4), although it may possibly be analysed as part of the *i*-declension. Occurring particularly frequently in this declension are plurals with an additional umlaut (Table 5.18), of which almost all are feminine, a considerable number of them relatively recent loans from other languages.

Table 5.18: *i*-declension nouns with umlaut


### 5 Nouns

As with the *a*-nouns, there are several *i*-nouns that have gone through vowel lengthening in their basic form but have kept a short vowel in their inflected form, as in the examples in Table 5.19. The origin of this declension is less straightforward as far as OIA is concerned, and the history of the large bulk of this declension may be found either in MIA or in rather recent developments. It is indeed home to a considerable amount of "modern" loans from Urdu.

Table 5.19: *i*-declension nouns with length alternation


The nouns for which I have been able to trace an OIA cognate and also am able to hypothesise on their development into the Palula form, as shown in (39), are either from the OIA declension with stems ending in a short vowel *i* or *u*, the declension with stems ending in a long vowel *ā, ī, ũ*, or are derived nouns in OIA, ending in *-ya*.

(39) *díṣṭ* 'hand-span' (f) < *diṣṭi-* 'a measure of length' (f) *ɡrheéṇḍ* 'knot' (f) < *\*ɡrheeṇḍí* < *\*ɡraaṇḍhí* < *ɡranthí-* 'knot, etc.' (m) *bheéṇ* 'sister' (f) < *\*bhaíṇ* < *\*bhaiṇi* < *bhaɡinī-* 'sister' (f) *dharaáṇ* 'ground, earth' (f) < *\*dharáṇ* < *dharáṇī-* 'ground' (f) *kúḍ* 'wall' (f) < *kuḍya-* 'wall' (n) *muúl* 'price, value' (f) < *mũlya-* 'original value, price' (n)

The proposed intermediary forms remain tentative, although most of the processes are evident from parallel comparative material, such as apocope, umlaut, forward shift of aspiration, *a*-lengthening and intervocalic lenition of plosives.

### **5.6.3** *m-***declension**

Nouns belonging to the *m*-declension typically form their plural with an *m*-suffix, but they do not inflect for oblique case. Examples are displayed in Table 5.20. According to my data, this declension is slightly smaller (at 16 per cent) than the

### 5.6 Declensions

*i*-declension. It is also the most homogenous of the main declensions, as all nouns in this class are feminine,<sup>7</sup> and their basic form ends in *i* (unaccented except in a few cases).


Table 5.20: *m*-declension nouns

This pattern is the result of a rather recent historical process that is only partly clear to me. The basic assumption is that the previous plural marking (as far as there was an overt marking at all) disappeared in a process of apocope that affected all final unaccented vowels, and the plural oblique (with an *m*) was (maybe compensatorily) extended to all nouns with a plural reference, hence becoming a plural-marker rather than a case-marker.

The development of this declension is to some extent a feminine parallel to that of the masculine *u*-ending nouns of the *a*-declension. A feminine derivational suffix *-ikā-* was used in OIA in much the same way as the above-mentioned  *aka-* (Whitney 1960 [1889]: 1222), while the NIA endings with their origin in *-ikā-* probably retain much more of the Sanskritic diminutive sense (Masica 1991: 222). OIA nouns with the ending *-ikā-* are part of the OIA declension 3 together with other nouns ending in a long vowel. Some examples of Palula *m*-declension nouns with probable OIA *ikā*-cognates are displayed in (40).

(40) *čhéeli* 'goat (she-goat)' (f) < *\*čhaali* < \*čhawali < *čhaɡalikā-* 'goat' (f) *déeṛi* 'beard' (f) < *\*daaṛi* < *dāḍhikā-* 'beard' (f) *béeǰi* 'heifer' (f) < *\*bíadzi* < *dvivatsikā-* 'two year old' (f)

However, there are nouns in the Palula *m*-declension for which we cannot be absolutely certain that they derive from OIA nouns with an *ikā-*suffix.<sup>8</sup> They

<sup>7</sup> A possible counter-example is the Pashto loan *malɡiri* 'friend, companion', with the plural form *malɡirim*, which seems to be assigned gender referentially rather than having an inherent and invariable gender.

<sup>8</sup> At least there are no *ikā*-formations occurring in the older literature, which of course is no proof they have never existed.

### 5 Nouns

may, for instance, derive from nouns ending in a long vowel *ī* (*séeti* 'thigh' (f) <? *sakthī-* 'thigh, thighbone' (f)) or nouns ending in a short i (*héeṛi* 'duck' (f) <? *ātí-* 'an aquatic bird' (f)) and have, for reasons I am presently unable to explain, come to declensionally converge with the above-mentioned *ikā-*derived nouns and not with the Palula *i*-declension nouns. Thus Masica (1991: 222) comments on the development of the NIA feminine marker (such as the ending *i* of the *m*declension nouns in Palula): "Its evolution was no doubt influenced, however, by the existence of a Feminine in *-ī*, at times restrengthened, in all periods of the language."

Although regular masculine–feminine pairings with the endings *aka* and *ikā* respectively seem to have been common already in OIA, I would not suggest that similar pairings so common in Palula all go back to nouns with these suffixes. It is, I think, much more likely that many of them were formed at a much later stage, maybe as feminine derivations of semantically generic masculine nouns, or are both adjectives that have come to be increasingly used as nouns. My hypothesis is that the more lexicalised pairs (including those where the feminine counterpart is primarily a diminutive, see Table 5.3) belong to an older layer, whereas simple male–female pairs (see Table 5.2) are formed to a large extent at a later stage, somewhat in analogy with the former.

Also nouns ending in unaccented *u*, such as the ones in Table 5.21, should probably be considered part of this declension. These, however, show some instability in their inflectional pattern (with an alternative paradigm: *pháapu, pháapa, pháapam*), and it is possible that in spite of their being feminine are becoming part of the subgroup of the *a*-declension, which drops the final *u* when inflecting. This may or may not be the first step towards a subsequent gender-shift. Historically, the few nouns of the *m*-declension that end in a short unaccented *u* probably derive from OIA feminines with a suffix *-ukā*, such as *práašu* 'rib' (f) < *paršukā-* 'rib' (f).


Table 5.21: *m*-declension nouns with ending unaccented *u*

5.6 Declensions

### **5.6.4 Smaller declensions and irregular nouns**

Apart from the main declensions, there are groups of nouns that, for various reasons, do not conform to any of the previously introduced declensions, although in many respects they seem to share features with one or the other of them.

### *ee***-declension**

A group of nouns, presently making up about eight per cent of my database, has a plural form ending with a second-mora accented *eé* (Table 5.22). Characteristically the nominative singular ends in an accented *á*, and the *eé* may be described as the result of hiatus between the stem *á* and the suffix *-í*, thus basically qualifying to be included in the *i*-declension. (The latter is even more apparent in the B dialect, where the plural form suffix *-í*, has fused to a much lesser extent with the final stem vowel: *kuṇḍá - kuṇḍaí*.) However, contrary to the typical *i*-declension nouns, these nouns do not distinguish between nominative and oblique singular, while that case distinction is upheld in the plural. As it is phonologically rather distinct from other *i*-declension nouns in the present stage of the language (at least in the A dialect), I prefer to treat it as a declension of its own. This declension includes a number of nouns with a long history in the language, but at the same time, it seems a rather productive one as far as incorporation of more recent loans are concerned.

All of the nouns from this declension for which gender is known are masculine. This should be compared with the group of nouns, already introduced under the *a*-declension, with a singular *ái*-ending and a plural *ée*-ending (i.e., first-mora accented). Although their plural forms are segmentally very similar to the ones discussed here, I hesitate to include them in this group, as they behave like *a*declension nouns (and in an alternative pronunciation receive a plural ending *-éea* or *-áya*) in most other respects. In addition, almost all of those are feminine.


Table 5.22: *ee*-declension nouns

### 5 Nouns

This group of nouns contains a large number of non-inherited words, some rather recent loans from Urdu or Pashto, others with a longer history in the language, but only a few that can be regarded as inherited from OIA. The only one I have been able to trace comes from the *a*-ending declension in OIA, and here it is likely that apocope along with stress on the remaining final segment has resulted in a second-mora accented long *aá* in Palula: *paalaá* 'leaf' (m) < *\*paalaawá- < pallava-* 'sprout, twig, blossom' (m, n).

Another, very limited group of masculine nouns, with ending *oó* in the singular (Table 5.23), behave in a very similar manner, and show similar forms in the plural, but they tend (with a few exceptions) to use the same form for the singular oblique as for the plural nominative. It makes most sense to consider these nouns as another subcategory of the *ee*-declension.

Table 5.23: *ee*-declension nouns with ending *oó*


This group is closely related to the *u*-ending nouns of the *a*-declension (in the closely related variety spoken in Sau, Afghanistan, these two build up a declension of their own vis-à-vis the other nouns in the *a*-declension). Like those, the *oo*-ending nouns derive from the OIA declension of *a*-ending nouns, as shown in (41).

(41) *haṇoó* 'egg' (m) < *\*haṇóo-* < *\*aaṇáa-* < *āṇḍaka-* 'egg; testicles' (n) *ǰhamatroó* 'son-in-law' (m) < *\*ǰamatróo-* < *\*ǰamaatráa-* < *ǰāmātraka-* 'daughter's husband' (m) *muuṣoó* 'elbow' (m) < *\*muuṣóo-* < *\*muuṣáa-* < *\*mũṣala-* 'muscle'

For the first two words we have to assume a change in accent-pattern in the development into modern Palula. A second-mora accented long *aá* would have kept the vowel quality, whereas there is a regular development of first-mora accented long *áa* into *óo*, so we assume that after the apocope had taken place we were left with a final long *áa* with a first-mora accent. This could be compared

### 5.6 Declensions

with two other (irregular) *oo*-ending nouns, where the *aa* is preserved in the inflected forms: *phoó* (nom.pl *phaayá*) 'boy' < *\*pháa*, and *bhroó* (nom.pl *bhraawú*) 'brother' < *\*bhráa*. To that should be added that we find some irregularities in this group that may constitute remnants of an earlier inflectional pattern, such as with words referring to 'descendant of so and so' as with *phaṭakoó* 'descendant of Paṭak': *phaṭakeé* nom.pl, *phaṭakeé* obl.sg, *phaṭakúm* obl.pl.

### *aan***-declension**

Although nouns forming plural with *-aán* (Table 5.24) in many ways could be described as a subcategory of the *a*-declension, it is an interesting group (making up less than five per cent of all nouns) because of its maximum formal number and case differentiation and also because of the semantics of these particular nouns. While in the main declensions there is either a formal collapse between the nominal plural and the oblique singular or a total absence of nominal–oblique case distinctions, there is a four-way contrast displayed for many of the *aan*nouns.


Table 5.24: *aan*-declension nouns

Almost all of the nouns in my database with *-aán* as a plural marker denote male higher animates, often referring to occupations, but there are apparent exceptions such as inanimate *aaluɡaán* 'potatoes', low animate *traambuaán* 'wasps' and the abstract noun *hamaliyaán* 'habits'.

This declension is heterogenous in the sense that many of the nouns have alternative forms. Many of the nouns are basically accent-shifting *a*-declension nouns, but have alternative plural forms with a plural suffix *-aán* and a plural oblique in *-aanóom* (or for some nouns *-aanúm*). To some extent this may be

### 5 Nouns

an effect of borrowing, where the plural form of some Pashto nouns along with its suffix has also been copied *saṇḍá* 'male buffalo' – *saṇḍaɡaán* 'male buffaloes'. It this case, it is not entirely clear whether the word is the one inherited from OIA (*sā́ṇḍa* 'uncastrated (of bull))' that later acquired its morphology through Pashto influence or as a whole has been reintroduced via Pashto (*saṇḍa* 'a male buffalo') rather recently. But whatever the origin may be in every single case, it has become a very common pattern, especially for nouns with male human reference, even with words that are very clearly inherited from OIA, such as *saaṇḍú* 'wife's sister's husband' (< *\*sāˊ(ṇ)ḍhu-*) – *saaṇḍuɡaán*. In the latter case, with its epenthetic *-ɡ-* in the plural, it is tempting to suggest a form-analogy with the already mentioned *saṇḍá – saṇḍaɡaán*. *-aan* is a common plural formation for nouns denoting human referents not only in Pashto but also in the locally influential languages Persian (Windfuhr & Perry 2009: 431) and Khowar (Endresen & Kristiansen 1981: 221–225).

### **Irregular nouns**

There is a small group of nouns, exemplified in Table 5.25, that either have unique paradigms or are otherwise highly irregular, and in some cases also display alternative forms for some categories. Most of these are kinship terms. In the case of *maámu/maamaá*, there is probably some sort of paradigm mixing going on, although there may have been a distinction in the past.


Table 5.25: Irregular nouns

## **6 Pronouns**

### **6.1 Introduction and overview**

Palula is essentially a 'two-person' language (Bhat 2004: 4–15), having distinct first- and second-person (singular, plural) pronouns, i.e., for the speech act participants, whereas third person is expressed by forms belonging to the category of demonstratives. Third person displays a number of properties (such as distinction in gender, distance and accessibility) not shared with the first- and secondperson pronouns, but that are shared to a large extent with a larger category of pro-forms, which build a number of symmetrical sets that cut across several word classes. Personal pronouns proper, other types of pronouns as well as demonstratives (pronominal and adnominal) and with them related forms are treated in this chapter.

### **6.2 Personal pronouns**

The personal pronouns, 1sg, 2sg, 1pl and 2pl, occur in two to four case forms each, as displayed in Table 6.1. <sup>1</sup> While the singular persons make a two-way nominative-genitive distinction, the plural persons make a four-way distinction between nominative, accusative, genitive and ergative.

That means that, whereas most nouns only distinguish between nominative and oblique case (beside the genitive forms), and the agent of transitive perfective clauses is expressed in the oblique, the plural personal pronouns have a unique ergative pronominal form; compare the ergative 1pl *asím* in example (2) with the nominative *be* and accusative *asaám* in example (1).

<sup>1</sup> Throughout this work, the pronouns are glossed according to formal distinctions only. That means, 1sg *ma* and 2sg *tu* are glossed as nom regardless of their functions (as subjects, direct objects or postpositional objects, etc.), and 1sg *míi* and 2sg *thíi* are glossed as gen regardless of their functions (as heads of possessive constructions or subjects of perfective transitive clauses, etc.).

### 6 Pronouns


Table 6.1: Personal pronouns

*<sup>a</sup> máa=the* 'to me' is special. As far as I know, it is only together with this particular postposition that the vowel of the pronoun is lengthened. What this shows is that the pronoun forms a phonological word with the postposition and has gone through regular *a*-lengthening.

(1) *be* 1pl.nom *eetáa* there.rem *yhóol-a* come.pfv-mpl *ta* when *hiimaál* avalanche *čhinǰ-í* strike-cv *asaám* 1pl.acc *híṛ-a* take.away.pfv-mpl

'When we came there, an avalanche hit and swept us away.' (A:ACR012)

(2) *asím* 1pl.erg *tu* 2sg.nom *na* neg *bulaḍíl-u* call.pfv-msg *hín-u* be.prs-msg 'We haven't called you.' (A:GHU030)

The first- and second-person singular pronouns on the other hand make fewer distinctions than most full nouns, with their two-way case contrast. As in Kohistani Shina, the oblique case has merged with the nominative (Schmidt & Kohistani 2008: 82). The nominative is thus used for intransitive subjects, as the 1sg *ma* in example (3), all direct objects, as the 1sg *ma* in example (5), as well as for transitive agents in the imperfective, while the genitive covers the possessor of a possessive construction, as the 1sg *míi* in example (3), as well as the agent of perfective clauses, as exemplified by 1sg *míi* in (4).


6.2 Personal pronouns

(5) *lo* 3sg.dist.nom *ma* 1sg.nom *kha-áan-u* eat-prs-msg *than-íi* say-3sg *de* pst 'He said, "It will eat me up".' (A:GHA018)

The form used for direct objects is also the form preceding postpositions. It would be possible to analyse some of those combinations of pronoun and postposition (examples (6) and (7)) as case forms, as the two tend to be phonologically slightly more fused as compared to what is observed with most combinations of full nouns and postpositions. However, for the sake of descriptive economy, I have here chosen to treat them as postpositions.


There is nothing comparable in Palula to the honorific levels in second-person pronouns found in regionally influential Urdu (Schmidt 1999: 17), and to a lesser degree in Pashto (own personal observations). The use of the second-person singular is in itself not perceived as derogatory or unsuitable, and neither is the second-person plural pronoun ever used with singular reference in Palula.

To personal pronouns can also be added a particle of emphasis or exclusiveness, *eé* (in B *e*), to be compared with exclusiveness as expressed with numerals §7.4.1. It seems to be most frequently used together with the genitive form of the firstand second-person singular pronouns. In example (8) it has the approximate meaning 'my *own*'.

(8) *fazelnuúr* Fazel.Noor *ba* top *míi* 1sg.gen *e* ampl *ɡaaḍubáabu* grandfather *de* be.pst 'And Fazel Noor was my own grandfather.' [Preceded by a genealogy starting with a distant forefather.] (B:ATI078)

However, in uses with nominative pronominal forms, as in (9), it takes on a meaning similar to 'I myself'. Note that this, in Palula, is a function distinct from that of the reflexive pronouns (see §6.4).

### 6 Pronouns

(9) *típa* now *aní* prox *mauǰudá* present *waqt-íi* time.gen *ma* 1sg.nom *eé* ampl *bakáara-m-ii* herd-obl-gen *kasúb* occupation *th-áan-u* do-prs-msg 'Nowadays I myself am engaged in shepherding.' (A:KEE079)

A reduplicated form of a pronoun and the suffix, as in (10), is used in approximately the same sense.

(10) *mam* 1sg.nom.red *eé* ampl *ɡíi* go.pfv.fsg *hín-i* be.prs-f *aḍaphaár* halfways *tií* until 'I myself went halfways.' (A:CAV010)

The particle *eé* can also be used with a demonstrative, as in (11).

(11) *dúu* two *ta* cntr *míi* 1sg.gen *ni* prox *eé* ampl *achíi-a ̣* eye-pl *dúu* two *ba* top *tóoruṇ-a* forehead-obl *wée* in 'Two were those very eyes of mine, and two were up on my forehead.' (A:HUA114)

### **6.3 Demonstratives**

### **6.3.1 Relationship to the larger pro-form system**

As mentioned in the introduction to this chapter, the demonstrative pronouns are part of a larger system of pro-forms (pro-NPs, pro-adverbs, pro-adjectives, etc.), displaying a certain degree of symmetry (Table 6.2). Most (but not all) of those pro-forms have similar-looking sets consisting of a proximal, a distal, a remote and an indefinite/interrogative member (see Haspelmath 1993: 187–188 for some strikingly similar sets in Lezgian).

Somewhat simplified, a consonantal element (*-n-, -ṛ-, -s-/-t-* or *k-/ɡ-*) regularly indicates the deictic function. Additional dimensions are relevant to certain subsets, such as case, number, gender and emphasis for the adnominally and nominally used demonstratives, animacy for indefinite/interrogative pronouns, and spatial specification (primarily) for locational pro-forms. For further discussion and examples, see §6.3.2, §6.3.7, §8.1.1–§8.1.2, §8.1.4–§8.1.5, §15.2.2, §15.2.4 and §15.3.2.

### 6.3 Demonstratives


Table 6.2: Correlations between pro-forms

*a* It is only with a following postposition that these forms have a locational interpretation. Used alone they (ergatively) code the subject of a perfective transitive clause.

### **6.3.2 Demonstratives and third person**

As already hinted at above, a basic three-way distance differentiation is used extensively with demonstratives and has at least to a certain degree been extended to the realm of unstressed third-person reference. I have chosen to regard all of these as demonstratives, of which many can and are indeed being used as thirdperson pronouns, some more so than others. It is even typologically speaking problematic to make any clear-cut distinction between personal pronouns and demonstratives, a fact pointed out, among others, by Himmelmann (1996: 206) and Kibrik (2011: 123–124), which is why it makes most sense to present them together. A distinction, however, that needs to be made within the category of demonstratives, is that between pronominal (or pro-NPs) and adnominal demonstratives. As pointed out by Himmelmann (1996: 206), pronominally used demonstratives tend, from a typological perspective, to occur with a lower frequency and in fewer contexts than adnominally used demonstratives. Since the pronominal demonstratives often are derived from the adnominal ones, the former are usually more complex.

As can be seen in Table 6.3, the basic adnominal demonstratives occur in only two different forms in each deictic set, as compared to the much more complex system of pronominal demonstratives that we will discuss shortly. The only contrast is one between the demonstrative functioning attributively with a mascu-

### 6 Pronouns

line singular nominative referent vs. a referent agreeing with its head in any other case, gender or number category. The forms are identical to the pronominally used masculine singular nominative and feminine singular nominative, respectively (see below), a situation not uncommon in languages in general (Himmelmann 1996: 214).


Table 6.3: Adnominal demonstratives

In example (12), the masculine noun *xooṛá* 'betrothal' first occurs in the nominative and is preceded by the adnominal demonstrative in its masculine singular nominative form, then it occurs in the genitive and is then preceded by the adnominal demonstrative in its other (non-nominative, plural or feminine) form.

(12) *khayí* which *dees-á* day-obl [*eeṛó* dist.nom.msg *xooṛá*]nom betrothal *hensíl-u* stay.pfv-msg *de* pst [*eeṛé* dist *xooṛá-ii*]gen betrothal-gen *ṭeem-í* time-obl *ǰhaamatroó* son-in-law *mooǰúd* present *na* not *hensíl-u* stay.pfv-msg *heentá…* would 'If, on the day that betrothal took place, the son-in-law were not present at the time of that betrothal…' (A:MAR116-7)

The relevant categories for the pronominally used (or substantivised) demonstratives, presented in Table 6.4, are number, case, gender and distance/visibility, the first three also found with Palula nouns.<sup>2</sup> We can also differentiate between so-called strong and weak forms of demonstratives (or at least for most of them). Initial *ee-* in the strong forms in A (the ones displayed in Table 6.4) regularly corresponds to B *ha-*. The pronouns in the weak remote sets of the demonstratives (*so*, etc.), are the default choice and most frequently used as typical third person pronouns, with easily accessible discourse referents (Diessel 2006: 432–433), and could be said to stand particularly close functionally to the personal pronouns proper.

<sup>2</sup> Some of the forms do not occur in text data but are affirmed through direct elicitation as possible forms.


Table 6.4: Pronominal demonstratives (Only markedly different B forms are cited in the table.)

### 6 Pronouns

The masculine–feminine gender distinction is found only in the singular nominative, compare (13) and (14), whereas masculine and feminine are syncretised in the plural, as is evident when comparing (15) with (16).


The primary function of the distance differentiation is to distinguish three basic degrees of distance from the speaker/experiencer, where the proximal is used with referents close at hand, the distal with referents further removed from the speaker, and the remaining category, here labeled "remote", is used to refer to something or someone out of sight.

The weak–strong opposition is related to accessibility. The strong forms tend to be picked for deictic or anaphoric use in order to keep track of less accessible discourse referents, or to contrast another referent or switch from one referent to another. They are also the ones used in relative-correlative clauses and in most cases as adnominal demonstratives. In the example sentence (17), the speaker has for some time been talking about making doors when building a house, but is here switching to talking about windows instead. The weak forms, on the other hand, are used almost exclusively non-contrastively and to pick out the most natural or accessible referent in a discourse. In example (18) the antecedent of the pronoun (the cupboard) is very close and there is no ambiguity in what is referred to.

(17) *seentá* when *eetás* 3sg.rem.acc *samóol-ii* build.pptc-gen *pahúrta* after *théeba* then *khiṛkí* window *bi* also *tarkaáṇ* carpenter

6.3 Demonstratives

*khooǰa-áan-u* ask-prs-msg 'Then when that has been built, the carpenter will ask about the window.' (A:HOW044)

(18) *aalmaaríi* cupboard.gen *bi* also *muxtalíf* different *dizeen-í* design-pl *yh-éend-i* come-prs-f 'A cupboard can also have different designs.' *tasíi* 3sg.gen *xaaneé* shelf.pl *muxtalíf* different *muxtalíf* different *yh-áand-a* come-prs-mpl 'It can have many different kinds of shelves.' (A:HOW049-50)

Thus, it is the weak form that functions most like a third-person pronoun. The contrast is similar to the distinction Givón (2001a: 417–419) makes between "stressed independent pronouns" and "unstressed anaphoric pronouns". Palula has also developed definite articles, formally identical to the nominative weak remote/non-visible demonstratives (see §6.3.6). This strong–weak continuum mirrors in many respects the relative anaphoric distance of Givón (2001a: 419), as well as the grammaticalisation cline described by Diessel (2006: 432). However, the absence of an absolute distinction between demonstratives and third-person pronouns across the board in Palula, and the presence of a number of borderline cases, may very well reflect an ongoing grammaticalisation (Himmelmann 1996: 213). The weak distal forms with *ṛ* and *l*, respectively, are mere pronunciation variants, of which only the *l*-variant is phonologically possible in B. Nor is there anything indicating any significance in the alternation between the strong distal forms beginning with *a-* and *ee-*, respectively, but more detailed analysis is needed to be certain.

There is also another continuum that cuts across the entire paradigm. While the remote category is primarily a phoric device (Saxena 2006: 131), with a high frequency in, for instance, narrative discourse, and only secondarily functions deictically, the opposite holds for the proximal category, which is primarily deictic and only rarely functions phorically. The mid-level category, the distal, is used extensively in both functions. An indirect effect of the differences in deictic scope is the ability to signal a contrast in the presence–absence of the third-person object or person in the speech situation. It is normally only when a person or object is present in the speech situation that the proximal and distal terms are used (exophorically or anaphorically), whereas the remote assumes their absence. The same holds for a distinction between narration and direct speech -

### 6 Pronouns

especially crucial in oral story-telling. Along with other quotative devices, the proximal and distal terms refer to the outside world as perceived by the quoted speaker, whereas the narrator (whether identical to the former or not) would use the corresponding remote terms to refer to the same entities.

Schmidt (2000: 204–205, 207–212) and Schmidt & Kohistani (2001: 134–136) describe similar effects obtained in the use of different sets of pronouns in some other Shina varieties. In Tileli there are two sets of third-person pronouns, where one set signals visibility and first-hand knowledge, whereas the other one codes what is invisible, unknown or indirectly inferred. In Kohistani Shina, a similar discrimination can be made between information derived by visual means and information received by other means, through contrasting a proximal set of demonstratives with a remote set.

Three particular functions (exophoric, anaphoric and discourse-deictic) will be further discussed and exemplified below (§6.3.3–§6.3.5), following the pragmatic taxonomy of demonstratives suggested by Himmelmann (1996: 205–254) and discussed by Diessel (2006: 432). I do not have any clear examples of recognitional use, a fourth category; however, there is evidence for the development of a definite, or "anaphoric article" (Juvonen 2006: 486), contrasting with a similarly developing indefinite article. These two will be discussed in §6.3.6. A secondary spatial specification will also be touched upon briefly in §6.3.7.

### **6.3.3 Exophoric use**

The exophoric, or situational, use orients the hearer in the outside world and points to an entity in the speech situation. This could be said to be the basic or original use of the demonstrative. In Palula, it is primarily the proximal and distal levels that enter into this function, and it is almost without exception the strong forms that are being used, whether pronominal or adnominal.

The proximal term *anú, aní* 'this' points to something or someone close at hand. In its most typical use, it is adnominal while pointing to an object, as in (19).

(19) *aní* prox *kakaríi* skull *kasíi* whose *thaní* qot '[He] said, "Whose skull is this?".' (A:WOM459)

In (20), on the other hand, it refers pronominally to a man who in the quoted speakers' presence has just defeated a fierce bear, what Himmelmann (1996: 222) describes as reference in the narrated situation. In example (21), it points, in

### 6.3 Demonstratives

a similar fashion, to a boy who is threatened with being captured and taken away from his stepfather, who is the speaker within the narrative, and who is standing right next to the boy. In (22), the location is the same as where the speaker finds himself in at the speaking moment, although talking about times long ago, and is therefore an instance of situational use in the actual utterance situation (Himmelmann 1996: 222).


The proximal term could also refer to temporal closeness, such as in (23), where it refers to the time of the utterance.

(23) *aní* prox *dees-óom* day-obl.pl *atsareet-á* Ashret-obl *wée* in *qariibán* about *čúur* four *zára* thousand.pl *kušúni* households *hín-a* be.prs-mpl

'In these days there are about 4,000 households in Ashret.'<sup>3</sup> (A:PAS007)

This stands in contrast to the past times that this particular narrative focuses on, which are elsewhere in this discourse referred to as *eesé waxtíi* 'of those times', thus using the remote term.

A simultaneous (and explanatory) reference to, for example, a body part of a character in a narrative and the body part of the speaker, in (24) (accompanied by a pointing gesture), is also done with the proximal term.

<sup>3</sup> It must be assumed that the speaker means '4,000' inhabitants, although he uses a word normally used for 'household, family'.

### 6 Pronouns

(24) *áak-ii* one-gen *ta* cntr *ḍheerdáṛ* stomach.pain *nikhéet-i* appear.pfv-f *áak-ii* one-gen *ba* top *aní* prox *phiaaṛmaǰ-í* side-obl *wée* in *breéx* pain *nikhéet-i* appear.pfv-f 'One of them felt pain in his stomach, and the other in this side.' (A:GHA059)

The distal term *aṛó* (B *haṛó*) 'that' points to something a little farther away, but still fully visible and identifiable, as in (25), where it refers to something in the actual utterance situation, and in (26), where it refers to something in the narrated situation.


It is probably futile to try and give absolute measures of when the proximal term is used as compared to when the distal term is used. It is more a matter of relative distance, as perceived by the speaker/experiencer in the situation. There are no apparent animacy restrictions on any of these terms when used exophorically, which seems to be the case with the corresponding terms in, for instance, Kohistani Shina (Schmidt & Kohistani 2001: 135). In as far as the remote term is used exophorically, it is in referring to an entity that is *not* present or visible in the utterance situation.

### **6.3.4 Anaphoric use**

The anaphoric, or tracking, use is used to keep track of referents in discourse, and can been seen as being derived from the exophoric use by analogical extension, from pointing in the outside world to "pointing" to referents within the discourse. The remote is used particularly often for this, but the distal sets are also frequently used, whereas the use of the proximal seems highly limited in this regard. Depending on how easily accessible the referents are, either the strong or the weak form may be used. The more accessible, the more likely it is that the

### 6.3 Demonstratives

weak (unstressed) form is chosen, whereas the greater the need to refocus the hearer's attention, the more likely it is that the strong (stressed) form is selected.

In a passage, exemplified in (27), belonging to a procedural discourse, the remote term is used, both in its weak and its strong form.

(27) *tsiip-áan-a* squeeze-prs-mpl *tsiip-áan-a* squeeze-prs-mpl *tsiip-áan-a* squeeze-prs-mpl *tas* 3sg.acc *díi* from *áa* idef *paṇáar-u* white-msg *šay* thing *nikh-áand-u* come.out-prs-msg *tas* 3sg.acc *the* to *man-áan-a* say-prs-mpl *iští* "ishti" *eetás* 3sg.rem.acc *matíl-u* churn.pfv-msg *seentá* condh *tasíi* 3sg.gen *bi* also *ɡhiíṛ* butter *bh-áan-u* become-prs-msg 'We keep squeezing, and from it comes out a white part. We call that/it "ishti". When that has been churned, it becomes butter.' (A:KEE040-1)

Although the strong form tends to be used as an anaphoric demonstrative – with a refocusing or contrastive function (Diessel 2006: 432) – and the weak tends to be used as a personal pronoun, the distinction between the two is not all that clear and is deserving of further research.

When the distal term is used anaphorically (without any additional exophorical function), it is usually coreferential with a main character or a referent with special focus. Whereas the other referents, in an exposé on the benefits of keeping goats, in (28), are referred to by the remote term, the goats are repeatedly – although not exclusively – referred to by the distal term (mostly in the singular).

(28) *eeṛé* dist *čhéeli* goat *seetíl-i* keep.pfv-f *seentá* condh */…/ las* 3sg.dist.acc *eendáa=the* like.this=do.cv *saat-eeṇḍeéu* keep-oblg *ki* comp *tariqá* method *baándi* by *wíi* water */…/ las* 3sg.dist.acc *the* to *beezáaya* extra *ṭoḍusóol-u* cut.pfv-msg *seentá* condh *bi* also *le* 3fsg.dist.nom *xaraáp* spoiled *bh-éen-i* become-prs-f 'This goat should be cared for like this… She should be given water at the proper time… Also, if she is fed too much, it is not good for her.' (A:KEE012-4)

When the term is used in this discourse, it is first in the strong form, but as it continues to be referred to, the weak form is used, thus functioning more like a regular third-person pronoun.

In a particular historical account, of which (29) is part, two groups of people are referred to, and while the remote term is used for either of them when it is

### 6 Pronouns

clear which one is intended, the distal is used at a point when there is a need for special emphasis or clarification.

(29) *ma* 1sg.nom *díi* from *xu* but *aṛanaám* 3pl.dist.acc *mheeríl-a,* kill.pfv-mpl *uɡheeṇíi-a* Pashtun-pl *mheeríl-a* kill.pfv-mpl 'They [the people you gave me] were killed, the Pashtuns.' (A:BEZ116-7)

When a referent is present and close by in the speech situation (in (30), a shepherd's flock of sheep and goats), it is referred to anaphorically with the promixal term.

(30) *aniaám* 3pl.prox.acc *wíi* water *keé-na* why-neg *pila-áan-u* make.drink-prs-msg *thaní* qot 'Why not let them/these drink water.' (A:PAS070)

The latter could alternatively be described as an instance of the proximal term being used as (an unstressed) third-person pronoun, since there is no competing antecedent, and the pronoun is not really used in order to refocus the listeners.

### **6.3.5 Discourse-deictic use**

The discourse-deictic use has to do with entire propositions being referenced rather than with tracking individual discourse participants. Its function is to combine chunks of discourse (Diessel 2006: 432), and is thus more abstract in its scope. In this rather specialised function, it seems the distal term is used most of the time. It may be either anticipatory, as in (31), or refer back to a rather long preceding chunk of discourse as in (32).


Although the typical use seems to be pronominal, it also occurs adnominally, as in (33). It usually occurs in its strong form, but it may occasionally be less emphasised, as in (34), where the weak form refers back to the preceding proposition, thus "naming" what has already been defined through discourse.

6.3 Demonstratives


If the demonstrative is immediately introducing a chunk of discourse, the proximal term can also be used, as in (35), probably with the additional effect that the listener is invited into a private exchange between a husband and his wife.

(35) *teeṇíi* refl *maǰí* among *eenú* prox.msg.nom *mašwará* consultation *thíil-u* do.pfv-msg 'Between themselves they made this plan…' (A:WOM474)

The same 'plan' or 'consultation' is a little later, in (36), referred to with the distal term, then from the perspective of a secret listener, standing outside the house where the exchange was taking place. The remote term does not seem to be frequently used discourse-deictically. A possible example may be the one in (37), where the demonstrative refers to a point in time in a narrative, the 'night' following the events described in the immediately preceding chunk of discourse. However, this rather specialised function, Himmelmann (1996: 225) defines as a subtype of discourse-deictic use.


### 6 Pronouns

Sometimes, as in (38), there is also an ambiguity as to whether the demonstrative refers anaphorically to a particular discourse referent or to a preceding chunk of discourse. Here, the planks as well as the totality are possible referents for the demonstrative.

(38) *aalmaarí* cupboard *wée* into *ba* top *páanǰ* five *paaw-á* quarter-pl *ḍhínɡar* wood *bhit* plank *bhakúl-u* thick-msg *laɡaiǰíl-u* be.put.pfv-msg *heentá* condl *eesó* 3msg.rem.nom *bíiḍ-u* much-msg *šóo* good.msg *bh-áan-u* become-prs-msg 'If five quarters of wooden planks are put into the cupboard, that will look very beautiful.' (A:HOW051)

### **6.3.6 Article-like uses**

Closely related to adnominally-used demonstratives, and still possibly a subcategory of them, are the forms *so* and *se* (formally the weak forms of the remote set), used for signalling definite or previously introduced and firmly established entities in the discourse. Their distribution (Table 6.5) is the same as that of the adnominal demonstratives: the form *so* is used preposed to a masculine singular nominative head, whereas *se* is used with a head in any other case, gender or number.

Table 6.5: Definite "articles"


Contrasting in definiteness and identifiability with *so/se* used in this way, is the "indefinite article" *áa* or *áak* (B *a* or *ak*), derived from, and in one of its forms still homonymous with, the numeral *áak* 'one'. While *áa* is used for introducing a participant, *áa míiš* 'a (certain) man', *so/se* is used in order to refer to or track an already earlier introduced participant, *so míiš* 'the/that man'. The examples (39)–(42) are all taken from one and the same story, where the two main characters are a man and a (male) monster. At the very beginning of the story, in (39), the man is introduced, with a full noun and the indefinite *áa*.

6.3 Demonstratives

(39) *muṣṭú* before *zamanáii* time.gen *áa* idef *míiš* man *de* be.pst 'Once upon a time, there was a man.' (A:THA001)

In the sentences that immediately follow (not included here), the man is referred to by a third-person pronoun only (i.e., by the remote demonstrative, as described above), and we learn that the man goes hunting, shoots a deer and takes it to a hut, where he prepares the meat and starts eating. Then, in (40), the next participant shows up, the monster, which is also introduced by a full noun preceded by the indefinite *áa* and a descriptive attribute, 'hairy'.

(40) *tíi* 3sg.obl *maǰí* on *áa* idef *ǰhaṭíl-u* hairy-msg *ṭhaaṭáaku* monster *yhóol-u* come.pfv-msg 'Meanwhile a hairy monster came along.' (A:THA005)

In the next sentence, in (41), the *ṭhaaṭáaku* 'monster' is referred to by a full noun only, without any determiner or attribute (which perhaps has some sort of intermediate status between the indefinite and the definite), while *míiš* 'man', when he is referred to again, is done so through the use of a full noun marked with the definite *so*.

(41) *ṭhaaṭáaku* monster *yhaí* come.cv *šíiṭi* inside *ačíit-u* enter.pfv-msg *ta* sub *so* def.msg.nom *míiš* man *mhaás* meat *khóo* eat.3sg *de* pst

'When the monster came inside, the man was eating.' (A:THA006)

In the same way, when the monster is referred to in the following sentence, in (42), after the man is mentioned, he too is referred to by using a full noun and the definite *se* (that form because the head is in the oblique case).

(42) *théeba* then *se* def *ṭhaaṭáak-a* monster-obl *bi* also *tas* 3sg.acc *sanɡí* with *khainií* eat.vn *široó* start *thíil-u* do.pfv-msg 'And the monster started eating with him.' (A:THA007)

Having exemplified this use of the "definite article" primarily as a tracking device, it should be pointed out that there are no clear and unambiguous examples of it being used as identifiable in the larger situation or associative-anaphorically, two uses that are typical for definite articles, while not generally associated with demonstratives in general (Juvonen 2006: 485; Himmelmann 1996: 233). Therefore, since the use of adnominal *so/se* is not dramatically different

### 6 Pronouns

from the anaphoric use of adnominal demonstratives, an alternative labelling of it would be "anaphoric article" (Juvonen 2006: 486), but it also comes close to what Himmelmann (1996: 230–239) includes in the recognitional use of demonstratives. What sets it apart from the (strong) anaphoric use (see §6.3.4), however, and thus points to it being further grammaticalised towards a more typical definite article interpretation as "identifiable" in general (Juvonen 2006: 485), is its frequent and near-obligatory presence with already introduced and well-established referents in a discourse, its weak, thus phonetically reduced, form and its paradigmatic contrast with the indefinite article.

Along the same lines, the use of *áa/áak* exemplied above does not set it apart as radically estranged from its use as a numeral, but it could be said to have gone through an initial grammaticalisation stage (as a phonetically reduced presentative marker) towards a full-blown indefinite article (Juvonen 2006: 486).

The full system of reference and deixis, needless to say, is deserving of much more in-depth research, and the sections above should be taken as a preliminary analysis and suggestion as to what devices are being used in Palula.

### **6.3.7 Spatial specification**

In addition to the basic differentiation between remote, distal and proximal, there are a number of further spatially defined specifications that can be made. To the demonstratives (particularly in their adnominal use) of the distal set, certain spatial adverbs (§8.1.2) can be added (or fused with the demonstratives) to indicate finer degrees of distance as well as vertical and horizontal position in relation to the speaker, as shown in Table 6.6.


Table 6.6: Secondary spatial specifications of distal demonstratives

6.4 Possessive pronouns

### **6.4 Possessive pronouns**

What is sometimes referred to as a category of distinct possessive pronouns were already introduced above as genitive case forms of the personal and demonstrative pronouns, and are exemplified in (43) and (44).


### **6.5 Reflexive pronouns**

There is one frequently used pronoun, *teeṇíi,* that can be described as reflexive. It occurs in this form only, as is evident from the examples (45)–(48). Usually, but not exclusively, it is the possessor in a possessive construction, and its referent is then identical with the clause subject (whether explicit or implicit).


'The spirit said to the woman, "Don't rattle your bracelets!"' (A:WOM643)

(48) *ǰanǰ* wedding.party *ɡúum* go.pfv.msg *teeṇíi* refl *sanɡí* with *bíiḍ-a* much-mpl *ba* top *xálak-a* people-pl *ɡhin-í* take-cv *ɡúum* go.pfv.msg

'He went to a wedding party, taking a lot of people with him.' (A:GHU008)

### 6 Pronouns

However, when the reflexive occurs in the object position, as in (49), it seems necessary to use the construction *teeṇíi zaán* 'own self', where the reflexive pronoun remains in the possessive.

(49) *karáaṛ-a* leopard-obl *díi* from *handáa=the* this.way=do.cv *teeṇíi* refl *zaán* self *bač* salvation *thíl-i* do.pfv-f 'This way he saved himself from the leopard.' (B:CLE381)

### **6.6 Reciprocal pronouns**

The pronoun *akaadúi* (literally 'one' + a segment *aa* + 'other') is reciprocal, as illustrated in examples (50) and (51).


The finite verb agrees in the plural, and if the agent/direct object is a pronoun that makes a case differentiation between agent and direct object, it occurs in the accusative as in (52).

(52) *eetanaám* 3pl.rem.acc *akaadúi* recp *mheeríl-a* kill.pfv-mpl 'They killed each other.' (A:MAB004)

The reciprocal pronoun may, however, also occur in case-inflected forms, such as the genitive form in (53).

(53) *dhuimeém* both.obl *akaaduéem-e* recp.obl-gen *haalat-í* condition-pl *khooǰéel-i* ask.pfv-f 'The two inquired about each other's well-being.' (B:FOY002) 6.7 Indefinite-interrogative pronouns

### **6.7 Indefinite-interrogative pronouns**

The area of indefinite pronouns and their distribution needs further research, but there are strong indications that most, if not all, indefinite pronouns are also used as interrogative pronouns (as further exemplified in §15.2.2). As mentioned above (§6.3.1), the indefinite-interrogative pronouns (Table 6.7) belong together with the demonstratives in the larger system of pro-forms. An animacy distinction is also made here, compare (54) and (55), not otherwise part of the pronominal system.

Table 6.7: Indefinite-interrogative pronouns


(54) *ɡokhíi-a* Chitrali-pl *asaám* 1pl.acc *the* to *ḍanɡarík* Dangarik *than-áan-a* call-prs-mpl *koó* some *ba* top *asaám* 1pl.acc *the* to *kaaláaṣ-a* Kalasha-pl *than-áan-a* call-prs-mpl

'The Chitralis call us Dangarik, and some consider us Kalasha.' (A:ANJ015)

(55) *tíi* 3sg.obl *wíi-a* water-obl *wée* in *ɡubáa* some *šay* thing *dhríṣṭ-u* see.pfv-msg *hín-u* be.prs-msg 'She has seen something in the water.' (A:SHY053)

See also §15.3.2 for a possible development of a set of negative indefinite pronouns (*ɡá=bi=na* 'nothing', *koó=bi=na* 'nobody', *ɡóo=bi=na* 'nowhere'), based on the basic indefinite ones.

### **6.8 Relative pronouns**

There are no distinct relative pronouns in Palula. Instead, the demonstratives and the indefinite-interrogative pronouns are used in relative constructions, or in the functional equivalents of relative clauses (see §14.6).

## **7 Adjectives and quantifiers**

This chapter is primarily about descriptive adjectives and their properties, but some other classes with close affinities to descriptive adjectives, particularly those used in quantifying noun heads, are included and discussed briefly. This is also the rationale for numerals being presented in this chapter. Additionally, there is some focus on semantic aspects of the adjective category and its status vis-à-vis nouns and verbs.

### **7.1 The adjective and its properties**

A core group of descriptive adjectives in Palula forms a lexical category clearly differentiated semantically, morphologically and syntactically from both nouns and verbs. It seems, however, size-wise to be a rather limited class when compared with nouns and verbs and also less uniform as a whole. To what extent adjectives should be considered an open word class will be discussed below.

Among the adjectives there are those that display a close affinity with nouns and another group that shares characteristics with some verb forms. A considerable overlap between adjectives and nouns, on the one hand, and between adjectives and participles, on the other, was a feature also of OIA (Whitney 1960 [1889]: 322, 967). There is another group of words that can be used either as adjectives or as adverbs. A number of properties that tend to be coded as adjectives in languages with large inventories, such as English and Japanese (Pustet 2006: 60), are either verbs or nouns in Palula. Those groups of potential adjectives roughly correspond to the two "swing-categories" suggested by Givón (1979: 321), based on their respective time stability as compared to the most typical members of the class.

Before dealing in more depth with semantic (§7.2) and morphological (§7.3) properties of adjectives, something will be said about their distribution and phonological structure. Normally adjectives do the job of modifying nouns, as in example (1), or that of descriptive predication, as in (2).

### 7 Adjectives and quantifiers


Occasionally, some adjectives may also occur on their own in noun phrases and then distributionally function as nouns (see §7.3.1). On the other hand, another noun may also modify the head noun of a noun phrase, in genitive as well as in its basic form, and also be used predicatively (see Chapter 11).

Structurally, inflected adjectives have a lot in common with nouns and some of the most common finite verb forms (those historically based on participles). Compare, for instance, the words highlighted in examples (3) and (4) (the first an adjective, the second a noun and the third a verb), all with an ending segment *éeli* (which illustrates the partial alliterative agreement found in Palula, see Corbett 2006: 87–88). However, while gender is an inherent property of nouns, gender marking of adjectives and verbs is only in the form of agreement with the gender of a noun.


The most typical Palula adjectives belong to the inflecting group and comprise a monosyllabic stem and an agreement suffix (see §7.3.1). The stem typically consists of a long first-mora accented vowel, maximally preceded by a two-consonant onset and followed by a two-consonant cluster: *tríimb-* 'thick', minimally without an onset but followed by a stem consonant: *áaḍ-* 'half'. Almost equally common is a monosyllabic stem consisting of a short accented vowel, maximally preceded by a two-consonant onset and followed by a two-consonant cluster: *tríṣṭ-* 'bitter', minimally without an onset but followed by a stem consonant: *úč-* 'little'. There are also monosyllabic stems with a long second-mora accented vowel: *mhoór-* 'sweet', *lhuúṇ-* 'salty'.

### 7.2 Semantic properties of adjectives

Adjectives belonging to the inflecting group, but with a two-syllable stem, are also frequent in the language. All such stems are accented on the vowel of the last syllable, which is either a short vowel – as in *bhakúl-* 'fat' – or a first-mora accented long vowel – as in *šidáal-* 'cold'. The first unaccented vowel may be either long or short. I do not have any evidence of three-syllable adjective stems in the inflecting group.

Non-inflecting adjectives may be monosyllabic or polysyllabic. A number of the monosyllabic ones have a CVC-structure: *šut* 'sour', *zeṛ* 'yellow', *paák* 'clean', whereas the polysyllabic ones are structurally more diverse, though almost exclusively accented on the last syllable: *muškíl* 'hard, difficult', *naawás* 'dangerous', *arzaán* 'cheap', *askóon* (B *askáan*) 'easy'.

### **7.2 Semantic properties of adjectives**

Properties or qualities that are typically coded or lexicalised as adjectives in many other languages generally occur as such in Palula as well. Dixon concludes from a comparative study that adjectives with the semantic content of dimension, colour, age and value are the most likely to occur, however small the class of adjective is in a language (Dixon 1982: 46). Along similar – but not identical – lines, Givón (2001a: 82) points out size, colour and a number of qualities perceived by human senses are most typically expressed with adjectives.

The inflecting adjectives exemplified in the following sections are given in their masculine singular form, and the invariant adjectives are consistently indicated with the abbreviation inv.

### **7.2.1 Dimensional adjectives**

Many of the adjectives denoting spatial dimensions come as antonym pairs. It is not uncommon that the choice between near-synonyms is dictated by animacy in one way or another. As can be seen in the list (Table 7.1), many of the adjectives here are of the inflecting type.

The antonym pair *ɡáaḍu–lhoóku* has the additional connotation of age – further extended to relative importance when applied to human beings – which the pair *ɡháanu–léku* lacks insofar as it simply refers to physical size. While the former word pair often occurs with humans, the latter pair tends to be used more frequently with animals and inanimates, but also for qualifying small children (as far as *léku* is concerned).

### 7 Adjectives and quantifiers

Table 7.1: A selection of dimensional adjectives (relevant antonyms placed on the same row)


The adjectives *čóolu, bhakúlu, thúlu* and *čaáx* all have to do with thickness, but here also additional semantic properties as well as the animacy of the noun being qualified govern their selection. Whereas *čóolu* is used invariably to define the thickness of inanimate objects, *čaáx* is never used to refer to anything other than animates, and is used primarily when talking about fat, strong and well-fed domestic animals. The group of nouns being qualified by *thúlu* is similar to that of *čaáx*, but with less emphasis on the strength and more on the actual dimensions, whereas the "in-between" adjective *bhakúlu* is very widely applied to humans and animals, as well as inanimates.

The opposite (and in this culture, negatively rather than positively associated) terms available in this case (*čúṇu* and *tróku*) are fewer, and as expected, each has a wider scope than any of the aforementioned antonyms; *tróku* is used at the upper end of the animacy continuum and *čúṇu* at the lower end. The pair *dhríɡu–khaṭáanu* refers to length as well as to vertical extension.

### **7.2.2 Colour adjectives**

Adjectives denoting brightness (dark/light, black/white) come in antonym pairs, whereas the other colour terms cover particular sections of the rainbow (Table 7.2). These terms form, like in many other languages, a complement set with a very restricted size (Dixon 1982: 19, 46). Many, but far from all, of these adjectives are of the inflecting type.

Including the terms for 'black' and 'white', there are four basic colour-terms in Palula: *k(r)iṣíṇu, paṇáaru, níilu, lhóilu.* For reasons we will return to below, *zeṛ* and *aaɡabháanu* must be considered less basic and less typical than other Palula adjectives, and among them *zeṛ* at least must be considered a recent Pashto

### 7.2 Semantic properties of adjectives


Table 7.2: Colour adjectives

loan (*ziyaṛ* 'yellow; brass', Raverty 1982 [1901]). The term *níilu* covers the whole spectrum referred to with English 'blue' and 'green', but the particular nuance can be specified with nominal derivations, such as *aaɡabháanu*. The pair *chiṇ- ̣ práal* act as rather marginal adjectives; they tend to be used as nouns, 'darkness' and 'light', respectively.

The occurrence of the four basic terms, as well as the acquisition of new terms, confirm the hierarchy and distributional restrictions suggested by Berlin & Kay (1969: 2–5). That yellow (and perhaps light colours in general) used to be associated with *paṇáaru* 'white' is supported by historical-comparative data, where the cognate of this in a number of NIA languages, in some MIA languages as well as in the OIA use of *pā́ṇḍara-* (in Šatapatha-Brahmāṇa), covers senses such as 'whitish-yellow, yellow, pale, white' (Turner 1966: 8047). That *paṇáaru*, at least in the not too distant past, was associated with the colour of maize is hinted at in the present use of the derived verb *paṇará* 'rinsing or peeling the ears of corn, white-wash'. Similarly for *k(r)iṣíṇu* 'black', historical-comparative data suggests a past usage covering dark colours in general, or more specifically 'dark blue, black' as attested for OIA *kr̥ṣṇá-* (in Rgveda, according to Turner 1966: 3451).

Composite categories of blue/black and yellow/white, respectively, unattested in the previous findings of Berlin & Kay (1969), have been attested in the more recent findings of Kay, Berlin & Merrifield (1991: 17). Following their reasoning, a first stage in the evolution of basic colour terms would consist of two composites (Kay, Berlin & Merrifield 1991: 19), one white/yellow/red and one green/blue/black, essentially embodying a distinction between "light" colours and "dark" colours and, in turn, would correspond to a universal or near-universal distinction between day (light time) and night (dark time) (Wierzbicka 1996: 288). These two super-categories may perhaps roughly correspond to a presumed historical "wider" usage of the cognates of *paṇáaru* and *k(r)iṣíṇu*, respectively.

That would also assume that the other two basic terms have entered into the system some time later and only gradually gained their status as basic and – in relation to the first two – completely complementary terms. The cognate of *níilu*

### 7 Adjectives and quantifiers

Table 7.3: Hypothetical evolution of Palula colour terms


is known also in OIA, *nīla-* (Turner 1966: 7563), which refers to 'dark blue, dark green, black', thus competing with OIA *kr̥ṣṇá-* but also used as a noun 'blue substance; indigo'. This probably means that it started out as a noun specifically referring to a particular plant and then gradually gained its more general use to signify the colour of that plant before becoming associated with the blue/green segment of the blue/green/black composite. The etymology for *lhóilu* seems somewhat shakier, but Turner (1966: 11168, 11158) connects it with an OIA root *lōhá-*, variously glossed as 'red, copper-coloured; made of copper; copper; iron'. This, at least, hints at a background as a noun even for this term, suggesting that it has gradually gained status as a basic colour term, possibly competing with a former white/yellow/red composite.

Somewhat speculatively we may therefore make some guesses as to the evolution of colour terms in previous stages of Palula, as displayed in Table 7.3. In addition to those by now rather established terms, it may be added that another "newcomer", *ɡulaabí* (from Urdu for 'pink, rose-coloured'), seems to be making headway into the red part of the spectrum. That the blue/green composite is a persistent one is confirmed by some languages in which this composite remains undissolved even after brown and purple have become thoroughly introduced into the system (Kay, Berlin & Merrifield 1991: 18).

It is obvious from this discussion that it is the less basic and relatively newly acquired colour adjectives that are invariant, such as *zeṛ* 'yellow', whereas the most basic and firmly established terms, such as *kriṣíṇu* 'black' and *paṇáaru* 'white', are inflecting.

### **7.2.3 Age adjectives**

The age-type of adjectives is, like in many other languages, size-wise very restricted (Dixon 1982: 46). Those that occur (Table 7.4) come in antonym pairs,

### 7.2 Semantic properties of adjectives

similar to the dimensional adjectives. The antonym pair *búuḍu* –*zuwaán* is perhaps used more as a pair of independent nouns in the sense of 'old man' and 'young man', respectively, especially the latter, than as adjectives modifying another noun.

Table 7.4: Age adjectives (relevant antonyms placed on the same row)


The pair *ɡhaḍeeró–lhookeeró* reveals a trace of an earlier comparative formation, those forms being old comparatives of *ɡáaḍu* 'big, grown, important', and *lhoóku* 'small, young' (see above). In this case, however, it is solely the age component that remains, as in *ɡhaḍeeró bhroó* 'elder brother'. In particular, *ɡhaḍeeró* is used as a noun, and often in a more restricted sense, meaning an 'elder' of an entire village, clan or tribe.

### **7.2.4 Value adjectives**

While Dixon (1982: 46) places a semantic type "value" among those most likely to belong to the adjectives, even when the class is small, Givón (2001a: 83) includes what he calls "evaluative" in the less prototypical group. In Palula, one of the more obvious characteristics of this type (Table 7.5) is that to a large extent it is made up of relatively recent loans, and many of them are morphologically invariant.

Table 7.5: A selection of value adjectives (relevant antonyms placed on the same row)


### 7 Adjectives and quantifiers

### **7.2.5 Physical-property adjectives**

Lacking a better term, I chose to use the term physical-property adjectives suggested by Dixon (1982: 16) for properties that in different ways parallel human sensory capacities. What is essentially the same class of potential adjectives, Givón (2001a: 82) divides into four groups: auditory qualities, shape, taste and tactile. For reasons having to do with their lower temporal stability, Givón (2001a: 83) places some of Dixon's physical properties under transitory states (e.g., temperature and external condition), a group thatGivón considers less prototypically coded as adjectives.



As can be seen in Table 7.6, and also pointed out by Dixon (1982: 19), the properties of this type that code taste seem to be members of a complement set akin to that of colour terms (which in themselves very well could be treated as physical properties that are visually perceived), whereas some of the other clusters (like the tactile suggested by Givón 2001a: 82) occur in antonym pairs in a fashion similar to that of dimensional adjectives. While all the aforementioned types (dimensional, colour, age and value) are among those typically occurring as adjectives, according to Dixon (1982: 46), this type occurs with less frequency in languages with smaller inventories of adjectives.

Whereas tactile- and taste-related properties are rather well-represented, auditory qualities and more specialised shape (i.e., visually perceived) properties are generally not coded as adjectives in Palula. Some of those signifying less temporally stable properties, such as *šúku* 'dry' and *páaku* 'ripe, cooked' are less adjectival; both are in fact formally identical to perfective (including Perfective

7.2 Semantic properties of adjectives

Participle) forms of the verbs *šuš-* 'dry' and *pač-* 'ripen, become cooked', respectively. However, it is interesting to note that all of the exemplified adjectives of this type are inflecting.

For an in-depth study of temperature terms in Palula, including adjectives such as *táatu* and *šidáalu*, the interested reader may consult the article *Facts, feelings and temperature expressions in the Hindukush* (Liljegren & Haider 2015a).

### **7.2.6 Speed adjectives**

It is uncertain whether this type really qualifies as a main adjective type, or rather ought to be included among physical-property adjectives, but in order to compare with Dixon's results I have kept it separate. As predicted by Dixon (1982: 46), it is a very restricted type (Table 7.7) with respect to size.

Table 7.7: Speed adjectives


It is doubtful how much these words are really used as adjectives. Mostly they are used as adverbs, qualifying verbs rather than nouns. Used as an adverb, *teéz*, for instance, has at least two near-synonyms, *lap* (or *lab*) and *táru*, both meaning 'quickly, fast'. Dixon (1982: 47–48) notes a certain conditional relationship between the physical property and the speed type: if the physical property type in a given language is associated with the verb type (which we saw above that it to some extent is), then the speed type will be associated with the class of adverbs.

### **7.2.7 Human-propensity adjectives**

A type of adjective that tends to be particularly numerous is one that is applied to human behaviour and states. Although adjectives of this type primarily qualify nouns referring to humans, this type is often extended to include all higher animates (Dixon 1982: 16, 46). The distinction between this type and value adjectives (see above) is at best approximate. Many of the properties found here (Table 7.8) correspond to Givón's (2001: 83) transitory states and states of living, both of which less prototypically occur as adjectives. Some of them have more or less obvious antonyms, but for others, there is not one clear opposite term, or it would simply not make sense to talk about an opposite trait.

### 7 Adjectives and quantifiers

Table 7.8: A selection of human-propensity adjectives (relevant antonyms placed on the same row)


Whereas languages with a very small class of adjectives tend to associate physical property with verbs, human propensity instead tends to be associated with nouns. In fact, a number of the Palula adjectives of this type that have to do with (more permanent) states, can equally well function as nouns and often do; e.g., *ṣíiṛu* could have the sense 'blind' when modifying a noun and the sense 'blind person' as the head of a noun phrase. On the other hand, some other adjectives presented here that are used for (more temporary) mental-internal states are closely associated – or even identical – with verb forms, e.g., *khíndu*, which is the perfective (or Perfective Participle) of *khínǰ-* 'become tired'.

### **7.2.8 Summary of findings**

The investigation in §7.2.1–§7.2.7 gives us a somewhat clearer idea how to view the class we have referred to as "adjectives" in Palula.

First, there is a core of typical adjectives, used exclusively or nearly exclusively as descriptive or qualitative modifiers of nouns. Many of those are found among words referring to dimension and colour.

Second, there are a number of overlaps, and some words from the following categories can also be used as modifiers of nouns: a) adverbs (i.e., as modifiers or adjuncts of verb phrases or adjective phrases), particularly words referring to speed or quantity; b) nouns, particularly words referring to age and human propensity, but also some of those referring to dimension, quantity, age and colour; c) verbs (or more specifically participles), particularly words referring to physical property but also some of those referring to quantity and human propensity.

### 7.3 Morphological properties of adjectives

Third, while the firmly established adjectives (used with relative freedom attributively as well as predicatively) tend to be of the inflecting type, most newly acquired words used as noun modifiers are invariant and usually more restricted in their distribution (more readily used predicatively than attributively), especially newly acquired evaluative words. The relative age of the colour terms discussed in §7.2.2, also illustrates this difference: while the older and most basic terms are inflecting adjectives, the relatively newly acquired colour terms are less basic, distributionally more restricted and invariant.

The relative closeness between important groups of adjectives and nouns on the one hand and verbs on the other suggests that the diachronic rise of agreement features within the noun phrase (which are to some extent alliterative, see Corbett 2006: 87) is to be sought either in the nominal paradigm or in the agreement features of participles, or perhaps both, thus reinforcing one another and possibly strengthened further through analogy.

### **7.3 Morphological properties of adjectives**

### **7.3.1 Inflectional morphology**

As has been mentioned already, adjectives can be divided into two main categories as far as morphology is concerned: inflecting and invariant (or non-inflecting) adjectives. When they qualify nouns, as modifiers or as predicative complements, inflecting adjectives show agreement in gender, number and case. Although the inflectional morphology in many ways mirrors that of nouns, adjectival inflection shows a considerably lower degree of declensional variation and fewer available forms. Case agreement is, for instance, part of the inflectional properties of adjectives, but it plays a minor part in the paradigm as compared to the nominal paradigm, and number contrast is only partially displayed.

### **Inflecting adjectives**

The great majority of inflecting adjectives occur in three forms ending in *-u*,  *a* and *-i*, respectively, as displayed in Table 7.9. (There is also a marginal fpl form ending in *-im*, although mainly realised when an adjective is being substantivised, §7.3.2.)

The first two forms are a direct reflection of the nominative singular and the nominative plural/oblique singular forms of the *u*-marked nouns of the *a*-declension, and the feminine form corresponds to the singular form of the *m*-declension nouns. Inflecting adjectives with an accent on a long *óo* (<*áa*) or *áa* (<*a*) in

### 7 Adjectives and quantifiers

nom.msg undergo vowel modification (umlaut) to *ée* in the feminine agreement forms with an *i*-suffix (Table 7.10).

Individual adjectives inflect differently (Table 7.11), but they are too few to analyse meaningfully as parts of separate declensions; however, it should be pointed out that they do bear resemblances to some of the other noun declensions and may possibly reflect earlier more widespread declensional differences among adjectives, similar to the adjective classes in Kohistani Shina (Schmidt & Kohistani 2008: 100–103).

### **Invariant adjectives**

Adjectives that do not inflect for agreement with a noun head either end in a consonant or a vowel other than *u* or *i*. Some examples of invariant (non-inflecting) adjectives are shown in Table 7.12.

Although most of these without doubt function as adjectives, there are a number of features that make at least a fair number of them slightly less prototypical: A large proportion of the invariant adjectives are fairly recent loans from other languages, and although some of them are old enough in the language to have acquired an indigenised phonology (such as *xaróob* from the Perso-Arabic *xarāb*), they have not yet developed the inflectional paradigm typical of most inflecting adjectives. Furthermore, many of these tend to be used predicatively, whereas the inflecting adjectives are used equally well attributively and predicatively. They can also to a larger extent be used as nouns and adverbs apart from their adjectival usage.

### **7.3.2 Substantivisation**

An adjective can occur on its own as the head of a noun phrase when substantivised, i.e., an adjective like *ɡáaḍu* 'grown, big, important' thus being used as in (5) in the sense 'the big one, the adult'. Apart from a slight semantic shift, being substantivised also means that case forms otherwise only applied to nouns have to be used (when applicable). This is also the realm where the otherwise rare feminine plural forms with *-im*, as in (6), are non-optional.

(5) *so* 3msg.nom *ɡáaḍ-am* big-mpl.obl *díi* from *náqal* copying *th-áan-u* do-prs-msg 'He is imitating the adults [the grown up men].' (A:SMO005)



### Table 7.10: Inflection (involving umlaut) of adjectives


Table 7.11: Irregularly inflecting adjectives


*a* In the Biori dialect this adjective has been reinterpreted as an invariant adjective: *šuy*.


### Table 7.12: A selection of invariant adjectives

### 7 Adjectives and quantifiers

(6) *se* def *éeḍ-im* half-fpl *bhíilam* fearfully *khonḍíl-im* speak.pfv-fpl 'The rest [the other half] of them spoke fearfully.' (A:BEZ022)

### **7.3.3 Comparison of degree**

As we saw above, there are some traces of an earlier inflection for comparison, and Morgenstierne (1941: 17) also points out some remains of the OIA superlative degree, but in the modern language these degrees are exclusively expressed periphrastically. The comparative is expressed by a standard of comparison in oblique case and the postposition *díi* 'from', as illustrated in (7) and (8), preceding the adjective functioning as the parameter of comparison, literally translatable as 'X is large from Y'.


A construction having a function close to the superlative of many European languages similarly uses the oblique form of the indefinite pronoun *buṭheé* 'all' and *díi*, see example (9), literally meaning something like 'X is more powerful than all'.

(9) *buṭhimeém* all.obl *díi* from *taaqatwár* powerful *hín-u* be.prs-msg *insaán* human 'Man is the most powerful [creature].' (A:KIN003)

### **7.3.4 Derivational morphology**

Apart from the general flexibility of some words, as we saw above (§7.2.8), being used as nouns *and* adjectives or as nouns *and* participles, etc., adjectives may also be derived morphologically from other parts of speech, particularly from nouns and adverbs. A commonly occurring adjectival derivative suffix applied to nouns, particularly referring to materials or substances, is the accent-bearing derivational suffix *-íil*, as shown in Table 7.13, to which agreement suffixes are subsequently added.

7.4 Quantifiers


Table 7.13: Adjectives derived from nouns

The difference between the noun-to-adjective derivation in *-íil* and the Perfective Participles with an *íl*-segment should be noted, as they may take on slightly different semantics when used attributively: *ǰhaṭíilu* 'made of fur' vs. *ǰhaṭílu* 'furclad, hairy'. A special group of adjectives is morphologically derived from the relatively small class of non-derived (primarily calendrical) temporal adverbs (see §8.1.3) by means of an accent-bearing suffix *-úk* (Table 7.14), to which agreement suffixes are added.

Table 7.14: Adjectives derived from temporal adverbs


### **7.4 Quantifiers**

The Palula numeral system is basically vigesimal, a system common to most languages of the region (Bashir 2003: 823). The numerals discussed in this section are cardinal numerals, substantivised numerals, and ordinal numerals.

### 7 Adjectives and quantifiers

### **7.4.1 Cardinal numerals**

The cardinals from twenty to one hundred are vigesimal, which means that twenty (and not ten) functions as a base, preceded by a multiplier. The word *bhiš* 'twenty' is essentially a noun in this construction, plural-inflected and preceded by a modifying lower numeral, as shown in Table 7.15.

Table 7.15: Vigesimal counting


The numbers between 20, 40, and 60, etc., are formed by coordination, whereby a coordinating element (compare with §14.2.1) is attached to 'twenty' and then followed by one of the numerals 1–19, as exemplified in (10) and (11). Phonologically, the complex is one word with one main accent. Note that the order is the reverse in some other languages in the region (compare with Kohistani Gawri, described by Baart 1999a: 57).


Table 7.16 presents the Palula numerals 1–40, the system after that being regular up to one hundred.

Although even the numerals one hundred and above can be given according to the vigesimal system, this is no longer in common use. Instead, as can been seen in Table 7.17, the numerals 100, 200, etc., 1 000, 2 000, etc., and all higher numbers are represented by loans from languages of wider communication, essentially words from Pashto. With a growing emphasis on education in the Palula-speaking area, even small children are familiar with Urdu numerals and use them in free variation with the native Palula words, especially for all numbers

### 7.4 Quantifiers


Table 7.16: Cardinal numerals 1-40

above ten. For indicating years according to the common era, Urdu numerals are used exclusively.

Reduplication, as in (12), is used to form distributive numerals.

(12) *tus aak-áak looṛíi-a aṭ-óoi*

> 2pl.nom red-one bowl-pl bring-imp.pl

'Go and get a bowl each [all of you]!' (A:KAT125-6)

Emphasis or exclusiveness can be added to a numeral with a particle *eé* (compare with emphasised pronouns, §6.1), as in (13), giving the meaning 'only ten, etc.'

### 7 Adjectives and quantifiers

Table 7.17: Higher cardinal numerals


(13) *eesé* rem *waxt-íi* time-gen *dáaš* ten *eé* ampl *kušúni* household *de* be.pst 'In those days there were only ten households.' (A:PAS010)

For inclusiveness, on the other hand, i.e., to give the meaning 'all ten, etc.', a form of the numeral with an accented ending *eé* and the numeral stem deaccented (compare with the ordinal numerals, §7.4.3) is used, as can be seen in (14).

(14) *šíin-ii* bed-gen *čureé* four.incl *šeenbóo-a* leg.of.bed-pl *phooṭíl-a* break.pfv-mpl 'All four legs of the bed broke.' (A:GHU024)

### **7.4.2 Substantivised numerals**

Apart from being used attributively, cardinal numbers, at least the lower ones, can also be used independently as heads of noun phrases (like many adjectives and demonstratives). Used this way they are inflected for case, see (15)–(16).


7.4 Quantifiers

### **7.4.3 Ordinal numerals**

Ordinal numerals (Table 7.18) are formed with an accented and somewhat variable suffix *-úma/-íma/-íima*, added to a de-accented form of the cardinal numerals. The ordinal 'first' is altogether suppletive (although a form *akúme* has been noted in the B dialect). It should be noted that ordinal numerals functionally and distributionally group with other determiners (such as the adnominal demonstratives and some of the other articlelike elements mentioned in §6.3), but are presented here due to their form relationship with the cardinal numerals.



### **7.4.4 Adjectival quantifiers**

Another group of Palula quantifiers is made up of non-numerally quantifying or somewhat existential expressions. Some examples are provided in Table 7.19. Some of them are distributionally rather flexible. Both *púuntu* 'full' and *puunǰí* 'full' are essentially verbal, the first being the Perfective Participle of *púunǰ-* 'become filled' and the second the converb of the same verb.

The quantifier *áaḍu* may, for instance, occur as the head of an adjective phrase, as in (17), but it may equally well occur as a substantivised head of a noun phrase, as in example (6) above. The word *bíiḍu* is also distributionally very flexible. It can be used as an adjectival modifier of a noun, as in (18), as well as a degree adverb modifying an adjective, as in example (1) above, or as a degree adverb modifying another adverb.


Table 7.19: A selection of adjectival quantifiers


## **8 Adverbs and postpositions**

### **8.1 Adverbs**

Hardly surprising, adverbs and adverbial expressions comprise a heterogenous category in Palula. Only a few adverbs are entirely non-derived (whether synchronically or diachronically speaking), and it is a relatively small word class, consisting of a number of subclasses, many of which are made up of closed rather than open sets. Most spatial and temporal adverbs are pronominal or nominal, most manner adverbs are verbal and some members of the small subclass of degree adverbs are also used as modifiers of nouns. Many spatial adverbs are also used as postpositions. Still, as described in §7.3.4, some adjectives are derived from the small group of non-derived (primarily calendrical) adverbs.

### **8.1.1 Symmetrical adverb sets**

The systematic differentiation between the categories proximal, distal, remote and indefinite-interrogative among pronouns (and other pro-forms; see §6.3.1) is to some extent carried over to and partly overlapping with certain sets of adverbial pro-forms. Although the symmetry is not complete, the relevant forms are given in Table 8.1, to be easily identified when referred to in the sections to follow.

For the spatial adverbs (also referred to as "adverbial demonstratives" or "locational deictics", Diessel 2006: 431), all four categories are represented, and as with the pronouns, there are in most cases two forms available in each category: a neutral (or highly accessible) form, and an emphasised form (used for referring back to something less accessible in the discourse, or correlatively, see §14.6), the latter with an initial *ee-*. As not to clutter the table more than necessary, only the emphasised forms are included (e.g., emphasised *eetáa* corresponds to neutral *táa*), and only the relevant A dialect forms of these adverbs (there is a regular correspondence between A *ee-* and B *ha-*; A *ɡóo* – B *ɡúu*; ablative forms A *-áai* – and B *-áauu*.)

Used adverbially, the oblique forms (also functioning as 3sg ergative pronouns) are always used along with a positional or directional postposition (hence the hy-

### 8 Adverbs and postpositions

phen), see §8.2.2, the only exception being the indefinite-interrogative *kíi*, which can be used alone meaning 'where to'. The locative and ablative can also be followed by a directional postposition, but can also stand alone as spatial adverbs. As will be further discussed and exemplified in §8.2, the members of the spatial set may also take on a temporal interpretation along with certain postpositions.


Table 8.1: Symmetrical adverb sets

The degree adverbs are primarily modifiers of adjectives, whereas the manner adverbs modify entire clauses. The form *eendáa* with the converb form of *bhe-* 'become' is used with intransitive clauses, while that with *the-* 'do' is used with transitive clauses. There is no obvious symmetrical correspondence to the indefinite-interrogative temporal adverb *kareé* 'when', but the nearest functional equivalent is *eetheél* 'then, at that time'.

8.1 Adverbs

### **8.1.2 Spatial adverbs**

Most adverbial expressions with spatial semantics are either pronominal/deictic or nominal in nature or derived from those word classes, often playing a syntactic role similar to that of noun phrases and postpositional phrases.

### **The deictic set** *índa – eeṛáa – eetáa – ɡóo*

As mentioned above, most of these adverbial deictics are related to demonstratives, and some may be used pronominally as well as adverbially. Typically they describe location, such as *índa* 'here', *eeṛáa* 'there (further away but still visible), *eetáa* 'there' (not visible), *ɡóo* 'where', the latter primarily interrogatively. It may, beside its primary static reading, as in (1), also imply goal, as in (2).


The members of the series *andóoii – eeṛáai – eetáai – ɡóoii* have an ablative function, expressing a movement away from a location, as in example (3), or an origin in a certain location.

(3) *eetáai* from.there *wheel-í* take.down-cv *ɡhróom-a* village-obl *phedóol-u* bring.pfv-msg 'Getting it [the corpse] down from there, we brought it to the village.' (A:GHA075)

The members of the locative series as well as of the ablative may be further specified (for goal or source) or emphasised by a following postposition, and for some postpositions, such as *wée* in (4), the members of the oblique series have to be used instead of those of the locative or ablative series (for further examples, see §8.2.2).

(4) *ṣiṣ* head *ba* top *aníi=wee* 3sg.prox.obl=in *bi* also *buṭ-í* plait-cv *aníi=wee* 3sg.prox.obl=in *bi* also *buṭ-í* plait-cv 'As for his hair, it was braided into many braids.' (A:JAN069)

### 8 Adverbs and postpositions

### **Various spatial adverbs**

*huṇḍ* **'up'.** This essentially nominal adverb is used for a location situated in a more or less vertical upward direction (not necessarily visible) from the point of reference. Used alone it can refer to a location 'up (there)', as in (5), as well as a movement to such a location.

(5) *huṇḍ* above *ta* cntr *chiítr=ee ̣* field=cnj *bhun* below *ba* top *ɡhaawaáz* stream.bed *de* be.pst 'The field was up above and the sand and stream were down below.' (A:JAN082)

From *huṇḍ*, a number of other semantically related adverbs or adverbial phrases can be derived: *huṇḍ the/húṇṭe* 'up (to)', as in (6); *huṇḍíi/huṇḍíi thíi* 'from above'; *huṇḍɡiraá* (A), example (7); or *huṇṭeɡiráak/huṇṭráak* (B) 'upward, uphill', *huṇḍaṛáa* (< *huṇḍ + eeṛáa*) 'up there'.


eat-prs-msg

'They were looking up and saw the bear sitting in the top of the tree eating raspberries.' (A:KAT145)

*bhun* **'down'.** This essentially nominal adverb is used for a location situated downward or below the point of reference. Used alone it can refer to a location, 'down (there)' as in (5), as well as a movement to such a location, as in (8). The following adverbs or adverbial phrases are examples of derivations from *bhun*: *bhún the/bhuná* 'down (to)', *bhuníi thíi* 'from below', *bhunɡiraá* (A) or *bhunteɡiráak/bhuntráak* (B) 'downward', *bhunimaá* '(on the way) downhill', exemplified in (9), *bhunaṛáa* (and other deictic forms based on it) 'down there'.

(8) *so* dem.msg.nom *méeš* man *hatí* 3sg.rem.obl *maǰí-e* in-gen *uṭik-í* jump-cv *bhun* down *whaí* come.down.cv *ba* top *uḍheew-í* flee-cv *wháat-u* come.down.pfv-msg 'Meanwhile, the man jumped down [from the tree] and fled down [to the village].' (B:CLE377)

8.1 Adverbs

(9) *tuúš* some *bhunimaá* downhill *wh-íi* come.down-3sg *ta* sub *karáaṛu* leopard *ukh-áand-u* come.up-prs-msg 'Getting some ways downhill, [he meets] the leopard coming up.' (A:KAT095)

*aǰá* **(B** *aǰé***) 'up (there)'.** This essentially deictic adverb is used for a location situated upstream or in a slightly upward location from the point of reference (compare with *huṇḍ* which implies a place more or less straight above the point of reference). Used alone it can refer to a location, 'up/over (there), upstream', as well as a movement to such a location, as in (10). Although not quite as productive as *huṇḍ* and *bhun*, this adverb is similarly used as a building block for other adverbs and adverbial expressions, such as *aǰimaá* 'upward', *aǰaṛáa* 'up/over there', *aǰí/aǰí thíi* (B) 'from upstreams/up-country'.

(10) *yáab* canal *ɡhaš-í* take-cv *aǰá* up *ɡúum* go.pfv.msg *ta* sub 'I went up along the [irrigation] canal.' (A:HUA053)

*túuri* **'down (there)'.** This adverb, otherwise mostly used as a postposition (see §8.2.2), is only occasionally used without a noun or pronoun as an argument. It seems to be contrasting with the aforementioned adverb in describing a location or movement downstream from the point of reference, such as in (11). It may also be followed by the postposition *wée* (see §8.2.2).

(11) *dúe* other *xálaka* people *ba* top *túuri* down *dhrúuk-a* gorge-obl *ǰe* down.to *ɡéa* go.pfv.pl 'The other people went down into the valley.' (B:AVA205)

*rhalá* **'up (above, on top)'.** This adverbial expression, related to the adjective *raál* (primarily occurring as as a host element in conjunct verbs with meanings such as 'raise, lift up'), is used for describing the position on top of a structure or object, as in example (12). In some sentences its function may rather be analysed as that of a manner adverb or a postposition.

(12) *dúi* other *hiimaál* glacier *whaí* come.down.cv *ṭópa* down *traác̣* host *de* give.cv *asaám* 1pl.acc *rhalá* up *ɡaḍíl-a* take.out.pfv-mpl 'Another avalanche struck came and brought us out [from under the snow].' (A:ACR015)

### 8 Adverbs and postpositions

*awaaɡír* **'up high, at a high elevation'.** This adverbial expression (possibly also adjectival) seems to be exclusively used for referring to a high elevation in the mountains, as in (13).

(13) *paṇáar-u* white-msg *táapaṛ* hill *thaní* qot *awaaɡír* high.up *qaribán* about *dúu* two *thiáaḍ-a* and.half-mpl *míil-a* mile-pl *táapaṛ* hill *hín-u* be.prs-msg

'There is a hill called the White Hill about two and a half kilometers up.' (A:HUA117)

*phará* **'yonder, over there'.** This deictic adverb (probably based on a spatial noun \**phaár* or \**paár*, compare with *aḍaphaár* 'halfways' below) is used for a location some distance away but still visible (and in a straight line) from the point of reference, as in (14), often in contrast with *oóra* 'on this side, over here' (see below). Like some of the other spatial adverbs above, it is used as a building block for other adverbs and adverbial expressions: such as *pharimaá* (A)/*pehrimaá* (B) '(some ways) forward, onward', *pharaṛáa* 'over there, yonder', *pharíi thíi* (A)/*parí thíi* (B) 'from some distance away', *pharaɡiraá* 'at some distance'. The deictic expression *pharaṛáa* may be further modified by vowel lengthening: *phaaraṛáa*, implying a location even further away than *pharaṛáa*. (Compare with the postposition *pharé*, §8.2.2).

(14) *ɡhoortsí* Ghortsi *bi* also *so* def.msg.nom *áa* one *kučúru* dog *bi* also *uḍheew-í* flee.cv *oóra* over.here *šan-á* roof-obl *so* dem.msg.nom *šay* thing *bi* also *tanaám* 3pl.acc *sanɡí* with *yhaí* come.cv *šan-á* roof-obl *ma* 1sg.nom *ba* top *phará khilaí dharíit-u*

yonder alone remain.pfv-msg

'Both Ghortsi [the name of a dog] and the other dog fled from that thing and went back to the roof over here. The thing followed them onto the roof, but I stayed by myself, away from there.' (A:HUA031-3)

*oóra* **'on this side, over here'.** As already mentioned and exemplified in (14), this deictic adverb is used in contrast with *phará*, implying a location close to the point of reference as opposed to a location further away. I am not aware of any derivations from this adverb.

### 8.1 Adverbs

*patú* **'behind, in back'.** This basic form is, according to my material, only used as a postpostion (in A) or conjunction (in B), see §8.2.2; however, a couple of (partly spatial, partly temporal) adverbial expressions are clearly derived from it: *patuɡiraá* (exemplified in (15))*/patuɡiróo/paturaá* (A)/*patuɡiráak* (B) 'back', *padúši (< patú + dúši)* 'behind', as in (16), the latter almost exclusively used in a temporal sense in B (compare with §8.1.3).


*muṣṭú* **(A only) 'in front, ahead, forward'.** As with the aforementioned adverbial group, *muṣṭú* has in addition to its basic spatial meaning, as in (17), acquired some largely temporal aspects (see §8.1.3). A nearly synonymous *muṣṭuɡiraá* is also derived from it.

(17) *tuúš* some *muṣṭú* in.front *bíi* go.3sg *ta* sub *áak* idef *luumái* fox *ḍhoó* sight *dít-i* give.pfv-f *hín-i* be.prs-f 'Going some distance forward, he saw a fox.' (A:KAT011)

*nhiáaṛa* **(B** *niháaṛa***) 'near, nearby'.** This adverb is often used in a so-called compound postposition (§8.2.3) but may also be used alone, see (18), as a spatial adverb.

(18) *tasíi* 3sg.gen *ba* top *páanǰ* five *ṣo* six *bhraawú* brother.pl *nhiáaṛa* near *hóons-an* stay-3pl *de* pst 'Five or six of her brothers were living nearby.' (A:HUA119)

*dhúura* **'far away'.** Like the aforementioned adverb, it is often used in a compound postposition (§8.2.3), but may also be used independently (19).

### 8 Adverbs and postpositions

(19) *dhúura* far.away *áa* idef *ɣar-í* peak-obl *učát* high *ɣar-í* peak-obl *eeṛáa* there.rem *piír* pir *hín-u* be.prs-msg 'Far away, on the top of a high mountain, there is a pir [a holy person].' (A:PIR014)

*šíiṭi* **'inside'.** This adverb, also used as a simple postposition (§8.2.2), as well as in postpositional sequences (§8.2.4), may be derived diachronically from a noun formerly denoting 'house' or 'home'<sup>1</sup> and a postposition 'to' (compare with Kalkoti *šíi* 'house', *šíiti* 'inside, into the house'), or it may have been borrowed as a whole or partly from a variety of Kohistani (compare with Gawri *šiṭ* 'house, home', *šiṭši* 'inside', *ši* 'in, into', Baart 1997: 119; 1999: 76). It is mostly used dynamically, as in example (20), as a spatial adverb.

(20) *ma* 1sg.nom *šíiṭi* inside *be* go.cv *tes* 3sg.acc *sanɡí* with *madád* help *th-áam* do-1sg 'I will go inside and help him.' (B:FOY060)

*aḍaphaár* **'halfway, midway'.** Composed of *\*aḍa > áaḍa* 'half' and *\*phaar* (see above) 'yonder', it is used in this essentially nominal form for the movement up to a point somewhere right between the point of origin and an expected goal, but it may occur in an oblique form *aḍaphará* with no obvious difference in meaning, a form to which the postposition *tií* (§8.2.2) can be added, rendering the approximate meaning 'as far as halfway'. It can also occur in the genitive *aḍapharíi* 'from the middle', as seen in example (21).

(21) *aḍapharíi* from.middle *huṇḍ* up *the* to *ta* cntr *ǰaláṣ* hairy *bhun* down *the* to *ba* top *lhíst-u* bald-msg 'It was half hairless and half covered in hair. The top half was hairy and the bottom half was hairless.' (A:HUA075)

### **8.1.3 Temporal adverbs**

Like spatial adverbs, many adverbial expressions with temporal semantics are nominal in nature, but there are also a number of synchronically non-derived temporal adverbs.

<sup>1</sup> For 'house' as well as 'home', *ɡhoóṣṭ* is used in today's Palula.

8.1 Adverbs

### **General deictic adverbs**

*típa* **'now'.** This freqently used and synchronically non-derived adverb is used for referring to the present moment, as illustrated in (22), as well as to 'nowadays' in general. There is also a rather little used form *tipaán tií* 'until now, even to this day'. Interestingly 'from now, after this' is expressed with the proximal ablative member of the spatial set, *andóoii pahúrta*.

(22) *típa* now *ba* top *ma* 1sg.nom *tasíi* 3sg.gen *paalaweeṇíi* strongmanship.gen *qiseé* story.pl *tháan-u* do.prs-msg 'Now I am telling the stories about his strongmanship.' (A:PAS029)

*muṣṭú* **(A)/***muxáak* **(B) 'before, in the past, once'.** The A adverb has spatial functions (see §8.1.2) along with its temporal ones. Both the A and the B adverb may take additional modifiers, such as degree adverbs (*bíiḍu muṣṭú* 'long before, a long time ago') or calendrical expressions (*dúu yúuna muxáak* 'two months ago'). The meaning 'since then, for a long time' can be expressed by *muṣṭúi niiɡiraá*. A sentence with the B adverb *muxáak* is given in (23).

(23) *muxáak* before *be* 1pl.nom *iskuul-í* school-obl *the* to *b-íia* go-1pl *de* pst 'Once, we were going to school…' (B:ANG001)

*eetheél/(eesé) waxtíi* **'at that time, in those days, then'.** Usually, for reference to a specific time (usually in the past), as in (24), the genitive of the noun *waxt* (or *waqt*) 'time' is used with a preceding demonstrative (if referred to contextually) or another identifying modifier (*kufurdóore waxtíe* 'in the pagan era' B), but also the synchronically non-derived adverb *eetheél*, possibly more widely used in the past, is still in use, as can be seen in example (25). The latter can refer to a point of time in the past as well as in the future (e.g., in *eetheél tií* 'before that time, up to that time'). The genitive of the English loan *ṭeém* 'time' is used in a way quite similar to *waxtíi*, taking various modifiers, such as *basandíi ṭeemíi* 'at spring time'.

(24) *neečíir* hunting *ba* top *eesé* rem *waxt-íi* time-gen *bíiḍ-i* much-f 'Hunting was a common custom at that time.' (A:HUA046)

### 8 Adverbs and postpositions

(25) *eetheél* that.time *maidaaní* of.field *ǰanɡ* war *de* be.pst 'In those days there used to be fighting in the fields.' (A:PIR005)

*aakatí waxtí (maǰí)/waxt bhe/padúši* **(B) 'later, after some time'.** A number of expressions, some of them containing forms of the noun *waxt*, are used when referring to a later point in time, in (26) with the converb of *bhe-* 'become' (compare manner adverbs formed with converbs, §8.1.4). Only in the B variety is *padúši* used in a clearly temporal sense (see §8.1.2).

(26) *khéli* quite.some *waxt* time *bhe* become.cv *dach-íi ̣* look-3sg *ta* sub 'Some time later he was looking.' (A:SMO016)

*waxtíi thíi* **'early'.** Again, the noun *waxt* is used, now in the genitive and with the directional postposition *thíi*. Also the adverb *raɣáṣṭi* has been noted in B.

*kareé* **'when, what time'.** This is a member of the series of indefinite-interrogative pronouns and other pro-forms (see §14.4.1 and §15.2.2).

Adjectival derivations with *-úk-* are also used in expressing similar temporaldeictic propositions, e.g., *tipúku* 'of now, "nowish"', *muṣṭúku* 'of the past, of old, "oldish"', as seen in (27), *eetheelúku* 'of that time' (see Table 8.4 on p. 178).

(27) *muṣṭúk-a* of.past-mpl *xálak-a* people-pl *dhii-á* daughter-obl *díi* from *na* neg *khooǰ-óon* ask-3pl *de* pst 'People in the old days were not asking their daughter [who she wanted to marry].' (A:MAR018)

### **Calendrical adverbs**

The most common calendrical adverbs (Table 8.2) used deictically are synchronically non-derived. In (28), the use of *dhoóṛ* 'yesterday' is illustrated.

(28) *dhoóṛ* yesterday *índa* here *kir* snow *dít-u* give.pfv-msg *de* pst *típa* now *bi* also *kir* snow *hín-u* be.prs-msg 'It was snowing here yesterday, and even now there is snow.' (A:CHE070320)

8.1 Adverbs


Table 8.2: A selection of basic calendrical adverbs

These adverbs also behave morphologically more like spatial deictic adverbs than nouns, when expressing notions such as 'before, until' or 'from, since'. While the basic form is used with postpositions otherwise taking oblique arguments, an inflection *-ii* or an idiosyncratic *-oo* (compare with B *hatáwuu* 'from there' and related forms) with an ablative function is used with *pahúrta* and *niiɡiraá*. Some multi-word calendrical expressions are displayed in Table 8.3.

Table 8.3: Calendrical expressions involving postpositions


Although *heeṇṣúka* 'this year' diachronically may be an adjectivally derived form (with *-úk-*), it is used like the other non-derived deictic adverbs, as in (29), without modifying a noun head.

(29) *heeṇṣúka* this.year *ma* 1sg.nom *čiiríit-u* be.delayed.pfv-msg 'This year I'm delayed.' (A:SHY026)

Belonging semantically to this group is also *rhootašíia* (with the alternative forms *rhoošíia* and *rhoošée*) 'tomorrow', used as in (30), which, however, is the oblique case of the noun *rhootašíi* 'morning'.

### 8 Adverbs and postpositions


Table 8.4: Calendrical cyclic expressions

(30) *ma* 1sg.nom *nis* 3sg.acc *aáǰ* today *kh-úum* eat-1sg *ta* sub *rhootašíi-a* morning-obl *ba* top *kanáa* like.what *bh-úum* become-1sg 'If I eat this today, what will then become of me tomorrow.' (A:HUB005)

As with the general deictic adverbs, adjectival forms in *-úk-* can be productively derived from most, if not all, of these calendrical adverbs: *aaǰúku* 'of today', today's', *bhiaalúku* 'of/from last night', *parúku* 'of/from last year, last year's', etc.

The identity of most calendrical cyclic expressions (Table 8.4), on the other hand, is clearly nominal, although the oblique case forms can be said to have been lexicalised as adverbs. In (31), the use of *heewandá, róota* and *deesá* can be seen.

(31) *heewand-á* winter-obl *tas* 3sg.acc *the* to *róot-a* night-obl *chóoṇ ̣* oak.twigs *lama-áan-a* hang-prs-mpl *dees-á* day-obl *har-í* take.away-cv *wíi* water *pila-áan-a* give.to.drink-prs-mpl *tas* 3sg.acc *šišáwi=the* beautiful=do.cv *šuí* good.f *zhay-íim* place-pl.obl *ɡhin-í* take-cv *ɡir-áan-a* turn-prs-mpl 'In the winter we hang oak-branches for her [the goat] during the night, and in the day we take her to drink plenty of water in beautiful places.'

(A:KEE090-1)

### 8.1 Adverbs

While the above oblique forms of the calendrical nouns express a "temporal location", a couple of these and a number of other temporal nouns (such as *kaál* 'year', *yúun* 'month', *haftá* 'week') may be used in various forms and derivations to express temporal quantities or frequences. The examples (32)–(41) all constitute adverbial phrases in Palula.


8 Adverbs and postpositions

(41) *dúi* other *kaal-á* year-obl *the* to 'another year, next year'

Most time-of-day expressions are nouns, quite a few of them loans from Pashto. They are like the calendrical cyclic ones above in that they occur in their oblique case form when functioning as time-of-day adverbs. Two exceptions are *rhoošnaám* 'morning' (in (42)) and *dhrumanaám* 'mid-afternoon', which are synchronically non-derived adverbs.

(42) *dhrumanaám* afternoon *ba* top *pašambeé* Pashambi *ta* cntr *chóoṇ-ii ̣* oak-gen *bháaru* load *ɡhin-í* take-cv *ma* 1sg.nom *ba* top *ɡúči*

free

'In the afternoon Pashambi carried fodder [to the goats] while I was free.' (A:PAS051)

### **Other temporal adverbs**

A few other adverbs with a temporal meaning, all poly-morphemic, are shown in Table 8.5.

Table 8.5: Other temporal adverbs


### **8.1.4 Manner adverbs**

Manner is mainly expressed by nonfinite verb forms, primarily the converb and the copredicative participle. The former of those is the basis for what can be considered a derivation of manner adverbs (or its nearest equivalent) from other parts of speech, particularly from adjectives.

### **Non-derived manner adverbs**

There is a small class of non-derived manner adverbs. Some of those are shown in Table 8.6.

8.1 Adverbs


(43) *ǰanɡibaazxáan-a* Jangibaz.Khan-obl *bhraáš* slowly *šukhaáu* coat *ɡaḍ-í* take.off-cv *čhúuṇ-u* put.pfv-msg 'Slowly, Jangibaz Khan took off his coat.' (A:JAN066)

### **The deictic set** *eendáa=bhe – kanáa=bhe – eendáa=the – kanáa-the*

This series exemplifies the derivation of manner adverbs. Here a deictic manner adverb (or adverbial phrase) is formed by adding the converb form of a verbaliser (*the-* 'do' or *bhe-* 'become') to the deictic adjective<sup>2</sup> *eendáa* (B *handáa* or the alternative non-emphatic form *andáa*). For modifying an intransitive predicate, as in (44), *bhe* is used, for modifying a transitive predicate, as in (45), *the* is used, and for questioning the manner by which something is done, the indefinite-interrogative *kanáa* and the relevant converb is used (see §15.2.2).


The clue to the interpretation of the deictic manner adverb is usually found in the immediate context of the utterance, either explicitly in the wider discourse, sometimes rendering 'thus' a good translation equivalent, or by extralinguistic means, such as gestures by the speaker.

<sup>2</sup> This can also be analysed as a pro-form of the host element in a conjunct verb (which does not necessarily need to be an adjective) or a predicate phrase (whether nominal or adjectival).

### 8 Adverbs and postpositions

### **Derivation of manner adverbs**

The derivational process described above is quite productively applied to words from various parts of speech (primarily adjectives and pronouns) in forming manner adverbial expressions, as illustrated in Table 8.7. To what extent these derivations are adverbs, adverbial phrases or adverbial (mini-)clauses is still open to further analysis, but considering their embedded status and the fact that they form a phonological word, we tentatively define them as derived adverbs. Neither is the relationship between Converb clauses with conjunct verbs and these manner adverbials entirely clearcut. An exact parallel to this formation is described by Schmidt & Kohistani (2008: 219) for Kohistani Shina.


Table 8.7: Examples of manner adverbial derivation

Also ideophonic expressions with no specific or standardised lexical source, such as those in (46), can be made manner adverbs this way. (The adverbs in the English translation in (46) are only approximate equivalents of the Converb cum reduplicated ideophone complexes.<sup>3</sup> )

(46) *se* def *insaan-á* man-obl *ráaǰ* rope *mučaá* open.cv *o* and *muṭ-á* tree-obl *sanɡí* with *so* dem.msg.nom *amzarái* lion *seéb* Sir *ram-raám=the* red-"firm"[idph]=do.cv *kaš-kaáš=the* red-"tight"[idph]=do.cv *ɡhaṇḍ-í* tie-cv *ɡaíl-u* throw.pfv-msg

'The man opened a rope and tied Lord Lion firmly and tightly to the tree.' (A:KIN023)

<sup>3</sup> A widespread covariation of so-called periphrastic 'do'-constructions and reduplicated stems has been noted by Jäger (2006).

8.1 Adverbs

### **8.1.5 Degree adverbs**

The function of degree adverbs is mostly to modify adjectives, but some of them also function as modifiers of other adverbs as well as of nouns.

### **Non-derived degree adverbs**

Forming a small class, non-derived degree adverbs are mainly used as modifiers of other adverbs but also as quantifiers of nouns, especially non-count nouns. The list in Table 8.8 is not meant to be exhaustive. See (9) for an example with *tuúš*.

Table 8.8: Non-derived degree adverbs


### **The deictic pair** *eetí* **(B** *hatí***) –** *katí*

These deictic degree adverbs intensify adjectives, meaning 'such, so', as in (47)*,* and 'what, how', as in (48). (Note that *katí* in particular also functions as a direct quantifying modifier of a noun head, meaning 'how much, how many'.)


### **Reduplication**

Reduplication is one strategy applied for degree modification of manner adverbs. The process can be full reduplication, as seen in (49), or reduplication of the first syllable. In example (50) both full and first-syllable reduplication are being used.

(49) *wháat-a* come.down.pfv-mpl *andáa=bhe* like.that=become.cv *bhraáš* red *bhraáš* slowly 'We came down like that, very slowly.' (A:GHA056)

### 8 Adverbs and postpositions

(50) *mhaamaǰaán* Mahmad.Jan *ba* top *la-láp* red-fast *la-láp* red-fast *khóo* eat.3sg *de* pst 'Mahmad Jan was eating very fast.' (A:MAH044)

### **Co-lexicalised intensifiers**

There is a number of more or less standard compounds with an adjective/adverb and a matching intensifying element, not much different from the effect other degree adverbs have on the modified constituent. Such an intensifier is either uniquely occurring with a particular adjective/adverb, or occurs only with a limited set of adjectives/adverbs. It seems those elements are mostly made up of a single closed syllable, as can be seen in Table 8.9.

Table 8.9: Examples of co-lexicalised intensifiers


Strikingly similar compounds have been observed in several other languages in the region, some of them even involving similar or identical forms as those found in Palula: e.g., in Dameli (Perder 2013: 163) and Khowar (Elena Bashir, pc, and own field notes).

### **8.1.6 Sentence adverbs**

Sentence adverbs (Table 8.10) modify entire utterances. They may for instance specify the speaker's attitude toward the event referred to, or emphasise its truthvalue, as in (51).

Table 8.10: A selection of sentence adverbs


8.2 Postpositions

(51) *so* 3msg.nom *ba* top *rištaá* really *xučháii* Khush.Shah.gen *so* def.msg.nom *máamu* uncle 'He was indeed Khush Shah's uncle.' (A:JAN056)

### **8.2 Postpositions**

All regularly used adpositions in Palula are postpositions. Most of them have spatial-temporal functions, but a few express some grammatical as well as spatialtemporal functions. Whenever a full noun is followed by a postposition it occurs in the oblique case (for those nouns that display a distinct oblique case), except for a couple of postpositions that take a full noun genitive argument. When a pronoun is followed by a postposition, there is a slightly higher degree of case differentiation (this will be commented on under each individual postposition presented) as compared to nouns followed by postpositions.

Apart from simple single word postpositions (§8.2.2), there are two types of complex postpositions, compound postpositions (§8.2.3) and postpositional sequences (§8.2.4). Some of the postpositions also function as adverbs (see above) or as heads of adverbial phrases.

### **8.2.1 Postpositions vis-à-vis case inflection**

Case inflection has been treated elsewhere (see §5.5), but since the differentiation between this category and adpositions (both following nouns) is not always obvious, particularly in IA languages, this should be commented on briefly.

One line of argumentation is phonological, where a string of segments with its own pitch accent would be considered a separate word, thus a postposition and not an inflection. This would definitely place the longer strings, i.e., those with two or more syllables, in the postposition category. The shorter, monosyllabic ones, however, are phonologically weak, and tend to cliticise to the preceding noun word or pronoun. Another line of argument is morphological, by which case-inflections occur close to the stem, whereas postpositions combine with already case-inflected forms of the noun (and mostly with non-nominative forms of pronouns), and thus are more peripheral. Using this argument, the genitive morpheme ends up as somewhat ambiguous in this respect (see §5.5.3).

A third line of argumentation may be more helpful, a mainly syntactic one, with the coordinating suffix or clitic *ee* as a diagnostic tool (see Baart 1999a: 77, for his analysis on case marking in Gawri). While inflectional suffixes obligatorily occur on both of two nouns coordinated with *ee*, a postposition is attached

### 8 Adverbs and postpositions

once, to the coordination as a whole, and does not occur inside the coordination. According to the latter diagnostics, the genitive along with the other case inflections (as described in §5.5) are clearly distinguished from the postpositions (as described below).

Many of the postpositional functions described below, especially those of the more central simple postpositions, are expressed by what are clearly inflectional case-suffixes rather than free postpositions in closely related Kohistani Shina (Schmidt & Kohistani 2001: 115–130).

### **8.2.2 Simple postpositions**

The order of the following presentation, without any claim of total comprehensiveness, reflects to a great extent the relative centrality and frequency of simple postpositions in Palula, proceding from the more central or frequent to the more specialised and less frequent ones. The postpositions *the* and *díi* are very common and express a number of grammatical as well as spatial-temporal functions.

*the* **'to, for'.** This postposition takes an oblique nominal argument, 'house' in (53), or an accusative pronominal argument, 'them' and 'us' in (52). There is a wide range of meanings, all having the core semantics of marking the recipient of a transaction or the goal of a movement.


Apart from marking the recipient in a typical "ditransitive" clause, *the* can also identify a beneficiary in a transitive clause and the recipient of an abstract entity, such as in the utterance in (54).

(54) *dhii-á* daughter-obl *mac̣hook-á* Machoke-obl *the* to *maníit-u…* say.pfv-msg 'The daughter told Machoke…' (A:MAA016)

### 8.2 Postpositions

If *the* marks a goal expressed as a spatial pro-adverb (also functioning as an inanimate/abstract demonstrative pronoun), the locative member of the set, *táa* 'there', as in (55), is used. However, if the goal of a movement carried out by a person is another person, the postpositions *khúna* or *kéeči* (see below) have to be used instead.

(55) *táa* there *the* to *misrí* mason *bulaḍ-eeṇḍeéu* call.for-oblg

'A mason is called to there [to see to that].' (A:KAT009)

Verbal Nouns in complements of permissive predicates as well as Verbal Nouns in purpose clauses are also taken as arguments by *the* (see §14.5.3 and §14.4.2). Some arguments with *the* are part of the valence pattern of some predicates, particularly non-nominative experiencers (§13.2.6) and objects of some conjunct verbs (§13.2.8). It is also used in some temporal expressions (see §8.1.3) as well as in specifying the direction of spatial adverbs (see §8.1.2). Also, see (§8.2.4), for the use of *the* in postpositional sequences.

*díi* **'from, (out) of, than'.** This postposition takes an oblique nominal argument, 'the Damelis' in (56), or an accusative pronominal argument. There is a wide range of meanings, all referring to the source of a transaction or the point of origin/reference.

(56) *ṛaním* 3pl.dist.erg *eeṛé* dist *riwaayát* tradition */…/ ɡiḍúuč-am* Dameli-obl.pl *díi* from *ɡhíin-i* take.pfv-f *hín-i* be.prs-f 'They have received [lit: taken] this tradition from the Damelis.' (A:MIT002-5)

A place as the source or starting point for a movement is almost always expressed with the genitive (along with the postposition *thíi*). In contrast, *díi* is primarily postposed to nouns denoting human sources, such as 'my grandmother' and 'my father' in (57). The transferred, or brought forth, entity, however, can be abstract, such as an utterance, as well as concrete, for instance in expressing the animate source of reproduction, as in (58).

(57) *míi* 1sg.gen *déedi* grandmother *díi* from *míi* 1sg.gen *ṣúunt-u* hear.pfv-msg *míi* 1sg.gen *báaba* father.obl *díi* from *míi* 1sg.gen *ṣúunt-u* hear.pfv-msg

'I heard it from my grandmother, and I heard it from my father.' (A:PAS005)

### 8 Adverbs and postpositions

(58) *khar-á* donkey-obl *díi* from *khar* donkey *ǰa-yáan-u* be.born-prs-msg 'A donkey is born by a donkey.' (B:PRB006)

Some arguments with *díi* are part of the valence pattern of some complex predicates (§13.2.8), and Verbal Nouns in complements of negative implicative predicates are also taken as arguments by *díi* (§14.5.2). The possessor in one main (primarily alienable, as the 'mason's hammer' in (59)) possessional construction is also marked with *díi*.

(59) *misrí* mason *yhóol-u* come.pfv-msg *seentá* when *misrí* mason *díi* from *tsaṭák* hammer *hóons-a* stay-3sg 'When the mason comes he will have a hammer [lit: When the mason has

come, from the mason a hammer will be present].' (A:HOW010)

Inalienable possession, on the other hand, is often expressed with the genitive case (a distribution parallel to that in Kohistani Shina, where the genitive case similarly is used for inalienable possession while the addessive case *-di* or *-idi* is used for alienable possession, see Schmidt & Kohistani 2008: 65, 69–70). The postposition *díi* is also used to express the comparative degree of the standard of comparison (§7.3.3).

*sanɡí* **'(along) with, at'.** This postposition takes an oblique nominal argument or an accusative pronominal argument. It typically expresses accompaniment, as in (60).

(60) *ḍaaku-aan-óom* robber-pl-obl *yhaí* come.cv *bakáara-m* flock-obl *sanɡí* with *tas* 3sg.acc *ɡhaš-í* take-cv *híṛ-u* take.away.pfv-msg *de* pst 'The robbers came and abducted him along with his flock.' (A:GHA005)

Used with an inanimate noun, such as the 'tree' in (61), it can have a further connotation of being attached to.

(61) *ma* 1sg.nom *tu* 2sg.nom *ráaǰ-a* rope *de* give.cv *ɡhaṇḍ-í* tie-cv *ɡal-áan-u* throw-prs-msg *aní* prox *muṭ-á* tree-obl *sanɡí* with 'I will tie you with ropes to this tree.' (A:KIN021)

### 8.2 Postpositions

Some arguments with *sanɡí* are part of the valence pattern of some intransitive verbs with a postpositional object (§13.2.4) as well as that of some complex predicates (§13.2.8), typically those coding events or actions involving two participants on some sort of equal basis. The postposition *sanɡí* may also take as an argument a Verbal Noun in a simultaneity clause (§14.4.1).

*ǰhulí* **'on (top of), on to, over, about, due to'.** This postposition takes an oblique nominal argument or (in most cases) an accusative pronominal argument in its essentially spatial sense 'on, onto' and an oblique pronominal argument in more abstract senses such as 'about, concerning'. It typically expresses a position on the immediate surface of something, such as the 'stone' in (62), or somebody. It is also used for the movement onto the surface or into the position on top of something or somebody.

(62) *se* def *baṭ-á* stone-obl *ǰhulí* on *se* def *kuṇaak-íi* child-gen *paaṇṭí* clothes *bi* also *heensíl-i* stay.pfv-f *de* pst 'On the stone were also the child's clothes.' (A:BER012)

A pronoun referring to an inanimate but still concrete entity can also (optionally it seems) occur in the oblique form with this postposition, the case otherwise used in pronominal reference to abstract entities (as in example (64)). Note that the oblique pronoun (which is the same as the 3sg agent in ergative alignment) here in fact is referring to the plural entity *muṣṭookhurá* 'forelegs'.

(63)

(64) *muṣṭookhur-á* foreleg-pl *dhraǰaá* stretch.out.cv *ba* top *tíi* 3sg.obl *ǰhulí* on *ṣiṣ* head *čhoor-í* put-cv *ba* top *bhéṭ-u* sit.down.pfv-msg 'Stretching out its forelegs, it put its head on them/there and sat down.' (A:PAS061)

Used with abstract nouns it can encode a whole range of meanings, some of them probably shading out into idiomatic expressions. The more common ones denote the topic of an utterance 'on, about, concerning', a reason for something to happen 'due to, with that', as in (65), or the means or attitude by which something is carried out.

### 8 Adverbs and postpositions

(65) *lhooméi* fox *teeṇíi* refl *mákar-a* cunning-obl *ǰhulí* on *askúun-a* ease-obl *baándi* on *kuhée* well.obl *díi* from *nikhéet-i* appear.pfv-f 'The fox easily got out of the well due to his own cunning.' (B:FOX033)

The reason reading is also the usual when *ǰhulí* takes a Verbal Noun as its argument (§14.4.3). In this abstract usage, another postpostion, *baándi*, also in (65), a loan from Pashto, is alternatively used, particularly in A. Some arguments with *ǰhulí* are part of the valence pattern of some predicates, particularly nonnominative experiencers and objects of some conjunct verbs (§13.2.6 and §13.2.8 respectively).

*maǰí* **'among, in, inside, during'.** This postposition takes an oblique nominal argument in its spatial (and temporal) sense 'in, into, inside', an animate oblique plural nominal or accusative plural argument in the sense 'among, out of', whereas the oblique form is used when its temporal sense is expressed with a pronoun. One of the basic uses of *maǰí* is to single somebody out as part of a group, as in example (66), 'of them' or 'among them'. The argument taken is always a plural entity or a collective expression.

(66) *tanaám* 3pl.acc *maǰí* among *áak* one *míiš* man *muṭ-á* tree-obl *ǰe* up *ukh-áai* ascend-inf *bhóo* be.able.3sg *de* pst 'Of them only one man was able to climb the tree.' (A:UNF007)

The other basic use of this postposition is to express a position inside of something, the 'bazaar' in (67), often in a certain geographical location.

(67) *a* idef *ɣaríb* poor *méeš* man *baazúur-a* bazaar-obl *maǰí* in *teeṇíi* refl *kuṇaák* child *bhanǰ-úu* beat-3sg *de* pst 'A poor man was beating his own child in [the middle of] the bazaar.' (B:ANG002)

It is also used for the movement of something 'into' another something, as in (68).

(68) *ak* idef *muṭ-á* tree-obl *wée* in *trúu* three *šúunɡ-am* branch-pl.obl *maǰí* in.between *so* def.msg.nom *mhaás* meat *čhúuṇ-u* put.down.pfv-msg 'He placed it between three branches in a tree.' (B:SHB735)

8.2 Postpositions

As mentioned above, *maǰí* can also be used in a temporal sense, 'while, at, during', with an oblique pronoun such as in (69), a nominal time expression or a Verbal Noun (§14.4.1).

(69) *tíi* 3sg.obl *maǰí* at *áa* idef *ǰhaṭíl-u* hairy-msg *ṭhaaṭáaku* monster *yhóol-u* come.pfv-msg 'Meanwhile a hairy monster came in.' (A:THA005)

*túuri* **'under, beneath, below'.** This postposition takes an oblique nominal argument, e.g., the 'deodar tree' in (70), or an accusative (alternatively oblique) pronominal argument. It describes a position which is in a purely spatial sense the opposite to that of *ǰhulí*, but it is not particularly frequent in my data. It typically expresses the position in or the movement into a position lower than or beneath something. It can also be used with abstract nouns such as 'agreement' in (71).


'They stayed under that agreement and remained here for some time.' (A:MAB014)

Either an oblique or an accusative pronoun can be taken as an argument of this postposition referring to an inanimate entity, the former exemplified in (72), and the latter in (73). Whether this reflects a dialectal difference or is just an example of free variation, is a matter for further study.


### 8 Adverbs and postpositions

*čhoor-í* put-cv *ba* top *khaṭúur-a* log-pl *d-áan-a* give-prs-mpl *tas* 3sg.acc *túuri* under 'The remaining water is then heated by building a fire under it with large pieces of wood.' (A:KEE048)

*wée* **'in, on, into'.** This spatial postposition takes an oblique nominal argument or an oblique pronominal argument. It is exclusively used as a postposition of nouns or pronouns with inanimate referents. In most cases it signals contact with the surface of something, as the fruit put into the bag in (74), or the penetration into a location or beneath the surface of an object, such as the hunter sitting in the tree in (75), i.e., inside the structure of tree branches. This latter characterisation is basically what differentiates it from the locative use of the oblique case of a noun or a locative pro-adverb as well as the use of the goal-specific postposition *the* with a noun. It seems, however, that many locative nominal expressions can occur almost interchangeably as oblique nouns and as nouns followed by *wée*. Perhaps the postposition serves a function of further emphasising the locative reading, especially when a movement is implied.

(74) *tíi* 3sg.obl *so* def.msg.nom *meewá* fruit *samaṭ-í* collect-cv *ba* top *booǰée* bag.obl *wée* in *de* put.cv *wheelíl-u* take.down.pfv-msg 'He collected the fruit, put it into the bag and brought it down [home].' (A:HUB010)

(75) *so* def.msg.nom *iškaarí* hunter *méeš* man *muṭ-á* tree-obl *wée* in *hín-u* be.prs-msg 'The man is in the tree.' (B:CLE375)

The noun taking this postposition may also be an abstract entity, such as 'our school class' in (76).

(76) *asíi* 1pl.gen *ǰameet-í* class-obl *wée* in *áaṣṭ* eight *kuṇaak-á* child-pl *hín-a* be.prs-mpl 'There are eight children in our class.' (A:OUR010)

*ǰe* **'up (into), up (along)'.** This rather specialised spatial postposition takes an oblique nominal argument or an oblique pronominal argument. It is exclusively

### 8.2 Postpositions

used with nouns or pronouns with inanimate referents. It is similar to *wée*, in that it signals contact with the surface of something or the penetration into a location, but it is restricted to an upward movement, such as that of the hunter making his way up into the tree in (77), and it occurs mostly with verbs having a connotation of upward movement. It seems that this postposition primarily has a connotation of movement, but there are examples of it being used in a static sense as well.

(77) *hasó* rem.msg.nom *iškaarí* hunter *ba* top *bhíilam-e* fear-gen *loomuṭ-á* deodar.tree-obl *ǰe* up *huṇṭráak* upward *ukháat-u* come.up.pfv-msg 'Due to fear, that hunter climbed into a deodar tree.' (B:CLE357)

*thíi* **'from'.** This spatial postposition takes a genitive nominal argument, but there is no evidence for any pronominal forms occurring with it. It is exclusively used with inanimate nouns (and as a directional specifier of spatial adverbs, see §8.1.2) and is in many ways used interchangeably with a locative nominal expression in the genitive case. However, it may serve a function of emphasising the movement away from (as it does when used with certain adverbs), whether in a concrete spatial sense, as in (78), or in an extended temporal sense, as in (79), although the latter is not very common.


*khúna/kéeči* **'(near) to, with'.** These two synomymous postpositions take an oblique nominal argument (81) or an accusative pronominal argument (80). They are almost exclusively postposed to nouns or pronouns with animate (and particularly human) referents. It can describe the position 'near to, at, with' as well as the movement '(near) to'. In the directional meaning, *khúna/kéeči* may also be followed by *the* 'to' in a postpositional sequence (see §8.2.4).

(80) *amzarái* lion *tas* 3sg.acc *khúna* near.to *yhóol-u* come.pfv-msg *hín-u* be.prs-msg 'The lion approached him.' (A:UNF013)

### 8 Adverbs and postpositions

(81) *tusím* 2pl.erg *lúuɡ-a* strange-obl *míiš-a* man-obl *kéeči* with *ma* 1sg.nom *baasaá* spend.night.cv *míi* 1sg.gen *bheezatí* disgrace *thawéel-i* cause.to.do.pfv-f 'You disgraced me by making me spend the night with a stranger.' (A:UXW051)

*šíiṭi* **'in, inside'.** This postposition, which is also used as a spatial adverb (see §8.1.2), takes an oblique nominal argument or an oblique pronominal argument. It is exclusively postposed to nouns or pronouns with inanimate referents, particularly those denoting enclosed areas or places, such as the 'house' in (82). It can describe the position 'in, inside' as well as the movement 'into, inside'.

(82) *eení* prox *ɡhooṣṭ-á* house-obl *šíiṭi* inside *ma* 1sg.nom *seé* sleep.cv *hín-u* be.prs-msg 'Inside this very house I was asleep.' (A:HUA014-5)

In the directional meaning, *šíiṭi* may also be followed by *the* in a postpositional sequence (§8.2.4).

*pharé* **(***B phará***) 'along, through, across, over'.** This postposition takes an oblique nominal argument or an oblique pronominal argument. It is exclusively used as a postposition of nouns or pronouns with inanimate referents. It expresses an outstretched contact of one entity with the surface of another entity or that the two are located parallel to one another. Typically this is the way of expressing path, as in (83).

(83) *hazrát* Lord *iisa=aleehisalaám* Isa=peace.be.upon.him *eesé* rem *páand-a* path-obl *pharé* along *bi-áan-u* go-prs-msg 'Lord Isa [Jesus], PBUH, came walking along that path.' (A:ABO033)

However, this postposition can capture a number of slightly different, but still related, positions and movements, such as the bread put on the mouth of the dead man in example (84).

(84) *tas* 3sg.acc *mheer-í* kill-cv *ɡal-í* throw-cv *zaalim-aan-óom* brute-pl-obl *dhút-a* mouth-obl *pharé* toward *ɡúuli* bread *bi* also *de* put.cv *ɡíia* go.pfv.pl *de* pst 'The brutes, who had killed him, had also put bread in his mouth and left.' (A:GHA076-7)

### 8.2 Postpositions

*dúši/ḍáḍi* **'toward, at, in the direction of'.** This postposition takes an oblique nominal argument or an accusative pronominal argument if animate and an oblique argument if inanimate and locative. While *dúši* is used in A as well as in B, *ḍáḍi* seems to be most common in B. It is used with nouns or pronouns with human or animate referents as well as with locative expressions, such as the 'village' in (85).

(85) *díiš-a* village-obl *dúši* toward *tilíl-i* walk.pfv-f *hín-i* be.prs-f 'She started to walk toward the village.' (A:KAT085)

*maxadúši* **(B** *muxadúši***) 'before, in front of'.** This diachronically complex postposition (with the above introduced *dúši* as one of its components) is also used as an adverb. It takes a genitive argument, *kuṇaak-íi* in (86), which in most cases has a human referent. The relation expressed is commonly of an abstract kind. For temporal precedence a compound postposition *díi muṣṭú/díi muxáak* 'before' is used (§8.2.3).

(86) *kuṇaak-íi* child-gen *maxadúši* in.front.of *šóo* good.msg *kráam* work *ta* cntr *th-eeṇḍeéu* do-oblg 'One should act properly in front of children.' (A:SMO022)

*patú* **'after, behind'.** This postposition takes an oblique nominal argument, as in (87), or an accusative pronominal argument. While the same word is almost exclusively used as a subordinate conjunction in B (corresponding to *pahúrta* in A), it primarily expresses a spatial relationship in A. It is also used in the sense of being 'in the pursuit of' something.

(87) *pal-í* hide-cv *áak* idef *ṭómb-a* trunk-obl *patú* behind 'I hid behind a tree trunk and…' (A:HUA072)

*tií* **'until, up to, as far as'.** This postposition takes an oblique nominal (and inanimate) argument, 'knees' in (89), or, if referring to a point in time or space, a locative spatial adverb, as seen with *eeṛáa* in (88). It shows that something extends to a specific point in space or time.

### 8 Adverbs and postpositions


*dapáara* **(A)/***pándee* **(B) 'for'.** This postposition, which expresses purpose (if inanimate) or beneficiary (if human), takes a genitive argument. The beneficiary use can be seen in (90).

(90) *ma* 1sg.nom *ba* top *aní* prox *pooštrá* fattened *abaíim* she.goat.pl *thíi* 2sg.gen *dapáara* for *saat-áan-u=ee* keep-prs-msg=q 'Do you think I have taken care of these fattened goats for you?' (A:PAS093)

When a purpose is expressed pronominally, it is the form of the pro-adverb that expresses a movement from a location (i.e., an ablative) that is being used, e.g., *eeṛáai* (91) in A, or *hatáawuu* (92) in B. The postposition *dapáara/pándee* may also take as an argument a Verbal Noun in a purpose clause, (see §14.4.2).


### **8.2.3 Compound postpositions**

Compound postpositions are phrases consisting of one of the simple directional postpositions, *díi* 'from, of' and *the* 'to', followed by an adverb. Although this list probably is far from exhaustive, this postpositional type does not seem to be extremely common in Palula. It should be noted that they need not occur immediately next to each other.

8.2 Postpositions

*the nhiáaṛa* **'close to, next to'.** The spatial adverb *nhiáaṛa* 'near, nearby' (see §8.1.2) occurs with the postposition *the* 'to', meaning 'near to, close to', as in example (93).

(93) *malikcheétr ̣* Malikchetre *thaní* qot *ɡhróom-a* village-obl *the* to *nhiáaṛa* near *áak* idef *chíitr ̣* field *hín-u* be.prs-msg 'Near to the village, there is a field called Malikchetre.' (A:JAN029)

The adverb component in the compound can be further modified, such modifications, in (94) *tuúš bi*, occurring between the adverb and the simple postposition.

(94) *se* def *wíi-a* water-obl *the* to *tuúš* a.little *bi* also *nhiáaṛa* near *ɡíi* go.pfv.f *hín-i* be.prs-f 'She went a bit closer to the water.' (A:SHY054)

If the argument is human, e.g., 'me' in (95), a compound *keeči the nhiáaṛa*, consisting of a postpositional sequence (§8.2.4) and an adverb, is used.

(95) *ma* 1sg.nom *kéeči* with *the* to *nhiáaṛa* near *bheš* sit.down.imp.sg 'Sit down next to me!' (A:AYB029)

*díi muṣṭú/muxáak* **'before, prior to'.** The temporal adverb *muṣṭú* (B *muxáak*) 'before, in the past' (see §8.1.3) occurs with the postposition *díi* 'from', meaning 'prior to', as in example (96). The compound *díi muṣṭú/muxáak* also occurs with Verbal Nouns in temporal subordination (§14.4.1).

(96) *asíi* 1pl.gen *dóodu* grandfather *choók ̣* Choke *kaṭur-á* Kator-obl *díi* from *dúu* two *sóo* hundred *kaal-á* year-pl *muṣṭú* before *yhóol-u* come.pfv-msg *hín-u* be.prs-msg 'Our ancestor Choke arrived two hundred years prior to the Kator [dy-

nasty].' (A:ASH047)

*díi dhuúra* **'far from, away from'.** The spatial adverb *muṣṭú* (B *dhuúra*) 'far away' (see §8.1.2) occurs with the postposition *díi* 'from', meaning 'far from, away from', as in example (97).

### 8 Adverbs and postpositions

(97) *taníi* 3pl.gen *chíitr ̣* field *ba* top *taníi* 3pl.gen *ɡhooṣṭ-á* house-obl *díi* from *taqriibán* about *tróo* three *čúur* four *kulumiṭer-á* kilometre-pl *dhúura* away.from *de* be.pst 'Their field was about three to four kilometres away from their house.' (A:WOM468)

### **8.2.4 Postpositional sequences**

Another kind of complex postposition consists of a sequence of two simple postpositions. The first in the sequence is the semantically more central one, whereas the second is a further fine-tuning or specification, usually as far as the direction is concerned. Exactly which postpositions can and which cannot be combined in this fashion is a matter of further research; (98)–(103) are only a few illustrative examples. It should be noted that the noun preceding the postpositional sequence is regularly assigned case by the first component of the sequence. The (nominally derived) postposition *dúši*, itself assigning oblique case to the preceding noun in (101), receives genitive marking from the second postposiion *thíi*.


8.2 Postpositions

(103) *ǰeep-í* pocket-obl *wée* in *yúu* from *se* def *ɡhambúri-m* flower-pl *ɡhaḍíl-im* take.out.pfv-fpl 'He brought out the flowers from inside his pocket.' (B:FLW794)

The second component in the sequence *wée yúu* in (103) does not occur as an independent postposition but is certainly related to the ablative function of this particular segment *(y)uu/w(uu)* occurring in a few spatial expressions, especially in the B variety: *hatáa-(w)uu* 'from there', *aṛáa-(y)uu* 'from there', *índee-(y)uu* 'from here', *ɡóo-(y)uu* 'from where', but also as *-oo* in some temporal adverbs in A (see §8.1.3 above). Schmidt & Kohistani mention a rare ablative suffix *-nyuu/nuu/uu* in Kohistani Shina with a very similar distribution (2001: 130).

## **9 Verbs**

### **9.1 The verb and its properties**

Distributionally, the verb normally occurs clause-finally, functioning as the main predicate of the clause. The verb in (1) is preceded by a single argument, a subject, whereas the verb in (2) is preceded by an oblique (postpositional) object, a subject and a direct object (see Chapter 12).


Verbs can also function as the predicate of a dependent clause, and depending on what type of clause this is, the verb occurs as an inflected finite verb (just like in the examples above) or in one of its non-finite forms (see §14.3–§14.6). Usually, the verb also then is clause-final.

Phonologically, the vast majority of all verb stems in Palula are either monoor disyllabic, the monosyllabic stem being the most commonly occurring nonderived stem.<sup>1</sup> Verb stems are without exception accented on their final syllable, either on a short vowel or on the first mora of a long vowel.

The typical monosyllabic verb stem has a CVC-structure, but CVCC, CCVC, CCCVC and VC also occur. A small class of (about ten) monosyllabic verbs has a CV, or a CCV, stem, among them some of the most frequently occurring verbs in the language. These different patterns are exemplified in Table 9.1.

<sup>1</sup> The distinction between derived and non-derived stems is not altogether straightforward, as, from a diachronic perspective, many transitive stems that can be considered derived causatives do not have any synchronic non-causative counterparts. For the most part in this work, I consider those as stems in their own right, i.e. non-derived. I refrain from distinguishing between root and stem, as that would raise more questions than could be answered here.

### 9 Verbs


Table 9.1: Examples of mono-syllabic verb stems

### Table 9.2: Examples of poly-syllabic verb stems


There is a variety of structures among disyllabic verb stems (Table 9.2), the most common being a CVCV stem, and for the small group of trisyllabic verb stems, CVCVCV is the most common.

Although verbs form a major word class in most, if not all, of the world's languages, the way events are encoded varies a great deal. One effect of this

### 9.1 The verb and its properties

is a dramatic variation in the number of verbs found in languages. While at one extreme, the main European languages, such as English, can boast 10,000 or more verbs, there are at the other extreme languages in other parts of the world with markedly different lexical structures that manage with minimal verbal systems of 10–40 simple verbs (Viberg 1994: 347–348, 2006: 409). In the light of that we will try to discern whether Palula has a verbal structure similar to European languages or should rather be included in the category of languages with minimal verbal systems or at least be said to share some lexical characteristics of either one.

Regardless of the size of the verb lexicon, the twenty most frequent verbs in any one language tend to have some characteristics in common, and a number of basic meanings coded as verbs are more or less bound to show up here (Viberg 2006: 209). The presence of the verbs GO, GIVE, TAKE, MAKE, SEE and SAY, Viberg (1994: 247) points out as unmarked, in this sense, occurring as highly frequent verbs in English and in a number of other European languages as well as in non-European languages.

In Palula, too, we find these verbs among the Top Twenty (Table 9.3), if we count *har-* 'take away' as roughly corresponding to TAKE and *daché- ̣* 'look' as equivalent to SEE, although in the latter case there is a more general perception verb *paš-*, glossed as 'see', that is 27th in the ranking. Viberg (2006: 409) refers to these "universal" verbs as *nuclear*, covering such basic semantic domains as motion, possession, production, verbal communication and perception. He also mentions the verbs HIT, COME, KNOW and WANT as possible candidates for this *nuclear* group, although with a typologically slightly more marked status. In the Palula Top Twenty list we also find HIT and COME, whereas the closest equivalent of WANT shows up in the 21–40 range as *dawá-* 'ask for' and KNOW (as a simple verb) with an even lower frequency.

Not all the Top Twenty verbs of European languages, such as English, are *nuclear* in the same sense and must therefore be defined as language- or area-specific. This is the case with BE and HAVE, the first one being the overall most frequent in almost all European languages, and the second a verb with few parallels outside Europe. In English, the modals *will, can, may, shall* and *must* are all among the twenty most frequent, and in many other European languages the modals CAN and MUST are found in this frequency range (Viberg 1994: 346– 349).

A similar tendency can be seen when studying the Palula Top Twenty list. As in European languages, an equivalent of BE tops the list by a wide margin, not wholly surprising considering the solidly Indo-European identity of this lan-

### 9 Verbs

guage. The suppletive and defective verb *hin-* 'is', with its invariable past-tense form *de*<sup>2</sup> (itself possibly a grammaticalisation of a participle form of GIVE) is the Palula copula as well as an important auxiliary participating in the formation of a number of periphrastic tense-aspect categories. The verb *háans-* 'live, exist' is in a similar manner grammaticalised in one of its uses, mainly "standing in" for the defective *hin-* in some of the TMA categories.

The two verbal communication verbs *thané-* and *mané-* interact in a languagespecific way and are, beyond their simple characteristic as utterance-verbs that take direct-quote complements (Givón 2001a: 155), also grammaticalised in one or more forms as quotatives and hearsay markers, respectively, see §10.2.4 and §14.5.1. <sup>3</sup> The high frequency of the verbs *bhe-* 'become', *the-* 'do, make', and *de-* 'give' reflects a subareal feature that we will have reason to return to in our discussion below, namely that in addition to their more literal meaning mainly are featured as verbalisers (Masica 1991: 368) or "dummy verbs" in so-called "conjunct verbs" (1991: 326), see §9.6.1.

Another language-specific or possibly subarea specific feature can be traced in the presence of at least four verbs coding events close to COME and GO among the Top Twenty: *be-* 'go', *yhe-* 'come', *whe-* 'get down' and *nikhé-* 'appear, get out'. While the first two of these four motion verbs are spatially (but not directionally) neutral, the two others include a spatial specification (along with a directional neutralisation), *whe-* coding a movement up-to-down, and *nikhé-* a movement inside-to-outside. Another verb belonging to this group, although with respect to frequency at a much lower ranking, is *ukhé-* 'get up', coding a movement reverse of *whe-*, i.e. down-to-up. This tendency finds a number of parallels in other areas of the lexicon (spatially "fine-tuned" adverbs, demonstratives, postpositions, etc).

Leaving the individual verbs, their lexical-grammatical characteristics, and the events they code aside, one of the more striking observations we can make has to do with the relative textual verb occurrence within certain frequency ranges. Viberg (2006: 409) claims that the twenty most frequent verbs tend to cover close to 50 per cent of the textual frequency of verbs in European languages. He compares that with the language Kalam in Papua New Guinea, with a total number of simple verbs around 100, of which fifteen verbs account for 90 per cent of the textual occurrences. As indicated in Table 9.4, Palula is located between

<sup>2</sup> The form *de* is alternatively analysable as a mere past-tense marker, see §9.3.12 and §13.1, and as such not part of any verb-specific paradigm.

<sup>3</sup> This, however, is not reflected to any greater extent in this frequency count where only finite verb forms are taken into account.


Table 9.3: Palula Verbs Top Twenty. The twenty most frequent verbs. (The percentage is calculated on occurrence of finite verb forms in the text corpus.)

these two, with close to 80 per cent accounted for by the Top Twenty verbs. That makes it significantly different from the European type, but it is still quite different from languages with minimal verbal systems.

Interestingly, half of the occurrences in text is accounted for by only the five topmost verbs, among them the two most productive verbalisers, *bhe-* 'become' and *the-* 'do, make', with the other fifteen verbs in the Top Twenty comprising another quarter of all verbs. The following twenty verbs account for a tenth, while the remaining 100 or so verbs only represent twelve per cent of the total number of verbs occurring in the text corpus. This does not mean that Palula

### 9 Verbs

Frequency range (number of verbs) % finite verb occurrence 1–5 (5) 51.5 6–20 (15) 26.2 21–40 (20) 10.0 41–138 (98) 12.3 All (138) 100.0

Table 9.4: Palula textual verb occurrence related to frequency ranges

has no more than 138 simple verbs (in fact, I have elicited more or less complete paradigms for nearly 400 verbs<sup>4</sup> ), but it suggests that the total number is likely to be in the hundreds rather than in the thousands, and that any verbs beyond these 138 are quite infrequent, although we must not make too hasty conclusions based on a rather small corpus that is somewhat limited in terms of genre. Can the lexical structure observed for Palula in some ways be related to lexical typology in general? Is it possible to trace any particular areal or subareal features responsible for some of the properties of the lexical structure in Palula? Are we observing the effects of an ongoing development of the lexical structure in one direction or the other? These are big and interrelated questions, and I only intend to hint at some possible explanations and give suggestions for further research.

Viberg (2006: 409) only contrasts two extremes as it seems; on the one hand, we have, from a European perspective, well-known languages with large verbal systems, comprising thousands of simple verbs, some with very specialised meanings; on the other, we have languages with minimal systems, comprising less than 100 simple verbs. In the latter case, a small set of simple verbs are used systematically as building blocks to form complex verbs with more specialised meanings equal to those of semantically specialised simple verbs in the languages of the former type. Palula, however, does not really fit into any of those two extremes. Instead it seems to belong to an intermediary type. Its verb inventory is quite rich, and some meanings are rather specialised, as we saw above with the spatially specified motion verbs, but we can probably count the total number in the hundreds rather than in the thousands. It shares with the "minimal type" the property of specialised event coding by means of complex verbs (or rather, complex predicates). Viberg (2006: 348) exemplifies two such strategies, one

<sup>4</sup> As simple verbs in this case, I also consider attested stems derived productively with the valency-increasing *á/awá*-suffix and the valency-decreasing *íǰ*-suffix respectively.

### 9.1 The verb and its properties

combining a noun or an adjective with a verb, and in effect forming new lexical units equal to simple verbs: noun+verb > complex predicate, or adjective+verb > complex predicate; another by combining two verbs: verb1+verb2 > complex predicate.

The former strategy (discussed below, §9.6.1) is well attested in Palula, and I will, in line with Masica (1991: 326), refer to this as the *conjunct verb* construction (while in some other traditions it is called a *light verb* construction). It is a productive and easily applicable strategy, especially for verbalising culturally new concepts, and we find, perhaps not surprisingly, a substantial number of loan words (primarily from Urdu or Pashto) in the so-called "host" slot of *conjunct verbs*. The *conjunct verb* as a phenomenon is neither language-specific nor areaspecific. Instead it seems to be a feature (although not exclusively) of languages in a larger "macro-area", possibly comprising a large part of Asia. Nevertheless, Masica notes "impressionistically" that it does appear to be more common in the northern part of South Asia than in the south, possibly owing something to Persian influence (1991: 368). As already noted above, there are primarily three verbs acting as verbalisers in Palula, each representing a particular argument structure (see §13.2.8). This strategy may very well be on the increase in Palula, partly due to influence from languages of wider communication. Since it is such a productive strategy, especially for incorporating culturally new concepts (Gambhir 1993: 85), it is not too farfetched to assume that once a model like this has been established even simple verbs with a specific meaning already existing in the language may be replaced by the more easily accessible *host* + *verbaliser* constructions. So far in Palula, however, it seems to be primarily a question of creating entirely new vocabulary to suit new situations, such as access to new technology, formal education and acquisition of new knowledge, and only to a limited extent a matter of replacing already existing basic vocabulary.

The latter is of course not in itself a threat to the Palula inherited lexicon, but it may eventually mean a rather radical restructuring. An example of this we can see in Persian, which in many ways stands as an "areal model" as far as conjunct verbs are concerned. In Persian, these "new" verb complexes have been gradually replacing "old" simple verbs for the last 700 years, resulting in the existence of numerous parallel complex and simple verbs corresponding to more or less the same verbal concepts. An additional effect of this phenomenon has been to provide a literary language or prestigious register in which the simple verb is used vis-à-vis an everyday language in which the corresponding complex predicate is applied to a much higher degree (Folli, Harley & Karimi 2005: 1369).

### 9 Verbs

The second strategy mentioned above, through which new predicates can be formed by combining two simple verbs, is not as well attested in Palula (some possible examples will be discussed under §9.6.2), while it is considered a typical – although not exclusive – IA feature (Ebert 2006: 559; Masica 2001: 250–252).

As to the structural properties of verbs, all the three morphological markers commonly found on verbs (Viberg 2006: 409) are present in Palula: TMA markers, agreement markers, and valency markers. These will be dealt with in §9.4– §9.5, but before discussing these inflectional categories, some important classes of verb stems will be introduced and exemplified.

### **9.2 Stems and verb classes**

As far as inflectional morphology is concerned, there are two main morphological verb classes in Palula, which I have chosen to refer to as L-verbs – an open, productive and large class – and T-verbs – a closed, non-productive and rather heterogeneous class. Additionally, there is a small group of verbs with stems that are suppletive to varying degrees (see Table 9.5). Also within each of the main classes there are variations in the inflectional paradigms (described in detail below) due to accent-position and the quality of stem vowels.

All of the inflectional suffixes are each associated with either a perfective or an imperfective stem, as illustrated in the partial paradigm (Table 9.5). For most verbs the perfective stem (stripped of its perfective "inflection") is identical to the


Table 9.5: Partial paradigm illustrating stems and main morphological verb classes

9.3 Morphological verb classes

imperfective stem (see §9.4.3), but in some of the classes the two are clearly distinguished as separate stems. For a more consistent treatment, however, I have chosen to indicate all perfective stems in this work inclusive of morphological perfectivity, regardless of it being present in a regular or predictable form or as stem modification/alternation only. Apart from imperfective-perfective stem alternations, some imperfective stems show vowel alternations between imperfective verb forms with an accent-bearing suffix and those imperfective verb forms that are formed with an accent-neutral suffix (as described in §3.4.3 and §3.5.1). What I cite as a stem (with a final hyphen) is an underlying form, mostly corresponding to the non-strengthened vowel quality/quantity found with accentbearing suffixes.

### **9.3 Morphological verb classes**

The following classification of verbs into different classes is based on formal criteria, taking both stem alternation and inflectional allomorphy into consideration. Only a few inflectional categories will be exemplified in the present section, especially as indicators of verb class membership, whereas a fuller description of verb inflection follows in §9.4.

### **9.3.1 Consonant-ending L-verbs**

As mentioned above, the L-class, so named because of its perfective ending in *íl, óol*, etc., includes the majority of verbs and is also the productive formation insofar as new simple verbs are formed (which, however, is synchronically quite rare). Within this large class, we can discern three important subclasses, each including a substantial number of verbs, primarily recognised on the basis of suffix vowels (or more correctly the vowels arising from the coalescing of final stem vowel and an actual suffix vowel). The first is verbs with consonant-ending stems (Table 9.6). The great majority of them are intransitive, and all morphological passives formed with *-íǰ* (see §9.5.2) also belong to this subclass, but a few non-derived transitive verbs are also found. For perfective stems, the accent always occurs on the final vowel *í* (of the segment *íl*), which is the single feature setting them apart from the imperfective stems. One of the more prominent features of this subclass is the consonant-ending singular imperative, for most of these verbs identical to the stem itself.

A few examples of the large subclass of consonant-ending L-verbs are shown in Table 9.7. Regular morphophonemic alternations *aa–oo–ee*, *ee–ii* and *a–aa* (as

### 9 Verbs

Table 9.6: Partial paradigm for consonant-ending L-verbs


Table 9.7: A selection of consonant-ending L-verbs


shown by the parenthetical forms in the lists with examples in this chapter) are results of historical strengthening/lengthening and umlaut (as described in §1.4 and §3.5).

### **9.3.2** *a***-ending L-verbs**

The next subclass comprises verbs with stems ending in an accented *á* (Table 9.8). Many, but not all, of these verbs are transitive, and among them we also find all causative and transitive verbs (productively) derived with *-á* or *-awá* from noncausative and intransitive verbs. The final vowel of the stem has coalesced with the suffix vowel and sometimes been subject to further historical strengthening or given rise to umlaut. Unique for this particular subclass are thus the future verb forms with *óo* and the perfective segments *óol* or *éel*.

### 9.3 Morphological verb classes


Table 9.8: Partial paradigm for *a*-ending L-verbs

The stem-*á* followed by the suffix *-áan* has resulted in a few parallel present tense surface realisations: /kʰaˈjaːnu/, /kʰiˈaːnu/, /ˈkʰaːjnu/, /ˈkʰajnu/. The form with the epenthetic *-y-* /j/ is the dominant one in the B variety, whereas the other forms are commonly heard in A.<sup>5</sup> This is one of the largest subclasses, and – again – the verbs listed in Table 9.9 are only a few examples.

Table 9.9: A selection of *a*-ending L-verbs


### **9.3.3** *e***-ending L-verbs**

The third important subclass of L-verbs (Table 9.10) includes both intransitive and transitive verbs. Here we find some transitive verbs clearly corresponding to or

<sup>5</sup> Morgenstierne (1941: 22) does not report any forms other than those with *y*, which is why we can assume that to be the more conservative pronunciation and the other ones to have evolved during the last few decades.

### 9 Verbs

derived from intransitive verbs in other verb classes, although often through less transparent (and synchronically unproductive) processes than the ones we found in the *a*-ending class.


Table 9.10: Partial paradigm for *e*-ending L-verbs

A few other examples of such verbs are displayed in Table 9.11. An underlying (and historical) final accented *é* is assumed here that also occurs in the surface form of B (but not A) imperative singular. Although the final *é* has been subject to deletion in most other forms, the future tense *-íi* results from the coalescence of this stem-*é* and the suffix-*e* (compare with the *-ée* in the closed syllable of the 2sg future verb form in B, *aṭ-éeṛ*, etc, where this strengthening process has not applied).

Table 9.11: A selection of *e*-ending L-verbs


Only some of the verbs in this class have a long accented vowel in the perfective forms with *íil*, while most are left with a short *íl*. I don't have any explanation

9.3 Morphological verb classes

for this, but interestingly, Morgenstierne (1941: 22–23) has documented alternative ("older") forms of *the-* 'do': *thiānu* (prs msg) and *thīelo* (pfv msg), which suggests that there may have been for all *e*-ending verbs an intermediate form with both the vowel quality of the stem and that of the suffix preserved. For this particular verb, the modern perfective forms differ between the dialects, so that the A form uses a long *íil*, whereas B has a short *íl*. Also, the converbs of *the-* and *bhe-* (two extremely frequent verbs, being the two most common verbalisers, see §9.6.1 and §13.2.8) are formed with *e* (*the* and *bhe*) instead of the *-í* of the other verbs of this subclass.

### **9.3.4 Other L-verbs**

Two other frequent L-verbs should be mentioned that do not easily fit into any of the above-mentioned subclasses: *yhe-* 'come' and *ru-* 'cry' (in Table 9.12). Both have (at least historically) stems ending in vowels that in various ways interact with the suffix-vowels.


Table 9.12: Partial paradigm for two vowel-ending L-verbs

Although the paradigm of *ru-* in many ways reminds us of that of the *e*-ending L-verbs, and therefore in some sense could be seen as a subgroup of that class, *yhe-* presents a more complicated case. For one thing, the present tense is not like anything we have seen in the classes presented above, with its *áand*-formation instead of the *áan*-suffix we have seen so far. There is reason to return to this issue shortly, as this present-formation is also a conspicuous feature of a group of T-verbs. The other problem has to do with the quality of the underlying stem vowel, giving rise to a future suffix akin to the *e*-ending verbs above, while the perfective is more what would be expected with a final *á*. Although I represent the stem with a final *e* for the time being, it is merely a vocalic placeholder; it

### 9 Verbs


Table 9.13: Partial paradigm for consonant-ending T-verbs

may very well go back to a stem ending in a diphthong *ái* or *éi*. In B there is also an unexpected alternation between *y* and *yh*, stem-initially, an alternation that cuts right through the imperfective realm. We will return to this issue of aspiration when discussing the other verbs with *-áand*, but for now we will only regard it as an irregularity among the L-verbs.

While the L-verbs within their respective three subclasses present a rather homogeneous picture, the paradigms of the T-verbs display a much higher degree of irregularity. In reality, these verbs, together with the suppletive verbs mentioned, could be seen as a continuum stretching from verbs with almost agglutinative morphology, through verbs that show an increasingly irregular correspondence between perfective and imperfective stems, and ending up at the other extreme with entirely suppletive verbs, without any segments in common and with irregular inflections.

### **9.3.5 Consonant-ending T-verbs**

The T-verbs form, together with the suppletive verbs, a closed class of verbs, comprising between 50 and 60 verbs altogether in my database. Many of these verbs form their perfectives with a plosive segment, in the clear cases a *t*-suffix, but often this has been assimilated with preceding stem segments, hence the name. The largest subclass of the T-verbs are those with a stem ending in a consonant (Table 9.13). Like the consonant-ending L-verbs, the imperative singular is identical to the imperfective stem, save for historical strengthening/lengthening. This class includes intransitive as well as transitive verbs, with no particular preference for one or the other.

### 9.3 Morphological verb classes


Table 9.14: A selection of consonant-ending T-verbs

In some cases, the imperfective and perfective stems are identical or nearly identical, save for a *t*-segment added to the perfective. In other cases, the stems are considerably different, involving stem vowel alternation, alternation between an aspirated and an unaspirated stem, metathesis or assimilation of a final consonant with the perfective *t*-element. In the list with additional examples of such verbs (Table 9.14), the imperfective stem is listed in the first column along with an alternating imperfective stem form within parenthesis, followed by the perfective stem in the second column.

### **9.3.6** *e***-ending T-verbs**

The next subclass, *e*-ending T-verbs (Table 9.15), is considerably smaller than the consonant-ending subclass, and is similar to the *e*-ending L-verbs. Even here we need to reconstruct a stem ending in an accented *é*, and again the B variety has regularly preserved this *é* in imperative singular. The perfective and imperfective stems are identical for all of these verbs, save for an ending segment *íit* or *ít* in the perfective.

### 9 Verbs

Some more examples are displayed in Table 9.16. Some of these verbs are to varying degrees grammaticalised in some of their uses, *de-* (and possibly also *ǰe-*) as a verbaliser, and *thané-* as a quotative, and *mané-* as a hearsay marker. Like *the-* 'do' and *bhe-* 'become' of the *e*-ending L-verbs, *de-* and *ǰe-* occur as converbs (see §13.2.8) in the forms *de* and *ǰe* instead of with the regular suffix *-í*.

### **9.3.7 Accent-shifting T-verbs**

Two small groups of T-verbs can be described as hybrids between consonantending and *e*-ending verbs. The verbs in one of them (Table 9.17) behave like the consonant-ending verbs in the imperfective and like the *e*-ending verbs in the perfective.

The verbs of the other group (Table 9.18) look like *e*-ending verbs in the imperfective but like consonant-ending verbs in the perfective.

Whereas the first group is all intransitive, with similar imperfective and perfective stems, the second are all transitive with some sort of unpredictable stem alternation. Most likely, *ɡalé-* and *ukualé-* have developed from being *e*-ending L-verbs (which they still are in B) to assimilating the two identical segments adjacent to one another and thus losing their characteristic *íl*-endings. The *e*-ending L-verb *whaalé-* 'take down' (presented above), which is structurally similar and semantically parallel to *ukualé-*, is accent-shifting with some speakers of the A variety.

### **9.3.8** *aand***-verbs**

As already mentioned above, there is a group of four (intransitive) motion verbs that all share the present-tense allomorph *-áand* vis-à-vis the usual *-áan*. One of them, *yhe-* 'come', an L-verb with a unique paradigm, was mentioned already. The remaining three (in Table 9.19) are T-verbs. As with *yhe-*, the final *e*-element is merely a vocalic placeholder as far as representation is concerned.<sup>6</sup>

Following Morgenstierne (1941: 22), the present-tense suffix is analysed as an old participle *-ant-*. What we do not know at this point is why it mostly occurrs in modern-day Palula as *-áan*, whereas only this small group of verbs, with its*-áand*, has preserved the consonant cluster (although with a lengthened vowel

<sup>6</sup> The paradigm of this verb is not entirely stable within the A dialect; it is sometimes treated as an L-verb.

<sup>7</sup> In B (and for some A-speakers) an *e*-ending L-verb.

<sup>6</sup> I have suggested an underlying *-éi* elsewhere (Liljegren & Haider 2011; 2015b). This is also reflected in the morphological representation in the sample texts.


Table 9.15: Partial paradigm for *e*-ending T-verbs

Table 9.16: A selection of *e*-ending T-verbs


Table 9.17: Examples of accent-shifting T-verbs, type 1


Table 9.18: Examples of accent-shifting T-verbs, type 2


*a* In B an *e*-ending L-verb.

*b* In B an *e*-ending L-verb.

### 9 Verbs


Table 9.19: Partial paradigm for T-verbs with -*aand* (present)

and with the plosive voiced in the position before the vocalic gender/number suffix, of which both are well-supported processes in the Palula diachronic development). One line of thought to be pursued is that aspiration, now segmentally at the onset of the first syllable, historically would have been located at the morpheme break, i.e. between the stem and the participial suffix *(ukháand- < uké+h+ánt)*, only later transposed leftwards (another well-attested process in the Palula lexicon). This is supported by the lack of aspiration in, for instance, the historical derivation *ukualé-* 'bring up' from *ukhé-* 'come/go/climb up'.

### **9.3.9** *i***-ending T-verbs**

Another subclass of equal size is one ending (at least underlyingly) with the vowel *i*, as seen in Table 9.20. They are also (Table 9.21), like the previous group, in some sense motion verbs, at least three of them describing upward motion. Two of them are transitive and two intransitive (including the already exemplified *uthí-*).

### **9.3.10** *u***-ending T-verb**

In my database, there is a single T-verb ending in a vowel *u* (or *o*), the verb *su-* 'sleep, fall asleep' (in Table 9.22).

### **9.3.11 Suppletive verbs**

There are only three radically suppletive verbs in the language. One, *paš-/dhriṣṭ-* 'see', was already introduced in Table 9.5. The other two are the verbs *be-/ɡ(a)-*

### 9.3 Morphological verb classes

Table 9.20: Partial paradigm for *i*-ending T-verbs


Table 9.21: Examples of *i*-ending T-verbs


'go' and the copula/auxiliary *hin-/de*. The latter will be presented in §9.3.12 along with other highly irregular or defective verbs. Due to its unique forms, the verb 'go' is presented in a more comprehensive paradigm in Table 9.23, although we will return to the inflectional categories themselves and their functions in the sections to follow.

### **9.3.12 Irregular verbs and verbs with highly grammaticalised functions**

### **The copula/tense auxiliary**

The copula/tense auxiliary has an incomplete paradigm, displayed in Table 9.24, using forms of the verb *háans-* 'live, stay' in the imperfective with non-present reference. The perfective (or more correctly, past, as far as this particular verb is concerned) *de* is invariable and most probably a (fairly recent) grammaticalisation of the conjunctive participle of *de-* 'give, put'. There is as far as I know no other Shina variety with a similar past-tense copula. Instead they all tend to go back to the same stem: Kalkoti *aas*, Sauji *al-*, Gilgiti *as-* (Radloff & Shakil w. Shakil 1998), Kohistani Shina *asílo-*, Drasi *asiló-*, Guresi *asúlu-* (Schmidt 2004a: 44–45). The latter is, in addition to the absence of an overt copula with many non-

### 9 Verbs


Table 9.22: Partial paradigm for the vowel-ending verb *su*-



verbal predicates, an argument for an alternative interpretation (as argumented for in §13.1.1) of *de* as simply a past tense marker. The use of these forms as tense auxiliaries will be further discussed and exemplified in §10.1.5–§10.1.8, and their copular use in §13.1.

### 9.3 Morphological verb classes


### Table 9.24: Paradigm for copula

### **Modal verbs and verb forms with highly grammaticalised functions**

*ṣáat-* **'start, begin'.** The perfective of the verb *ṣač-* 'climb, rise', functions, apart from its lexical use, as a modal verb with a complement predicate in the infinitive (see §13.2.7 and §14.5.2).

*bha-* **'can, be able to'.** This *a*-ending L-verb is an entirely modal verb, taking complement predicates in the infinitive (see §13.2.7 and §14.5.2).

*thaní* **'called, said' (labelled quot).** Although the verb *thané-* 'call, say' does occur (but rarely) in other forms, it has become grammaticalised as a converb, *thaní*, and as such has come to function as a quotation marker (see §10.2.4) and may potentially develop into an auxiliary or clitic with an even more restricted syntactic distribution.

*maní* **'it has been told' (labelled hsay).** Like *thaní, maní* is a converb, related to the verb *mané-* 'say', and has in this particular form become grammaticalised as a hearsay marker (see §10.2.4). It is also distributionally more restricted and even more specialised than *thaní*, and can therefore alternatively be described as a special type of auxiliary or verb clitic. It can also be said to have become enough separate from the "regular" use of the verb *mané-*.

Although not necessarily analysable as verbs any more, two other important modality words or markers should be mentioned:

*heentá* **'would, might, were' (labelled condl).** Invariable in this form, it occurs in constructions primarily expressing conditionality and is therefore as much to be regarded as a subordinate conjunction (functioning like 'if'-'then') as a verb form. See §14.4.4.

9 Verbs

*seentá (B síinta)* **'should, shall, will' (labelled condh).** Like *heentá* it is formally invariable, occurring in Conditional constructions, where the connection between cause and result essentially is factual and temporal (functioning similarly to 'when'-'then'). See §14.4.4.

### **9.4 Inflectional categories**

Verbs are primarily inflected for tense and argument agreement, whereas aspect in the present analysis is considered as already included in a perfective or imperfective stem. The stem itself may be analysed as composed of a base and a valency specification (see §9.5). The focus in the rest of this chapter (§9.4-§9.6) is on verb forms (whether formed inflectionally or derivationally), whereas verbal categories as defined functionally (capitalised throughout this work) is the topic of §10. Some of the latter (such as the TMA-categories Future and Present) directly correspond to inflectional categories (§10.1.1-§10.1.4, §10.2.1, §10.2.3, and §10.3), others (such as Perfect and Past Imperfective) are expressed periphrastically, i.e. outside the actual system of inflectional morphology (§10.1.5-§10.1.8, §10.2.2, and §10.2.4).

Argument agreement occurs immediately outside the stem, except for when it occurs subsequent to inflectional tense. Two different kinds of agreement are part of the paradigm, person agreement, and gender/number agreement, the former confined to the agreement marking directly attached to the imperfective stem. Gender/number agreement occurs twice in the case of the Perfect, as main verb inflection as well as auxiliary verb inflection.

The singular imperative is basically an accented form of the verb stem, whereas the plural is formed with a unique suffix, without any direct parallels in the agreement system at large. Apart from finite inflectional categories, there are a number of important non-finite forms. Table 9.25, with *til-* 'walk' as an example verb, gives an overview of verbal inflectional categories, including finite as well as nonfinite forms.

### **9.4.1 Agreement morphology**

As mentioned above, two different types of argument agreement are present in the language: a) person agreement, and b) gender/number agreement. A similar situation where different sets of agreement markers are being used in one and the same language is not at all uncommon in IA languages, where the "primary" person forms almost always descend from the OIA present active – although not

### 9.4 Inflectional categories


Table 9.25: Verb forms (*til*- 'walk')

necessarily every single suffix as such – while the "secondary" gender/number forms are adjectival and connected with verb forms built on participles, which have only later come to function as finite verbs (Masica 1991: 259–260).

While person agreement is always with the intransitive subject or the transitive agent, the gender/number agreement follows an ergative pattern in the perfective and an accusative pattern in the imperfective. For details on agreement patterns, see §12.1.

### **Person agreement**

Person agreement (Table 9.26) occurs with the non-tense marked imperfective stem, and more specifically with the Future and the Past Imperfective categories. All person-agreement suffixes added to the stem, except the 1pl, are accent-neutral.

The consonant segments in 1sg, 2pl and 3pl are fully recognisable from the ancient present ending (*-āmi, -asi, -ati, -āmas, -atha, -anti*), and are also, including 3sg, somewhat expected considering the general diachronic loss of final syllables. 9 Verbs

Table 9.26: Person-agreement suffixes


The Palula 2sg and 1pl, however, remain largely unexplained. Similar person agreement suffixes are found in other Shina varieties, such as those of the Kohistani Shina subjunctive: *-am, -ii/-ee, -ee, -ooṇ, -at, -an/-en* (Schmidt & Kohistani 2008: 114).

Due to interaction with a final stem-vowel, however, whether synchronically realised or underlying, the actual future endings may take a number of different shapes, as can be seen in Table 9.27, but they all go back to the same basic suffixes, and there is therefore in a strict sense only one "conjugation" (Masica 1991: 261) in Palula.

Table 9.27: Person-agreement allomorphs


The first form displayed in each cell in Table 9.27 is the unassimilated suffix following a consonant-ending verb stem. The second form displayed is the person-suffix fused with the final vowel of an *a*-ending stem, and the third form is used with *e*-ending stems. 1sg uses the second form for *a*-ending and *e*-ending verbs alike, and 1pl uses the first form for consonant-ending and *e*-ending verbs.

Verb stems ending in *u* make use of the third ending for second and third persons, and the second ending for both 1sg and 1pl. That is also the case with the *áand*-verbs, except that there is a special 1pl form *-éea* or *-éya* in the B dialect with these verbs (as in *nikh-éea* 'we will appear' and *y-éya* 'we will come'). The

### 9.4 Inflectional categories

Singular Plural m *pil-áan-u* 'I, you, he, it (msg) am/are/is drinking' *pil-áan-a* 'we, you, they (mpl) are drinking' f *pil-éen-i* 'I, you, she, it (fsg) am/are/is drinking' *pil-éen-im* 'we, you, they (fpl) are drinking'

Table 9.28: Gender/number agreement with the present

verb *the-* 'do' shows some variation in its 1pl, *th-íia* or *th-áaya*, and there is similar variability in the paradigms of the *i*-ending verbs; a 1sg form *uṛ-íim* has, for instance, been noted in the B dialect for the verb *uṛí-* 'pour'.

The deviant pattern of suppletive *be-* 'go' has already been displayed in Table 9.23, with all its person agreement forms included.

### **Gender/number agreement**

Gender/number agreement occurs with the perfective stem and with present tense. All these categories are historically participial categories, hence also referred to as adjectival agreement (Masica 1991: 260). The interaction of gender and number results in four possible agreement suffixes, all familiar from noun and adjective morphology. The suffixes are the same, but with the present (Table 9.28) there is always the additional property of vowel alternation (umlaut) occurring inside the preceding present tense suffix, when the agreement suffix includes a high front vowel (as in the two feminine suffixes).

In some sense, the *-m* of the feminine plural could be seen as a plural marking added to the (for plural otherwise unmarked) feminine *-i*, as a more peripheral layer, thus being a "tertiary" (later added<sup>7</sup> ) element to use Masica's (1991: 260– 261) terminology. Its use is indeed also less stable or optional (as compared to the non-optional opposition msg vs. mpl vs. f). It has a direct parallel in the nasal element – marking feminine plural – in Urdu-Hindi verb morphology. Its origin is probably to be found in noun morphology rather than in adjective morphology (see §5.6.3).

With the perfective, the occurrence of such vowel alternation in the preceding segments of the word varies between different verb classes. With the most com-

<sup>7</sup> Compare the perfective number/gender forms of 'go' (Table 9.23), which lacks this four-way contrast.

### 9 Verbs

monly occurring L-verbs with a consonant-ending stem the agreement suffixes are the sole reflexes of agreement (Table 9.29).

However, accented *áa* and *óo* (in B *úu,* historically derived from *áa*) in the perfective stem, as in Table 9.30, are subject to vowel alternation (umlaut), just parallel to the present verb forms.

In some of the periphrastic categories agreement occurs twice, shown in Table 9.31, first in the main verb, then also in the auxiliary (although there is a strong tendency for at least one *-m* in the fpl agreement forms to be dropped).

### **9.4.2 Verb forms derived from the imperfective stem**

**Imperative.** All verbs form distinct singular and plural imperatives, for example *čhín* (sg) – *čhínooi* (pl) 'cut!' More examples are provided in Table 9.32. The imperative singular is in the typical case identical to the imperfective stem, and it always carries the accent on its last syllable. For nouns with a consonantending stem, the imperative singular is always the same as the stem, but with additional lengthening of an accented stem-vowel *á* to *áa* (in the A dialect only), and strengthening of an accented stem-vowel *áa* to *óo* (in B *úu*), and *ée* to *íi*. In the *e*-ending verbs, the underlying final *é* has (in the A dialect) become *á* in imperative singular and are thus formally identical to the imperative of *a*-ending verbs.

The imperative plural is invariably formed with a suffix *-ooi* (in B *-uui*). This suffix receives the accent with vowel-ending verb stems (replacing its final vowel with the imperative suffix), whereas consonant-ending stems keep the word-accent on the stem-vowel, often with a subsequent weakening of the suffix to *-oi* /-oj/ (B *-ui* /-uj/). See §10.2.1 for a description of the Imperative as a TMA-category.

**Future.** The tense-unspecified imperfective, is always realised as the imperfective stem directly followed by one of six person agreement suffixes (see §9.4.1 and Table 9.26), and has without any further specification a (primarily) future meaning in Palula, while it forms the Past Imperfective when used together with the auxiliary *de*.

Historically this is the old present tense, which in the modern language has become confined to the non-present imperfective realm. While in many IA languages the old present has been driven out by newer formations and been left with less central functions, some of them with vague future or subjunctive meanings, it is especially in the north-western IA languages where it has come to func-


Table 9.29: Gender/number agreement with the perfective

Table 9.30: Vowel alternation related to gender/number agreement


Table 9.31: Double gender/number agreement


### 9 Verbs


tion as the future *per se* (Masica 1991: 288). It is interesting, however, that these forms seem rather marginal even in the closest relatives of Palula, namely Sauji and Kalkoti. In Sauji, they are used as subjunctives, but very infrequently, and only the 1sg forms *-um/-om/-aam* correspond with any greater precision with the Palula forms. The form *-iyee* is used for 3sg, 1pl and 3pl alike, and most closely resembles the 1pl of Palula. On the other hand, a particular verb form built up with the 1sg subjunctive, an *n*-element and a gender/number agreement suffix (giving *dumnoo, dumnee, dumni, dumne* 'will give') is used as a future in Sauji. This morphological construction is obviously a Sauji-specific innovation; it is very likely that the combination *n* + gender/number agreement is somehow related to the copula *(hinoo, hinee, hini, hine): dumnoo < dum hinoo*. For the Past Imperfective (§10.1.6), the other instance where the imperfective stem + person agreement forms are used in Palula (with the past tense marker *de*), Sauji uses a construction with the present imperfective verb form (*-aan*) followed by a suffigated form of *aalo* 'was (msg)' or one of its gender/number alternants: *thaanaloo* 'was doing', etc. The ancient agreement pattern (in person) has therefore been given up almost exclusively in favour of the new gender/number pattern in Sauji (Buddruss 1967: 46–54).

The near absence of any forms related to the Palula future in Kalkoti tends to point in the same direction. Elicitation of future propositions tends to produce the same imperfective verb forms as for most present tense propositions. I only have a few examples where some of the forms with person agreement are used: 1sg *-am/-um* (as in *ma ɡuwaa th-am* 'What should I do?'), 2/3sg *-ä*, and 3pl *-ään*, rather closely corresponding to the Palula 1sg, 3sg and 3pl forms, respectively. Although the form sharing between the 2sg and 3sg forms of the aorist is also found in Kohistani Shina (Schmidt 2004a: 39), I would rather assume a loss of con-

### 9.4 Inflectional categories

trast, probably as the result of sound changes in a more distant past in Kalkoti (and Kohistani Shina), than to suggest a split into 2sg and 3sg in Palula, although the actual form of the modern and somewhat "mysterious" 2sg in Palula is an innovation peculiar to this Shina variety. See §10.1.2 for a description of the Future as a TMA-category, and §10.1.6 for a description of the Past Imperfective.

**Present.** The frequently used present tense is regularly formed with a suffix *–áan* (see Table 9.28), which invariably carries the accent.

The origin of this element is not an entirely uncontroversial issue. Both Morgenstierne (1941: 22) and Buddruss (1967: 48) clearly state that the *-áan* of Palula as well as the virtually identical element in closely related Sauji goes back to the OIA present active participle *-ant* or *-antaka*, even this with numerous parallels (involving various degrees of reduction) in other NIA languages (Masica 1991: 270–271). What complicates the picture is the small class of verbs in Palula, already mentioned, that form their present tense not with *-áan*-agr but with *-áand*-agr. Of course, we may decide that these are a few residual forms, occurring only with a group of high-frequency motion verbs, which for some reason or another have resisted a change affecting all other verbs (further supported by the fact that the form is missing altogether in Sauji and Kalkoti), but a question of separate origins is equally justified. We have in any case no difficulty in showing a development *-ant > -áand*, as it is entirely parallel to other examples of *a*-lengthening and voicing in the language (compare with OIA *vansantá > \*basant >* Palula *basaánd* 'spring'). However, if we want to propose that the form *-áan* is merely a reduced form of *-áand*, we would want to explain why neither *dáanda* 'teeth' nor *páanda* 'path' (both going back to forms containing an OIA *ant*-segment) have been similarly reduced to *\*\*daana* and *\*\*paana* in the modern languages. We also have a particular problem posed by the imperfective/present verb forms of Kalkoti with three different gender/number allomorphs: *-uun, -iin, -aan*. I see it as very unlikely that the different vowel qualities would be straightforward examples of umlauts triggered by a now lost final vowel of an agreement suffix. Although umlaut is a feature of the historical development of Kalkoti, we do not have any parallel cases where *a* or *aa* has developed into *uu* or *ii* elsewhere in this variety. Another possible origin is in a gender/number agreeing auxiliary *hino* or *hano*, etc. 'is' being attached to the aspectually unmarked verb stem (as suggested by Ruth Laila Schmidt, pc).

This historical participial form, regardless of its exact origin, has probably entered the TMA system as an aspectual marker with a rather limited imperfec-

### 9 Verbs

Table 9.33: Present formation with a-ending L-verbs


tive – probably progressive – use, but has steadily gained ground within the imperfective realm, marginalising the former present tense to the non-present imperfective, thus establishing itself as the sole marker of present-tense reference. An argument for regarding this as primarily a tense category and the suffix as a present-tense marker and not an aspect marker is that the unmarked imperfective as well as the perfective verb form can be further specified with tense auxiliaries (see §10.1.5), whereas this is normally not the case with the present.<sup>8</sup>

The suffix *-áan* never appears as the final segment of a verb, but to it is always added a number/gender suffix, agreeing with the subject argument (see §9.4.1). The present tense inflection undergoes an umlaut-process when one of the feminine agreement suffixes (with a high-front vowel) is attached, with the form *-éen* as a result. As already pointed out above, a handful of verbs use the present tense inflection *-áand* instead, and even undergo umlaut formation with the feminine suffixes resulting in the forms *wh-éend-i* 'she is coming down', *wh-éend-im* 'they (pl) are coming down', etc. In some speech varieties (particularly within the A dialect), a third variant is presenting itself in one of the verb classes, as the result of interaction with the final vowel of the *a*-ending verbs, potentially creating a new causative present-tense marker, as shown in Table 9.33. The temporal versus aspectual character the TMA-category Present, using this particular verb form, is further discussed in §10.1.3.

<sup>8</sup> There are a few exceptions in my data where the present-tense auxiliary is used with the present verb form, but it is very unusual and perhaps only used when the verb form is interpreted as a participle rather than as a finite verb. The past auxiliary *de*, however, is considered entirely ungrammatical following this form.

### 9.4 Inflectional categories


**Obligative.** This chiefly modal category has an invariant suffix-accented form *-aiṇḍeéu* (in the A dialect mostly pronounced *-eeṇḍeéu*), added to basically any verb stem (Table 9.34). It seems that it in most cases simply substitutes for the vowels in vowel-ending stems. This exemplifies a verb form that seems to "fall between two stools" as far as finiteness is concerned (see §10.2.3 for a further discussion). Schmidt's (2003: 139) "injunctive" in Kohistani Shina, with its invariant form *-óoṇṭha* and its similar semantics, is probably related to the Palula obligative.

**Infinitive.** The infinitive is formed with *-áai* (in B *-ái*) 9 , leaves the stem unaccented, and except for in a few cases (stems with final *u* and in Biori *bayái* 'to go' and *khayái* 'to eat') replaces the final stem-vowel (see Table 9.35). It appears to be in free variation with the form *-áa*.


Table 9.35: Infinitive formation

<sup>9</sup> It could be argued that it is altogether deaccented, either functioning as a clitic or together with the next morpheme constituting a phonological word.

### 9 Verbs

It is, however, still somewhat doubtful whether what I have labelled here an infinitive really is to be considered an independent verb form and a free morpheme rather than a secondary stem formation (see §10.3.6).

**Converb.** Undoubtedly the converb (or the "conjunctive participle" as it often is called when referring to IA languages) is the most frequent and important nonfinite form, not only in Palula but in IA languages in general (Masica 1991: 323). There are two frequently occurring and regular forms of the converb, an accented suffix *-í* added to consonant-stems, *e*-ending stems and (replacing the final vowel of) *i*-ending stems, and an ending *aá* replacing the final vowel of *a*-ending verb stems. Verb stems with other vowel endings (*su-* 'sleep' and *ru-* 'cry') have gone through assimilation to *-eé*. However, a few verbs (the *áand*-verbs, see above) keep a vowel-segment of the stem and add *-í* (without assimilation or forming a diphthong). In addition, a small group of verbs (*the-*<sup>10</sup> 'do', *de-* 'give', *ǰe-* 'hit', *be-* 'go', *bhe-* 'become') have a converb identical to the stem, i.e. with a short *-e*. Examples are provided in Table 9.36. For further examples and a description of Converb uses, see §10.3.1, §14.3.1, and §14.4.1.


Table 9.36: Converb formation

**Copredicative participle.<sup>11</sup>** This participle seems to be invariably formed from the verb stem with an accent-bearing suffix *-íim* (Table 9.37). It is formally reminiscent of the instrumental noun suffix, with which it has obvious semantic parallels.

<sup>10</sup> The converb of this verb, *the*, is under certain circumstances lengthened to *thée* (especially as lexicalised with a meaning approximately corresponding to 'then', often beginning a new utterance, referring back to the immediately preceding context, often along with a postposed

### 9.4 Inflectional categories


Table 9.37: Copredicative participle formation

Due to its adverbial character (see §10.3.5 and §14.4.1 for details), there are rather severe semantic restrictions on what verbs can occur as copredicative participles. For further examples of the use of the Copredicative Participle, see §10.3.5 and §14.4.1.

**Verbal noun.** The verbal noun is probably in some sense a secondary verb form, based (at least in the A dialect) on the infinitive to which the invariable suffix *-nií* is added (Table 9.38). This nominalising suffix always carries the accent, while the infinitival part is deaccented (and its vowel subsequently shortened).


Table 9.38: Verbal noun formation

As a non-finite verb form with noun-like qualities, it may inflect for case and perhaps also for number (although there is no strong evidence for that in my data). For further examples and a description of the various uses of the Verbal Noun, see §10.3.3, §14.4-§14.5, and §14.6.6.

clitic *ba*: *thée=ba* 'And then…').

<sup>11</sup> See Haspelmath (1995: 17-20) for a more precise definition.

9 Verbs

**Agentive verbal noun.** The agentive verbal noun is, like the verbal noun, a nominalised form, but with a more restricted application than the latter (see §10.3.4). It is formed with a suffix *-áaṭ* (or its umlaut variant *-éeṭ*) followed by a gender/number suffix.

### **9.4.3 Verb forms derived from the perfective stem**

**Perfective.** This is (when used alone) the single most frequently used verb form in narrative types of discourse, mainly occurring in its Simple Past function (see §10.1.3). With an added auxiliary, *hín-* 'is' it forms the periphrastic category Perfect (§10.1.7), and with the auxiliary *de* 'pst' it forms the Pluperfect (§10.1.8).

The most common perfective stems end with an *l*-element, also being the defining feature of the class referred to as L-verbs. The final syllable of the stem (which alternatively could be analysed as a perfective suffix) is always accented and thus carries the accent of the whole word. Like the present-tense inflected stem, the perfective stem must be followed by a number/gender suffix (see §9.4.1 and Table 9.29), agreeing with the intransitive subject or the transitive direct object.

The ending elements of the L- and T-based perfective verb forms are obviously quite old. The perfective forms belonging to the class of T-verbs with a discernable *t*-element, sometimes realised as *d* or *ṭ*, most certainly derive from an OIA past (passive) participle *-ta* (Whitney 1960 [1889]: 952), representing an early development of a perfectivity category, contrasting initially with an aspectually unmarked plain verb stem. This element has a number of parallels in other NIA languages (Masica 1991: 269, 272). The other perfectivity marking element, with *l*, is a more recent development and can be traced back to the Prakrit *-illa* (Ruth Laila Schmidt, pc); outside Shina, it mostly occurs in NIA languages in the eastern and southern parts of the subcontinent (Masica 1991: 270). The T-forming class is clearly a kind of residual and closed category in Palula as well as in its closest-related varieties (particularly in Sauji and Kalkoti), and it is limited in number and productivity, whereas the paradigm of the L-forming class (and its subclasses) has become the normative or system-defining structural property (McMahon 1994: 104) for verbs, evidenced by the inclusion of rather recent (although few and far between) loans into this class, such as B *newešíl-* 'wrote' (from Khowar *niweš-*). The class as such has been subject to much more levelling and innovation as compared to the older T-class (Ruth Laila Schmidt, pc).

The non-finite Perfective Participle (see §10.3.2) is formally identical to the finite Simple Past (Table 9.29) to which it is diachronically related. Perfective Participles are essentially adjectival but may also function as nominal forms and therefore be additionally inflected for case. These are distinguished from the 9.5 Valency-changing morphology

perfective used for finite Simple Past only by their distribution, and when they happen to be inflected like nouns.

### **9.5 Valency-changing morphology**

Although mechanisms involved in stem-derivation are derivational rather than inflectional, the area of valency is a morphologically important and complex issue in Palula like in many other IA languages (as pointed out by Masica 1991: 315). The two primarily morphological processes of valency reduction and valency addition, each with its unique suffix, interact with transitivity, causativity, and in a more limited sense with voice. Palula verb stems come with, or are normally reserved for, a certain degree of valency – they are either intransitive or transitive – but this basic valency can be changed morphologically, resulting in derived stems with either an increased valency (§9.5.1) or a reduced valency (§9.5.2). Of the two, the former process seems synchronically slightly more productive and more frequently occurring in the language.

### **9.5.1 Valency addition**

The productive (and regular) way of increasing or adding valency is by adding an accented suffix *–á* or *–awá*<sup>12</sup> to another (imperfective) verb stem, as can be seen in Table 9.39. The verbs derived in this way all belong to the *a*-ending Lclass. The valency-increasing suffix added to an intransitive imperfective stem *(núuṭ-)* thus derives a transitive imperfective verb stem *(nuuṭá-)*, and when added to an already transitive stem *(the-)*, a causative verb is derived *(thawá-)*. As this process is a matter of stem formation, any inflectional morphology, as described above, occurs after the derivational suffix: *pašáan-a* 'are seeing' > *pašawa-áan-a* /paɕaˈwajna/ 'are showing', as is also apparent from the examples below.

For the perfective stems, which were analysed as including, e.g., an *l*-element, valency addition takes place before that element (thus within the stem itself), compare with the stems in examples (3) and (4).


<sup>12</sup> Which really can be understood as a secondary or doubled causative *-á-á*.

### 9 Verbs


Table 9.39: Regular valency addition

There are also some non-productive (or irregular) derivations (Table 9.40), reflecting historical or alternative patterns, with root vowel alternation, aspiration alternation, and a derivative segment *-aal*. The alternation *már-/mhaaré-* is related to the OIA causative *-áya*, which was accompanied by a strengthened grade of the root (Masica 1991: 316–321). While the causative suffix itself was reduced to *-ē* already in MIA (which the underlying *-e* in the *e*-ending L-verbs may be descended from), the root vowel alternation survived, at least in a few verbs. The more productive *-á* and *-awá*, on the other hand, most probably go back to the OIA causative pseudo-allomorph (to borrow Masica's term) *-āpaya*, which in Sanskrit became a productive form of the causative. It has since eroded substantially, phonetically speaking. The history of the segment *-aal* (occurring in a few verbs) is much less certain. It may have a connection with the regular causative marker *-ar* in other Shina varieties (Radloff & Shakil w. Shakil 1998: 26) and/or the irregular causative formation of 'bring up' from 'come up' with *-l* (*ukāɡ > ukālũɡ*) found also in Kohistani Gawri (Baart 1999a: 88); there are also a few causatives in Urdu-Hindi (Schmidt 1999: 87) and in Siraiki (Shackle 1976: 74) that are formed with a suffix containing an *l*-element.

Table 9.40: Irregular valency addition


### 9.5 Valency-changing morphology

In the examples above, the semantics is not radically altered in the process. That, however, is not always the case, and the respective stems have sometimes, since being derived from others lived their own lives, so to speak, shifting semantic focus or developing secondary senses. Some have obviously drifted apart more than others, as can been seen in Table 9.41.

Table 9.41: Valency addition and semantic shifts


As a great number of verbs in the *a*-ending class are either inherently transitive (without any apparent intransitive counterpart that it has been derived from) or are derived transitives, the final *-á* has in itself become a transitivity marker of some sort. The causative or valency-increasing formation, as described above, can apply to intransitive and transitive, as well as stems already derived by the valancy-increasing formation.

### **Transitives derived from intransitive verbs**

First, the formation can be applied to intransitive verbs. Although this, strictly speaking, is not part of our focus here, we can see that the same construction is used regardless of the degree of voluntary control that the causee exercises: compare (5) with (6), (7) with (8), and (9) with (10).


<sup>13</sup> The form without *-u-* has been noted for Biori Palula.

9 Verbs


While all the verb stems thus regularly derived causatively end up in one and the same morphological class (Table 9.42), namely the *a*-ending L-class, the corresponding intransitive verb could come from virtually any verb class, although the most common source is the consonant-ending class of L-verbs.


Table 9.42: Valency addition and verb class membership

### **Causatives derived from transitive verbs**

The same process of valency addition can also be applied to transitive verbs. Here we start out with a verbal event involving two participants (as in (11)), a subject 9.5 Valency-changing morphology

and a direct object. Through the process of causation we end up with a verbal event involving three participants (as in (12)): a) a causer, b) a manipulee (the person actually carrying out the verb act on behalf of the ultimate causer), and c) a direct object. The manipulee is marked as such with a special marker, a grammaticalised converb of *ṣaawá-* 'dress, turn on'.


By the same process some transitive verbs can also be made into the equivalent of ditransitive verbs, like *pilá-* 'feed, give to drink (especially used with children and domestic animals as indirect objects)' in example (13), from *pil-* 'drink'.

(13) *aníi* 3sg.prox.obl *wée* in *wíi* water *hín-u* be.prs-msg *činaaróom* Chinarom *aniniaám* 3pl.prox.acc *wíi* water *keéna* why.not *pil-a-áan-u* drink-caus-prs-msg *thaní* qot ' "Here is water, why not have them [the goats] drink in Chinarom", he said.' (A:PAS070)

### **Second causatives**

It is also possible to form so-called second causatives, with a doubled causative suffix *-awá (-á + -á* with an epenthetic *-w-* in between), as in (14). Primarily its function is to derive causatives from derived transitives. Here the person uttering the sentence caused or manipulated his son to make the guests drink tea, thus the initiator of the action is one step further removed from the action as compared to (13), making the verb in (14) a four-argument verb.

(14) *míi* 1sg.gen *teeṇíi* refl *putr-á* son-obl *ṣaawaá* manip *teeṇíi* refl *práač-am* guest-pl.obl *čái* tea *pil-a-wéel-i* drink-caus-caus.pfv-f (< *pil-á-á-íl-i*)

'I had my son give the guests tea to drink.' (A:HLE2506)

### 9 Verbs

### **9.5.2 Valency reduction**

Although seldom as elaborate as valency addition, the corresponding valencydecreasing process is very similar. That, too, is regularly carried out by a suffix, in this case an accented *-íǰ* added to the verb stem (or replacing the final vowel of an *a*-ending verb stem), as can be seen in Table 9.43. The verb stems so derived are without exception part of the consonant-ending L-class, and a majority of the corresponding non-derived verbs originate in the a-ending L-class.


*de-* 'give' > *da-íǰ-* 'be given'

Table 9.43: Regular valency reduction

The valency-decreasing suffix added to a transitive stem *(bilá-)* thus derives an intransitive (or passive) verb *(bilíǰ-)*. Again, like with valency addition, any inflectional morphology occurs after the derivational suffix: *paš-áan-a* 'are seeing' > *pašiǰ-áan-a* 'are seen', etc. (Compare the transitive (perfective) verb stem *ḍhanɡóol-* 'bury' in (15) with its corresponding valency-descreased stem *ḍhanɡiǰíl-* 'be buried' in (16)).


'160 years later she was buried by my hand.' (A:PAS133)

Historically (Masica 1991: 316–317), the *íǰ*-suffix undoubtedly goes back to an OIA passive in *-yá*, which in some MIA dialects was phonetically strengthened to *-iǰǰa*. A similar form is found in a number of modern IA languages, e.g., *-íiǰ* in Gilgiti Shina (Radloff & Shakil w. Shakil 1998: 116). There are also a number

### 9.6 Complex predicates

of intransitive verbs in Palula with a *ǰ*-element which may originally have been derived from transitive verb stems, but for which the valency-reduced member is now either the sole survivor, or has due to a semantic shift lost its productive connection with a transitive counterpart. There is no valency-reducing process comparable to the doubled causative suffix.

### **9.6 Complex predicates**

Two other constructions (which were alluded to in the beginning of this chapter) that in some sense also may be described as derivational, although not morphological in a formal sense, are the "conjunct verb" and the "compound verb". Although two separate phenomena, they are similar in that they are both typically (but not exclusively) Indo-Aryan, and both are complex constructions in the sense that they use "building material" from more than one lexeme and still in many ways function lexically like a single verb stem. The two terms are in themselves confusing and prone to be mixed up, and possibly it would make more sense if they were swapped for one another (as already pointed out by Masica 1991: 326), but in order to avoid adding even more confusion to the discussion, I will primarily use what has become – more or less – standard terminology in South Asian linguistics, but also under each heading mention and discuss some alternative terms that are also widely used, some of them for the same or closely related phenomena elsewhere in the world.

### **9.6.1 Conjunct verbs**

*Conjunct verbs* are complex constructions that function as lexical units, usually consisting of a verb preceded by a noun or an adjective (but also words from other parts of speech are possible). There are also some cases where it is not obvious to what part of speech the non-verb component (which will be referred to as a *host*) belongs, sometimes because it does not occur in the language outside of this particular conjunct verb construction.

The verb in such a construction comes from a small set of verb stems that I will refer to as *verbalisers*. This verbaliser does not contribute to the construction with much more than the general "verbness" (including the ablility to be inflected), whereas it is the host that tends to contribute the main semantic content to the complex. The verb is here merely a "dummy" without any semantic weight of its own. Some conjunct verbs formed through such host-verbaliser

### 9 Verbs

combinations are exemplified in Table 9.44. Usually the host element occurs immediately preceding the verbaliser, as can be seen in examples (17)–(18).


Table 9.44: Derivations of conjunct verbs


The three verbalisers exemplified in the table, 'do', 'give' and 'become', are the only ones that seem to be used productively to construct conjunct verbs (although there are a few infrequent combinations with other verbs that come close). It is, however, no mere coincidence that we find these particular verbs in the "dummy" role; as we shall see, they are functionally precisely what it takes.

So, what is the function? First, it is a derivational process, deriving verbs from other parts of speech, primarily – but not exclusively – from nouns. A wide range of phenomena that are lexicalised as nouns, adjectives or something else can thus easily be depicted as events in Palula, as illustrated in examples (19) and (20).

(19) *be* 1pl.nom *teeṇíi* refl *maǰí* inside *ǰarɡá* council *th-íia* do-1pl 'We will discuss it among ourselves.' (A:MAR015)

9.6 Complex predicates

(20) *kuṛíina* woman.pl *ta* cntr *támbul-am-ii* drum-pl.obl-gen *ǰ-íin* beat-3pl *rhoo-á* song-pl *d-íin* give-3pl 'The women were beating the drums and singing.' (A:JAN034)

That is further achieved in an economical way, limited strictly to only a few inflectional paradigms. An extended advantage of this "verbal flexibility" is the relative ease with which entirely new verbs can be made or brought into the language, sometimes as totally novel constructions, sometimes as calques of conjunct verb constructions in languages of wider communication. This does not necessarily mean that a new verb will be made just to fill a gap or to replace one that is already there; instead these constructions could elaborate on or refine the meaning of other verbs. We also find, perhaps not surprisingly, a substantial number of loan words (primarily from Urdu or Pashto) in the host slot of conjunct verbs, such as *daawát* and *muqaabilá* in (21).

(21) *deeúl-ii* Dir-gen *xálak-am* people-pl.obl *daawát* invitation *dít-i* give.pfv-f *atsharíit-am* Ashreti-pl.obl *the* to *ki* that *muqaabilá* contest *th-íia* do-1pl 'The people from Dir invited the the Ashreti people to have a competition

with them.' (A:CHA001)

But also lexical material from a structurally and culturally very different language like English can thus be made into "new" verbs (usually via Urdu, but people with direct access to English have no problem applying it on the spot), as exemplified in (22) and (23).


(A:CHN070110)

Having this type of mechanism, we need to answer the question why Palula needs three different "dummy" verbs when one would be the most economical.

### 9 Verbs

The answer lies in what the verb contributes in addition to its "verbness", namely valency: *bhe-* 'become' provides an intransitive frame, *the-* 'do' a transitive, and *de-* 'give' a transitive frame with place for an oblique object.

There are, however, some complexities connected with argument structure, and to what extent the host element takes part in it; this will be discussed further on (see §13.2.8). This, and related issues, is also described at greater length in a separate article (Liljegren 2010).

### **9.6.2 Compound verbs**

The other strategy for expanding the verb lexicon, commonly found in IA languages, uses two verb stems, a construction sometimes referred to in South Asian linguistics (Masica 1991: 326) as the *compound verb* construction. The first verb stem (verb1), occurring in a non-finite form, carries the main semantic content, while the second (verb2), drawn from a small set of verbs, carries the inflections as a finite verb. The second verb is usually subject to semantic bleaching but without being fully grammaticalised. The second component (verb2) has variously been referred to as a *vector*, an *intensifier*, an *operator* and an *explicator*. I will use *vector*. This is certainly not a verb derivation in the same sense as the *conjunct verb* construction (although for instance Butt (1993: 31; 2010: 49) considers the verb2 of compound verbs as well as the verbaliser of conjunct verbs as "light verbs", a particular class of verbs being used in forming complex predicates, whether the other – in IA, preceding – component in the complex is a verb, a noun or an adjective). Masica (1991: 326–330) argues that the construction primarily functions as a specification of Aktionsart as far as IA languages are concerned, whereas in other descriptions the relexicalisation taking place is being emphasised, i.e., the meaning of the complex is not really predictable from its components but must be learnt (Schmidt 1999: 143). Although a true innovation of IA languages (Masica 1991: 326; Hook 1977) in more recent times (in NIA, or possibly MIA), it has several parallels in other parts of the world (Hook 1977: 348– 349), where they have sometimes been labelled *serial verb constructions* (Ansaldo 2006), as well as in other parts of Asia (Ebert 2006: 559).

Schmidt (2004b: 20) points out that the compound verb construction is a far more marginal feature in Kohistani Shina than in Urdu-Hindi and Punjabi, and my present assumption is that the same is true of Palula, but this issue will need further research. The inventory of vectors commonly include: 1. directionals ('go', 'come'), 2. disposals ('throw', 'send', 'put aside'), 3. verbs expressing suddenness or unexpectedness ('fall', 'rise'), and 4. auto- and other benefactives ('give', 'take'). Only categories 1 and 2 seem to appear in what can be termed as

9.6 Complex predicates

possible examples of compound verbs in Palula, as shown by the examples (24) and (25), but even then relatively infrequently.


The first verb occurs as a non-finite converb, and carries the main semantic content, while the second is a regularly inflected finite verb but is semantically light when considering the meaning of the sentence as a whole. It is, however, not always easy to make a clear differentiation between possible instances of compound verbs and the regular use of a Converb, the latter in which it is the head of a dependent clause.

I am in no position to say whether this is a construction on the rise, due to contact with lowland languages where this is a conspicuous feature, or rather is one dwindling in importance and productivity. It should be noted that the construction also exists in neighbouring Kalasha, where it reinforces already morphologically expressed inferentiality contrasts (Bashir 1993: 1–4).

## **10 Verbal categories**

### **10.1 Tense-aspect categories and their functions**

There are seven frequently occurring TMA-categories in the language. Their respective paradigms are exemplified with the verb *til-* 'walk' in Table 10.1, and described in further detail in §10.1.2-§10.1.4, §10.1.6-§10.1.8, and §10.2.1. A number of additional verbal categories, some of them less frequent (§10.2.2-§10.2.4), others non-finite (§10.3), are also introduced in the chapter. All of these verbal categories are capitalised (Future, Simple Past, etc.) to indicate that they are language-specific labels, defined functionally and only partly correspond to the, noncapitalised, inflectional categories (future, perfective, etc.) that were introduced in §9.4, on the one hand, or to grammatical terms applied cross-linguistically (future [tense], perfective [aspect], etc.), on the other hand.

Tense differentiation is in Palula, as in many other IA languages (Masica 1991: 262), secondary as compared to aspectual differentiation. Whereas aspect – or to be more precise, perfectivity – as an inflectional element occurs next to the stem, followed by agreement suffixes, tense categories are, at least historically speaking, latecomers, and are still not fully part of the verb in Palula. There is possibly one very visible and unexpected exception to this rule: the Present, which occurs with a unique TMA-marking morpheme that for various reasons should be viewed as a marker of tense rather than aspect. Insofar as tense other than that is deemed relevant, it is indicated periphrastically by auxiliaries positioned after the finite verb (§10.1.5).

### **10.1.1 Basic tense-aspect categories**

As shown in Figure 10.1, there are three non-periphrastic categories that can be considered basic: Future, Present and Simple Past.

### **10.1.2 Future**

Within the imperfective, the Future category in Palula is the morphologically unmarked counterpart of the Present, but it is also the periphrastically unmarked counterpart of the Past Imperfective (see §10.1.6).


Table 10.1:TMA categories and their formations (*til*- 'walk')


Table10.1:TMAcategoriesandtheirformations(*til*-'walk'),(continued) 10 Verbal categories

Figure 10.1: Basic non-periphrastic tense-aspect categories (bold type: inflectionally realised forms; gen/num: gender/number agreement; erg: ergative alignment; acc: accusative alignment)

This category is almost exclusively used in reference to future, i.e., not yet realised, situations, as in (1) and (2), whether close or distant in time, and covers intention as well as prediction (Dahl 1985: 105–108):


'Then my father will die, either you will kill him or he himself will die.' (A:MAA013)

While Present (§10.1.3) is used for actions started but not yet seen as completed at the time of the utterance, the Future is used with reference to all future situations that have not yet commenced, such as the ones in (3), regardless of the degree of probability or volitionality. There is no other major construction in the language competing with this verb form.

(3) *ma* 1sg.nom *aní* prox *šaak-á* wood-pl *háar-um* take-1sg *ɡhooṣṭ-á* house-obl *har-í* take-cv *kúṛi* wife *anɡóor* fire *ǰheel-í* light-cv *tanaám* 3pl.acc *the* to *ɡúuli* bread *th-íi* do-3sg

'I will bring this wood to the house, and my wife is going to make a fire and bake bread for them.' (A:KIN017)

10.1 Tense-aspect categories and their functions

Future is also often part of a Conditional expression as the apodosis verb ('may eat' in (4)) and is sometimes used to express hopes and congratulations (as in (5)), i.e., optative modality (Bybee, Perkins & Pagliuca 1994: 179).


### **10.1.3 Present**

The only non-periphrastic category that is primarily a tense (and not an aspect or mood) category is the Present, neither standing in a direct morphological contrast with the Future nor with the Simple Past. Somewhat surprisingly, both from the perspective of languages in general (Dahl 1985: 103–128) and IA comparative studies (Masica 1991: 282, 288–289), Present is the only category morphologically marked for tense (at least if analysed from a synchronic perspective). It is as such one of the most frequently used verb forms and covers, within the temporal realm of the present, a range of aspectual areas.

Present typically refers to progressive situations, as illustrated in examples (6)–(7),whether durative or punctual, with present time reference.


However, it also covers habitual situations, such as those in examples (8) and (9), which can be considered repetitions on a regular basis, at or after the time of the utterance.

### 10 Verbal categories


The Present is also used for general statements, as in (10), or sayings, as in (11).


Quite infrequently, Present is used in historical narratives, especially if describing a situation from the perspective of a role character. In (12), the narrator switches from the regular narration tense (i.e., the Simple Past) to the Present.

(12) *eetáa* there *wháat-u* come.down.pfv-msg *ta* sub *tsak=tsák* red=idph *thaní* qot *áak* idef *šay* thing *wíi* water *tsak-áan-u* lick-prs-msg 'Coming down there he sees a thing making sounds licking the water.' (A:PAS037)

Present is used rather than the Future for imminent future reference, for instance, when someone is just about to do something, as when the speaker in (13) is just beginning to tell a story.

(13) *típa* now *ba* top *ma* 1sg.nom *tasíi* 3sg.gen *paalaweeṇ-íi* strong.man-gen *qiseé* story.pl *th-áan-u* do-prs-msg 'Now I'm going to tell you the story about this strong man.' (A:PAS029) 10.1 Tense-aspect categories and their functions

### **10.1.4 Simple Past**

The Palula Simple Past is primarily an aspectually defined category, referring to events and actions completed before the time of the utterance. Although it alternatively could be labelled "Perfective" (as reflected in the glossing), it is indeed restricted to past time reference (Dahl 1985: 79), and it contrasts clearly with the Past Imperfective (see §10.1.6) and other periphrastic categories that have relevance for more than one point in time, such as the Perfect and the Pluperfect (see §10.1.7–§10.1.8).

The Simple Past is used to refer to completed events in the past, whether remote or close in time. Examples (14)–(18) illustrate the use of the Simple Past, from the very remote event in (14), then gradually decreasing in remoteness in the following examples to the very close at hand reference in (18).


It is also the category used for the main story line in narratives (whether historical or fictional), referring to sequences of events in the past, such as in (19).

### 10 Verbal categories

(19) *deeúlii* Dir.gen *xálak-am* people-obl.pl *daawát* invitation *dít-i* give.pfv-f *atsharíit-am* Ashreti-obl.pl *the* to *ki* comp *muqaabilá* competition *th-íia* do-1pl *thaníit-u* say.pfv-msg *ta* sub *atshareet-íi* Ashret-gen *paalawaaṇ-aán* strong.man-pl *ɡíia* go.pfv.pl *paalawaaṇ-aán* strong.man-pl *be* go.cv *deeúli* Dir *phedíl-a* arrive.pfv-mpl 'The people from Dir invited the Ashreti people to have a competition with them. When the Ashreti strongmen heard this, they left, and arrived at Dir.' (A:CHA001-3)

Depending on the analysis, the Simple Past could also be said to include most types of conditionality. However, I have chosen to describe the Conditional under non-indicative categories (see §10.2.2).

### **10.1.5 Periphrastic tense-aspect categories**

What I refer to as periphrastic tense-aspect categories are those that are formed by a finite inflected verb and one or more auxiliaries. While the so central aspectual distinctions "at the heart of the NIA verbal system" (Masica 1991: 262) are very often made morphologically and close to the verb stem, tense and mood are normally more peripherally marked functions in most modern IA languages. This is also the case with Palula, possibly with the exclusion of the morphologically marked Present (as was pointed out above).

Interesting to note is that this may be a relatively recent (or early stage of) grammaticalisation. We find the same principle applied in the two closely related varieties Sauji and Kalkoti, i.e., the addition of elements to the basic aspectual categories to obtain tense contrasts and to form perfects and pluperfects (Buddruss 1967: 48–55; Liljegren 2009: 48–53). Partly, however, we find other elements than those used in Palula, and in some cases, the former auxiliaries have become part of inflectional morphology and even fused with aspect suffixes. In Kalkoti, for instance, an *s*-suffix (most certainly derived from the past copula *aas* 'was, were') is added to the imperfective (mostly corresponding to Palula Present) to make it past imperfective, *čuṇ-uun-s* 'was writing', and to the perfective (mostly corresponding to Palula Simple Past) to make it pluperfect, *čuṇ-i-s* 'had written'.<sup>1</sup> There is, generally speaking, a great variety among the Shina varieties in how the less central TMA categories have been grammaticalised (Liljegren 2013: 144–156).

<sup>1</sup> The final consonant of the perfective *čuṇ-il* 'wrote' has been dropped in phonological developments involving apocope and cluster reduction.

### 10.1 Tense-aspect categories and their functions

Schmidt (2004a: 38) provides examples of forms of 'be', 'go' as well as 'come' as origins of such grammaticalisations in the respective varieties.

Tense, for instance, is mainly auxiliary, occurring outside the actual verb morphology (although this "outside" is not to be taken as an absolute, as will be observed below when pointing out an ongoing grammaticalisation process). The present tense copula *hin-* in its different agreement forms is used as a present tense auxiliary (be.prs) and the (invariable) Past tense copula (alternatively past tense marker) *de* is used as a Past tense auxiliary (pst).

Supplementing Figure 10.1, we now add the periphrastic marking (at least as far as tense is concerned) to get the fuller picture of the tense-aspect dimensions relevant for the language, as illustrated in Figure 10.2. This characterisation is, however, a slight simplification, as there are also resultative categories that are not taken into account as well as some dialectal differences between A and B that will be discussed in the following sections.

Figure 10.2: Basic and periphrastic tense-aspect categories (bold type: inflectionally realised forms; *bold italics*: periphrastically realised forms; gen/num: gender/number agreement; erg: ergative alignment; acc: accusative alignment)

### **10.1.6 Past Imperfective**

The Past Imperfective is formed by means of the person-inflected imperfective verb stem (i.e., the future verb form) followed by the invariable marker *de*. It has two main functions, as a past habitual, see (20), and a past progressive, see (21).

### 10 Verbal categories


'He [an evil creature] made the child sit, and there he was taking off its clothes. When he had taken them off and put them down, he ate the child while it was crying.' (A:BRE005-6)

An indication of an ongoing grammaticalisation or (more correctly) morphologisation, by which the Past copula may become part of the verb morphology as a past-tense suffix *(d-íi-de)*, is the fact that *de* phonologically rather behaves like a clitic than an independent word, without independent word-accent. In beginning attempts at writing the language, the inflected main verb has often been observed as written together with the copula auxiliary as if it were one word,<sup>2</sup> and it has become the topic of discussion among Palula writers whether this should be done or not.

### **10.1.7 Perfect**

The most common way of expressing the perfect, i.e., a reference to a past (completed) event with current relevance, is by combining the perfective verb form with the present tense copula *hin-*. <sup>3</sup> Both the perfective verb and the auxiliary agree in gender and number with S, as in (22), or O, as in (23) and (24), according to an ergative pattern.

<sup>2</sup> This is interesting in view of the fact that the more common tendency when writing Palula with an Arabic -based script is splitting rather than lumping.

<sup>3</sup> This is at least the case in the A dialect, while there is reason to return to the situation in the B dialect.

10.1 Tense-aspect categories and their functions


An extended use of the Perfect (like in a number of other languages, Dahl 1985: 152) is inferential. In (25), it has actually stopped snowing, but it is inferred from looking at the ground on the morning after a snowfall, that it must have been snowing during the night.

(25) *kir* snow *dít-u* put.pfv-msg *hín-u* be.prs-msg 'It has been snowing.' (A:CHE070320)

The Perfect is also used, instead of the Simple Past, in some narratives, perhaps indicating the hearsay character – similar to the use of *maní* (see §10.2.4) – of the narrated content, as in (26), which is part of a narrative where the Perfect is used throughout.

(26) *so* 3sg.nom *utrap-í* run-cv *muṭ-á* tree-obl *ǰe* up *ukháat-u* go.up.pfv-msg *hín-u* be.prs-msg 'He ran and climbed up into a tree.' (A:UNF008)

There is also what appears to be a parallel perfect form, used in a very similar manner to the regular Perfect, seen in (27)–(29). This is formed by combining the converb (cv) with the present tense copula *hin-*.

### 10 Verbal categories


It is possible that the focus is on the resulting state rather than on the preceding event leading up to it in this construction, as it seems more commonly used with stative verbs and verbs with a low degree of volitionality. It may therefore make more sense to see this as a more lexically limited resultative construction (Dahl 1985: 135) than a pure perfect. In example (29), the temporal aspect is perhaps more that of a pluperfect, i.e., 'had stretched out'.

Example (30) clearly illustrates the use of this resultative constructing. Here the resulting absence of the husbands, rather than their going away, is focused. The translation 'our husbands are gone' may in fact be better than 'our husbands have gone'. Note also how it interacts with the Past Imperfective as the tenseaspect category used for the waiting of the women and children at home, and the Future is used for the hoped for outcome in the form of fresh meat being brought home.

(30) *díiš-a* village-obl *ba* top *baalbač-á* child-pl *kuṛíina* women.pl *táma* waiting *th-éen* do-3pl *de:* pst *asée* 1pl.gen *bharíiw-a* husband-pl *be* go.cv *hín-a* prs-mpl *neečíir* hunt *the* do.cv *whaal-éen* bring.down-3pl *mhaas-á* meat-pl *whaal-éen* bring.down-3pl 'In the village, women and children were waiting, [saying] "Our husbands have gone hunting. They will bring back meat".' (B:AVA218)

When used with a motion verb such as 'get up', the converbal construction brings forth the adjectival meaning 'is standing', as in (31).

(31) *ak* idef *čhéeli* goat *maidúun-a* field-obl *wée* in *du=khur-áam* two=foot-obl.pl *ǰhulí* on *uth-í* get.up-cv *hín-i* be.prs-f 'A goat has got up [i.e., is standing] on two feet in the field.' (B:SHB721)

### 10.1 Tense-aspect categories and their functions

The interesting thing, however, is that the regular Perfect as described in the beginning of this section, seems to be missing altogether in the B dialect. Instead, the converbal construction is widely used with intransitive verbs, as in the examples (28) and (30)–(31), and probably marginally also with transitive verbs, as in (32)–(33).<sup>4</sup>


'I have not written this [referring to some suspicious letters], he said.' (B:LET027)

Note that this construction, again focusing on the result rather than on the event itself, is based on a transitive imperfective stem and therefore requires an accusative agreement pattern, while the agreement with the regular Perfect, using the perfective transitive stem is ergative. But there is also another (possibly more "regular" or lexically less restricted) perfect used in this dialect formed with the converb and *heensíl-* (the perfective of the intransitive verb *háans*- 'stay, live, exist'), as in (34).

(34) *ma* 1sg.nom *khaá* eat.cv *heensíl-u* stay.pfv-msg 'I have eaten.' (B:HLE1034)

Even in the A variety, this construction (but with an added present copula) is used for hearsay, as in (35), or inferentially (by seeing water on the ground and subsequently reporting it to another person who has not been able to infer it), as in (36). However, it may still be in free variation with the regular Perfect.

(35) *so* 3msg.nom *tas* 3sg.acc *khúṭ-a* leg-obl *wée* in *čuk-í* bite-cv *heensíl-u* stay.pfv-msg *hín-u* be.prs-msg 'It [so he said] bit him in the leg.' (A:TAQ178)

<sup>4</sup> My B data, however, is too limited to the genre of narrative discourse to provide a clear indication.

<sup>5</sup> An answer to a question or an invitation to eat.

### 10 Verbal categories

(36) *aáǰ* today *peexáur-a* Peshawar-obl *muč-í* rain-cv *heensíl-u* stay.pfv-msg *hín-u* be.prs-msg 'Today it has been raining in Peshawar.' (A:CHE070320)

### **10.1.8 Pluperfect**

While the regular Perfect is the perfective with relevance for the present situation, the Pluperfect is the perfective with relevance for a past situation. This is also reflected in the formal expression: Perfect is formed with the perfective verb and a present-tense copula, and the Pluperfect is formed with the perfective verb and a past tense-copula (or past-tense marker) *de* 'was, were'.

The contrast between these is illustrated in (37) and (38). Example (37) refers to an event with present time relevance and a still visible result, a recently painted house, hence Perfect, while (38) has past-time relevance only, i.e., a house that was built but had been torn down prior to the utterance, hence Pluperfect.


In narratives, the Pluperfect is used to refer to an event taking place prior to, and thus contrasting with, the time line of the story (which normally would be related with the Simple Past), as in (39).

(39) *atáa* there *ɡíia* go.pfv.pl *ta* sub *dóodu* grandfather *ayaan=miír* Ayan=Mir *ta* cntr *ǰáand-u* alive-msg *hín-u,* be.prs-msg *xaamaár* dragon *ba* top *mheeríl-u* kill.pfv-msg *de* pst

'When they came there, grandfather Ayan Mir was alive, but the dragon had been killed.' (A:AYB044)

There seems to be a drift towards a use of Pluperfect as a general remote past, not an uncommon development generally (Dahl 1985: 147), and in the region in particular. However, that is not to say that the two constructions – Simple Past and Pluperfect – would therefore be in totally free variation when referring to remote past, although I am not able to pinpoint exactly where the dividing line

### 10.1 Tense-aspect categories and their functions

goes. The utterance in (40) is an answer to the question whether the speaker had met the other person recently, and it could perhaps be that it is the temporal adverbial phrase that somehow favours the Pluperfect, but this is so far only speculative on my part.

(40) *áak* one *haftá* week *muṣṭú* back *dhríṣṭ-u* see.pfv-msg *de* pst '[I] saw him a week ago.' (A:CHN070110)

Another possible indication of which category to use seems to be related to whether or not a measurable chunk of time (exactly how long is probably not that important) can be considered to have elapsed between the event – or the end-point of the event if durative – and the time of reference, as contrasted with the event coinciding too closely with the time of reference. This again points to the Pluperfect as developing a secondary sense of temporal remoteness. This can be seen when comparing examples (41) and (42) (both part of the questionnaire prepared by Dahl (1985) and applied to Palula), where the former uses the Simple Past and the latter the Pluperfect.


However, it is also possible to describe the difference as being one of perspective. In (41), the utterance can be said to describe what took place during the time before my arrival, whereas the utterance in (42) describes the resulting situation at my arrival.

Again, although this particular form of the pluperfect does occur in the B dialect, it seems markedly less frequent than in the A dialect. There is also another construction in B, illustrated in (43) and (44), somewhat parallel to the converb+*heensíl*-construction above, but with an additional past-tense copula,

### 10 Verbal categories

which certainly shares at least some of the semantic characteristics of the regular Pluperfect in A (the phonologically reduced alternative form *khasúlde* in (44) may point to an ongoing grammaticalisation).


### **10.2 Non-indicative finite categories and their functions**

Not all of the following categories are equally central or to the same extent grammaticalised, but they are all somehow part of the modal realm within the TMA system and thus deserve to be mentioned together in one place.

### **10.2.1 Imperative**

The bare imperative, in its two forms, imperative singular, as in (46) and (47), and imperative plural, as in (45) and (48), is mainly used for commands, as in (45), and not overly polite requests, as in (46)–(48).


10.2 Non-indicative finite categories and their functions

(48) *man-éen-i* say-prs-f *ki* comp *ma* 1sg.nom *patú* after *yh-óoi* come-imp.pl *ma* 1sg.nom *saat-óoi* take.care-imp.pl 'It [the goat] says, "Come after me! Look after me!" ' (A:KEE007-8)

Normally, it seems, the second person is not explicitly expressed with a pronoun in imperative clauses, but occasionally it does occur, as in (49).

(49) *míi* 1sg.gen *maníit-u* say.pfv-msg *ki* comp *šóo* good.msg *tus* 2pl.nom *b-óoi* go-imp.pl 'I said, "That's fine, you may leave!" ' (A:ACR010)

The plural form is always used with reference to more than one person, but a more polite request could be done by adding the particle *neé* or *na*, to any of the two forms, to the singular in (50) and to the plural in (51).


'Open the door, please! [uttered by a little girl addressing older women, supposedly mother and grandmother, in the household]' (B:HLN1020)

The Imperative is regularly negated with the negation particle *na*, as in (52), and there is therefore no prohibitive category distinct from the Imperative in the language.

(52) *kúṛi* woman *the* to *maníit-u* say.pfv-msg *ki* compl *teeṇíi* refl *banɡleé* bracelet.pl *na* neg *širinɡá* rattle.imp.sg 'He said to the woman, "Don't rattle your bracelets!" ' (A:WOM643)

### **10.2.2 Conditional**

Conditionality is essentially expressed with the protasis verb in a Conditional clause in the Simple Past (or one of the periphrastic categories built on the perfective), while the apodosis verb may be in the Present or the Future, a "split tense/aspect reference" not uncommon in languages overall (e.g., in Arabic, Dahl 1985: 80). It is commonly, in such constructions, used together with an additional element that can be analysed alternatively as a subjunction, alternatively

### 10 Verbal categories

as an auxiliary. There are two such elements in Palula, one *seentá* (B *síinta*) which is used to form constructions of assumed conditionality, *yhóolu seentá* 'if he comes, when(ever) he comes' (53) and another, *heentá*, which usually carries a hypothetical meaning, *paás bhíla heentá* 'if we pass, were we to pass' (54). Further details on conditionality can be found in §14.4.4.


'If we pass the test, we will go to the village school.' (A:OUR015)

### **10.2.3 Obligative**

Whether or not the verb form coding the Obligative should be seen as a finite or a non-finite verb form is not as straightforward a matter as it may first seem. It is analysable as both, and it is possible that we are witnessing a still ongoing development from one to the other.

The function of the Obligative is to express obligation, need or predestination to carry out a particular action. This is the agent-oriented modality expressed in languages in general, though not necessarily inflectionally (Bybee, Perkins & Pagliuca 1994: 177–187). Often in IA languages this particular semantics is connected with what Masica (1991: 322) refers to as a *future passive participle (of obligation)*, which behaves like an adjective, but may in some languages be formally identical to a verbal noun, as is the case with the Urdu-Hindi infinitive formed with *-nā*.

In Palula, the Obligative is used both in the personal sense 'scheduled to, is to' (55) and in a more general or impersonal sense of 'ought to, one should' (56).


### 10.2 Non-indicative finite categories and their functions

Sometimes the future reading of 'will inevitably happen' is clearly present, as in (57), while in other cases there is no such future sense, only an instructional one with the sense of 'the way it should be/usually is being done', as in (58).


The dual syntactic character of the construction is seen if we compare (55) with (57). In (55) it is noun-like ('Munir's inevitable going to Drosh'), with the agent in the genitive, whereas in (57) it is adjective-like ('the inevitably finishing meat'). It is also in the first instance where the Obligative is most finite, whereas its finiteness can be questioned in the second one.

It seems the agent is generally avoided with obligative transitive verbs, and as such it is also becoming an alternative way of expressing a passive meaning (compare with §9.5.2), with a sense of ability (or inability) rather than one of obligation. I would still hold that Obligative is primarily modality and only secondarily voice. Possibly related to the degree of obligation is an alternation between the coding of the agent in the nominative, as in (59), versus the genitive coding we saw above in (55).

(59) *so* 3sg.nom *típa* now *b-eeṇḍeéu* go-oblg

'He is about to leave./He must leave now.' (A:Q9.0003)

In my data I do not have any clear examples of transitive verbs showing the same alternations, but as far as the agent is at all explicitly mentioned, she or he always occurs in the genitive. However, Morgenstierne (1941: 24) gives examples of a few transitive obligatives where it seems the nominative and the genitive are equally possible, while his subjects of intransitive obligatives invariably occur in the nominative:<sup>6</sup> *mī / ma krām thäiṇḍēu* 'I must(?) do the work'; *mī / ma wī piläiṇḍēu (hinu)* 'I can(?) drink water'; *āǰ ma yɛṇḍēu hinu ̄* 'I can (or must) come today'.

<sup>6</sup> Although we can exclude the genitive, we in fact do not know, due to the absence of such examples, whether any pronouns other than the 1sg (which lacks a nominative/accusative contrast) would have been coded in the nominative or accusative in those days (in the 1920s).

### 10 Verbal categories

### **10.2.4 Hearsay and quotative**

Hearsay, formed with *maní*, is grammaticalised to a much lower extent than the hitherto mentioned categories, and is thus only in a peripheral sense part of the TMA system. When it occurs, it occurs at the end of a sentence, like an auxiliary, after the inflected finite verb form (whether simple or periphrastically formed) to indicate reported (but not self-experienced) information. It is chiefly (but optionally) used in narratives, especially those of a legendary character, and mostly at the beginning of such narratives, which is seen in (60).<sup>7</sup>

(60) *bíiḍ-u* much-msg *muṣṭú* before *áak* idef *miiš-íi* man-gen *áak* idef *dhií* daughter *de* be.pst *maní* hsay 'A long time ago there was a man who had a daughter.' *se* 3fsg.nom *áak* idef *bakaraál* shepherd *phoó* boy *the* to *ašáx* love *de* be.pst *maní* hsay 'She was in love with a shepherd boy.' *so* def.msg.nom *phoó* boy *bi* also *tás* 3sg.acc *the* to *bíiḍ-u* much-msg *šóo* good.msg *ǰhóon-a* know-3sg *de* pst 'The boy was also very fond of her.' (A:SHY001-3)

As can be seen in example (60), *maní* is used only in the two first sentences, but there are also narratives where this is used much more frequently, in the extreme case after almost all sentences. It is also used when presenting legendary or questionable facts in a narrative, as in (61), otherwise unmarked in this respect.

(61) *tasíi* 3sg.gen *áaṣṭ* eight *zára* thousand.pl *kuṇaak-á* child-pl *heensíl-a* stay.pfv-mpl *de* pst *maní* hsay 'He had 8,000 children [it has been said].' (A:ABO008)

The marker *maní* is, however, not restricted to narrative discourse. It may be used in any everyday conversation, as in (62), here along with the Perfect. Also in this utterance it is only after the first finite verb form that the hearsay marker occurs.

<sup>7</sup> I am inclined not to consider it a story starter as such; that function is instead carried by *bíiḍu muṣṭú, muṣṭú zamaanáii*, *muxáak zamaanée*, or similar more or less standardised story-starters. The initial sentence's occurrence of the participant-introduction *heensílu de* 'there was' is also typical of the Palula storytelling discourse.

10.3 Non-finite forms and their functions

(62) *asíi* 1pl.gen *atshareet-á* Ashret-obl *bíiḍ-u* much-pl *kir* snow *dít-u* give.pfv-msg *hín-u* be.prs-msg *maní* hsay *hiimeel-í* glacier-pl *bi* also *whéet-im* come.down.pfv-fpl *hín-i* be.prs-f '[I have been told that] a lot of snow has fallen in our [village] Ashret, and there have been avalanches as well.' (A:CHN070320)

Related to this issue, is the use of the quotation marker *thaní*, with a similar etymology, which may develop into an auxiliary. Triggered by the verb-final word order, it may become restricted to the position after the finite verb and become narrowed down to this particular verb form.<sup>8</sup> In the process, it may also become a marker of a more grammaticalised inferentiality category, as such contrasting with Hearsay, rather than now as chiefly a marker of direct quotation/perception (see §14.5.1). In (63) it has taken on a function beyond mere quotation.

(63) *so* 3msg.nom *búd-u* understand.pfv-msg *ki* comp *míi* 1sg.gen *kúṛi* woman *dúi* other *koó* someone *ɡaḍ-í* pull-cv *ɡhin-í* take-cv *ɡíia* go.pfv.pl *thaní* qot 'He understood that someone else had taken his wife away.' (A:WOM656)

The grammaticalisation of Hearsay as well as other types of inferentiality is documented for a number of other languages in the region, especially in the geographically adjacent Khowar and Kalasha, where it is particularly well-developed and is an integral part of verb morphology (Bashir 1996a).

### **10.3 Non-finite forms and their functions**

Most of the non-finite verb forms have implications for syntax and are hard to define functionally without discussing syntactic structure. The following sections therefore only aim at giving a basic characterisation of their function; where to find them in the grammatical system and what their further inflectional properties are, wherever the latter is relevant. For further details in this regard, the cross-references given at the end of each subsection should be consulted.

<sup>8</sup> It is still alternatively expressed with other verb forms.

### 10 Verbal categories

### **10.3.1 Converb (conjunctive participle)**

The Converb – defined by Haspelmath (1995) as "a non-finite verb form whose main function is to mark adverbial subordination" – is essentially a perfective adverbial participle (compare with the Copredicative Participle, §10.3.5, which is an imperfective adverbial participle). There is a similarly-functioning participle in many other IA languages, often referred to as a conjunctive participle.<sup>9</sup> It is a characteristic, very frequently occurring, and syntactically important nonfinite verb form (Masica 1991: 323, 397–401). Like in many of those other languages, the Palula Converb carries an approximate meaning of 'having done, etc.' It is (as Haspelmath's definition also spells out) especially important in adverbial subordination and in expressing perfective sequentiality, as illustrated in (64).See §14.3.1 and §14.4.1 for further details.

(64) *aḍaphará* halfways *whaí* come.down.cv *ba* top *damá* pause *thíil-u* do.pfv-msg 'Halfway down, we rested.' (A:GHA057)

It is also a component in some resultative constructions (§10.1.7) as well as in compound verbs (§9.6.2), and is possibly the source of some postpositions with a verbal origin.

### **10.3.2 Perfective Participle**

Since the Perfective Participle is formally identical to the Simple Past (and as such has become a finite verb form), it only occurs marginally in what must be interpreted as a non-finite function. When it does, such as in the (perfective) nominalisations below, it can also (but not consistently so) be inflected for case, genitive in (65) and oblique in (66).


<sup>9</sup> The identification of the corresponding verb form in Kohistani Shina with the converb, according to Haspelmath's (1995) definition, is carried through in Schmidt 2003.

10.3 Non-finite forms and their functions

The Perfective Participle is also, as seen in (67), used attributively in some relativisations (see §14.6.6 for further details).

(67) *phaí* girl *teeṇíi* refl *háat-am* hand-ins *čooṇṭéel-i* embroider.pptc-f *rumiaál* handkerchief *dít-i* give.pfv-f *hín-i* be.prs-f 'The girl gave him the handkerchief which she had embroidered herself.' (A:SHY031)

Often however, as can be seen in (68), the Perfective Participle will occur with an additional adjectiviser *bhaáu* or *wáandu/wéendi*, maybe to mark it off as different from the otherwise formally identical Simple Past.

(68) *muṣṭú* back *zamanáii* period.gen *paṇardóoṛ-a* elder-pl *búuḍ-a* old-pl *ɡáaḍ-a* big-pl *xalk-íim* people-pl.obl *díi* from *ṣuunt-u=wáand-u* hear.pptc-msg=adj-msg *áa* idef *qisá* story *eeteeṇ-ú* such-msg *hín-u* be.prs-msg *ki* comp 'Once, a story goes, which has been heard from the elders, the old and the powerful people….' (A:KIN001)

Some Perfective Participles have been fully lexicalised as adjectives: for example *šúku* 'became dry' > 'dry', *páaku* 'ripened' > 'ripe', and *khíndu* 'became tired' > 'tired'.

### **10.3.3 Verbal Noun**

The Verbal Noun is in many ways the imperfective counterpart of the Perfective Participle, although their respective usages do not exactly mirror each other. The verbal noun is the form that most clearly in the eyes and ears of educated mother-tongue speakers essentialises the verb, the form first given or normally used when talking metalinguistically about functions and meanings of individual verbs. It is also regarded as the most natural choice as a citation form in word lists.

Although textually not as frequent as the Converb, it is a category used in a number of dependent clauses, such being the predicate in many complement clauses (for details, see §14.5), as in (69), as well as in some clauses with adverbial functions (see §14.4). It may also, as in (70), although quite infrequently, be used attributively (see §14.6.6 for a discussion).

(69) *pirsaahib-á* Pir.Sahib-obl *ba* top *inkaár* refusal *thíil-i* do.pfv-f *so* def.msg.nom *phoó* boy *deníi* give.vn *díi* from 'Pir Sahib refused to hand over the boy [lit: from giving the boy].' (B:ATI062)

### 10 Verbal categories

(70) *rhootašíi-a* morning-obl *ǰhambréeṛi* bride *har-ainií* take.away-vn *waxt* time *yhóol-u* come.pfv-msg *ta* sub 'In the morning, at the time of taking the bride…' (A:GHU010)

The Verbal Noun can also occur with nominal inflection, although, again, it is rather unusual (and in the A dialect often phonologically neutralised), as in the causality clause in (71), where it takes a genitive case inflection.

(71) *ǰeníi-e* hit.vn-gen *iṇc̣* bear *zaxmát* wounded *bhíl-u* become.pfv-msg 'The bear was wounded from the firing.' (B:BEL320)

### **10.3.4 Agentive Verbal Noun**

The Agentive Verbal Noun is another nominalisation but in many ways it is secondary as a nominal formation and is semantically rather restricted. It is a way of singling out an agent of an action (provided with a gender/number ending) but often reflects a more idiomatic or lexicalised meaning than the semantics of the verb itself. When used as nouns they inflect as nouns, see (72), but they can equally well function as adjectival modifiers of other nouns, as in (73).


'This girl has a bad character [she is "loose"].' (A:HLE2084)

The Agentive Verbal Noun also occurs rather systematically in a special construction with *bhílu hínu* 'has become'. In this case, the whole sentence communicates either the beginning of an event (possibly leading up to some sort of climax) or the co-occurrence of two events, as in (74). This may be related to the *vālā* of Urdu-Hindi, which, added to the oblique form of the infinitive,<sup>10</sup> captures a meaning of imminent action (Schmidt 1999: 139).

<sup>10</sup> The Urdu verb form with an ending *-nā* is traditionally called an infinitive but occurs in a number of constructions, often as a verbal noun and as such being subject to case inflection (Schmidt 1999: 132–142).

10.3 Non-finite forms and their functions

(74) *deerá* room *šíiṭi* inside *siɡréṭ* cigarette *dhrak-í* pull-cv *ba* top *tasíi* 3sg.gen *maxadúši* front.of *wée* in *tasíi* 3sg.gen *putr* son *tas* 3sg.acc *the* to *dac̣h-áaṭ-u* look-ag-msg *bhíl-u* become.pfv-msg *hín-u* be.prs-msg 'As he smoked a cigarette in the room, in front of him, his son had started to look at him.' (A:SMO001-2)

### **10.3.5 Copredicative Participle**

This (imperfective or aspectually non-specified) participle form functions as a modifier of the main verb in a clause, often functionally equal to that of a manner adverb. There is probably a higher degree of lexicalisation as far as combinations of head verbs and adverbial participles are concerned than with most other participle forms. In many ways, the kind of interaction of the two verb forms, the participle and the finite verb, as can be seen in (75) and (76), is with a resulting specified meaning not too different from the function of the Converb. For more details of use, see §14.4.1.


### **10.3.6 Infinitive**

The Infinitive occurs solely, as far as my data goes, as the head of the complement of two modal verbs, *bha-* 'be able to, can', as in (77), and *ṣáat-* 'began', as in (78), see also §13.2.7. In this combination, the Infinitive tends to be more or less deaccented and phonologically fused with the matrix clause finite verb.

(77) *dúi* other *ta* cntr *ɡaṭíl-u* defeat.pfv-msg *áak* one *ḍaaku-waan-óom-ii* robber-pl-obl-gen *qilaá* fort *tíi* 3sg.obl *na* neg *ɡaṭ-aa=bhóol-u* win-inf=be.able.to.pfv-msg

'He defeated them all, only one den of thieves he was not able to defeat.' (A:PIR008)


## **11 Noun phrases and non-verbal agreement**

### **11.1 Noun phrase properties**

### **11.1.1 Types of noun phrases**

A noun phrase can consist of: A pronoun, as *taním* and *ɡa* in (1).

(1) *taním* 3pl.erg *ɡa* anything *na* neg *láad-u* find.pfv-msg 'They didn't find anything.' (A:DRA003)

A nominalised clause, as the one headed by the Verbal Noun *thainií* in (2).

(2) *putríi* son-gen *ǰhaní* marriage *thainií* do.vn *zarurí* necessity *bháanu* become.prs-msg 'It is necessary to get one's son married [lit: Making a son's marriage becomes a necessity].' (A:MAR009)

Or a noun head, with or without preceding modifiers, as can be seen in (3).

(3) *ṣiṣ-íi* head-gen *so* def.msg.nom *tuúš* some *tukṛá* piece *mhaás* flesh *aṭ-í* bring-cv *qábur* grave *the* to *aṭ-í* bring-cv *tíi* 3sg.obl *wée* in *ḍhanɡóol-u* bury.pfv-msg

'[He] brought some piece of the flesh to the grave and buried it there.' (A:BER014)

It is the latter type that will be the main focus of this chapter. Pronouns (see Chapter 6), in the sense of pro-noun phrases, are not treated here, other than very briefly in connection with apposition; and nominalised clauses, as described in §14.4 and §14.5, have the internal syntax of clauses and are thus not examples of noun phrase syntax proper.

### 11 Noun phrases and non-verbal agreement

A number of modifiers can precede (and exceptionally follow) a noun head within the noun phrase, some of them single words, others phrases or even clauses in themselves: adjectives/adjective phrases, genitive phrases, quantifiers/ quantifier phrases, determiners and relative clauses. As will be clear in the discussion below (§11.1.2) some words can fill more than one such modifier function, and in some cases even extend into adverbial modification. The differentiation between the classes of modifiers is not always clearcut.

It is also possible for a (substantivised) modifier alone to function as the head of a noun phrase. This is particularly common with adjectives, and to a lesser extent with cardinal numerals. The ability of demonstratives to function as modifiers as well as pro-NPs is of course another example of the same tendency.

### **11.1.2 Modifiers in noun phrases**

**Adjectives or adjective phrases** are descriptive in nature, often capturing inherent properties or qualities of the entity referred to by the head noun. They may consist of a single adjective, as *ɡáaḍu* in (4).

(4) *so* 3msg.nom *ba* top *ɡáaḍ-u* big-msg *maidóon* field 'It was a big field.' (A:JAN030)

An adjective phrase can also be complex. It can consist of two or more adjectives, symmetrically related to each other, such as *dhríɡi* and *bhakúli* in (5), both pointing to certain characteristics of the noun they modify.

(5) *se* def *insaan-á* human-obl *áa* idef *dhríɡ-i* long-f *bhakúl-i* thick-f *lhaléemi* stick *ɡhin-í* take-cv 'The man took a long, thick stick…' (A:KIN024)

The adjective head can also itself be modified, usually by a scalar quantifier, such as *bíiḍu* modifying *mhoóru* and *šišáwu* in (6), assuming that the noun can have more or less of the quality implied by the modifying adjective phrase.

(6) *bíiḍ-u* very-msg *mhoór-u* sweet-msg *bíiḍ-u* very-msg *šišáw-u* good-msg *šay* thing *bh-áan-u* become-prs-msg 'It is a very sweet and a very good thing.' (A:KEE052)

A repeated adjective, on the other hand, such as *muxtalíf* in (7), usually signifies a distributive meaning.<sup>1</sup>

<sup>1</sup> Although the adjective phrase in the example is predicative, it still illustrates the point.

11.1 Noun phrase properties

(7) *tasíi* 3sg.gen *xaaneé* shelf.pl *muxtalíf* different *muxtalíf* different *yháand-a* come.prs-mpl 'There are different kinds of shelves.' (A:HOW050)

Finally, Perfective Participial clauses with the external syntax of APs, can also function as modifiers of a head noun (see §14.6.6).

**Genitive phrases** are very common as modifiers. Although this is also a matter of specifying quality in some sense, it is usually a specification of origin, relatedness or material, as illustrated in (8)–(11). They therefore also share some of the identifying characteristics of demonstratives and relative clauses (see below). A modifying genitive phrase can also be a nominalised clause with a Verbal Noun in the genitive case as its head (see §14.6.6).


**Quantifiers or quantifier phrases** are as a modifier category rather heterogeneous and comprise a few subclasses. One easily distinguishable subclass is cardinal numerals. Used alone, they simply specify quantity, as in (12). They can also, as seen in (13), be modified themselves by e.g., *taqriibán* (often pronounced *qariibán*) 'about, approximately', a Perso-Arabic loan, and thereby become a bit more relative.

(12) *so* 3msg.nom *ta* cntr *bač* saved *bhil-u* become.pfv-msg *trúu* three *ǰáan-a* person-pl *ba* top *hiimeel-í* glacier-obl *híṛ-a* take.away.pfv-mpl

'He was saved, but three persons were taken by the avalanche.' (B:AVA214)

### 11 Noun phrases and non-verbal agreement

(13) *qaribán* about *bhiíš* twenty *kaal-á* year-pl *maxadúši* before 'About twenty years ago…' (A:GHA048)

While cardinal numerals can be used only to modify count nouns, scalar quantifiers, such as *tuúš* 'some' (14) and *úča* 'a few' (15) are used to quantify mass nouns as well as count nouns in plural. Some of these are also freely used in adverbial constructions, for instance to modify adjectives, adverbs or entire clauses. One particularly frequent multipurpose modifier is *bíiḍu* 'much, many, very'.


Another strategy for quantification, seen in (16) and (17), is by means of a partitive noun phrase. It specifies the quantity of the head noun, often itself preceded by or modified by a cardinal numeral. Typically, but not exclusively, the nouns used in such partitive phrases denote containers or measuring terms of various kinds. In many ways it would make sense to describe higher numerals (such as 20, 100, 1000) as heads of partitive phrases, modified by the cardinal numerals 1–19 to express the numbers 21–39, etc. (see §7.4).


**Determiners** and their use in noun phrases have been described elsewhere (see §6.2 and §6.3.6). This class includes all words or phrases that have the function of singling out a referent in contrast to other referents. It primarily identifies the 11.1 Noun phrase properties

referent of the noun head among a number of potential referents. Whereas the genitive phrase modifier defines or introduces a referent, the determiner points out a particular referent, 'this house' in (18), and 'the man', 'another day', and 'other people' in (19), whether already defined or not.


'Next day the man called the king, the prime minister and other people together and made an announcement.' (A:UXW060)

**Relative clauses** (see §14.6) serve a very similar purpose of identifying the referent, as the 'brutes' in (20), and the 'stories' in (21). For many relative clauses, however, it is unclear to what extent they should be considered part of the noun phrase at all or rather be analysed as entirely paratactic constructions. The latter is especially true of the so-called co-relative or relative-correlative constructions common in IA languages.


'He had the people tell him what had happened [lit: stories which had passed].' (A:UXW057)

### **11.1.3 Apposition**

Apposition is another phenomenon with seemingly fuzzy borders, and it is not always obvious what is to be analysed as noun phrase syntax and what as noun derivation. The following are a few different types of noun phrases consisting

### 11 Noun phrases and non-verbal agreement

of a final noun head and one or more modifying (or further specifying) noun phrases (or what would at least constitute noun phrases if used independently). The head and the modifier have the same referent.

The head can be a title or a designation and the preceding apposition is usually a proper name. Only the head, in this case *hakím* 'ruler' in (22), and *ṣoó* 'king' in (23), is inflected.


A complex proper name (i.e., many names referring to the same person), as in (24) or in (25), functions in much the same way, as its last component is more prominent. Quite often this component is (or has been) a title of some kind. For short complexes, it is probably a matter of word formation rather than apposition. The included names constitute a single phonological word, with the last syllable of the final component receiving the main accent (or the suffix according to shiftaccent rules applied), but the preceding parts are entirely or partly deaccented.


'Mullah Mahmad Said knew that…' (A:MAH011)

A special case is when the obligatory string *aleehisalaám* 'peace be upon him' (PBUH) is added after the name of a prophet has been uttered, according to Islamic tradition. Then the final syllable of that standard string receives the accent, as is seen in (26), and any inflections are attached to the end of the entire phrase. 11.1 Noun phrase properties

(26) *xu* but *eesé* dist.obl *waqt-íi* time-gen *peeɣambár* prophet *hazrát* lord *iliaás* Elijah *aleehisalaam-íi* peace.be.upon.him-gen *beet-í* word-pl *káaṇ* host *na* neg *th-íi* do-3sg *de* pst 'But he was not listening to Lord Ilyas [Elijah], PBUH, the prophet of the time.' (A:ABO011)

The head as well as the preceding apposition can also be common nouns. Although the apposition functions as the modifier, and as such takes the place of an adjective phrase, both nouns together contribute almost equally to the identification or specification of the referent, 'female person' and 'women' in (27), and 'shepherd' and 'boy' in (28). Each of them can separately and independently function as a referring noun phrase.

(27) *tas* 3sg.acc *sanɡí* with *čúur* four *páanǰ* five *ǰéeni* female.person *kuṛíina* woman.pl *áa* one *ba* top *míiš* man */…/ phray-áan-a* send-prs-mpl 'They send with her four or five women and one man.' (A:MAR082-3) (28) *se áak bakaraál phoó the ašáx de*

3fsg.nom idef shepherd boy to in.love be.pst 'She was in love with a shepherd boy.' (A:SHY002)

In other cases, the head is a proper name and the preceding apposition is a kinship term. In (29), both the noun head, 'Mullah Mahmad Said', and the head of the apposition, 'grandfather', take inflections, thus displaying a higher degree of independence (and symmetry) than the aforementioned types.

(29) *míi* 1sg.gen *se* def *dóod-a* grandfather-obl *mulaa=mhaamad=seed-á* Mullah=Mahmad=Said-obl *the* to *ba* top *eesó* rem.msg.nom *paalawaáṇ* strong.man *maalúm* knowledge *heensíl-u* stay.pfv-msg *hín-u* be.prs-msg 'My grandfather Mullah Mahmad Said knew this strong man.' (A:MAH027)

The head can also be a common noun and the preceding apposition a pronoun. The role of the pronoun, 'we' in (30) and 'you' in (31), is not much different from that of a determiner.

11 Noun phrases and non-verbal agreement


### **11.2 Word order in the noun phrase**

Although it is possible to use more than two modifiers in the same noun phrase, it is not too common in natural speech, and therefore the following description of the relative word order should be regarded as a presentation of tendencies more than as hard and fast rules without exceptions. The clearest tendencies can be seen in the order of genitive phrases, determiners, quantifiers and adjective phrases with respect to each other. The following order seems to be more or less fixed: Genitive NP + Determiner + Quantifier + AP + Head. The determiner, *se* in (32) and *aní* in (33), precedes the quantifier.


The determiner, *se* in (34) and *ɡóo* in (35), also precedes the adjective phrase.


11.2 Word order in the noun phrase

The quantifier, *dúu* in (36), precedes the adjective phrase *ɡéeḍi*. Although numerals and adjective phrases only rarely co-occur in the same noun phrase, there is a strong feeling about the grammaticality of this order vis-à-vis the opposite one.

(36) *dúu* two *ɡéeḍ-i* big-f *durbaṭ-í* pot-pl *(\*ɡéeḍi dúu durbaṭí)* 'two big pots' (A:HLE2474)

The genitive NP, regardless of its internal complexity, generally precedes all other modifiers whenever they occur in the same noun phrase: adjective phrases, as in (37), quantifiers, as in (38), indefinite determiners, as in (39), and definite determiners, as in (40).


Another modifier preceding the head of the genitive NP, the definite *se* in (41) and the indefinite *ak* in (42), is normally interpreted as part of the genitive NP and as such a modifier of the genitive noun rather than the head of the main noun phrase.


### 11 Noun phrases and non-verbal agreement

However, the order between the genitive NP and other modifiers is not entirely fixed, and although the preferred order is genitive NP first, there are a few cases, such as (43) and (44), where the preceding modifier most likely is a direct modifier of the main noun head, 'owner' and 'fort' respectively, rather than being a part of the genitive NP. This interpretation is based partly on agreement features, partly on context, and it opens the construction up to a certain degree of syntactic ambiguity but seldom with any real risk of semantic misinterpretation.


Occasionally a genitive modifier is heavy-shifted to a position after the noun head, as *míišii práačamii* in (45). This may alternatively be interpreted as an example of afterthought.

(45) *áak* idef *ɡáaḍ-u* big-msg *haál* hall *míiš-ii* man-gen *práač-am-ii* guest-pl.obl-gen 'a big hall for the man's guests' (A:SMO021)

It is harder to make any generalisations about the position of relative clauses vis-à-vis other modifiers, partly due to their questionable status as an integral part of the noun phrase (as already mentioned), and partly because of the lack of any clear evidence as far as co-occurrence of relative clauses and other modifiers is concerned.

### **11.3 Agreement patterns**

There are two main types of agreement within the noun phrase, a lower-differentiating determiner agreement and a higher-differentiating adjectival agreement. Agreement between a predicate adjective phrase and the subject noun phrase basically follows the same principles as attributive adjectival agreement. In addition to those patterns, there is an extended partial agreement between a scalar quantifier used adverbially and the head of a noun phrase.

### 11.3 Agreement patterns

### **11.3.1 Determiner agreement**

For the most common determiners, there is one form agreeing with a nominativemasculine singular head, within each determiner set, and another form used with all other heads, as far as case, number and gender are concerned. This kind of agreement is displayed in Table 11.1.



All determiners with more than one form have: a) one that agrees with the nominative-masculine singular head, ending in an accented *ó* or *ú*, such as *so* agreeing with the nominative singular masculine head in (46), and: b) another form ending in an accented *é* or *í*, for example, *se* agreeing with the non-masculine head in (47), *se* agreeing with the non-nominative head in (48), and *se* agreeing with the non-singular head in (49).


### 11 Noun phrases and non-verbal agreement

(49) *se* def *xálaka* people[nom.mpl] *qalánɡ* tax *na* neg *d-áa* give-inf *bhaá* be.able.to.cv *ba* top *uḍheew-í* flee-cv *ɡíia* go.pfv.mpl 'The people were not able to pay the taxes, so they left, fleeing.' (A:MAB030)

### **11.3.2 Adjectival agreement**

Also, for most cases of adjectival agreement, there is a unique nominative masculine singular form ending in *u*, but there is an additional differentiation between masculine and feminine, as shown in Table 11.2.

Table 11.2: Adjectival agreement (*dhríɡ-* 'tall, long')


The great majority of inflecting adjectives occur in the three forms ending *-u, -a* and *-i* (and an additional but marginally used feminine plural *-im* in predicative sentences, see §11.3.3 below). There is thus agreement with the noun head in gender, number and case. Whereas agreement in gender is consistently maintained, compare (50) with (51), number agreement (compare (52) with (53)) and case agreement (compare (54) with (55)) is mostly neutralised.


11.3 Agreement patterns

*hín-a* be.prs-mpl 'There are small innocent children in my house.' (A:KIN017)


However, as was pointed out in Chapter 7, there is a category of invariable (non-inflecting) adjectives, not showing any kind of overt agreement at all with the head of the noun phrase, such as *muxtalíf* in (56) and *taaqatwár* in (57). It should be noted that although such adjectives do occur attributively, they are more readily used predicatively and some of them exclusively so.


### **11.3.3 Predicate agreement**

Just as adjectival attributes agree with their heads, the heads of predicate phrases agree in gender and number with the head of the subject noun phrase, as displayed in Table 11.3. That is in as far as the particular adjective belongs to the inflecting class. It should be noted, however, that the large majority of adjectives used in the predicative function are of the invariable type.

As with attributively used adjectives, there is a non-optional inflectional contrast between adjectives agreeing with a masculine-singular head, as in (58), and adjectives agreeing with a masculine-plural head, as in (59).

### 11 Noun phrases and non-verbal agreement

### Table 11.3: Predicate agreement



A secondary (and optional) feminine singular-plural inflectional differentiation seems to be permitted in at least copula-less sentences, as can be seen when comparing the adjectives in (60) and (61).<sup>2</sup>


Apart from a larger variety of adjectives and adjective phrases allowed predicatively rather than attributively, even quantifiers may occur in a predicative function. Quantifiers with the ability to inflect agree with the subject noun phrase in gender/number, as can be seen with *bíiḍ-* 'much, many' in (62) and (63), and *biǰóol-* 'several' in (64).

<sup>2</sup> The status of agreement with *-im* requires further investigation. The form as relating to plurality has probably originated (relatively recently) in the class of feminine nouns (see Chapter 5) ending in *-i*, and has subsequently spread analogically to the verbal and adjectival paradigms, although not yet consistently applied in the latter two.

11.3 Agreement patterns


### **11.3.4 Extended agreement**

The adjectival agreement is also extended or copied to inflecting adjectives used as adjuncts of other adjectives (primarily *bíiḍ-* 'much'), i.e., as adverbs. That means that the gender and number of the noun that the adjective head agrees with (whether attributively or predicatively) is also lended in agreement to the adjective adjunct, this regardless of the ability of the adjective head itself to inflect (compare with the agreement patterns of adverbs in Gujarati, Hook & Joshi 1991). This is seen in the forms of *bíiḍ-* in (65)–(69) and *šóo* in (70).


### 11 Noun phrases and non-verbal agreement


This is also the case with adjuncts in adverbial phrases used predicatively in which the adjunct agrees in gender and number with the noun head of the subject. Again, it is primarily the scalar modifier *bíiḍ-* 'much' that is being used. In (71), *bíiḍ-* agrees in feminine gender with *iskuúl* 'school', and in (72) it agrees in masculine singular with *ɡhoóṣṭ* 'house'.


A form of incomplete agreement pattern is seen between an argument in a nominalised complement clause and the adjunct of the complement-taking adjective. In example (73), the adjunct *bíiḍ-* agrees with the feminine direct object of the nominalised verb *xat* 'letter', whereas in (74) and (75) the choice of the masculine singular seems to be due to its default value rather than one triggered by agreement with any particular argument. This certainly deserves more indepth study.<sup>3</sup>


<sup>3</sup> Note that *askóon* 'easy' is a non-inflecting adjective.

11.3 Agreement patterns

Adjectival agreement of this kind may be extended even to clause-level adverbial modification, as can be seen in examples (76) and (77).


'The North Wind blew as hard as she could.' (A:NOR005)

## **12 Grammatical relations**

Palula, like many related IA languages, displays an intricate and quite complex relationship between the grammatical cases expressed, the particular case forms and agreement patterns available and the various functions a noun phrase can have in a given utterance (Masica 1991: 230–231). A complicating factor is the type of split ergativity displayed, related to aspect on the one hand and the nature of the NP on the other. The former is a rather straightforward matter, with ergativity being a feature only of perfective clauses, whereas the relationship between properties of the NP and ergativity is less transparent, with several different cutoff points, some of them less expected from a typological standpoint. Word order is unmarkedly intransitive subject – verb and transitive subject – object – verb, but it allows for quite a deal of pragmatic flexibility. As the two other factors weigh heaviest as far as grammatical relations and alignment are concerned, word order will not enter into the present discussion. This also goes along with the general observation that the presence of the other mechanisms in a language correlates to a relative flexibility in the basic word order (Blake 2001: 14–15).

Following Dixon (1994: 6–8)/Dixon (2010: 76–77) and Bickel (2011: 402), I will be using the following abbreviations for the three grammatical relations (also syntactic primitive relations): S – intransitive subject, A – transitive subject, and O – transitive object.<sup>1</sup> This means that a verb with one core noun phrase is intransitive and the sole argument relation we refer to as S, a verb with two core noun phrases is transitive and the two argument relations we call A and O. In Palula, the unmarked order of the arguments is S-V*itr* and A-O-V*tr*, respectively.

### **12.1 Verb agreement**

As already described in §9.4.1, grammatical relations can be reflected in the marking of the predicate itself, i.e., by verb agreement. In Palula, the verb can only display agreement with a single argument in one and the same clause. While

<sup>1</sup> O corresponds to P (for patient) in some of the typological literature (Comrie 1989; Croft 2003).

### 12 Grammatical relations

agreement in person is limited to the non-present imperfective (Future and Past Imperfective), the main type of agreement is one in gender and number, found in the perfective categories and in the Present (see §10.1). However, as far as grammatical alignment is concerned, the dividing line goes between the perfective and all non-perfective TMA categories: In the perfective there is ergative verb agreement, and in the non-perfective there is accusative verb agreement.

### **12.1.1 Accusative alignment**

In the non-present imperfective – Past Imperfective in (1) and Future in (2) –, the verb always agrees in person in accordance with an accusative alignment, i.e., with the subject, whether S, as in the intransitive clause in (1), or A, as in the transitive clause in (2).

(1) *ak* idef *praší* slope *phará* along *se* 3pl.nom **S** *b-éen* go-3pl *de* pst

'They were moving along a slope.' (B:AVA211)

(2) *se* def **A** *zinaawur-aán* beast-pl *xu* but *ma* 1sg.nom **O** *kh-óon* eat-3pl 'The beasts will eat me.' (A:KAT059)

An accusative agreement pattern is also applied in the Present, as can be seen in examples (3)–(4), although here in the form of number and gender agreement.


### **12.1.2 Ergative alignment**

When, on the other hand, any of the TMA categories based on the perfective are used, the verb always agrees – in accordance with an ergative alignment – with

### 12.2 NP case differentiation

S, as in (5), or O as in (6) and (7), whereas it never agrees in gender and number with A.

(5) *čhéeli* she.goat[fsg] **S** *eetáa* there *the* to *ɡíi* go.pfv.f *de* pst *ta* sub

'The goat had gone there.' (A:CAV026)

(6) *ínc-ạ* bear[msg]-obl **A** *čhéeli* she.goat[fsg] **O** *khéel-i* eat.pfv-f

'The bear ate the goat.' (A:PAS056)

(7) *kúṛi* woman[fsg] **A** *teeṇíi* refl *ɡhúuṛu* horse[msg] **O** *deechinéeti ̣* right *ḍáḍi* side *nuuṭóol-u* turn.pfv-msg 'The woman turned her horse to the right.' (A:UXW028)

Even when O occurs in a non-nominative form (described in detail in §12.2.2), as in (8), the verb still agrees in gender and number with that particular noun phrase argument.

(8) *tíi* 3sg.obl **A** */…/ tanaám* 3pl.acc **O** *bíiḍ-a* much-mpl *tanɡ* narrow *thíil-a* do.pfv-mpl *hín-a* be.prs-mpl

'He has troubled them a lot.' (A:KIN003)

### **12.2 NP case differentiation**

Three sub-instances of case differentiation will be exemplified below, as they relate to grammatical relations and alignment: inflectional case marking of nouns, pronominal case differentiation and NP-internal marking.

### **12.2.1 Inflectional case marking**

As we saw in §5.5, the central case distinction made inflectionally is that between the nominative and the oblique cases. As for the relations we are interested in, the nominative is used for S, A and O alike in the non-perfective categories, whereas

### 12 Grammatical relations

in the perfective, A, as in (11), is singled out as coded in the oblique case (*kaṭamušá*) versus the nominatively<sup>2</sup> coded (*kaṭamúš*) S in (9) and O in (10).

(9) *kaṭamúš-***ø** Katamosh **S** */…/ sóon-a* pasture-obl *dúši* toward *ɡúum* go.pfv.msg *hín-u* be.prs-msg 'Katamosh set out to the high pasture.' (A:KAT010) (10) *íṇc-ạ* bear-obl **A** *kaṭamúš-***ø** Katamosh **O** *aamúuṣṭ-u* forget.pfv-msg *hín-u* be.prs-msg 'The bear forgot about Katamosh.' (A:KAT140) (11) *kaṭamuš-á* Katamosh-obl **A** *ɡábina* nothing **O** *khóol-u* eat.pfv-msg *hín-u* be.prs-msg

'Katamosh didn't eat anything.' (A:KAT065)

This, however, is a somewhat simplified picture. In actual fact, not all nouns make the distinction between the nominative and the oblique, and some make it in the plural and not in the singular. The forms themselves occurring as morphological markers of ergativity also differ between nouns belonging to different declensions (see §5.6).

Although there is a form syncretism between the oblique singular and the nominative plural in the large *a*- and *i*-declensions (Table 12.1 and Table 12.2), that does not distort the nominative-oblique contrast *per se*. Here, a suffix *-a* or *-i* is the morphological reflex of ergativity in the singular and a suffix -*óom* or -*íim* in the plural. In the *m*-declension (Table 12.3), on the other hand, there is no differentiating ergative case marking available at all. For some *aan*-declension nouns (Table 12.4), the case differentiation is neutralised in the singular but maintained in the plural, while for a few others (Table 12.5) there are four distinct forms available: nominative singular, nominative plural, oblique singular and oblique plural. The *ee*-declension nouns (not displayed here) largely make the same distinctions as *a*- and *i*-declension nouns. With respect to frequency, the "full contrast pattern" represents the large majority of all Palula nouns (the *a*-declension and the *i*-declension together making up 70 per cent of all nouns), masculine as well as

<sup>2</sup> In the description of a system with a consistently ergative case alignment, this would be referred to as absolutive, a term I prefer not to use when describing the split ergative system of Palula.


Table 12.1: Case and number differentiation in the *a*-declension (perfective): *putr* 'son' (m)

Table 12.2: Case and number differentiation in the *i*-declension (perfective): *preṣ* 'mother-in-law' (f)


Table 12.3: Case and number differentiation in the *m*-declension (perfective): *méemi* 'grandmother' (f)


Table 12.4: Case and number differentiation in the *aan*-declension, V-ending (perfective): *baačaá* 'king' (m)


Table 12.5: Case and number differentiation in the *aan*-declension, C-ending (perfective): *anɡreéz* 'Brit' (m)


### 12 Grammatical relations

feminine, while the "no contrast pattern" is relatively small (about sixteen per cent), comprising exclusively feminine nouns.

### **12.2.2 Pronominal case differentiation**

The noun phrase slot could of course also be filled with a pronoun, and here too we have different forms bearing a relation to case. Apart from the singling out of A in the perfective, *asím* in (13), we also have pronominal forms particular to O, *asaám* in (14), vis-à-vis S, *be* in (12) and A (in the perfective and the nonperfective alike).

(12) *rhootašíia* morning *be* 1pl.nom **S** *ɡíia* go.pfv.pl

'In the morning we left.' (A:GHA006)

(13) *asím* 1pl.erg **A** *ǰinaazá* corpse **O** *khaṣeel-í* drag-cv *wheelíl-u* take.down.pfv-msg *de* pst

'We dragged the corpse down.' (A:GHA044)

```
(14) karáaṛu
      leopard
      A
              asaám
              1pl.acc
              O
                       kh-úu
                       eat-3sg
     'The leopard will eat us.' (B:FOY025)
```
This, again, is only part of the whole picture. Starting with the personal pronouns proper, these do not uniformly have the same number of forms or make the same distinctions formally, as can be seen in Table 12.6. The first- and second-plural personal pronouns make a three-way distinction, with unique ergative forms, *asím* and *tusím*, whereas the first- and second-singular only have two forms each (with nominative-accusative neutralisation as well as an ergative-genitive neutralisation). The demonstratives, which are used as third-person pronouns, see Table 12.7, show differentiation to the same extent as the plural personal pronouns, i.e., with a three-way case contrast.

### **12.2.3 NP-internal marking**

Case marking is also relevant for dependents in the noun phrase, although it has a much more limited scope (for details, see §11.3). There are two kinds of

### 12.3 The split system summarised


Table 12.6: Personal pronouns and case differentiation in the perfective

Table 12.7: Demonstrative case differentiation in the perfective (only the remote set represented)


NP-internal agreement: a) determiner agreement and b) adjectival agreement. Determiners occur in a maximum of two forms, one of them occurring only with a singular masculine noun head in the nominative, the other occurring elsewhere, i.e., with non-nominative singular heads, feminine and plural heads. Adjectives of the inflecting category display three different forms to reflect properties of the noun head: one for a singular masculine head in the nominative, another for nonnominative singular and plural masculine head, and a third for feminine heads, regardless of case or number. The agreement displayed by dependents within the noun phrase is therefore not adding anything to the differentiation already made explicit by the case-inflected head as far as case is concerned.

### **12.3 The split system summarised**

Summarising the findings in §12.1–§12.1, we have two dimensions on which ergative vs. accusative alignment and their expressions depend in Palula. First and foremost, the presence of ergative alignment is aspectually determined. While accusative properties can be present regardless of the TMA category realised, it is in the perfective only that an (additional) ergative pattern is found. A consistent correlation (see Figure 12.1) exists between perfective aspect and ergative verb agreement, and an accusative verb agreement and non-perfective categories.

### 12 Grammatical relations


Figure 12.1: Correlations between aspect and alignment in verb agreement (shading represents verb agreement)

Much less straightforward is the relationship between the nature of the NP and case marking. Even within the same aspectual category (the perfective) we have examples of non-differentiation (ASO all the same as far as case marking is concerned), a two-way differentiation (A marked differently from S and O) as well as a tripartite differentiation (A, S and O all distinguished by case marking). Table 12.8 illustrates how case differentiation is displayed for four different categories of NPs in Palula:



12.4 Alignment and split features in the region and beyond

### **12.4 Alignment and split features in the region and beyond**

How do the features summarised above relate to those found in the surrounding region and in related languages? As far as the presence of (morphological) ergativity is concerned, the situation in Palula is far from unique, neither among NIA languages in the region (Èdel'man 1983; Skalmowski 1974; Liljegren 2014) nor beyond (Deo & Sharma 2006; Klaiman 1987; Stroński 2009; Verbeke 2011), but its manifestations and more precise characteristics are manifold and quite diverse in what Masica (2001: 250) describes as an "ergative belt" stretching from the northeastern part of the subcontinent all the way to Caucasus, with modern Persian as one major exception in the middle of it. This belt includes Indo-Aryan, Iranian and Tibeto-Burman, as well as the isolate Burushaski and some of the language families represented in the Caucasus.

While ergative case marking is conditional in Palula, it is applied across the board in Burushaski and in the Shina varieties spoken adjacent to it (as far as case marking is concerned<sup>3</sup> ). Although the latter is due to substratum effects from Burushaski according to Masica (2001: 248), another phenomenon, termed "dual ergativity" by Hook & Koul (2004: 213), is observed in certain Eastern (including Kohistani) Shina varieties, where there is a TMA-related (imperfective vs. perfective according to Schmidt & Kohistani 2008: 51–53) alternation between ergativity markers of IA origin and an ergativity marker supposedly of Tibetan origin (Hook & Koul 2004: 214; Bailey 1924: 211).

However, far more common in NIA languages as well as in Tibeto-Burman, is some sort of TMA split between ergative patterns and accusative patterns (Masica 2001: 248), usually between perfective and non-perfective tenses (Masica 1991: 342–343). This may be manifested, as in Palula, in the agreement of transitive verbs with O in the perfective along with a distinctive case marking of A. That is the case in Urdu-Hindi (Schmidt 1999: 124) as well as in many of the other major NIA languages of the subcontinent (Masica 2001: 248). Geographically closer to Palula, this is also observed for the Kohistani languages (Baart 1999a: 136; Hallberg & Hallberg 1999: 34; Lunsford 2001: 93–95<sup>4</sup> ). While ergativity is seen in the case marking of A in other Shina varieties, verb agreement with O is not a feature of Gilgiti or Kohistani Shina. Instead, as in neighbouring Dameli and Gawarbati (personal observations), the verb agrees (accusatively) with S or

<sup>3</sup> As pointed out to me by Carla Radloff (pc), there is in Gilgit Shina a clearcut split between accusatively aligned verb agreement and ergatively aligned case marking (invariably with *-se* or *-s*).

<sup>4</sup> In the case of Torwali, ergativity is also manifested in the future tense.

### 12 Grammatical relations

A and never with O, whether or not there are other manifestations of ergativity or accusativity.

A number of Iranian languages in the region also display split ergativity (Payne 1980), although with certain peculiarities. For one, Pashto exemplifies a tense split rather than an aspectual split, with verb agreement with O and ergative case marking of A in past tenses, regardless of aspect (Tegey 1977: 4–5; Lorenz 1979: 71–72). In addition, a class of intransitive verbs (expressing involuntary activity) also require an ergatively marked A (Babrakzai 1999: 112), a phenomenon also described by Hook & Koul (2004: 217) for Indo-Aryan Kashmiri.<sup>5</sup>

While most languages within this so-called ergative belt show some ergative features, there are nevertheless some where they are entirely absent. In the immediate vicinity of Palula, the most notable examples are Kalasha and Khowar (Bashir 1988: 41). While this absence is a retention feature of Kalasha and Khowar, in some languages in other parts of the subcontinent, such as Bengali, Oriya and Sinhalese, a former ergative construction has probably been replaced only later by a consistent accusative alignment (Masica 1991: 343–344).

A number of different patterns are observed in the region as far as case marking, case syncretism and various types of NP splits being realised. Those languages manifesting verb agreement with O in some TMA categories, also tend to case mark A distinctively vis-à-vis S and O, but there are also those languages that maintain a tripartite S vs. O vs. A differentiation, if not for nouns, at least for the pronouns or a subset of them. In Punjabi, there is a shared 1sg nominative/oblique form, whereas 2sg, 1pl, as well as 2pl and the third-person pronouns differentiate between these two cases (Bhatia 1993: 229). In Gawri, the 1sg and 2sg make a subject vs. object/oblique/agent distinction, the 1pl and 2pl a subject/agent/object vs. oblique distinction, a subject vs. agent vs. object/oblique distinction in the 3sg, and a subject/object/oblique vs. agent distinction in the 3pl (Baart 1999a: 39).

In neighbouring Dameli (Morgenstierne 1942 and own observations), there is a nominative vs. accusative/ergative differentiation<sup>6</sup> in first- and second-singular as well as in plural, but where 1sg and 1pl nominative somewhat surprisingly have merged. For the demonstratives functioning as third-person pronouns, the situation is further complicated by animacy distinctions. In Gawarbati (Morgenstierne 1950 and personal observations), there seems to be an almost complete nominative vs. accusative vs. ergative differentiation upheld in all persons and

<sup>5</sup> This is also found in Urdu to a limited extent (Schmidt 1999: 168).

<sup>6</sup> This rather unexpected grouping of A with O vs. S is sometimes referred to as "double-oblique" (Payne 1980).

### 12.4 Alignment and split features in the region and beyond

in singular and plural (with nouns and pronouns alike), but only in so far as the NPs are definite and occur in the perfective. However, due to the lack of a more comprehensive study of the language sufficient to base any conclusions on, the analysis remains tentative.

For a more comprehensive treatment of alignment patterns and areality in the region, see Liljegren (2014).

## **13 Simple clauses and argument structure**

In this chapter, I present the different types of simple clauses found in Palula. In the first section (§13.1), nonverbal or nominal clauses are introduced, and the presence vs. absence of an overt copula is discussed. In the second section (§13.2), verbal clauses are introduced, with a focus on subclassification of verbs based on argument structure and transitivity. At the end of that section (§13.2.8), a tentative analysis of conjunct verbs and the syntactic role of the host element in these constructions is offered.

### **13.1 Nonverbal predicates**

Nonverbal predication corresponds to a few different forms or constructions. The particular type of construction and the copular form a nominal predication typically occur with is primarily a question of phrasal identity (whether it is another NP, an adjective phrase, or a locative expression). The choice is also related to the semantic distinction between equation/identification, property and location, as well as to tense.

Although Table 13.1 is a generalisation that does not take every single instance into account, it shows how the phrasal categories found in my data – and the semantic relationships coded by them – are mapped to certain copular forms and expressions. Each of these three types is described in detail below (§13.1.1- §13.1.3).

There is also a special affinity between copular clauses with locational expressions and two other constructions: existentials and possessives (§13.1.4). Besides the standard copular forms, some alternative copulas that occur are presented under each section.

### **13.1.1 Copular clauses with nominal predicates**

As seen in examples (1)–(3), predicate noun phrases typically occur without an overt copula in the present tense. Masica (1991: 337) mentions this feature in

### 13 Simple clauses and argument structure

Eastern NIA languages and in Sinhalese, where NP + NP as well as NP + AP clauses without a copula are normative, but he also remarks that it is not found in the more central Urdu-Hindi. Baart (1999a: 118–122), likewise, describes leftout copulas as possible in some Gawri clauses expressing identity, which also seems to be permitted, or even typical, in Khowar (own observations). The copula may be left out in Palula regardless of the phrase being used for classification (ascriptively) or identification.


The unmarked word order in such clauses is sbj - prd, as in the examples, but a reverse word order, for focus and various other discourse purposes, is also possible with this type of clause, as in (4) and (5). There are also clauses with discontinuous predicates, as in (6).

(4) *muṣṭuí* of.past *niiɡiraá* since [*mal-íim-ii* property-pl.obl-gen *kasubɡár*]prd professional [*ma*]sbj 1sg.nom 'I have been a shepherd for a long time.' (A:KEE003)

Table 13.1: Distribution of standard copular forms and expressions


13.1 Nonverbal predicates


The absence of an overt copula is, as noted above, normally confined to the present tense, whereas clauses of the above type with past tense reference, as in (7) and (8), appear with a Past tense copula *de*, reflecting a widespread pattern for languages that allow so-called copula dropping (Pustet 2003: 34; Givón 2001a: 120) as well as other South Asian or neighbouring languages where copula-less sentences are allowed or normal (Masica 1991: 339; Baart 1999a: 121). However, it should be pointed out that an equally plausible analysis (although not adopted in this work) is to regard the form *de*, which bears no formal resemblance to present tense *hin-*, as a past tense marker without any actual copular meaning attached to it, regardless of its occurrence with nonverbal predicates (as described in this section) or its auxiliary use in periphrastic tense-aspect formations (see §10.1.5).


However, the presence of a copula in the past tense is not an absolute, as pasttense copula-less sentences do occur, especially in list-like discourses such as in (9).

(9) [*mhamadíin-e* Mahmuddin-gen *putr*]sbj son *ba* top [*xaeerudíin*]prd Khairuddin [*xaeerudíin-e* Khairuddin-gen *putr*]sbj son *ba* top [*ɣeyratxaán*]prd Ghairat.Khan

'And Mahmuddin's son was Khairuddin, and Khariuddin's son Ghairat Khan.' (B:ATI016-7)

### 13 Simple clauses and argument structure

Copular clauses involving a change of state can be formed with *bhe-* 'become' (not to be confused with the superficially similar *bhe*-conjuncts, see §13.2.8), as in (10).

(10) [*ma*]sbj 1sg.nom [*míiš*]prd1 man *na* neg *de,* be.pst *bálki* however *maǰburí* necessity *ki* by [*míiš*]prd2 man *bhíl-i* become.pfv-f *de* pst 'I was not a man, but out of necessity I had become a man [uttered by a woman in a story who had on a previous occasion dressed up like

a man].' (A:UXW061)

Another (normally intransitive) verb *ɡir-* 'turn' may rarely be used in a similar way, carrying the approximate meaning 'X turns into Y'.

The above mentioned *bhe-* 'become' and *ɡir-* 'turn', along with a few other verbs, such as *dhar-* 'remain' and *yhe-* 'come', as they are used in some clauses, seem to occupy an intermediate position between copular and full verbs, and can thus be described as semi-copulas (Pustet 2003: 5–6).

A special type of subjectless copular clause is illustrated in (11)–(12), thus referring to a temporal setting introduced earlier. Often such expressions occur at the beginning of a story. The standard Past copula *de* is used.


### **13.1.2 Copular clauses with adjectival predicates**

While clauses with a predicate adjective phrase do occur without an overt copula, they seem to do so with less regularity than those with a predicate noun phrase. While it is left out in some fixed expressions, such as in greetings, a present-tense form *hin-* (agreeing in gender and number with the NP) of the standard copula is normally present in running discourse, as seen in examples (13) and (14).

13.1 Nonverbal predicates

(13) *aṛó* dist.msg.nom *ǰinaazá* corpse *asaám* 1pl.acc *díi* from *ɡhašá* take.out.imp.sg [*zhaáy*]sbj place [*naawás*]prd dangerous *hín-i* be.prs-f 'Take this dead body from us [i.e., help us], the place is dangerous.' (A:GHA039)

(14) [*moosúm*]sbj weather *típa* now [*šuy*]prd good *hín-u* be.prs-msg 'The weather is good now.' (B:VIS245)

The nearly obligatory copula dropping with nominal predicates compared to the optional or variable pattern with adjectival predicates may imply that the present-tense copula *hin-* is not entirely devoid of meaning (Pustet 2003: 8, 31, 66). This is also supported by the existential use of *hin-* (§13.1.3), but that is a matter for further research.

Examples (15) and (16) show that with past-tense reference, *de* is used, just as with the nominal predicates.


When simple-present (*hin-*) or past (*de*) time reference is not sufficient, the relevant forms of the existential (see §13.1.4) verb *háans-* 'stay, remain, find oneself, be present' is used in place of the standard copula. In (17), the longer duration of the happiness needs to be expressed with the Past Imperfective of *háans-*.

(17) [*se*]sbj 3pl.nom *heewand-á* winter-obl [*bíiḍ-a* very-pl *xušaán*]prd happy *hóons-an* stay-3pl *de* pst 'They were [remained] very happy during the winter.' (A:SHY008)

As with noun phrase predicates, adjective phrase predicates, as in example (18), also occur with *bhe-* 'become' as a copula.

13 Simple clauses and argument structure

(18) *aní* prox *čhoot-á* cheese-pl *kha* eat.imp.sg *ta* sub [*tu*]sbj 2sg.nom [*čaáx*]prd fat *bh-íiṛ* become-2sg 'Eat this cheese and you will become fat and healthy.' (A:KAT075)

Some predicative adverbial phrases can function just like predicative adjective phrases. However, in such clauses the adverbial phrase, 'alone' in (19), in actual fact denotes a property of the subject.

(19) [*be*]sbj 1pl.nom [*khilaí*]prd alone *ba* top *na* neg *de* be.pst 'We were not alone.' (A:ACR017)

### **13.1.3 Copular clauses with locative expressions**

In clauses with a predicate locative phrase, the copula is always overt. With present-time reference the standard copula is used in its present-tense form *hín-* (agreeing in gender and number with the subject NP). The semantics of such clauses, examples (20)–(21), is to declare something or someone referred to by a definite NP as present (or absent) in a certain location.


With past tense reference the Pluperfect of the verb *háans-* 'stay, remain, find oneself, be present', as in (22), is used.

(22) *dach-íi ̣* look-3sg *ta* sub [*kúṛi*]sbj woman *ba* top [*ɡhooṣṭ-á* house-obl *šíiṭi*]prd inside *na* neg *heensíl-i* stay.pfv-f *de* pst 'He looked and saw that the woman was not in the house.' (A:WOM656)

Normally the subject precedes the locative predicate phrase, but this can be reversed, as in (23), when the focus is shifted, although the order locative-subject is more typical of the closely related existential construction (see §13.1.4).

13.1 Nonverbal predicates

(23) [*se* def *baṭ-á* stone-obl *ǰhulí*]prd on [*se* def *kuṇaak-íi* child-gen *paaṇṭí*]sbj clothes *bi* also *heensíl-i* stay.pfv-f *de* pst [*teewiz-í*]sbj amulet-pl *bi* also *heensíl-i* stay.pfv-f *de* pst 'On the stone were the child's clothes and also his amulets.' (A:BRE012)

### **13.1.4 Other copular or copula-like expressions**

**Existentials.** Although similar to copular clauses with a locative expression, the function of existentials is not to specify the location of a known entity but to assert the existence of some previously unintroduced entity.

Therefore, while the subjects of local copular clauses are definite, the subjects of existential clauses are indefinite. Normally the word order is also the reverse of most local copular clauses, i.e., the locative expression precedes the subject noun phrase, and can probably not be considered a predicate in the same sense as in the copular locative expressions. The verb used is either *háans-* 'stay, remain, find oneself, be present', especially with past time reference (25), or the present form (24) – and occasionally the Past form – of the standard copula.


In many existential expressions, (26)–(28), there is no overt location at all.


13 Simple clauses and argument structure

(28) *eesé* rem *dášum* ten.obl *maǰí* among [*dúu* two *bhraawú*]sbj brother.pl *de* be.pst 'Among them there were two brothers.' (A:PAS011)

The (normally) intransitive verb *yhe-* 'come' could in some sentences be analysed as having a function similar to *háans-* with the approximate meaning 'X comes into existence', such as in (29).

(29) *tasíi* 3sg.gen *watan-í* country-obl [*qaatí*]sbj famine *yhéel-i* come.pfv-f 'There was a famine in his country.' (A:ABO019)

Another similarly functioning verb is *lhaíǰ-* 'to be found', the passive or valency-reduced form of *lhay-* 'find'.

**Possessives.** There are two types of possessive constructions, both of which are similar in structure to the existential expressions, where the possessor is expressed either as a genitive NP or postpositionally (with 'from'). There is an approximate but not absolute correspondence between the first construction, in (30)–(32), and inalienable possession and the second construction, in (33)–(34), and alienable possession.


'Each tribe and clan has its own marriage customs [lit: Every tribe's and every clan's custom of marriage is present].' (A:MAR001)

(32) *muṣṭóoi* of.past *zamaná-ii* time-gen *áak* idef *míiš-ii* man-gen *áak* idef *lhéṇḍ-u* bald-msg *putr* son *de* be.pst *maní* hsay 'Once upon a time a man had a bald son [lit: In the past there was a man's bald son].' (A:KAT001)

13.2 Verbal predicates

(33) *ma* 1sg.nom *díi* from *paiseé* money.pl *náhin-a* neg.be.prs-mpl 'I don't have any money [lit: From me money is not.].' (B:ANG008)

(34) *misrí* mason *yhóol-u* come.pfv-msg *seentá* condh *misrí* mason *díi* from *tsaṭák* hammer *hóons-a* stay-3sg 'When the mason comes he will have a hammer [lit: When the mason has come, from the mason a hammer will be present].' (A:HOW010)

**Transitive copular clauses.** The verb *the-* 'do' can be used as the transitive equivalent of the adjectival copular *bhe-* 'become', as shown in example (35).

(35) *ɡhaḍeerá* elder.obl *tas* 3sg.acc *peerišaán* worried *thíil-u* do.pfv-msg *de* pst 'The elder had made him worried.' (A:Q9.0088)

### **13.2 Verbal predicates**

### **13.2.1 Argument structure and transitivity**

Like most IA languages, there is a strict distinction between intransitive and transitive verbs in Palula. Almost without exception, a verb stem is either intransitive or transitive and cannot be ambivalent or polyvalent as far as transitivity is concerned. There is, on the other hand, fairly productive valency-changing morphology (as described in §9.5) by which a stem can increase or decrease its valency. A verb's transitivity is primarily diagnosed on the basis of two factors: 1) the absence/presence of ergative morphology in the perfective and 2) the aspectual shift between accusative alignment and ergative alignment as far as verb agreement is concerned (see §12.1).

Within each of these two main categories, intransitive verbs and transitive verbs, there is another distinction made between simple intransitive/transitive verbs and intransitive/transitive verbs with an indirect object. I am here using a very broad definition of an indirect object, as a non-nominative/non-ergative argument, usually coded by a postposition.

The four resulting argument structures (in Table 13.2) cover a large majority of all verbs. Also a few verbs displaying a non-standard pattern are discussed, as well as complement-taking verbs and the somewhat analytically challenging conjunct verbs.

### 13 Simple clauses and argument structure



### **13.2.2 Simple intransitive verbs**

The typical pattern for intransitive verbs is to take a single argument in the form of a subject noun phrase coded in the nominative: **NPsbj V**. This pattern is exemplified in (36) and (37) with the verbs 'die' and 'break', respectively.


'When that happened, all four legs of the bed broke.' (A:GHU024)

Many such verbs (Table 13.3) are process verbs and the subject noun phrase has a semantic role that (following Givón 2001a: 125) could be described as a patientof-change, whether human, animate or inanimate.

Table 13.3: Examples of simple intransitive verbs taking a patient-of-change subject


Other verbs (Table 13.4) can take as a single argument an agent subject or a patient-of-state subject.

13.2 Verbal predicates

Table 13.4: Examples of simple intransitive verbs taking an agent-subject or a patient-of-state subject


Example (38) illustrates the use of *uthí-* 'stand up, get up'.

(38) [*ma*]sbj 1sg.nom *roošnaám* morning *raɣáṣṭi* early *uth-áan-u* get.up-prs-msg 'I get up early in the morning.' (B:MOR001)

It should be noted that these verbs primarily are process verbs with a punctual interpretation, closely corresponding to English 'stand up, sit down' etc. The corresponding stative meaning 'stand, sit' is derived through a resultative construction (see §10.1.7). Compare examples (39) and (40), where the verb *bheš-* 'sit (down)' in the latter is expressed resultatively, and hence receives a stative (durative) interpretation.


### **13.2.3 Simple transitive verbs**

The typical pattern for transitive verbs is to take as arguments: a) one subject noun phrase, always coded in the nominative in the imperfective, while in the perfective some NPs are non-optionally coded in a non-nominative case (oblique or ergative), and b) one direct object noun phrase coded in the nominative or the accusative (again depending on the nature of the NP, but regardless of aspect, see §12.2). The unmarked word order is subject preceding direct object: **NPsbj NPdo V**.

### 13 Simple clauses and argument structure

This pattern is exemplified in (41) and (42) with the verbs 'eat' and 'kill', respectively.


The direct object in many transitive verbs of this type (Table 13.5) has a patient-of-change role corresponding rather closely with the role of the subject of the corresponding intransitive verb, for example *mar-/mhaaré-* 'die/kill', *phooṭé- /phooṭá-* 'break(itr)/break(tr)'.

Table 13.5: Examples of simple transtive verbs taking a patient-of-change object


Following Givón (2001a: 127), such verbs could be further classified according to several types of change: a) creation (e.g., *samá-*), b) destruction (e.g., *kha-*), c) change in physical condition (e.g., *phooṭá-*), d) change in surface conditions (e.g., *pičhá-*), e) change in internal qualities (e.g., *taapé-*). It is especially in the categories c. and e. that we find the most parallels between the direct object and the subject in typical intransitive verbs.

Some less prototypically transitive verbs (Table 13.6) also conform to this pattern, e.g., those with experiencer-subjects rather than agent-subjects as in the verbs exemplified above.

Table 13.6: Examples of simple transitive verbs taking an experiencer-subject

13.2 Verbal predicates

Example (43) illustrates the use of *paš-* 'see'.

(43) *eetíi* 3sg.rem.obl *maǰí* in [*luumée*]sbj fox.obl [*kaṭamuš-íi* Katamosh-gen *lhéṇḍ-i* bald-f *kakaríi*]do scalp *dhríṣṭ-i* see.pfv-f *hín-i* be.prs-f 'Meanwhile the fox noticed Katamosh's bald scalp.' (A:KAT152)

### **13.2.4 Intransitive verbs with an indirect object**

The next major type of intransitive verbs take, in addition to a subject, an indirect object (the term here used in a broad sense, which will become even more obvious when discussing the parallel situation with transitive verbs taking an indirect object). The indirect object occurs mostly as a postpositional phrase, but occasionally as a non-nominative noun phrase. In the typical cases the subject is an agent and the indirect object a locative: **NPsbj PP/NPio V**.

This pattern is exemplified in (44) and (45) with the verbs 'enter' and 'reach' respectively.


Many verbs of this type (Table 13.7) code events of motion.

Table 13.7: Examples of intranstive verbs with an indirect object


For some of these verbs, especially those which are already spatially defined (e.g., *whe-* and *ukhé-*), the argument status of the indirect object is somewhat

### 13 Simple clauses and argument structure

doubtful, and they may alternatively be classified as simple intransitive verbs with possible (but optionally occurring) locative complements or elaborators (Allerton 2006: 304–305).

Whether the existence of verbs whose meanings include vertical directional senses 'up', 'down', 'level', is a feature of an area comprising several languages, particularly a mountainous region such as the Hindukush, is a matter for further research. It is in any case a feature pointed out by Noonan (2003: 9) and summarised as the presence of "vertical case and vertical verbs" in some languages of the Himalayas. In Palula, spatial and vertical differentiation among pronouns and adverbs can certainly be seen as part of the same phenomenon.

### **13.2.5 Transitive verbs with an indirect object**

The contrast between simple intransitive verbs and intransitive verbs with an indirect object is more or less parallel to a contrast between simple transitive verbs and transitive verbs with an indirect object. We recognise here quite a few verbs as transitive (or causative) counterparts of the verbs presented in the section above and sometimes they also happen to be transparent morphological derivations of those (compare with *lanɡ-/lanɡá-* 'cross/take across').

To the pattern of simple transitive verbs is added an indirect object, occurring mostly as a postpositional phrase but alternatively as a non-nominative noun phrase. Again, the indirect object is a very broadly defined argument type that includes a whole range of postpositional phrases and noun phrases coded as noncore participants. In the typical case the subject is an agent, the direct object a patient whose physical location is being changed, and the indirect object a locative: **NPsbj PP/NPio NPdo V**.

This pattern is exemplified in (46) with the verb 'throw'.

(46) [*tíi*]sbj 3sg.obl *bi* also [*teeṇíi* own *zaán*]do self [*wíi-a*]io water-obl *ɡaíl-i* throw.pfv-f *hín-i* be.prs-f 'He, too, threw himself into the water.' (A:SHY062)

The movement can also be extended into the abstract realm, as with the verb *ɡaḍé-* 'take off, extract' in (47).

(47) [*ḍaaktar-á*]sbj doctor-obl [*bidráaɡu*]do ill.person [*xatrá* danger *díi*]io from *ɡaḍíl-u* take.off.pfv-msg 'The doctor brought the patient out of danger.' (A:Q9.0594)

### 13.2 Verbal predicates

Many verbs of this type (Table 13.8) code events of movement (concrete as well as abstract) caused by a human agent, some of them inclusive of a vertical specification (as some of the intransitive verbs with an indirect object).

Table 13.8: Examples of transitive motion verbs with an indirect locational object


Other events (Table 13.9), coded with verbs traditionally labelled bi- or ditransitive, can easily be seen as extensions of these, where the indirect object instead of being a locational goal (often) is a human "goal" with a benefactive role. There is, not surprisingly, some overlap between these two, so that some of the verbs may be used with a locative non-human goal as well as with a benefactive human goal.

Table 13.9: Examples of transitive verbs with an indirect benefactive object


Examples (48) and (49) illustrate the use of *phedá-* 'take, make reach' and *phrayé-* 'send', respectively.


'The ruler of Shishi has sent that enemy of yours in the direction of the prince [of Dir].' (B:ATI031)

### 13 Simple clauses and argument structure

In fact, not even *de-* 'give', the verb most typically fitting the description ditransitive in a comparative perspective, is essentially or inherently ditransitive. It is merely a transitive verb with an indirect object which has a certain preference for the goal-role, but which shows quite some variability in this respect (compare with §13.2.6, §13.2.8). Baart (1999a: 43) notes a similar range of usage and variability in the valency-pattern of GIVE in Gawri.

The pattern is extended to other verbs with "movements" (Table 13.10) that only allow for an abstract interpretation.

Table 13.10: Other transitive verbs with an indirect object


Example (50) illustrates the use of *mané-* 'tell, read'.

(50) [*so*]sbj 3sg.nom [*nis*]do 3sg.prox.acc [*har* every *áak-a* one-obl *the*]io to *man-áan-u* tell-prs-msg 'He's telling everybody about it.' (A:Q9.575)

With some verbs of this kind, an alternation in coding between the two objects is possible; each corresponds to the particular perspective taken describing what is virtually one and the same event.


In (51), the content, i.e., the 'water', is coded as a direct object (agreeing with the finite verb) and the vessel as a locative indirect object, whereas in (52), the vessel is coded as a direct object (agreeing with the finite verb) and the content in the genitive.<sup>1</sup>

<sup>1</sup> It is equally possible to interpret the genitive NP 'of water' in this sentence as a dependent

13.2 Verbal predicates

Some verbs coding transactions, such as 'ask for' in (53) and (54), seem to take a direct object and two potential indirect objects, although in actual usage it seems only one of these indirect objects at a time appears as an independent sentence argument.


### **13.2.6 Non-standard valency patterns**

**Dummy-subject or subjectless verbs.** Although Palula has a tendency to avoid subjectless clauses, there are a few instances, particularly in weather expressions (Table 13.11), when that is possible.

Table 13.11: Examples of verbs with an optional subject


These two verbs are exemplified in (55) and (56).


of 'pot', rather than an "indirect object", thus making the sentence a simple transitive one; however, that is a less likely interpretation in the sentence *tíi baalṭí wíiyii puuríli* 'She filled the bucket with water', where the direct object instead precedes the genitive NP.

### 13 Simple clauses and argument structure

In both of these cases, the verb is strictly limited to certain forms (those with masculine singular agreement) and each is, generally speaking, only used with one particular noun, *beedhré-* with *aaɡhaá* and *muč-* with *báaṣ (*B *baṣ)* 'rain'. In the latter case, the noun needs to be explicitly included in the A variety, whereas it is often left implicit in B. In any case, it is clear that even when the subject is present, it does not add anything semantically to the clause that is not already implicit in the meaning of the verb.

Some other weather expressions make use of the verb *de-* 'give' (Table 13.12), which normally is a transitive verb with an indirect object, but appears here as formally intransitive (rendering 'fall' a better gloss for *de-*). However, it would also be possible to analyse these expressions as subject-deprived, resulting in the object formally filling the subject position, thus agreeing with the verb in the imperfective as well as in the perfective.

Table 13.12: Examples of weather expressions with the verb *de-*


The expression *kir de-* is exemplified in (57).

(57) *panǰ* five *phuṭ-í* foot-pl *kir* snow *dít-u* give.pfv-msg 'We had five feet of snow.' (B:AVA200)

**Intransitive verbs with a non-nominative experiencer.** A special case of intransitive verbs with an indirect object (§13.2.4) can be said to be used for a number of constructions where a sensation is coded as the nominative subject – agreeing with the verb – whereas the human or animate experiencer appears as a nonnominative (often coded like or perceived as a locative) NP or PP. Some examples can be seen in Table 13.13. Formally the pattern is equal to §13.2.4, but due to the special status of the non-nominative argument, this particular construction deserves special treatment here. Note that the verbs used in these constructions have more generic meaning and scope when used in other, non-experiential, clauses: *de-* 'give', *ṣač-* 'adhere to, climb', *yhe-* 'come', *ḍhak-* 'touch'. Some of these are illustrated in examples (58)–(60).

### 13.2 Verbal predicates

Table 13.13: Examples of verbs with a non-nominative experiencer


Similar expressions that code an experiencer non-nominatively and the sensation nominatively, often referred to as "dative subjects", are found in various South Asian languages (both in IA and non-IA languages<sup>2</sup> ), as shown by Hook (1990b: 326–330) and Abbi (1990: 256–263). However, a construction involving (formal) causatives for expressing involuntary experience, found in Kalasha (Bashir 1990: 310) as well as in Gilgiti Shina (Hook & Zia 2005), is not at all evidenced in Palula.

In some sense, the conjunct verb *ḍhoó/ḍhoowá de-* 'see, notice/appear' (see §13.2.8) can also be described along the same lines, where the experiencer (if explicit in the clause) occurs as a non-nominative NP or a PP, while what appears or is seen is a nominative NP agreeing with the verb, as can be seen in (61).

(61) *akáaš* eleven *kaal-á* year-pl *padúši* after [*tas* 3sg.acc *the*] to [*dúu* two *ziaarat-í*] shrine-pl *ḍhoowá* host *dít-im* give.pfv-fpl

'Eleven years later he came upon two graves.' (B:FOR021)

<sup>2</sup> Outside IA, however, non-nominative experiencers are primarily found in Dravidian (Abbi 1990: 260–263), a reason for Dravidian to be suggested as a possible source of the construction (Hock 1990: 136). It is not a typical feature of Tibeto-Burman (Abbi 1990: 260; Bickel 2004: 82), and they are only found in those Tibetan languages in close contact with IA (Bickel 2004: 83, 88; Noonan 2003: 8–9).

### 13 Simple clauses and argument structure

Another construction in which the experiencer also is "demoted" to non-nominative coding is found with e.g., some bodily sensations, as in (62), where the body part occurs, in this case 'teeth', as the head of the subject NP, and the experiencer appears as the genitive modifier 'my'.

(62) *míi* 1sg.gen *dáand-a* tooth-pl *šila-yáan-a* hurt-prs-mpl 'I've got a toothache [lit: My teeth are hurting].' (B:DHE5112)

**Intransitive verbs with a postpositional object.** What can be viewed as a metaphoric extension of the pattern of intransitive verbs with an indirect object (see §13.2.4) is the pattern of intransitive verbs with a dative or patient indirect object (Table 13.14). Such indirect objects are coded with oblique case and a postposition although they in a semantic sense can be said to be primary objects.

Table 13.14: Examples of intrantive verbs with a postpositional object


Example (63) illustrates the use of *khóṇḍ-* 'talk with'.

(63) [*ma*]sbj 1sg.nom [*tas* 3sg.acc *sanɡí*]io(primary object) with *khóṇḍ-um* talk-1sg *de* pst *ta* sub *bi* also *so* 3sg.nom *kráam* work *th-íi* do-3sg *de* pst 'He continued working while I talked to him.' (A:Q9.1105)

Some conjunct verbs with *bhe-* (§13.2.8) display the same pattern: *ašáq bhe-* 'fall in love (with)', *milaáu bhe-* 'meet'.

**Transitive verb with a genitive object.** One verb, the transitive *ǰe-* 'hit, beat', is exceptional in that it codes the direct object – or at least what seems to be the only explicit object – in the genitive (instead of the usual nominative/accusative), as seen in (64).

(64) [*kuṛíina*]sbj woman.pl *ta* cntr [*támbul-am-ii*]gen drum-pl.obl-gen *ǰ-íin* beat-3pl 'And the women would beat the drums.' (A:JAN034)

13.2 Verbal predicates

That we are indeed dealing with a formally (and not only logically) transitive verb is confirmed by the ergative pattern in the perfective (i.e., non-nominative coding of the subject and non-subject verbal agreement). Here, however, another anomaly shows up, namely that the verb agrees, seemingly by default, in the feminine singular, even when the explicit object is masculine, as in (65), or plural.

(65) [*aṛé* dist *míiš-a*]sbj man-obl [*aní* prox *dúu* two *lhoók-a* small-obl *kučúr-am-ii*]gen dog-pl.obl-gen *ǰít-i* beat.pfv-f 'That man hit these two small dogs.' (A:ADJ80)

Baart (1999a: 43), who observes the same anomaly in Gawri with a verb meaning 'hit, beat' (where it happens to be one sense of the verb 'give'), suggests by pointing to a synonymous construction that this particular pattern is historically elliptic, leaving out a feminine singular noun meaning 'stroke', which is itself the direct object being modified by the genitive noun phrase.

The genitive coding does not show up when *ǰe-* is used as a verbaliser in a conjunct (§13.2.8).

### **13.2.7 Verbs with clausal complements**

The syntax of verbs that take a clause as their complement will be treated in detail elsewhere (§14.5). As a part of this chapter, only some major subtypes, as they relate to argument structure, are pointed out and exemplified.

**Modality verbs.** With modality verbs (Table 13.15), the subject of the main clause headed by a finite modality verb is coreferent with the subject of the complement clause. The latter is left unexpressed. The complement-clause verb appears either as an Infinitive or a Verbal Noun. Quite a few modality verbs are conjunct verbs (the ones in the right column).



<sup>3</sup> This modality verb, which as a non-modal verb corresponds to the imperfective stem *ṣač-* of a very polysemous verb with meanings such as 'climb, argue, be lit', also implies aspectuality and occurs in my data only in the perfective, hence the perfective stem given.

### 13 Simple clauses and argument structure

Examples (66) and (67) illustrate the use of *bha-* 'can/could, be able' and *ṣáat-* 'begin', respectively.


The scope of these verbs, the grammatical categories available to them, and the degree of their grammaticalisation varies extensively. For example, some of them are modality verbs only, as is the case of *bha-*, whereas some other verbs in this group, such as *ṣáat-*, display multiple membership. Although they sometimes occur in otherwise perfectly parallel constructions (as far as surface valency is concerned), their inherent transitivity/intransitivity and argument coding is kept intact. That is, with the transitive modality verb *bha-*, the direct object of the complement clause is coreferent with the direct object of the main clause (68), whereas with the intransitive modality verb *ṣáat-,* the direct object of the complement clause stands outside the argument structure of the modality verb (69).


Complement-taking conjunct verbs (such as *iraadá the-*) are a rather fluid and open class, with the potential of easily introducing new modality senses through direct calques from e.g. Urdu.

**Manipulation verbs.** The (human) agent-subject of a main clause headed by a manipulation verb manipulates the behaviour of a manipulee. The latter is coreferential with the agent of the complement clause. The complement-clause verb often appears as a Verbal Noun coded as an indirect object, *bainií the* in (70), of the main clause.

13.2 Verbal predicates

(70) *ɡhaḍeer-á* elder-obl *xálak-a* people-pl *bainií* go.vn *the* to *uṛíit-a* let.pfv-mpl 'The elder allowed people to leave.' (A:Q6.13.03)

There is also a special causative construction, using *ṣaawaá*, the converb of *ṣaawá-* 'dress (someone), turn on (e.g., a light or a fire)' preceded by the manipulee coded as an indirect object. The inflected finite verb in this construction could be any verb derived causatively (see §9.5.1). The causative verb in (71) is derived from *aṭé-* 'bring'.

(71) *míi* 1sg.gen *tas* 3sg.acc *teeṇíi* refl *putr-á* son-obl *ṣaawaá* manip *aṭawóol-u* cause.to.bring.pfv-msg 'I made my son bring him.' (A:HLE2589)

**Perception, cognition and utterance (PCU) verbs.** The subject of a PCU verb (Table 13.16) perceives or cognises a state or event, or utters a proposition concerning a state or event. The complement clause corresponds to what is perceived, cognised or uttered. Some such verbs are only or primarily used with a clausal complement, whereas a few of them display multiple membership.

Table 13.16: Examples of PCU verbs


There are reasons for not regarding the complement clause as a direct object, one being the non-verb-final position of the complement clause, the other being that the structure stays the same regardless of the "inherent" argument structure of the PCU verb being used. In the examples (72) and (74), *mané-* 'say' and *khooǰá-* 'ask' are both transitive verbs with indirect objects (although explicit only in (74)), whereas *buǰ-* 'understand' in (73) is intransitive. Transitive PCU verbs with a complement display masculine-singular agreement in the perfective, by default. This is similar to the "mismatch" between argument coding and the valency patterns with the intransitive vs. transitive modality verbs above.

### 13 Simple clauses and argument structure


would eat them [lit: the leopard will eat us].' (B:FOY025)

(74) [*ṣiúul-a*]sbj jackal-obl [*lhooméea* fox.obl *díi*]*io* from *khooǰúul-u* ask.pfv-msg *ki* comp [*ée* o *lhooméi…*]cpl fox 'The jackal asked the fox: "O fox…" ' (B:FOY005)

Normally the complement clause is preceded by the complementiser *ki*, as in the examples above, but there are also other (and additional) strategies available, some of them with relevance only for individual PCU verbs (see §14.5.1).

### **13.2.8 Valency patterns of conjunct verb constructions**

Conjunct verbs (see §9.6.1) are special in that the host (i.e., the element that combines with a verbaliser to form a conjunct verb) in some of them stands in no grammatical relation whatsoever to other parts of the sentence, whereas with others it functions as a direct object of the verbaliser and as such controls agreement in the perfective. Using various diagnostics, some scholars (Verma 1993: 201; Mohanan 1993: 165) have suggested that only some of these combinations are "true" conjunct verbs (particularly those of the former kind), while the rest are normal syntactic combinations of nouns and verbs.

This, however, seems somewhat oversimplified and would possibly exclude (as pointed out by Masica 1993: 160) the more productive types in the language as well as obscure the mechanisms by which this pattern has come about in the first place. Therefore I prefer to regard all of them, at least preliminarily, as conjunctverb constructions, but on different levels of lexicalisation or grammaticalisation, and with a possible subclassification into: a) conjuncts with a host contributing to the argument structure (what I refer to as "non-incorporating" below) and b)

### 13.2 Verbal predicates

conjuncts with hosts playing no role in argument structure and agreement patterns (referred to below as "incorporating"). The latter is supported by a distinction suggested by Jäger (2006: 69–74) between light-verb constructions (by him defined more narrowly than Butt 2010) and periphrastic constructions (discussed by Jäger particularly with reference to so-called 'do'-periphrasis), also acknowledging that the two constructions in fact form a cross-linguistic continuum and may share functional properties in individual languages.

### **Conjunct verbs with** *the-* **'do'**

The most productive means of forming conjunct verbs is with *the-* 'do', and we accordingly notice a wide variety of constructions and possible argument structures for conjuncts formed this way. For a large part of them, but far from all, the host is a relatively recent loan from another language. All of these constructions are transitive, which is obvious from the ergative alignment in the perfective. While Verma (1993: 201) suggests a subclassification of conjuncts into three types – lexically complex, syntactically complex and a purely analytical sequence (the latter only superficially looking like a conjunct verb) – I see no reason to make a distinction between the two latter categories. Instead I will distinguish between *incorporating conjunct verbs* and *non-incorporating conjunct verbs* only, basically following Haig (2002) in his work on complex predicates in Kurdish.

**Incorporating** *the***-conjuncts (without host-verbaliser agreement).** These conjuncts (examples listed in Table 13.17) are internally complex but externally unitary. While being phonologically two words (each carrying its own accent), the complex is syntactically equal to simple transitive verbs.

This also means that the direct object in such constructions is an argument of the complex as a whole, and the complex can be analysed as a single predicate. It is still possible, however, for the host to be separated from the verbaliser by, for instance, a negation. All conjunct *the*-verbs with an adjective host, such as *tanɡ the-* exemplified in (75) – or at least what is an adjective in other uses or in a donor language – belong to this particular type. What is more interesting, however, is that we also find complexes of this type formed with a noun as a host (at least in a historical-etymological sense) that in no way participates as an NP in the argument structure. The conjunct *káaṇ the-* in (76) is an example.


Table 13.17: Incorporating *the*-conjuncts


I suggest that a host element, which may be homonymous with another noun in the language – such as *káaṇ* 'ear' in *káaṇ the-* 'listen' – with which it is historically related, has in fact ceased to constitute an NP in this construction. And in

### 13.2 Verbal predicates

in *široó the-* 'start' – it may never even have been interpreted as an NP from the very start (which is probably the case with the English origin element *rekaáḍ* in *rekaáḍ the-* 'record' as well). In such a case it does not matter how much of a noun it is in the donor language. This process of lexical incorporation (Verma 1993: 203) is simply a convenient way of deriving new verbs, combining the semantics (or some aspects of it) of the host element with the "verbness" of the verbaliser with the resulting verb behaving, as far as argument structure is concerned, in no way different from any other simple transitive verb (compare with Haspelmath 1993: 286 for Lezgian). In some cases it is not entirely unlikely that the complex as a whole is a calque from another language, without even going through a process of lexical incorporation, as there are several close parallels to Palula conjuncts in Urdu as well as in Pashto. The conjunct verb of this type as a whole is lexical in nature (Verma 1993: 199).

**Non-incorporating** *the***-conjuncts (with host-verbaliser agreement).** In the next type of conjunct verb, the verbaliser does agree with the nominal host (in the perfective), thus, if holding on to the conjunct idea, meaning that the predicate agrees with an element internal to itself (Mohanan 1993: 168). Examples of conjuncts of this kind are given in Table 13.18.

As with the former type, many of the host elements are transparent loans, such as *madád* in (77) and (78), but there are also a good number of hosts with a longer history in the language. Quite a number of those, however, are not actually used, or have any clear homonyms, outside their conjunct use.


One solution to the somewhat puzzling situation of the host being part of the argument structure is to analyse the noun in the complex as itself assuming the role of a predicate (Verma 1993: 204–212; Mohanan 1993: 164–170). This predicate chooses a suitable verbaliser to team up with and contributes itself to the total number of arguments of the clause, but at the same time it assigns roles and case marking to its "own" arguments. I hold, however, that it is sufficient and far easier to see the valency pattern displayed (e.g., the host coded as a direct object and the

### 13 Simple clauses and argument structure


Table 13.18: Non-incorporating *the*-conjuncts

occurrence of indirect objects, i.e., certain postpositional phrases) as a property (or lexical specification) of each individual construction in its entirety.


As far as perfective agreement, the verbaliser agrees in the same manner with the host as any transitive verb does with its direct object, while the subject is ergatively coded. A remaining argument (which in that case could be considered a "logical object" vis-à-vis the host) has to be coded as an indirect (postpositional) object, e.g., *ɣaribaanóom the* 'to the poor' in (79) and *tusaám ǰhulí* 'on you' in (80). The postposition used for that argument is (as noted above) specified by each conjunct. Some of these are displayed in Table 13.19.

13.2 Verbal predicates


Table 13.19: Postpositions in the valency pattern of some *the*-conjuncts

The complexes considered "purely analytical" sequences by Verma (1993: 201) and not "true" conjuncts only differ from the ones already discussed in that they do not include a postpositional phrase in their valency pattern. To treat them as different phenomena on the basis that they are logically intransitive, i.e., not having a "logical object" in addition to the direct object-coded "formal" host, is not a very convincing argument; they are in fact only "logically intransitive" in the trivial sense of translation equivalence (as pointed out by Masica 1993: 157). These would in Palula be conjuncts such as *kráam the-* 'work', *nimóos (*B *nimáas) the-* 'pray', exemplified in (81), and *hiǰrát the-* 'migrate'.

(81) [*ak* idef *buzrúɡ*]sbj wise.man *teeṇíi* refl *ɡhooṣṭ-á* house-obl *maṇḍayí* porch.obl [*nimáas*]do prayer *th-íi* do-3sg *de* pst 'A wise man was praying on the porch of his [own] house.' (B:THI001)

Just as with the other non-incorporating conjuncts, the verb agrees with the nominal host in gender/number (in the perfective), the latter thus being part of its argument structure in the form of a direct object.

Looking at the conjunct *hiǰrát the-* 'migrate' in (82), we also notice that this indeed specifies a particular valency pattern, including a source (realised by *taayúu* 'from there' in the example) in addition to the subject and the direct object.

(82) [*tíi*]sbj 3sg.obl *taayúu* from.there [*hiǰrát*]do migration *thíil-i* do.pfv-f 'He migrated from there.' (B:ATI002)

The distinction suggested by Verma becomes even more superficial when considering a conjunct like *bhootíi the-* 'plough', which can code the non-nominative argument denoting 'field' alternatively as a locative NP (oblique case), as in (84), or a postpositional phrase, as in (83).

13 Simple clauses and argument structure


Another argument for treating all these instances of *the*-conjuncts as a single phenomenon is the means of passivisation/valency reduction. While other transitive nouns are amenable to morphological passivisation with the affix *-íǰ* (see §9.5.2), *the*-conjuncts – regardless of subtype – are passivised or valency-reduced by substituting *the-* for *bhe-* 'become', the other main verbaliser (see below), as shown in (85)–(87).


### **Conjunct verbs with** *bhe-* **'become'**

The other productive means of forming conjunct verbs is with *bhe-*, for which some examples are given in Table 13.20. Here we do not observe the same variety of construction types and argument structures as with the *the*-conjuncts, the most obvious reason being that all complexes with *bhe-* are intransitive and thus bound to have fewer potential arguments. For the most part, but again not exclusively, the host element is often an element copied from another language.

While verbaliser-host agreement seemed to be relatively common with *the*conjuncts, this is, as far as I have been able to observe, not found with *bhe*conjuncts outside of passive constructions derived from *the*-conjuncts with host agreement exemplified above in (85)–(87).

### 13.2 Verbal predicates


Table 13.20: *bhe*-conjuncts

Although a number of these conjuncts are made up of the same host element as some of the *the*-conjuncts, such as *teér bhe-* in (88) and *ǰamá bhe-* in (89) they are not simply derived from these, as the meaning of the corresponding *bhe*-conjunct is not necessarily passive, but rather they stand in the same kind of relationship to one another as any other transitive/intransitive pairs with or without a common base segment.


The historical-etymological identity of the host element as far as part of speech is concerned is not unitary: a) *teér* originates in Pashto where it is an adjective 'passed, gone by, done, accomplished, spent (as time), over, lapsed, that has been', and even there is used in a conjunct, in some tenses morphologically incorporated into the verb (*teredal*), with a meaning very similar to the Palula conjunct;

### 13 Simple clauses and argument structure

b) *ǰamá* is of Perso-Arabic origin and is listed as a noun or verbal noun in Urdu, Pashto as well as in Persian (the three most likely donor candidates) with an approximate meaning 'collection, assembly, congregation, whole'; c) *raál* is used in Palula outside of this construction as an adverb meaning 'high'. The crucial point is that it is not interpreted as an NP, even when the origin happens to be nominal.

As with many of the *the*-conjuncts, some *bhe*-conjuncts are also specified for an argument structure involving postpositionally coded participants, such as 'Machoke' in (90), and 'he' in (91). Some of these indirect objects would fit the description "human dative".


### **Conjunct verbs with** *de-* **'give' and other verbalisers**

The situation is in many ways much more complex when it comes to conjunct verbs formed with verbalisers other than the above mentioned *the-* and *bhe-*, as they tend to fade out into the area of idiomatic expressions and it is difficult to specify any grammatical conditions for choosing any of them instead of the transitively triggered *the*-conjuncts or the intransitively triggered *bhe*-conjuncts. While DO and BE/BECOME verbalisers in IA languages do not to any larger extent contribute to the semantics of the complexes, that is less true when it comes to other verbalisers (Gambhir 1993: 78–79), "shading off", as Masica (1993: 157) phrases it, through GIVE and TAKE into individual combinations of certain verbs and nonverbal elements of a mostly idiomatic character.

Nevertheless, some of the observations and categorisations outlined above for *the-* and *bhe-* are in some ways relevant even for conjuncts with *de-* 'give'. As already noted above, *de-* is already as a simple verb rather polysemous and as such is associated with a few different valance patterns, something that further blurs the picture.

**Incorporating** *de***-conjuncts (without host-verbaliser agreement).** These conjuncts (examples given in Table 13.21) are internally complex but externally unitary, as we saw with the incorporating *the*-conjuncts above.

13.2 Verbal predicates


Table 13.21: Incorporating *de*-conjuncts

Often no exact meaning can be assigned to the host element, and the basic meaning of *de-* has been considerably bleached, with the result that the complex meaning seldom can be deduced or guessed from its individual components. The two conjuncts in Table 13.1 are exemplified in (92) and (93).


These complexes are intransitive, but other non-nominative arguments may appear, which also opens up the possibility for analysing e.g., *ḍhoó/ḍhoowá de*as taking a non-nominative experiencer subject (either explicitly or implicitly), see §13.2.6.

**Non-incorporating** *de***-conjuncts (with host-verbaliser agreement).** In other *de*-conjuncts (Table 13.22), the verbaliser does agree with the nominal host (in the perfective), and is therefore an argument of the clause.

It is obvious from (94) and (95) that the host element in this case is perceived as an NP, even when it does not appear synchronically as a noun in the language outside of the particular conjunct. (Although *póo* is related to the word for foot in the protolanguage, *pā́da-* (Turner 1966: 8056),<sup>4</sup> the word in current use with the meaning 'foot' is *khur*.)

<sup>4</sup> Compare with Palula *naalpoó* 'barefoot'.

### 13 Simple clauses and argument structure

Table 13.22: Non-incorporating *de*-conjuncts



These complexes are transitive with an indirect object, although what is coded as the indirect object is semantically really the (most affected) patient.

**Extended uses of GIVE or non-incorporating** *de***-conjuncts?** For one group of *de*-conjuncts, which in a formal sense behaves no differently than the conjuncts above, it can be argued that the basic meaning of *de-* as 'give' is preserved to such an extent that it would be trivial to consider them conjuncts. But we can also argue that these constructions (where the receiver is coded with the postposition *the*) happen to coincide with the basic semantics of sentences where a simple verb *de-* is used to describe how concrete objects are being handed over to someone. Examples of such conjuncts (or standard combinations of an abstract object and GIVE) are *húkum de-* 'order' (96), *daawát de-* 'invite', *xabaár de-* 'inform' and *baát de-* 'tell' (97).


13.2 Verbal predicates

Most conjuncts of this type, such as *daawát de-* 'invite' in (98), can also take a clausal complement, thus function as PCU-verbs (see §13.2.7).

(98) [*deeúl-ii* Dir-gen *xálak-am*]sbj people-pl.obl [*daawát*]do invitation [*dít-i*] give.pfv-f [*atsharíit-am* Ashreti-pl.obl *the*]io to *ki* comp [*muqaabilá* contest *th-íia*]cpl do-1pl 'The people from Dir invited the Ashreti people to have a competition with them' (A:CHA001)

**Conjuncts with other verbs.** There are a few other combinations of a verb and a host element that may be considered conjuncts (Table 13.23), some of them already pointed out as participating in non-standard valency patterns (§13.2.6). In particular the transitive *ǰe-* 'hit' used as a verbaliser should be mentioned.


Table 13.23: Conjuncts with other verbs

Most, or maybe all, such conjuncts are of the non-incorporating (host-agreeing) type, as can be seen in (99) and (100).

(99) *haṛíiwee* in.there [*peereṇíi-a*]sbj fairy-pl *har* every *deés* day [*laambú*]do bath *ǰ-éen* hit-3pl *de* pst 'Every day the fairies used to take a bath in it.' (B:FOR029)

(100) [*míiš-a*]sbj man-pl *ba* top [*axsí*]do "akhsi" *ǰ-íin* hit-3pl 'And the men would play "akhsi".' (A:JAN035)

However, this again borders on the idiomatic, and these possible conjuncts are in any case less central and productive than the ones formed with *the-* and *bhe-*.

## **14 Complex constructions**

### **14.1 Introduction and overview**

The absolute binary distinction between coordinate and subordinate clauses is not very helpful when describing complex (i.e., multi-clause) constructions in Palula. Instead it makes more sense to regard inter-clausal dependence as a continuum, as suggested by Givón (2001b: 327–328), ranging from the least dependent coordinate clauses expressing cross-event coherence through various intermediate stages – including chained clauses and adverbial clauses – to clauses with a high degree of dependence and event integration, as exemplified in some types of complementation.

In the following sections, this gliding scale is reflected, both in the way typical functions relate to particular constructions and in the way different subtypes are categorised within these sections. In §14.2, on coordination, the constructions described are to a large extent symmetrical, whereas §14.3 exemplifies a type I call clause chaining. The latter shows a higher degree of asymmetry but not enough to be labelled subordinate, and therefore an intermediate category "cosubordinate" seems better suited. This naturally leads over to the structurally closely related contents of §14.4, which deals with clauses with adverbial functions, themselves displaying a continuum from a low degree of asymmetry to clearly asymmetrical constructions involving nominalised clauses. §14.5, on complementation, contains the same kind of internal continuum, ranging from a comparatively low degree of event integration to clause union. §14.6 covers relative clauses and presents a special analytical challenge as far as the distinction between parataxis and embedding is concerned, and here too we witness a continuum from clauses displaying mere discourse coherence to those involving various types of nominalisation.

### **14.2 Coordination**

Although coordinators in the traditional sense of the term are used in Palula, there seem to be stronger constraints on their use than in the more familiar Eu-

### 14 Complex constructions

ropean languages. Perhaps the alternative strategies for coordination, such as juxtaposition, discourse particles and suffixes, represent the more native means for conjoining phrases and sentences. However, as will be seen in the next main section (§14.3), many sequences that in English are coded with the coordinating conjunction *and* correspond to clause chaning in Palula.

In §14.2.1–§14.2.4 I follow the semantic subdivisions and their associated terms as used and discussed by Payne (1985): *conjunction* (p and q), *postsection* (p and not q), *presection* (not p and q), *disjunction* (p or q) and *rejection* (not p and not q).

### **14.2.1 Conjunction**

Palula conjunctions make use of a number of different coordinating devices. Following Haspelmath's (2007) classification, they include asyndesis (juxtaposition), monosyndesis (using a single marker), as well as bisyndesis (using two markers).

**Clitical** *ee***-conjoining.** A postposed conjunction clitic *ee* is primarily used to conjoin two noun phrases, as can be seen in examples (1)–(3). It is cliticised to the right of the first constituent, following on any other word class specific inflections.


Infrequently the use of this conjunction is extended to the conjoining of two constituents of other kinds, such as two verb phrases that are intimately linked and expressing something symmetrical or simultaneous, as in (4) and (5) (compare with §14.4.1 and simultaneity expressed with a Verbal Noun followed by what is possibly the same marker *ee*). This should, however, not be seen as a matter of sentential coordination in a strict sense, as the verb phrases have a subject in common.

14.2 Coordination


**Juxtaposition.** Conjunction of more than two noun phrases is often expressed by juxtaposition, in a list-like manner, (6)–(7), but can occasionally be used with two noun phrases only, as in example (8).


Juxtaposition is also applied to closely interrelated verb phrases or clauses, for instance when a particular verbal action or event is identical for several subjects, as in (9), or when the verb phrases, as in (10), have an identical subject as well as an identical recipient or beneficiary. A sequence of events with the same subject, such as the one in (11), would in most cases be coded with clause chaining (§14.3), but for some reason, maybe reflecting how upset the speaker is, juxtaposition is used instead.

### 14 Complex constructions


**The conjunction** *o/oór***.** A special conjunction, *o* alternatively *oór*, is used sparingly and sometimes in addition to other conjunction strategies. It is not wholly unlikely that they owe something to Pashto and Urdu influence, respectively, being close in form and function to the most important conjunctions in each of these languages. The latter form is, in particular, found primarily in data obtained from younger persons (in A) and in written discourse (in B).

As a conjunction between noun phrases, it is mainly found where more than two noun phrases are conjoined, as in (12) and (13), and then the position is always and only between the last two phrases.

(12) *se* def *míiš-a* man-obl *dúi* other *dees-á* day-obl [*baačaá* king *wazíir* minister *o* and *dúi* other *xálak-a*] people-pl *samaṭ-í* gather-cv *ilaán* announcement *thíil-i* do.pfv-f 'Next day the man called the king, the prime minister and other people

together and made an announcement.' (A:UXW060) (13) *mheeríl-ii* die.pptc-gen *pahúrta* after [*tasíi* 3sg.gen *híṛu* heart *tasíi* 3sg.gen *ǰhanɡaár* liver *oór* and *tasíi* 3sg.gen *aandáara*] intestines *ɡaḍ-í* take.out-cv *khóol-a* eat.pfv-mpl

'After he had killed her, he took out her heart, her liver, and her intestines, and ate them up.' (A:WOM651)

14.2 Coordination

As we saw above, this kind of stringing together of several noun phrases can also be done by mere juxtaposition. In fact, when a dialect adaptation was made (from B to A) of a story that (7) is taken from, the same sentence was equipped with the conjunction *o* between the two last noun phrases.

A conjunction word is also used sometimes to conjoin closely interrelated verb phrases or same-subject clauses, as the ones exemplified in (14) and (15). Often, the second proposition or phrase provides additional information about the action or event mentioned in the first. The two are sometimes but not necessarily co-temporal.


Rather infrequently a conjunction is also used to link two different-subject clauses, such as the ones in (16).

(16) *nhiáaṛee* near *ba* top [*áaṇc-ii ̣* raspberry-gen *áak* idef *muṭ* tree *heensíl-u* dwell.pfv-msg *hín-u* be.prs-msg *o* and *áaṇc-ạ* raspberry-pl *ba* top *pač-í* ripen-cv *šiiy-í* fall-cv *bi-áaṭ-a* go-ag-mpl *bhíl-a* become.pfv-mpl *hín-a*] be.prs-mpl 'Nearby a raspberry bush grew and its berries had become ripe.' (A:KAT128)

However, as far as temporal sequencing is concerned, the overwhelmingly most common ways of conjoining clauses are by using cosubordinate converb clauses and subordinate *ta*-clauses in so-called "clause chains" (as described in §14.3).

There are also examples of predicate adjective phrases being thus conjoined. An example can be seen in (17).

### 14 Complex constructions

(17) *so* 3sg.nom *bíiḍ-u* very-msg [*uxiaár* wise *o* and *hileéṛ*] brave *de* be.pst *maní* hsay 'He was very wise and brave.' (A:UXB004)

**Conjunction with the "separating" particle** *bi***.** An added emphasis on the separateness of two or more conjoined units is given by means of a particle *bi* (approximately 'too, also'), as in examples (18)–(20). It is primarily added immediately after the constituents in focus, which in most cases are noun phrases, whereas the coordination itself is sentential rather than phrasal (which is not obvious with copular clauses lacking an overt copula, as in (18)). The particle used this way is added after each of the conjoined constituents, rendering it 'X as well as Y' or 'both X and Y'. The use of bisyndesis for emphatic coordination is known from a number of other languages (Haspelmath 2007: 15–17).

(18) *ma* 1sg.nom [*tarkaáṇ* carpenter *bi* also *misrí* mason *bi*] also

'I'm a carpenter as well as a mason.' (A:HOW009)

(19) *ma* 1sg.nom *ba* top *uth-í* stand.up-cv *ba* top [*toobaák* gun *bi* also *heensíl-i* be.present.pfv-f *de* pst *khanɡaár* sword *bi* also *heensíl-u* be.present.pfv-msg *de* pst *ṭhónɡi* axe *bi* also *heensíl-i* be.present.pfv-f *de*] pst 'I woke up, and there I had a rifle as well as a sword and an axe.' (A:HUA021- 2)

(20) *pha-íi* girl-gen *báabu* father *ǰhaamatreé* son.in.law.obl *díi* from [*xarčá* expenses *bi* also *dawa-áan-u* ask.for-prs-msg *ǰandeé* goat.pl *bi* also *khéli* numerous *dawa-áan-u*] ask.for-prs-msg 'The girl's father demands expenses [to be paid] as well as numerous goats.'

(A:MAR032-3)

The particle *bi* is sometimes also used in combination with coordinating *ee*, as in (21), or the conjunction *o*, as in (22).

(21) *aṛé* 3fsg.dist.nom [*šišáki* "shishaki" *bi* also *man-áan=ee* say-prs-mpl=cnj *ruikúṛi* "ruikuli" *bi* also *man-áan-a*] say-prs-mpl 'We call her a "shishaki" or a "ruikuli".' (A:PAS110)

14.2 Coordination

(22) *dhrúuk-a* stream-obl *be* go.cv *dach-íin ̣* look-3pl *ta* sub [*manɡái* water.pot *čhoor-í* put-cv *hín-i* be.prs-f *o* and *áak* idef *rumiaál* handkerchief *bi* also *čhoor-í* put-cv *hín-i*] be.prs-f 'Going to the stream, they saw the water pot and a handkerchief lying there.' (A:SHY059)

In Kohistani Shina (Ruth Laila Schmidt, pc; Schmidt & Kohistani 2008: 252) a particle *-ɡa* has a function comparable to Palula *bi*, but its scope is apparently much larger than that of the Palula particle, as it conjoins various types of phrases and clauses (much like Palula *-ee*), not necessarily with an additional "separating" semantics.

**Conjunction with the "adversative" construction** *ta … ba***.** An adversative meaning with a semantic contrast implied, translatable with 'but, whereas, while' (but sometimes 'and' is better still), is indicated with the use of a contrast particle *ta* placed immediately following on the first (and thus topicalised) constituent of the first clause,<sup>1</sup> and another particle *ba* (also with other discourse functions, see §15.4) placed immediately after the first (and similarly topicalised) constituent of the next clause with which the first-mentioned is contrasted. As is apparent from examples (23)–(25), the coordination itself is, like the use of *bi*, sentential rather than phrasal.

	- pst

'The men were shaking them [the walnuts] down, while the women were collecting them.' (A:JAN017)

(24) *aaɡhaá* sky *ta* cntr *dhúura* distant *dharán* ground *ba* top *ḍanɡ* hard

'The sky is far and the ground is hard.' (B:PRB009)

<sup>1</sup> The contrast marker *ta* is almost exclusively used along with the topic marker *ba*. In some of the cases where it does occur on its own, it is not entirely clear whether it should instead be understood as an unusual use of the homophonous general subordinator *ta*.

### 14 Complex constructions

(25) *so* 3msg.nom *ta* cntr *bač* saved *bhíl-u* become.pfv-msg *trúu* three *ǰáan-a* person-pl *ba* top *hiimeel-í* glacier-obl *híṛ-a* take.away.pfv-mpl

'He was saved, but three of them were lost in the avalanche.' (B:AVA214)

The construction *ta…ba* is sometimes, as in (26), used along with the coordinating marker *ee*.

(26) *huṇḍ* above *ta* cntr *chíitr= ̣ ee* field=cnj *bhun* below *ba* top *ɡhaawaáz* stream.bed *de* be.pst 'The field was above and the stream down below.' (A:JAN082)

**Conjunction with the "adversative" particle** *xu***.** Another type of adversative, with a denial of expectation implied, is formed with a particle *xu* (also functioning as a marker of evidentiality).

(27) *ɡhaḍeerá* elder.obl *mána* prevention *thíil-u* do.pfv-msg *xu* but *so* 3msg.nom *áahola* repeatedly *tool-íi* measure-3sg *eé* ampl<sup>2</sup> *de* pst

'The older [brother] forbid him to do that, but he [the younger one] continued his measuring.' (A:DRA019)

### **14.2.2 Presection and postsection**

The "adversative" particle *xu* (see above) is also used together with the negation *na* in typical sentential pre- and postsection (as defined in §14.2) constructions.

**Presection with** *na…xu***.** In this construction, exemplified in (28), the particle *xu* occurs between the two conjoined parts, and the negation word *na* in its regular preverbal position in the first part.

<sup>2</sup> In my material the amplifier is a rarely occurring post-verbal marker. It appears, at least in this sentence, to have a continuative meaning. It is probably identical to the (*eé*) of exclusiveness or emphasis as it occurs postposed to pronouns or numerals.

14.2 Coordination

(28) *se* 3pl.nom *bi* even *čúur* four *páanǰ* five *bhe* become.cv *ma* 1sg.nom *na* neg *ṭinɡ* challenge *th-áai* do-inf *bhóon* be.able.3pl *de* pst *xu* but *míi* 1sg.gen *se* def *preṣ* mother.in.law *phed-í* arrive-cv *ba* top *chéeṇi ̣* stick *ɡhin-í* take-cv *ǰ-íi* hit-3sg *de* pst *seentá* condh *bhíi-um* be.afraid-1sg *de* pst 'Not even these four or five people could hold me, but when my motherin-law came and started hitting me with a stick, I became afraid.' (A:HUA123-4)

**Postsection with** *xu…na***.** In this construction, exemplified in (29), the particle *xu* occurs between the two conjoined clauses or propositions, and the negation word in its regular preverbal position in the second proposition.

(29) *sóon-a* pasture-obl *ba* top *so* def.msg.nom *phoó* boy *koošíš* attempt *th-áan-u* do-prs-msg *xu* but *waxt* time *milaáu* meeting *na* neg *bh-áan-u* become-prs-msg

'At the high pasture, the boy tried to find the time to meet her, but he couldn't.' (A:SHY042)

**Postsection with final** *na***.** Postsection can also be expressed with a sentencefinal negative particle *na*, as in (30) and (31), without the aid of *xu*.


'If it is put too low people will hit their head, but if it is put too high nobody will reach up.' (A:HOW069-70)

### **14.2.3 Disjunction**

**Juxtaposition.** The least emphasised kind of disjunction is by mere juxtaposition, normally of two numerals (as in the juxtaposition of 'four' and 'five' in

### 14 Complex constructions

(28) above) or of two noun phrases with numeral attributes, as in (32), expressing an approximate figure rather than constituting an alternative in an absolute sense.

(32) *díiš-e* village-gen *xálak-a* people-pl *ǰáma* collection *bhe* become.cv *kir* snow *kareé=ɡale* when=ever *bíiḍ-u* much-msg *dít-u* give.pfv-msg *síinta* condh *ba* top [*panǰ* five *phuṭ-í* foot-pl *ṣo* six *phuṭ-í*] foot-pl *kir* snow *dít-u* give.pfv-msg *síinta* condh *ba* top 'The village people used to gather when much snow had fallen, like 5-6 feet of snow.' (B:AVA198)

**Phrasal disjunction with** *yaá***.** A slightly more emphasised disjunction uses *yaá* between the phrases, as seen in (33).

(33) *eesé* rem *phaíi* girl.gen [*bhróo* brother *yaá* or *máamu*] uncle *koó* who *eesé* rem *ṭeem-íi* time-gen *haazír* present *na* neg *heensíl-u* stay.pfv-msg *heentá,* condl *tasíi* 3sg.gen *mux-íi* face-gen *nikh-eeṇḍeéu* come.out-oblg *bh-áan-u* become-prs-msg 'If the girl's brother or her uncle wasn't present at that time, he has to be greeted.' (A:MAR047-9)

**Sentential disjunction with** *yaá=ba***…***yaá=ba***.** Although there are few examples, it seems a construction with a repeated *yaá* combined with the particle *ba* (probably forming one phonological word), as in (34), is preferred for sentential disjunction, possibly with an added emphasis on the mutual exclusivity of the alternatives thus presented.

(34) *típa* now *lo* dist.nom.msg *šay* thing *phed-í* arrive-cv *asaám* 1pl.acc [*yaá=ba* or=top *kh-óo* eat-3sg *yaá=ba* or=top *na* neg *kh-óo*] eat-3sg 'Now this thing may or may not come and eat us.' (A:GHA063)

**Sentential and phrasal disjunction** *ki***.** A semantically equivalent construction presenting mutual exclusivity is also found, as in (35)–(36), with a particle *ki*.

14.2 Coordination


'I couldn't recognise the thing, whether it was a leopard or a bear or something else.' (A:HUA038-9)

Kohistani Shina uses a word borrowed from Punjabi, *paa∼wee∼* 'or', to express disjunction (Ruth Laila Schmidt, pc).

### **14.2.4 Rejection**

**Rejection with** *na…na***.** Rejection can be expressed simply with a phrase- or clause-initial negation *na*, as exemplified in (37) and (38).

(37) *na* neg *zinaawur-á* beast-obl *tas* 3sg.acc *the* to *ɡa* what *asár* effect *thíil-i* do.pfv-f *de* pst *na* neg *ɡhrast-á* wolf-obl *thíil-i* do.pfv-f *de* pst

'No wolf or any other wild animal had touched him.' (A:GHA014)

(38) *se* 3fsg.nom [*na* neg *kasíi* anyone.gen *xaadí* happiness *dach-éen-i ̣* look-prs-f *na* neg *kasíi* anyone.gen *marɡ* death *dach-éen-i ̣* ] look-prs-f

'She doesn't have a care for anyone's happiness or sorrow.' (A:KEE005-6)

Two other strategies available for expressing rejection are both combinations of negation with conjunctions already discussed above.

**Rejection using "adversative" conjunction and negation,** *na ta…na ba***.** It should be noted that the word order (the particle vis-à-vis the contrasted constituent) is the reverse compared with the one found in the conjunctive construction with *ta…ba*. It could also be classified as an "emphatic negative coordination" (Haspelmath 2007: 17–19). Examples are given in (39) and (40).

### 14 Complex constructions


'Then, neither did he pray, nor did he utter any curses.' (A:PIR038-9)

As can be seen in (41), the strategy also occurs in combination with the conjunctive marker *ee*.

(41) *méeǰi* between [*na* neg *ta* cntr *ṣoo-íi* king-gen *tarap-íi* direction-gen *ɡa* any *faaidá=ee* benefit=cnj *na* neg *ba* top *barawulxáan-ii* Barawul.Khan-gen *tarap-íi* direction-gen *ɡa* any *faaidá*] benefit

'There were no benefits attached, neither from the king's side, nor from Barawul Khan.' (A:JAN007-8)

Rejection using "separative" conjunction and negation, *bi na…bi na*, is illustrated in (42)–(43).


Kohistani Shina expresses rejection with *nee∼…nee∼* 'no…no', while the regular clause negation in this variety is *na* (Ruth Laila Schmidt, pc).

### **14.3 Clause chaining**

One of the more salient features of Palula discourse (especially of a narrative kind) is its use of clause chaining, thus fitting into the category of "chaining" languages, as described by Thompson, Longacre & Hwang (2007: 242) and Longacre

14.3 Clause chaining

(2007: 374–376). Such clause chains express sequences of events or actions carried out and consist of one or several non-final clauses (note the distinction made between (non-)*final* and (non-)*finite*), followed by one final clause. Each non-final clause relates to the one immediately preceding it and the one following it, but not necessarily to the final clause. I have adopted the term *cosubordinate* (see explanation below) to describe this type of clause linking which is somewhat intermediate between coordination and subordination.

### **14.3.1 Same-subject chaining**

In examples (44)–(46), the construction used in the non-final clauses is a nonfinite converb (cv), a verb form without any participant agreement, and the subject left implicit. A literal translation of (44) would render something like 'The father, having killed a goat, having removed its skin, put the skin on him'.


(A:DRA016)

As pointed out by Givón (2001b: 327-8), although traditional grammar has made a binary distinction between dependent/subordinate adverbial clauses, on the one hand, and independent/coordinate main clauses, on the other, it is evident that dependency, as well as subordination, is a matter of degree rather

### 14 Complex constructions

than being discrete properties of clauses (compare with the discussion in Cristofaro (2005)). Haspelmath (1995: 20–27), following the same reasoning, suggests a *cosubordinate* category intermediate between coordination and subordination, which would thus capture the status of non-final verbs in the above-mentioned chains, in some traditions also described as *medial* verbs. Whereas the connection between a cosubordinate clause and another cosubordinate clause, or between a cosubordinate clause and its corresponding main/final clause, often will have to find its translation equivalent in a coordinate clause (for instance in English), the non-final clause is still to a large extent dependent on the independent clause for its TMA specification or participant reference.

However, an argument in favour of still viewing Converb clauses as subordinate would be that they sometimes seem to appear inside (i.e., embedded into) the final clause. Example (44) is ambiguous in this respect, since we can analyse the ergatively marked *báaba* 'father', either as the explicit subject of the first nonfinal transitive clause (as is suggested by the square brackets) or as the explicit subject of the final transitive clause.

In example (46) on the other hand, the final clause is transitive and the first converb is intransitive, and therefore the ergatively marked subject must be analysed as the subject of the final (and finite) verb, and therefore the first Converb clause in its entirety has to be analysed as embedded into the final clause (as is also the second Converb clause, which has an instrumental reading, see §14.4.1). It is at this point that clause chaining shades over into the realm of more asymmetrical adverbial functions.

In (45), however, sentence-initial *ti* 'he' cannot be the subject (S) of the intransitive final verb 'go'. Since it is marked for ergative alignment, it must be the subject (A) either of transitive *whaalé-* 'take down' or transitive *ɡhin-* 'take', both of these non-final converbs in this sentence. Simple embedding is therefore not really the case here, and again cosubordination seems a suitable categorisation.

Some overlap between medial verbs – as used in chained or so-called cosubordinate clauses – and converbs coding more typical adverbial functions is indeed to be expected (Haspelmath 1995: 26), and we find that the very same mechanism is put to work in many clauses corresponding to distinct adverbial-clause types (see §14.4.1–§14.4.5) in the familiar European languages, with a number of semantic relationships implied between the dependent clause and the main clause.

14.3 Clause chaining

### **14.3.2 Different-subject chaining**

While the chaining exemplified above expresses sequencing of events or actions with one and the same subject, another construction type is used when there is a subject switch after a non-final clause: a non-final clause as the ones found in (47) and (48), in which a finite verb is followed by the general subordinator *ta*, which is in its turn followed by another (non-final or final) clause.


Sequences of two or more non-final clauses can include same-subject (SS) as well as different-subject (DS) clauses, as can be seen in (49) and (50).


Note that, while the converbs in the SS constructions do not show agreement, the verbs in the DS constructions do. In example (49), the perfective verb *ṣaawóolu* agrees in (masculine) gender and (singular) number with the direct object *púustu*, and in example (50), the verb *bhíli* agrees in (feminine) gender with the subject *dhií*.

### 14 Complex constructions

In a great many cases of chaining with *ta,* the non-final clause is most naturally translatable as a 'when'-clause in English, and a number of other semantic relationships can be implied as well, the latter something that will become obvious when discussing different constructions expressing adverbial functions (§14.4). As was the case with SS-chaining using the Converb construction, DS-chaining by means of the *ta*-construction also shades over into even more asymmetrical and more typical adverbial functions.

### **14.4 Clauses with adverbial functions**

The types of clauses using the Converb construction (SS) and the *ta*-construction (DS) would certainly qualify as "absolutive" adverbial clauses following Thompson, Longacre & Hwang's (2007: 264–266) definition, due to their wide scope and their lack of explicit signals as to the exact semantic relationship between non-final and final clauses. Their function in the discourse is often inferred from the context. As was pointed out above, it is not always very obvious where to draw the line between such semi-independent chained clauses and clauses that to a larger extent are interpretable as adverbial and dependent vis-à-vis one particular main clause.

Like non-final chained clauses, clauses with adverbial functions generally precede their head. They usually precede the entire head clause although embedding into the head clause or head proposition also occurs quite frequently. Only rarely does the modifying clause occur subsequent to the verb of the head clause. All such clauses with adverbial functions are in essence dependent but, as was reflected earlier in the discussion on clause chaining above, not equally and unquestionably subordinate in nature.

A number of different construction types are available for clauses functioning adverbially, some of them with a more general application, and others more restricted to certain subfunctions:


### 14.4 Clauses with adverbial functions


These constructions will be discussed as they occur in the different adverbial functions presented below.

### **14.4.1 Temporality and related functions**

**Subsequence.** Many dependent clauses with obvious temporal readings are in fact not much different from the sequential chaining discussed above (§14.3). Especially when it is a matter of subsequence, and the subject remains the same, the Converb construction (i) is used, as in (51)–(52). Sometimes, especially in narratives, an immediately preceding finite verb is repeated but as a converb, thus not adding any new information but instead highlighting the temporal relationship between the two events. This latter use of the converb is more typically adverbial – and the Converb clause to a greater extent dependent – than its use in sequential chaining as laid out above.


'He migrated from there. When he had migrated, he came to Drosh.' (B:ATI003-4)

### 14 Complex constructions

When the subject in the dependent clause is different from the subject in the main clause, as was the case in clause chaining, the corresponding construction is a finite *ta*-clause (construction (ii)). That can be seen in example (53) and (54). Again, the verb in the preceding finite clause is often repeated.


It appears to be possible to leave out the subordinator *ta*, as in (55), and retain a different-subject reading, although that is not particularly common.

(55) *naɡarǰúti* Nagarjuti *yhéel-i* come.pfv-f *tas* 3sg.acc *díi* from *khooǰóol-u* ask.pfv-msg *kanáa* what *bhíl-i* become.pfv-f 'When Nagarjuti came, he asked her what happened.' (MAB029)

The subsequential reading is contextually implicit rather than explicit in the examples given so far. However, a separate temporal conjunction *kareéɡale* or *kareé ɡalé ki* 'when' can also (especially in B), as seen in (56), be added to the *ta*-construction to make this reading obvious. This distinguishes it more clearly from simple clause chaining.

(56) *kareé=ɡale* when=ever *se* 3pl.nom *múṛ-im* die.pfv-fpl *ta* sub *asím* 1pl.erg *tenaám* 3pl.acc *ḍhanɡéel-im* bury.pfv-fpl 'When they died, we buried them.' (B:FOR037)

This construction has characteristics of a relative clause (see also §14.6.2), like a number of similar temporal adverbial constructions in the world's languages (Thompson, Longacre & Hwang 2007: 246–247), which is even more obvious in (57).

(57) *hasó* 3msg.nom *kareé=ɡale* when=ever *ki* comp *sastíil-u* heal.pfv-msg *ta* sub *ṣuú* king *itlaá* message *phedíl-i* arrive.pfv-f 'When he had become healthy, the king heard about it.' (B:ATI059) 14.4 Clauses with adverbial functions

A finite temporal clause with *kareé* and without *ta* (58) is also possible.

(58) *kareé* when *ǰéel-i* give.birth.pfv-f *chiír ̣* milk *taap-áan-a* heat.up-prs-mpl 'When it [the goat] has given birth, its milk is heated.' (A:KEE025)

Perfective Participle clauses with a postposition, such as *pahúrta, baád*, *patú,* all glossed as 'after', are also used in temporal subsequence (construction (iv)), thereby giving them an explicit "temporal anteriority" reading (Cristofaro 2005: 159). *pahúrta* (in (59)) and *baád* (in (60)) each govern the genitive case of the participle, whereas *patú* (in (61)) takes the nominative. The latter seems to occur only in B.


**Subsequence cum instrument.** The temporal subsequence reading of the Converb construction shades off into what can be seen as expressing instrument or supplying the main verb with a manner specification and is especially relevant for converbs formed from transitive verbs of motion, as in examples (62)–(65), such as *ɡhín-* 'take', *de-* 'put, give', *ɡaḍé-* 'take out', *ɡhašé-* 'take hold of, grab', *ɡalé-* 'throw'. With some of these converbs we may very well witness a further grammaticalisation into postpositions or, alternatively, the development of compound verbs (see §9.6.2).

(62) *ɡhaḍeerá* elder.obl *phed-í* arrive-cv *laṣ* completely *čax* swiftly *kaṭéeri* knife *ɡhin-í* take-cv *se* def *taáǰ* crown *čhiníl-i* cut.pfv-f *čhin-í* cut-cv *rumeel-í* handkerchief-obl *maǰí* in *de* put.cv *ǰeep-í=ee* pocket-obl=into *dít-i* put.pfv-f

'The older [brother] came, took a knife, and cut off the crown. Having cut

### 14 Complex constructions

it off, he wrapped it in a handkerchief and then put it into his pocket.' (A:DRA016)

(63) *maačís* matches *ɡaḍ-í* take.out-cv *tasíi* 3sg.gen *laméeṭi* tail *anɡóor* fire *ṣaawóol-u* put.on.pfv-msg *ta* sub *so* 3sg.nom *uḍheewíl-u* flee.pfv-msg

'He took matches, and when he had put fire to its tail, it fled.' (A:HUB009)

(64) *yhaí* come.cv *tasíi* 3sg.gen *háat-ii* hand-gen *ɡhaš-í* grab-cv *ḍúkur-a* hut-obl *šíiṭi* inside *the* to *ɡhin-í* take-cv *ɡíia* go.pfv.pl *hín-a* be.prs-mpl

'Having come there, they took his hand and brought him inside the hut.' (A:KAT062)

(65) *so* def.msg.nom *darwóoza* door *kaná* shoulder.obl *wée* on *ɡal-í* throw-cv *dharéndi* outside *nikháat-u* appear.pfv-msg 'He came out, with the door on his shoulder.' (A:GHU028)

**Accompanying circumstance.** The Converb construction can also express an accompanying circumstance, or specification of means with reference to the same subject and the same event. In (66)–(68) a subsequential interpretation is even less of an issue.


dragged, brought down] the dead body….' (A:GHA044)

14.4 Clauses with adverbial functions

(68) *tas* 3sg.acc *wheel-í* bring.down-cv *atshareet-á* Ashret-obl *phedóol-u* bring.pfv-msg 'We reached [lit: having taken down, brought it] Ashret with it.' (A:GHA081)

However, an accompanying circumstance, or further specification of the action carried out by the subject, is sometimes, as in (69) and (70), expressed with the more restricted Copredicative Participle (construction (vii) above).


**Simultaneity.** Simultaneity involves two overlapping events and includes what Givón (2001b: 330) refers to as "point coincidence". The backgrounded event (Thompson, Longacre & Hwang 2007: 254–255) is usually expressed with a Verbal Noun (construction (v) above) in Palula. If the subject is the same in both the foregrounded and the backgrounded events, the Verbal Noun occurs alone, as in (71), or with a postposition *maǰí* 'in, among' (72) or *sanɡí* 'with'. If there is a subject shift between the foregrounded and the backgrounded event, the Verbal Noun (or the postposition) occurs, as in (73) and (74), with a suffix *-eé*, 3 which functions as a switch-reference marker similar to the contrast between *ta*-constructions vis-à-vis Converb constructions as described above (§14.3).

<sup>3</sup> Probably it is accent-bearing and possibly identical to the suffix coding inclusivity, used with numerals, see §7.4.1.

### 14 Complex constructions

(71) *bachúuṛu ̣* calf *be* go.cv *ba* top *patuɡiraá* back *yh-ainií* come-vn *ba* top *páand* path *na* neg *léed-i* find.pfv-f *se* def *bachúuṛ-a ̣* calf-obl

'The calf went [in], but as it was coming back, it didn't find its way, the calf.' (A:CAV015)


man eat-vn to

'As he reached the peak, she came running down [from it], ready to eat the man.' (A:PAS121-2)

(74) *aǰdahaá* dragon *katoolíi-a* fodder.sack-obl *wée* in *ač-aníi* enter-vn *sanɡi-eé* with-incl *lhooméea* fox.obl *se* def *míiš-a* man-obl *ḍáḍi* toward *išaará* hint *thíil-u,* do.pfv-msg *thanaáu* string *dhrak-é* pull-imp.sg 'Just as the dragon went into the sack, the fox signaled to the man to pull the string.' (B:DRB036)

A somewhat parallel strategy is the use of the Agentive Verbal Noun and *bhe-* 'become', (75)–(76), literally 'became an onlooker, became goers', etc., referred to above as construction (vi). With this construction, however, it is not always obvious which event is backgrounded and which one is foregrounded.


14.4 Clauses with adverbial functions

*hín-a* be.prs-mpl *naaɡhaaní* suddenly *áak* idef *amzarái* lion *nikháat-u* appear.pfv-msg *hín-u* be.prs-msg 'One day, as they were walking through a forest, suddenly a lion appeared.' (A:UNF005-6)

Less commonly, simultaneity or temporal overlap is expressed with a perfective verb form with a postposed *ee*-conjunction, as in (77). There is no conclusive answer at the moment as to whether the perfective verb form is finite or nonfinite in this case.

(77) *so* 3sg.nom *ta* cntr *ɡúum=ee* go.pfv=cnj *ma* 1sg.nom *ba* top *patuɡiraá* back *yhóol-u* come.pfv-msg 'It went away and I went back [home].' (A:HUA043)

**Precedence.** In the temporal clauses mentioned so far, there is either an implied simultaneity or a sequence of events where the subordinate (or cosubordinate) 'when'-clause refers to an event taking place before the event named in the main clause. In so-called precedence clauses (Givón 2001b: 327), on the other hand, the event in the 'before'-clause has not yet been realised in relation to the event mentioned in the main clause (Thompson, Longacre & Hwang 2007: 247–248). This is also referred to as "temporal posteriority" (Cristofaro 2005: 159). For such clauses a Verbal Noun with the complex postposition *díi muṣṭú* 'before' (in B *díi muxáak*) is used, as can be seen in (78)–(80).


### 14 Complex constructions

### **14.4.2 Purpose**

Dependent clauses expressing the purpose for an event to be carried out usually have an implicit human subject coreferential with the (normally) explicit subject of the main clause (Givón 2001b: 337). The construction used, a Verbal Noun followed by the postposition *the* 'to', parallels such "dative" purpose clauses found in many other languages (Thompson, Longacre & Hwang 2007: 251–252). The finite verb in sentences, such as (81)–(82), is almost invariably an intransitive motion verb *ye-* 'come', *ukhé-* 'come up', *whe-* 'come down', *be-* 'go', etc.


Often, the verb itself is implicit in the purpose clause, and instead the direct object argument, *looṛíia* 'bowls' in (83), receives the dative marking.

(83) *théeba* then *se* 3pl.nom *iṇc̣* bear *kaṭamuš-á* Katamosh-obl *the* to *óol* watch *thawaá* make.do.cv *looṛíi-am* bowl-pl.obl *the* to *ɡíia* go.pfv.mpl *hín-a* be.prs-mpl

'Then they made the bear stay with Katamosh and went to get bowls [lit: for bowls].' (A:KAT127)

Purpose is also quite often explicitly expressed by the postposition *dapáara* 'for, for the sake of' (a Pashto loan) following the Verbal Noun, as in (84) or (in B), the postposition *pándee* 'for, for the sake of' following the Verbal Noun in the genitive, as in (85).

(84) *bhraawéeli* brotherhood *th-ainií* do-vn *dapáara* for *tanaám* 3pl.acc *maǰí* among *áak* one *qabúl* consent *th-áan-u* do-prs-msg *ki* comp *ma* 1sg.nom *bhíiru* he.goat *ɡhin-í* take-cv *thíi* 2sg.gen *ɡhooṣṭ-á* house-obl *the* to *yh-úum* come-1sg 'In order to form a brotherhood, one of them agrees to bring a male goat 14.4 Clauses with adverbial functions

to the other person's house.' (A:MIT013)

(85) *ɡhooṣṭ-á* house-obl *baás* shelter *de* give.cv *ba* top *patuɡiraák* back *tenaám* 3pl.acc *dac̣h-aníi-e* look-vn-gen *pándee* for *ɡhireé* again *ɡáu* go.pfv.msg

'After spending the night at home, he went back again to look at them.' (B:BEL335)

### **14.4.3 Causality**

As noted by Thompson, Longacre & Hwang (2007: 247), it is no surprise that we find causal relationships expressed in the same way as temporal relationships, or rather that the constructions themselves are neutral between a time and a cause interpretation. Often in Palula, dependent clauses, such as the different-subject *ta*-construction (see §14.4.1), can thus imply an immediately preceding event in time as well as a cause for the event or action in the main clause to happen or be carried out.

Example (50), quoted above, may, for instance be given a sequential interpretation as well as a causal one: 'When a daughter of yours was born, your wife took her and went up into the forest' (sequential), or: 'Because a daughter of yours was born, your wife took her and went up into the forest' (causal).

As further pointed out by Givón (2001b: 335), cause (along with reason) can itself be divided into a number of subdistinctions, such as agentive external, nonagentive external, eventive external, non-eventive external, eventive internal and non-eventive internal.

For external causality, the preferred strategy is, as can be seen in (86), the already mentioned finite different-subject construction with *ta*.

(86) *ǰaanɡul-á ma bhanǰóol-u ta ru-áan-u*

> Jangul-obl 1sg.nom beat.pfv-msg sub cry-prs-msg

'Jangul beat me, so therefore I am weeping.' (A:HUA104)

A construction involving a Verbal Noun,<sup>4</sup> also used to imply simultaneity (see §14.4.1), can also be used to signal causality, as in (87)–(89).

<sup>4</sup> To what degree any of the case suffixes in the examples from Biori correspond to the (possibly conjunctive) clitic *ee* in the Ashret example remains at this point an unanswered question.

### 14 Complex constructions


For most cases of internal causality, where the same-subject condition holds, the Converb construction is used, as in (90) and (91).


Causality (perhaps internal in particular) can also be expressed in a more explicit way, for instance as in (92), through a relativisation of *wáǰa* 'reason', a noun borrowed from Pashto, or a dependent clause with a Verbal Noun and a postposition, the latter exemplified in (93).

(92) *so* def.msg.nom *wáǰa* reason *ba* top *eeṛó* 3sg.dist.nom *ki* comp *so* def.msg.nom *lhookeer-ó* younger-msg *brhóo* brother *so* def<sup>5</sup> *kaná* shoulder.obl *pharé* along *ɡal-í* throw-cv *wheél-u* bring.down.pfv-msg *de* pst *ḍóo* host 14.4 Clauses with adverbial functions

*the* do.cv *eetasíi* 3sg.gen *asár* effect *tas* 3sg.rem.acc *the* to *phedíl-i* arrive.pfv-f 'Because [lit: for the reason that] he had carried his younger brother on his shoulders, he was afflicted.' (A:DRA028-9) *na čit-aníi ǰhulí ɡhrast-í iṣkáar bhíl-i*

(93) *čhéeli* goat neg think-vn on wolf-gen prey become.pfv-f 'Since the goat didn't really think, she fell prey to the wolf.' (B:FOX034)

### **14.4.4 Conditionality**

As is the case in many languages (Thompson, Longacre & Hwang 2007: 257–258), in Palula there is no absolute distinction made between temporality and conditionality. Although there are distinct morphological markings available to signal conditionality, there is, as we shall see, some overlap in coding between clauses with an implied temporality (and to some extent causality) vis-à-vis clauses with conditional semantics, an observation lining up with more general, cross-linguistic tendencies (Cristofaro 2005: 161).

It is also possible, as noted by Thompson, Longacre & Hwang (2007: 255–260), as well as by Givón (2001b: 333–334), for languages to make explicit finer gradations along a scale of verisimilitude, an observation relevant for the analysis of Palula conditionals. We can discern at least three degrees of verisimilitude: a) generally true, b) possibly or probably true, and c) not true. These will be discussed and exemplified under the following headings: assumed conditionality, hypothetical conditionality, and counterfactuality.

**Assumed conditionality.** Conditionals under this heading partly match what Thompson, Longacre & Hwang (2007: 255–256) refer to as reality conditionals. Here we find the conditionally marked clauses that are most closely related to temporality and causality as described above. Often the conditionally marked clause is equally translatable with 'when…' and 'if…' (and, especially with pasttime reference, 'whenever'). The typical construction is a finite (often perfective) 'when'/'if'-clause with the conditionality marker *seentá (B síinta)*, followed by an imperfective 'then'-clause. They describe a condition that usually holds, especially in a more generic sense, or used to do so in the past. Examples are given in (94)–(97).

<sup>5</sup> I do not have any explanation for this form to appear. If it indeed is a determiner of the oblique case-marked *kaná*, the form *se* is expected.

### 14 Complex constructions


(A:UXB016)

(97) *ac̣húuṛi-m* walnut.tree-pl *ṭhak-ainií* shake.down-vn *waxt* time *bhíl-u* become.pfv-msg *seentá* condh *yh-íin* come-3pl *de* pst

'When it became the season for picking walnuts, they would come here.' (A:JAN015)

Sometimes the clause-final conditional marker also co-occurs, as in (98), with a temporal conjunction *kareé* 'when'.

(98) *míiš-ii* man-gen *putr* son *kareé* when *zuaán* young.adult *bhíl-u* become.pfv-msg *seentá* condh *tas* 3sg.acc *the* to *kúṛi* woman *dawainií* ask.vn *bandubás* arrangement *th-áan-u* do-prs-msg 'When someone's son comes of age, he arranges for his marriage.' (A:MAR003-4)

Some clauses with a predictive semantics are coded in the same manner, either with the 'when'/'if'-clause in the perfective (99) or in the future (100), although most predictions tend to be coded as hypothetical conditionals (see below). Again, the choice between the two is most likely to do with the degree of predictability, although in some predictive contexts the two codings are used more or less interchangeably.

14.4 Clauses with adverbial functions


'When I come, at whatever time during the night, and knock, you should come out, and we will go to work.' (A:WOM628)

In some clauses, such as in (101), *síinta* clearly implies cause rather than condition, particularly in B it seems.

(101) *koó* someone *se* def *muṭ-á* tree-obl *túuri-e* below-gen *thíi* from *baṭ-áam* stone-ins *ǰ-áan-a* hit-prs-mpl *šíinta* condh *ba* top *kúuk-a* crow-pl *na* neg *bheš-ái* sit.down-inf *bhay-áan-a* be.able.to-prs.mpl

'Someone was throwing stones from under the tree, keeping the crows from being able to sit down.' (B:SHB741)

In some pieces of discourse, *seentá/síinta* is also used clause-initially, as a single element, either without any obvious reference to any preceding clause and with a rather vague semantics, glossed as 'then' or 'well', or with a very general causal reference to a larger chunk of discourse, functioning as a conjunction with the meaning 'so', 'therefore' or 'it means'. An example of the latter is (102).

(102) *seentá* so *asíi* 1pl.gen *iskuúl* school *bi* also *asaám* 1pl.acc *the* to *bíiḍ-i* much-f *dhúura* distant *hín-i* be.prs-f 'So, we also have a long way to school.' (A:OUR016)

**Hypothetical conditionality.** These conditionals belong mainly to the imaginative type, following Thompson, Longacre & Hwang's (2007: 259–260) classification. Examples (103)–(104) express conditions of a lower verisimilitude than the assumed conditionals above. The conditionally-marked clause of this kind, often translatable with 'if…', is a finite (often perfective) clause with the conditionality

### 14 Complex constructions

marker *heentá*, followed by an imperfective 'then'-clause. They describe a condition that is probable or possible, although to a certain extent imagined. Most conditionals referring to the future are also coded this way.


'If two persons like one another, they become brothers, they form a brotherhood.' (A:MIT010-12)

The same structure is found in negative Conditionals (105) as well.

(105) *thíi* 2sg.gen *ninaám* 3pl.prox.acc *na* neg *phedúul-a* take.pfv-mpl *heentá* condl *qeaamatée-e* judgement-gen *dees-á* day-obl *ma* 1sg.nom *tu* 2sg.nom *díi* from *khooǰ-áam* ask-1sg 'If you don't take these [to her], I will ask you on the day of judgment.'

(B:FLW800)

As with assumed conditionality, the clause-final conditional marker also cooccurs with a temporal conjunction in some instances of hypothetical conditionality, as can be seen in (106).

(106) *maṇḍáu* veranda *kareé* when *thíil-u* do.pfv-msg *heentá* condl *maṇḍawíi* veranda.gen *ɡhareé* again *ḍhínɡar* wood *ɡhooṣṭ-á* house-obl *díi* from *tuúš* a.little *muxtalíf* different *bh-áan-u* become-prs-msg

'If you were to make a veranda, the wood used for the veranda should be a bit different from that used for the house.' (A:HOW073)

Besides the 'then'-clauses in (103)–(106) with Future and Present, respectively, the Imperative as well as the Simple Past are possible to use in future-referring 'then'-clauses, both occurring in (107).

14.4 Clauses with adverbial functions

(107) *ki* comp *anú* 3msg.prox.nom *múṛ-u* die.pfv-msg *heentá* condl *ḍhanɡ-á* bury-imp.sg *anú* 3msg.prox.nom *ǰáand-u* alive-msg *ba* top *dharíit-u* remain.pfv-msg *heentá* condl *asím* 1pl.erg *tu* 1sg.nom *the* to *baxíl-u* donate.pfv-msg 'If he dies, then bury him, if he instead stays alive, then he will remain

with you [lit: we entrusted him to you].' (B:ATI052)

Quite infrequently, the 'if'-clause occurs in a TMA category other than the perfective categories. The utterance in (108) is in effect a threat, while the utterance in (109) is a polite request.


'Would you be so kind as to give us our person [the bride].' (A:MAR059)

Apart from the explicitly marked Conditional clauses exemplified above, we also find conditionality implied in Converb clauses, as in (110), as well as in *ta*clauses, as in (111). The Past Imperfective in the 'then'-clause in (111) implies unfulfilment.


<sup>6</sup> This is not the expected form of a determiner, since the head is a masculine singular nominative.

14 Complex constructions

*de*

pst

'If I told my companions, they may run away and leave the dead body.' (A:GHA047)

**Counterfactuality.** In counterfactual expressions, as in (112)–(113), both the 'if' clause and the 'then'-clause occur with the conditionality marker of low verisimilitude, *heentá*, i.e., the one also used in the hypothetical expressions above. The word *áɡar* in (113) is a subordinating expression also occurring in Urdu conditionality (Schmidt 1999: 101–103).


(A:HLE3086)

The proposition in the 'if'-clause of (113) is contrary to fact, i.e., the person speaking knows for sure that the person referred to is dead. Therefore the 'then' clause here expresses what would have been the case if that had been otherwise. If the verisimilitude value, in constrast, is pending (Givón 2001b: 332), as is the case with the hypothetical conditionality exemplified already, the proposition of the 'then'-clause may still be true. A hypothetical conditionality contrasting with the counterfactuality in (113) can be seen in the almost parallel sentence in (114).

(114) *áɡar* if *tasíi* 3sg.gen *dóodu* grandfather *ǰáand-u* alive-msg *hín-u* be.prs-msg *heentá* condl *tasíi* 3sg.gen *úmur* age *čuurbhišá* eighty *kaal-á* year-pl *hóons-a* stay-3sg

'If his grandfather is still alive, he would be 80 years old.' (A:HLE3086)

### 14.4 Clauses with adverbial functions

**Concessive conditionality.** Concessive conditionality is basically expressed by the same means as hypothetical conditionality, i.e., with a finite (often Simple Past) 'if'-clause with the conditionality-marker *heentá*. However, as seen in (115) and (116), the particle *bi* 'also, even' is added to emphasise the contrary-to-expectation reading.


(A:CHE071114)

### **14.4.5 Clauses with other adverbial functions**

**Concessive clauses.** There is probably no unique strategy for forming concessive clauses. A conjunction with "adversative" reading (see §14.2.1–§14.2.2) can, for instance be used for this purpose, with (117) or without (118) the contrary-toexpectation particle *bi* 'also, even'.

(117) *so* 3sg.nom *šúi* good.f *paaṇṭí* clothes *bi* even *ṣaá* put.on.cv *heensíl-u* stay.pfv-msg *de* pst *xu* but *bač* saved *na* neg *bhíl-u* become.pfv-msg

'Even though he was wearing good clothes, he did not survive.' (A:CHE071114)

(118) *míi* 1sg.gen *parčá* paper *ta* cntr *dít-u* give.pfv-msg *xu* but *ma* 1sg.nom *paás* passed *ba* top *na* neg *bh-eeṇḍeéu* become-oblg

'Although I turned in my paper, I will not pass.' (A:CHE071114)

### 14 Complex constructions

**Substitutive clauses.** As with concessive clauses there is no single dedicated substitutive construction. Instead a few different constructions are used to express the semantics of a substitutive.

First, there are constructions involving the noun *zhaáy* 'place'. The one exemplified in (119) has a subordinate clause almost analogous to the English 'instead of…', with a Verbal Noun qualifying the oblique form of *zhaáy*. The 'place'-noun in (120) is on the other hand only qualified by a possessive phrase and is therefore not a subordinate clause at all.


It is also possible to use a comparative construction 'X-ing a, is better than X-ing b'. In this case, as seen in (121), both the subject head and the head of the predicate noun phrase are Verbal Nouns.

(121) *pulusá* police *the* to *man-ainií* say-vn *díi* from *máa=the* 1sg.nom=to *man-ainií* say-vn *šóo* good.msg *de* be.pst 'Rather than telling the police, you should have told me [lit: The telling to me is better than the telling to the police].' (A:CHE071114)

Although implicit to a larger extent, a substitutive reading of an utterance-complement with the direct-quote marker *thaní* (see §14.5.1), is also possible ((122)– (123)) and may very well be the preferred way of expressing substitution. In (123) this interpretation is supported by the word *ulṭá* 'upside-down'.


14.5 Complement clauses

*whaal-éen* bring.down.3pl *thaní* qot *hasé* rem *dees-á* day-obl *ulṭá* upside.down *hasé* rem *trúu* three *ǰinaazeé* corpse.pl *ɡhin-í* take.cv *wháat-a* come.down.pfv-mpl *xálak-a* people-pl

'In the village, children and women were waiting for the menfolk who had gone hunting, expecting them to bring back meat. Instead, they brought back three corpses.' (B:AVA218-20)

**Additive clauses.** There is not much evidence of any entirely unique construction expressing additive semantics. Instead the conjunction with *bi … bi* (see §14.2.1) seems to cover some occurrences of "addition". Possibly the construction found in (124), with a subordinate Verbal Noun followed by the postposition *sanɡí* 'with' along with *bi* in the main clause, represents a more explicit expression of "addition".

(124) *ma* 1sg.nom *bhanǰ-ainíi-a* beat-vn-obl *sanɡí* with *áak* one *the* to *tíi* 3sg.obl *míi* 1sg.gen *paiseé* money.pl *bi* also *palóol-a* steal.pfv-mpl 'Besides beating me up, he stole my money.' (A:CHE071114)

It is uncertain what role the adverbial phrase *áak the* plays in signalling "addition". In (125), now with *ak ta*, a similar reading is possible, although the two clauses here are symmetric rather than one of them being subordinate to the other.

(125) *haṛó* 3msg.rem.nom *hateeṇ-ú* such-msg *deés* day *de* be.pst *ki* comp *ak* one *ta* cntr *kir* snow *bíiḍ-u* much-msg *de* be.pst *dhuíime* second *xálak-a* people-pl *adrax-í* forest-obl *heensíl-a* stay.pfv-mpl

'It was a day on which there was both a lot of snow, and a lot of people in the forest .' (B:AVA216)

### **14.5 Complement clauses**

Palula utilises a number of different strategies for sentential complementation, mainly determined by the complement-taking verb. However, due to the degree of semantic bond between the complement-taking predicate and the verbal element of the complement (Givón 2001b: 39-40), most of them fall, quite

### 14 Complex constructions

clearly, into one of two main types: a) sentence-like complements, and b) nonfinite complements. While the complements of the first type are usually (but not exclusively) extraposed and syntactically relatively independent from the main clause, the complements of the second type are almost invariably embedded into the main clause and have a time reference entirely determined by the main clause.

The specific strategies found in the S-like main type are:


An important feature of S-like complements with strategies (i)–(iii) is that they always take a quoted-speaker's/quoted-experiencer's perspective as far as pronominal reference and deixis are concerned. This also means that there are hardly any constructions that are indirect; probably no reported speech ever occurs as indirect.

Strategy (ii) used with reported discourse is often combined with strategy (iii), in which case *thané-* mainly functions as a (end-of) quotation marker (see discussion in §10.2.4).

The specific strategies found in the non-finite type of complement clause are:


14.5 Complement clauses

(vii) a causative construction in which the complement-taking predicate is entirely co-lexicalised with the complement.

Strategy (vii) represents the top of the complementation scale (Givón 2001b: 74) and is at best only underlyingly analysable as two clauses, but is included here as there is no other natural context for it to be discussed.

A few additional minor strategies are exemplified below, but due to their isolated occurrence in the data, they have not been listed above.

In the following we will exemplify and discuss these strategies as they occur within some broadly and functionally defined classes or categories of complement-taking predicates (following Givón 2001b: 40–59 and Noonan 2007: 120– 145). However, as will be apparent in the discussion, the strategies outlined above are not randomly applied, but represent a language-internal differentiation along a continuum. This is a continuum that, on the one hand, branches from complement-taking PCU (Perception-Cognition-Utterance) predicates with a weak bond between the two events to modality predicates with a strong bond between them. On the other hand, it is a continuum branching from PCU predicates with a weak bond to manipulation predicates with a strong bond between the two events.

### **14.5.1 Complement-taking PCU predicates**

**Utterance predicates.** It is only with utterance predicates that the zero strategy ((i), above) may be used alone, and even then it occurs sparingly. According to my data, it is only the verb *mané-* 'say, call, tell', as in (126)–(127), that seems to allow it.


'A woman told him, "When a daughter had been born, your wife took her with her and went up into the forest."' (B:FOR013-4)

### 14 Complex constructions

More frequently, the *ki*-strategy ((ii) above) is used, with *mané-* as well as with a number of other utterance predicates (or predicates that in context imply an utterance): (128)–(130). Whenever *ki* is used as a complementiser, the complement is heavy-shifted to the end.


Related utterances in Palula always occur as direct quotations, as can be seen in the examples (126)–(130), or more correctly, they are always presented from a quoted-speaker's or experiencer's (i.e., the subject of the matrix clause) perspective, as far as deixis, pronominal and time reference is concerned. Non-oral perception or cognition<sup>7</sup> is treated in this way too, i.e., thoughts, as in (131), are presented from the thinking agent's perspective, as if they were uttered orally (a marginal exception to this is found with what I refer to below as "immediate perception predicates").

(131) *áa* one *deés* day *se* def *phoó* boy *xiaál* opinion *thíil-i* do.pfv-f *híni* be.prs-f *ki* comp *tuúš* some *čhoót* cheese *míi* 1sg.gen *se* def *yaar-íi* friend-gen *rumeel-í* handkerchief-obl *maǰí* in *ɡhaṇḍ-í* tie-cv *wíi-a* water-obl *keé-na* why-neg *ɡal-úum* throw-1sg

'One day the boy thought, "Why don't I put some cheese in my friend's handkerchief and throw it into the water?"' (A:SHY043)

<sup>7</sup> It is probably equally possible to classify the predicate in a sentence like this as desiderative – i.e., expressing a desire that the proposition be realised – rather than as an uttererance predicate.

### 14.5 Complement clauses

While the complement of the utterance predicates mentioned so far are all extraposed (vis-à-vis the basic SOV word order), that is never allowed with any form of the verb *thané-* 'say, call' (strategy (iii) above), as in (132)–(133). Note that, in example (133), the sentence with *thaní* is part of a longer clause chain (see §14.3).


'The Chitrali people became very happy, saying "He has saved our reputation [lit: made our name live]", and they also distributed gifts and sweets.' (A:BEW007-9)

The verb *thané-*refers straightforwardly and literally back to what was actually uttered, and especially when combined with another utterance predicate and/or *ki* preceding the utterance, which is the case in (134) and (135). It becomes, in effect, an end-of-quotation marker, particularly in its converb shape *thaní*. 8

(134) *deeúlii* Dir.gen *xálak-am* people-pl.obl *daawát* invitation *dít-i* give.pfv-f *atsharíit-am* Ashreti-pl.obl *the* to *ki* comp *muqaabilá* contest *th-íia* do-1pl *thaníit-u* say.pfv-msg *ta* comp *atshareet-íi* Ashret-gen *paalawaaṇ-aán* wrestler-pl *ɡíia* go.pfv.pl 'The people fom Dir invited the Ashreti people, "Let's have

a competition!" And [having said that] the Ashreti strongmen left.' (A:CHA001-2)

<sup>8</sup> That *thaní* is becoming grammaticalised to a large extent is reflected in its being used rather often, contrary to expectation even when there is a subject switch, while a converb normally signals same-subject reference across clauses (as described elsewhere) without having any clear connection with the following discourse.

### 14 Complex constructions

(135) *míi* 1sg.gen *ba* top *ṭeekíl-u* call.pfv-msg *ki* comp *óo* o *kaakaá* uncle *óo* o *kaakaá* uncle *ínc̣-a* bear-obl *čhéeli* goat *khéel-i* eat.pfv-f *thaní* qot *ta* sub *bhraáš* slowly *pašambeé* Pashambi *so* def.msg.nom *bháaru* load *čhúuṇ-u* put.pfv-msg 'I called out, "O uncle, the bear ate the goat", whereby Pashambi slowly put down his load.' (A:PAS056-7)

**Propositional attitude predicates.** Closely related to and sometimes difficult to distinguish from utterance predicates are those predicates that express an attitude regarding the complement proposition, especially concerning its verisimilitude. The particle *xu* (in examples (136) and (137)) expresses certainty on the part of the subject of the matrix clause concerning the immediately preceding argument or phrase.

It seems also that the range of strategies available is much the same as for utterance predicates. For both utterance and propositional attitude predicates, the complement is presented in direct (reported) discourse, often signalled by the end-of-quotation marker *thaní*.

(136) *se* def *míiš-a* man-obl *soóč* thinking *thíil-i* do.pfv-f *ki* comp *anú* 3msg.prox.nom *xu* however *típa* now *ma* 1sg.nom *páta* surely *kh-óo* eat-3sg *thaní* qot *ba* top

'The man thought that he would surely eat him too [lit: The man thought, "He will surely eat me too"]. Having said that…' (A:THA009-10)

(137) *ée* o *míi* 1sg.gen *xudaayaá* God.voc *ni* 3pl.prox.nom *xu* evidently *ux-íi* camel-gen *rhaíi* foot.prints *hín-i* be.prs-f *xéer* well *béem* go.1sg *ni* prox *rhaíi* foot.prints *ɡhaš-í* catch-cv *ɡubáa* what *ta* cntr *thaní* qot 'I thought, "O my God! It's the footprints of a camel! Well, in any case, I'll find out."' (A:HUA061-2)

Propositional attitudes may also, as in (138), be expressed as noun complementation, the general structure remaining the same.

(138) *asíi* 1pl.gen *xiaál* opinion *ki* comp *ɡóo* perhaps *mheeríl-u* kill.pfv-msg *heentá* condl *khóol-u* eat.pfv-msg *thaní* qot 'We thought that perhaps he had been killed and eaten [lit: Our opinion was that, "If he has been killed, he has been eaten"].' (A:GHA011)

### 14.5 Complement clauses

**Commentative predicates.** Commentative (or factive) predicates provide a comment on a complement proposition that is assumed to be real (Noonan 2007: 127– 129). Again, the distinction between these and utterance predicates is not always very obvious, as the complement often occurs as reported discourse, and there is probably quite some overlap. Typically, the complement clause is formed as a question, with *keé* 'why' in (139) or *kateeṇí* 'what kind' in (140).


When the subject of the complement clause and the matrix clause are coreferent, as in (141), the complement can occur as a Verbal Noun (strategy (v)).

(141) *lhoomée* fox.obl *teeṇíi* refl *taalím* education *man-ainíi* say-vn *šárum* shame *dach-íi ̣* look-3sg *de* pst 'The fox was ashamed of talking about his own education.' (B:FOY006)

**Predicates of knowledge acquisition.** Many complement-taking predicates having to do with acquisition of knowledge use the same strategies as for utterances, and again the complement is presented as an utterance (either by the agent of the matrix clause, as in (142), or by some other (explicit or implicit) source participant, as in (143) and (144)), whether it is literally and verbally uttered or is merely cognitively processed, i.e., "uttered" mind-internally.

(142) *búd-a* understand.pfv-mpl *hín-a* be.prs-mpl *ki* comp *phaí* girl *wíi-a* water-obl *ɡíi* go.pfv.fsg *thaní* qot 'They understood that the girl must have thrown herself into the water.' (A:SHY060)

### 14 Complex constructions

(143) *ak* one *yúun* month *padúši* after *so* 3msg.nom *xabár* informed *bhíl-u* become.pfv-msg *ki* comp *šišíi-e* Shishi-gen *hakim-á* ruler-obl *thíi* 2sg.gen *so* def.msg.nom *duṣmán* enemy *nawaab-á* prince-obl *ḍáḍi* side *phreyíl-u* send.pfv-msg *thaní* qot

'A month later he [the king] was informed that the ruler of Shishi had helped his [the king's] enemy to get to [the territory of] the prince [of Dir].' (B:ATI030-1)

(144) *tíi* 3sg.obl *ṣúunt-u* hear.pfv-msg *de* pst *ki* comp *zanɡal-íi* forest-gen *zinaawur-óom* beast-pl.obl *díi* from *insaán* human *thaní* qot *áa* idef *šay* thing *hín-u...* be.prs-msg *thaní* qot 'He heard from the beasts of the forest, that a thing called "man"…' (A:KIN002-4)

In examples (142)–(144), the complement is presented as an already uttered proposition. Even when the – assumedly verbal or cognitive – knowledge is still to be obtained, as perceived by the subject of the matrix clause, the same structure is utilised, which can be seen in (145) but now without the final *thaní*.

(145) *so* def.msg.nom *insaán* human *bulaḍ-íim* find.out-cprd *bíi* go-3sg *de* pst *páand-a* road-obl *pharé* along *ki* comp *so* def.msg.nom *insaán* human *kateeṇ-ú* what.kind-msg *šay* thing *hín-u* be.prs-msg 'He went ahead to find out what kind of thing man is.' (A:KIN004)

**Immediate perception predicates.** With at least two verbs coding directly perceived events, *dach-̣* 'look, see' (in (146)–(147)) and *dhay-* 'watch, note' (the latter only in B, (148)), another construction is regularly used (strategy (iv)), in which the finite complement-taking predicate occurs with a particle *ta*, followed by a finite complement clause, i.e., the same construction used in the chaining of different-subject clauses (see §14.4).

(146) *huṇḍɡiraá* upward *dach-íin ̣* look-3pl *ta* sub *iṇc̣* bear *muṭ-íi* tree-gen *phúṭi* top *ǰhulí* on *bheš-í* sit-cv *áaṇc̣-a* raspberry-pl *kha-áan-u* eat-prs-msg

'They looked up and there was the bear sitting in the top of the tree eating raspberries.' (A:KAT145)

14.5 Complement clauses


(B:FOR030)

It should be especially noted that while all the complements with PCU-predicates exemplified in the previous sections represent direct discourse, the perceived event presented with this particular construction is indirect. Had it been presented as from the view of the experiencer in the matrix clause, the pronoun used in (149) would have been *míi* 'my' and not *tasíi* 'her'.

(149) *so* def.msg.nom *šay* thing *ɡaḍ-í* take.off-cv *dach-íi ̣* look-3sg *ta* sub *tasíi* 3sg.gen *dit-i=whaáu* give.pptc-f=adj *rumiaál* handkerchief *hín-i* be.prs-f

'Taking up the thing and looking at it, she saw that it was the handkerchief she had given.' (A:SHY054)

### **14.5.2 Complement-taking modality predicates**

**Modal predicates.** As modality is part of the TMA system as a whole and is expressed morphologically, there are few examples of complement-taking predicates with a purely modal meaning. However, one such predicate is *bha-* 'be able to, can', illustrated in (150). This verb regularly takes an infinitival complement, a strategy (vi), confined, as far as my data is concerned, to this modal verb and the phasal verb *ṣáat-* (see **Phasal predicates** below).

(150) *se* 3pl.nom *har* every *deés* day *akaadúi* recp *paš-áai* see-inf *bhóon* be.able.to.3pl *de* pst 'They were able to meet every day.' (A:SHY007)

### 14 Complex constructions

That this verb is relatively morphologised as a potentialis marker is indicated by its loss of inherent transitivity (otherwise highly unusual in the language). It is instead the complement verb that makes the whole sentence transitive or intransitive and also governs case assignment, while morphological verb agreement is part of the modal segment. In the perfective transitive sentence (151), the agent of 'eating' is coded in the ergative, and *bhá-* agrees in gender and number with 'meat' (masculine singular) as if it were a transitive verb. In (152), which is also perfective transitive, the verb agrees in gender and number with the agreeing host element 'endurance' (feminine singular). In (153), however, the sentence as a whole is (perfective) intransitive, and the subject of 'going' is coded in the nominative and agrees intransitively with *bhá-* in gender and number.


from his friend.' (B:FLW777)

That this exemplifies clause union rather than a full lexical union is supported by the fact that it is still possibile to split the two predicates and insert other constituents between them, as in (154).

(154) *ɡa* anything *man-áa* say-inf *ba* top *koó* who *na* neg *bh-óon* be.able-3pl *de* pst 'No-one could tell them.' (A:JAN021)

Necessity or ablility can also be expressed with complement clauses of complement-taking nouns. In those cases, the Verbal Noun strategy (v) is used. This is exemplified in (155)–(158).

14.5 Complement clauses


**Achievement predicates.** Although it is sometimes difficult to draw a definite line between modal predicates of the ability type presented above and achievement predicates, the main difference in the typical cases lies in implicativity. While the modals above are non-implicative, achievement predicates are implicative, i.e., signalling successful vs. failed performance/realisation. In any case, the Verbal Noun strategy seems to be, if not the only possible, at least the preferred one. Examples are given in (159)–(160). Some achievement predicates require a postposition with the Verbal Noun.


### 14 Complex constructions

Normally the Verbal Noun complement is embedded into the matrix clause. The complement in (160), however, occurs after the matrix clause and is thus added explanatorily rather than being extraposed in a strict sense.

As is clear from examples (162)–(164), achievement (or non-achievement) can also be expressed with complement clauses of complement-taking adjectives.


**Phasal predicates.** As was also pointed out above concerning modals, many phasal notions having to do with termination and continuation, are indicated by means other than complementation, and probably mainly so. Inception, however, can be expressed with at least two different complementation strategies.

One of those strategies is infinitival (vi), whereby the perfective form *ṣáat-* of the polysemous verb *ṣač-* 'climb, etc.' is utilised (165).

(165) *káakeṛ* Kakel *teeṇíi* refl *kúṛi-e* woman-gen *pándee* cause *kaantar-í* go.mad-cv *ba* top *biabeen-íim* wilderness-pl.obl *ɡir-áa* turn-inf *ṣáat-u* begin.pfv-msg 'Kakel went mad due to this with his wife and began wandering about in the wilderness.' (B:FOR020)

Unlike the Infinitive-taking modal *bha-* (see **Modal predicates** above), *ṣáat-* is confined to the perfective and remains intransitive, regardless of the transitivity of the complement predicate. In (166), with an intransitive complement, as well as in (167), with a transitive complement, the phasal verb agrees in gender and number with the subject of the matrix clause.

14.5 Complement clauses


Although the construction is a possible example of clause union, *ṣáat-* seems not to have morphologised to the same extent as *bha-*, evidenced also in the parallel use of a Verbal Noun (168) with it instead of an Infinitive.

(168) *ak* idef *čoór* thief *tesée* 3sg.gen *ɡhooṣṭ-á* house-obl *ačíit-u* enter.pfv-msg *ɡhoóṣṭ* house *laṭ-ainíi* search-vn *ṣáat-u* begin.pfv-msg 'A thief entered his house and started to search through the house.' (B:THI002-3)

With another nearly synonymous phasal predicate *široó the-*, exemplified in (169), the Verbal Noun strategy is the only possible one.

(169) *se* def *ṭhaaṭáak-a* monster-obl *bi* also *tas* 3sg.acc *sanɡí* with *kh-ainií* eat-vn *široó* start *thíil-u* do.pfv-msg 'The monster also started to eat together with him.' (A:THA007)

There is evidence, though scanty, that the phasal notion of continuation can be expressed through two (for the purpose) less usual strategies with the complement-taking predicate *dhar-* 'remain', one being a converb (170) and the other a Copredicative Participle (171).


### 14 Complex constructions

**Desiderative predicates.** The Verbal Noun strategy is also put to use in subjectcontrolled complements of desiderative predicates, as in (172)–(176). The Verbal Noun may also be inflected (174) or followed by a postposition (176).


'We have agreed on the division of our game.' (B:FOY065)

Finally, as pointed out already, some utterance predicates, using the *ki*-strategy, especially those concerning "thinking", can equally well be considered desiderative.

### **14.5.3 Complement-taking manipulation predicates**

**Permissive predicates.** When *uṛí-* 'pour, let out' is used as a complement-taking verb, it occurs in two different constructions, both of them with permissive semantics. One is the Verbal Noun strategy ((v)), also occurring in various other complement clauses (see §14.5.1), followed by the postposition *the* 'to', as shown in (177) and (178). This is used in the indicative.

14.5 Complement clauses

(177) *uchí ̣* lift.up.cv *ba* top *se* def *čúti-m-e* paw-pl-gen *zaríia* means *baándi* by *so* def.msg.nom *baṭ* stone *húṇṭraak* upward *chuɡal-áan-u ̣* hurl-prs-msg *se* def *kúuk-a* crow-pl *se* def *muṭ-á* tree-obl *bheš-aníi* sit.down-vn *the* to *na* neg *uṛ-áan-u* let-prs-msg *mhaás* meat *kh-ainíi* eat-vn *the* to *na* neg *uṛ-áan-u* let-prs-msg

'After picking them up, it was throwing the stones upward with the help of his paws, not allowing the crows sit in the tree, not allowing them to eat the meat.' (B:SHB752-6)

(178) *ɡhaḍeeró* elder *miṭínɡ* meeting *xátum* finished *bhíl-ii* become.pptc-gen *pahúrta* after *asaám* 1pl.acc *b-ainií* go-vn *the* to *uṛ-íi=ee* let-3sg=q

'Will the elder let us leave after the meeting is over?' (A:Q6.29.07)

Note that the manipulee of the main verb is coreferent with the agent of the complement verb but is assigned case by the main verb, thus coded accusatively in (178).

The other strategy, involving the particle *ta* (iv), is used in the Imperative, with the literal rendering 'let x (out), and x will do y'. Whereas in the strategy above the complement clause is embedded, the complement clause in this construction, seen in (179), is S-like and extraposed. The main verb does not assign case to the manipulee/complement clause agent in this construction.

(179) *ṛo* 3sg.dist.nom *uṛí* let.imp.sg *ta* sub *rhoó* song *d-íi* give-3sg 'Let him sing!' (A:Q6.15.02)

**Causative predicates.** While the most important means of expressing causativity is morphological – a matter we will return to shortly – manipulation that involves a lower degree of agentive control (Givón 2001b: 45) is often expressed through what are basically utterance predicates, also using the same strategies. Examples are given in (180)–(182). Hereby the manner of causation or persuation is also made explicit (Noonan 2007: 136). The complement clause is either imperative or hortative.

### 14 Complex constructions


'Coming there, he signalled to the child to come to him.' (A:BRE003)

(182) *deeúl-ii* Dir-gen *xálak-am* people-pl.obl *daawát* invitation *dít-i* give.pfv-f *atsharíit-am* Ashreti-pl.obl *the* to *ki* comp *muqaabilá* contest *th-íia* do-1pl

'The people from Dir invited the Ashreti people to have a competition with them.' (A:CHA001)

When, however, agent control is strong, the preferred strategy is to utilise the morphological causative construction *per se*. This can either be in the form of direct causation, in essence deriving a transitive verb from an intransitive (such as 'make sit down, seat' from 'sit down' in (183) and 'hang (something)' from 'hang (by itself)' in (184)), or in the form of indirect causation (such as 'have (someone) drink' from 'drink' in (184), 'have someone tell' from 'tell something' in (185) and 'have someone educated' from 'educate (oneself)' in (186)). With indirect causation, an animate manipulee is marked by *ṣaawaá*, a grammaticalisation of the converb of *ṣaawá-* 'put on, dress, turn on'.


14.6 Relative clauses

*the* do.cv *šuí* good.f *zhay-íim* place-pl.obl *ɡhin-í* take-cv *ɡir-áan-a* turn-prs-mpl

'In the winter they need to be given oak-branches during the night, and in the day they need to be given water to drink, and to be fed grass in nice places.' (A:KEE091)

(185) *xalk-íim* people-pl.obl *ṣaawaá* manip *teér* passed *bhíl-a* become.pptc-mpl *qiseé* story.pl *th-awóol-a* do-caus.pfv-mpl

'He had the people tell him all that had happened.' (A:UXW057)

(186) *áa* idef *putr-á* son-obl *ṣaawaá* manip *ma* 1sg.nom *taalím* education *th-awa-áan-u* do-caus-prs-msg 'My son can become educated [lit: I make my son study]' (A:KEE082)

In this construction, clause union is complete, and we can also define it as a total lexical union (Noonan 2007: 86) or co-lexicalisation of the "cause" and the "caused event". Note that this is therefore not in a strict sense an instance of complementation. It is only included here for comparative reasons.

### **14.6 Relative clauses**

Relativisation in Palula offers an unusual analytical challenge, partly due to the many different strategies that seem to be available, and partly due to their relative scarcity in natural discourse, whether spoken or written.

Generally speaking, the unmarked relative clause (or its functional equivalent) is preposed to the modified NP as well as the main clause in its entirety. Although there is a preference for co-relative (also referred to as relative-correlative) constructions, as described by Downing (1974), most of them are not of the kind typical of major IA languages (Masica 1991: 410–415), which is why I have chosen not to use the term other than in comparison with the more "typical" IA pattern. Neither does Palula share the general Shina preference for participial relative clauses (Carla Radloff, pc), possibly a result of the synchronic non-distinctiveness in Palula between many participles and the finite verb forms historically derived from the former.

There is no attempt on my side to reflect the relative frequency of occurrence by the order with which the various strategies and types of relativisation are outlined below. The number of examples of relative clauses is basically too small to

### 14 Complex constructions

present any reliable statistics. Extraposition with *ki* (§14.6.7) is not strictly a separate type of relative clause but rather an additional, but nevertheless frequently used, strategy, primarily related to presentational structure and other pragmatic functions.

Restrictive and non-restrictive relative clauses overlap to a large extent, although it seems that extraposed clauses are favoured with non-restrictive clauses.

### **14.6.1 Relative clauses with a full NP**

Most similar to the typical Urdu-Hindi relative-correlative construction are those sentences that have a full NP in a preposed modifying clause corresponding to a full NP in the main clause. In (187)–(188), the otherwise interrogatively used *khayí* 'which' is used as an adnominal relative pronoun, correlating with a demonstrative from the remote set. Although the noun can be repeated in full in the main clause along with a demonstrative, it can equally well be coreferenced by a demonstrative alone.

Whereas the modifying clause is dependent on the main clause for its interpretation, the main (indicative and finite) clause can in most cases stand alone as a complete sentence. The modifying (relative) clause occurring on its own (which is not always possible due to a non-basic word order), however, would be interpretable as a content question rather than as a statement. Following Givón (2001b: 182), a construction like this is essentially paratactic, the linking device between the two clauses being deictic rather than a matter of subordination or embedding.

(187) *báaba* father.obl *khayí* which *xarčá* expenditure *dawóol-u* ask.for.pfv-msg *hín-u* be.prs-msg *eesó* rem.msg.nom *xarčá* expenditure *ɡhin-í…* take-cv

'Taking the reimbursement that the father has demanded…' (A:MAR037-8)

(188) *se* def *baaqimaandá* additional *mehmaan-aán* guest-pl *khayí* which *hín-a* be.prs-mpl *práač-a* guest-pl *khayí* which *hín-a* be.prs-mpl *eetanaám* 3pl.acc *the* to *ɡúuli* bread *d-áan-a* give-prs-mpl 'We feed those other guests that are present.' (A:MIT024)

Some modifying clauses with adnominal *khayí* are possibly embedded rather than preposed, as in (189), where we may further suggest that a noun *xálak* 'people' is the implicit head. However, the analysis is still pending in anticipation

14.6 Relative clauses

of more natural data of a similar kind. It should also be noted that a pronoun *ni* 'they (here)' from the proximal set is being used and not one from the usual remote set.

(189) *tipúka* nowadays *ni* 3pl.prox.nom *ba* top *khayí* which *hín-a* be.prs-mpl *dhii-á* daughter-obl *díi* from *khooǰ-ainií* ask-vn *zarurí* necessity *bh-áan-u* become-prs-msg 'Nowadays there are those for whom it is necessary to ask their daughter [who she wants to be married to].' (A:MAR019)

In (190), the relative clause is probably internal. Note that the construction as a whole is extraposed to the temporal clause.

(190) *paturaá* back *nuuṭíl-u* return.pfv-msg *ta* sub *khayí* which *zhay-í* place-obl *wée* in *asím* 1pl.erg *ǰinaazá* corpse *khaṣeel-í* drag-cv *wheelíl-u* bring.down.pfv-msg *de* pst 'When I returned to the place where we had dragged the dead body…' (A:GHA044)

### **14.6.2 Indefinite-conditional relative clauses**

Quite similar to, and possibly not altogether distinct from, the relative clauses above are those that contain an interrogative pronoun, often in combination with (or even fused with) an indefinite word or particle *ɡalá (B ɡalé)* meaning 'ever'. The resulting meaning is something like 'whoever, whatever', referring to one in a group of potential referents having a non-specific reference. The relative clause is often a Conditional clause, and is therefore more clearly subordinate than the above described construction, and the correlative in the main clause is referred to with a demonstrative or personal pronoun from the remote set (if at all explicit). Examples are given in (191)–(193).


'The one [of us] whose neck can reach the sheaf shall eat.' (B:SHI016)

### 14 Complex constructions


The relativised NP can thus be a possessor, an ergative subject or a direct object. It does not even need to be a noun phrase that is relativised in this way; the same basic construction is used in some temporal clauses with *kareé=ɡala* 'whenever' (see §14.4.1), in this context always without an explicit correlative.

When the noun phrase being relativised is a nominative subject, as in (194), inclusion of *ɡalá* does not seem to be necessary.

(194) *koó* who.nom *eesé* dist *ṭeem-íi* time-gen *haazír* presence *na* neg *heensíl-u* stay.pfv-msg *heentá* condl *tasíi* 3sg.gen *mux-íi* face-gen *nikh-aaṇḍeéu* come.out-oblg *bh-áan-u* become-prs-msg

'Anyone not present at the time, will have to come and visit [later].' (A:MAR048-9)

A similar construction (without *ɡalá*, and not explicitly conditional) can also be used with manner adverbials, as exemplified in (195).

(195) *kanáa* like.what *húuši* wind *ziaát* most *bh-íi* become-3sg *de* pst *eendáa* like.that *so* def.msg.nom *musaafár* traveller *šukhaáu* coat *teeṇíi* refl *huǰut-í* body-obl *pharé* toward *pail-óo* fold-3sg *de* pst

'The stronger the wind blew, the more closely did the traveller fold his cloak around him.' (A:NOR006)

<sup>9</sup> The form of the determiner is not the expected one, since the head is non-nominative.

14.6 Relative clauses

### **14.6.3 Gapped relative clauses**

What seems to be a major strategy for relativisation is a preposed gapped modifying clause, i.e., one in which the relativised NP is not explicitly present at all. Even here the correlative NP in the main clause often occurs with a demonstrative from the remote set, thus in essence being of the same kind as the relativecorrelative construction we have seen examples of already.

The modifying clause may be indicative finite, as in examples (196)–(199), with the positions direct object, oblique object coding source and transitive subject, as well as the oblique object coding recipient being relativised.


The modifying clause may also occur as a preposed Conditional clause, in (200) relativising the possessor NP.

(200) *áaṛu* knot *heensíl-u* stay.pfv-msg *heentá* condl *eesó* rem.msg.nom *ḍhínɡar* timber *aaindá* future *dapáara* for *nakaám* failed *bh-áan-u* become-prs-msg 'Such timber that has knots will be worthless in the future.' (A:HOW032)

Another possibility, seen in (201) and (202), is to use preposed dependent Converb clauses for relativisation.

### 14 Complex constructions


A reflexive pronoun, as in (203), can also occur in the modifying clause.

(203) *teṇ* red *teeṇíi* refl *násel* descent *pašaá* show.cv *ba* top *muxáak* before *hasó* 3msg.rem.nom *tshaak-íi* taste-3sg 'The one who is able to prove his noble heritage may taste first.' (B:SHI006)

Some constructions that technically are relativisations with a preposed Converb clause are lexicalised to a large extent, and they function as standard presentational constructions when new participants or places are introduced into a discourse. Note in (204) that the coreferential noun phrases in the main clause are indefinite.

(204) *muxáak* before *zamanée* time.gen *čakaaḍhám* Chakadham *than-í* call-cv *ak* idef *díiš-a* village-obl *kaakeṛ* Kakel *than-í* call-cv *ak* idef *méeš* man *heensíl-u* stay.pfv-msg *de* pst

'Once there was, in a village called Chakadham, a man called Kakel.' (B:FOR001)

While most, if not all, of the exemplified relative clauses above are restrictive, the postposed modifying finite clause in 205 is clearly non-restrictive; it merely adds supplementary information about *míiš* "man" (who, incidentally, is identified with another, preposed, modifying *thaní*-clause).

(205) *áa* one *ba* top *habibulaxaán* Habibullah.Khan *thaní* qot *míiš* man *de* be.pst *mar-í* die-cv *hín-u* be.prs-msg 'And one was a man named Habibullah Khan, who is dead now.' (A:ACR023-24)

14.6 Relative clauses

### **14.6.4 Gapped relative clauses with a complementiser**

Sometimes in gapped relative clauses, a particle *ɡa* is used as a complementiser or relativiser, otherwise occurring as an indefinite ('some', 'any') or an adnominal interrogative ('what', 'what kind of') pronoun. This particle or word is invariable and thus is not coreferential with the modified NP. Instead it seems to merely mark the modifying clause vis-à-vis the main clause. Examples are given in (206)– (209).

There does not seem to be any significant covariation between the presence of this marker and any particular position being relativised, although it possibly is used more readily (and maybe even non-optionally) when an oblique constituent is relativised, such as the location in (208) and (209).


'This woman, who is taking care of my daughter, brought all this firewood' (A:REQ17)

(208) *máa=ee* 1sg.nom=cnj *tu* 2sg.nom *ɡa* rel *bheš-í* sit.down-cv *hín-a* be.prs-mpl *eení* prox *ɡhooṣṭ-á* house-obl *šíiṭi* inside

'In the house where you and I are sitting…' (A:HUA014)

(209) *henrík* Henrik *de* put.cv *ɡa* rel *ɡíia* go.pfv.pl *eeṛáa* there *rhalá* above *ba* top *qilaá* fort *de* be.pst 'Above [the place] where you took Henrik, there was a fort.' (A:CAV004)

The position of *ɡa* is probably preverbal, but further investigation is needed to back up this observation. In (210), the modifying clause is postposed, and here *ɡa* occurs clause-initially.

(210) *ée* oh! *kučúru* dog *ɡa* rel *so* def.msg.nom *wíi* water *ǰhuṭá* dirty *thíil-u* do.pfv-msg 'It must be the dog that has contaminated the water' (A:PAS038)

### 14 Complex constructions

In (211), the modifying string 'who used to walk with the millstone' seems to function simultaneously as a postposed non-restrictive relative clause to 'the Dameli strong man' and as a preposed restrictive relative-correlative clause to 'that (one) didn't take up the challenge'.

(211) *so* def.msg.nom *ɡiḍúuču* Dameli.man *paalawaáṇ* strong.man *so* def *yambaáṭ* mill.stone *ɡa* rel *ɡhin-í* take-cv *b-íi* go-3sg *de* pst *eesó* 3msg.rem.nom *na* neg *ṣandíl-u* take.up.challenge.pfv-msg *maní* hsay 'The strong man of Damel, who used to walk with the millstone, apparently didn't take up the challenge' (A:MAH054)

The modifying clause in (212), beginning with *ɡa*, takes as a whole the place of the transitive subject. Tentatively it is analysed as a (headless) internal relative clause.

(212) *bhunaṛíi* down.below.obl *wée* in *ɡa* rel *šáali* Shali *ṭópa* downside *dít-i* give.pfv-f *šáali* Shali *khéel-i* eat.pfv-f 'Down there, the one that caught Shali ate her.' (A:PAS059)

### **14.6.5 Pronominal relative clauses**

Some data (213)–(215)) suggests the possibility of regular anaphoric (personal or demonstrative) pronouns being used in the modifying clause, whether preposed or postposed. The evidence, however, is too scanty to draw any definite conclusions. Suffice it to say that the positions being thus relativised are low on the accessibility hierarchy (postpositional object and possessor, respectively), which is fully in line with typological predictions (Keenan 1985: 147–148; Andrews 2007a: 226). On the other hand, we have examples of relativisation of recipient (199) as well as possessor (200) by gapping.

(213) *anú* 3msg.prox.nom *dhoóṛ* yesterday *míi* 1sg.gen *putr-á* son-obl *hanís* 3sg.prox.acc *the* to *kitaáb* book *dít-i* give.pfv-f *hasó* rem.msg.nom *phoó* boy

'This is the boy who my son gave a book to yesterday.' (B:DHE5367)

14.6 Relative clauses

(214) *lawrái* Lowari *thaní* qout *zhaáy* place *hín-i* be.prs-f *eetasíi* 3sg.rem.gen *se* def *ɣar-í* peak-obl *the* to *asím* 1pl.erg *tas* 3sg.acc *phedóol-u* bring.pfv-msg 'We reached to the peak of a place called Lowari with him [lit: a place,

named Lowari, to whose peak we brought him].' (A:GHA029)

(215) *asím* 1pl.erg *so* def.msg.nom *méeš* man *ɡhašiǰíl-u* be.caught.pfv-msg *baazúur-a* bazaar-obl *phará* along *so* def.msg.nom *kuṇaák* child *ba* top *atesée* 3sg.rem.gen *ɡhooṣṭ-á* house-obl *the* to *phreyíl-u* send.pfv-msg 'We sent the child to the house of the man who had been caught in the bazaar.' (B:ANG018)

Pronominals used in the modifying clause are otherwise, as we shall see, the normal case with so-called extraposed *ki*-constructions (see §14.6.7).

### **14.6.6 Nominalisation and the use of participles**

A less-accessible noun phrase may be expressed through nominalisation, whereby a Verbal Noun becomes the possessor of the relativised entity. This seems to be especially common with time, location and means, as seen in (216)–(219). The genitive case of the Verbal Noun is only explicitly present in the B variety (*-e*), whereas the genitive has been levelled in A.


'He is going to kill me instead of rewarding me [lit: in the place of giving a reward to me].' (B:DRB020)

(218) *bhíiru* he.goat *uṛ-ainií* let.out-inf *tartíb* method *ba* top *eeṛé* dist 'That is the method of letting out the male goat [to the female goats].' (A:KEE073)

### 14 Complex constructions

(219) *ma* 1sg.nom *tu* 2sg.nom *the* to *ɣeer-í* cave-obl *šíiṭi* inside *be* go.cv *neečíir* game *aṭaníi-e* bring.inf-gen *baát* word *d-áam* give-1sg

'Once we have got into the cave, I will tell you about the catching of game [lit: give the word of bringing game].' (B:FOY048)

Less commonly, Perfective Participles may be used attributively, as in (220)– (222), and in those cases, the participials relativise the direct object. As an additional means of distinguishing the participle from the formally identical finite perfective verb, an explicit (but optional) participial marker (also referred to as an adjectiviser in this work), such as *bhaáu* in (222), can be used.


A corresponding (headless) construction for relativising an agent subject is by using the Agentive Verbal Noun, as seen in (223). That, however, does not seem to be a very common strategy.

(223) *teewiz-í* amulet-pl *th-áaṭ-u* do-ag-msg *le* dist *peeriaán* djinn.pl *ɡaḍ-í* take.out-cv *ṣeeka-áaṭ-u* lead.out-ag-msg 'He was an expert in making amulets and providing deliverance from djinns.' (A:HUA129)

Alternatively, the present-tense verb is itself used without an overt head, as in (224). This has an implicit habituality read into it, whereas the corresponding

14.6 Relative clauses

non-habitual meaning would have to be expressed with an explicit correlative head (225).


### **14.6.7 Extraposed** *ki***-constructions**

The possibility of extraposing a modifying clause is facilitated by the particle or complementiser *ki* (see §14.5.1). The sentences in (226)–(228) are all elicited, and are thus not necessarily the most natural way of expressing these ideas. However, regarding them as a general indication of grammaticality, we can conclude that the construction works for a number of different positions. Note also that the emphatic or demonstrative pronoun is used in the main clause, whereas a nonemphatic anaphoric pronoun is used in the modifying *ki-*clause, although the latter is optional for the subject position.

(226) *anú* 3msg.prox.nom *eesó* rem.msg.nom *míiš* man *ki* comp *níi* 3sg.prox.obl *se* def *kúṛi* woman *mheeríl-i* kill.pfv-f 'This is the man who killed the woman.' (A:HLE2617) (227) *anú* 3msg.prox.nom *eesó* rem.msg.nom *míiš* man *ki* comp *nisíi* 3sg.prox.gen *ɡhooṣṭ-á* house-obl *ma* 1sg.nom *hín-u* be.prs-msg 'This is the man whose house I am [living] in.' (A:HLE2618) (228) *anú* 3msg.prox.nom *eesó* rem.msg.nom *míiš* man *ki* comp *(nu)* 3msg.prox.nom *dhoóṛ* yesterday *yhóol-u* come.pfv-msg *de* pst 'This is the man who came yesterday.' (A:HLE2618)

### 14 Complex constructions

It seems that the construction is particularly favoured in presentational discourse, and a number of correlative expressions, belonging to various parts of speech (see below), can be linked to an extraposed *ki*-clause, expressing a variety of pragmatic- or discourse-related functions (some of them touched upon when discussing clauses with adverbial functions (§14.4) and complementation (§14.5)). In most cases, however, the *ki*-clause as a whole corresponds to the correlative in the main clause, rather than to a particular relative word in it.

**Nominal.** *ki*-clauses that correlate to a nominal element in the main clause are primarily heavy-shifting, as can be seen in (229)–(231).


(231) *íṇci-e ̣* bear-gen *xiaál* opinion *haṛé* dist *bhíl-i* become.pfv-f *ki* comp *karáaṛu* leopard *máa=the* 1sg.nom=to *uṭik-í* jump-cv *de* give.cv *hín-u* be.prs-msg *thaní* qot 'The bear thought that the leopard had attacked him.' (B:BEL320)

**Adjectival.** *ki*-clauses that correlate to an adjectival element in the main clause express comparison or provide exemplification or specification. Some examples are given in (232)–(235).

(232) *anú* prox.msg.nom *míiš* man *eetí* so.much *maldaár* wealthy *ki* comp *nisíi* 3sg.prox.gen *ɡhooṣṭ-á* house-obl *čailúṭii* sparrow.gen *c̣hiír* milk *bi* also *lhayiǰ-áan-u* be.found-prs-msg 'This man is so rich that you even find sparrow's milk in his house.' (A:DHN6691)

14.6 Relative clauses

(233) *dhríɡ-u* tall-msg *ba* top *eetí* such *ki* comp *loomuṭ-íi* deodar.tree-gen *aḍaphaár* halfways *tií* until *phed-í* reach-cv *hín-u* be.prs-msg

'It was so tall that it reached halfway up a deodar tree.' (A:HUA076)

(234) *baazí* a.few *xálak-a* people-pl *hateeṇ-á* like.that-mpl *hín-a* be.prs-mpl *ki* comp *se* 3pl.nom *bakareelí* shepherding *th-áan-a* do-prs-mpl

'A few people are engaged in goat and shepherding.' (B:DHN5263)

(235) *áak* idef *eeteeṇ-ú* like.that-msg *waxt* time *yhóol-u* come.pfv-msg *ki* comp *atshareet-á* Ashret-obl *wée* in *xálak* people *biǰóol-a bhíl-a*

several-mpl become.pfv-mpl

'This was the time when the population in Ashret had started to increase.' (A:GHU001)

Although not examples of relativisation in a strict sense, it should be noted that the same construction is used with clause adverbials (as has already been illustrated many times before). In (236) the *ki*-clause corresponds to a reason, and in (237) it serves as an explanation.


## **15 Sentence modification**

### **15.1 Introduction and overview**

Some of the more salient sentence modifications are presented in this chapter, especially those that are related to particular markers or constructional types. They include interrogative sentences, negation and switch-topicality.

### **15.2 Interrogative sentences**

The two main types of interrogative sentences, polar and constituent interrogatives (König & Siemund 2007: 290–303), are discussed and exemplified in the sections below.

### **15.2.1 Polar interrogatives**

Polar interrogatives, or 'yes/no' questions, are formed by adding a sentence final particle *ee* (B *aa*) to a declarative sentence without any subsequent change in the word order. This sentence-final particle is accompanied by a slight rising intonation. Examples are shown in (1)–(3).


The clitical nature of the particle is evident in that it attaches to the sentencefinal word, regardless of its part of speech, and in a phonological sense becomes

### 15 Sentence modification

part of that word; for instance a final vowel /a/ coalesces entirely with the particle. As an SOV language, the sentence normally ends in a verb, and therefore the particles attach to the verb, but that is not the case in most nominal sentences lacking an overt copula, as can be seen in (4)–(6).


Apart from ordinary polar questions, there are at least two different ways of forming tag questions, one (7) with a sentence-final *ɡa* 'what' (see §15.2.2), and another (8) with *nheé* (possibly derived from the negation *na* and the question particle -*ee*).


Alternative questions are expressed with *ki na* 'or not', as in (9), optionally with a repetition of the verb, as in (10). This, too, can be used as a tag question, as in (11).


15.2 Interrogative sentences

(11) *búd-u* understand.pfv-msg *ki* or *na* neg 'You understood, didn't you? [lit: Did you understand or not?]' (A:SHA033)

### **15.2.2 Constituent interrogatives**

Constituent interrogatives, or parametric questions, are formed by replacing a questioned constituent with an indefinite-interrogative pronoun. Its position is usually immediately preverbal or (in the case of adnominal constituents) occurs in the phrase immediately preceding the verb.

Palula has a rather large inventory of interrogative words that may replace various constituents.

**Questioning core constituents of the clause.** Examples (12)–(14) illustrate the questioning of the nominative subject (*koó* 'who'), the oblique (i.e., ergative) subject (*kií* 'who') and the accusative direct object (*kaseé* 'whom'), respectively, when referring to human beings.


When questioning non-human (in particular inanimate) core constituents, *ɡubáa* 'what' is used instead. As can be seen in (15)–(17), the form stays the same regardless of its syntactic role (although questioned inanimates only seem to appear either as intransitive subjects in copular clauses or as direct objects in verbal clauses).

(15) *amzarée* lion.obl *thíi* 2sg.gen *káaṇ-a* ear-obl *wée* in *ɡubáa* what *maníit-u* say.pfv-msg 'What did the lion whisper in your ear?' (A:UNF015)

### 15 Sentence modification


**Questioning postpositional constituents of the clause.** For questioning postpositional human constituents, as in (18)–(20), the accusative form *kaseé* (the same as the direct object form) is used.


Postpositional or non-core inanimate constituents can also be questioned, as in (21)–(22), again using *ɡubáa* 'what', supplied with any case inflection required by the postposition in question.


**Questioning constituents of the noun phrase.** Various modifiers of a noun can be questioned, such as a possessor: *kasíi (B kasée)* 'whose' (23); a cardinal numeral or quantifier: *katí* 'how many, how much' (24); an ordinal numeral: 15.2 Interrogative sentences

*katíma* 'in which' (25); an adjective *kateeṇú* 'what kind of (person)' (26) (adjectivally inflected) or *ɡa* 'what kind of (thing)' (27); and a determiner *khayú* 'which, what' (28) (demonstratively inflected) or *khayú áak*/*khayáak* 'which one' (29).


'How are you doing? [lit: what position is there]?' (A:CHN070110)


While the interrogative words exemplified above function attributively and occur before the noun they modify, predicatively functioning modifiers being questioned occur immediately preverbally, as can be seen in (30)–(32). Note that predicatively used *kateeṇí* corresponds to attributively used *ɡa* and *kateeṇí* alike. In (31), an additional adjectival interrogative *katíiti* is exemplified, questioning the size of an inanimate noun.

### 15 Sentence modification


A rather non-specific *kanáa* 'what, like what' is used to question a predicate phrase (see also below), nominal or adjectival, in (33)–(36). It may also be interpreted as questioning the host position of conjunct verbs (especially those with *the-* 'do').


'When Nagarjuti came, he asked her, "What happened [to you]?"' (A:MAB029)


**Questioning adverbials.** Various types of adverbials can be questioned using indefinite-interrogative words.

For questioning temporal and local adverbials, *kareé* 'when' and *ɡóo* (B *ɡúu*) 'where' respectively are used, as in (37) and (38).

15.2 Interrogative sentences


As *ɡóo* only codes location *per se*, goal and source need to be specified by postpositions and case, *ɡóo=the* (B *ɡúu=the*) 'to where' (39) and *ɡóoii* or *ɡóoii thíi* (B *ɡúue thíi*) 'from where' (40)–(41). For questioning the goal an alternative distinct indefinite-interrogative adverb, *kíi* (in B *kési*) 'whither' is often used, as in (42).


'Having stood up, he said, "O bald one, where did you come from?"' (A:KAT030)

(42) *íṇc-ạ* bear-obl *maníit-u* say.pfv-msg *hín-u* be.prs-msg *ki* comp *ée* o *lhéṇḍu* bald.one *kíi* whither *bi-áan-u* go-prs-msg 'The bear said, "O bald one, where are you going?"' (A:KAT020)

Adverbials corresponding to clauses, rather than to single adverbs or adverbial phrases, can be questioned by e.g., *keé* 'why', as in (43), and *kanáa=bhe/=the* 'how, by what means' (see §8.1.4), (44)–(45). Note that *kanáa=bhe*, which consists of *kanáa* 'like what' and the converb of the verbaliser *bhe-* 'become', is used in intransitive clauses, and *kanáa=the*, which is *kanáa* and the verbaliser *the-* 'do', is used in transitive clauses.

(43) *kúṛi* wife *ma* 1sg.nom *díi* from *khooǰóol-u* ask.pfv-msg *ki* comp *keé* why *ru-áan-u* cry-prs-msg *thaníit-u* say.pfv-msg *ta* sub 'My wife asked me, "Why are you crying?"' (A:HUA101)

### 15 Sentence modification


### **15.2.3 Subordinate interrogative clauses**

As described in Chapter 14, indirect discourse is extremely limited in the language. There is subsequently no category of subordinate or indirect questions clearly distinct from direct questions. When questions occur in reported speech or reported perception, they always appear as if uttered by the quoted speaker or experiencer, as in (46).

(46) *so* def.msg.nom *insaán* human.being *bulaḍ-íim* search-cprd *b-íi* go-3sg *de* pst *páand-a* path-obl *pharé* along *ki* comp *so* def.msg.nom *insaán* human.being *kateeṇ-ú* what.kind-msg *šay* thing *hín-u* be.prs-msg

'He left in search of that man along the road to find out what sort of thing he was [lit: He left to search the man along the the road: What kind of thing is that man?]' (A:KIN004)

### **15.2.4 Interrogatives in exclamative use**

Some sentences that essentially are interrogative sentences in form may also be used in an exclamative function, with or without special interjections utteranceinitially. Some examples are provided in (47)–(50).

(47) *dun-áaṭ-u* think-ag-msg *bhíl-u* become.pfv-msg *hín-u* be.prs-msg *ki* comp *aní* 3fsg.prox.nom *ba* comp *kateeṇ-í* what.kind-f *ǰuánd* life 'He started thinking, "What a life!" ' (A:KAT057)

15.3 Negation


### **15.3 Negation**

The main strategy for negation is by means of a separate and invariable negative particle, *na*.

### **15.3.1 Basic sentence negation**

The pragmatically unmarked position of the negative particle is preverbal, as is evident from (51)–(53), regardless of the TMA categories reflected in the predicate.


Also when there is an auxiliary verb present, as in the periphrastically expressed TMA categories, the negative particle precedes the main verb. This can be seen in examples (54) and (55).

### 15 Sentence modification


This is also true of passives (56) and non-finite verb forms (57).


Predicate noun phrases without an overt copula are negated by the negative particle alone, thus occurring clause-finally, as seen in (58) and (59).


As for the position being strictly preverbal or being before the entire verbal group, the data shows some variability.

With the modality verb *bhá-* 'be able to', the negative particle almost always precedes the modal as well as the main verb, as in (60)–(61).

(60) *dúi* other *ta* cntr *ɡaṭíl-u* win.pfv-msg *áak* idef *ḍaaku-aan-óom-ii* robber-pl-obl-gen *qilaá* fort *tíi* 3sg.obl *na* neg *ɡaṭáa* win.inf *bhóol-u* be.able.to.pfv-msg 'When he had won everything else, there was a den of thieves that he could not capture.' (A:PIR008)

15.3 Negation

(61) *taníi* 3pl.gen *báaba* father.pl *tu* 2sg.nom *na* neg *kháai* eat.inf *bh-óon* be.able.to-3pl 'Not even their fathers will be able to eat you.' (A:KAT074)

However, there are occasional exceptions, as example (62) shows, where the negative particle occurs between the modal and the main verb.

(62) *patráak* back *nikhái* get.out.inf *na* neg *bháam* be.able.to.1sg 'I won't be able to get back out [of the well].' (B:FOX006)

With conjunct verbs there is even more variability. Although in most cases the negative particle occurs right before the main verb and after the host element, there are some (incorporating) conjuncts where the negative particle occurs sometimes before the whole conjunct, (63)/(65), sometimes between the host element and the inflected verb, (64)/(66). Further research may reveal certain conditions, pragmatic or of some other kind, that must be met for each to occur.

(63) *ma* 1sg.nom *na* neg *ṭinɡ* challenged *bhíl-u* become.pfv-msg 'I could not stand up against him.' (A:HUA108)

(64) *nis* 3sg.prox.acc *the* to *koó* anybody *ṭinɡ* challenged *na* neg *bh-áan-a* become-prs-mpl 'Nobody could stand up against him.' (A:JAN062)


time.' (A:ABO011) Although it was stated at the beginning of this section that the negative particle is invariable, it may nevertheless fuse phonologically with an adjacent morpheme, such as with the present-tense form of the copula (*na hínu > náinu* etc.,

### 15 Sentence modification

(67) *yéei* mother *uth-í* stand.up-cv *anɡaá* conscious *bhe* become.cv *dach-íi ̣* look-3sg *ta* sub *kuṇaák* child *náin-u* neg.be.prs-msg *darák* appearance *náin-i* neg.be.prs-f *índa* here *dít-i* give.pfv-f *eeṛáa* there *dít-i* give.pfv-f 'The mother woke up and could not see the child or any sign of him whereever she turned.' (A:BRE007)

Variations resulting from referential and pragmatic factors are discussed in the following sections.

### **15.3.2 Negative pronouns/particles**

Negation is generally not 'permeable' (Ramat 2006: 563), i.e., a negative morpheme occurs only once in a negated clause, as in (68)–(70).


The pronoun *ɡa* belongs to a set of indefinite-interrogative pronouns and is in itself neutral (rather than negative). It could, however, be argued that the combination *ɡa na* phonologically is one word, as occurring in (68) and (69), and as such constitutes a negative pronoun. In any case we have only one morpheme with a clearly negative semantics in sentences like these.

The possible emergence of a set of negative pronouns is even more obvious with combinations indefinite-interrogative pronoun + *bi* 'also' + *na*, as in examples (71)–(73). Neither do these "negative compounds" occur with an additionally negated verb; instead the entire predication lies in the scope of this negative pronoun, itself in preverbal position in the clause. Probably the morpheme *bi* contributes an added emphasis to the negation, approximately corresponding to 'at all, else'.

15.3 Negation


A marginal case where double negation may be argued to occur is when a particle *hiǰ* 'at all' (derived from Persian or Pashto where it has a clearly negative value) is added to the indefinite-interrogative *ɡa* in a negated clause, such as in (74) and (75). We may on the other hand say that it is used in a way very similar to *bi*, thus primarily adding emphasis to the already negative expression.


### **15.3.3 The scope of negation**

**Negation of subunits.** In the examples, so far, the entire predicate lies in the scope of the negation. But it is also possible to negate only a phrase or a subunit of a clause, as in (76)–(78). Here, however, the negated unit is especially marked or extraposed, and the negation does not occur in the "regular" immediately preverbal position, and therefore the scope of the negation has to be interpreted as narrowed down.

(76) *ɡúum* go.pfv.msg *táa* there *the* to *róot-a* night-obl *dees-á* day-obl *na* neg *róot-a* night-obl *ɡúum* go.pfv.msg 'He went there, during the night, not during the day.' (A:PIR015-7)

### 15 Sentence modification


The exact nature of and the mechanisms available for negating subunits is an area needing further research.

**Negation in complex constructions.** With one complex construction involving the modal *bha-* 'be able to', already touched upon briefly above (see examples (60)–(62)), we observed that the negative particle tends to precede both the complement-taking verb and the infinitival complement. This underlines the high degree of clause integration pointed out in §14.5.2. Although it is in fact the ability that is negated, the negation occurs closest to what is formally the subordinate verb, and it is not even possible to negate the subordinate verb only.

In other complex constructions, with a Verbal Noun in the complement and where the bond is not quite so tight between the complement-taking predicate and the verbal element of the complement, it is obvious that either of the two clauses can be negated, either that coded by the matrix verb, as in (79) and (80), or the one coded by the Verbal Noun, as in (81) and (82).


15.3 Negation

### *mhaás* meat *khainíi* eat.vn *the* to *na* neg *uṛ-áan-u* let-prs-msg

'After picking them up, it was throwing the stones with the help of its paws, not allowing the crows to sit in the tree, not allowing them to eat the meat.' (B:SHB752-6)


As far as Conditional constructions are concerned, I only have clear examples of the 'if'-clause being negated, as in (83), in which case the negative particle occurs immediately preverbally.

(83) *thíi* 2sg.gen *ninaám* 3pl.prox.acc *na* neg *phedúul-a* take.pfv-mpl *heentá* condl *qeaamatée-e* judgement-gen *dees-á* day-obl *ma* 1sg.nom *tu* 2sg.nom *díi* from *khooǰ-áam* ask-1sg

'If you don't take these [to her], I will ask you on the day of judgment.' (B:FLW800)

In a number of coordinate constructions where one or more elements are negated, the negative particle does not occur in preverbal position, but instead appears in more or less fixed positions according to the particular construction in question, including sentence-finally in postsection constructions (see §14.2.2) and clause (or phrase) initially or as parts of the strings *bi na…bi na* and *na ta…na ba* in rejection constructions (see §14.2.4).

### **15.3.4 The pragmatics of negation**

A couple of observations on the pragmatics of negation should be pointed out in particular.

### 15 Sentence modification

The first concerns possessive negation. Just as one main strategy of expressing possession is by means of an existential construction, the negated counterpart is in the form of a denial of existence, whether alienable as in (84) or inalienable as in (85).

	- 'He had no sons [lit: His sons were not].' (B:FOR003)

Such a possessive clause can also, as in (86), include an indefinite-interrogative pronoun.

(86) *tasíi* 3sg.gen *ba* top *ɡa* what.kind *wasá* capacity *na* neg *heensíl-u* stay.pfv-msg 'He had no strength [lit: His any capacity was not].' (A:GHA017)

The other comment concerns so-called Obligative constructions. The positive (non-negated) Obligative codes necessity or obligation (see §10.2.3), especially with transitive verbs. The obligative verb form negated, however, has a primarily prohibitive reading, corresponding to 'it is not advisable, one should not, one should avoid'. This is particularly the case with transitive verbs, as in (87) and (88), whereas a negated intransitive Obligative (89) can imply non-ability.


15.4 Switch-topicality

A simple denial of an obligation, on the other hand, is expressed by other constructions, such as those in (90) and (91).


go.pfv.msg

'Since his brother was going to Peshawar, he didn't go [i.e., it was not necessary any longer].' (A:CHE080304)

### **15.3.5 Prohibitive negation**

As was mentioned in §10.2.1, there is no prohibitive category morphologically distinct from the Imperative. Instead prohibitive negation is formed by the same means as indicative negation, i.e., by the imperative verb form being immediately preceded by the negative particle *na*, as shown in (92) and (93).


### **15.4 Switch-topicality**

Although pragmatic- and discourse-related functions are only marginally part of this work, at least one very frequently occurring particle, *ba*, which has a rather wide scope, will need some brief and tentative comments.

We have already come across *ba* as it occurs together with another particle, *ta*, in coordinate contrasting or adversative expressions (see §14.2.1), but it also occurs alone as an expression of topicality or emphasis. While a clearly identifiable

### 15 Sentence modification

or recently referred to subject (with its expected topicality) normally is not particularly marked for topicality, it seems most other entities need to be identified as such by the specific postposed switch-topic (Andrews 2007b: 149) marker *ba*.

In example (94), the subject of the first sentence is a particular witch; then in the next sentence the man Pashambi, who, as the main character of this historical account, has been previously introduced but not recently referred to, is reintroduced as the topic noun phrase and is thus marked with the switch-topic marker *ba*.

(94) *úuč-a* Uch-obl *se* 3fsg.nom *be* go.cv *heensíl-i* stay.pfv-f *hín-i* be.prs-f *pašambeé* Pashambi *ba* top *bakáara* flock *ɡhin-í* take-cv *úuč-a* Uch-obl *the* to *ɡúum* go.pfv.msg *hín-u* be.prs-msg 'She had gone to live in Uch. [Now it so happened that] Pashambi was going with his flock to Uch.' (A:PAS113-4)

Many times when using the marker *ba*, an explicit contrast with an immediately preceding subject is obtained, as in (95) and (96), which is not very different from the use of 'while' or 'however'/'though' in English.

(95) *míi* 1sg.gen *ɡhoóṣṭ* house *lookúṛi* Lokuri *hín-u* be.prs-msg *iskuúl* school *ba* top *asíi* 1pl.gen *kaṇeeɡhaá* Kanegha *hín-i* be.prs-f

'My house is in Lokuri, while our school is in Kanegha.' (A:OUR004)

(96) *tus* 2pl.nom *aakáak* one.each *looṛíi-a* bowl-pl *aṭ-óoi* bring-imp.pl *iṇc̣* bear *ba* top *kaṭamuš-á* Katamosh-obl *the* to *óol* watch *bh-íi* become-3sg

'Go and get a bowl each! The bear, though, will stay here and watch Katamosh.' (A:KAT125-6)

Sometimes, although still being a contrast of sorts, *ba* serves primarily as a signal that similar-looking or otherwise somehow related topics are non-identical, for instance in lists, as the ones in (97) and (98), or genealogical accounts, as in (99).

(97) *koó* who *ba* top *paiseé* money.pl *d-áan-a* give-prs-mpl *koó* who *ba* top *toobaák* gun *d-áan-a* give-prs-mpl *koó* who *ba* top

15.4 Switch-topicality

*ṭeép* tape.recorder *d-áan-a* give-prs-mpl

'Some give money, others give guns, others tape-recorders.' (A:MAR091-3)

(98) *tus* 2pl.nom *hakim-í* ruler-gen *buṭheé* all *putr-á* son-pl *mhaar-úuy-a* kill-imp.pl-q *tes* 3sg.acc *bi* also *mheer-í* kill-cv *kuṛíina* woman.pl *ba* top *ɡhaš-í* take-cv *ukaal-úuy=a* bring.up-imp.pl=q *ɡhooṣṭ-áam* house-pl.obl *ba* top *anɡáar* fire *ṣaa-wúuy=a* put.on-imp.pl=q

'Kill all of the ruler's sons, kill him, [then] take all the women up here, and [then] set their houses on fire!' (B:ATI033-6)

(99) *míi* 1sg.gen *nóo* name *aaxuunseéd* Akhund.Seyd *míi* 1sg.gen *báabii* father.gen *nóo* name *ba* top *ɡulseéd* Gul.Seyd *míi* 1sg.gen *dóodii* grandfather.gen *nóo* name *ba* top *ɣulaamseedmalák* Ghulam.Seyd.Malak *ɣulaamseedmalak-íi* Ghulam.Seyd.Malak-gen *báabii* father.gen *nóo* name *ba* top *sahibǰií…* Sahib.Jee 'My name is Akhund Seyd, my father's name was Gul Seyd, my grandfather's name was Ghulam Seyd Malak, and Ghulam Seyd Malak's father's name was Sahib Jee…' (A:ASH019-20)

This particle, along with its topic-switching function, may also be seen as a device for signalling natural continuity, 'and then… and then', itself having a conjunctive function, connecting one piece of discourse with the next, which is obvious when looking at example (100).

(100) *tarkaáṇ* carpenter *teeṇíi* refl *the* do.cv *bheénš* main.beam *ɡal-íi* put.in-3sg *rhalá* on.top *bheenš-á* main.beam-obl *ǰhulí* on *ba* top *čauráts* cross.beam *ɡal-íi* put.in-3sg *čaurats-í* cross.beam-obl *ǰhulí* on *ba* top *bhít-a* plank-pl *ɡal-íi* put.in-3sg

'The carpenter himself puts up the main beam, and then on top of the main beam he puts in the cross-beams, and then on the cross-beams he puts in planks.' (A:HOW016-7)

If not explicitly contrasted with any particular or easily definable entity in the preceding utterances, the reading is rather one of special emphasis put on the phrase thus marked by *ba* in (101)–(104), sometimes corresponding to English

### 15 Sentence modification

'as for', other times corresponding to what would be intonationally signalled as somehow outstanding. Example (105) may be described as a cleft construction, where *ba* marks focus that precedes the background.


This can be used also in questioning about the general whereabouts of a particular person, as in (106).

(106) *o* o *méeš* aunt.voc *kaṭamúš* Katamosh *ba* top 'O auntie, what about Katamosh?' (A:KAT112)

Sometimes, it is difficult to see exactly what *ba* does other than signal a switch in referentiality. That can be seen in how the entity pronominally referred to by the first *tasíi* 'his' is not the same as that referred to by the second *tasíi* 'his' in example (107).

(107) *tasíi* 3sg.gen *áak* idef *putr* son *de* be.pst *ɣaazisamadxaán* Ghazi.Samad.Khan *ba* top *tasíi* 3sg.gen *nóo* name *de* be.pst 'He had a son whose name was Ghazi Samad Khan.' (A:GHA004)

A topic-marked entity, as in (108) and (109), can also be further expanded in an extraposed *ki*-construction.

15.4 Switch-topicality


see-vn to go-prs-mpl

'The custom of bride-inspection, then, is the following: The relatives, men and women from the boy's house, go there, to see their daughter-in law.' (A:MAR104)

Apart from the cross-referencing between a topic marked with *ba* and the content of a *ki*-clause, it seems that the marking of a non-subject entity with *ba*, allows for one of the other arguments to be extraposed to a postverbal position, as is the case in (110) and (111). Although (110) corresponds to a passive construction in English, it is not passive in Palula, which is seen in the otherwise regular ergative case marking of the extraposed agent subject. The exact information status of the extraposed argument is a matter for further research.


Switch-topic marked entities are by no means confined to participants in the clause or even noun phrases. Almost any word or phrase can be highlighted and brought to the foreground by *ba*: a noun modifier, as in (112) and (115), an adverbial (113) or a locative expression (114).

### 15 Sentence modification


'There are two teachers, and *forty* children in our school.' (A:OUR011)

This includes clauses in complex constructions. Same-subject clauses, as in (116)–(117), as well as different-subject clauses with adverbial functions, as in (118)–(119), can be marked with *ba* (see §14.4).


(119) *phedóol-ii* arrive.with.pptc-gen *pahúrta* after *ba* top *hukumát* government *xabaár* informed *bhíl-u* become.pfv-msg

> 'After that [they had brought it there], the government found out about it.' (A:GHA08)

<sup>1</sup> It is not entirely clear why the verb shows singular agreement here even though the direct object is in the plural.

## **Sample text: Ashret dialect**

### **Across the Lowari Pass**

This is an account given by Muhammad Hussain, son of Noor Majid, recorded on 23 July 2003 by Naseem Haider – instructed by the author –, in the former's home in Čhíni in the upper Ashret Valley. Muhammad Hussain, father of four sons and three daughters, and the late elder of the Kučurooṭeé clan, was born in 1920 and passed away in 2007. He served as a postman during the time of the *ṣóo*<sup>1</sup> (the ruler of Chitral), carrying letters across the Lowari Pass between Chitral and Dir. Muhammad Hussain was not formally educated but spoke Pashto and Khowar. The narrative is a recollection of a dramatic event taking place at the Lowari Pass. While carrying the prime minister's wife in a sealed carriage through the pass, Muhammad Hussain and his comrades were swept away by a sudden avalanche but managed to get out of the snow and come to the rescue of the prominent lady. The annotated text was previously published in Liljegren & Haider (2015b: 131–137). The version occurring here has been revised. While that version was also supplied with an Arabic-based orthographic representation, that has been excluded here.


tus 2pl.nom kháṇ-a mountain-obl the to be-ooi go-imp.pl and that we should therefore head to the Lowari Pass.

<sup>1</sup> *Mehtar* in Khowar.


(15) *dharí* dhar-í remaincv *dúi* dúi another *hiimaál* hiimaál glacier *whaí* whéi-í come.down-cv *ṭópa* ṭópa down *traác̣* traác̣ host *de* de-í give-cv *asaám* asaám 1pl.acc *rhalá* rhalá up *ɡaḍíla* ɡaḍé-íl-a pull.out-pfv-mpl Then another avalanche came and brought us out.

429


## **Sample text: Biori dialect**

### **A Shepherd and a Leopard**

This narrative by Haji Muhammad Atiqullah, son of Wazir, was recorded by the author in Drosh on 29 June 2000. Atiqullah, whose home is situated in Dhamareét, the middle village of Biori, belongs to the Mulaakoór lineage. He is the most learned person of Biori Valley, working for many years as a school teacher at Drosh Higher Secondary school and later serving as headmaster of the middle school in his home village. Atiqullah is a poet and language promoter, and he takes a deep interest in all matters related to language and education. He is a fluent speaker of Pashto, Khowar, Urdu and English, and is also knowledgeable in Arabic and Persian. The present narrative is the account of an unnamed shepherd who lived long ago in Dhamareét. It is about the shepherd's encounter with a *karáaṛu*, a snow leopard, while taking his flock to graze on the slopes high above Biori Valley. The annotated text was previously published in Liljegren & Haider (2015b: 182–187). The version which appears below has been revised.

(1) *muxáak* muxáak before *zamanéee* zamaná-e time-gen *asée* asée 1pl.gen *díiše* déeš-e village-gen *ak* ak idef *bakaraál* bakaraál shepherd *de* de pst *so* so 3msg.nom *bakáara* bakáara flock *ɡhiní* ɡhin-í take-cv *saaraá* saaraá wilderness *the* the to *ɡáu* ɡáu go.pfv.msg Once a long time ago in our village, there was a shepherd. He took his herd and went into the wilderness.

(2) *bakáara* bakáara herd *praší* praš-í slope-obl *phará* phará along *čáaren* čar-en graze-3pl *de* de pst *so* so 3msg.nom *ba* ba top *ṭéka* ṭéku-a hilltop-obl *bheší* bheš-í sit.down-cv *ba* ba top *tenaám* tenaám 3pl.acc *the* the to *dachíi ̣* daché-e ̣ look-3sg *de* de pst

The goats were grazing on the slope. He was sitting on the mountain watching the goats.


He was surprised, and wondered why the goat was standing on two legs.

(5) *šuy* šuy good *tíiṇu* tíiṇ-u sharp-msg *bhe* bhe-í become-cv *dachíi ̣* daché-e ̣ look-3sg *ta* ta sub *dhayíi* dhayé-e note-3sg *ta* ta sub *karáaṛu* karáaṛu leopard *tesée* tesée 3sg.gen *teeṇíi* teeṇíi refl *čútim* čúti-m paw-pl *baándi* baándi by *tesée* tesée 3sg.gen *tes* tes 3sg.acc *húṇṭraak* húṇṭraak upward *raál* raál high *the* the-í do-cv *ba* ba top *tesée* tesée 3sg.gen *rat* rat blood *čuuṣáanu* čuuṣé-án-u suck-prs-msg

When he looked carefully, he saw a leopard that, with the help of his paws, had lifted up the goat and was sucking its blood.


*púustu* púustu skin *ɡhaḍí* ɡhaḍé-í take.off-cv *so* so def.msg.nom *mhaás* mhaás meat *púusta* púustu-a skin-obl *wée* wée into *čhúuṇu* čhúuṇ-u put.down.pfv-msg

He came down, slaughtered the goat, skinned it, and put the meat inside the skin.

(8) *čhoorí* čhooré-í put-cv *ba* ba top *so* so def.msg.nom *púustu* púustu skin *harí* har-í take.away-cv *ba* ba top *muṭá* muṭ-a tree-obl *ak* ak idef *muṭá* muṭ-a tree-obl *wée* wée in *trúu* tráa three *šúunɡam* šáanɡ-am branch-pl.obl *maǰí* maǰí between *so* so def.msg.nom *mhaás* mhaás meat *čhúuṇu* čhúuṇ-u put.down.pfv-msg *čhoorí* čhooré-í put-cv *ba* ba top *teeṇíi* teeṇíi refl *bakáara* bakáara flock *ɡhiní* ɡhin-í take-cv *ɡáu* ɡáu go.pfv.msg *húṇṭraak* húṇṭraak upwards

He carried the skin and placed it between three branches in a tree. After that, he took his goats uphill.


When he came near and looked, he saw the leopard sitting under the tree, picking up stones with its paws.

(13) *uchí ̣* uchí-í ̣ lift-cv *ba* ba top *se* se def *čútime* čúti-m-e paw-pl-gen *zaríia* zaríia means *baándi* baándi by *so* so def.msg.nom *baṭ* baṭ stone *húṇṭraak* húṇṭraak upward *chuɡaláanu ̣* chuɡalé-án-u ̣ hurl-prs-msg *se* se def *kúuka* káak-a crow-pl *se* se def *muṭá* muṭ-a tree-obl *bhešaníi* bheš-anií sit.down-vn *the* the to *na* na neg *uṛáanu*

uṛí-án-u

let.go-prs-msg

After picking them up, it was throwing the stones upward with the help of his paws, not allowing the crows to sit in the tree,

(14) *mhaás* mhaás meat *khainíi* kha-anií eat-vn *the* the to *na* na neg *uṛáanu* uṛí-án-u let.go-prs-msg *so* so def.msg.nom *bakaraál* bakaraál shepherd *hairán* hairán surprised *bhílu* bhe-íl-u become-pfv-msg *ki* ki comp *dhayé* dhayé-í note-cv *aṛáa* aṛáa there.dist *ṣaá* ṣaá out.of

not allowing them to eat the meat. The shepherd was surprised when he saw this.

(15) *típa* típa now *anú* anú 3msg.prox.nom *hatí* hatí such *uxiaár* uxiaár clever *haiwán* haiwán creature *ki* ki comp *haṛé* haṛé dist *kúuka* káak-a crow-pl *ba* ba top *aṛé* aṛé dist *kháan* kha-en eat-3pl *thaní* thaní qot *ba* ba top *mhaás* mhaás meat *kháan* kha-en eat-3pl *thaní* thaní qot *ba* ba top *típa* típa now *índe* índe here *yooí* yooí from(?) *wáal* wáal on.guard *páta* páta surely *bhe* bhe-í become-cv *hínu* hin-u be.prs-msg

"Now, this is a clever animal, guarding the tree so that the crows cannot eat the meat."

(16) *bhaarúuṛi* bhaarúuṛi scream *ǰeníie* ǰe-anií-e beat-vn-gen *so* so def.msg.nom *karáaṛu* karáaṛu leopard *ɡáu* ɡáu go.pfv.msg *tíi* tíi 3sg.obl *ba* ba top *bhun* bhun down *whayí* whéi-í come.down-cv *ba* ba top *so* so def.msg.nom *mhaás* mhaás meat *muṭí* muṭ-í tree-gen *bhun* bhun below *wheelí* whaalé-í take.down-cv *ba* ba top *teeṇíi* teeṇíi refl *ɡhooṣṭá* ɡhoóṣṭ-a house-obl *the* the to *ɡhiní* ɡhin-í take-cv *ɡáu* ɡáu go.prv.msg

When he shouted, the leopard left. He came down and took down the meat from the tree and went to his house.

## **References**


### References


*of Language Structures Online*. Leipzig: Max Planck Institute for Evolutionary Anthropology.


### References


Pakistan 1), 65–80, 193–205. Islamabad: National Institute of Pakistani Studies & Summer Institute of Linguistics.


Elfenbein, Josef. 1997. Pashto phonology. In Alan S. Kaye (ed.), *Phonologies of Asia and Africa (including the Caucasus)*, 733–760. Winona Lake, Ind.: Eisenbrauns.

Emeneau, Murray Barnson. 1965. India and historical grammar. Annamalainagar.


### References


### References


ley (eds.), *On diversity and complexity of languages spoken in Europe and North and Central Asia* (Studies in Language Companion Series 164), 133–174. Amsterdam: John Benjamins Publishing Company.


as a linguistic area). In Peri Bhaskararao & Karumuri Venkata Subbarao (eds.), *The yearbook of South Asian languages and linguistics 2001*, 205–267. New Delhi: Sage Publications.

Matthews, Peter H. 1991. *Morphology*. Cambridge: Cambridge University Press.


*International Hindu Kush Cultural Conference*, 250–262. Karachi: Oxford University Press.


## **Name index**

Abbi, Anvita, 321 Akhunzada, Fakhruddin, 9, 25 Allerton, David J., 316 Andrews, Avery D., 396, 420 Ansaldo, Umberto, 244 Baart, Joan L. G., 1, 17, 20, 22, 72, 109, 162, 174, 185, 236, 299, 300, 304, 305, 318, 323 Babrakzai, Farooq, 300 Bailey, Thomas Grahame, 103, 299 Bashir, Elena, 1, 13, 15, 22, 161, 245, 267, 300, 321 Berlin, Brent, 151, 152 Bhat, D. N. Shankara, 70, 125 Bhatia, Tej K., 300 Bickel, Balthasar, 45, 291, 321 Biddulph, John, 1, 3 Blake, Barry J., 291 Buddruss, Georg, 2, 15, 18, 30, 39, 101, 115, 228, 229, 254 Butt, Miriam, 244, 327 Bybee, Joan, 251, 264 Cacopardo, Alberto M., 1, 3, 4, 9–13, 18 Cacopardo, Augusto S., 1, 3, 4, 9–13, 18 Cardona, George, 53 Comrie, Bernard, 291 Cooper, Gregory R., 23

Corbett, Greville G., 46, 99, 101, 102, 148, 157 Cristofaro, Sonia, 352, 357, 361, 365 Croft, William, 83, 291 Dahl, Östen, 21, 99, 100, 250, 251, 253, 257, 258, 260, 261, 263 David, Anne Boyle, 24 Decker, Kendall D., 2–4, 7, 9, 18, 22– 27, 30 Decker, Sandra J., 19 Degener, Almuth, 22, 24, 39 Deo, Ashwini, 299 Diessel, Holger, 130, 133, 134, 137, 138, 167 Dixon, Robert M. W., 83, 149, 150, 152–155, 291 Downing, Bruce T., 389 Ebert, Karin, 21, 208, 244 Èdel'man, Džoi Iosifovna, 2, 22, 299 Elfenbein, Josef, 24 Emeneau, Murray Barnson, 21, 22 Endresen, Rolf Theil, 69, 124 Fennig, Charles D., 13, 24 Folli, Raffaella, 207 Fussman, Gérard, 22 Gambhir, Vijay, 207, 334 Givón, T., 83, 133, 147, 149, 153–155, 204, 305, 312, 314, 339, 351,

### Name index

359, 361–363, 365, 370, 373, 375, 387, 390 Haider, Naseem, 31, 32, 37, 83, 155, 216, 427, 433 Hallberg, Calinda E., 70, 299 Hallberg, Daniel G., 70, 299 Harley, Heidi, 207 Haspelmath, Martin, 128, 233, 268, 329, 344, 349, 352 Heegård, Jan, 1, 22, 23 , 70 Himmelmann, Nikolaus, 129, 130, 133–135, 139, 141, 142 Hock, Hans Henrich, 321 Hook, Peter Edwin, 39, 103, 244, 287, 299, 300, 321 Hwang, Shin Ja J., 350, 354, 356, 359, 361–363, 365, 367 Iggesen, Oliver A., 45 Jettmar, Karl, 12 Joshi, Dayashankar M., 287 Juvonen, Päivi, 134, 141, 142 Jäger, Andreas, 182, 327 König, Ekkehard, 403 Karimi, Simin, 207 Kay, Paul, 151, 152 Keenan, Edward L., 396 Kibrik, Andrej, 129 Klaiman, Miriam H., 299 Kohistani, Razwal, 1, 2, 39, 55, 71, 126, 134, 136, 158, 182, 186, 188, 199, 224, 299, 345 Koptjevskaja-Tamm, Maria, 21 Koul, Omkar N., 299, 300 Kristiansen, Knut, 69, 124 Lehmann, Winfred Philipp, 70

Lewis, M. Paul, 13, 24 Liljegren, Henrik, 19, 31, 32, 83, 155, 216, 244, 254, 299, 301, 427, 433 Liljegren, Maarit, 9, 25 Longacre, Robert E., 350, 354, 356, 359, 361–363, 365, 367 Lorenz, Manfred, 300 Losey, Wayne E., 70 Lunsford, Wayne A., 72, 299 Luraghi, Silvia, 53 Mørch, Ida Elisabeth, 22, 70 Maddieson, Ian, 43 Masica, Colin P., 2, 9, 14, 21, 39, 97, 98, 101, 106, 111, 112, 115, 119, 120, 204, 207, 208, 223–225, 228, 229, 232, 234–236, 240, 241, 244, 247, 251, 254, 264, 268, 291, 299, 300, 303, 305, 326, 331, 334, 389 Matthews, Peter H., 107 McMahon, April M. S., 234 Merrifield, William, 151, 152 Mock, John, 13 Mohanan, Tara, 326, 329 Morgenstierne, Georg, 1, 2, 6, 9, 15, 18, 22–24, 30, 31, 55, 61, 71, 101, 115, 160, 211, 213, 216, 229, 265, 300 Nichols, Johanna, 45 Ningomba, M. S., 70 Noonan, Michael, 316, 321, 375, 379, 387, 389 Pagliuca, William, 251, 264 Payne, John R., 300, 340 Perder, Emil, 1, 22, 24, 184

### Name index

Perkins, Revere, 251, 264 Perry, John R., 124 Pischel, Richard, 113 Pustet, Regina, 147, 305–307 Radloff, Carla F., 20, 39, 60, 103, 219, 236, 240 Ramat, Paolo, 414 Raverty, Henry G., 151 Rehman, Khawaja A., 17, 20, 37 Rensch, Calvin R., 19 Robson, Barbara, 107 Saeed, Ahmad, 11 Sagar, Muhammad Zaman, 37 Saxena, Anju, 133 Schachter, Paul, 83 Schmidt, Ruth Laila, 1, 2, 16, 17, 39, 55, 71, 107, 126, 127, 134, 136, 158, 182, 186, 188, 199, 219, 224, 228, 231, 236, 244, 255, 268, 270, 299, 300, 345, 370 Shackle, Christopher, 70, 236 Shakil, Shakil Ahmad, 219, 236, 240 Sharma, Devyani, 299 Shopen, Timothy, 83 Siemund, Peter, 403 Siiger, Halfdan, 23 Simons, Gary F., 13, 24 Skalmowski, Wojciech, 22, 299 Strand, Richard F., 2, 9, 11, 13, 15, 19, 22–24 Stroński, Krzysztof, 299 Szemerényi, Oswald J. L., 70, 73 Tegey, Habibullah, 107, 300 Thompson, Sandra A., 350, 354, 356, 359, 361–363, 365, 367 Tikkanen, Bertil, 22

Toporov, Vladimir Nikolayevich, 22 Trail, Ronald L., 23 Trask, Robert Lawrence, 19 Turner, Ralph Lilley, 3, 71, 112, 115, 151, 152, 335 Verbeke, Saartje, 299 Verma, Manindra K., 326, 327, 329, 331 Viberg, Åke, 203, 204, 206, 208 Whitney, William Dwight, 112, 113, 115, 119, 147, 234 Wierzbicka, Anna, 151 Windfuhr, Gernot, 124 Wälchli, Bernhard, 21 Zia, Muhammad Amin, 39, 103, 321 Zoller, Claus Peter, 14, 19, 54

## **Language index**

Arabic, 27, 28, 36, 39, 54, 158, 256, 263, 275, 328, 334 Balti, 21 Bashkarik, *see* Gawri Bateri, 16 Brokskat, 16 Burushaski, 21, 22, 299 Chilisso, 15 Chitral, 15, 23 Dameli, 15, 22–24, 28, 36, 184, 187, 299, 300, 396 Dardic, 13, 14, 21, 22, 30 Dravidian, 21, 55, 321 English, 10, 21, 27, 31, 33, 36, 37, 40, 54, 147, 151, 175, 182, 203, 243, 313, 329, 340, 352, 354, 372, 420, 421, 423 Gandhari, 14 Gawarbati, 15, 299, 300 Gawri, 11, 15, 19, 20, 72, 109, 162, 174, 185, 236, 300, 304, 318, 323 Gojri, 70 Gowro, 15 Grangali, 15 Gujarati, 287 Indo-Aryan, 1, 13, 21, 22, 30, 31, 241, 299, 300

Hindukush Indo-Aryan, 14, 15, 17, 55, 72 Middle Indo-Aryan, 14, 113, 118, 151, 236, 240, 244 New Indo-Aryan, 14, 55, 70, 97, 98, 106, 111, 112, 115, 119, 120, 151, 229, 234, 244, 254, 299, 304 Old Indo-Aryan, 17, 53, 70, 71, 98, 101, 103, 112–115, 118– 120, 122, 124, 147, 151, 152, 160, 222, 229, 234, 236, 240 Kalasha, 4, 11, 13, 15, 22, 23, 25, 28, 29, 31, 70, 145, 245, 267, 300, 321 Kalkoti, 12, 16, 17, 19, 20, 174, 219, 228, 229, 234, 254 Kamviri, 6, 24, 25 Kashmiri, 16, 300 Katarqalai, 15 Kati, 24 Khowar, 1, 3, 4, 9–11, 15, 22–29, 31, 69, 124, 184, 234, 267, 300, 304 Kohistani, 15 Dir Kohistani, *see* Gawri Indus Kohistani, 15, 70 Kalam Kohistani, *see* Gawri Kunar, 15 Kundal Shahi, 17, 20 Nuristani, 6, 10, 11, 18, 21, 23, 24, 31,

Language index

35 Pamir, 21 Pashai, 15 Pashto, 4, 7, 23–29, 32, 33, 35, 37, 82, 105, 107, 119, 122, 124, 127, 150, 162, 180, 190, 207, 243, 300, 329, 333, 334, 342, 362, 364, 415 Pech, 15 Pech-Kunar, 15 Persian, 10, 22, 28, 124, 207, 299, 334, 415 Prakrit, 14, 234 Punjabi, 244, 300, 349 Sauji, *see* Sawi Sawi, 17–20, 219, 228, 229, 234, 254 Shekhani, *see* Kamviri Shekhwar, *see* Kamviri Shina, 1, 3, 11, 12, 16, 17, 19–21, 39, 60, 65, 71, 102, 103, 113, 126, 134, 136, 158, 182, 186, 188, 199, 219, 224, 228, 229, 231, 234, 236, 240, 244, 254, 268, 299, 321, 345, 349, 350, 389 Shumashti, 15 Sinhalese, 300, 304 Siraiki, 70, 236 Tibetan, 299, 321 Tirahi, 15 Torwali, 16, 72, 299 Turkic, 21 Urdu, *see* Urdu-Hindi Urdu-Hindi, 25–28, 33, 36, 37, 54, 107, 118, 122, 127, 152, 162,

163, 207, 225, 236, 243, 244,

264, 270, 288, 299, 300, 304, 324, 329, 334, 342, 370, 390 Ushojo, 16, 17, 19, 20

accent, 28, 39, 40, 44, 61, 62, 71–78, 80, 86, 112–115, 122, 123, 157, 160–162, 208, 209, 216, 223, 226, 229, 232–234, 256, 278, 327, 359 accompanying circumstance, 358, 359 accusative, 45, 49, 113, 125, 144, 145, 186–191, 193, 195, 223, 250, 255, 265, 292, 296, 297, 299, 300, 311, 313, 322, 405, 406 accusative agreement, 259, 292 additive, 373 adjectival agreement, 225, 282, 284, 287, 297 adjective, 47–49, 86–88, 147–150, 155–161, 165, 171, 181, 184, 207, 225, 241, 244, 264, 265, 274, 279–282, 285–288, 303, 306–308, 327, 333, 343, 407 adverb, 88, 92, 155, 165, 170–176, 184, 187, 192, 195–197, 334, 409 affricate, 40, 43, 54, 58, 67 alignment, 49, 50, 250, 255, 291–294, 297, 300, 301, 311 anaphoric, 90, 132–134, 136, 137, 142, 396, 399 apposition, 273, 278, 279 approximant, 44, 60, 63, 64, 66 argument structure, 207, 244, 303, 323–327, 329, 331, 334

aspiration, 43, 58, 59, 66, 69–72, 78, 114, 118, 214, 218, 236 auxiliary, 204, 219, 221, 222, 226, 229, 230, 234, 255, 256, 264, 266, 267, 305, 411

causality, 270, 363–365


conditionality, 221, 254, 264, 365, 367–371 conjunct verb, 181, 241, 243, 321, 326, 327, 329 conjunction, 93, 173, 195, 221, 340, 342–344, 349, 350, 354, 356, 361, 366–368, 371, 373 conjunctive participle, 219, 232, 268, 354 consonant cluster, 68, 148, 216 Converb, 48, 51, 182, 223, 232, 245, 268, 269, 271, 352, 354, 355, 357–359, 364, 369, 393, 394 coordination, 162, 186, 339, 340, 344, 345, 349, 351, 352 Copredicative Participle, 233, 268, 355, 359, 385 copula, 51, 84, 85, 87, 204, 219, 228, 254–257, 259–261, 286, 303–309, 344, 404, 412, 413 copular clause, 306 correlative clause, 396 counterfactuality, 365, 370 declension, 45, 51, 107, 109, 111–123, 157, 294 definite, 133, 134, 140–142, 281, 301, 308, 309, 383, 396 degree adverb, 165 deictic, 22, 46, 90, 128, 129, 132–134, 138, 139, 169–172, 175–178, 181, 183, 390 demonstrative, 84, 89, 91, 128–130, 134, 137, 139–141, 143, 175, 187, 390, 391, 393, 396, 399 derivation, 83, 85, 97, 99, 115, 161, 180, 181, 218, 235, 244, 277 derivational, 1, 47, 81, 101, 115, 119, 160, 182, 235, 240–242

determiner, 91, 141, 277, 279, 280, 283, 365, 369, 392, 407 determiner agreement, 282, 297 diphthong, 214, 232 direct object, 49, 50, 86, 95, 107, 144, 201, 234, 239, 288, 313, 314, 316, 318, 319, 322–327, 329– 331, 353, 362, 392, 393, 398, 405, 406, 425 discourse, 31, 84, 90, 91, 93, 130, 132– 140, 142, 167, 181, 234, 259, 266, 304, 306, 339, 340, 342, 345, 350, 354, 367, 374, 377– 379, 381, 389, 394, 400, 410, 419, 421 disjunction, 347–349 distal, 90, 91, 128, 132–134, 136–139, 142, 167 ditransitive, 186, 239, 317, 318 ergative, 45, 46, 49, 107, 110, 125, 167, 223, 256, 259, 292, 294, 296– 300, 311, 313, 323, 382, 392, 405, 423 ergative alignment, 189, 250, 255, 292, 297, 311, 327, 352 existential, 165, 307–310, 418 exophoric, 134, 136 finite, 29, 47, 49, 85, 88, 97, 144, 148, 201, 204, 206, 222, 223, 230, 232–235, 244, 245, 247, 250, 254, 255, 264–268, 271, 318, 323, 325, 351–357, 361–363, 365, 367, 371, 374, 380, 389, 390, 393, 394, 398, 412 fricative, 43, 44, 53, 54, 58, 67, 68 Future, 47–50, 85, 208, 210–214, 217– 220, 222, 223, 226, 229, 247,

248, 250–252, 255, 258, 263, 292, 368 gender, 22, 46, 47, 49, 63, 84, 87, 88, 90, 91, 95–103, 107, 110, 111, 119–121, 125, 128, 130, 132, 140, 148, 157, 218, 223, 225, 228–230, 234, 256, 270, 283–288, 292, 293, 306, 308, 331, 353, 382, 384 gender assignment, 46, 99, 102, 114 gender/number agreement, 85, 222, 223, 228, 250, 255 genitive, 11, 45, 46, 62, 78, 84, 96, 107– 109, 113, 125–127, 130, 143– 145, 148, 174–176, 185–188, 193, 195, 196, 198, 265, 268, 270, 274, 275, 277, 280–282, 296, 298, 310, 318, 319, 322, 323, 357, 362, 397 grammatical relations, 106, 291, 293 Hearsay, 92, 266, 267 host element, 171, 242, 244, 303, 328, 329, 332, 333, 335, 337, 382, 413 Imperative, 47, 48, 50, 208, 210–214, 217–220, 223, 226, 228, 249, 263, 368, 387, 419 Imperfective, 47–50, 85, 208, 210– 214, 217–220, 222, 223, 226, 228, 229, 247, 248, 253, 255, 258, 292, 307, 369 imperfective, 85, 86, 126, 208, 209, 214–216, 219, 222, 223, 226, 228–230, 235, 247, 254, 255, 259, 268, 269, 271, 292, 299, 313, 320, 323, 365, 368

indefinite, 52, 89–91, 93, 128, 134, 140–142, 145, 160, 167, 168, 176, 181, 281, 309, 391, 394, 395, 405, 408, 409, 414, 415, 418 indirect object, 311, 312, 315–320, 322, 324, 325, 336 Infinitive, 48, 223, 231, 233, 271, 323, 374, 384, 385 inflecting adjective, 288 inflection, 62, 96, 106, 109, 112, 157, 160, 177, 185, 208, 209, 222, 230, 270, 406 inflectional, 29, 47, 59, 85, 86, 97, 103, 106, 111, 116, 120, 123, 157, 158, 185, 186, 208, 209, 219, 222, 235, 240, 243, 247, 254, 267, 285, 286, 293 intensifier, 184 interrogative, 52, 84, 89, 90, 93, 128, 129, 145, 167, 168, 176, 181, 391, 395, 403, 405, 407–410, 414, 415, 418 Intransitive, 312, 320, 322 intransitive, 48, 50, 85, 107, 126, 168, 181, 189, 209–212, 214, 216, 218, 223, 234, 235, 237, 238, 240, 241, 244, 259, 265, 291, 292, 300, 306, 310–312, 314– 317, 320, 322, 324, 325, 331– 333, 335, 352, 362, 382, 384, 388, 405, 409, 418 invariant adjective, 159 juxtaposition, 51, 340, 341, 343, 347, 374 L-verb, 208, 216, 217, 221 lateral approximant, 60

423

noun declensions, 51, 96, 111, 158

locative, 45, 75, 110, 168, 169, 187, 192, 193, 195, 303, 308, 309, 315– 318, 320, 331, 423 manipulation verb, 324 manner adverb, 171, 181, 271 modality verb, 323, 324, 412 mood, 47, 84, 85, 251, 254 mood marker, 92, 354 nasal, 44, 54, 59, 61, 66, 67, 72, 225 nasalisation, 43, 61, 67, 72 negation, 263, 327, 346, 347, 349, 350, 403, 404, 411, 414–419 nominalisation, 270, 339, 397 nominative, 45–47, 70, 78, 84, 90, 91, 96, 101, 105–108, 110, 112– 116, 121, 122, 125–127, 130, 132, 133, 140, 142, 145, 157, 185, 187, 190, 265, 283, 284, 293, 294, 296–298, 300, 311– 313, 315, 316, 320–323, 331, 335, 357, 369, 382, 392, 405 nonfinite, 30, 180 noun, 45, 46, 48, 49, 51, 68, 74, 75, 78, 84, 86–89, 91–93, 95– 100, 102–110, 113, 115, 117, 123, 130, 140, 141, 147, 148, 150, 152, 153, 156–158, 161, 162, 164, 165, 169, 171, 172, 174–177, 183, 185, 188, 192, 198, 207, 223, 225, 232–234, 241, 244, 264, 265, 269, 270, 273–282, 284–288, 291, 293, 296, 297, 303, 306, 307, 309, 312, 313, 315, 316, 320, 323, 327–329, 334, 335, 340–344, 348, 364, 372, 378, 390, 392, 394, 397, 406, 407, 412, 420,

numeral, 21, 105, 140, 142, 161–164, 276, 348, 406 Obligative, 48, 223, 231, 264, 265, 418 oblique, 45, 46, 62, 70, 75, 78, 84, 92, 96, 103, 105–112, 114, 117– 119, 121–123, 125, 126, 141, 145, 157, 159, 160, 167–169, 174, 177–180, 185–195, 198, 201, 244, 268, 270, 293, 294, 298, 300, 313, 322, 331, 365, 372, 393, 395, 405 ordinal numeral, 406 participial, 218, 225, 229, 389, 398 passive, 33, 75, 86, 234, 240, 264, 265, 310, 332, 333, 423 PCU, 325, 326, 337, 375, 381 Perfect, 29, 47–49, 85, 222, 234, 249, 253, 255, 257, 259, 260, 266 perfective, 29, 45, 47–51, 75, 81, 85, 86, 107, 108, 125, 126, 129, 154, 156, 208–210, 212–216, 219, 221–223, 225, 226, 230, 234, 235, 240, 247, 254, 256, 259, 260, 263, 268, 291–294, 296–299, 301, 311, 313, 320, 323, 325–327, 329–331, 335, 353, 361, 365–367, 369, 382, 384, 398 Perfective Participle, 154, 156, 165, 234, 268, 269, 355, 357 person agreement, 85, 222–226, 228 phrase, 89, 92, 94–97, 106, 148, 156– 158, 165, 181, 261, 273–282, 285, 286, 291, 293, 296, 297, 303, 304, 306–309, 312, 313,

315, 316, 323, 343, 372, 373, 378, 392, 397, 405, 406, 408, 415, 417, 420, 421, 423 pitch, 71, 72, 88, 184 pitch accent, 17, 44, 73, 74, 185 plosive, 40, 43, 44, 53, 54, 66–68, 86, 214, 218 Pluperfect, 47–49, 85, 234, 249, 253, 255, 260–262, 308 postposition, 48, 92, 107, 108, 110, 126, 127, 129, 160, 167–169, 171– 174, 176, 185–199, 311, 322, 330, 336, 355, 357, 359, 361, 362, 364, 373, 383, 386, 406 postpositional phrase, 315, 316, 331 postpositional sequence, 193, 194, 197, 198 postsection, 346, 417 precedence, 195, 361 predicate, 51, 52, 84–87, 95, 97, 181, 201, 207, 221, 269, 282, 285, 291, 303, 306, 308, 309, 327, 329, 343, 372–376, 380, 381, 384, 385, 408, 411, 415, 416 Present, 47–51, 85, 208, 210–214, 217–223, 229, 230, 247, 248, 250–252, 254, 255, 263, 292, 304, 368 prohibitive negation, 419 pronominal, 50, 84, 125, 127, 129, 134, 138, 145, 167, 169, 186–189, 191–195, 293, 296, 374, 376 pronoun, 46, 52, 90, 91, 126–128, 132, 133, 137, 138, 141, 143, 144, 160, 171, 185, 187, 189–191, 263, 273, 279, 296, 381, 391, 395, 399, 405, 414, 418 proper noun, 102

proximate, 205 purpose, 36, 39, 187, 196, 277, 362, 371, 385, 406 quantifier, 91, 92, 165, 274, 275, 280– 282, 406 quotative, 134, 216 reciprocal pronoun, 90, 144 reduplication, 94, 183 reflexive pronoun, 90, 144, 394 rejection, 349, 350, 417 relative clause, 356, 389–391, 396 relative pronoun, 390 remote, 4, 90, 91, 128, 130, 132–137, 139–142, 167, 253, 260, 390, 391, 393 retroflex, 17, 21, 22, 37, 43, 53–55, 58, 59 sex, 46, 99, 100 Simple Past, 47, 49, 85, 234, 235, 247, 248, 250–255, 257, 260, 261, 263, 268, 269, 368, 371 simultaneity, 189, 340, 361, 363 spatial adverb, 173, 174, 194, 195, 197 subject, 30, 31, 45, 49–51, 67, 68, 84, 90, 95, 107, 115, 129, 143, 201, 210, 212, 223, 226, 230, 234, 238, 244, 270, 282, 285, 286, 288, 291, 292, 300, 308, 309, 312–316, 319, 320, 322–325, 330, 331, 335, 340, 341, 343, 351–353, 355, 356, 358, 359, 362–364, 372, 376–380, 382, 384, 386, 392, 393, 396, 398, 399, 405, 420, 423, 424 subsequence, 355, 357 substitutive, 372

syllable, 29, 44, 59, 62, 64, 66, 68–70, 74, 81, 96, 115, 149, 183, 184, 201, 212, 218, 226, 234, 278 syllable structure, 28, 43, 78 T-verb, 208, 218 temporal adverb, 168, 197 tense, 40, 47, 49, 51, 64, 85, 92, 204, 211–213, 216, 219, 220, 222, 223, 225, 226, 228–230, 234, 247, 251, 252, 254–258, 260, 261, 263, 299, 300, 303, 305– 308, 398, 413 topicality, 403, 419, 420 transitive, 45, 48–50, 85, 86, 107, 108, 125, 126, 129, 168, 181, 186, 201, 209–211, 214, 216, 218, 223, 234, 235, 237–241, 244, 259, 265, 291, 292, 299, 311, 313–316, 318–320, 322–325, 327, 329, 330, 332, 333, 336, 337, 352, 357, 382, 384, 388, 393, 396, 409, 418 umlaut, 81, 87, 117, 118, 158, 210, 225, 226, 229, 230, 234 utterance predicate, 377 valency, 206, 208, 222, 235–237, 240, 241, 244, 310, 311, 318, 324, 325, 329, 331, 332, 337 valency addition, 235, 238, 240 verb agreement, 35, 291, 292, 297, 299, 300, 311, 382 verb class, 209, 238 verbal agreement, 49, 323 Verbal Noun, 97, 189–191, 196, 233, 269, 270, 273, 275, 323, 324, 340, 355, 359–364, 372–374, 379, 382–386, 397, 398, 416

verbaliser, 181, 216, 241, 242, 244, 323, 326, 327, 329, 330, 332–335, 337, 409 vocative, 107

# 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 Palula

This grammar provides a grammatical description of Palula, an Indo-Aryan language of the Shina group. The language is spoken by about 10,000 people in the Chitral district in Pakistan's Khyber Pakhtunkhwa Province. This is the first extensive description of the formerly littledocumented Palula language, and is one of only a few in-depth studies available for languages in the extremely multilingual Hindukush-Karakoram region.

The grammar is based on original fieldwork data, collected over the course of about ten years, commencing in 1998. It is primarily in the form of recorded, mainly narrative, texts, but supplemented by targeted elicitation as well as notes of observed language use. All fieldwork was conducted in close collaboration with the Palula-speaking community, and a number of native speakers took active part in the process of data gathering, annotation and data management.

The main areas covered are phonology, morphology and syntax, illustrated with a large number of example items and utterances, but also a few selected lexical topics of some prominence have received a more detailed treatment as part of the morphosyntactic structure. Suggestions for further research that should be undertaken are given throughout the grammar.

The approach is theory-informed rather than theory-driven, but an underlying functional-typological framework is assumed. Diachronic development is taken into account, particularly in the area of morphology, and comparisons with other languages and references to areal phenomena are included insofar as they are motivated and available. The description also provides a brief introduction to the speaker community and their immediate environment.