# **Borrowed preverbs and the limits of contact-induced change in aspectual systems**

*Peter Arkadiev*

#### **1.** *Introduction***<sup>1</sup>**

In this article I address the issue of the degree of contact-induced influence on aspectual systems mediated by borrowing or calquing of morphological material, primarily verbal prefixes (preverbs). By surveying four cases of such influence with a Slavic or Baltic language as a donor and a non-Slavic language (Romani, Livonian, Yiddish and Istroromanian) as a recipient, I show that even in situations of intense language contact manifested by heavy borrowing of preverbs or their functions, the recipient languages fail to transfer from the donor language the grammatical category of aspect and end up with systems either considerably less grammaticalized than those of the donor languages or organized according to different principles, even if based on borrowed material.

The paper is organized as follows. In section 2 I provide a brief characterization of the domain of inquiry, i.e. "Slavic-style" aspect and contact-induced grammatical change, couched in terms of recent typologically-oriented approaches. In sections 3 and 4 I discuss the empirical data, starting with the case of (Eastern) Yiddish, which has reorganized its originally Germanic system of verbal prefixes on the basis of the Slavic model, then passing to languages which have borrowed Slavic or Baltic prefixes, and ending with the rather special case of Istroromanian. In section 5 I summarize my conclusions.

FUP Best Practice in Scholarly Publishing (DOI 10.36253/fup\_best\_practice)

<sup>1</sup> I am grateful to the participants of the conference *The role of prefixes in the formation of aspect and related categories. Problems of grammaticalization* (Padua, 28-29 September 2015) for the useful feedback on the presentation this article is based on, and to Rosanna Benacchio for her kind invitation to participate. I also thank Kirill Kožanov, Anton Tenser and Björn Wiemer, as well as two anonymous reviewers and the editors of the volume for their comments on the first version of the article. All faults and shortcomings remain mine. The work has been supported by the Russian Foundation for the Humanities, grants # 14-04-00580 and 17-04-00444.

Peter Arkadiev, *Borrowed preverbs and the limits of contact-induced change in aspectual systems*, pp. 1-21, © 2017 Author(s), CC BY 4.0 International, DOI 10.36253/978-88-6453-698-9.03

#### **2.**  *Theoretical preliminaries*

#### **2.1.** *Slavic-style aspect*

"Slavic-style" aspect is a term coined by Dahl (1985: 84-89) to cover the aspectual systems sharing the following characteristics, primarily associated with Slavic languages (hence the term), but also exhibited elsewhere:


Such aspectual systems are attested mainly in the vicinity of the Slavic languages, e.g. in Baltic (Lithuanian and Latvian), Hungarian, Ossetic, and Kartvelian (Georgian, Svan, Mingrelian, Laz), but are also found in other geographical areas, e.g. in Sino-Tibetan (Qiangic and rGyalrongic languages), in Austronesian (some Micronesian languages), in Chadic (e.g. Margi), Quechuan and Aymaran (see e.g. Breu 1992, Tomelleri 2008, 2009, 2010, Plungjan 2011, Arkadiev 2014, Arkad'ev 2015, Arkad'ev, Šluinskij 2015). In many of these languages, e.g. in Margi and Quechuan, perfectivization is expressed by suffixes rather than prefixes, although the latter are not limited to the geographical area close to Slavic, being attested e.g. in the Sino-Tibetan languages.

The aspectual categories of the Slavic languages have been considered highly grammaticalized (to the extent that in Russian aspectology there is a tradition to treat at least some of the aspectual oppositions as inflectional, see Gorbova 2014, 2015 for a recent discussion) by the following criteria (cf. Vimer 2001, Lehmann 2004, Wiemer, Seržant 2017):


sistent use of the imperfective with phasal verbs in all standard Slavic languages), and the lack of a common "semantic denominator" for all the uses of each of the aspects, as evidenced by the rather futile attempts at formulating the putative "semantic invariant" of the aspects as well as the frequent characterization of the Imperfective aspect as "unmarked" (cf. e.g. Fortuin, Kamphuis 2015).

However, even if highly grammaticalized, aspectual categories show considerable and principled variation across Slavic languages, as has been shown by a number of recent studies such as e.g. Stunová (1993), Petruchina (2000), Dickey (2000, 2005, 2008, 2015), Wiemer (2008), Vimer (2015), Fortuin, Kamphuis (2015), among many others. This variation reveals itself in such parameters as differences in the productivity of (secondary) imperfectivization (generally increasing from West to East, cf. Petruchina 2000), differences in the choice and productivity of "empty prefixes" (see e.g. Dickey 2005, 2008), if this notion is valid at all, cf. Janda *et al.* (2013), and, finally, variation in the distribution of aspects in many contexts, implying differences in the semantics of aspects among individual languages.

Moreover, as recent studies such as Tomelleri (2008, 2009, 2010) and Arkadiev (2014), Arkad'ev (2015) have shown, not all of the features traditionally associated with Slavic aspect are found in the languages with a similar kind of aspectual system (e.g. the pan-Slavic ban on perfective verbs under phasal predicates is conspicuously absent in Baltic and Hungarian; likewise, many languages lack secondary imperfectivization etc.), which does not necessarily render the non-Slavic systems "not grammaticalized" – rather, they are grammaticalized in a different way. In fact, the adequate assessment of the place of the Slavic and "Slavic-style" aspectual systems in the typology of aspect and verbal systems in general is still largely lacking, crucially requiring much more empirical work on languages with derivational aspectual categories (see Arkad'ev, Šluinskij 2015 for a pilot typological study).

Finally, the grammaticalization of "aspect" as a rather abstract morphosyntactic or morphosemantic feature (in the sense of Dahl 2004: Ch. 9) manifesting itself on the levels of morphology, syntax, semantics and pragmatics, should be distinguished from the grammaticalization of formal means of expression of perfectivity and imperfectivity in their different flavors, such as e.g. aspectual prefixes. As has been shown in Vimer (2001) and Arkad'ev (2008) for Lithuanian, productive employment of prefixes to mark perfectivity does not necessarily entail grammaticalization of aspect.

#### **2.2.** *Contact-induced grammatical change*

In this paper I distinguish two major types of borrowing or transfer from the donor language to the recipient language (see e.g. Sakel 2007, Gardani *et al*. 2015): MATter borrowing, i.e. "direct replication of morphemes and phonological shapes from a source language" (Matras, Sakel 2007: 829) and PATtern borrowing, which the same authors define as "re-shaping of language-internal structures" when "it is the patterns of distribution, of grammatical and semantic meaning, and of formal-syntactic arrangement .. that are modeled on an external source" (*ibidem*: 829-30). These phenomena have been long known as borrowing proper and calquing, respectively, but the unified terminology proposed by Sakel and Matras is more appealing (cf. also Johanson 1999, 2008 on "global" vs. "selective" code-copying).

The influence of languages with "Slavic-style" aspect on their neighbours has been studied from the perspective of language contact, see e.g. Wexler (1964, 1972), Talmy (1982), Šišigin (2014, 2015, 2016) on Yiddish, Ariste (1973), Igla (1998), Rusakov (2001), Schrammel (2002, 2005), Kožanov (2011) on Romani, Klepikova (1959) and Hurren (1969) on Istroromanian, Kardelis, Wiemer (2002), Pakerys, Wiemer (2007), Wiemer (2009) on Lithuanian dialects. The aim of this article, rather than to contribute new empirical material, is to offer a synthetic overview of several contact situations with "Slavic-style" aspect in the model language on the basis of existing literature and to demonstrate the limits of contact-induced change in the domain of grammatical aspect as distinct from perfectivization by means of preverbs.

#### **3.** *PAT-borrowing of prefixes*

It is well known that Eastern varieties of Yiddish have restructured the inherited Germanic system of preverbs under the influence of the neighbouring Slavic languages such as Polish, Belorussian, Ukrainian and Russian, see Wexler (1964, 1972), Talmy (1982) and a recent book-length study by Šišigin (2016). Notably, the Yiddish preverbs have acquired the systematic perfectivizing function (or perhaps have retained this function from Old German, in contrast to Modern High and Low German, which have largely lost prefixal perfectivization; see Eroms 1997, Wischer, Habermann 2004 on older stages of German).

Contact influence on the system of Yiddish verbal prefixes is mostly manifested in the copying of polysemy patterns, cf. examples (1)-(3). Note that some of the uses of Yiddish prefixes correspond not only to the Slavic ones, but to the German ones as well, while many other differ from them markedly.

(1) Yiddish

preverb *iber*- 'over'

*ibershraybn* 'copy, rewrite' ~ Rus. *perepisat'*, Germ. *überschreiben* 'id.' *iberton zikh* 'change clothes' ~ Rus. *pereodet'sja* vs. Germ. *übertun* 'overdo' *ibervinken zikh* 'wink to each other' ~ Rus. *peremigivat'sja* vs. no Germ. analogue (Talmy 1982: 243)

(2) preverb *far*- 'resultative'

*farblijen* 'start blooming' ~ Rus. *zacvesti* vs. Germ. *verblühen* 'cease blooming' *fartrinken* 'drink down after' ~ Rus. *zapit'* vs. Germ. *vertrinken* 'spend money on drink' *farshraybn* 'write down' ~ Rus. *zapisat'* vs. Germ. *verschreiben* 'prescribe' (Wexler 1972: 99-100)

(3) preverb *unter*- 'under' *untergebn* 'add' ~ Pol. *poddać* 'id.' vs. Germ. *untergeben* 'subordinate' (adjective) *untergisn* 'pour more' ~ Rus. *podlit*' vs. Germ. *hinzugießen*

*unterzogn* 'prompt, give a cue' ~ Rus. *podskazat'* vs. Germ. *untersagen* 'prohibit' (Šišigin 2016: 139-140)

Copying of the Slavic more abstract Aktionsart and even of the "purely perfectivizing" uses of prefixes has also occurred, cf. the examples in (4). It is worth noting that the Germanic prefixes in such functions occur both with native Yiddish and Slavic borrowed verbs.

(4) preverb *on*- 'surface' *onshraybn* 'write' ~ Rus. *napisat' ontseykhenen* 'paint' ~ Rus. *narisovat' onkormen* 'feed' ~ Rus. *nakormit' ontrinken* 'let drink' ~ Rus. *napoit'* (Šišigin 2016: 84, 87)

The issue of grammatical aspect in Yiddish has been subject to much controversy, cf. Aronson (1985) and Gold (1999) for critical overviews. On the one hand, prefixes can have clear aspectual impact, cf. examples (5a) and (5b), where the absence resp. presence of the prefix correlates with the imperfective (simultaneity) vs. perfective (precedence) interpretation.


On the other hand, the use of prefixed verbs in perfective contexts is not obligatory in Yiddish, and neither are prefixed verbs banned from imperfective contexts (Gold 1999: 75; cf. Aronson 1985), cf. examples (6)-(8); the last example shows that prefixed verbs in Yiddish can combine with phasal predicates.

(6) *Shpor-n zey op fun di kleyne fardinst-n*. save-prs.3pl they.nom pvb from def.pl small earning-pl 'They save from their small earnings.' (Gold 1999: 75)

(7) *Vi a fish lig-t er op gantse sho-en* like indf fish lie-prs.3sg he.nom pvb many hour-pl *unter-n vaser*. under-def water 'He lies under the water like a fish for many hours.' (ibid.) (8) *Es hob-n on-ge-hoyb-n aroys-gey-n etlekhe pedagogishe*  it have-prs.3pl pvb-ptcp-start-ptcp pvb-go-inf several pedagogical *zhurnal-n*.

magazine-pl

'There started being published (lit. go out) several pedagogical magazines.' (Corpus of Modern Yiddish <http://corpustechnologies.com:8080/YNC>)

As Talmy (1982: 242) concludes, "[w]hereas the Slavic prefix indicates … that the end point of a process is actually **reached** (unless countermanded by a secondary suffix), the Yiddish prefix indicates, rather, that the end point of a process is **in view**" (emphasis in the original). Thus, the use of the Slavicized prefixes in Yiddish has more to do with telicity, i.e. lexically encoded actional properties, rather than with perfectivity (grammatical aspect) per se, which, especially in the light of the observation by Gold (1999: 72) that "whether a verb is classified as perfective or imperfective has no other ramifications for the language", suggests a weaker degree of grammaticalization.

## **4.**  *MAT-borrowing of preverbs*

In this section I will present several case studies of languages which have borrowed whole systems of Slavic or Baltic verbal prefixes and use them as verbal modifiers and perfectivizers.

## **4.1.** *Romani<sup>2</sup>*

Wholesale borrowing of Slavic and Baltic (as well as Hungarian) verbal prefixes has occurred in a number of Romani varieties. Thus, North-Russian Romani uses Russian-origin prefixes as lexical modifiers of verbs, example (9), as well as perfectivizers, example (10).

North Russian Romani (Rusakov 2001: 315–316)

(9) *ot*- 'away': *otdes* 'give away' ~ Rus. *otdat' vy*- 'out'; *vydes* 'give out' ~ Rus. *vydat'*

<sup>2</sup> I am grateful to Kirill Kožanov and Anton Tenser for their generous help with the Romani data.

*raz*- 'apart': *rozdes* 'distribute' ~ Rus. *razdat'*  (10) *po*-: *popuchne* 'they asked' ~ Rus. *poprosili u-*: *uchorde* 'they stole' ~ Rus. *ukrali* 

Likewise, Latvian Romani uses the Latvian-origin prefixes in comparable functions (Ariste 1973, Manuš-Belugin 1973), cf. (11), and Bulgarian Romani uses prefixes originating in Bulgarian (Igla 1998: 67-70), cf. (12).


According to Rusakov (2001: 314), massive borrowing of the Russian verbal prefixes has induced a restructuring of the North-Russian Romani tense-aspect system: the original system with an inflectional distinction between the perfective and imperfective past tenses has been replaced by a symmetrical system with tense and aspect expressed independently of each other (cf. Table 1).


**Table 1.**  Restructuring of the tense-aspect system in North-Russian Romani

However, the actual situation is less straightforward. As Rusakov (2001: 315- 316) notes, there is no direct match between the aspectual functions of Romani and Russian prefixed vs. simplex verbs, rather Romani verbs exhibit a degree of free variation "without any obvious difference in meaning". Cf. the following pair of examples, where both the prefixed and the simplex verbs are used as equivalents to the Russian prefixed perfective *uznali* 'got to know':

(13) North Russian Romani (Rusakov 2001: 315)


Likewise, prefixed verbs are attested in unequivocally imperfective contexts in conjunction with unprefixed verbs, as in example (14) from an original literary text, notably featuring the "old" imperfective past in the habitual function.

(14) Soviet Romani (Kožanov 2015: 3)


According to Rusakov (2001: 316-317) and Kožanov (2015: 4), the use of "purely perfectivizing" prefixes with Romani verbs depends on the lexical actional characteristics, so that accomplishment and achievement (telic) verbs normally do not take prefixes even in perfective contexts, while activity (atelic) verbs have a greater propensity for prefixation, as can be seen in example (15).

(15) North Russian Romani (Rusakov 2001: 316–317)


Thus, it is legitimate to conclude that although North-Russian Romani has a system of borrowed prefixes used as lexical and aspectual modifiers of verbs, their role as perfectivizers is not yet fully grammaticalized and depends on lexical and contextual factors.

#### **4.2.** *Livonian*

Wholesale borrowing of prefixes also occurred in Livonian, a now extinct Finno-Ugric language which has been in intense contact with Latvian for at least a millennium (cf. Wälchli 1996, 2000 for an overview of Livonian-Latvian contact phenomena). The prefixes of Latvian origin in Livonian have been studied by de Sivers (1971), cf. also Sjögren (1861: 43-45), Wälchli (2001: 418), Ernštreits, Kļava (2014: 83-85). They occur as spatial as well as non-spatial lexical modifiers of verbs, as in examples (16) and (17), and as perfectivizers, as in example (18).


However, as in Romani, the aspectual uses of the Latvian loan preverbs in Livonian seem to be even less systematic than those of their Latvian prototypes (on the far from straightforward situation with aspect in Latvian see Hauzenberga-Šturma 1979, Holvoet 2000). Prefixed verbs can occur in imperfective contexts, as in example (19), and simplex verbs can be found in perfective contexts, as in example (20).


The cases of both North Russian Romani and Livonian clearly demonstrate that Slavic or Baltic prefixes are mostly borrowed into contact languages as lexi-

<sup>3</sup> I am grateful to Bernhard Wälchli for kindly providing interlinear glosses for this example.

cal modifiers of verbs with concrete semantic content, spatial as well as non-spatial, while when "aspectual" prefixes (or rather aspectual functions of prefixes) are also borrowed, their use does not become systematic, let alone obligatory. Hence, even the borrowing of whole systems of preverbs does not lead to the emergence of grammatical aspect in the recipient languages. In the next section we will see that even in a language which has apparently acquired a "Slavicstyle" aspectual system via contact, i.e. Istroromanian, the result is not a mere copy of the Slavic prototype.

#### **4.3.** *Istroromanian*

Istroromanian is a unique case of a language, which has borrowed from Slavic (Čakavian Croatian) not only a system of perfectivizing verbal prefixes, but the imperfectivizing suffix -*va* as well (Klepikova 1959, Hurren 1969, Kovačec 1966: 70–74, 1971: 125–130). This has been probably due to a large number of borrowed lexical verbs acquired by Istroromanian during several centuries of intense and asymmetric influence from Slavic (see Dahmen 1989: 448–453, Katunar 2008 on the history and sociolinguistics of Istroromanian). Istroromanian uses Slavic prefixes both as lexical modifiers of verbs, as in example (21), and as perfectivizers, as in example (22). In the perfectivizing function we find not only the prefixes of Slavic origin, but also the few inherited Romance preverbs (Dahmen 1989: 455), cf. *ănviså* 'dream'.

(21) *rez*- 'apart': *lega* 'tie' ~ *rezlega* 'untie', cf. Hrv. *razvezati* (Klepikova 1959: 45) *na*- + reflexive 'enough': *durmi* 'sleep' ~ *nadurmi (se)* 'sleep enough', cf. Hrv. *naspati se* (*ibidem*: 39)

*ze-* 'start': *plănje* 'weep' ~ *zeplănje* 'burst into tears', cf. Hrv. *zaplakati* (ibid.: 45) (22) *ćira* ~ *poćira* 'have supper', cf. Hrv. *povečerati* (*ibidem*: 38)

*parti ~ resparti* 'divide', cf. Hrv. *razdijeliti* (*ibidem*: 43)

The borrowed imperfectivizing suffix -*va* occurs both with simplex bases, as in example (23), and with prefixed bases, as in example (24), thus serving as a secondary imperfectivization device.


Istroromanian is claimed to have a grammaticalized aspectual opposition involving different morphological relations between imperfective and perfective verbs (Kovačec 1966: 71-72; Hurren 1969), summarized in Table 2.


**Table 2.**  Relations between imperfective and perfective verbs in Istroromanian

However, the distribution of simplex vs. prefixed vs. suffixed verbs in Istroromanian is not fully similar to that found in Slavic, in particular since many simplex verbs, of both Romance and Slavic origin, are treated as perfectives and form imperfective counterparts by suffixation, cf. example (25) with a Romance-origin verb and example (26) with a Slavic-origin verb<sup>4</sup> . It is possible that the recategorization of many simplex verbs of both Romance and Slavic origin as perfective has been due to the analogical extension of the Slavic-style model of prefixal perfective vs. suffixal secondary imperfective "pairs", which must have become productive in Istroromanian due to both lexical and morphological borrowing.


It has to be noted that simplex verbs with the imperfectivizing suffix have not only the iterative, but also the durative or progressive value (Klepikova 1959: 50, 52; cf. Hurren 1969: 70), see examples (27) and (28).

<sup>4</sup> Zegrean (2012: 126), evidently following Kovačec (1971: 125-130), states that the simplex vs. suffixed verbs are opposed as "imperfective" vs. "iterative", but does not provide confirming examples; examples given by Klepikova (1959) clearly indicate that the opposition is rather "perfective" vs. "imperfective".

(27) *Prevt-u jos mai jos cad-avei-t.* priest-sg down more down fall-ipfv-ptcp 'The priest was falling lower and lower.' (Klepikova 1959: 50) (28) *Prende ie trec-avei-t=a, ţel-a cral'-u* where he.nom pass-ipfv-ptcp=have.prs.3sg this-nom.sg.m king-sg *dobi-vei-t.* win-ipfv-ptcp 'Where he passed, this king, he kept winning.' (*ibidem*: 52)

My hypothesis about the makeup of the Istroromanian aspectual system is summarized in Table 3. From the data at hand it appears that telic and atelic base verbs behave differently with respect to perfectivizing prefixation: simplex verbs denoting telic processes and punctual events are usually perfective and form imperfective "partners" by suffixation, while simplex atelic verbs denoting states and activities, like in Slavic, are imperfective and admit perfectivizing prefixation and formation of suffixal iteratives. As to lexical modification by prefixes, which also induces perfectivization, it is available to all verbs, just as secondary suffixal imperfectivization.

**Table 3.**  The Istroromanian aspectual system


Thus, Istroromanian has borrowed from Slavic both the formal means of expressing perfectivity and imperfectivity and the more abstract aspectual opposition itself, but the resulting system is markedly different from the Slavic prototype, to the extent that Slavic originally imperfective verbal loans have been reinterpreted as perfective. This clearly shows that such complex linguistic phenomena as aspectual systems comprizing both functional and morphological oppositions arise via long-term language-internal processes (cf. Dahl 2004 on "mature" linguistic features) rather than by (even extreme) borrowing.

## **5.**  *Conclusions*

In this article I have shown that, on the one hand, borrowing of both matter (perfectivizing prefixes and, more rarely, the imperfectivizing suffix) and pattern (polysemy and the perfectivizing function of prefixes) from languages with "Slavic-style" aspect into languages with very different verbal systems is a well-attested phenomenon, but, on the other hand, even extensive borrowing does not lead to the creation in the recipient languages of aspectual categories grammaticalized to a degree similar to those of the donor languages. This is in line with the observation by Heine (2012: 132) that "replica categories are generally less grammaticalized than the corresponding model categories". This observation is corroborated by the fact that verbal aspect is conspicuously absent from the well-known lists of "balkanisms" (Aronson 1981, Rusakov 2007: 86- 87, Rusakov, Sobolev 2008: 28-29), i.e. linguistic features that have arisen due to diffusion and contact. Indeed, extensive language contact and influence of non-Slavic languages does not seem to have had any effect on the development of the "Slavic-style" aspect in Bulgarian and Macedonian, and neither have the aspectual systems of the latter served as models for replication in the non-Slavic Balkan languages (cf. e.g. Igla 1998: 70 on the aspectual system of Bulgarian Romani, which has experienced influence from the Slavic languages, but has by no means fully converged towards the Bulgarian system, notably, not employing the borrowed prefixes for aspectual purposes).

Even in the case of "extreme" borrowing as in Istroromanian, the resulting system, arguably highly grammaticalized, is a result of language-internal development and refunctionalization of borrowed material, rather than a direct copy of any Slavic system, as evidenced by the non-trivial treatment of many simplex borrowed verbs, which are imperfective in the source Slavic language but perfective in Istroromanian. This is also not surprising, since perfect cross-language alignment of highly grammaticalized morphosyntactic patterns is at best infrequent even in cases of the so-called "metatypy" (Ross 2007), i.e. restructuring of basic morphological and syntactic patterns induced by prolonged and intense contact, cf. Wiemer, Wälchli (2012: 37).

To conclude, contact-induced influences in the domain of aspect (at least, of the "Slavic-style" aspect) are to a large extent restricted to both matter- and pattern-borrowing of formally transparent and functionally loaded elements (Weinreich 1953: 34-35, Winford 2003: 91-92, Gardani 2008), i.e. Aktionsarten (including telicity) rather than highly abstract aspectual oppositions, and lexically and semantically, rather than morphosyntactically, determined categories.

#### *Abbreviations*


14 *Peter Arkadiev*


# *Bibliography*








20 *Peter Arkadiev*

## *Abstract*

#### Peter Arkadiev

#### *Borrowed preverbs and the limits of contact-induced change in aspectual system*

In this article, I survey several cases of contact-induced influence from the Slavic and Baltic languages with productive prefixal perfectivization on such languages as Yiddish, Romani, Livonian and Istroromanian. I show that, despite the fact that both matter and pattern borrowing of entire systems of Slavic or Baltic verbal prefixes is attested, grammatical aspectual categories similar to those of the donor languages do not arise in the contact languages. Even Istroromanian, which has borrowed from Slavic not only the perfectivizing prefixes but the imperfectivizing suffix as well, has developed an aspectual system markedly different from that of the Slavic prototype. This indicates that abstract grammatical oppositions such as Slavic or "Slavic-style" aspect are immune to direct transfer in language contact.

*Keywords:* Aspect, verbal prefixes, perfectivity, language contact.