Borrowed Morphology 9781614513209, 9781614515562, 9781501515873

Citation preview

Francesco Gardani, Peter Arkadiev, Nino Amiridze (Eds.) Borrowed Morphology

Language Contact and Bilingualism

Editor Yaron Matras

Volume 8

Borrowed Morphology Edited by Francesco Gardani Peter Arkadiev Nino Amiridze

ISBN 978-1-61451-556-2 e-ISBN (PDF) 978-1-61451-320-9 e-ISBN (EPUB) 978-1-5015-0037-4 ISSN 2190-698X Library of Congress Cataloging-in-Publication data A CIP catalog record for this book has been applied for at the Library of Congress. Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available on the Internet at http://dnb.dnb.de. © 2015 Walter de Gruyter, Inc., Berlin/Boston/Munich Typesetting: Compuscript Ltd., Shannon, Ireland Printing and binding: CPI Books GmbH, Leck ♾ Printed on acid-free paper Printed in Germany www.degruyter.com

Contents Francesco Gardani, Peter Arkadiev and Nino Amiridze Borrowed morphology: an overview  1 Part I  Theory Sarah G. Thomason When is the diffusion of inflectional morphology not dispreferred? Yaron Matras Why is the borrowing of inflectional morphology dispreferred?

 27  47

Part II  Borrowing of derivation Eleanor Coghill Borrowing of verbal derivational morphology between Semitic languages: the case of Arabic verb derivations in Neo-Aramaic  83 Metin Bağrıaçık, Angela Ralli and Dimitra Melissaropoulou Borrowing verbs from Oghuz Turkic: two linguistic areas  109 Martine Robbeets Common denominal verbalizers in the Transeurasian languages: borrowed or inherited?  137 Part III  Borrowing of inflection Brigitte Pakendorf A comparison of copied morphemes in Sakha (Yakut) and Ėven

 157

Felicity Meakins From absolutely optional to only nominally ergative: the life cycle of the Gurindji ergative suffix  189 J. Clancy Clements and Ana R. Luís Contact intensity and the borrowing of bound morphology in Korlai IndoPortuguese  219 Françoise Rose Innovative complexity in the pronominal paradigm of Mojeño: a result of contact?  241

vi 

 Contents

Thomas Stolz Adjective-noun agreement in language contact: loss, realignment and innovation  269  303 Index of subjects Index of languages  307

Francesco Gardani, Peter Arkadiev and Nino Amiridze

Borrowed morphology: an overview 1 Introduction

Borrowing has traditionally occupied a prominent role in historical linguistics, as it has been viewed as one of the main sources of language change, besides sound change and analogy. While lexical borrowing has attracted particular interest, the borrowing of morphology has generally attracted less attention in the literature. There is no doubt that this can be explained in terms of the apparent relative infrequency of morphological borrowing. At the turn of the 20th century, two schools of thought dominated this debate. On one hand, advocates of a retentionist view (Müller 1862; Meillet 1921; Sapir 1921; Jakobson 1938) claimed that the borrowing of inflectional morphemes is most unlikely. Most explicitly, Meillet maintained that “il n’y a pas d’exemple qu’une flexion comme celle de j’aimais, nous aimions ait passé d’une langue à une autre” (1921: 86). On the other hand, Schuchardt, one of the proponents of the opposed diffusionist view (to whom scholars such as Whitney 1881 and Trubetzkoy 1939 also belong), claimed that there are no completely unmixed languages and that morphological borrowing exists (Schuchardt 1884: 9). The first analytical framework for the study of language contact in general, and borrowing in particular, was provided by Weinreich (1953), who observed that derivational affixes are more easily transferable from one language to another than inflectional affixes, while at the same time reporting instances of inflectional morphemes that were transferred from one language to another (Weinreich 1953: 31–33). Following Weinreich’s seminal work, and based on the apparent resistance of bound morphology to contact-induced change, linguists have interpreted the borrowing of morphology as a reflex of very strong social pressure that one language, the source language (SL), exerts over another, the recipient language (RL). In order to seize different degrees of borrowability, linguists have developed a number of borrowing scales (e.g. Whitney 1881: 19–20; Haugen 1950: 224; Moravcsik 1978: 110–113; Thomason and Kaufman 1988: 74–76; Field 2002: 36–37). All currently accepted hierarchies deem a high intensity of contact to be necessary for morphological borrowing to occur (Matras 2007; Matras 2009: 153–165 and Wohlgemuth 2009: 11–17, provide useful overviews). The last decade has seen an increased interest in contact-induced morphological change, and several publications reflect this tendency, such as Borrowing of inflectional morphemes in language contact (Gardani 2008), Copies versus cognates

2 

 Francesco Gardani, Peter Arkadiev and Nino Amiridze

in bound morphology (Johanson and Robbeets 2012), and Morphologies in contact (Vanhove et al. 2012). All in all, a number of attempts have been made to put this field of research on both a theoretical and an empirical footing. Today, thanks to both the compilation of grammars of previously undescribed languages and the publication of studies on structural borrowing based on large cross-linguistic data (e.g. Matras and Sakel 2007a; Wohlgemuth 2009), a substantial number of instances of morphological borrowing are known, and useful comparative analyses – in terms of putatively universal tendencies – have been proposed. Thus, while progress has been made on the empirical side in terms of an extension of the number of attested instances of morphological borrowing, on the theoretical side things have proceeded more slowly. That is, despite the fact that linguists have recognized in their approaches the potential of contact-induced morphological change as a source of evidence for the structure of grammar (see, e.g. Myers-Scotton 2002, 2006; Gardani 2008, 2012; Meakins 2011a), more theoretically inspired work needs to be pursued in order to get deeper insights into the matter and be able to formulate more valid generalizations. The present book presents advancements in research in morphological borrowing, addressing the need for improving the conceptual and methodological basis of this field of linguistics. The contributions to this volume reflect heterogeneous theoretical and methodological tools, based on the editors’ belief that only a variety of approaches can help capture the array of diverse phenomena with which the data confront us. In the sections that follow, we will sketch the state-of-the-art of current research in morphological borrowing and situate the volume’s articles within the research landscape. Among the issues addressed in the volume, one fundamental question concerns the borrowability of morphology. Is morphological borrowing an infrequent phenomenon in cross-linguistic terms, or is it not as rare as is often purported in the literature? A scientific treatment of this question requires, first and foremost, an elaboration of several fundamental distinctions, such as the questions about what is borrowed in terms of matter versus pattern (Section 2), and which type of morphology, derivational or inflectional, is borrowed (Section 3). A further central query relates to the relationship (or distinction) between morphological borrowing sensu stricto and phenomena such as code-switching, creolization, and the genesis of mixed languages. Pursuing this last question requires a better understanding of the interplay between sociolinguistic and cognitive conditioning factors of interlinguistic transfer, on the one hand, and different degrees of borrowing, on the other. These issues are treated in Section 4. On the methodological side, the investigation of morphological borrowing is of great importance to historical-comparative linguistics, as correspondences between inflectional and derivational morphemes have often been taken

Borrowed morphology: an overview  



 3

as strong indicators of, or even diagnostic evidence for, genetic relatedness (see, e.g. Meillet 1921; cf. the discussion in Ross and Durie 1996: 7) (Section 5). No less important are the understanding of cross-linguistic tendencies in morphological borrowing as well as its linguistic and social motivations for linguistic typology and the study of language universals, because morphological borrowing, especially pattern borrowing, is among the principal factors responsible for the diffusion of structural traits and the development of linguistic areas (see, e.g. Ross 1999 or Donohue 2012).

2 mat-borrowing versus pat-borrowing Adopting the terminology of Sakel (2007) and Matras and Sakel (2007b), we distinguish between two types of borrowing: the borrowing of concrete phonological matter (mat-borrowing); and the borrowing of functional and semantic morphological patterns (pat-borrowing) from a SL into a RL. (Both types are compatible with borrowing derivation and borrowing inflection; on this, see Section 3.) This distinction is by no means new, and looks back at a rich terminological history. The first type has traditionally been referred to as “borrowing”, “direct transfer”, “direct diffusion”, “transfer of fabric”; the second type has often been called “replication”, “indirect transfer”, “indirect diffusion”, “loan-formation”, “calque”. See also Johanson’s (1999, 2008) terms of “global copying” (roughly corresponding to mat-borrowing) vs. “selective copying” (roughly corresponding to pat-borrowing). Morphological pat-borrowing implies that a RL rearranges its own inherited morphological structure in such a way that it becomes structurally closer to the SL. An instance of pat-borrowing from derivational morphology is found in Basque, which replicates a Romance pattern to form deverbal verbs through a prefix expressing repetition. The Basque formative that replicates the Romance pattern expressed by re‑ (cf. Spanish reproducir ‘to reproduce’) is bir‑ (or its ­allomorph berr‑), as in (1a), compared to the corresponding Spanish lexemes in (1b) (Basque data from Jendraschek 2006: 158–159). (1) Basque a. aztertu ‘examine’ berr-aztertu ‘re-examine’

Spanish b. examinar re-examinar

In nominal morphology, a pertinent example of pat-borrowing is the use of the category of nominal past in Mawayana (Maipurean, Guyana), which has emerged

4 

 Francesco Gardani, Peter Arkadiev and Nino Amiridze

because of contact with the Cariban languages. In Mawayana, the form ‑ba is suffixed to a nominal element and replicates the Cariban obligatory marking of the nominal past, used to express former possession, deceased persons, gone objects, or pity (Carlin 2006: 322–325). See the use of the suffix to express a former possession in Mawayana, in (2a), and compare it to the Trio (Cariban) equivalents of the nominalized form, in (2b). (2) a. njee katabi-ke-ba jimaaɗa (Mawayana) human.being catch-ag.nmlz-pst jaguar ‘Jaguar used to catch people.’ (lit. jaguar was a catcher of people) b. wïtoto apëi-ne-npë teese kaikui (Trio) human.being catch-ag.nmlz-pst he.was jaguar ‘Jaguar used to catch people.’ In this volume, Thomas Stolz provides a fascinating cross-linguistic study on ­ at-borrowing, with a focus on NP-internal agreement (concord). Based on a p wealth of cross-linguistic data, Stolz proposes and exemplifies three scenarios of change in adjective-noun agreement in contact situations: (1) loss of agreement (Armenian in contact with Turkic); (2) reshaping of agreement on the model of the SL (Nahuatl in contact with Spanish); and (3) rise of agreement (Baltic-Finnic in contact with Indo-European languages). A subtype of pat-borrowing is contact-induced grammaticalization or, in Heine and Kuteva’s (2003) terminology, “replica grammaticalization”, which, as they claim, involves the replication of a process of grammaticalization rather than of a fixed pattern. For a recent reassessment of contact-induced grammaticalization, see Wiemer et al. (2012) and in particular, Gast and van der Auwera (2012). For example, based on the model of neighboring Ewe, Likpe (both belong to different branches of the Kwa family in the Niger-Congo phylum, Ghana, Western Africa) has developed plural-marking on a subset of kin terms (the ego’s parents’ generation) and proper names (Ameka 2006: 126–127). The pluralizing suffix ‑mə́, in (3a), has the same form and meaning as the 3pl pronoun mə́, in (3b) (Ameka 2006: 130). (3) Likpe a. éwú éwu-mə́ grandmother grandmother-pl b. mə lə́ ntí 3pl loc midst ‘among them’ The evolution from (3b) to (3a) parallels the Ewe patterns in (4). In Ewe, wó is both a plural clitic on nouns (4a) and a third person plural pronoun (4b).



Borrowed morphology: an overview  

 5

(4) Ewe a. ame (eve má=)wó ko person two dem=pl only ‘only (those two) people’ b. wó-dzo (wó) 3pl-fly 3pl ‘They flew (them).’ As we have already mentioned, mat-borrowing concerns the concrete phonemic matter that an RL takes from an SL. An appropriate example of mat-borrowing has been described by Breu (1991) for Bulgarian, Macedonian, and other Balkan languages. Throughout the Balkan Sprachbund, the formative ‑s‑ is productively used as a loanverb marker. It was borrowed from the Greek verbalizer ‑iz‑, such as in alat‑íz‑o ‘to salt’ from aláti ‘salt’. In Macedonian, for example, the Turkish verb bit-mek ‘to finish’ has been integrated as a composite stem biti‑s‑, to which the stem-building formative and the inflections apply (data from Breu 1991). (5) Macedonian biti‑s‑uv‑a finish‑lvm‑suff-prs.1sg ‘I finish’ In our volume, the issue of borrowed loanverb formatives is taken up by Metin Bağrıaçık, Angela Ralli and Dimitra Melissaropoulou, who analyze it in areal terms. Two distinct Turkic suffixes borrowed into several typologically distinct languages are used to create “input forms” (Wohlgemuth 2009, Ch. 5) to accommodate loanverbs from Oghuz Turkic. The distribution pattern of the borrowed suffixes enables the authors to identify two separate linguistic areas. The perfect/inferential marker -mIş (accompanied by a light verb) is found in the area including Eastern Asia Minor, Transcaucasia, and Transoxiana, while the past marker ‑D(I) (with no light verb present) occurs in borrowed Turkic verbs in various languages of the area encompassing the Balkan peninsula and Western Asia Minor. Crucially, the paper shows that structural reasons are at hand for the selection of either formative: in the case of ‑D(I), the selection is determined by the type of base that “is operative in the recipient language for word-formation purposes”, whereas in the second area, the selection of ‑mIş is guided by the independent existence of both perfect grams and the use of a light verb strategy to create denominal verbs in the RLs. The illustration of mat-borrowing through example (5) suffices to fill the space of this brief overview, because nine out of ten papers of the volume focus

6 

 Francesco Gardani, Peter Arkadiev and Nino Amiridze

on mat-borrowing. This preponderance has been the editors’ explicit choice, not least because to date publications have been focusing on pat-borrowing rather than on mat-borrowing (see Mithun 2012, for a very recent paper, and many articles in Matras and Sakel 2007a; for contact-induced grammaticalization, see Grandi 2002; Heine and Kuteva 2003, 2005, 2006; Gast and van der Auwera 2012; Wiemer et al. 2012). Another issue that we placed on the agenda of research on morphological borrowing and which is not systematically represented in the present volume (though cf. the contribution by Felicity Meakins, who investigates the functional development of the Gurindji ergative marker in Gurindji Kriol) is the question of the degree of semantic-functional matching between a borrowed morpheme in the RL and its counterpart in the SL. For example, the study of “relabelling” in creoles and mixed languages (Lefebvre 2008), that is, the process whereby phonetic strings drawn from the lexifier language replace original forms expressing the same concept in the substrate language(s), has shown that the new lexeme has the same semantic and syntactic properties of the original one, but its phonological representation is different. Conversely, lexical borrowing need not involve the transfer of the full polysemy of the SL’s lexical items (see, e.g. Weinreich 1953: 55–56; Rohde et al. 1999). Finally, Heine (2012) claims that, in contact-induced grammaticalization, the replica element or construction in the RL almost invariably occupies a less advanced stage of functional-semantic development than its model in the SL. There is thus no reason to assume that mat-borrowed grammatical morphemes in a RL take over the full gamut of functions of their sources, as is implied, e.g. in Johanson’s notion of global copying. As has been repeatedly shown by different scholars (see Winford 2003: 91–92, for an overview), if interlinguistic transfer of morphemes occurs at all, it is the morphemes with a higher degree of functional transparency that are borrowed more frequently. From this, it follows that morphemes that are polyfunctional in the SL, are borrowed into the RL primarily with their more concrete and transparent functions. This claim is supported by studies on the borrowing of Slavic and Germanic verbal prefixes and particles into various contact languages, such as varieties of Romani (see, e.g. Rusakov 2001; Schrammel 2002) or Balkan Romance languages. For instance, in the varieties of Romanian spoken in Serbia, the prefix do- borrowed from Slavic denotes the attainment of the final point of motion or activity (Petrović Rignault 2008), as the following example shows.  Serbian (6) Vlach Romanian a. do-facu     b.  do-jesti prv-do:pst.3sg  prv-eat:inf ‘S/he finished doing sth.’  ‘to finish eating’



Borrowed morphology: an overview  

 7

Importantly, however, in Vlach Romanian, the borrowed prefix does not have the perfectivizing role characteristic of the SL, Serbian, as well as of the Slavic verbal prefixes, in general. That means that the more abstract function realized by the morphemes of the SL, or even associated not just with particular morphemes but with the whole make-up of the verbal system, has not been introduced into Vlach Romanian. This example alone shows that the semantic aspect of morphological borrowing is de facto quite complex and deserves much more attention than, to our knowledge, it has received so far. In addition to mat-borrowing and pat-borrowing, there seems to be a type of morphological transfer that lies in between. In our volume, Eleanor Coghill portrays a complex issue of verbal derivational patterns borrowed from Arabic by three distinct modern Aramaic languages. The distribution of phonological material in the Semitic verbal stem is organized by segmental morphology and more abstract structural templates. Coghill shows that Arabic loan derivations are first largely limited to Arabic loanverbs, but can subsequently spread to the inherited Aramaic lexical stock, giving rise, for example, to a new mediopassive category in Western Neo Aramaic.

3 Borrowability of morphology It is common knowledge that morphology is a cover term for a rather wide range of phenomena, roughly including compounding, derivation, and inflection, which seem to be processed in different areas in grammar. Accordingly, claims have been made that different areas of morphology show different degrees of propensity for borrowing, which is reflected in the various borrowability scales mentioned above. Generally, it is assumed that derivation is borrowed more frequently than inflection: for example, Thomason and Kaufman (1988: 74–75) situate derivational borrowing (der-borrowing) at level 3 of their borrowing scale, whereas inflectional borrowing (inf-borrowing) ranks at level 4, the highest level. With respect to inflection, Gardani (2008, 2012) has shown that variance in the degree of borrowability of inflectional formatives correlates with their classification as realizing either inherent inflection or contextual inflection, according to Booij’s (1994, 1996) famous dichotomy. In this connection, the borrowing of formatives that realize features of inherent inflection (i.e. context-autonomous inflection), such as nominal number or semantic case, verbal voice, tense, aspect, negation, mood, or evidentiality, largely outweigh the borrowing of formatives that realize contextual inflection (i.e. inflection induced by obligatory syntactic

8 

 Francesco Gardani, Peter Arkadiev and Nino Amiridze

government or agreement), such as nominal grammatical case or verbal person, number, and gender.1 As a prototypical value of inherent inflection, nominal plural has a higherthan-average borrowing rating (Gardani 2012). As a case in point, consider the case of Bolivian Quechua nouns ending in a vowel (the vast majority), which realize the plural via a suffix ‑s borrowed from the contact language, Spanish [data in (6a) from Muysken (2012: 33), based on Urioste (1964)].   Spanish (7) Bolivian Quechua a. algu b. perro ‘dog’ algu-s   perro‑s ‘dogs’ An example of the rare cases of borrowed formatives that realize contextual inflection is found in Megleno-Romanian, a Balkan Romance language spoken in south-eastern Macedonia and northern Greece. In the Megleno-Romanian ­varieties spoken in the villages of Nănti, Ošinj, Lundzinj, and Kupă, some verbs, belonging to a specific theme vowel class (‑a‑) and ending in the consonant cluster muta cum liquida, display the formative ‑ş, for the 2sg of the indicative present, which is added to the corresponding native Romance formative -i on inherited Romance bases (a antra ‘to enter’ < Latin intrare) [data in (8a, b) from Capidan (1925: 159) and Atanasov (1990)].2 The formative ‑ş has been borrowed from south-eastern Macedonian dialects. In (8), the verb form with borrowed formative (8a) is contrasted with the corresponding form both of the same verb in the standard variety of Megleno-Romanian (8b) and of the Macedonian verb gali ‘to caress’ (8c).

1 There is no disguising the fact that the distinction between inflection and derivation is neither obvious nor uncontroversial, and so is the distinction between morphological compounding and formation of phrases in syntax (see, e.g. Booij 2005, 2010). Both dichotomies are to a large extent language-specific. Born out in linguistic studies focusing on Indo-European languages, the distinction between inflection and derivation has proved “particularly elusive” to capture (Laca 2001: 1215). Some scholars (e.g. Bybee 1985; Dressler 1989; Plank 1994) have advocated a non-discrete, gradual distinction along a continuum which matches that ranging from the syntax to the lexicon, while others, like Bauer (2004), have proposed a more refined typology of morphological processes with several, instead of just two or three, major types. Still others (e.g. Behrens 1996; Haspelmath 2013) challenge the validity of this distinction as a universally applicable comparative concept. See Laca (2001: 1215–1218), for an insightful discussion. 2 While an explanation of the phenomenon in terms of an internal Romance development is conceivable, too, the explanation in terms of the influence of Macedonian on the MeglenoRomanian dialects cannot be ruled out completely (see Friedman 2012: 324–328).



Borrowed morphology: an overview  

 9

(8) Megleno-Romanian Macedonian a. antri‑ş    b. antri    c. gal‑iş enter:2sg-2sg enter:2sg caress-2sg ‘you enter’ ‘you enter’ ‘you caress’ While there seems to be a consensus that inflectional borrowing is a relatively rare phenomenon (although its actual frequency in different contact situations is still to be determined in a world-wide typological study), derivational morphology seems to be more susceptible to borrowing. The amount of data collected in a wealth of publications indicate this tendency, although to date there has been no comprehensive survey of the great amount of borrowed derivational morphology in the world’s languages. The general consensus about this claim rests ultimately on the abundance of derivational borrowings in the most studied language of the world – English – dating back to the time when (Middle) English extensively borrowed from French. Recent works, such as the papers collected in Matras and Sakel (2007a), Matras (2009: 209–212) and, especially, Seifart’s (2013) newly published A world-wide survey of affix borrowing (AfBo) have provided a collection of numerous instances of derivational borrowing. A superficial look at AfBo shows that adjectivizers, diminutives, and nominalizers rank highest among the borrowed derivational affixes. This conforms to the long-held opinion that categories which carry out “concrete” meaning are more prone to borrowing. We exemplify this with a case from Tetun Dili, an Austronesian language spoken in East Timor, which has borrowed the agentive suffix ‑dór (9a) from Portuguese (9b) and applies it to native roots, as in the following example from Hajek (2006: 172). (9) Tetun Dili Portuguese a. hemu-dór    b. descobri‑dor ‘someone who likes to drink’ ‘discoverer’ On the basis of the currently available evidence and the publications mentioned above, we propose the following tentative borrowability scale for morphology: derivation > inherent inflection > contextual inflection (an idea originally developed in Gardani, in press). Further empirical research and theoretical insights are certainly needed in order to test and refine this generalization and especially to provide a principled explanation for the “differential access” of different kinds of morphology to borrowing, grounded in identifiable cognitive factors rather than in the rather vague ­ yers-Scotton’s and elusive dichotomy between inflection and derivation (cf., e.g. M 4M-model as a possible approach to this issue, see Myers-Scotton 2002). In this volume, the general issues of borrowability of morphology are addressed from different perspectives by two leading experts in the field of contact ­linguistics.

10 

 Francesco Gardani, Peter Arkadiev and Nino Amiridze

Sarah Thomason argues that inflectional borrowing is “considerably more common than one might guess from the general language contact literature” and shows that the borrowing of inflectional matter is especially common in situations characterized by intense contact and by close relatedness of languages and varieties of the same language. In contrast, Yaron Matras argues that cognitive, communicative, and sociocultural constraints inhibit the borrowing of morphological matter, ­especially inflectional morphology. He maintains that “[s]traightforward cases of borrowed inflectional morphemes are hard to find” and addresses the issue of the differential susceptibility of derivational vs. inflectional morphology to borrowing from the viewpoint of his “activity-oriented” approach (Matras 2009, 2012). Matras – in our view similarly to Myers-Scotton’s theory – considers inflection to be indicative of the language choice made by the bilingual speaker and related to their identity, whereas derivational morphology, because of its heavier semantic load, is in charge of constructing and modifying meanings. In Matras’ terms, “the purpose of borrowed derivational morphology is to replicate procedures of meaning derivation from the source language in the recipient language”, while “the purpose of borrowed inflectional morphology is to re-draw social boundaries”, and thus the borrowing of inflectional morphology, having considerably more far-reaching effects on both the language system and the social identity of the speakers, is “strongly dispreferred”. In an attempt to reconcile Thomason’s and Matras’s proposals on the borrowability of inflectional morphology in situations of ordinary language contact (for extreme borrowing and language mixing, see Section 4), one might hypothesize that, given a disparity between linguistic communities in terms of prestige, speakers of the less prestigious language who strive for a higher social status may be more prone to borrowing inflectional matter the higher the degree of structural similarity between the languages is. Obviously, only the investigation of morphological borrowing based on the largest possible number of contact situations, diverging in terms of degree of genealogical relatedness, structural congruity of the languages involved, and sociolinguistic scenarios, will allow for robust generalizations and principled explanations of what are preferred and dispreferred types of morphological borrowing.

4 Extreme borrowing and mixed language genesis Different language contact situations can give rise to different linguistic ­processes and results. Thomason (2001: 60) proposes a three-fold outline, based on the structural effects induced by language contact, including contact-induced language change, extreme language mixture, and language death. With respect to morphological borrowing, some scholars (e.g. Thomason and Kaufman 1988) treat



Borrowed morphology: an overview  

 11

ordinary contact-induced change and mixed languages as separate ­phenomena, based on the argument that languages which have undergone contact-induced change, but are clearly traceable back to a single ancestor from which they descend, cannot be treated on a par with new languages that emerged from the mixture of two or more SLs under specific social circumstances. However, other scholars (e.g. Field 2002; Meakins 2011a), do not differentiate between “ordinary” contact-induced change and language mixing and use “borrowing” to describe both developments due to ordinary contact-induced language change and occurrences of SL-derived formatives in creoles and mixed languages. Clearly, the type of morphology that is found in mixed languages differs – in both quantitative and qualitative terms – from that found in languages that have undergone ordinary contact-induced morphological change. As a matter of fact, the status of a language as a mixed language is acknowledged precisely because it shows (complete) paradigm borrowing. The transfer of entire inflectional paradigms has long been regarded as the last challenge to morphological borrowability. Weinreich (1953: 44) had observed that the adoption of a full set of morphemes “has apparently never been recorded”. The later-reported case of the Russian-derived finite verbal paradigm in Mednyj Aleut, which was spoken east of Kamchatka in Russia, has been dealt with as a case of borrowing of entire inflectional paradigms (Menovščikov 1969; Golovko and Vakhtin 1990; Thomason 1997). But considering this to be a case of borrowing is not uncontroversial, for the very fact that Mednyj Aleut is a mixed language. In (10), we exemplify paradigm borrowing in Mednyj Aleut, comparing the present tense forms of Mednyj Aleut (10a) with those of Russian (10b) and of Bering Aleut (10c) (data from Thomason and Kaufman 1988: 234–235; based on Menovščikov 1969: 132). (10) Mednyj Aleut Russian3 Bering Aleut a. b. c. 1sg uŋuči-ju stro-ju uŋuči-ku-q 2sg uŋuči-iš stro-iš uŋuči-ku-x̣t 3sg uŋuči-it stro-it uŋuči-ku-x̣ 1pl uŋuči-im stro-im uŋuči-ku-s 2pl uŋuči-iti stro-iti uŋuči-ku-x̣t-xičix 3pl uŋuči-jat stro-jat uŋuči-ku-s ‘I/you/he/she/we/they sit’ ‘I etc. build’ ‘I etc. sit’

3 We have replaced the Russian example given by Thomason and Kaufman through a verb from a more productive inflectional class showing more direct correspondences to the Mednyj Aleut borrowed morphemes; Russian wordforms are given in a broad phonological transcription.

12 

 Francesco Gardani, Peter Arkadiev and Nino Amiridze

In our volume, Brigitte Pakendorf reports on a case of paradigm borrowing in Lamunkhin Ėven. This endangered Western dialect of Ėven (Northern Tungusic), spoken in the village of Sebjan-Küöl in the Republic of Sakha (Yakutia, Siberia, Russia), has heavily borrowed from the Turkic language of Sakha (a.k.a. Yakut). Two paradigms, the necessitative and the assertive, are established borrowings, while two other paradigms, the indicative present tense and the hypothetical mood, are potentially ongoing borrowings. Also, Pakendorf compares morphological borrowing in Lamunkhin Ėven with that in Sakha (Turkic) and shows that, while affixes borrowed from Sakha into Lamunkhin Ėven are mostly inflectional, the majority of Mongolic morphological borrowings in Sakha are derivational suffixes. The author demonstrates that it is not similar structural preconditions, such as structural congruence, and similar contact situations (both RLs are subordinate in socio-political terms), but divergent sociocultural characteristics of the two contact situations that have played a determining part in the borrowing asymmetry found. While molecular anthropological material reveals no intimate contact between the Sakha ancestors and Mongols, genetic evidence discloses an intimate, long-lasting social and physical interaction between the Ėven and the Sakha, which makes the existence of contact on a close level (through intermarriage and bilingual families) a probable factor boosting inflectional borrowing. In the discussion of structural changes in language mixing, the case of Gurindji Kriol (northern Australia) is of particular relevance, because it is – to date – the only mixed language to have been documented diachronically from the mid-1970s through to today (McConvell 1988; McConvell and Meakins 2005; Meakins 2011b), although with serious lacunae in the 1980–90s, when extensive code-switching between the Australian language Gurindji and the English-based Kriol started to stabilize, giving rise to the mixed language (Meakins 2011b: 145). Meakins shows that, in Gurindji Kriol, the ergative and dative case-formatives are Gurindji-derived. In the following example, (11a) shows a Kriol-derived noun (cf. English pussycat), to which the Gurindji-derived ergative suffix applies (Meakins 2011a: 68), while (11b) shows a noun marked by the same source ­formative, in Gurindji (Meakins 2011b: 14). (11) Gurindji Kriol a. pujikat-tu-ma    b. ngakparn-tu cat-erg-top frog-erg In our volume, Meakins deepens her research on Gurindji Kriol and shows that the ergative case suffix not only applies to transitive subjects only optionally



Borrowed morphology: an overview  

 13

but can also mark (again optionally) intransitive subjects and is thus ­analyzable as an optional marked nominative rather than an ergative. Meakins shows that the development of the case morpheme from an obligatory and exclusively ergative marker in Gurindji to a pragmatically loaded nominative marker in Gurindji Kriol, is linked to differences in word order and syntactic encoding of information structure between Gurindji and Kriol. All this makes the paper an important contribution not only to the study of mixed languages and morphological borrowing but also to the diachronic typology of case systems. A topic that has so far never been addressed in the language contact literature is the borrowing of so-called “autonomous morphology” (Aronoff 1994; Maiden et al. 2011).4 In our volume, Clancy Clements and Ana Luís investigate a peculiar case of morphological borrowing in Korlai Indo-Portuguese, a Portuguese-based creole language in contact with Marathi, which, in addition to the three conjugation classes inherited (in reduced form) from Portuguese, has created a new inflectional class, specifically for integrating Marathi loan verbs. By investigating the Marathi verbal paradigm, the authors show that the source of the new inflectional class in Korlai Indo-Portuguese was the non-finite form occurring in the Marathi negative imperative construction, and that this inflectional affix was reanalyzed as an inflectional class marker in the RL. Language mixing under special sociolinguistic conditions seems to be the only known type of language contact in which contextual inflection, such as nominal case markers or verbal person-number inflections, is systematically transferred from one language to another. Going back to Matras’s claim that the borrowing of inflection is a function of “redrawing of social boundaries”, we should relate the fact that the transfer of contextual inflection does happen in situations of language mixing to the sociolinguistic frame in which language mixing occurs. According to a well-established assumption, language mixing can be motivated by dissociation of identity, that is, the wish of one of the groups involved in the contact situation to establish a new social and linguistic identity, distinct at least from the identity of the sociolinguistically and culturally dominating group language. From this, we can infer that the conditions under which

4 Autonomous or pure morphology is a kind of morphology that has relevance only for the morphological component of grammar; it doesn’t serve the syntax, nor contribute any kind of meaning, and thus stands by itself. Inflectional class formatives, e.g. in the Slavic and Romance languages, are examples of autonomous morphology.

14 

 Francesco Gardani, Peter Arkadiev and Nino Amiridze

language mixing occurs, favor the transfer of formatives that realize contextual inflection (on this issue, see, e.g. Croft 2003; Matras and Bakker 2003: 13–15; Meakins 2013: 181, 184–185).

5 Borrowing or inheritance? As already claimed by Boas (1917, 1920), language contact can affect languages to such an extent that genealogical traceability becomes impossible (see also Swadesh 1951, for the so-called Boas-Sapir controversy). Thomason and Kaufman (1988, especially chapter 8) have proposed to assess genealogical affiliation in situations of language contact in terms of a distinction between “normal” language transmission and abnormal or “interrupted” transmission. In “normal” language transmission, the genealogical traceability of a language is possible even in cases of strong contact influence, whereas in abnormal transmission, usually involving imperfect adult language acquisition, genealogical ties are disrupted and can pave the way for the emergence of mixed languages. In particular, Thomason and Kaufman argue that languages in which the lexicon and the morphology come from different sources are suspect of having undergone abnormal transmission sometime in their history (see Donohue 2013, for a discussion of this issue with respect to the genealogical classification of some Melanesian languages as either “Austronesian” or “Papuan”). Recently, Karatsareas (2009, 2014) has established clarity in the much debated and intricate study of Cappadocian Greek as an extreme case of language change. Focusing on the loss of the Greek traditional tripartite gender distinction into masculine, feminine, and neuter, the author demonstrates that this innovation is not the result of language contact with Turkish, as has been generally claimed in the literature, but rather the result of a series of internal analogical processes, which were probably boosted, but not triggered, by language contact. Extreme contact situations and their outcomes notwithstanding, recent publications have shown how important it is to distinguish between cognate and borrowed morphological elements (Johanson and Robbeets 2012). In cases in which the genealogical relatedness of languages is still a matter of debate (as, e.g. in the Transeurasian or Altaic hypothesis), this distinction is fundamental, because the identification of certain formally and functionally corresponding elements as inherited from a common antecessor language, and not as of a result of borrowing, provides strong evidence for common genealogical ancestry. Thus, the study of morphological transfer has proved a useful heuristic tool in



Borrowed morphology: an overview  

 15

investigations of the genealogical relatedness of languages or language groups (see Robbeets 2012, for a fruitful example with respect to the T ­ ranseurasian ­hypothesis; and Whaley 2012, for a study of genealogical relatedness, based on evidence from derivational morphology).5 No less relevant is the distinction between “copies” and “cognates” when the focus is both on the degree of contact-induced convergence between genealogically related languages and the origins of morphemes common to two or more languages (see, e.g. Bowern 2013; Epps 2013; Law 2013; Mithun 2013, for morphological change induced by contact between genealogically related languages). A particularly telling example of difficulties that possibly arise in the study of morphological borrowing between closely related languages comes from the Slavic-Baltic contact area (see Koptjevskaja Tamm and Wälchli 2001; Wiemer 2003, 2004, 2009; Wiemer et al. 2014). For example, both Slavic and Baltic have verbal prefixes expressing various spatial and non-spatial modifications of the verbal lexical meaning, as well as Aktionsart and perfectivity. Most of these prefixes are cognate, but instances of mat-borrowing are also attested. In some cases, the Slavic and Baltic prefixes show non-trivial regular phonological correspondences, which makes their status as cognates seem indisputable, cf. Russian na- vs. Lithuanian nu- or Russian v(o)z- vs. Lithuanian už-. In the event that formally close morphemes violate the independently established phonological correspondences, borrowing is the only way to account for the presence of a prefix both in Slavic and Baltic, as in the case of Russian razand Lithuanian dialectal raz- (regular sound change would predict Lithuanian *arž-). However, regular phonological development can produce (nearly) homonymous forms, such as Russian pri- and Lithuanian pri- or Russian po- ([pa-] in unstressed position) and Lithuanian pa- (both pairs of prefixes are considered cognates), or Russian do- ([da-] when unstressed) and Lithuanian da-.6 When this happens, cognates cannot be distinguished from potential borrowings on formal grounds only, and more intricate methods should be invoked in order to determine the origin of a particular morpheme; for instance, the distribution

5 As a word of warning, the methods of contact linguistics and historical linguistics should not only be applied soundly but also complement each other. The long-held belief that morphology is the most reliable basis for genealogical classification of languages (the so-called “Ludolf’s rule”) may lead to wrong analyses, for example, if borrowings are not recognized as such (see Grant 2008: 166). 6  Lithuanian linguists regard the Lithuanian da- as a borrowing from Slavic and therefore exclude it from the standard language; this matter is, however, not uncontroversial, see, e.g. Kozhanov (2014) for an assessment.

16 

 Francesco Gardani, Peter Arkadiev and Nino Amiridze

of its type frequency with respect to the geographical areas with more or less contact between the languages in question and the degree of similarity between the semantic profiles (polysemy patterns) of morphemes in these languages (see Kozhanov 2014 on the application of these methods to the abovementioned prefix da- in Lithuanian). In our volume, the problem of “copy vs. cognate” in morphology is addressed in the articles authored by Martine Robbeets on denominal verbalizers in the Transeurasian languages and by Françoise Rose on pronominal paradigms in Arawakan. Both papers provide detailed empirical analyses and arrive at negative conclusions concerning the possibility that morphological borrowing has occurred. Robbeets investigates “suspect” derivational markers in the languages traditionally known as “Altaic” (Turkic, Mongolic, Tungus-Manchu, Korean, and Japanese) and tests four phonologically similar, denominal verbalizers against a set of criteria designed to determine whether bound morphemes are borrowings or cognates. The diagnostics of borrowing include, among others, the forms’ restriction to bases shared between the languages in question (which are thus possibly lexical borrowing themselves); unilateral morphological complexity of morphemes; morphological, phonological, and functional mismatches; and, limitation of the shared morpheme’s distribution to contact zones. Based on a wealth of data, the author concludes that the correspondences and similarities between the studied forms indicate that they are inherited rather than borrowed. Robbeets’s article is not only rich in empirical evidence but, crucially, provides linguistics with a whole array of criteria to apply to other disputed cases in the historical-comparative research. By contrast, Rose investigates just one language, Mojeño, and portrays an innovation occurred in its rich pronominal paradigm, in which a distinction based on the gender of the speaker has developed. This distinction is not attested in Proto-Arawak, Old Mojeño, or elsewhere in the Arawak family. Other innovations in Mojeño are the introduction of a non-human third person category and the development of a non-specified third person category. Rose considers all possible genesis explanations, both contact-induced and internally motivated, and shows that none of the potential candidates as SLs of Mojeño’s innovative pronominal forms stands the test. She concludes that there is not sufficient evidence for a borrowing scenario. Even this negative result is, in our view, extremely valuable in terms of methodology, because it makes explicit that the postulation of morphological borrowing requires an extremely accurate examination of the existing linguistic and socio-historical facts, and that the burden of proof for a borrowing hypothesis may sometimes be insurmountable.



Borrowed morphology: an overview  

 17

6 Conclusion This introductory chapter portrays the current trends of research in contactinduced morphological change and shows how far the articles published therein advance the field. As we have seen, the volume enriches the research landscape in terms of: i. empirical evidence, by providing data on morphological borrowing from a large number of languages; ii. methodology, by explicitly addressing cases that are hard to attribute to internal or external causation, or even to the result of multiple causation, and by providing instruments to test the probability of borrowing rather than genealogical inheritance; iii. theory, by showcasing what types of morphology are borrowed in the languages of the world and how different degrees of borrowability are explainable, owing to different approaches to language contact, both structurally, functionally, and sociolinguistically inspired. The ten chapters published in the present volume are thematically divided into three parts. In Section I – Theory – two articles (by Thomason and Matras) discuss the ease of borrowability of morphology. The other three sections of the book are defined by the theoretical issues focused upon in the respective articles. Section II (Coghill; Bağrıaçık, Ralli and Melissaropoulou; and Robbeets) deals with the borrowing of derivational morphology in terms of mat-borrowing, and Section III discusses the borrowing of inflectional morphology, in terms of both mat-borrowing (Pakendorf; Meakins; Clements and Luís; Rose) and pat-borrowing (Stolz).

Acknowledgments The editors strongly believe in the advantages of serious selection procedures. Ten papers, out of fourteen submitted, have been selected for publication in this book. The review procedure was three-fold, consisting for each submitted paper of a blind peer review by external experts, an internal review by the fellow authors, and the editors’ review. We wish to thank Evangelia Adamou, Mark Aronoff, Walter Breu, Petra Eccarius, Lenore Grenoble, Eitan Grossman, Ilja Gruntov, Martin Haspelmath, Brian Joseph, Petros Karatsareas, Yury Lander, Marianne Mithun, Pieter Muysken, Stefan Schnell, Eva Schultze-Berndt, Florian Siegl, Donald Winford, Fernando Zúñiga, Inge Zwitserlood, as well as all contributors to the volume for their invaluable help. Thanks are due to Yaron Matras for reviewing the whole manuscript. Usual disclaimers apply.

18 

 Francesco Gardani, Peter Arkadiev and Nino Amiridze

Abbreviations in glosses 1 2 3 ag dem erg inf lvm nmlz pl prs prv pst sg suff top

first person second person third person agentive demonstrative ergative infinitive loanverb marker nominalizer plural present perfective past singular suffix topic

References Ameka, Felix K. 2006. Grammars in contact in the Volta Basin (West Africa): On contact-induced grammatical change in Likpe. In Alexandra Y. Aikhenvald & Robert M. W. Dixon (eds.), Grammars in contact: A cross-linguistic typology, 114–142. Oxford: Oxford University Press. Aronoff, Mark. 1994. Morphology by itself: Stems and inflectional classes. Cambridge, Mass.: MIT Press. Atanasov, Petar. 1990. Le mégléno-roumain de nos jours: Une approche linguistique (BalkanArchiv, Beiheft 7). Hamburg: Buske. Bauer, Laurie. 2004. The function of word-formation and the inflection-derivation distinction. In Henk Aertsen, Mike Hannay & Rod Lyall (eds.), Words in their places: A festschrift for J. Lachlan Mackenzie, 283–292. Amsterdam: Vrije Universiteit. Behrens, Leila. 1996. Lexical rules cross-cutting inflection and derivation. Acta Linguistica Hungarica 43. 33–65. Boas, Franz. 1917. Introductory. International Journal of American Linguistics 1(1). 1–8. Boas, Franz. 1920. The classification of American languages. American Anthropologist 22(4). 367–376. Booij, Geert E. 1994. Against split morphology. In Geert E. Booij & Jaap van Marle (eds.), Yearbook of morphology 1993, 27–50. Dordrecht: Kluwer. Booij, Geert E. 1996. Inherent versus contextual inflection and the split morphology hypothesis. In Geert E. Booij & Jaap van Marle (eds.), Yearbook of morphology 1995, 1–16. Dordrecht: Kluwer. Booij, Geert E. 2005. Compounding and derivation: Evidence for Construction Morphology. In Wolfgang U. Dressler, Dieter Kastovsky, Oskar Pfeiffer & Franz Rainer (eds.), Morphology



Borrowed morphology: an overview  

 19

and its demarcations: Selected papers from the 11th Morphology Meeting, Vienna, February 2004, 109–132. Amsterdam & Philadelphia: John Benjamins. Booij, Geert E. 2010. Construction morphology. Oxford & New York: Oxford University Press. Bowern, Claire. 2013. Relatedness as a factor in language contact. Journal of Language Contact 6(2). 411–432. Breu, Walter. 1991. Abweichungen vom phonetischen Prinzip bei der Integration von Lehnwörtern. In Klaus Hartenstein & Helmut Jachnow (eds.), Slavistische Linguistik 1990: Referate des XVI. Konstanzer Slavistischen Arbeitstreffens Bochum/Löllinghausen 19.-21.9.1990, 36–69. München: Otto Sagner. Bybee, Joan L. 1985. Morphology: A study of the relation between meaning and form. Amsterdam & Philadelphia: John Benjamins. Capidan, Theodor. 1925. Meglenoromânii. I, Istoria şi graiul lor (Academia Română. Studii şi Cercetări). Bucarest: Cultura Naţională. Carlin, Eithne B. 2006. Feeling the need: The borrowing of Cariban functional categories into Mawayana (Arawak). In Alexandra Y. Aikhenvald & Robert M. W. Dixon (eds.), Grammars in contact: A cross-linguistic typology, 313–332. Oxford: Oxford University Press. Croft, William. 2003. Mixed languages and acts of identity: An evolutionary approach. In Yaron Matras & Peter Bakker (eds.), The mixed language debate: Theoretical and empirical advances, 41–72. Berlin & New York: Mouton de Gruyter. Donohue, Mark. 2012. Typology and areality. Language Dynamics and Change 2(1). 98–116. Donohue, Mark. 2013. Towards a Papuan history of languages. Journal of the Linguistics Society of Papua New Guinea 31(1). 24–41. Dressler, Wolfgang U. 1989. Prototypical differences between inflection and derivation. Zeitschrift für Phonetik, Sprachwissenschaft und Kommunikationsforschung 42(1). 3–10. Epps, Patience. 2013. Inheritance, calquing, or independent innovation? Reconstructing morphological complexity in Amazonian numerals. Journal of Language Contact 6(2). 329–357. Field, Fredric W. 2002. Linguistic borrowing in bilingual contexts. Amsterdam & Philadelphia: John Benjamins. Friedman, Victor A. 2012. Copying and cognates in the Balkan Sprachbund. In Lars Johanson & Martine I. Robbeets (eds.), Copies versus cognates in bound morphology, 323–336. Leiden & Boston: Brill. Gardani, Francesco. 2008. Borrowing of inflectional morphemes in language contact. Frankfurt am Main: Peter Lang. Gardani, Francesco. 2012. Plural across inflection and derivation, fusion and agglutination. In Lars Johanson & Martine I. Robbeets (eds.), Copies versus cognates in bound morphology, 71–97. Leiden & Boston: Brill. Gardani, Francesco. In press. Morphology and contact-induced language change. In Anthony Grant (ed.), The Oxford handbook of language contact. Oxford: Oxford University Press. Gast, Volker & van der Auwera, Johan. 2012. What is ‘contact-induced grammaticalization’? Examples from Mayan and Mixe-Zoquean languages. In Björn Wiemer, Bernhard Wälchli & Björn Hansen (eds.), Grammatical replication and borrowability in language contact, 381–426. Berlin & Boston: De Gruyter Mouton. Golovko, Evgenij V. & Nikolai B. Vakhtin. 1990. Aleut in contact: The CIA enigma. Acta linguistica Hafniensia 22(1). 97–125. Grandi, Nicola. 2002. Morfologie in contatto: Le costruzioni valutative nelle lingue del Mediterraneo. Milano: Franco Angeli.

20 

 Francesco Gardani, Peter Arkadiev and Nino Amiridze

Grant, Anthony. 2008. Contact-induced change and the openness of ‘closed’ morphological systems: Some cases from Native America. Journal of Language Contact 2(1). 165–186. Hajek, John. 2006. Language contact and convergence in East Timor: The case of Tetun Dili. In Alexandra Y. Aikhenvald & Robert M. W. Dixon (eds.), Grammars in contact: A crosslinguistic typology, 163–178. Oxford: Oxford University Press. Haspelmath, Martin. 2013. An inflection/derivation distinction on the other side of the globe? http://dlc.hypotheses.org/388. Haugen, Einar. 1950. The analysis of linguistic borrowing. Language 26(2). 210–231. Heine, Bernd. 2012. On polysemy copying and grammaticalization in language contact. In Claudine Chamoreau & Isabelle Léglise (eds.), Dynamics of contact-induced language change, 125–166. Berlin & Boston: De Gruyter Mouton. Heine, Bernd & Tania Kuteva. 2003. On contact-induced grammaticalization. Studies in Language 27(3). 529–572. Heine, Bernd & Tania Kuteva. 2005. Language contact and grammatical change. Cambridge: Cambridge University Press. Heine, Bernd & Tania Kuteva. 2006. The changing languages of Europe. Oxford: Oxford University Press. Jakobson, Roman. 1938. Sur la théorie des affinités phonologiques entre les langues. In Actes du quatrième Congrès International de Linguistes, tenu à Copenhague du 27 août au 1er septembre 1936, 48–59. Copenhagen: Einar Munskgaard. Jendraschek, Gerd. 2006. Basque in contact with Romance languages. In Alexandra Y. Aikhenvald & Robert M. W. Dixon (eds.), Grammars in contact: A cross-linguistic typology, 143–162. Oxford: Oxford University Press. Johanson, Lars. 1999. The dynamics of code-copying in language encounters. In Bernt Brendemoen, Elizabeth Lanza & Else Ryen (eds.), Language encounters across time and space: Studies in language contact, 37–62. Oslo: Novus forlag. Johanson, Lars. 2008. Remodeling grammar: Copying, conventionalization, grammaticalization. In Peter Siemund & Noemi Kintana (eds.), Language contact and contact languages, 61–79. Amsterdam & Philadelphia: John Benjamins. Johanson, Lars & Martine I. Robbeets (eds.) (2012). Copies versus cognates in bound morphology. Leiden & Boston: Brill. Karatsareas, Petros. 2009. The loss of grammatical gender in Cappadocian Greek. Transactions of the Philological Society 107(2). 196–230. Karatsareas, Petros. 2014. On the diachrony of gender in Asia Minor Greek: The development of semantic agreement in Pontic. Language Sciences 43. 77–101. Koptjevskaja Tamm, Maria & Bernhard Wälchli. 2001. The Circum-Baltic languages: An arealtypological approach. In Östen Dahl & Maria Koptjevskaja Tamm (eds.), The Circum-Baltic languages. Typology and contact, vol. 2: Grammar and typology, 615–748. Amsterdam & Philadelphia: John Benjamins. Kozhanov, Kirill. 2014. Priešdėlio da- semantika lietuvių kalboje. In Artūras Jūdžentis, Tatjana V. Civjan & Marija V. Zavjalova (eds.), Baltai ir slavai: dvasinių kultūrų sankirtos, 254–274. Vilnius: Versmė. Laca, Brenda. 2001. Derivation. In Martin Haspelmath, Ekkehard König, Wulf Oesterreicher & Wolfgang Raible (eds.), Language typology and language universals, vol. 1, 1214–1227. Berlin: Walter de Gruyter. Law, Danny. 2013. Inherited similarity and contact-induced change in Mayan Languages. Journal of Language Contact 6(2). 271–299.



Borrowed morphology: an overview  

 21

Lefebvre, Claire. 2008. Relabelling: A major process in language contact. Journal of Language Contact 2(1). 91–111. Maiden, Martin, John C. Smith, Maria Goldbach & Marc-Olivier Hinzelin. 2011. Introduction: The concept of autonomous morphology. Background and rationale. In Martin Maiden, John C. Smith, Maria Goldbach & Marc-Olivier Hinzelin (eds.), Morphological autonomy. Perspectives from Romance inflectional morphology, 1–9. Oxford: Oxford University Press. Matras, Yaron. 2007. The borrowability of structural categories. In Yaron Matras & Jeanette Sakel (eds.), Grammatical borrowing in cross-linguistic perspective, 31–73. Berlin & New York: Mouton de Gruyter. Matras, Yaron. 2009. Language contact. Cambridge: Cambridge University Press. Matras, Yaron. 2012. An activity-oriented approach to contact-induced language change. In Claudine Chamoreau & Isabelle Léglise (eds.), Dynamics of contact-induced language change, 17–52. Berlin & Boston: De Gruyter Mouton. Matras, Yaron & Peter Bakker. 2003. The study of mixed languages. In Yaron Matras & Peter Bakker (eds.), The mixed language debate: Theoretical and empirical advances, 1–20. Berlin & New York: Mouton de Gruyter. Matras, Yaron & Jeanette Sakel (eds.) (2007a). Grammatical borrowing in cross-linguistic perspective. Berlin & New York: Mouton de Gruyter. Matras, Yaron & Jeanette Sakel. 2007b. Introduction. In Yaron Matras & Jeanette Sakel (eds.), Grammatical borrowing in cross-linguistic perspective, 1–13. Berlin & New York: Mouton de Gruyter. McConvell, Patrick. 1988. MIX-IM-UP: Aboriginal code-switching, old and new. In Monica Heller (ed.), Codeswitching: Anthropological and sociolinguistic perspectives, 97–150. Berlin: Mouton de Gruyter. McConvell, Patrick & Felicity Meakins. 2005. Gurindji Kriol: A mixed language emerges from code-switching. Australian Journal of Linguistics 25(1). 9–30. Meakins, Felicity. 2011a. Borrowing contextual inflection: evidence from northern Australia. Morphology 21(1). 57–87. Meakins, Felicity. 2011b. Case-marking in contact: The development and function of case morphology in Gurindji Kriol. Amsterdam & Philadelphia: John Benjamins. Meakins, Felicity. 2013. Mixed languages. In Peter Bakker & Yaron Matras (eds.), Contact languages: A comprehensive guide, 159–228. Boston & Berlin: De Gruyter Mouton. Meillet, Antoine. 1921. Linguistique historique et linguistique générale. Paris: Champion. Menovščikov, G. A. 1969. Aleutskij jazyk. In P. J. Skorik (ed.), Jazyki narodov SSSR, vol. 5: Mongol’skie, tunguso-man’č žurskie i paleoaziatskie jazyki, 386–406. Leningrad: Nauka. Mithun, Marianne. 2012. Morphologies in contact: Form, meaning, and use in the grammar of reference. In Martine Vanhove, Thomas Stolz, Aina Urdze & Hitomi Otsuka (eds.), Morphologies in contact, 15–36. Berlin: Akademie Verlag. Mithun, Marianne. 2013. Challenges and benefits of contact among relatives: Morphological copying. Journal of Language Contact 6(2). 243–270. Moravcsik, Edith A. 1978. Language contact. In Joseph H. Greenberg, Charles A. Ferguson & Edith A. Moravcsik (eds.), Universals of human language, vol. 1: Method & theory, 93–122. Stanford: Stanford University Press. Müller, Max. 1862. Lectures on the science of language, delivered at the Royal Institution of Great Britain in May, April, and June 1861, 2nd edn. New York: Scribner. Muysken, Pieter. 2012. Root/affix asymmetries in contact and transfer: Case studies from the Andes. International Journal of Bilingualism 16(1). 22–36.

22 

 Francesco Gardani, Peter Arkadiev and Nino Amiridze

Myers-Scotton, Carol. 2002. Contact linguistics: Bilingual encounters and grammatical outcomes. Oxford & New York: Oxford University Press. Myers-Scotton, Carol. 2006. Multiple voices: An introduction to bilingualism. Malden, Mass.: Blackwell Publishing. Petrović Rignault, Marijana. 2008. DO-: étude d’un préfixe verbal en valaque. In Biljana Sikimić & Tijana Ašić (eds.), The Romance Balkans: Collection of papers presented at the international conference The Romance Balkans, 4–6 November 2006, 261–288. Belgrade: Serbian Academy of Sciences and Arts, Institute for Balkan Studies. Plank, Frans. 1994. Inflection and derivation. In Robert E. Asher (ed.), The encyclopedia of language and linguistics, vol. 3, 1671–1678. Oxford: Pergamon Press. Robbeets, Martine. 2012. Shared verb morphology in the Transeurasian languages: Copy or cognate? In Lars Johanson & Martine I. Robbeets (eds.), Copies versus cognates in bound morphology, 427–446. Leiden & Boston: Brill. Rohde, Ada, Anatol Stefanowitsch & Suzanne Kemmer. 1999. Loanwords in usage-based model. In Sabrina J. Billings, John P. Boyle & Aaron M. Griffith (eds.), Chigaco Linguistic Society 35-1: Papers from the main session, 265–275. Chicago: Chicago Linguistic Society. Ross, Malcolm. 1999. Exploring metatypy: How does contact-induced typological change come about? Keynote paper presented at the Annual meeting of the Australian Linguistic Society, Perth, 1999. Ross, Malcolm & Mark Durie. 1996. Introduction. In Mark Durie & Malcolm Ross (eds.), The comparative method reviewed: Regularity and irregularity in language change, 3–38. New York & Oxford: Oxford University Press. Rusakov, Alexander Y. 2001. The North Russian Romani dialect. Interference and code-switching. In Östen Dahl & Maria Koptjevskaja Tamm (eds.), The Circum-Baltic languages, vol. 1: Past and present, 313–337. Amsterdam & Philadelphia: John Benjamins. Sakel, Jeanette. 2007. Types of loan: Matter and pattern. In Yaron Matras & Jeanette Sakel (eds.), Grammatical borrowing in cross-linguistic perspective, 15–29. Berlin & New York: Mouton de Gruyter. Sapir, Edward. 1921. Language: An introduction to the study of speech. New York: Harcourt, Brace and company. Schrammel, Barbara. 2002. The borrowing and calquing of verbal prefixes and particles in Romani dialects in contact with Slavic and German. Graz: University of Graz MA thesis. Schuchardt, Hugo. 1884. Dem Herrn Franz von Miklosich zum 20. Nov. 1883. Slawo-Deutsches und Slawo-Italienisches. Graz: Leuschner & Lubensky. Seifart, Frank. 2013. AfBo: A world-wide survey of affix borrowing. Leipzig: Max Planck Institute for Evolutionary Anthropology. Available online at http://afbo.info. Swadesh, Morris. 1951. Diffusional cumulation and archaic residue as historical explanations. Southwestern Journal of Anthropology 7(1). 1–21. Thomason, Sarah G. 1997. Mednyj Aleut. In Sarah G. Thomason (ed.), Contact languages: A wider perspective, 449–468. Amsterdam & Philadelphia: John Benjamins. Thomason, Sarah G. 2001. Language contact: An introduction. Edinburgh & Washington, DC: Edinburgh University Press and Georgetown University Press. Thomason, Sarah G. & Terrence Kaufman. 1988. Language contact, creolization, and genetic linguistics. Berkeley: University of California Press. Trubetzkoy, Nicolai S. 1939. Gedanken über das indogermanenproblem. Acta Linguistica 1(1). 81–89. Urioste, Jorge L. 1964. Transcripciones quechuas I–VII. La Paz: Instituto de Cultura Indígena.



Borrowed morphology: an overview  

 23

Vanhove, Martine, Thomas Stolz, Aina Urdze & Hitomi Otsuka (eds.) (2012). Morphologies in contact. Berlin: Akademie Verlag. Weinreich, Uriel. 1953. Languages in contact: Findings and problems. New York: Linguistic Circle of New York. Whaley, Lindsay. 2012. Deriving insights about Tungusic classification from derivational morphology. In Lars Johanson & Martine I. Robbeets (eds.), Copies versus cognates in bound morphology, 395–409. Leiden & Boston: Brill. Whitney, William D. 1881. On mixture in language. Transactions of the American Philological Association 12. 5–26. Wiemer, Björn. 2003. Dialect and language contacts on the territory of the Grand Duchy of Lithuania from the 15th century until 1939. In Kurt Braunmüller & Gisella Ferraresi (eds.), Aspects of multilingualism in European language history, 105–143. Amsterdam & Philadelphia: John Benjamins. Wiemer, Björn. 2004. Population linguistics on a micro-scale. Lessons to be learnt from Baltic and Slavic dialects in contact. In Bernd Kortmann (ed.), Dialectology meets typology: Dialect grammar from a cross-linguistic perspective. Berlin: Mouton de Gruyter. Wiemer, Björn. 2009. Zu entlehnten Präfixen und anderen morphosyntaktischen Slavismen in litauischen Insel- und Grenzmundarten. In Lenka Scholze & Björn Wiemer (eds.), Von Zuständen, Dynamik und Veränderung bei Pygmäen und Giganten: Festschrift für Walter Breu zu seinem 60. Geburtstag, 347–390. Bochum: Brockmeyer. Wiemer, Björn, Ilja Seržant & Aksana Erker. 2014. Convergence in the Baltic-Slavic contact zone: Triangulation approach. In Juliane Besters-Dilger, Cynthia Dermarkar, Stefan Pfänder & Achim Rabus (eds.), Congruence in contact-induced language change: Language families, typological resemblance, and perceived similarity, 15–42. Berlin: De Gruyter. Wiemer, Björn, Bernhard Wälchli & Björn Hansen (eds.) (2012). Grammatical replication and borrowability in language contact. Berlin & Boston: De Gruyter Mouton. Winford, Donald. 2003. An introduction to contact linguistics. Malden, Mass.: Blackwell. Wohlgemuth, Jan. 2009. A typology of verbal borrowings. Berlin & New York: Mouton de Gruyter.

Part I Theory

Sarah G. Thomason

When is the diffusion of inflectional morphology not dispreferred? 1 Introduction The common consensus among historical linguists has always been that morphology – in particular inflectional morphology – is the grammatical subsystem least likely to be affected by language contact. The most popular explanation for this fact has been that foreign elements cannot easily make their way into the inflectional morphology because its tightly interconnected paradigmatic structures form a barrier (see Thomason and Kaufman 1988: 14–15 for discussion and references). As with so many generalizations in historical linguistics, this one needs some shading when it is confronted with the evidence from a wide range of contact situations. In this paper I argue that there is no global dispreference for morphological diffusion. In certain types of contact situations, even inflectional morphology passes readily from one language to another. I do not mean to suggest that inflectional morphology is transferred as frequently as other structural features and lexicon; it isn’t. My goal, instead, is to show that the diffusion of inflectional features is considerably more common than one might guess from the general language-contact literature. The structure of the paper is as follows. Section 2 has two purposes: first, it covers preliminary issues that set the stage for the discussions to follow; and second, it character­izes the kinds of contact situations in which morphological transfer is most likely to occur. Section 3 focuses on borrowing situations with low levels of typological distance, Section 4 explores morphological transfer in instances of shift-induced interference, and in Section 5 we will consider cases involving deliberate morphological change. Section 6 is a brief summing-up conclusion.

2 A few preliminaries Several distinctions must be made in order to address the issue of morphological diffusion between languages. Most importantly, a distinction needs to be drawn between two quite different kinds of intense contact situations and their linguistic outcomes (as outlined in Thomason and Kaufman 1988: chapter 3), one is morphological borrowing in a narrow sense of “borrowing”, where bilingual speakers transfer grammatical morphemes and/or morpho­logical patterns from

28 

 Sarah G. Thomason

one of their languages into the other; the other type of diffusion is shift-induced interference, where (thanks to imperfect learning) shifting speakers introduce morphological features from their original L1, their heritage language, into the target lan­guage, the language they are shifting to. In this paper I will use the term “borrowing” only in the narrow sense. By including both of these types of contact-induced change, I depart from the definition of “morphological borrowing” adopted by Gardani (2012), which excludes shift-induced interference. The distinction between interference via borrowing and interference via shift correlates robustly with a difference in the kinds of linguistic features transferred from one language to the other. In borrowing, the predominant interference features are words, with structural features lagging behind and confined to intense (rather than casual) contact situations. In shift-induced interference, by contrast, the predominant interference features are structural, specifically phonological and syntactic; there may be few transferred lexical items in a shift situation. (Although these correlations are robust, they are not exceptionless. In some cultures, for instance, lexical borrowing is considered inappropriate; when such cultures come into intense contact with other cultures, therefore, the linguistic outcome might involve structural borrowing without lexical b ­ orrowing – for examples, see e.g. Kroskrity 1993 and Aikhenvald 2002.) The reason this distinction is important in the present context is that contactinduced morphological changes are most likely to occur in two distinct kinds of situations: first, cases of borrowing (in my narrow sense) in which the languages in contact are very closely related, perhaps even dialects of the same language; and second, cases of shift-induced interference in which the shifting speakers’ version of the target language displays numerous learners’ “errors”. In the borrowing situations, the typological distance between the source language and the receiving language, at least at the relevant structure point(s), is minimal. As we will see in Section 4 below, typological distance appears to be less important in shift situations than in borrowing situations. It is also important to note that borrowing is more likely than shift-induced interference to include the transfer of morphemes in addition to structure. In shift-induced interference it is often (though by no means always) the case that only structures are transferred, so that the innovative structures are expressed by native morphemes in the receiving language. In this paper I consider both of these types of morphological interference – that is, with and without accompanying transfer of morphemes (for discussion of this distinction, see Heath 1978 on direct vs. indirect diffusion, e.g. p. 125, followed by later authors, e.g. Thomason and Kaufman 1988, e.g. p. 32; Matras and Sakel 2007). This is again in contradistinction to the approach adopted in Gardani 2012, where only morphological borrowing that includes transferred morphemes is considered. My main



When is the diffusion of inflectional morphology not dispreferred?  

 29

reason for being more inclusive is that general claims about the possibilities and probabilities of morphological interference can (in my opinion) contribute to our overall understanding of processes and results of contact-induced change only if they encompass the entire range of contact situations. Another reason for including shift situations and structure-only contact-induced changes is that in many or most shift situations, borrowing and shift-induced interference occur simultaneously, mediated by different agents; and it is not always possible to determine which process(es) has/have produced a given innovation. An example is the Chantyal-Nepali contact situation described by Noonan (2008: 84–85), in which Chantyal speakers used Nepali regularly and were far outnumbered by Nepali speakers, but in which many speakers of Nepali (and several other languages) learned Chantyal because of its speakers’ advantageous economic position. The sociolinguistic complexities of this situation most likely promoted both borrowing and shift-induced interference, including the transfer of morphemes such as a comparative, a benefactive “for”, and a comitative (Noonan 2008: 95). Obviously, given the relevance of typological distance in morphological borrowing and its probable relevance in shift-induced morphological interference, we must also distinguish between pairs of languages in contact according to typological congruence: contact-induced morphological changes, especially in the inflectional morphology and especially in borrowing situations, occur much more frequently at typologically congruent structure points than at typologically disparate structure points. (The general argument about typological congruence facilitating structural borrowing was made as early as Weinreich 1953 and Heath 1978; the same argument has also been made by numerous later authors, e.g. Thomason and Kaufman 1988. The greater importance of typological congruence in borrowing situations as opposed to shift situations is discussed in more detail in Thomason 2014). The notion of typological congruence, as I am using it here, is not necessarily self-evident, so I will be explicit about two crucial aspects of the phenomenon: it is gradient, and it is structure-specific. A structure in one language may be more or less similar typologically to a functionally comparable structure in another language; the more similar two morphological structures are, the more likely they are to be the target of morphological transfer. Assessing the degree of typological congruence in a given instance is often not a straightforward either/or decision. In efforts to explain the linguistic outcomes of language contact, it is not useful to compare (for instance) morphosyntactic systems as wholes: the fact that languages A and B have more or less completely dissimilar verbal systems, for instance, does not make it less likely that cases can diffuse from A to B if the languages have overlapping categories in noun declension. Although typological distance is relevant for both borrowing and shift situations, its role is clearer in borrowing than in shift-induced interference. Given sufficiently

30 

 Sarah G. Thomason

intense contact – which, in shift situations, usually means that shifting speakers greatly outnumber target-language speakers – morphological innovations in the shifting group’s version of the target language can include typologically novel features. This can also happen even in borrowing situations, but it is relatively rare (see Thomason 2001b for discussion). Another relevant distinction concerns the processes, or mechanisms, of contact-induced change. When morphology is borrowed, the mechanisms may (but need not) differ from the mechanisms in play in shift-induced morphological transfer. Borrowing of morphology between closely-related systems may proceed by passive familiarity, whereby speakers who speak only one of the varieties in contact adopt features from a variety they understand but do not speak. This won’t happen in shift-induced interference, because by definition shifting speakers do not have full familiarity with the target language. In shift-induced interference, the dominant mechanism will almost certainly be second-language acquisition strategies, usually combined with negotiation (for discussion of these and other mechanisms of interference, see Thomason 2001a: chapter 6). Finally, in all analyses of contact effects we must consider social factors of various kinds, because social factors provide both motivation and opportunity for contact-induced changes of all kinds, including morphological interference. The importance of social factors is most obvious in the distinction between borrowing and shift-induced interference, which is ultimately a sociolinguistic dichotomy; most strikingly, however, it is seen in instances of deliberate language change, where linguistic factors such as typological congruence between source language and receiving language appear to play no role in determining the linguistic results of contact (see Section 5 below). As with internally-motivated language change, we cannot hope to arrive at a deterministic predictive theory of contact-induced change: there are too many social and linguistic variables that can affect the fate of any given innovation.

3 Borrowing, little typological distance It is especially easy to find cases of morphological transfer in contact situations involving closely-related languages or dialects of the same language. The main linguistic reason for this surely lies in the fact that the morphological systems in such contacts tend to be typologically congruent to a very considerable degree. The main sociolinguistic reason, of course, is that closely-related languages/ dialects are also fairly likely to be in close contact. In the typical case, a language expands to the point where constant communication across the entire



When is the diffusion of inflectional morphology not dispreferred?  

 31

speech community is reduced so that dialects diverge, and eventually those dialects split into separate languages, all the while remaining in more or less close contact. Other causes of language split – out-migration of a subsection of a speech community to a distant place and in-migration of another language whose new territory divides the original speech community in two – are less likely to result in close contact between the diverging dialects; but those events are also less common than dialect divergence in contiguous (or continuous) territories. Geographical closeness, given other appropriate sociolinguistic circumstances, makes morphological transfer in closely-related varieties possible and sometimes even likely. By contrast, even intense contact between languages that are very different morphologically is much less likely to result in morphological transfer. One set of examples of morphological borrowing between closely-related varieties is found in contacts between East Slavic languages, for instance Surzhyk, a mixture of Russian and Ukrainian (Grenoble 2010: 592–593). According to Grenoble, Surzhyk has Ukrainian as a matrix language with inserted Russian features. Russian intrusions into Ukrainian inflectional morphology include the genitive singular of masculine nouns (usually -a as in Russian rather than -u as in Ukrainian), the dative singular of masculine animate nouns (typically Russian -u rather than Ukrainian ‑ovi), and a first person plural imperative formed on the Russian model. These features are of course accompanied by other Russian borrowings, especially loanwords and lexical calques and also syntactic calques. Most of the morphological bor­rowings from Russian noted by Grenoble involve morphemes transferred into the Ukrainian matrix, but in one instance – the 1pl imperative – apparently only the Russian structural pattern was transferred, not the Russian morpheme. A second East Slavic mixed variety re­ported by Grenoble (2010: 593–594) is Trasjanka, which combines Russian and Belarusian and “typically has Belarusian phonetics and intonation, a mixed morphology and mixed lexicon” (p. 594). Probably the most common cases of morphological transfer affect dialects of the same language, in particular standard dialect influence on nonstandard dialects. One example is in the noun declensional system of the Croatian dialect of the island Hvar, which came under the influence of Standard Serbo-Croatian (Hraste 1935: 22). In the early 1930s, at least some older speakers on Hvar still had the original dialect system of oblique plural declension in masculine nouns of the old o-stem class, in which the genitive-locative suffix ‑ih differed from suffixes in the other oblique plural cases (dative and instrumental); this system contrasted with that of Standard Serbo-Croatian, in which the genitive plural had one suffix (‑ā) and the other three oblique plural cases shared another suffix, ‑ima. Younger speakers of the Hvar dialect, who (unlike their elders) had

32 

 Sarah G. Thomason

presumably been systematically exposed to the standard dialect, had replaced the older Hvar system with the standard configuration of oblique plural cases: Before: Hvar

gen/loc.pl ‑ih vs. dat/inst.pl - ?

After:  Hvar

gen.pl

‑ih vs. dat/inst/loc.pl

-ima

Cf. Stand. Serbo-Croatian: gen.pl

‑ā vs. dat/inst/loc.pl

-ima

As in Surzhyk, the borrowing strategies are heterogeneous: younger Hvar speakers keep the original Hvar dialect suffix ‑ih but confine it to the genitive plural, thus borrowing only the function of the standard dialect genitive plural ‑ā; but they have borrowed both the form and the function of the dative-instrumental-locative plural suffix ‑ima from the standard dialect. (The suffix -ima in the dative and instrumental plural was probably borrowed before the genitive plural and locative plural split in Hvar; Hraste does not provide earlier forms for these two cases, listing only the suffix ‑ima and its occasional variant -iman.) These examples of morphological borrowing between closely-related systems are typical of contact effects in situations with little or no typological distance between the systems in contact: the categories and functions all match closely. Not surprisingly, studies of ongoing dialect mixing contain comparable examples. For instance, Peter Trudgill, in Dialects in contact (1986), cites a study that includes morphological accommodation of a variety of Swedish spoken in Norway to the surrounding Norwegian – a contact situation in which the two languages, though they are officially separate, “have a very high degree of mutual intelligibility” (1986: 24). Among the study’s examples of morphological accommodation are two inflectional features of adjectives, both of which appeared more or less randomly (rather than systematically) in the speech of twenty-two Swedish women living in Bergen, Norway: first, the women sometimes adopted the Norwegian pattern of neuter adjectival inflection, namely, a zero suffix, replacing the neuter -t of Swedish; and second, plural adjectives sometimes received a suffix -e only in attributive position, not (as in Swedish) in predicate position as well (Trudgill 1986: 26–27, citing Nordenstam 1979). As Trudgill emphasizes, Nordenstam’s variable results do not point to a systematic change in the Bergen variety of Swedish, in part because the accommodation features show no overall tendency toward accommodation to the Swedish patterns. The features do, however, illustrate a part of a (potential) process by which morphological borrowing can occur when the systems in contact are very similar. Inflectional borrowing involving languages that are not closely (if at all) related is not as common as inflectional borrowing between closely-related languages/dialects, but it isn’t difficult to find examples. In the great majority of cases, the source and receiving languages match rather closely, in a typological



When is the diffusion of inflectional morphology not dispreferred?  

 33

sense, at the structure point where morphological transfer occurs. The Iranian language Ossetic and the Kartvelian language Georgian, for instance, share several noun cases that neither language inherited from an ancestor (Belyaev 2010); the innovative cases are apparently expressed by native morphemes in each language – certainly they are not formally similar (Belyaev 2010: 310–311) – but the systems as wholes show significant convergence. That is, this is an instance of the transfer of inflectional patterns but not of inflectional morphemes. Belyaev, following such scholars as Abaev (1949) and Axvlediani (e.g. 1960), argues for the existence of centuries-long intensive contact between the two languages, with mutual bilingualism (2010: 317). He has no evidence about the direction of influence; it may well have been mutual (see Lindstedt 2000 and Thomason 2001a: 142–146 for discussion of processes of mutual contact-induced change via “negotiation”). The new cases increase the number of cases in (at least) the Ossetic declensional system, but since Ossetic already had two or more cases, there is little typological change – especially in view of the fact that the native Ossetic cases, like the borrowed cases, are expressed by suffixes. Johanson (e.g. 2010: 666) cites several instances of Turkic influence on nonTurkic languages and vice versa: the Uralic language Mari (formerly called Cheremis) has borrowed case suffixes from Turkic; some varieties of the Indo-European languages Tajik (Iranian) and Anatolian Greek have borrowed – or copied, in Johanson’s terminology – case and person-number markers; some Iranian Turkic varieties also have a suffix -i borrowed from Persian, where it is “considered to mark Persian nominals for “indefiniteness”” (Johanson 1998: 329); Iranian Turkic varieties have a comparative suffix -tar borrowed from Persian (Johanson 1998: 329), while Tajik has a comparative suffix borrowed from Turkic (Johanson 2010: 666). Johanson argues that the frequency of suffix copying from Turkic into other languages “may be due to the agglutinative nature, the low level of fusion, of Turkic morphology” (2010: 266). In all these examples, the languages in contact are not typologically congruent throughout their structures, but the inflectional categories that are affected are structurally and functionally similar. Both Mari and Turkic have rich case systems expressed by suffixes, both Turkic and Greek verbs have person-number suffixes, and the comparative is formed by suffixes in both Tajik and Turkic. It’s true that Greek and Iranian have fusional inflection while Turkic inflection is agglutinative, but clearly that difference is not sufficient to block morphological transfer. In recent decades Anatolian Greek has been the focus of intensive research, following on the famous 1916 study by R. M. Dawkins, and additional examples of inflectional borrowing from Turkish have been identified, especially in Cappadocian dialects of Greek. Hovdhau­gen (1976: 146) says that “Most Cappadocian dialects have got a set of suffixed possessive pronouns which closely resembles

34 

 Sarah G. Thomason

the system of [suffixed] possessive pronouns in Turkish”; according to Janse, this system is restricted to South Cappadocian dialects, where “the Turk­ish pattern is generalized” to all persons and both numbers (2004: 16). Another inflectional feature, the differential marking of definite and indefinite objects, is also widely attributed to Turkish influence (e.g. by Janse 2004: 16); in Turkish, the accusative suffix is added only to definite or specific objects (Janse 2004: 4, citing Kornfilt 1997: 214). All authors agree that, as Janse puts it, the “Cappadocian agglutinative declension is quite unique among the Greek dialects” (2004: 21), and that this pattern is borrowed from Turkish. Agglutinative patterns also appear in certain verbal paradigms in some varieties, and at least one Cappadocian dialect, Semenderé, has added 1pl and 2pl Turkish suffixes to present and imperfect Greek verb forms that already have 1pl and 2pl Greek suffixes (Janse 2009: 104–105) – that is, these forms are doubly marked for these two person/number combinations (see Thomason 1988 for discussion of double marking in contact-induced morphological change). Here again, although there is considerable typological distance between Turkish and Greek, the particular categories that are targeted by these instances of morphological transfer are shared by the two languages, and their positions in the word are also shared. In the doubly-marked 1pl and 2pl verb forms, for example, the Turkish suffix is simply added after the isofunctional Greek suffix. Also as in the previous examples in this section, these examples of Turkic contacts show heterogeneous borrowing strategies: sometimes morphemes and structure are borrowed and sometimes only structure is borrowed. There does not appear to be a preference for one strategy over another. In all the examples presented in this section, morphological transfer occurs in intense contact situations where imperfect learning plays no role. Among other things, intensity of contact in a borrowing situation (as op­posed to a shift situation) involves widespread, though not necessarily universal, bilingualism among borrowing-language speakers. The explanations of contact-induced syntactic conver­gence that propose a (subconscious) motive in lightening the cognitive burden of using two languages (e.g. Silva-Corvalán 1994) would apply to instances of morphological borrowing as well.

4 Shift-induced interference We have already seen several examples of borrowing in case systems; examples of case transfer are fairly easy to find in shift situations as well. Lithuanian (Indo-European, Baltic subbranch) acquired three new cases – an illative, an allative, and an adessive (and perhaps also, more controversially, a fourth case, an inessive) – as a result of interference from shifting Finnic speakers. Only one



When is the diffusion of inflectional morphology not dispreferred?  

 35

of the three uncontroversial new cases, the illative, remains in modern Standard ­Lithuanian, and then only in restricted usage; but all three are well attested from the 16th and 17th centuries, and some southeastern dialects of Lithuanian retain all three cases (Thomason and Kaufman 1988: 242–243, citing Senn 1966: 92 and Fairbanks 1977: 117). Two of the three cases are expressed with suffixes derived from a Lithuanian postposition (Peter Arkadiev, p.c. 2013); the origin of the third suffix, the illative, is controversial (Senn 1966; Koptjevskaja-Tamm and Wälchli 2001: 671–672; Seržant 2004); it may be a borrowed morpheme, but recent work suggests that the morpheme is inherited (Seržant 2004). Both Finnic languages (Uralic family) and Lithuanian have case systems expressed by suffixes. Finnic has much richer systems, with case proliferation primarily in the locative category. Adding a few more locative cases to Lithuanian, with its single inherited locative case, caused no significant typological change. A somewhat similar example is found in Russian noun declension, notably in the so-called second genitive. The Russian second genitive is a partitive construction expressed by a suffix ‑u in nouns belonging to the major masculine/ neuter noun declension, the old Indo-European o-stems. An example is the noun phrase čaška čaj-u ‘a cup of tea’ (i.e. some tea), which contrasts with the general genitive of o-stem nouns, which has the suffix ‑a, as in e.g. cena čaj-a ‘the price of tea’. This partitive construction, like the innovative Lithuanian cases, is surely due to the influence of shifting Finnic speakers. The suffix ‑u is native to Russian: it is the genitive singular suffix of the old u-stem declension. Presumably Finnic speakers learning Russian, when confronted with two different genitive singular endings that were functionally identical, reinterpreted them as functionally distinct – specifically, as expressing a genitive vs. partitive distinction that was native to the shifting speakers’ native Finnic language(s). In this instance the typological distance might be considered slightly greater than with the new Lithuanian cases. Subdividing a locative case category seems to be a less dramatic change than creating an entirely new case in the partitive. Shift-induced morphological interference from Uralic in Slavic and Baltic languages is not confined to the case systems. Northern Russian dialects, for example, have developed postpositive definite articles, probably under the influence of Komi-Zyrian definite-possessive suffixes (Grenoble 2010: 584, citing Tiraspol’skij 1998). Other examples listed by Grenoble (2010: 584) are the transfer of the Finnish causative suffix ‑(i)tta from Finnish; the comparative of nouns, e.g. berežee ‘closer to the shore’ (from bereg ‘shore’); and the nominative object (used instead of the accusative in certain constructions, as discussed in Thomason and Kaufman 1988: 242, citing Timberlake 1974: 220 and other sources). The latter two examples are due to the influence of Uralic languages that cannot be identified (perhaps because all their speakers shifted to Russian). The origin(s) of

36 

 Sarah G. Thomason

the Russian nominative object construction, in particular, has/have been much debated for decades, and there may still be no consensus on this matter; but a convincing recent detailed analysis of their status concludes as follows: “Whatever their initial origin may be, their geographically distributed peculiarities ... must be taken as a result of prolonged linguistic contacts between the Russian and Finnic (Karelian and Ludian) populations (see especially Sarhimaa 1992)” (Koptjevskaja-Tamm and Wälchli 2001: 663). All the changes mentioned in this paragraph, with the possible exception of the nominative object, cannot be said to be typologically trivial. Because most nouns in Russian and other Slavic languages do have identical nominative and accusative singular forms, innovating a nomi­native object in declension classes that do distinguish the two cases brings those declensions partly into line with declension classes that don’t. But Slavic languages did not inherit either a suffixed causative formation or morphological comparison in nouns, so these two innovations do mark a typological departure from the pre-existing structures. Another set of examples of shift-induced interference in the context of limited typologi­cal distance is found in the Ethiopian linguistic area, where speakers of Cushitic languages shifted in large enough numbers to the languages of Semitic-speaking newcomers that they left significant traces of their original languages in their new languages (see Thomason 2001a: 111–113, 127 for discussion of the Ethiopian linguistic area and sources on its features). Cushitic and Semitic languages are related, but distantly: they form two branches of the AfroAsiatic language family. Unsurprisingly, the two groups display both typological similarities and sharp ty­pological distinctions. Two shared morphological categories are the negative perfect con­struction and the causative construction. In Cushitic languages, which are solidly SOV with morpheme-ordering patterns typical of SOV languages, both of these constructions are marked by suffixes – a single negative perfect suffix and a doubled causative suffix. In Semitic languages each of these constructions is expressed by a single prefix, as one would expect in a VSO language: Proto-Semitic is reconstructed with VSO word order (e.g. Kaufman 1974: 132), together with the typical complex of morpheme ordering features, and the earliest attested Semitic language of Ethiopia, Ge’ez, was VSO. (Modern Arabic dialects, however, are SVO.) Modern Ethiopic Semitic languages are SOV, thanks no doubt to the influence of shifting Cushitic speakers. Both the negative perfect and the causative have undergone shift-induced interference in Ethiopic Semitic, though with rather different results. The negative perfect is expressed by an ambifix: the original Semitic prefix is copied after the verb stem, yielding an affix1‑verb­-affix1 construction. The doubled suffix that expresses the Cushitic causative is mirrored by doubling of the original Semitic prefix, yielding an Ethiopic Semitic causative construction prefix1-prefix1-verb.



When is the diffusion of inflectional morphology not dispreferred?  

 37

No morpheme transfer occurs in either construction; only the Cushitic pattern is copied, and it is copied only partially. The categories themselves match in Cushitic and Semitic, but the affix ordering differs sharply between the two groups. These two changes are therefore not typologically trivial; but at least they do not involve new categories. Most of the examples of shift-induced morphological interference that we have examined so far cause little if any typological disruption in the receiving language’s morphological system. As with morphological borrowing, most of the innovative features fit fairly well into the target language’s morphology – new cases in an already rich case system, new ways of expressing pre-existing morphological categories, and so forth. But we have also seen examples of shift-induced morphological diffusion that creates new categories in the receiving language. It is not surprising that this should occur in a case of group language shift where shifting speakers greatly outnumber target-language speakers. The shifting speakers have relatively few native-speaker models as they develop their version of the target language (TL), and as a result they both carry over heritagelanguage structures into the TL and fail to learn certain TL structures. Examples of both phenomena can be seen in the innovative compar­ative noun construction in northern Russian dialects and in the loss of grammatical gender in Latvian (a Baltic language) under the influence of shifting Livonian (Uralic) speakers (Comrie 1981: 147; see Koptjevskaja-Tamm and Wälchli 2001 for a more detailed discussion of gender loss in the area). Other shift situations present a larger number of striking examples of morphological innovations that would be disfavored if typological closeness were a requisite for shift-induced morphological interference. In the Indic language Shina, for instance, a number of morphological innovations have arisen under the influence of shifting Burushaski speakers (Thomason and Kaufman 1988: 137–138, citing Lorimer 1937). Among Lorimer’s examples are a new singulative category (expressed by a native Shina morpheme), a new ablative case suffix (also using native Shina material), a new reciprocal formation that reduplicates the Shina word for “one”, a vigesimal numeral system (instead of the usual Indic decimal systems), and the use of a suffix ‑a at the end of a verb to mark certain ques­tions. All of these, and other examples as well, represent novel morphological categories in Shina, and together they comprise evidence of quite significant typological change. Cases like this make me hesitate to assert confidently that typological congruence is a strong factor in facilitating morphological innovations in shift situations; it seems likely that intensity of contact is considerably more important here. This contrasts with borrowing situations: it is much easier to find examples of morphological borrowing between typologically congruent systems. It is easy to find examples of morphological interference between

38 

 Sarah G. Thomason

typologically con­gruent structures in shift situations too, but it isn’t a great deal harder to find shift-induced morphological changes that change the typology of the receiving language’s morphology.

5 Deliberate morphological change Traditionally, historical linguists have assumed that the vast majority of linguistic changes are subconscious and inexorable, not under the control of either individual speakers or a speech community (see Thomason 2007 for references). Certainly it is true that we have no evidence for any element of consciousness or deliberation in the vast majority of changes, including contact-induced changes. For this reason I am skeptical of suggestions like Matras’s that “[w]e become, as speakers, semi-conscious of the fact that we have replicated not just a word, but also a morphological procedure, once we borrow another word that utilises the same procedure ...” (2009: 212). Such statements seem overly speculative. Nevertheless, there are well-established examples of contact-induced (and other) changes that appear to have happened as a result of speakers’ conscious decision to change their language; and some of these changes affect the inflectional morphology. The Estonian language reformer Johannes Aavik introduced many innovations into his language, and a sizable number of these were adopted by his countrymen and became part of the language. Most of the new features cannot be traced to influence from other languages, but others – among them a synthetic superlative construction inspired by the German su­perlative – are indeed contact-induced changes. As Oksaar observed, Aavik’s successful innovations “are proof that arbitrarily coined new derivational and inflectional morphemes and new grammemes—such as the synthetic superlative—can be wholly accepted by the language users and ... incorporated into the language” (1972: 491). It is well known that some contact-induced changes increase the distance between two languages (or dialects) rather than shrinking it. Some distancing changes must be deliberate, in part because they happen too fast to be attributed to ordinary gradual language change. Typically the motivation seems to be to establish, or emphasize, the group’s identity as distinct from that of other groups. Perhaps the most spectacular example that involves the morphology is the gender reversal reported for the Uisai dialect of the Buin language that is spoken on Bougainville Island, Papua New Guinea. According to Laycock (1982), all nouns that are feminine in Uisai are masculine in other dialects of Buin and vice versa, so that all the elaborate gender agreement markers are reversed in Uisai sentences. Now, this Uisai change, and other examples of deliberate morphological distancing



When is the diffusion of inflectional morphology not dispreferred?  

 39

changes, don’t contribute to our understanding of morphological borrowing, much less shift-induced morphological transfer. They do, however, show that inflectional morphology can be manipulated deliberately by speakers, and that speakers sometimes do deliberately change morphological structure in intense contact situations. A more relevant deliberate morphological change – a potential change in this case, not one that actually occurred – is the deliberate shift from polysynthetic to analytic verb structure that was made by just one Montana Salish speaker on just one occasion. This shift illustrates a type of convergent change that is well known to fieldworkers engaged in primary language documentation. It is impossible to know how often accommodations of this type lead to change in an entire language, but it is certain that bilingual speakers can and do sometimes modify their speech deliberately to make their L1 utterances more similar to L2 structure. This particular example comes from a single speaker of a moribund language on a single occasion and is surely not a sign of incipient change – not even in this speaker’s L1. (In my view, even a single speaker’s one-time innovation is a potential language change, though of course most one-time innovations never become fixed in the speech of one speaker, much less in a whole speech community.) The example is from Montana Salish, the easternmost member of the Southern Interior branch of the 23-language Salishan family (Washington, Oregon, British Columbia, and extreme western Montana). It was provided by an elderly fluent speaker of the language during an elicitation session that focused on ditransitive sentences. I asked for translations of sentences like ‘Johnny stole a (deer) hide from Mary’ and repeatedly received Salish translations like (1): (1) Čoní naqw’ t q’ett tl’ Malí Johnny steal obl hide from Mary ‘Johnny stole a hide from Mary.’ Other than the particle and the lexicon, this sentence is very close to the English version: SVO word order, the verb in an intransitive bare root form, and a preposition marking the indirect object. The particle t ‘oblique’ obligatorily marks the patient of an intransitive verb form or the agent of a transitive verb form, and it has no English translation. But this is very far from what would be expected as a natural Salish sentence, which in a neutral context would have VOS word order and a fully inflected transitive verb. Agents are normally sentence-initial only if they receive special pragmatic emphasis, and an intransitive verb with a patient (much less with both a patient and an “indirect object”) is typically used only in particular limited discourse contexts. In isolation, these sentences were bizarre. When I finally asked my consultant why he was giving me such sentences, he was surprised; he said he thought I wanted English-like translations, since I was

40 

 Sarah G. Thomason

asking in English. So I asked how he would normally say these things, and he at once started providing much more natural sentences, as in (2), ‘Johnny stole Mary’s hide’: (2) Naqw’-m-ł-t-s Malí q’ett-s t steal­-der.trans-rel-trans-he Mary hide-3sg.poss obl Čoní Johnny ‘Johnny stole Mary’s hide.’ The sentence in (2) has the expected ditransitive verb form (der.trans = ‘derived transitive’, a stem formative; rel = ‘relational, second object’) and VOS word order. In offering examples like (1), this elder was deliberately accommodating his speech to English structure, producing Salish sentences which, though fully grammatical, were extremely strange outside the appropriate context. In the present context, the most significant element in the Montana Salish sentential calque in (1) is of course the verb form. It’s not that anything in the “Englishy” sentences is entirely new in Montana Salish; as noted above, all the “Englishy” sentences are grammat­ical Montana Salish utterances. They are deviant only in a pragmatic/discourse sense. But if we project from this one day’s examples to a future state in which such sentences have become the norm, the morphological change in the direction of English word structure would have to be considered profound. From that perspective, what this speaker did in constructing translations was to adjust his speech so that it was as close as possible to English structure. His calques certainly comprised an importation of English word (and sentence) structure into Montana Salish. This was not an established change even in this one speaker’s speech; but the fact that he could and did make the adjustment is a strong indi­cator that such changes are possible, if social factors make speakers want to move in that direction. It is comparable, in fact, to the convergent syntactic changes that Silva-Corvalán found in the SpanishEnglish contact situation she investigated (1994), although the morphological consequences of the potential Montana Salish change are much more profound than anything Silva-Corvalán found in her data. The final type of deliberate morphological change is what we find in bilingual mixed languages. These languages arise in intense contact situations and are created by fluent bilinguals, whether adults or children – maybe usually adults, although the emerging mixed language Light Warlpiri is being created by young children who receive code-switching input from older members of their community (O’Shannessy 2012). Because so few bilingual mixed languages have been discovered, and because even fewer have been analyzed in depth, no solid



When is the diffusion of inflectional morphology not dispreferred?  

 41

generalizations can be drawn about the nature of such languages. In particular, it may be difficult or even impossible to assign primary or secondary roles to the two languages that contribute to the structure of a bilingual mixed language (though this is not true of all bilingual mixed languages; Meakins 2011a, b has shown that the main matrix language in the code-switching stage that preceded the genesis of Gurindji Kriol was Kriol). And if we want to discuss morphological transfer, there has to be a language into which foreign morphological material has been transferred – a matrix language that adopts material from another language. This means that any doubt about which language is the matrix language precludes interpreting any of the specific morphological structures as borrowings. (They would have to be borrowings rather than shiftinduced interference, because all the available evidence suggests that bilingual mixed languages arise in highly bilingual environments, not in shift situations.) Mednyj Aleut is a case in point. Spoken on Mednyj (Copper) Island, one of the two Commander Islands off the far eastern coast of Russia, the language comprises primarily Aleut structure, including the nominal system (though with some borrowings from Russian) and the nonfinite verb inflection; but the entire finite verb inflection is Russian. As this description suggests, some analysts (e.g. Thomason 1997) consider Aleut to be the matrix language, with Russian finite verb inflection replacing the original Aleut finite verbal inflec­tional system. Other analysts, however, coming from a theoretical perspective that views (only finite?) verb inflection as the core of a language’s grammar, consider Russian to be the matrix language, in which case all the Aleut material might be best treated as borrowed into a Russian matrix. (Myers-Scotton [2002: 258–265] proposes, for instance, that Mednyj Aleut reflects a “Matrix Language turnover”, with Russian as the matrix language after the turnover.) On either of these views, however, it seems to me to be odd to treat the combining of such large chunks of two languages’ grammatical structures as a case of morphological transfer in either direction. An element of deliberation was almost certainly involved in the creation of Mednyj Aleut, both because of the short time period that saw its emergence and because of the social and linguistic factors that set it apart from all other known languages, including other bilingual mixed languages. As with the elicited Montana Salish sentences that followed the English model closely, the creation of Mednyj Aleut shows that speakers’ deliberate modifications of their speech in a contact situation depend not on their ability to make such changes but on the social context. I do not claim that deliberate morphological changes are in general a routine or common type of change. I do claim that speakers sometimes like to make deliberate morphological changes, just as they often like to invent new words. Admittedly speakers act on the latter inclination far more often than on the former.

42 

 Sarah G. Thomason

6 Conclusion The main argument of this paper is that transferring morphology from one language to another is not somehow weird in language contact situations, or even always unexpected. As with other contact phenomena, the likelihood of morphological transfer depends on a complex mix of linguistic and social factors. There are various linguistic factors that make morphological transfer less common than other types of structural interference, including the traditional one of the tightlyintegrated, hard-to-infiltrate nature of inflectional paradigms as well as typological mismatch and such proposed cognitive barriers as the relative difficulty of separating out a deeply embedded morpheme from a polymorphemic word. As far as social factors are concerned, there is no essential sociolinguistic difference between transferring morphology and transferring any other structural elements: the likelihood of transfer depends on how intense the contact situation is – the level of bilingualism in borrowing situations, the availability of native-speaker models in shift situations, and other things that factor into intensity of contact. The social (and, to a lesser degree, linguistic) factors that promote morphological interference are the same as those that promote syntactic and phonological interference; the difference is a matter of degree, not of kind. The most common situations in which inflectional morphology is transferred are intense borrowing contexts where the typologies of source and receiving language match closely. Within this general category, closely-related languages and dialects of the same language borrow morphology more freely than distantlyrelated or unrelated languages do, though there are many examples of the latter in situations where relevant parts of the morphology do match closely. The role of typology is less clear in shift situations, although here too it seems rather likely that typologically congruent morphological features introduced by shifting speakers are more likely than typologically mismatched features to be adopted by original target-language speakers and thus to become a permanent part of the modified target language. In the data I have examined, I find no significant sociolinguistic or structural difference between instances of morpheme transfer and instances in which only morphological patterns, but not the morphemes themselves, are transferred. Both types occur both in borrowing situations and in shift situations, though pattern-only transfer is more common than morpheme transfer in shift situations. In agreement with Gardani (2008: 88), I also find no evidence that a gap-filling motive plays any role in morphological transfer; more often, morphological transfer either rearranges the functions and/or the expression of existing categories (as with the plural cases in Hvar Croatian and the Ethiopic Semitic causative) or introduces new items within an existing category (as with



When is the diffusion of inflectional morphology not dispreferred?  

 43

new cases added to already existing case systems in various borrowing and shift situations). Someone might be tempted to propose that creating an entirely new morphological category through contact-induced change must fill a gap in the receiving language’s morphological system, but such a proposal would stretch the imagination a bit too far in a case like the Burushaski-inspired singulative in Shina. Finally, various authors have argued that a change in the frequency of a particular structure is not a change at all. Heine, for instance, represents this view both in his paper’s title, “Contact-induced word order change without word order change”, and in the opening sentence of its conclusion (2008: 56): In the survey data discussed in this paper we have not come across a single case where speakers really produced a word order entirely alien to the language concerned – hence, there is not really a change from one word order to a new order.

Authors who espouse this view do not explain why a change in frequency should not be considered a change in the language. Presumably what they have in mind is that language change must produce something entirely new to the language. This would mean that a change like the replacement of polysynthetic Montana Salish verbal structure by a completely analytic verbal structure, if that one speaker’s innovations were to become the community’s norm, would not count as a change in the language, in spite of the fact that its effect on the language’s morphosyntactic typology would be sweeping. The elevation of a highly-marked construction that occurs relatively rarely in naturally-occurring speech to the main (or only) verb construction in the language surely deserves a prominent status. By the reasoning of the frequency-change-is-not-­language-change school of thought, a sound change that devoiced all word-final obstruents would not count as a sound change at all if the language already had some voiceless obstruents (as all languages do). Unless someone can explain why a change in frequency and context should be real change in the phonology but not in the morphosyntax, I will continue to believe that a change in frequency is change. This paper will have achieved its goal if readers are convinced that morphological transfer is not always surprising and, when it does occur, not particularly exotic. Bilingual speakers readily transfer morphological patterns and morphemes from one of their languages into the other at structure points where the two languages are typologically congruent; shifting speakers fairly often transfer morphological features from their heritage language into their target language. And in cases of deliberate change, changes affect complex morphology without apparent typological barriers: it is vital not to underestimate speakers’ ability to manipulate linguistic structures.

44 

 Sarah G. Thomason

Acknowledgments I am very grateful to Felicity Meakins, Francesco Gardani, Peter Arkadiev, and an anonymous reviewer for many helpful comments on a previous draft of this paper; their suggestions have led to considerable improvement in the final draft. Any remaining errors or infelicities are of course my responsibility.

Abbreviations 3 sg dat der gen inst loc obl pl poss ptl rel trans

third person singular dative derived genitive instrumental locative oblique plural possessive particle relational, second object transitive

References Abaev, Vasiliy I. 1949. Osstinskij jazyk i fol’klor. Moscow-Leningrad. Aikhenvald, Alexandra Y. 2002. Language contact in Amazonia. Oxford: Oxford University Press. Axvlediani, G. S. 1960. Sbornik izbrannyx rabot po osstinskomu jazyku. Tbilisi. Belyaev, Oleg. 2010. Evolution of case in Ossetic. Iran and the Caucasus 14. 287–322. Comrie, Bernard. 1981. The languages of the Soviet Union. Cambridge: Cambridge University Press. Dawkins, R. M. 1916. Modern Greek in Asia Minor: A study of the dialects of Sílli, Cappadocia and Phárasa with grammar, texts, translations and glossary. Cambridge: Cambridge University Press. Fairbanks, Gordon. 1977. Case inflections in Indo-European. Journal of Indo-European Studies 5. 101–131. Gardani, Francesco. 2008. Borrowing of inflectional morphemes in language contact. Frankfurt am Main: Peter Lang.



When is the diffusion of inflectional morphology not dispreferred?  

 45

Gardani, Francesco. 2012. Plural across inflection and derivation, fusion and agglutination. In Lars Johanson & Martine Robbeets (eds.), Copies versus cognates in bound morphology, 71–97. Leiden & Boston: Brill. Grenoble, Lenore A. 2010. Contact and the development of Slavic languages. In Raymond Hickey (ed.), The handbook of language contact, 581–597. Oxford: Wiley-Blackwell. Heath, Jeffrey. 1978. Linguistic diffusion in Arnhem Land. Canberra: Australian Institute of Aboriginal Studies. Heine, Bernd. 2008. Contact-induced word order change without word order change. In Peter Siemund & Noemi Kintana (eds.), Language contact and contact languages, 33–60. Amsterdam & Philadelphia: John Benjamins. Hovdhaugen, Even. 1976. Some aspects of language contact in Anatolia. University of Oslo Working Papers in Linguistics 7. 142–160. Hraste, Mate. 1935. Čakavski dijalekat ostrva Hvara. Južnoslovenski Filolog 14. 1–59. Janse, Mark. 2004. Animacy, definiteness, and case in Cappadocian and other Asia Minor Greek dialects. Journal of Greek linguistics 5. 3–26. Janse, Mark. 2009. Watkins’ Law and the development of agglutinative inflections in Asia Minor Greek. Journal of Greek Linguistics 9. 93–109. Johanson, Lars. 1998. Code-copying in Irano-Turkic. Language Sciences 20. 325–337. Johanson, Lars. 2010. Turkic language contacts. In Raymond Hickey (ed.), The handbook of language contact, 652–672. Oxford: Wiley-Blackwell. Kaufman, Stephen A. 1974. The Akkadian influences on Aramaic. Chicago: University of Chicago Press. Koptjevskaja-Tamm, Maria & Bernhard Wälchli. 2001. The circum-Baltic languages: An arealtypological approach. In Östen Dahl & Maria Koptjevskaja-Tamm (eds.), The circum-Baltic languages: Typology and contact, vol. 2: Typology and grammar, 615–748. Amsterdam & Philadelphia: John Benjamins. Kornfilt, Jaklin. 1997. Turkish. London: Routledge. Kroskrity, Paul. 1993. Language, history, and identity: Ethnolinguistic studies of the Arizona Tewa. Tucson: University of Arizona Press. Laycock, Donald C. 1982. Melanesian linguistic diversity: A Melanesian choice? In R.J. May & Hank Nelson (eds.), Melanesia: beyond diversity, 33–38. Canberra: Australian National University Press. Lindstedt, Jouko. 2000. Linguistic Balkanization: Contact-induced change by mutual reinforcement. In Dicky Gilbers, John A. Nerbonne, J. Schaeken & Tjeerd d. Graaf (eds.), Languages in contact, 231–246. Amsterdam: Rodopi. Lorimer, D. L. R. 1937. Burushaski and its alien neighbors: Problems in linguistic contagion. Transactions of the Philological Society 36. 63–98. Matras, Yaron. 2009. Language contact. Cambridge: Cambridge University Press. Matras, Yaron & Jeanette Sakel. 2007. Investigating the mechanisms of pattern replication in language convergence. Studies in Language 31. 829–865. Meakins, Felicity. 2011a. Case marking in contact: The development and function of case morphology in Gurindji Kriol. Amsterdam & Philadelphia: John Benjamins. Meakins, Felicity. 2011b. Borrowing contextual inflection: Evidence from northern Australia. Morphology 21. 57–87. Myers-Scotton, Carol. 2002. Contact linguistics: bilingual encounters and grammatical outcomes. Oxford: Oxford University Press.

46 

 Sarah G. Thomason

Noonan, Michael. 2008. Contact-induced change: The case of the Tamangic languages. In Peter Siemund & Noemi Kintana (eds.), Language contact and contact languages, 81–106. Amsterdam & Philadelphia: John Benjamins. Nordenstam, Kerstin. 1979 Svenskan i Norge. Gothenberg: University Press. Oksaar, Els. 1972. Bilingualism. In Thomas Albert Sebeok (ed.), Current trends in linguistics, vol. 9: Linguistics in Western Europe, 476–511. The Hague: Mouton. O’Shannessy, Carmel. 2012. The role of code-switched input to children in the origin of a new mixed language. Linguistics 50. 305–340. Sarhimaa, Anneli. 1992. Karelian Sprachbund? Theoretical basis of the study of Russian/ Baltic-Finnic contacts. Finnisch-Ugrische Forschungen 50. 209–219. Senn, Alfred. 1966. Handbuch der litauischen Sprache. Heidelberg: Carl Winter. Seržant, Ilja A. 2004. Einige Bemerkungen zur Geschichte des Illativs. Baltu Filoloǵija 13. 113–121. Silva-Corvalán, Carmen. 1994. Language contact and change: Spanish in Los Angeles. Oxford: Oxford University Press. Thomason, Sarah G. 1988. Double marking in morphological change. In Ann Miller & Joyce Powers (eds.), ESCOL 87: Proceedings of the Fourth Eastern States Conference on Linguistics, 296–305. Columbus: The Ohio State University. Thomason, Sarah G. 1997. Mednyj Aleut. In Sarah G. Thomason (ed.), Contact languages: A wider perspective, 449–468. Amsterdam & Philadelphia: John Benjamins. Thomason, Sarah G. 2001a. Language contact: an introduction. Edinburgh & Washington, DC: Edinburgh University Press and Georgetown University Press. Thomason, Sarah G. 2001b. Contact-induced typological change. In Martin Haspelmath, Ekkehard König, Wulf Oesterreicher & Wolfgang Raible (eds.), Language typology and language universals, Sprachtypologie und sprachliche Universalien: An international handbook, 1640–1648. Berlin & New York: Walter de Gruyter. Thomason, Sarah G. 2007. Language contact and deliberate change. Journal of Language Contact 1. 41–62. Thomason, Sarah G. 2014. Contact-induced language change and typological congruence. In Juliane Besters-Dilger, Cynthia Dermarkar, Stefan Pfänder & Achim Rabus (eds.), Congruence in contact-induced language change: Language families, typological resemblance, and perceived similarity, 201–218. Berlin: De Gruyter. Thomason, Sarah G. & Terrence Kaufman. 1988. Language contact, creolization, and genetic linguistics. Berkeley: University of California Press. Timberlake, Alan. 1974. The nominative object in Slavic, Baltic, and West Finnic. Munich: Verlag Otto Sagner. Tiraspol’skij, Gennadij I. 1998. Finno-ugorskoe vlijanie na russkij jazyk kak prognostičeskij material [Finno-Ugric influence on the Russian language as prognostic material]. Obščie problemy prepodavanija jazykov: tezisy meždunarodnoj naučno-metodičeskoj konferencii MAPRJaL. 60–62. Trudgill, Peter. 1986. Dialects in contact. Oxford: Blackwell. Verstraete, Jean-Christophe. 2012. Contact-induced restructuring of pronominal morphosyntax in Umpithamu (Cape York Peninsula, Australia). Diachronica 29. 326–358. Weinreich, Uriel. 1953. Languages in contact. The Hague: Mouton.

Yaron Matras

Why is the borrowing of inflectional morphology dispreferred? 1 Introduction It has often been pointed out that bound morphology and especially inflectional morphology is more rarely borrowed than unbound morphemes, be they lexical or grammatical. Among the explanations offered is the “paradigmaticity” of bound morphemes (van Hout and Muysken 1994) as well as their more abstract semantic value (Moravcsik 1978; Field 2002) and the assumption that speakers are somehow less conscious of morphology (Mithun 2012: 15). While these traits may well be characteristic of structural material that is less prone to borrowing, they do not offer a direct explanation as to why they should make morphemes more resistant to change in language contact situations. In order to address this question we must firstly establish what it is that motivates grammatical borrowing in the first place. I will therefore begin by sketching the main principles of a user-oriented theory of borrowing (Section 2) that is outlined in more detail in Matras (2009). I then review the borrowing of various categories of bound morphology and attempt to link borrowing patterns with speakers’ motivation to borrow. I understand “motivation to borrow” as bilingual speakers’ motivation to blur the demarcation boundaries between different portions of their overall repertoire of linguistic structures and to generalise a form or structure across this repertoire as a whole and use it irrespective of setting, addressee, topic and so on. This – control over the multilingual repertoire – is, I propose, at the heart of the process that may ultimately lead to contact-induced language change. The choices that bilinguals make in using their repertoire of structures are ultimately responsible for the possible spread of elements from one speech community into another, which is what we identify in historical-diachronic perspective as “borrowing”. Below I cite some evidence that allows us to identify several types of morphological borrowing. The first involves replication of inflectional morphology patterns without actually any replication of formal structures. This kind of morphological convergence is illustrative of the efforts that speakers make in multilingual constellations to actually avoid the replication of bound inflectional morphemes from one language in another (Section 3). Next I look briefly at commonly attested morphological borrowing involving derivational morphology and the marking of nominal plurals (Section 4). I then examine what

48 

 Yaron Matras

I call “Morphological Analogy” – a process by which inflectional morphemes are replicated from a donor language on the basis of some kind of similarity that they show to native morphemes that are functionally related (Section 5). The final type involves what I call “Morphological Compartmentalization” (Section 6). Here, inflectional morphology is replicated along with lexical word forms from another language in situations in which speakers embrace and flag a bilingual identity. These are exceptional processes, which are very much confined, as far as we can tell, to a small number of communities in which a rather radical process of identity re-negotiation is or has been underway. They are the exceptions that prove the rule. Even in cases of morphological compartmentalization, the role of “borrowed” inflectional morphology is, I argue, to authenticate the replication of lexical word forms and phrases from a donor or contact language. Diffusion of inflectional morphology into inherited or native lexical material is usually blocked here too. Cases of morphological compartmentalisation are sometimes cited as examples that there are no constraints on borrowing and that “anything goes”. However, even these cases show that bilingual speakers are in fact quite uneasy about lifting the demarcation boundaries among languages in their repertoire when it comes to inflectional morphology. As a result we can conclude that the borrowing of inflectional morphology is certainly dis-preferred. The reasons for this are primarily, in my view, not structural, but functional. Inflectional morphology serves as the carrier for initiating and anchoring the predication (on the verb and its auxiliaries) and for identifying the predicate’s argument structure (on the noun and its attributes). In a multilingual communication setting, the choice of “language” amounts to the choice of structures used to anchor the predication and its arguments (verb-inflectional morphology such as person, tense, modality and aspect). The choice of, e.g. lexical material, modifiers, or prosody is less crucial in this respect, and so these are more easily “transferrable” from one language to another; or, in a formulation that is consistent with the functional, usage-based model followed here, they are more easily generalised by bilinguals across their repertoire of structures irrespective of setting, context, or addressee. The predication and its argument structure may therefore function as the last resort through which speakers identify and flag language choice. It allows speakers to maintain some kind of mental boundary between “languages” and consequently a social boundary between sets of interaction contexts – which is what the demarcation of languages represents. This boundary is only compromised in exceptional circumstances, where a speaker community embraces its diverse (plurilingual) repertoire of structures, forms and interaction settings as one single whole, and where language “choices” therefore become less instrumental in negotiating social identity.



Why is the borrowing of inflectional morphology dispreferred?  

 49

2 Toward a usage-based model of borrowing Descriptive and historical linguistics have traditionally viewed “contact” as a useful metaphor through which to depict the circumstances under which languages change by absorbing influences from other languages. Although Weinreich (1953), the pioneer of language contact studies, had remarked that the true locus of language contact is the bilingual individual, most contemporary research in the field is based on the assumption that linguistic “systems” influence one another in contact situations. Consequently, constraints are sought in the shape and nature of systems in order to explain or predict borrowing patterns (cf. Moravcsik 1978; Field 2002). Thomason and Kaufman’s (1988) well-cited borrowing scale emphasises the relevance of the intensity of cultural contacts to the level of structural borrowing, re-focusing on the socio-cultural dimension as a necessary angle in the assessment of contact. Nevertheless, the scale lacks an explanation as to why some categories – function words, for example – should be easier to borrow through shallower cultural contacts, while others – such as word order – require prolonged and more intense contact. I propose an alternative approach to language contact (see Matras 2009 for more detail). It is based first of all on a view of language as the practice of communicative interaction and of grammatical categories as triggers and operators of language processing tasks that are involved in communication. According to this approach, the selection of structures by a speaker is not arbitrary, but directly derived from the linguistic task-schema that the speaker wishes to carry out. This, in turn, is subordinated to the goal-oriented activity that the speaker pursues by means of verbal communication, organized at the level of discourse. Such a view of language is compatible with a wide range of theoretical approaches to communication and discourse (e.g. Sacks, Schegloff, and Jefferson 1974; Rehbein 1977; Gumperz 1980; Ehlich 2007) as well as to speech production (e.g. Green 1998; Paradis 2004). Next, the model followed here presupposes that “borrowing”, even in the less controversial sense of the term (for discussions of the codeswitchingborrowing distinction see e.g. Myers-Scotton 1993, 2002; Backus 1996; Muysken 2000; for the codeswitching-borrowing continuum see Matras 2009: 110—114), begins in situations in which speakers of a language must communicate in certain contexts in which their own “native” language does not constitute a fully adequate means of communication. In order to do this, speakers must extend their repertoire of linguistic structures, even if only in a rudimentary way. Becoming “bilingual” is therefore, even in the most basic sense, an extension of an individual’s settings of communicative interaction, and as a result, an extension of that individual’s repertoire of communicative structures. This

50 

 Yaron Matras

does not necessarily mean that borrowing is always deliberate or conscious, but it does mean that it is purposeful and functional. Becoming bilingual from infancy means early exposure to a complex repertoire. This requires gradually sorting out the sets of contexts and contextual conditions under which various sets of structures from within this repertoire are considered appropriate. Thus, even bilinguals-from-birth do not acquire two language “systems”. Rather, they acquire a repertoire of linguistic structures and forms and are left to gradually master the rules on appropriate context-bound selection of one form over another as part of a process of linguistic socialisation (see Lanza 1997; Matras 2009: chapter 2). Following from this assumption, I propose that bilinguals – even “full”, “fluent” or “balanced” bilinguals – do not, in fact, organise their communication in the form of two “languages”or “linguistic systems” (for more detail and a discussion of the empirical basis of this proposal see Matras 2009, 2012a). Rather, bilinguals have an enriched and extended repertoire of linguistic structures. As part of their linguistic socialisation, they learn when to select which word-form, construction, or prosody pattern as appropriate in a given setting or context of interaction. Some settings or contexts allow greater flexibility of choices. This is where bilinguals may make most effective use of their full repertoire, exploiting nuances as well as contrasts between variants of equivalent or near-equivalent meaning. Other interaction contexts are more exclusive. The existence of selection rules that are part of the bilingual’s communicative competence triggers a series of associations between a particular subset of structures and interaction context set A, between another and interaction context B, and so on. This association is what we identify as our socially constructed notion of a “language” or a “language system”. It is thanks to this sociallybroadcast notion that bilingual children learn, around the age of 3, that they speak two “languages”; until then, their use of word-forms and constructions is governed by a prolonged process of trial and error, usually unaccompanied by any explicit analytical labelling or other overt classification of the elements of their repertoire. It is important to note that such clear association between structure and set of interaction contexts does not necessarily exist for each and every element of the linguistic repertoire. German-English bilinguals, for example, have only one single word-form for concepts such as internet, download, computer or even baby (subjected of course to embedding in different phonological and morphosyntactic environments). By the same token, speakers of numerous Romani varieties have only one operator word-form in their repertoire that represents the function of the contrastive conjunction “but” and often of other discourse particles and connectors (see Matras 1998), and speakers of Jerusalem Domari have



Why is the borrowing of inflectional morphology dispreferred?  

 51

an entire inventory of conjunctions, prepositions, comparative adjectives, most higher numerals, modality markers and more that are shared with their principal contact language, Arabic (see Matras 2012b). Such category-specific inseparability among the subsets in a bilingual’s repertoire is part of the definition of “borrowing” that I pursue in this paper. How does borrowing come about? And how is it linked to other contact phenomena? Language contact phenomena are seen in the model outlined here as the outcome of function-driven choices through which speakers license themselves, while interacting in a context of type B, to select a structure (word-form, construction, meaning, phonological features, etc.), despite its association primarily with interaction context set A. When claiming that choices are function-driven, I am not suggesting that selection of A-structures in B-contexts is necessarily always conscious, deliberate, or strategic. In fact, I propose that contact phenomena are arranged on a continuum, from those that are in fact not at all voluntary, indeed even counter-strategic in their origin, to those that are conscious and deliberate (for a detailed discussion see Matras 2009). All, however, are functional in the sense that they are the product of language processing in goal-oriented communicative interaction. The susceptibility of certain structural categories to contact-related change is therefore not accidental, but inherently bound to the function that those categories have and the way they support language processing in discourse. Contact phenomena must in this respect be seen as enabling rather than as interfering with communicative activity. In the center of the approach that I follow is thus the assumption that it is not languages that borrow structural material, but rather bilingual speakers that license themselves to employ the same or similar sets of structures in different communicative settings, thereby allowing sub-sets of their linguistic repertoire to undergo fusion, i.e. to be generalized irrespective of the choice of “language” in a given interaction setting. The key to understand borrowing is to understand bilingual speakers’ motivations to allow such fusion of sub-sets within their repertoire. Different motivations may affect different functional categories in different ways. A detailed mapping of borrowing motivations to categories is beyond the scope of this paper (but see Matras 2007, 2009, 2011 for details) and I will instead limit myself to just a few generalizations. Let us begin with the borrowing of vocabulary. The Loanword Typology project (Haspelmath and Tadmor 2009) delivers an interesting dataset that allows us to identify the hierarchical nature of lexical borrowing in cross-linguistic perspective. Loanwords have traditionally been explained as motivated by cultural innovations, by taboos (against the use of established words), and otherwise by the prestige of the donor or source language. But for those

52 

 Yaron Matras

borrowings that are replacements of existing concepts rather than labels for new concepts, no explanation is offered as to why prestige should motivate loans in one domain but not in another, or why certain semantic constraints should work to resist borrowing. Loanwords are bilingual speakers’ way of adjusting their overall repertoire of lexical words and re-negotiating the constraints on the selective use of words in certain settings, or with certain interlocutors. The data provided in the contributions to the Loanword Typology volume (Haspelmath and Tadmor 2009) provide us with an opportunity to explore how this process of re-negotiating the bilingual lexical repertoire is related to the conceptualization of reality. It appears that “uniform” lexical material (i.e. lexical items that are generalised throughout the bilingual’s repertoire and used irrespective of interaction setting or interlocutor, i.e. “borrowed”) is symbolic primarily of activities that are shared with another, neighbouring linguistic community: These may include commerce, religion, administration, and technology (whether the concepts are new, or established but replaced by loanwords). By contrast, personal and family experiences (body, emotions, space) remain conceptually protected and individualised, and this is reflected in the enduring separation of language-specific sets of relevant linguistic expressions within the bilingual repertoire, i.e. in the low rate of borrowing for these domains. The borrowing of grammatical structures is representative of a somewhat similar conceptualization of contrasts, as can be inferred from the crosslinguistic sample presented in the contributions to Matras and Sakel (2007); see discussion in Matras 2007, 2009) as well as from other samples (e.g. Elšík and Matras 2006). Functions that serve to negotiate attitudes among the participants in the interaction and which convey evaluations, assessments, the processing of presuppositions, or emotions, are particularly prone to borrowing: This includes information structuring at the level of the discourse and clause, the expression of modality and evidentiality in the verb and verb phrase, indefiniteness in the noun phrase domain, prosody in phonetics and phonology, discourse particles and phasal adverbs in lexical morphology, and comparative and superlative marking in attributes – all of which are hierarchically more susceptible to ­contact-induced structural change within their respective grammatical categories or paradigms. They represent bilingual speakers’ need to align the emotional and presupposition-oriented side of negotiating communicative interaction across interaction settings. This need is, of course, constrained by the social norms of communication in the speech community and the extent to which innovations may be licensed and consequently propagated. Invariably, a community whose bilingualism is unidirectional and which uses an in-group language primarily in informal



Why is the borrowing of inflectional morphology dispreferred?  

 53

and private or domestic settings, is more likely to adopt a more lax attitude toward contact-induced innovations. The frequently cited notions of “prestige” or “dominance” capture this realisation. Macro-level social constraints are thus crucial to determining whether individual innovations will lead to language change. But they are only of secondary relevance when it comes to predicting and explaining which innovations are likely to occur in the casual speech of bilinguals, and so which changes stand a chance of being propagated in the first place. The hierarchical nature of contact-induced change in both lexicon and grammar (cf. Matras 2009, chapters 7–8; Matras 2011) reveals that the motivation for innovation is functional in the first instance, and that it is driven by the role that categories play in triggering mental processing tasks in communicative interaction. In the lexicon, borrowing represents the fusion of structural material that represents shared concepts and values. In grammar, borrowing represents in the first instance fusion of the operational procedures through which speaker and hearer gauge attitudes to propositional content and monitor and control participant roles in interaction. It is thus around the more gesturelike, evaluative and cooperative aspects of communication that bilingual speakers find themselves most tempted to eliminate the burden of having to select among structures in different interaction settings – in other words, to maintain a separation of “languages” – and where fusion or uniformity of form-function representation seems most beneficial.

3 Alternatives to the borrowing of morphemes In light of this view of contact-related language change, we must consider what functional roles are played by morphological paradigms, and what motivations speakers in bilingual situations may have to allow the sub-sets of their repertoire of structures – their “languages” – to undergo fusion in the morphological domain. The mere import of lexical material does not necessitate the replication of morphological “matter” (for the distinction between “matter” and “pattern” replication see Matras 2009), as borrowed lexical items are most commonly morphologically integrated. Indeed, morphological integration has famously been proposed as a key indicator of borrowing (Poplack, Sankoff, and Miller 1988), while other approaches view language mixing as inherently constrained by a separation of lexical content words and grammatical or “system” morphemes that are said to be resistant to mixing (Myers-Scotton 1993, 2002). Thus, English often assigns its own inherited adverb derivation procedure to French-derived adjectives and nouns: common‑ly, care‑fully, ­ comfortabl‑y.

54 

 Yaron Matras

Creative morpho-syntactic procedures may also be applied to borrowed lexical material. Thus, the Hebrew equivalent of “intelligently” is be-ófen inteligénti, literally ‘in an intelligent manner’. Maltese and Lovari Romani, on the other hand, tend to rely on borrowing (i.e. matter-replication) of full adverbial wordforms from the respective contact language rather than apply creative wordclass derivation processes: Maltese has ‑ment in Italian loans such as verament ‘truly’, speċjalment ‘especially’, while Lovari Romani has ‑no/‑nje in Slavic loans like specijalno ‘especially’, sistematičnje ‘systematically’, and objektivno ‘objectively’. Domari offers a somewhat extreme but interesting case of “bilingual suppletion” that is employed as an alternative to the productive borrowing (matter replication) of derivational morphology expressing degree (comparative and superlative) in adjectives. Its contact language, Arabic, employs a morphophonological template áCCaC to derive comparative/superlative forms from consonantal roots: kbīr ‘big’, ákbar ‘bigger’; zġīr ‘small’, ázġar ‘smaller’. This template cannot easily be isolated or integrated into the agglutinativeinflectional morphology structure of Domari, nor is it simple or even possible to break down Domari adjectives such as tilla ‘big’ or kištota ‘small’ into triconsonantal roots for insertion into the Arabic-based derivation template. The solution adopted by Domari speakers is to borrow the full Arabic word-form for all comparative/superlative forms, resulting in complete borrowing-based ­suppletion of the inventory of adjectives: tilla ‘big’, ákbar ‘bigger’; kištota ‘small’, ázġar ‘smaller’. Structural factors relating to the transparency and analysability of morphemes thus play a role in constraining the productive transfer of ­morphemes from one language to another in this case. However, at the same time speakers adopt an alternative solution that serves the same purpose. The motivation to syncretise operational procedures that are carried by morphological structures across languages can thus be satisfied in ways other than a direct import of productive morphemes. This principle is nicely illustrated by the prominence of morphological pattern replication, meaning the matching of pivotal functions of s­ emantic constructions to corresponding morphemes across languages (see Matras 2009). Morphological pattern replication rests on the role of morphological features in structuring and delivering constructions. In terms of motivation, it is a ­by-product of the trend to syncretise the inventory of constructions across the languages in a bilingual’s repertoire. In structural and functional terms, pattern replication facilitates the generalisation of constructions across the repertoire while maintaining the overt separation of form. It thus allows speakers to continue to conform to and to flag the separation of language-specific interaction contexts and settings.



Why is the borrowing of inflectional morphology dispreferred?  

 55

Khuzistani Arabic is a good example. It develops a remote past (pluperfect), a construction that is not inherited from Arabic, by calquing the use in the contact language Persian of an auxiliary “was” (Matras and Shabibi 2007): (1) a. Khuzistani Arabic mәn rәħ‑әt lә-l-bīet, huwwa mā-rāyәħ čān when went-1sg to-def-home he neg-going.sg.m was.3sg.m b. Persian vag  ti raft-am xūne, ūn na-rafte būd when went-1sg home he neg-gone was.3sg.m ‘When I went home, he had not [yet] gone away’ Northeastern Romani dialects – especially Russian, Lithuanian and Latvian Romani – have developed a construction that mirrors the distribution of functions among nominal cases in Russian. It is often found alongside the more conservative, inherited Romani construction (cf. also Tenser 2008): (2) a. Northeastern Romani (conservative/inherited) me na somas khere 1sg.nom neg was.1sg home b. Northeastern Romani (convergent) man na sys khere 1sg.obl neg was.3sg home c. Russian menya ne bыlo doma 1sg.acc/gen neg was.3sg.neutr home ‘I was not at home’ Heine and Kuteva (2005) identify contact-induced grammaticalization as one of the major processes of historical change in language contact situations. Contactinduced grammaticalization is understood as a change in distribution patterns and an extension of meaning and functional environment, which replicate the features of a model construction in another language, which is the target of imitation (Heine and Kuteva 2005: 80). The above examples show precisely this kind of process for the extension of the Khuzistani Arabic copula-auxiliary čān ‘was’ as a pluperfect marker and for the appearance of the Northeastern Romani oblique pronoun man ‘me’ and the impersonal copula sys ‘was.3SG’ in the past-tense locative construction. Pattern-replication may, however, give rise also to new structures of bound morphology. Bakker (2006) discusses the parallel grammaticalization of location

56 

 Yaron Matras

expressions into nominal case affixes in Sri Lankan Portuguese and Sri Lankan Malay. Kurdish and Neo-Aramaic, two contiguous languages of Mesopotamia, have both developed split verb alignment structures that distinguish between transitive and intransitive agreement morphology in the past tense. This distinction can still be seen in some dialects of the respective languages, such as Kurmanji Kurdish of the northern Kurdish regions and the Jewish Neo-Aramaic dialect of Saqqez in Iran (own fieldwork): (3) a. Kurmanji ez rabû-m û min 1sg.nom stood.up-1sg and 1sg.obl

derî   vekir door  opened.ø

b. Saqqez Neo-Aramaic qīm-na, tara-kē plix-li stood.up-1sg.itr door-the opened-1sg.tr ‘I stood up and opened the door’ Here, both languages make use of a person-morpheme derived from the historical present-tense copula form – Kurmanji 1sg -m, Aramaic 1sg.m -na – to indicate subject agreement on the past-tense intransitive verb. At the same time, both languages develop distinct structures to indicate the subject of the past-tense transitive verb, each derived from a structure that encodes the oblique or non-­ nominative agent. In Kurmanji, the form is the independent oblique pronoun min. In Aramaic, it is the historical person-inflected dative preposition l‑i *‘for-me’, which is now synthesised to the verb to express the agent of the transitive verb. The corresponding constructions in Kurmanji and Neo-Aramaic are thus not isomorphemic. Rather, they share the mapping of pivotal functions onto morphemes that have related meanings and which are exploited for the purpose of syncretising the formal organisation of a semantic unit. The Mesopotamian (Kurdish and Aramaic) example shows that the process can lead not only to changes in the distribution and semantic meaning of forms, as seen in the cases of Khuzistani Arabic and Northeastern Romani, but also to the emergence of new inflectional morphemes. Mithun (2012: 35) explains similarities in the morpho-syntactic organization of patterns across neighbouring languages in Northern California by referring to the sociolinguistic setting that was characterised by widespread multilingualism and flexibility in the choice of language, and the absence of strong differences in prestige among the languages of the area. Under these circumstances, the choice of speaking a particular language was manifested primarily in the lexicon, while abstract patterns of expressions are said to have been less likely to be under conscious control. This kind of scenario is well in line with the model outlined in the previous section. However, it does not



Why is the borrowing of inflectional morphology dispreferred?  

 57

explain specifically why certain inflectional morphemes are not borrowed if speakers are, as Mithun (2012: 15) asserts, “rarely conscious of bound morphemes”. Clearly, there is a difference between the roles played in language processing by more abstract constructions, on the one hand, and by individual morphemes, especially inflectional morphemes, on the other. Speakers’ ability to control the selection of items within a repertoire therefore differs. The selection of inflectional morphology seems to be very much a part of speakers’ conscious choice in favour of one language or another in a given interaction context, even if one might argue that individual bound morphemes, unlike lexical items, are not consciously identified and selected on a one-by-one basis. Consider the following examples from Domari – both transcriptions of natural speech – along with the equivalent translations into the contact language, colloquial Palestinian Arabic (for a discussion of data sources see Matras 2012b): (4) a. Domari aktar min talātīn xamsa ū talātīn sana more from thirty five and thirty year ma lak‑ed‑om‑is neg see‑past‑1sg‑3sg.obl b. Arabic aktar min talātīn xamsa ū talātīn sana more from thirty five and thirty year ma šuf‑t‑hā neg see.past‑1sg‑3sg.f ‘It has been more than thirty, thirty five years since I’ve seen her.’ (5) a. Domari hāda/ kān ʕumr‑om this was.3sg.m age‑1sg sabʕa snīn seven years

yimkin sitte maybe six

b. Arabic hāda/ kān ʕumr‑ī yimkin sitte this was.3sg.m age‑1sg maybe six sabʕa snīn seven years ‘This/ I was maybe six or seven years old.’

snīn years

snīn years

Domari borrows a massive amount of lexical material, function words, morphosyntactic organization patterns and phonological features from Arabic. The

58 

 Yaron Matras

density of shared material can clearly be seen by the comparison of the Domari data with the (constructed) Arabic translations in (4)–(5). The respective utterances are nearly identical. Yet Domari speakers, who are all bilingual, have no hesitation in identifying segments (4a) and (5a) as Domari, rather than Arabic. At the same time, they have no alternative way of expressing these sentences in Domari without drawing on Arabic-derived material. The similarities between the two languages in these two examples are therefore inherent to the (shared) structures of the two languages (through borrowing), rather than the product of context-bound stylistic choices (code-switches). What, then, does the choice of Domari as the language of the utterance rest upon in these two examples? In example (4a) it is limited to a single lexical item, lah‑ ‘to see’, which is accompanied by Domari (i.e. Indic, inherited) inflectional morphology indicating tense and subject and object agreement. In a way, this example is in line with Mithun’s (2012) assertion that lexical choice may provide the more obvious manifestation of language choice, but with the caveat that all other lexical material in this particular utterance is actually shared with Arabic and therefore not in itself distinctively constitutive of speaking “Domari”. Rather, it is the inflectional morphology indicating tense and person agreement that is uniquely and distinctively Domari and which carries the full weight of indicating language choice in this utterance. This is even more clearly visible in example (5), where the two segments with equivalent meaning, the Domari original and its Arabic translation, differ solely in the choice of the 1sg possessive marker on the word ʕumr‑ ‘age’, which is ‑om in Domari but‑ī in Arabic. This shows that there is a functional motivation to avoid the blurring of boundaries around inflectional morphology, as long as there is a motivation to maintain some sense of language separation in distinct interaction settings and contexts. As I stated in my opening remarks, inflectional morphology serves as the carrier for anchoring the predication and for identifying the predicate’s argument structure. It is thus the backbone of the delivery mode (rather than the propositional content) of the utterance. As such, it lends the utterance its identity in regard to the multilingual speaker’s efforts to accommodate to the expectations of language choice in a given communicative setting. The integrity of inflectional morphology, along with, to be sure, the preservation of some basic, distinctive lexicon, is the key to flagging language distinctness and thus the key to maintaining separation between distinct sets of communicative interaction settings, for example group-internal and group-external. In the case of Domari, a moribund language that is not being passed on to the next generation, it also distinguishes generation-internal from cross-generation communication. Meaning extension and grammaticalisation of morphemes, leading in some cases to the emergence of new bound morphology, enable speakers to maintain language separation while



Why is the borrowing of inflectional morphology dispreferred?  

 59

achieving maximum cross-language syncretisation among syntactic-semantic constructions. Much like the maintenance of basic lexicon, the preservation of morphemes serves to flag language loyalty. Having established that, I shall now devote the remainder of the discussion to the motivations and circumstances under which we do find substance or “matter” borrowing of bound morphology.

4 Common morphological borrowings The functional constraint on the borrowing of inflectional morphology – a reluctance to blur the fundamental way of distinguishing between languages in a situation of bilingualism – does not hold for bound morphology that is derivational in nature and which serves primarily to modify semantic meaning. Recent studies of morphological borrowing – e.g. Chamoreau (2012) on Spanish diminutive markers in Mesoamerican languages such as Yucatec Maya, Tosco (2012) on a French agentive marker in Piedmontese – add to a large pool of evidence of the borrowability especially of nominal derivation morphemes such as agentives and diminutives. Colloquial Modern Hebrew, for example, has adopted a series of agentive and diminutive suffixes from Yiddish. They include ‑ist (bitsu’íst ‘doer’, from bitsúa ‘implementation’), ‑er (širyon‑er ‘tankist’, from širyon ‘armour’), and ‑le (xamúdale, xamúdile ‘sweety (f/m)’, from xamud/á ‘sweet/cute’), as well as markers that had been borrowed into Yiddish from its Slavic contact languages, such as ‑nik (kibútsnik ‘member of a Kibbutz’) and ‑čik (baxúrčik ‘a [nice, adorable] young man’, from baxúr ‘young man’). Hebrew is an interesting case due to its particular history as a language that has been re-vernacularised initially as the second language of the first generation of speakers in the early 1900s. Borrowed derivational markers are common in colloquial speech, but less so in formal, written styles of Modern Hebrew. The first generation of speakers of Modern Hebrew licensed themselves to make full use of certain word-derivational resources of their multilingual repertoires irrespective of interaction context; when speaking Hebrew they continued to draw on the same procedures as in their native languages. The prominence of agentive and diminutive markers among borrowed bound morphemes can be attributed to the saliency of meaning differentiation in nouns, which in turn is connected to the prominent role of nouns as signifiers of topical entities, in particular of animate topics. But the borrowing of other derivational morphemes, for example those that are responsible for word-class changing derivation, is also well attested. Many Romani dialects show the suffix ‑(i)mos from Greek, which derives abstract nouns from Romani verbs and adjectives, e.g. sastimos ‘health’ from sasto ‘healthy’, marimos ‘struggle’ from mar‑ ‘to fight’.

60 

 Yaron Matras

Another suffix of Greek origin in Romani, ‑to (from Greek ‑tos, which also derives adjectives from verbs), derives ordinal numerals from cardinal numerals: dujto ‘second’, from duj ‘two’. Both affixes are productive with the pre-European (Indic or Asian) lexical component of the language. The English suffixes ­‑able and ‑(e) ous, both of Romance origin, are used to derive adjectives from nouns and verbs within both the Romance and Germanic lexical components: feasible alongside loveable; courteous alongside righteous, while ‑ment derives nouns from verbs, as in both argument, and Germanic-based bereavement. The borrowing of a more subtle meaning derivation procedure is attested in the central and eastern European dialects of Romani, which adopt so-called Slavic Aspect markers and apply them to pre-European Romani verbal roots. This is modelled on the derivational procedures that are applied to the corresponding verb roots with equivalent meaning in the contact languages such as Polish, Slovak, and Russian: Polish Romani za-pindžkirel ‘to introduce’ from pindžkirel ‘to recognise’ (Polish za‑poznać and poznać), do-resel ‘to obtain’ from resel ‘to arrive’ (Polish do‑stąpić and stąpić), pše-džal ‘to cross, climb over’ from džal ‘to go’ (Polish prze‑chodzić and chodzić). Here we have a wholesale adoption of the structural procedures of meaning derivation and a fusion of the two languages in contact in this domain, without disturbance to the inflectional morphology and so while maintaining the coherence and integrity of the verbal predication. In other areas we come across morphological borrowing in the domain of meaning derivation that does, however, have potential implications for syntactic processes such as agreement marking in the clause, and which thus infringes on the role of inflectional morphology. Vietnamese, Korean and Japanese have all borrowed nominal classifiers from Chinese. Rose (2012) reports on the borrowing of Cariban markers of plurality and collective number into Tupi-Guarani languages. Some Neo-Aramaic dialects borrow the Kurdish indefinite article, and some Romani dialects borrow indefinite articles from their respective contact languages Italian and Albanian. Plurality markers on the noun are probably the most frequently attested inflectional morphemes that are borrowed (cf. Gardani 2012). While in English plurals like phenomena and fungi are limited to a marginal inventory of borrowed lexemes (cf. discussion in Kossmann 2010), in Vlax Romani the Romanian-derived plural form ‑uri is productive and accompanies loans from subsequent contact languages, while in Jordanian Domari the Arabic derived feminine plural ending -āt diffuses into inherited material, irrespective of gender: lāčiy‑āt ‘girls’, putr‑āt ‘boys’ (see Matras 2012: 17). Plural markers belong to the class of grammatical markers that MyersScotton and Jake (2000) define as “early system morphemes”. It is hypothesised that they emerge “earlier” in the production process of the utterance and that they therefore occupy an intermediate position between lexical meaning (“content



Why is the borrowing of inflectional morphology dispreferred?  

 61

morphemes”) and sentence-level inflectional morphology (“late system morphemes”). The nomenclature does not offer an obvious explanation for the tendency of plurality markers (as well as other number markers and classifiers, and markers of definiteness) to be more prone to borrowing than, for instance, case or person morphology. But if we consider the information value of number and definiteness, we can regard them as categories that modify the content meaning of a lexical item – independently, potentially, of that item’s role in the predication as a whole. Fusion of meaning modification procedures across languages does not, as we have established, interfere with the language-specific integrity of the predication. One piece of evidence in support of this impression is the treatment of plurals as part of the lexical stem (rather than as an inflectional ending) in borrowed words. Jerusalem Domari adds its own inherited plural formation to Arabic-derived plural nouns: singular zálame ‘man’ (Arabic singular zálame), plural zlām‑é (Arabic plural zlām). Simango (2000: 494) reports on a similar reinforcement of English plural forms through native plural affixes in Chichewa: ma-refugee‑s ‘refugees’. A similar phenomenon is the doubling of definite articles in Spanish loanwords from Arabic: Spanish el arroz ‘the rice’ (Arabic ar-ruzz ‘the rice’). A case can be made therefore that inflectional morphology is more likely to be borrowed if it is re-interpreted as derivational, i.e. as modifying meaning rather than syntactic role. An excellent example is provided by the adoption of Greek tense-aspect markers into Romani. The Greek system distinguishes, for individual inflection classes, between a present and a past or aorist stem of the verb: Greek jir‑íz‑o ‘I return’, jír‑is‑a ‘I returned’. This stem distinction is carried over into Romani, where, however, it is reinforced by inherited (Indic) Romani tense-aspect morphology, which similarly distinguishes a present stem (by default, through the absence of any stem modification) and a past or perfective stem (represented by a past or perfective marker, often ‑d‑, ‑l‑ or ‑j‑ depending on inflection class and dialect). The Romani rendering of the Greek-derived loan verb “to return” is thus jir‑iz‑av ‘I return’, jir‑is‑áj‑l‑om ‘I returned’. It preserves the Greek tense-aspect distinction in the stem (‑iz‑/‑is‑). It then adds to it Romani subject agreement markers (1sg present tense ‑av, 1sg past tense ‑om) and, in the past tense, a past-perfective marker ‑l‑ that attaches to what appears to have served as a light verb integration strategy for loan verbs, based on the verb ‑a(v)‑ ‘to come/to become’ (for details see Matras 2002: chapter 6). This strategy appears to have emerged during the Early Romani period, in contact with Byzantine Greek. It is still found in the dialects of Romani that are spoken in Greece (by those who have been settled in the country continuously), such as the Romani dialect of Parakalamos: vojt‑iz‑av ‘I help’, vojt‑is‑áj‑l‑om ‘I helped’. But it was also retained in other regions of the Balkans following the fall of the Byzantine Empire and the shift to other contact languages. The dialects of some of

62 

 Yaron Matras

the Muslim Romani groups in Bulgaria, such as the Kalajdži and Kalburdžu, use this same Greek-derived template to adopt past-tense stems of Turkish verbs into the Romani inflectional paradigm: anlad‑iz‑av ‘I understand’, anlad‑is‑áj‑l‑om ‘I understood’. Some of the Romani dialects of southeastern Romania, such as the Spoitori, use it to integrate Romanian-derived verbs, as in konduč‑iz‑av ‘I drive’, konduč‑is‑áj‑l‑om ‘I drove’, while in some of the Macedonian Romani dialects, like Kovački of Skopje, it is used for Macedonian-derived verbs, as in piš‑iz‑ava ‘I write’, piš‑is‑áj‑(l)‑um ‘I wrote’. Other Romani dialects, including those that are spoken outside the Balkans, either preserve a similar system, drawing on the same or other Greek-derived tense-aspect markers (often present-tense ‑in‑), or simplify the system, usually maintaining some derivational morpheme that indicates the adaptation of a loan verb stem into the language and which mediates between that stem and the inherited Romani tense-aspect and person inflection. How can we explain the Romani patterns in terms of our usage-based model of language contact and contact-induced language change? The integration of borrowed verb forms into the inherited Romani tense-aspect and person agreement paradigm indicates that even in a situation of intense bilingualism in a geographically dispersed minority language (spoken by a population with no political power, few institutional resources, and low social prestige), flagging language identity through the choice of predication grammar was sufficiently important to maintain the inherited inflectional morphology with such loans. The adoption of Greek-derived tense-aspect marking and its continuing productivity long after contact with Greek was interrupted suggests that Greek tense-aspect morphology was associated with the context-bound meaning of the verb, rather than with the initiation procedure of the predication itself. Subsequently it began to serve as an integration template, bridging between the lexical content of borrowed verb roots and the language’s predication grammar. What began in Greek as an extension of the verb that identified both inflection class and was itself a carrier of tense-aspect inflection, was adopted into Romani purely as a derivational marker that identified the Greek (and later European) origin of the verb root. Although the markers continue to be sensitive to tense-aspect, at least in some Romani dialects, they are always reinforced by inherited Romani tense-aspect markers, and so their original inflectional potential is lost.

5 Cross-language morphological analogy Few cases of borrowed inflectional morphology are convincingly attested in the literature. This is not to say that there are no exceptions. Having now established



Why is the borrowing of inflectional morphology dispreferred?  

 63

why speakers are less motivated to borrow inflectional morphology (compared to their motivation to borrow other, emotional-attitudinal categories of language), and why speakers are, in addition, inhibited to borrow inflectional morphemes (seeking to avoid ambiguity in the identity of the predication, which in a bilingual situation is both symbolic and constitutive of language choice), I now turn to the more exceptional cases. Exceptions do not question the validity of the generalisations made above about borrowing. Rather, they confront us with cases that merit special investigation with a view toward identifying the special factors that might override what are common, function-oriented constraints and principles, or which might facilitate borrowing even in structural environments that are normally less susceptible to contact-induced change. One such factor is the coincidental similarity between the form-structure of functionally corresponding elements in the two languages, and a perception, arising from this similarity, that these corresponding forms are in fact identical or near identical. Bilinguals’ spontaneous choices are sometimes influenced by formal similarities among functionally related elements in their languages. This can be illustrated by the following examples (from own fieldwork): (6)

German (spoken by an English-dominant child, 7 years old) Er ist grösser denn mir he is bigger part me.dat ‘He is bigger than me’

(7) Hebrew (spoken by a child with frequent exposure to Arabic, 5 years old) ra’íti et ha-xayá ha-zot saw.1sg acc the-animal the-this ‘I saw this snake’ In (6), the child (who speaks German to his mother in what is otherwise an English-speaking environment) uses the words denn mir for ‘than me’ instead of German als ich. The choice is facilitated, of course, by a lack of confidence in German grammar and lack of exposure to the specific task-routine in German, which requires this particular German construction. In his attempt to replicate the English construction in German, the child draws on functionally related elements in German that have a similar formal shape: the discourse particle denn (which in spoken German has primarily prompting functions), and the dative pronoun mir. In (7), the child is a native speaker of Hebrew who attends an Arabic-speaking nursery. He is referring to a snake (Arabic ħayya), but selecting the formally similar Hebrew word xaya ‘animal’. In both examples, the corresponding pairs are formally similar but also semantically or functionally related in some way (and in each of the language pairs also etymologically related).

64 

 Yaron Matras

Formal similarities among functionally related inflectional morphemes may trigger similar analogies that can facilitate borrowing. In Maltese, adjectives that are borrowed from Italian are generally adopted into the inherited (Semitic) inflection patterns and retain gender and number agreement with borrowed Italian nouns. Maltese adjectives follow the noun, as they normally do in the principal contact language, Italian (and Sicilian). Like Italian, Maltese too has two genders. Borrowed adjectives that end in a consonant take the inherited (Semitic) inflection pattern, in which the m.sg is treated as default and has no identifiable vocalic ending, while the f.sg ends in ‑a: (8) a. Maltese f’kuntest in context

modern modern.m

b. Italian in un contesto modern‑o in a context.m modern‑m ‘in a modern context’ (9) a. Maltese poeżija modern‑a poetry modern‑f b. Italian poesia modern‑a poetry modern‑f ‘modern poetry’ Both the form and the position of the Maltese feminine singular adjective happen to agree with the form and position of the feminine singular adjective in Italian. This triggers an analogy with the plural, where the Italian inflectional ending ‑i (Italian m.pl) is preserved on adjectives that are borrowed from Italian: toroq modern‑i ‘modern roads’ (toroq being a Semitic noun). Stolz (2012) discusses the case of Spanish loan adjectives in Chamorro. Here too, there is no indication that bound inflectional morphemes from Spanish are diffused “backwards” into the inherited Austronesian vocabulary. While no obvious analogy is at work in the Chamorro cases, the retention of adjective inflection on Spanish loans is a case of morphological compartmentalisation (see below). Third person pronouns in some Romani dialects adopt the plural inflection markers of the contact languages, a process that is triggered by the coincidental similarity between the form of the inherited pronoun in Romani and the respective pronominal forms of the contact languages Hungarian, Slovene,



Why is the borrowing of inflectional morphology dispreferred?  

 65

and Turkish, which are all unrelated either to one another or to Romani. The original Romani third person pronouns are ov/oj ‘he/she’, and on ‘they’; they appear in most Romani dialects, sometimes with minor phonological stem modifications. The form of the third person plural pronoun in some varieties of Hungarian Romani (Romungro) is on‑k, which replicates the plural ending of the Hungarian third person pronoun (singular ő, plural ő‑k). In Slovenian Romani (Dolenjski), the form is on‑i, replicating the pronominal form in Slovene (singular on, plural on‑i), and in some varieties of Thracian Romani that are or were in contact with Turkish (Kalburdžu, Xoraxane and others) the form is on‑nar, replicating the Turkish structure (singular nominative o, singular oblique on‑, plural on‑lar). It is noteworthy that in all three contact languages, the ending that is used to indicate plurality on the pronoun is also the ending that is used to indicate plurality on nouns. It is thus arguably situated in the intermediate domain between meaning-derivational and inflectional morphology in the sense discussed above. A comparable case might be Asia Minor Greek (Dawkins 1916: 59), where the Turkish 2pl ‑iniz, which in Turkish is used to mark 2pl possession on nouns and 2pl in both the present copula and past-tense lexical verbs, is added to the Greek past-tense paradigm of inchoatives and medio-passives. Another development in Romani dialects, found in Crimean Romani and in some of the Romani dialects of northern Bulgaria, shows the borrowing of parts of the person concord set from Turkish (cf. Elšík and Matras 2006: 136). This results from an analogy that is based on chance similarities between the inherited Romani and Turkish conjugations. The inherited Romani pasttense concord markers contain the consonant ‑m in the first person (singular ‑om/‑em/‑im, depending on dialect, plural -am) and a consonant ‑n in the second person (singular ‑an, plural ‑en). They resemble the corresponding Turkish singular forms 1sg ‑Vm and 2sg ‑Vn (with variation subject to vowel harmony). The Turkish plural pronouns are augmented forms of the singular morphemes: 1pl ‑VmVz, 2pl ‑VnVz. By analogy, these Romani dialects form a past-tense 1pl concord marker ‑amus and a past-tense 2pl marker ‑enus. Here too, the agglutinative marking of plurality in the contact language makes the marker ‑us analysable. It is replicated in Romani with inherited verbs, replacing the original marker (which is preserved in other dialects of the language). Slovene Romani (Dolenjski), too, shows this kind of process. Here, the original Romani past-tense 1pl marker ‑am has been replaced by the corresponding Slovene affix ‑amo on the basis of the formal resemblance between the two. The analogy is then extended to the Romani 2pl (originally ‑an or -en), for which the Slovene affix ‑ate is adopted (which has no formal resemblance to the original Romani form).

66 

 Yaron Matras

6 Morphological compartmentalization Above I defined language contact as the conventionalization of bilingual practices through which bilingual speakers generalize a structure or an entire category of structures across their linguistic repertoire, leading to “fusion” or the inseparability of languages around that particular structure or category. The motivation behind the process is to syncretise language processing procedures in the bilingual’s languages, in other words, to reduce the need to select between competing sets of functionally equivalent or near-equivalent structures in different interaction settings or contexts. From a user perspective, fusion is thus functionally motivated. The motivation to syncretise procedures across languages is balanced off and constrained by a sense of language loyalty, which, at the same time, motivates bilingual speakers to maintain adequate means of separating their languages as a way of flagging accommodation to distinct interaction settings. I have argued that inflectional morphology plays a central role in flagging such separation. It encodes the predication and its arguments and so it serves as the delivery mode for the proposition as a whole. It is the predication that reflects, represents and indeed which verbally constructs an interaction setting as flavoured in a particular way. Where alternational codeswitching is involved, it is the predication that flags an utterance as set against the contextual expectations – what Gumperz (1980) has called “metaphorical juxtaposition of languages”, and what Myers-Scotton (1993, 2002) refers to as “marked” language choices. The borrowing of inflectional morphology is constrained due to the role that inflectional morphology plays in maintaining language differentiation in bilingual situations. My final point concerns a further exception to this norm: the compartmentalisation of inflectional morphology. I begin with examples from Jerusalem Domari. Here, extensive borrowing from Arabic leads to the wholesale adoption into Domari of, among other structures, almost all the forms that participate in indicating modality and “relevance” (in the sense discussed by Blakemore 2002). One can therefore speak of a fusion of the procedures for marking modality in Domari and Arabic. From the point of view of the pragmatics of communication, “Domari” speech is characterized and is identifiable through the choice of various structures, but these do not include modality markers; much like for a German-English bilingual, the distinction between the languages is manifested by the choices made around numerous grammatical and lexical structures, but not around the words internet or baby, which are identical in the two languages. What is remarkable (though not unparalleled) is that Domari adopts Arabic modal expressions along with their Arabic inflectional morphology, which remains fully



Why is the borrowing of inflectional morphology dispreferred?  

 67

productive in Domari (data from Matras 2012b; Arabic-derived modal elements are in bold face): (10) ama bidd-ī dža-m I want‑1sg go‑1sg ‘I want to go home.’

kurya‑ta house-dat

(11) putr‑or ḍall‑o      fumn‑ar‑i ben‑im son‑2sg remain‑3sg.m hit‑3sg‑prg sister‑1sg.obl ‘Your son keeps beating my sister.’ (12) ṣār‑u fēr‑and‑i baʕḍ baʕd‑ē‑san waṭ‑an‑ma began‑3pl hit-3pl-prg refl refl-pl-3pl stone‑obl.pl‑loc ‘They started to throw stones at one another.’ The Arabic person-agreement markers 1sg -ī , 3sg.m -o, and 3pl ‑u accompany the nominal modal expressions bidd‑ ‘want’ and ḍall‑ ‘keep’ and the verb ṣār‑ ‘to begin’, respectively. Their antecedents are all part of the Domari utterance or conversation context, and in examples (10) and (11) they are even verbalized explicitly through inherited (Indic) Domari expressions (ama ‘I’ and putror ‘your son’, respectively). The wholesale borrowing of Arabic modal and auxiliary expressions also extends to the marker of habitual aspect, which, as in Arabic, draws on the inflected Arabic past-tense copula kān- ‘was’: awa-ri nkī-s kull yōm (13) pandži kān-at 3sg was‑3sg.f come-3sg at-3sg every day ‘She used to arrive at his house every day.’ Note that the auxiliary form kān-at in (13) agrees with the third person pronoun subject pandži in both number and gender. However, Domari pandži does not encode gender. The adoption of Arabic inflection patterns in the copula thus introduces a gender agreement pattern into Domari which otherwise would not be encoded in the language. Since the full inflection patterns are adopted from Arabic – including person agreement and tense (full modal verbs such as ṣār‑ ‘to begin’ are also inflected for tense) – Domari has, in effect, a split in its verb inflection paradigms. Lexical verbs, whether inherited (pre-Arabic) or borrowed from Arabic, draw on the paradigms of one of several pre-Arabic (Indic) verb inflection classes. Modal expressions (with the exception of inherited sak‑ ‘to be able to’) draw on one of two principal Arabic inflectional paradigms, verbal or nominal. The language thus exhibits morphological compartmentalisation. Nonetheless, even in this particular case it would be a simplification to speak of the “borrowing” of Arabic inflectional morphology into Domari, since the plain term “borrowing” does not

68 

 Yaron Matras

quite capture the extent to which speakers rely on their Domari-Arabic bilingualism when using these forms: From a communicative perspective, what Domari speakers are doing is failing or avoiding differentiation between the procedures for indicating modality in their two languages. In their bilingual repertoire there exists only a single format for marking modality (for all semantic relations except for ‘to be able to’). Arabic inflectional morphology remains confined strictly to this function: it anchors modal and aspectual auxiliaries in relation to the lexical predication. The morphological marking of the lexical predication (verb inflection) and its arguments (nominal inflection) continues to be carried exclusively by the coherent set of inherited (Indic) inflectional morphology. Domari is not the only language in contact that adopts modal expressions along with their inflection. The Romani dialect of Parakalamos in the northwestern Greek province of Epirus shows a similar tendency. Here, however, the split in morphological paradigms is being extended through the incipient use of Greek inflection also with Greek-derived lexical verbs (items carrying Greek inflection are in bold face):1 (14) na bor‑o te diavaz‑o soske prepi te vojt‑iz‑av neg can-1sg comp study-1sg because must comp help‑loan‑1sg me daj‑a my.obl mother‑obl ‘I cannot study because I have to help my mother.’ Parakalamos Romani thus resembles Jerusalem Domari in that procedures that modify lexical predications – modals – are fused with the contact language. In examples (14) this included the impersonal prepi ‘must’ and the modal verb bor‑ ‘to be able to’, which takes productive Greek verb inflection. The treatment of Greek-derived lexical verbs is mixed. Some, like vojt‑iz‑av ‘I help’, are integrated into the inherited (Indic) Romani inflection paradigm. Others, in this example diavaz‑o ‘I study’, retain their Greek inflection. This appears to indicate an incipient breakup of the integrity of the Romani predication and a blurring of the distinction between interaction settings and their mapping onto “languages” in the bilingual community’s repertoire. This is very much in line with the community’s self-presentation in narrative interviews and informal conversations, where their Romani identity is strongly aligned with their identity as settled Greek musicians (cf. Theodosiou 2004). The community has few links to other Romani populations, and indeed members tend to distance themselves from the segregated

1  Data on Romani dialects are taken from the Romani Morpho-Syntax Database: http://romani. humanities.manchester.ac.uk/rms



Why is the borrowing of inflectional morphology dispreferred?  

 69

Romani minority. They regard themselves as integrated into the local Greek village community, and loyalty to their group language does stand in contradiction to this view. The discretion that speakers appear to be using in the domain of constructing the predication – allowing Greek predications to be mixed with Romani predications for no obvious stylistic effect – seems to be both a product and a symbolic token of this language ecology (see Matras 2008). The long-term effect of this kind of process can be seen in other Romani dialects of the Balkans, notably the dialects of Muslim Roms who have been living amongst Turkic-speaking minorities (see Matras 2002, 2008; Adamou 2012). In these varieties, bilingual speakers regularly draw on Turkish verb inflection when using Turkish-derived lexical verbs. Unlike the Parakalamos example, in the dialects of the Muslim Roms of the Balkans this pattern is highly conventionalised and is not subject to speakers’ choices. It is not, as in Parakalamos Romani, merely a token of the relaxation of the rules on the integrity of the predication and boundaries of “languages”, but an integral part of the structure of their Romani dialect. Turkish verb inflection is formalised with all Turkish-derived lexical verbs. The following examples (cf. Matras 2008) come from the Kalburdžu Romani dialect of Sindel in Northeastern Bulgaria (Turkish-derived inflected verbs are in bold face): (15) teara kan bittir‑iim adaja buki o zaman tomorrow fut finish‑1sg this work then bitaa kan dža‑v an e ga‑ete again fut go‑1sg in the village‑loc ‘Tomorrow I will finish this job. Then I will go back to the village.’ (16) pandž‑e daka‑en‑da sona bašla‑də te konušu‑i five-obl minute-obl.pl‑abl after began‑3sg comp talk‑3sg ‘After five minutes he started to talk.’ (17) ana etišt‑im othe dikhl‑em ani naj khere when arrived-1sg there saw-1sg that is.not home ‘When I got there, I saw that s/he wasn’t home.’ The Turkish-derived bittir‑ ‘to finish’, bašla‑ ‘to begin’ and etiš‑ ‘to arrive (punctually)’ carry (dialectal) Turkish person inflection (1sg –im, 3sg ­‑ə) as well as Turkish tense-aspect inflection (past tense –t‑/‑d‑). The Turkish verbs in (15) and (16) are both part of the extended domain of modal auxiliaries, yet their morphological inflection pattern in the Romani dialect is Turkish-derived, just like that of all Turkish-derived lexical verbs such as etiš‑ ‘to arrive (punctually)’. Kalburdžu Romani thus displays consistent compartmentalisation in its verb inflection morphology. Turkish inflection is used consistently with Turkishderived verbs, but it is contained within this particular group of verbs and does

70 

 Yaron Matras

not infiltrate the inherited domain of pre-European (or even Slavic- or Greekderived) lexical items. These carry inherited Romani (Indic) inflection. Note that Kalburdžu Romani retains, by and large, the normal word order of Balkan Romani dialects, and does not adopt verb-final order from Turkish. What might motivate speakers to adopt Turkish verb inflection despite the constraints discussed above on maintaining predication integrity in bilingual situations? It appears that the key to the process is the acceptance of Romani-Turkish bilingualism as a constituting aspect of group identity. Both languages are spoken interchangeably in the home as well as within the immediate community. When asked in the majority language, Bulgarian, to answer a question in ciganski (i.e. “the Gypsy language”), consultants working with fieldworkers of the Romani Project in Manchester (http://romani.humanities.manchester.ac.uk) often responded in Turkish rather than in Romani. This full acceptance of bilingualism and the fact that there is, effectively, no domain separation and no contextual demarcation between Romani and Turkish in community-internal interaction, blurs the boundaries within the bilingual repertoire. It grants speakers a license to initiate the predication at least in groupinternal communication in either language, regardless of any stylistic affect or accommodation-driven goal. When choosing a particular lexical verb, speakers are thus at liberty to employ the finite verb inflection system that is most easily associated with that verb. Full bilingual proficiency throughout the community supports these choices and helps to maintain a “double” system, at least for a certain period. In the Romani dialect of Ajia Varvara in Greece (Igla 1996), Turkish verbs retain Turkish inflection several generations after emigration from Turkey and loss of competence in Turkish. However, the number of verbs conjugated in this way remains small. This is also the case in Crimean Romani, which had been influenced by another Turkic language, Tatar. Even these rather exceptional cases do not exemplify the diffusion of borrowed morphological inflection markers into the inherited component of a recipient language. Instead, what we have here is a compartmentalization within the language’s morphology with borrowed morphology being incorporated on a wholesale rather than selective basis, but at the same time limited to just a particular set of lexical items, namely those that are themselves borrowed. Full community bilingualism and a strong sense of a hybrid identity – as “local Greek Gypsy musicians”, and as “Turkish Gypsies”, respectively – are the sociolinguistic triggers and apparently also the pre-conditions for the relaxation of the constraints that usually prevent any significant compromise on the integrity of inflectional morphology in the predication. Romani shows at least two other noteworthy cases of borrowed inflectional morphology. All Romani dialects use the Greek-derived nominal ending ‑o(s)



Why is the borrowing of inflectional morphology dispreferred?  

 71

and some also ‑i(s) with Greek-derived masculine nouns as well as with masculine loan nouns that are borrowed from subsequent European contact languages (through contacts that followed the dispersion of Romani populations from Byzantium around the late fourteenth century), e.g. prezidentos ‘president’. Some Romani dialects replicate the Greek third person singular conjugation ending ‑i with borrowed verbs, in a similar fashion. The adoption of the borrowed markers for use with borrowed vocabulary from new contact languages testifies to their productivity in the language. Nonetheless, in both cases they serve to maintain the compartmentalisation of the language’s lexical components. In this way, somewhat paradoxically, the blurring of the boundaries between Romani and its earlier contact language Greek (which enabled the borrowing of these Greek markers) now actually serves to sustain the boundary between Romani and its subsequent contact languages by identifying non-inherited vocabulary as belonging to a distinct inflectional class (cf. the Korlai Portuguese case discussed by Clements and Luís, this volume). My final example concerns another type of morphological compartmentalisation. It resembles the first, exemplified by the Romani dialects of the Balkans (examples 15–17), in two aspects. Firstly, here too borrowed morphology is used on a wholesale basis by adopting a full morphological paradigm rather than just individual inflectional markers. Second, here too borrowed morphemes appear in very strict distribution, accompanying only borrowed lexical items. The examples are a sub-set of what is now commonly referred to as “Mixed Languages” (cf. Bakker and Mous 1994; Bakker and Matras 2003). The candidates are a number of small community languages from the Canadian Prairies (Michif), the North Pacific (Copper Island Aleut), and Australia’s Northern Territory (Gurindji Kriol and Light Walpiri). Unlike the cases of the Romani dialects of the Balkans, these languages show a tendency toward an etymological compartmentalisation of inflectional morphology among word classes and not just among lexical items within the same verb class. In all four cases the division is between nominal and verbal inflectional morphology. In Michif (Bakker 1997), lexical verbs and verb inflection derive from Cree, while lexical nouns and nominal inflection derive from French. In Copper Island Aleut (Golovko and Vakhtin 1990), the speakers’ stronger language appears to be Russian, but in their in-group variety they integrate Aleut nouns and nominal c­ onstructions along with their Aleut nominal inflection. Gurindji Kriol (McConvell and Meakins 2005; Meakins, this volume) and Light Walpiri (O’Shannessy 2005) both employ verbs and verb inflection from Kriol, an English-based Australian creole, but they integrate nouns and nominal constructions along with their nominal inflection markers from the respective Australian Aboriginal languages, Gurindji and Walpiri.

72 

 Yaron Matras

Before discussing examples for this kind of compartmentalisation we need to briefly review what Mixed Languages are and how they come about. One of the varieties often cited in the literature as a Mixed Language is so-called “Angloromani” (users usually refer to this form of speech as “Romanes”, “Romani”, or “English Romanes”). Structural compartmentalisation in Angloromani is quite straightforward (cf. Matras 2010): Grammatical structures are exclusively based on English. The exceptions are occasional relaxation of some of the rules of English morpho-syntax, in particular omission of definite articles and of the present-tense copula in short utterances that convey mostly warnings or directives, and use of an independent negator kek in pre-verbal position without the auxiliary “do” (kek jins ‘doesn’t know’, literally ‘no knows’). The principal characteristic feature of Angloromani is the insertion of Romani-derived lexical vocabulary into English utterances: (18) Maw rokker, let mandi rokker, til ya chib! ‘Don’t talk, let me talk, cut your tongue!’ (19) We call a bad rakya what likes loads of mushes a ‘chikla luvni’ ‘We called a bad girl what likes loads of men a ‘dirty whore’.’ (20) Jel cause mandi’s gonna del dobba akai! ‘Go away cause I’m gonna hit this one here!’ Thomason and Kaufman (1988: 103) had described Anglo-Romani as a case of “inherited vocabulary, borrowed grammar”, where “the entire grammar has been borrowed (in effect) from English” (Thomason and Kaufman (1988: 7). This depiction can only be supported if under ‘borrowed’ we were to understand a process by which a population maintains its ethnic identity but loses its ethnic community language. Thus the English Gypsies, who until the second half of the nineteenth century spoke Romani as their community language in domestic contexts, but then shifted to English, would be described as having “borrowed” English (but maintained sporadic use of Romani-derived vocabulary). This definition is tricky. If followed consistently, it could lead us to characterise the population of Ireland as having “borrowed English grammar and vocabulary” but maintained Irish prosody and some elements of Irish information structure, along with individual vocabulary items from Irish. The more attractive way to understand Angloromani is as a case of language shift. There is no evidence of any wholesale borrowing of English grammar into English Romani. Documentation of inflected Romani in Britain from before the mid-nineteenth century shows borrowing of English lexicon, some derivational morphology, and a few semi-bound ­morphemes such as genitive ‑s as well as the noun plural ending -s (see Matras 2010). This extent of grammatical borrowing is well attested in other Romani dialects as well.



Why is the borrowing of inflectional morphology dispreferred?  

 73

Angloromani emerged when Romani was abandoned by a younger generation of speakers. These speakers retained a selection of Romani vocabulary as a stylistic device. (Individual middle-aged users now know on average around 350–450 vocabulary items.) It is used as a kind of group-internal emotive mode of speech, sometimes but not exclusively in order to conceal the content of the conversation from non-Romani bystanders. Most importantly, users of Angloromani employ Romani-derived lexical items by choice and at their discretion within their English discourse. Angloromani is therefore not a case of wholesale grammatical borrowing from English. It is rather a case of occasional and selective insertion from among an inventory of special vocabulary items, derived largely from Romani, into English conversation. We might interpret the emergence of other so-called Mixed Languages along similar lines. The younger generation has adopted the language of the surrounding speech community as its principal everyday language, but retains some knowledge of the heritage language used by the older generation. Younger speakers develop a mode of speech through which they flag their loyalty to their community heritage, inserting lexical items from this language into their default everyday language. These items are accompanied by inflectional morphology that is replicated from the source language. Nouns appear to be, as stable referents, more attractive targets for regular emblematic insertions of this kind. They are accompanied by nominal morphology, which eventually, becomes the conventionalized grammatical feature of the mixed variety: (21) Light Walpiri (O’Shannessy 2005: 49–50) (Walpiri-derived items are in bold face) fence-rla yu-rra shat‑im‑ap ngula‑j fence-loc 2sg‑nfut shut‑tr‑up anaph‑foc ‘Lock that one up inside the fence.’ (22) Gurindji Kriol (McConvell and Meakins 2005: 11) (Gurindji-derived items are in bold face) nyawa-ma wan karu bin plei-bat pak-ta nyanuny this-top one child past play-cont park-loc 3sg.dat warlaku-ywaung-ma dog-having-top ‘This one kid was playing at the park with his dog.’ Where speakers have access to the inflectional morphology of the “older” heritage language, its use allows them to authenticate lexical insertions from that language. In Copper Island Aleut, the effort to flag selective replication of the heritage language in a situation of language shift is characterised by an additional social feature, namely the gender split in the parent generation (the mothers

74 

 Yaron Matras

are Aleut-speakers, the fathers are Russian). The founder generation of Michif speakers based their utterances on Cree, the language of their mothers, for verbal predications, while selectively replicating nominal lexicon from French, the language of their fathers. This French lexicon was authenticated by French-derived nominal morphology. In Michif this compartmentalised mode of speech was eventually conventionalised and exceptionally transmitted further across several generations. While the sociolinguistic profiles vary, in all these cases of morphological compartmentalisation based on word class, it is the nominal or argument morphology that is replicated from the “weaker” language, while the verb or predication morphology appears consistently in the “stronger” language: Gurindji Kriol and Light Walpiri represent cases of language shift from the “old” heritage language to a “new” community language, Kriol, formed during the process of socio-cultural immersion and a partial erosion of traditional ethnic community structures. Copper Island Aleut represents the adoption of Russian colonial culture in the region. Michif represents the prevalence of a first nation identity in a community that absorbed a small population of French settlers. Angloromani, of course, matches this pattern, too: Verb inflection is consistently English, the target of language shift. The question arises why, in those cases where morphological compartmentalisation occurs, is it specifically nominal morphology that is incorporated from the “weaker” language? The answer lies, in my view, once again in the role of the predication as the carrier of the propositional content of the utterance. The language of the predication is the language of the utterance. When multilingual speakers are under pressure to make a choice rather than maintain stable bilingualism, they flag that choice through the grammar of the predication. The predication language flags their preferred sense of belonging in the subtle hierarchy of competing identities. We find compartmentalized inflectional morphology in a few communities in which language practices symbolise speakers’ reluctance to renounce their hybrid identity entirely. Loyalty toward a second heritage language is flagged by authenticating lexical insertions through inflectional morphology in a way that does not rupture the integrity of the predication. Remarkably, there is no known Mixed Language that shows an etymological split among tenses, persons, aspects or other features of the grammar of lexical predications, or where verbal and nominal inflection paradigms are not compartmentalized but randomly mixed from both etymological sources (some Mixed Languages, like Angloromani, Media Lengua, and Maa, are of course consistent in showing inflectional morphology from just one source language and derive just their lexicon from another). Even the extraordinary structural profiles of Mixed Languages are therefore not random outcomes of “lenient” attitudes



Why is the borrowing of inflectional morphology dispreferred?  

 75

toward language mixing, but products of constraints on language processing and the functional roles that structural categories have in communication. Given the motivation to align the grammar of the predication (represented by inflectional morphology on the verb) with the language that is gaining or maintaining stronger ground, speakers are left to draw on nominal inflectional morphology if they wish to consciously authenticate insertions as a way of flagging ethnic hybridity or split cultural loyalties. In the case of Angloromani, where a similar motivation may have been behind the emergence of a mixed code with English as the predicate language, the option of authenticating lexical insertions by drawing on Romani nominal inflection was not available, since English Romani had lost most of its nominal inflection even before Romani was abandoned as the default language of domestic interaction (cf. Matras 2010).

7 Concluding remarks Straightforward examples of morphological borrowing are French-derived -able in English loveable, Yiddish-derived ‑le in Modern Hebrew xamuda‑le ‘sweety’, Polish-derived do‑ in Polish Romani do-resel ‘to obtain’, Arabic-derived ‑āt in Jordanian Domari lačiy‑āt ‘girls’, Greek-derived ‑iz‑ in Spoitori Romani konduč‑iz‑av ‘I drive’, and many more. All these examples involve either derivational affixes, or borderline affixes with both derivational meaning and inflectional function (as in the case of the Arabic plural marker in Domari), or affixes of inflectional origin but derivational function (as in the case of the Greek tense-aspect affix serving as loanverb adaptation marker in Romani). Straightforward cases of borrowed inflectional morphemes are hard to find, and the examples discussed above are the exceptions that prove the rule. Borrowed inflectional morphemes are usually limited in their distribution to borrowed vocabulary and do not diffuse to inherited lexemes. Where diffusion of individual borrowed morphemes to inherited lexemes is attested, it is due to a close structural similarity between the borrowed form and the corresponding inherited affix. In effect these are cases of structural modifications of inherited forms based on analogy, rather than actual adoptions and replications of borrowed morphemes. Usually, borrowed morphemes are employed on a wholesale basis either with a closed class of items (such as borrowed modal auxiliaries), or with a particular word class (such as nouns), or with borrowed lexemes belonging to a particular word class (e.g. borrowed lexical verbs or borrowed adjectives). Many of the cases described above, especially those of Mixed Languages, have been discussed before and the structural facts of their morphological admixture are well described. But the implications remain controversial. The small class of “Mixed Languages” has been given this label precisely because the

76 

 Yaron Matras

languages involved display structural mixtures that are extraordinary. Yet even Mixed Languages do not offer examples of straightforward borrowing of inflectional morphemes. Rather than point to Mixed Languages and a modest number of other exceptional cases as proof that inflectional morphology can be borrowed, it seems more interesting and more challenging to try and explain why these cases are so exceptional and why even here we do not encounter the kind of borrowing behaviour that is attested for derivational morphology. The sociolinguistic situation, language loyalty, and “linguistic ecology” are often cited as key factors toward understanding the particular contact behaviour of some of the cases discussed above (e.g. Adamou 2012). There is no doubt that language attitudes and community norms license and regulate the structural choices that speakers make in conversation. This kind of approach was strongly supported above. Yet community attitudes and norms do not provide a sufficient explanation as to why derivational morphology should behave so differently from inflectional morphology. In order to answer this question we must consider the role of morphological paradigms in regulating language-processing tasks in communicative interaction. This, their “inner” function, provides the only link between the social reality of communicative settings (and attitudes and norms that accompany them), and the structural changes that individual linguistic categories undergo. Derivational morphology is a tool that modifies meaning and shapes lexical representations. Inflectional morphology plays a key role in initiating and anchoring the predication in the interaction context (in relation to the interaction role of participants and to presuppositions). In this way it also encodes and signals the language choices that bilingual speakers make in response to the interaction context: The choice of language is represented in the first instance through the choice of predication grammar. The need to protect the integrity of the predication grammar arises from the need to maintain clear language choices. Speakers may license themselves to abandon such boundaries and to flag a hybrid identity by authenticating lexical insertions. To this end, they may replicate inflectional morphology from the source language to accompany such insertions. But even in situations where boundaries are deliberately or consciously blurred, a hierarchical relationship is still maintained among the languages in the repertoire. This hierarchy is represented by the compartmentalisation of morphological components, which continues to protect the integrity of the predication grammar. The motivation to “borrow” inflectional morphology is inherently linked to re-negotiating language boundaries, which in turn is part of a process of re-negotiating identity. Whereas the purpose of borrowed derivational morphology is to replicate procedures of meaning derivation from the source language in the recipient language, ­ oundaries. the purpose of borrowed inflectional morphology is to re-draw social b

Why is the borrowing of inflectional morphology dispreferred?  



 77

The outcome of borrowing processes of inflectional morphology is therefore quite different from that of processes of borrowing that affect other structural categories, including the borrowing of non-inflectional (derivational) bound morphology. The principal distinction can be seen in the need to maintain a wholesale alignment between sets or paradigms of inflectional morphology and word classes or lexical sets. The borrowing of individual inflectional morphemes is, for this reason, strongly dis-preferred.

Abbreviations abl acc anaph‑foc comp cont dat def f fut gen itr loan loc m neg neutr nfut nom obl part past pl refl sg top tr

ablative accusative anaphora-focus complementiser continuous dative definite feminine future genitive intransitive loan verb marker locative masculine negation neuter near future nominative oblique particle past tense marker plural reflexive singular topic transitive

78 

 Yaron Matras

References Adamou, Evangelia. 2012. Verb morphologies in contact: Evidence from the Balkan area. In Martine Vanhove, Thomas Stolz, Aina Urdze & Hitomi Otsuka (eds.), Morphologies in contact, 143–162. Berlin: Akademie Verlag. Backus, Ad. 1996. Two in one. Bilingual speech of Turkish immigrants in the Netherlands. Tilburg: Tilburg University Press. Bakker, Peter. 1997. A language of our own. The genesis of Michif – the mixed Cree-French language of the Canadian Métis. New York: Oxford University Press. Bakker, Peter. 2006. The Sri Lanka Sprachbund: The newcomers Portuguese and Malay. In Yaron Matras, April McMahon & Nigel Vincent (eds.), Linguistic areas. Convergence in historical and typological perspective, 135–159. Houndmills: Palgrave. Bakker, Peter & Maarten Mous (eds.). 1994. Mixed languages. 15 case studies in language intertwining. Amsterdam: Institute for Functional Research into Language and Language Use. Bakker, Peter & Yaron Matras (eds.). 2003. The mixed language debate. Theoretical and empirical advances. Berlin: Mouton de Gruyter. Blakemore, Diane. 2002. Relevance and linguistic meaning. The semantics and pragmatics of discourse markers. Cambridge: Cambridge University Press. Chamoreau, Claudine. 2012. Spanish diminutive markers -ito/‑ita in Mesoamerican languages. A challenge for acceptance of gender distinction. In Martine Vanhove, Thomas Stolz, Aina Urdze & Hitomi Otsuka (eds.), Morphologies in contact, 71–90. Berlin: Akademie Verlag. Dawkins, R. M. 1916. Modern Greek in Asia Minor. A study of the dialects of Sílli, Cappadocia and Phárasa with grammar, texts, translations and glossary. Cambridge: Cambridge University Press. Ehlich, Konrad. 2007. Sprache und sprachliches Handeln. Berlin: Walter de Gruyter. Elšík, Viktor & Yaron Matras. 2006. Markedness and language change: The Romani sample. Berlin: Mouton de Gruyter. Field, Fred W. 2002. Linguistic borrowing in bilingual contexts. Amsterdam & Philadelphia: John Benjamins. Gardani, Francesco. 2012. Plural across inflection and derivation, fusion and agglutination. In Lars Johanson & Martine I. Robbeets (eds.), Copies versus cognates in bound morphology, 71–97. Leiden & Boston: Brill. Golovko, Evgeni V. & Nikolai Vakhtin. 1990. Aleut in contact: The CIA Enigma. Acta Linguistica Hafniensia 22. 97–125. Green, David. 1998. Mental control of the bilingual lexico-semantic system. Bilingualism: Language and Cognition 1. 67–81. Gumperz, John. 1980. Discourse strategies. Cambridge: Cambridge University Press. Haspelmath, Martin & Uri Tadmor (eds.). 2009. Loanwords in the world’s languages: A comparative handbook. Berlin: De Gruyter Mouton. Heine, Bernd & Tania Kuteva. 2005. Language contact and grammatical change. Cambridge: Cambridge University Press. Igla, Birgit. 1996. Das Romani von Ajia Varvara. Deskriptive und historisch-vergleichende Darstellung eines Zigeunerdialekts. Wiesbaden: Harrassowitz. Kossmann, Martin. 2010. Parallel System Borrowing. Parallel morphological systems due to the borrowing of paradigms. Diachronica 27(3). 459–487. Lanza, Elizabeth. 1997. Language mixing in infant bilingualism. A sociolinguistic perspective. Oxford: Clarendon.



Why is the borrowing of inflectional morphology dispreferred?  

 79

Matras, Yaron. 1998. Utterance modifiers and universals of grammatical borrowing. Linguistics 36. 281–331. Matras, Yaron. 2002. Romani: A linguistic initroduction. Cambridge: Cambridge University Press. Matras, Yaron. 2007. The borrowability of grammatical categories. In Yaron Matras & Jeanette Sakel (eds.), Grammatical borrowing in cross-linguistic perspective, 31–74. Berlin: Mouton de Gruyter. Matras, Yaron. 2008. Cultural hybridity, contact, and contrast: Romani and Domari. Paper presented at the workshop on Language evolution and social ecology, Lacito Paris, October 2008. Video available on: http://lacito.vjf.cnrs.fr/colloque/ecologie/videos_ gestion/ym.htm Matras, Yaron. 2009. Language contact. Cambridge: Cambridge University Press. Matras, Yaron. 2010. Romani in Britain. The afterlife of a language. Edinburgh: Edinburgh University Press. Matras, Yaron. 2011. Universals of structural borrowing. In Peter Siemund (ed.), Linguistic universals and language variation, 204–233. Berlin & New York: De Gruyter Mouton. Matras, Yaron. 2012a. A grammar of Domari. Berlin: Mouton de Gruyter. Matras, Yaron. 2012b. An activity-oriented approach to contact-induced language change. In Isabelle Leglise & Claudine Chamoreau (eds.), Dynamics of contact-induced change, 1–28. Berlin & Boston: De Gruyter Mouton. Matras, Yaron & Jeanette Sakel (eds.). 2007. Grammatical borrowing in cross-linguistic perspective. Berlin: Mouton de Gruyter. Matras, Yaron & Maryam Shabibi. 2007. Grammatical borrowing in Khuzistani Arabic. In Yaron Matras & Jeanette Sakel (eds.), Grammatical borrowing in cross-linguistic perspective, 137–149. Berlin: Mouton de Gruyter. McConvell, Patrick & Felicity Meakins. 2005. Gurindji Kriol: A mixed language emerges from code-switching. Australian Journal of Linguistics 25. 9–30. Mithun, Marianne. 2012. Morphologies in contact: Form, meaning and use in the grammar of reference. In Martine Vanhove, Thomas Stolz, Aina Urdze & Hitomi Otsuka (eds.), Morphologies in contact, 15–36. Berlin: Akademie Verlag. Moravcsik, Edith. 1978. Universals of language contact. In Joseph H. Greenberg (ed.), Universals of human language, 94–122. Stanford: Stanford University Press. Muysken, Pieter. 2000. Bilingual speech. A typology of code-mixing. Cambridge: Cambridge University Press. Myers-Scotton, Carol. 1993. Duelling languages. Grammatical structure in codeswitching. Oxford: Oxford University Press. Myers-Scotton, Carol. 2002. Contact linguistics. Oxford: Oxford University Press. Myers-Scotton, Carol & Janice Jake. 2000. Four types of morpheme: Evidence from aphasia, codeswitching, and second language acquisition. Linguistics 38. 1053–1100. O’Shannessy, Carmel. 2005. Light Walpiri: A new language. Australian Journal of Linguistics 25. 31–57. Paradis, Michel. 2004. A neurolinguistic theory of bilingualism. Amsterdam & Philadelphia: John Benjamins. Poplack, Shana, David Sankoff & Chris Miller. 1988. The social correlates and linguistic processes of lexical borrowing and assimilation. Linguistics 26. 47–104. Rehbein, Jochen. 1977. Komplexes Handeln. Elemente zur Handlungstheorie der Sprache. Stuttgart: Metzler.

80 

 Yaron Matras

Rose, Françoise. 2012. Borrowing of a Cariban number marker into three Tupi-Guarani languages. In Martine Vanhove, Thomas Stolz, Aina Urdze & Hitomi Otsuka (eds.), Morphologies in contact, 37–69. Berlin: Akademie Verlag. Sacks, Harvey, Emanuel A. Schegloff & Gail Jefferson. 1974. A simplest systematics for the organization of turn-taking in conversation. Language 50. 696–735. Simango, Silvester Ron. 2000. ‘My Madame is fine’: The adaptation of English loans in Chichewa. Journal of Multilingual and Multicultural Development 21. 487–507. Stolz, Thomas. 2012. Survival in a niche. On gender-copy in Chamorro (and sundry languages). In Martine Vanhove, Thomas Stolz, Aina Urdze & Hitomi Otsuka (eds.), Morphologies in contact, 93–135. Berlin: Akademie Verlag. Tenser, Anton. 2008. The Northeastern dialects of Romani. Manchester: University of Manchester PhD dissertation. Theodosiou, Aspasia. 2004. ‘Be-longing’ in a double-occupied space: The Parakalamos Gypsy musicians. Romani Studies 14. 25–58. Thomason, Sarah G. & Terrence Kaufman. 1988. Language contact, creolization and genetic linguistics. Berkeley: University of California Press. Tosco, Mauro. 2012. Swinging back the pendulum: French morphology and de-Italianization in Piedmontese. In Martine Vanhove, Thomas Stolz, Aina Urdze & Hitomi Otsuka (eds.), Morphologies in contact, 247–262. Berlin: Akademie Varlag. van Hout, Roeland & Pieter Muysken. 1994. Modelling lexical borrowability. Language Variation and Change 6. 39–62. Weinreich, Uriel. 1953. Languages in contact. The Hague: Mouton.

Part II Borrowing of derivation

Eleanor Coghill

Borrowing of verbal derivational morphology between Semitic languages: the case of Arabic verb derivations in Neo-Aramaic 1 Introduction While bound morphology in general is often acknowledged to be relatively resistant to borrowing, borrowing of derivational nominal morphology is common (Matras 2009: 209–210). The borrowing of verbal derivational morphology, on the other hand, is thought to be more limited (Matras 2009: 211). An area for which it has been established is the borrowing of aktionsart-derivational preverbs by Romani dialects from Slavic and Baltic languages, German and Greek (Ariste 1973; Schrammel 2005; Matras 2009: 211 etc.).1 Nevertheless a fuller picture of this cross-linguistically diverse area is not yet available. This paper will present cases of verbal derivational morphology that have been borrowed from Arabic by three distinct modern Aramaic languages. It will go on to show that each one represents a different stage in the process of integrating borrowed derivations into the native verbal system. Semitic languages, such as Aramaic and Arabic, present possibly unique challenges in the borrowing of verbal derivational morphology, due to their unusual non-concatenative root-and-pattern system. In this system, each verb lexeme is composed of a tri- or (less often) quadri-consonantal2 root (√C1-C2-C3, or √C1-C2-C3-C4, e.g. Arabic √ktb, √zxrf, North-Eastern Neo-Aramaic √bšl, √šxlp) and a derivation (known in different traditions as “stem”, “form”, “binyan” or “theme”). Derivations are associated with values such as transitivity, causativity,

1 Some examples in Russian Romani are: dava ‘I give’, dodava ‘I add’, obdava ‘I embrace’, otdava ‘I confiscate’ (Matras 2009: 211). Lithuanian dialects and Livonian have also borrowed such preverbs, from Slavic and Latvian respectively (Sivers 1971; Wiemer 2009). I am grateful to Peter Arkadiev for bringing to my attention the literature on borrowed preverbs. 2 Semivowels /y/ and /w/ can also count as radicals: in some forms they are realized as ­vowels. There is controversy over the traditional representation of Semitic morphology as root-based rather than word-based (for a general discussion see Ussishkin 2006; for the word-based view, e.g. Bat-El 1994 and Ussishkin 1999; for a more root-based view, e.g. Bolozky 2003; Prunet 2006; Faust and Hever 2010). I follow the latter (as well as traditional Semitists) in adopting the root-based analysis.

84 

 Eleanor Coghill

passivity etc. They tend to be of low productivity – for each root the existence of a given derivation is not usually predictable. The precise meaning can also be arbitrary, although tendencies can be identified. To form a verb, the root is fitted into a template, based on a vowel pattern, as well as prefixes and suffixes. The template of the stem is chosen according to the derivation and “tense” (conjugation expressing tense/aspect/mood or TAM). The root consonants (“radicals”) are slotted into the template. Typically person is expressed by prefixes and suffixes, selected according to tense. The basic tenses of earlier Semitic languages are known as the Prefix Conjugation (personal inflection marked by prefix and suffix) and Suffix Conjugation (personal inflection marked only by suffix). These conjugations have been replaced by newer ones in some modern Aramaic dialects. To illustrate the root-and-pattern morphology, we will show the inflection of the North-Eastern Neo-Aramaic (NENA) root √bšl in two of the three triradical NENA derivations. Verbs in NENA (Telkepe dialect) are mainly formed on two stems, the Past Base (expressing past perfective) and the Present Base (typically expressing present, irrealis or imperfective). In derivation I these have the templates CCəCand CaCC- respectively; in derivation II, they take the forms mCuCəC- and mCaCC-. The two bases each take different person inflection (known as L- and S-suffixes). The verb bšl I (i.e. derivation I) means ‘to cook (intr.)’, while the derived verb bšl II means ‘to cook (tr.)’. Thus, ‘it (3ms.) cooked (intr.)’ would be stem bšəl- plus 3ms. L-suffix ‑lə, i.e. bšəllə. ‘I (m.) cook (tr.)’ would be formed from the indicative prefix k(ə)-, stem mbašl- plus 1ms. S-suffix ‑ən, i.e. kəmbašlən. When a verb lexeme is borrowed into a Semitic language, a tri- or quadriradical root must first be identified and extracted.3 What complicates the picture is that the root must then be assigned to one of the derivations. When verbs are borrowed from non-Semitic languages, the root, once extracted, must be given a derivation. For instance, English charge has been borrowed into a NENA dialect

3 Another strategy bypasses this problem altogether: a borrowed (verbal) noun is combined with an inherited light verb. In Neo-Aramaic, this strategy is mostly attested with verbs which were already in this form in the donor language, such as Persian, Kurdish or Turkish compound verbs. For example, Sorani Kurdish seyr kirdin ‘to look’ (lit. ‘to make a look’, Qazzaz 2000: 378) has been bor-rowed into Jewish Sulemaniyya Neo-Aramaic as sayr ʾwl I ‘to look’, lit. ‘to make sayr’: the noun is borrowed but the light verb is Aramaic (Khan 2007: 209–210). In Maltese, on the other hand, newer loanverbs take Maltese inflection for the most part, but verbal stems are not adapted to the root-and-pattern morphology (see, e.g., Hoberman and Aronoff 2003).



Borrowing of verbal derivational morphology between Semitic languages 

 85

by speakers in Detroit as črj II ‘to charge’. The following example is a Present Base form, with the derivation II stem mCaCC-: (1) NENA (Telkepe dialect) kə-mčarj-i-lə. ind-charge.pres-3pl-obj.3ms ‘They charge it up.’ (adapted from author’s fieldwork) When verbs are borrowed from one Semitic language to another, as in the case of Arabic loanverbs in Neo-Aramaic dialects, they already belong to a derivation. In many cases there is a (morphologically or functionally) similar – and most likely cognate – derivation in the Aramaic dialect, to which the loanverb could be assigned. However, the Semitic languages have far from identical inventories of verbal derivations, and even those which are historically cognate may vary considerably in form and in function. For instance, Arabic has ten derivations, excluding the very rare, while most North-eastern Neo-Aramaic dialects have only three. When a verb lexeme is borrowed from an Arabic derivation which has no equivalent in Neo-Aramaic, one of two things happens: either the root is inserted into a native derivation, or the verb lexeme is borrowed with its derivational morphology. Thus in (2a), Arabic frj v4 (tafarraja) ‘to watch’ is borrowed into the non-cognate but morphologically somewhat similar NENA derivation II: that is, its root is borrowed and fitted into the templates of the native derivation. In (2b), however, the derivational morphology itself, that is the infixed ‑t‑ of the Arabic derivation viii, is imported with the lexeme: (2) NENA (Telkepe dialect) a. mfarj-ux! ( frj II