Aspects of Language Contact: New Theoretical, Methodological and Empirical Findings with Special Focus on Romancisation Processes 9783110206043, 9783110195842

Citation preview

Aspects of Language Contact

≥

Empirical Approaches to Language Typology 35

Editors Georg Bossong Bernard Comrie Yaron Matras

Mouton de Gruyter Berlin · New York

Aspects of Language Contact New Theoretical, Methodological and Empirical Findings with Special Focus on Romancisation Processes edited by Thomas Stolz Dik Bakker Rosa Salas Palomo

Mouton de Gruyter Berlin · New York

Mouton de Gruyter (formerly Mouton, The Hague) is a Division of Walter de Gruyter GmbH & Co. KG, Berlin.

앝 Printed on acid-free paper which falls within the guidelines of the 앪 ANSI to ensure permanence and durability.

Library of Congress Cataloging-in-Publication Data is available at http://catalog.loc.gov. Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data is available in the Internet at http://dnb.d-nb.de.

ISBN 978-3-11-019584-2 ISSN 0933-761X © Copyright 2008 by Walter de Gruyter GmbH & Co. KG, D-10785 Berlin. All rights reserved, including those of translation into foreign languages. No part of this book may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording or any information storage and retrieval system, without permission in writing from the publisher. Printed in Germany.

Preface Language contact is short for contact between speech communities and the occurrence of multilingualism. Such contact, if sufficiently intense and widespread, will influence the lexicon and grammar of the languages involved, more or less bi-directionally when there is socio-economic equivalence between the speech communities, and mainly uni-directionally if this is not the case. The existence of modern nation-states, often founded with a specific ‘national’ language as one of the binding factors, may create the impression that most speech communities to date are essentially monolingual, and most speakers are confronted with just one language on an everyday basis. And in fact, virtually all 192 countries recognized by the United Nations as we are writing have one official language, Paraguay and Malta being among the few exceptions. But with almost 7000 languages still extant, this means that there are around 35 non-official languages per country on average. This is not considering the countless dialects that are often fundamentally different from the official or major dialect to the extent of not being mutually intelligible. It also means that for the speakers of at least some 97% of the world’s languages some form of diglossia and bilingualism is the norm rather than the exception. The lexicon and structure of these languages are under continuous influence of one or more other languages, either genetically related or not. Over time, this must have caused a tremendous amount of language change. In all likelihood, such has been the situation for hundreds if not thousands of years, with languages in the role of donor or receiver or both. We may therefore safely assume that there exists virtually no ‘pure’ language in the world in the sense that the only changes were internal and diachronic. Areal, i.e. contact induced aspects of languages, be they substratal, superstratal or adstratal may therefore well turn out to be as crucial to establish language typologies as are genetic and diachronic aspects, be it that there may be considerable differences between languages, and between the different components of their grammars and lexicons in this respect. And on a more speculative note, one might also expect differences in what may and what may not be borrowed from other languages depending on the original typologies of the source and the target language in a specific contact situation. In order to discuss the impact that contact may have on language structure we organised two conferences on this topic, the first one in Amsterdam (May 2004; Dik Bakker and Jorge Gómez Rendón) and the second one in

vi Preface Bremen (May 2005; Thomas Stolz). Both conferences focussed on contact situations in which one of the languages is of Romance ancestry – more specifically French, Italian, Portuguese and Spanish, thus controlling in a way for the typology of at least one of the contributing languages. This might make the analyses of the results of these contacts more comparable, and may give a better idea of what may and may not be borrowed by different types of languages. Many of the articles in this volume are based on papers presented at one of these conferences. Other articles originating from one of these meetings appear in Stolz, Bakker and Palomo (2008a) – exclusively on borrowing from Spanish – and Stolz, Bakker and Palomo (2008b), on borrowing from Romance languages in Africa. The contributions to the present volume fall in two categories. The first seven articles are of a more general nature and discuss language contact and borrowing as a result of it from a more theoretical angle. In his introductory article, Thomas Stolz defines Romancisation – borrowing from a Romance language – in its prototypical form as ‘every process which leads to at least the stable and not individual partial copying of an overt or covert feature of an ideally identifiable Romance donor language in an identifiable and already existing non-Romance recipient language, regardless of the pre-history of the Romance feature itself.’ Using the method of parallel corpora, in this case by comparing translations of the same text in a number of non-Romance languages, he gives a number of examples of borrowing from one of the Romance languages, both lexical and non-lexical. He proposes a large scale project of this type to answer the major questions about Romancisation, and borrowing in general. Martin Haspelmath explores the systematicity of lexical borrowing. He argues that one of the central tasks of the diachronic study of language is to determine the constraints on borrowing. The lexicon, and more specifically the part of it that may be considered as basic, takes a central position in such studies. After providing definitions for the different kinds of lexical borrowing, he discusses the criteria for building a list of semantic fields and word meanings which can serve to study the borrowability of basic lexical items. Examples are given of the Loan Word Typology project (LWT), organized by the Max Plank Institute in Leipzig. In the course of this project word forms were collected for around 40 areally spread languages on the basis of an extended Swadesh type list of around 1500 meanings. In contrast to LWT, which concentrates on lexical borrowing, Jeanette Sakel and Yaron Matras describe a project on Language Convergence and Linguistic Areas (LCLA) which attempts to establish a model for contactinduced change in the grammar. For these authors there is a crucial distinc-

Preface

vii

tion, not always made in the literature on language contact, between the borrowing of matter, i.e. actual linguistic items from a source to a recipient language (MAT) and patterns, abstract structures from a model to a replicating language (PAT). The LCLA collects MAT and PAT elements from a number of language contact situations. The method is demonstrated on the basis of two borrowing pairs: Mosetén/Spanish and Romani/Rumanian. In their contribution, Søren Wichmann and Jan Wohlgemuth aim to establish a descriptive framework for studying the way in which verbs from a source language may be borrowed by some recipient language. They distinguish four major strategies: direct insertion (adoption of bare verbal stems), indirect insertion (integration via derivation), light verb strategy (‘do’ support), and paradigm transfer (borrowing of complete verbal paradigms). Using a sample with data from over 80 languages in 112 donor-recipient combinations, the authors come to the conclusion that it is hard to find strong evidence for a typological explanation of the choice for one or more of the distinguished borrowing strategies for verbs. Areal phenomena and levels of bilingualism are mentioned as potential explanatory factors. Wolfgang Wildgen discusses models for language contact studies. Earlier models were typically inspired by the static view on language acquisition championed by the generative tradition. Wildgen argues that these are much too simplistic. Focussing on lexical borrowing, he shows that a finite state model based on sets of phonological and semantic features on the one hand and standard transformation rules on the other hand is not adequate to describe the complexities involved in the transfer of a lexical element from one language to another. As in studies in biological evolution a noncontinuous and dynamic approach should be taken when modeling sociolinguistics and more in particular language contact. Wildgen argues for a model based on dynamic systems theory, which takes a multitude of relevant domains into consideration. The last contribution in the theoretical section of the book comes from Klaus Zimmermann. He presents a constructivist theory on language contact. The basic assumption is that all factors that contribute to linguistic behaviour are neurophysiological, and should get a purely cognitive explanation. There are two overall factors that determine the outcome in bilingual situations. First, the different sets of experiences individuals have when building up their knowledge of the languages spoken in their community. And second, the overall desire of speakers to reduce the difference between these individual systems. Together they explain for the occurrence and the specific shapes of interference in L2, borrowing in L1, and blends as in code switching and code mixing. Zimmermann exemplifies this take on

viii Preface language contact with some cases of Romancisation, and more specifically Hispanization. After these contributions that concentrate above all on the theoretical aspects of language contact and borrowing follow seven articles that give more weight to the empirical side of the matter. Dik Bakker, Ewald Hekking and Jorge Gómez Rendón introduce a framework for studying lexical and structural borrowing as a result of language contact. The results of fieldwork on three typologically different languages from the Americas – Otomí (Mexico), Quichua (Ecuador) and Guaraní (Paraguay) – involving over 100 native speakers were collected in a database. Employing a computer program specifically developed for this kind of analysis the authors argue that there are differences between the three languages in their borrowing behavior from Spanish that can not be explained on the basis of chance alone. Taking insights from linguistic typology as a point of departure, they show that much of the variation between the languages may be explained on the basis of typological differences. The observations get support from the fact that the borrowing behavior is consistent over different dialects of the respective languages. Peter Bakker and Robert Papen discuss the influence of French on the native languages of Canada and adjacent parts of the USA. They give a sketch of the early contacts between the original population of these areas and the French settlers who entered from around 1600 onwards, and the spread and later decline of the French language. The authors continue with an overview of the current sociolinguistic situation of the native languages of these territories. Then follows an inventory of the typical French borrowings that are reported in the literature, with a characterization of the phonological transformations that these elements undergo in the respective languages. They close with a discussion of code switching observed in these contact situations, and a characteristic of Michif, a mixed language that more or less exclusively pairs Cree verbs to French nouns. Moving back to South America, Stefan Dienst investigates contact phenomena in Kulina, an indigenous language from the Madihá branch of the Arawan family spoken in Brazil near the Peruvian border. The situation of Kulina is rather unique since, different from the majority of the indigenous languages of the area, most speakers are monolingual today, and only a few have a good command of Portuguese. Nevertheless, Dienst shows that the language has undergone changes due to contact. The first important influence was from Nheengatu, the lingua franca of the area well into the 19th century. This is mainly attested in the lexicon. More influence comes from the new national language, Portuguese. This is mainly restricted to the lexicon (nouns,

Preface

ix

verbs and numerals) and a few additions to the relatively simple phonological system. Given the current sociolinguistic status of the language, it is expected that most of the borrowings will be incorporated into the Kulina system. In his contribution, John Holm discusses the fate of inflection in the process of creolization. Until recently, the general assumption among creolists was that original inflectional morphemes from the lexical donor language typically did not make it to the creole language which was based on it. However, some examples of early borrowing make clear that this position is not tenable. Holm wonders how this myth may have come into existence. Studying five typologically different cases of Portuguese based creoles, he shows that in at least two of them, aspects of Portuguese gender, number and tense marking have survived, and are not the result of decreolization. The cause for the original erroneous assumption may have been the fact that the bulk of our knowledge is still based on Caribbean and WestAfrican creoles, of which either the superstrate or substrate languages have very restricted inflectional characteristics close to being isolating. Claire Moyse-Faurie directs the focus on language contact in Oceania. She gives an overview of the contact history of the area from the 16th century onwards. First came the Portuguese and the Spaniards, then the Dutch, and finally the French and the English. The author shows that the different objectives of the respective expeditions – conversion; trade; colonization – is reflected in the types of borrowings found in the indigenous languages and the creoles of the area. She then discusses the effect of borrowing on a number of Polynesian languages. The phonological and morphological integration of French and English loanwords, the introduction of phonemes from these two languages and the typical semantic domains of borrowing are highlighted. Finally, there follows a characteristic of Tayo, a Frenchbased creole from Saint-Louis, and Kayafou, a version of French created by young Kanaks from New Caledonia. The article by Françoise Rose and Odile Renault-Lescure takes the reader to French Guyana, the home of two Amerindian languages, Kali’na and Emérillon, belonging to the Carib and Tupi families respectively. Living near the coast, the speakers of Kali’na had earlier and more intensive contact with Portuguese, Spanish, English, Dutch and French than the Emérillon, who live deeper into the rainforest. During the last century, both languages underwent a certain amount of Francisation. Furthermore, over the whole post-Columbian period, several creole languages played a role in the interaction with other speech communities, both indigenous and European. The authors study the similarities and differences in the borrowing processes that are the result of all these contact situations.

x Preface In the next contribution, Mauro Tosco takes a view at a much lesser studied contact situation. Late-comers on the stage of colonization, the Italians established themselves in today’s Eritrea and Somalia in the 1880’s, and even more recently in Ethiopia. The influence of Italian was strongest in Somalia, where it even was the second language for some time. Tosco analyzes several cases of Italian influence on Somali and Tigrinya. This is mainly found in the phonology and the lexicon of these languages. Interestingly, lexical borrowing seems to be restricted to nouns. Some Italian loans are now replaced by Arabic or English forms. A further analysis is made of Italian as it is still spoken in these countries. Some traits of pidginization are visible. The final contribution, by Christel Stolz, discusses the fate of gender distinctions on nominal borrowings, more particularly from Romance languages into German. It appears that borrowed nouns are subjected to the same set of semantic and formal criteria that co-determine the gender of native nouns. It is not very easy to establish a clear ranking pattern among these criteria. Apart from this strategy, there are also quite a few cases of gender copy, i.e. the original gender in the source language appears to be decisive. Apart from borrowing in Standard German, several cases are studied of borrowing by German varieties spoken in countries where a Romance language is the majority language. Especially in these cases, gender copying seems to be the favoured strategy. The organizers of the Amsterdam conference express their gratitude for the financial support received from the Dutch Research Council (NWO). The Bremen conference was made possible by a generous grant from the Volkswagen Foundation (Hannover). The editors are also grateful to Sonja Kettler and Cornelia Stroh whose technical expertise and editorial skills made this volume become a reality. Rachel Montague helped us to nativize the English of the contributions written by linguists with a first language other than English. A word of thanks is due to Mouton de Gruyter – especially Ursula Kleinhenz – for their non-bureaucratic way of handling things. The editors of Empirical Approaches to Language Typology deserve our gratefulness for accepting the book in their well renowned series. Georg Bossong painstakingly reviewed each and every contribution to this volume and made very helpful comments. Thomas Stolz, Dik Bakker, Rosa Salas Palomo Bremen (Germany), Lancaster (United Kingdom), Mangilao (Guam) November 2007

Preface

xi

References Stolz, Thomas, Dik Bakker and Rosa Salas Palomo (eds.) 2008a Hispanisation. The impact of Spanish on the lexicon and grammar of the indigenous languages of Austronesia and the Americas. Berlin / New York: Mouton de Gruyter. 2008b Romancisation in Africa. The influence of French, Italian, Portuguese and Spanish on the indigenous languages of Africa. Bochum: Brockmeyer.

Contents

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

v

Romancisation world-wide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thomas Stolz

1

Loanword typology: Steps toward a systematic cross-linguistic study of lexical borrowability. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Martin Haspelmath

43

Modelling contact-induced change in grammar . . . . . . . . . . . . . . . . . . . . . . Jeanette Sakel and Yaron Matras

63

Loan verbs in a typological perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Søren Wichmann and Jan Wohlgemuth

89

Why we need dynamic models for sociolinguistics and language contact studies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123 Wolfgang Wildgen Constructivist theory of language contact and the Romancisation of indigenous languages. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141 Klaus Zimmermann Spanish meets Guaraní, Otomí and Quichua: A multilingual confrontation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking French influence on the native languages of Canada and adjacent USA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239 Peter Bakker and Robert A. Papen Portuguese influence on Kulina . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287 Stefan Dienst Creolization and the fate of inflections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299 John Holm

xiv

Contents

Borrowings from Romance languages in Oceanic languages . . . . . . . . . . 325 Claire Moyse-Faurie Contact-induced changes in Amerindian languages of French Guiana . 349 Françoise Rose and Odile Renault-Lescure A case of weak romancisation: Italian in East Africa . . . . . . . . . . . . . . . . . 377 Mauro Tosco Loan word gender: A case of romancisation in Standard German and related enclave varieties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 399 Christel Stolz

Index of subjects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 441 Index of languages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 456 Index of toponyms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 462 Index of authors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 464 Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 470

Romancisation worldwide1 Thomas Stolz

1. Introduction The Romance phylum boasts a long and variegated history of language contacts which presumably began long before the oaths of Strasbourg were taken in AD 8422. In Europe, an extended border of several thousand kilometres not only separates, but also brings together Romance languages and members of the Celtic, Germanic, and Slavic phyla. Albanian, Greek and non-Indo-European languages such as Hungarian (Uralic), Maltese (AfroAsiatic) and the genetic isolate Basque have been involved in rather intensive contacts with various Romance languages. In addition to these neighbourhood relations on European soil, there are also long standing contacts across the Mediterranean Sea in which the Afro-Asiatic languages of the Semitic and Hamitic phyla participate. Romance languages are involved in various Sprachbünde: French, together with Dutch and German, belongs to the core of the SAE (Standard Average European) languages, the so-called Charlemagne Sprachbund. The other Romance languages outside the Balkans are also bona fide SAE languages (Haspelmath 2001). Rumanian however joins its non-Romance neighbours Bulgarian, Macedonian, Albanian and Greek in the Balkan Sprachbund (Sandfeld 1968). Practically all members of the Romance phylum partake in the Mediterranean Sprachbund, which is comprised of all Circum-Mediterranean languages and some other languages of the immediate hinterland (Ramat and Stolz 2002). From the 15th century onwards, four major Romance languages – Portuguese, Spanish, French, and relatively late, Italian – diffused into their respective zones of influence overseas during the first era of colonialism. Contact was made with indigenous languages in Africa, the Americas, South to East Asia and Oceania. Sometimes the Romance languages were affected by the adstratal, substratal and/or superstratal impact of their contact partners and through this distinct local varieties of Romance have developed.3 Moreover, in various places, new languages arose from the contact between Romancespeaking colonisers and speakers of non-European languages. Today, the number of Creole languages with a predominantly or partially Romance lexicon is impressive. Romance languages have also contributed substan-

2 Thomas Stolz tially to the re-shaping of already existing languages without creating a new Creole language. Directly or indirectly, Romance languages are the sources of a wide range of genetically unjustified traits in literally hundreds of languages worldwide. According to an informal count by myself, Romanisms of various kinds can be found in at least 17 of the world’s 20 macrophyla. The indirect way of Romancisation consists of, among other things, socalled international words and neologisms which come in a Romance shape, or of itinerant features. The direct way, on the other hand, requires speakers of Romance languages to interact with speakers of non-Romance languages under spatio-temporally well-defined circumstances which induce the non-Romance speakers to introduce Romance features into their own language – features which in turn are accepted by other members of the same non-Romance speech-community who do not necessarily have any competence in a Romance language. Thus, a comparative in-depth study of Romance languages functioning as donors in a variety of contact situations promises to contribute substantially to the search for an answer to Curnow’s (2001) question “What can be borrowed?”. On the one hand, Romance languages are sufficiently similar to each other on all levels of language to justify the idea that they offer basically the same options when recipient languages copy features from them. At the same time, the Romance phylum is internally heterogeneous enough to allow for at least low-level variation among the potential donors and thus provides a suitable criterion with which to determine whether or not differential behaviour of Romance donors can be attributed to the differences which exist between one Romance language and its next-of-kin. Boretzky and Igla (1994a) and even more convincingly, Johanson (2000) have demonstrated that it makes good sense methodologically to work with one constant and several variables when comparing language contact situations (cf. Bakker, Gómez Rendón, and Hekking, this volume). In these earlier studies, the authors looked at one and the same donor or recipient language or group of languages in as many contact constellations as possible. This procedure has yielded results on which generalisations about the contact behaviour of Romani and Turkic languages, respectively, can be based. Nothing remotely similar exists for Romance languages although the ground has already been prepared for a large-scale comparison of Romance donor language behaviour. A thorough check of how Romance languages fulfil the function of donor in contact situations constitutes a valuable contribution to the progress of contact-related linguistics both empirically and theoretically as it enables us to compare the new findings with the ones derived from the above mentioned research projects. However, before the new task can be

Romancisation worldwide

3

tackled, two issues have to be clarified. Firstly, an investigation of Romancisation in a global perspective cannot be carried out solely within the framework of Romance Philology as it requires the intensive co-operation of specialists from many philological disciplines (with their own often largely incommensurable traditions such as Oriental Studies, African Studies, Oceanic Linguistics, Dravidian Linguistics, etc.) and several sub-disciplines of general linguistics. Secondly, one must be aware of the fact that the term Romancisation itself is by no means self-explanatory. It needs to be explored in some detail (cf. also Zimmermann [this volume]). This is what this contribution is meant to achieve. For the present purpose, I employ an open sample of languages. For the donors, I draw mainly on the major standard varieties of Romance although an occasional example also stems from regional or minor varieties. The recipient languages have been chosen according to the criterion of maximal diversity, i.e., they cover the widest possible range of phyla, areas and types. In section 2, I discuss the notion of Romancisation. For practical reasons, I focus on lexical copies throughout section 2. In section 3, I address a small selection of phenomena of general interest which go beyond the realm of the lexicon. In section 4, I present conclusions which include a synoptic outline of what a project on Romancisation should be about. The bulk of my data are extracted from two large parallel corpora, viz. the translations of volume 1 of the Harry Potter series (= HP with numbers referring to pages) and the original and translations of Antoine de Saint-Exupéry’s Le petit prince (= LPP with Roman numbers referring to chapters and Arabic numbers to sentences). Whenever translations of these sample texts are not available for a given language or when the translations pose problems, I resort to the usual descriptive material. In terminological matters, I mostly make use of well established notions which do not need special explanations. There are two exceptions. I employ the term copy coined by Johanson (2000) in order to avoid the much too suggestive traditional term borrowing. Admittedly, copy is itself not always the most fortunate solution – especially if language contact leads to the loss of a feature in the recipient language. It sounds awkward to say that the recipient language has copied the absence of X of the donor language (with “the absence of X” being the feature being copied). As I do not dwell on loss via language contact in this contribution, the terminological choice is largely unproblematic. In my wording, features of language X are copied by language Y during contact. Feature is a handy cover term for all kinds of traits from phonology via morphosyntax to discourse-pragmatics etc. be they overt or covert. Copied features thus, may involve matter (= segmental chains) and/or pattern (= construction templates/form-function pairings) –

4 Thomas Stolz to use the terminological dichotomy introduced by Sakel and Matras (this volume). 2.

What is Romancisation and what is Romancised?

2.1. Starter kit Romancisation encompasses all those processes which involve the copying of Romance features into a non-Romance recipient language. Romancisation constitutes one of two important categories within the framework of Romance language contacts – with De-Romancisation representing the logical (though terminologically perhaps rather unfortunate) complement of Romancisation. Thus, the study of Romancisation will not tell us the whole story of what happens when Romance languages come into contact with other languages. Nevertheless, the phenomenology of Romancisation is so rich that investigating it more closely promises many important insights, not only into the mechanisms of Romance language contacts specifically, but also into the flexibility of human language(s) in general. The process of Romancisation with which linguists are most familiar, and of which even philological laymen are often aware, is simple lexical copies. Words of Romance origin abound in many languages with different genetic backgrounds worldwide. To start, I only consider words from the four major Romance languages outside the Balkans. Unsurprisingly, Romance-derived lexemes turn up in genetically, typologically and areally so diverse languages as, e.g., Albanian, Huave, Kabiyé and Indonesian, which are spoken in different regions of Europe, Africa, America and Asia.4 The donor language is French for (1), Spanish for (2), Portuguese for (3) and Italian for (4). This exemplification is not meant to be exhaustive – what it nevertheless shows is the contemporary, worldwide distribution of lexical Romanisms. (1)

Donor: French – West Romance, Gallo-Romance a. Yiddish (Indo-European, Germanic; Israel) [LPP Yiddish XXVII, 34] dos iz far mir der shenster un troyerikster peyzazh landscape this is for me the beautiful:SUP5 and sad:SUP oyf der velt on the world ‘For me, this is the most beautiful and the saddest landscape in the world.’

Romancisation worldwide

5

b. Kabiyé (Niger-Kordofanian, Gur; Togo) (Lébikaza 1992: 82) p yal k
m-á m n-t frère come-PERF my-POR brother son ‘My nephew (= son of my elder brother) has come.’ c. Vietnamese (Austro-Asiatic, Vietnamese; Vietnam) [HP Vietnamese, 49] thC không dán tem letter NEG EMPH stamp ‘There was no stamp on the letter.’ d. Malagassy (Austronesian, Malayan; Madagascar) (Rasoloson 1997: 35) tàiza no nìsy mpivàrotra biè ticket where REL EXI vendor ‘Where was there a ticket vendor?’ (2)

Donor: Spanish – West Romance, Ibero-Romance a. Huave (Amerindian, Huavean; Mexico) (Stairs and Stairs 1983: 60) ngome aaga mwestre neh a hwan NEG this teacher he the Juan ‘Juan is not the teacher.’ b. Rapanui (Austronesian, Polynesian; Chile/Rapa Nui) (Makihara 2001: 197) e acuerda no ‘& au e k/ era IMPERF remember LIM PROG I IMPERF say PPD ‘I said that I still remembered.’ c. Bubi (Niger-Kordofanian, Bantu; Equatorial-Guinea) (Bolekia Boleká 1999: 44) ë sikwèlla si tyí hállo AUG school NEG not_be there ‘The school is not there.’ d. Basque (isolate; Spain/France) [HP Basque, 7] Dursley jauna Grunnings izeneko enpresa Dursley Mr.:DEF Grunnings name:GEN firm bateko zuzendaria zen one:GEN director:DEF be.PAST.3SG ‘Mr Dursley was the director of a firm named Grunnings.’

6 Thomas Stolz e. Tagalog (Austronesian, Philippinian; Philippines) (Ramos 1971: 52) kinuha ng bata ang libro get:GOAL.FOC NON.FOC child FOC book ‘The child got the book.’ (3)

Donor: Portuguese – West Romance, Ibero-Romance a. Swahili (Niger-Kordofanian, Bantu; Tanzania) (Legère 1989: 117) ki-talu cha Asha ki-na-pendez-a CL.5SG-garden CL.5SG.of Asha CL.5SG-PROG-please-INDIC ‘Asha’s garden is liked.’ b. Tariana (Amerindian, Arawak; Brazil) (Aikhenvald 2002: 177) Lucia-nuku nu-pekaru-ka nu-manika Lucia-TOP.NON.A/S 1SG-joke-REC.P.VIS 1SG-play ‘I am playfully making fun of Lucia.’ c. Indonesian (Austronesian, Malayan; Indonesia) [HP Indonesian, 10] Mr Dursley selalu duduk membelakangi jendela di kantornya Mr Dursley always sit turned_around window in office di lantai sembilan in floor nine ‘Mr Dursely always sat with his back to the window in his office on the ninth floor.’ d. Afrikaans (Indo-European, Germanic; South Africa) [HP Afrikaans, 42] hy swaai die sambreel met ‘n swiepende beweging he swing the umbrella with a sweeping movement deur die lug through the air ‘He swung the umbrella through the air with a sweeping move.’

(4)

Donor: Italian – East Romance, Italo-Romance a. Maltese (Afro-Asiatic, Semitic; Malta) [LPP Maltese I, 10] jiena wrejt il-kapulavur tieg-i lill-kbar I show.PERF:1SG the-masterpiece of:1SG to:the-big.PL ‘I showed my masterpiece to the adults.’ b. Albanian (Indo-European, Albanian; Albania) [HP Albanian, 27] motoçiklet-at nuk fluturojnë motor_cycle-DEF.PL NEG fly:3PL ‘Motor-cycles don’t fly.’

Romancisation worldwide

7

The etymological sources of the Romance items are: – (1) a. b. c. d.

Yiddish peyzazh < French paysage ‘landscape’, Kabiyé frère ‘elder brother’ < French frère ‘brother’, Vietnamese tem < French timbre ‘stamp’6, Malagassy biè < French billet ‘ticket’;

– (2) a. Huave mwestre < Spanish maestro ‘teacher’, b. Rapanui acuerda ‘remember’ < Spanish (se) acuerda ‘s/he remembers’7, c. Bubi sikwèlla < Spanish escuela ‘school’, d. Basque enpresa < Spanish empresa ‘undertaking; firm’, e. Tagalog libro < Spanish libro ‘book’; – (3) a. b. c. d.

Swahili kitalu < Portuguese quintal ‘garden’8, Tariana pekaru ‘joke’ < Portuguese pecado ‘sin’9, Indonesian jendela < Portuguese janela ‘window’, Afrikaans sambreel ‘umbrella’ < Portuguese sombreiro ‘sun hat’10;

– (4) a. Maltese kapulavur < Italian capolavoro ‘masterpiece’, b. Albanian motoçikletë < Italian motocicletta ‘motor-cycle’. In none of the above cases does code-switching apply – they all are bona fide instances of full-blown copies, i.e., the Romance-derived words form part of the regular lexicon of native speakers of the recipient languages.11 Some, at times only minor, changes on the phonological, morphological and/or semantic level notwithstanding, these words are still clearly identifiable as overt copies from a Romance donor language. These copies involve only content words – predominantly nouns from the donor language. In terms of contact-linguistic typologies, this is nothing spectacular because the presence of copied nominal content words in a recipient language is the expected picture for the least intensive stage of language contact (Thomason 2001: 70). What makes lexical copies from Romance donor languages an interesting research topic is the possibility to check whether or not there are recurrent patterns which select certain lexico-semantic fields when copying Romance content words.12 Answering this question is a task which closely resembles the goals of the Leipzig project on Loanword Typology (Haspelmath, this volume). Romance languages have also contributed to various cases of massive borrowing. For Philippinian languages such as Tagalog, the Romance-derived share of the general lexicon oscillates around the 25% mark, whereas it exceeds 50% by wide margins in languages such as Chamorro – the indigenous Austronesian language of the Marianas which draws much of its lexicon from Spanish (Rodríguez-Ponga 1995; Stolz

8 Thomas Stolz 1998), Tetun Dili – the most widespread Austronesian language of East Timor for which Portuguese is the major provider of lexical items (Williamsvan Klinken et al. 2002), and Maltese which derives more than half of its lexicon from Italian and related Romance varieties (Brincat 2000). In the case of Mitchif (Cree-French) and Media Lengua (Quechua-Spanish), the combination of Romance-derived lexicon and non-Romance grammar has yielded exemplars of intertwined languages (Bakker 1997; Muysken 1988). In a manner of speaking, massive borrowing and language intertwining border on each other (Stolz 2003) and a close study of the former within the framework of research on Romancisation might reveal that there is a continuum which links the two categories of language contact. A continuum is also postulated by Rodríguez-Ponga (1999a), who views modern Chamorro as a kind of backlash development from a stage of incipient creolisation, presumably reached in the first half of the 19th century when the Spanish impact on Austronesian Chamorro increased considerably. According to his hypothesis, the creolisation of Chamorro (partially via Spanish-oriented relexification) was a gradual process which came to a halt when the reform of Spain’s colonial system failed to yield any positive results for the colonised people. An in-depth study of Romancisation is likely to shed more light on the validity of models such as gradual creolisation. Before I proceed with evaluating the implications of the continua, a discussion of other, more problematic aspects of Romancisation is in order.

2.2.

Two etymological paradoxes

2.2.1. Non-Romance etyma Romancisation is not always as straightforward and uncontroversial as it is shown in the cases presented above. In (5), a constellation is presented which suggests that the definition of Romancisation has to be formulated rather cautiously. (5)

Contacts in the Balkans a. Rumanian (Indo-European, [East] Romance; Rumania) [LPP Rumanian XXIV, 60] trebuie s! ocrotim cu grij! l!mpile SUB protect:1 PL with care lamp:PL.DEF must ‘We have to protect the lamps with care.’

Romancisation worldwide

9

b. Kaldera (Indo-European, Indo-Aryan; Ex-Yugoslavia) (Boretzky and Igla 1994b: 283) trubul te dav mange must SUB go:1SG I:DAT ‘I have to go.’ c. Croatian (Indo-European, Slavic; Croatia) [HP Croatian, 53]13 treba joj platit to nam je donijela novine need her pay:INF what for be.3SG bring:PTCPL:F newspaper ‘It [= the owl] needs to be paid for it brought the newspaper.’ Rumanian is responsible for many Romance elements in those varieties of Romani which belong to the so-called Vlax branch (Matras 2002: 199). Therefore, it is highly likely that Kaldera has copied its expression of necessity – the impersonal trubul ‘must’ – from Rumanian trebuie ‘must’. However, words of similar phonological shape are attested in various languages of the Balkan region (Matras 2002: 202). The etymon itself is not part of the Latin heritage of Romance but ultimately stems from one of the various Slavic languages with which Rumanian or its immediate BalkanRomance predecessor came into contact. Actually, cognates of Rumanian a trebui ‘must’ can be found throughout the Slavic phylum – Croatian is only one of many examples. If we discount the possibility that at some stage Romani copied the expression directly from a Slavic language, the diachronic development looks like this: first, Rumanian copied from Slavic and then Romani copied from Rumanian. The Romance language thus functions as a kind of mediator between a distant Slavic source and a recipient language which did not have direct contact with the Slavic donor. This scenario is not unique to Rumanian, but is the leitmotif of Albalá Hernández (2000, 2001) who studies the diffusion of words of Amerindian origin into Austronesian languages of the West Pacific and Southeast Asia. The two language groups were never in direct contact. Mexican Spanish – which had already integrated Amerindian terms in huge numbers – acted as a linguistic go-between. Superficially, these and similar cases fail to qualify as instances of Romancisation because – strictly speaking – the items are not properly Romance. Thus, Rumanian contributed to the Slavicisation of Romani and Spanish made Austronesian languages resemble Amerindian ones. However, I claim that we are nevertheless dealing with Romancisation. It would be naive to picture Romance languages as being composed only of Latin-derived features (and of Latin being ‘pure’). Every member of the Romance phylum combines features of

10 Thomas Stolz an, at least, rather loose Latin background with features which cannot be explained in terms of inheritance from the Vulgar Latin ancestor. Latin-based or not, all these features together make up a Romance language – and thus each feature is automatically a Romance feature. Which means that each process of Lusitanisation, Hispanisation, Francisation, Italianisation, Rumanisation, etc. is also a process of Romancisation no matter how un-Romance the prehistory of a given feature turns out to be (but see section 2.3). Nothing precludes the possibility of one and the same process representing two different kinds of foreign influence. The copy of Rumanian trebuie by varieties of Romani constitute Romancisation-cum-Slavicisation. Note that this combination is asymmetric as the recipient language copies directly from Rumanian and only indirectly from Slavic. With a view to distinguishing terminologically, one could classify the Romancisation as a process of 1st order and the epiphenomenal Slavicisation as a process of 2nd order. This distinction leads us to a sideways glance at 2nd order Romanicisation. 2.2.2. Non-Romance donor Romance-derived features – especially words – may reach areas where Romance speakers hardly ever dare to go, in a manner of speaking. We encounter words with a Romance etymology in languages which have not been in direct contact with any Romance language. A pertinent case is given in (6) where the (globally rather successful) French term billet ‘ticket’ has been introduced into the lexicon of Greenlandic via the intermediary Danish. (6)

Contacts in the Far North a. Greenlandic (Eskimo-Aleut, ) [HP Greenlandic, 110] tamarmik billeti-mik allaqqapput all_this:INS ticket-INS write-PASS-3PL.INDIC ‘All these (bits of information) are written on the ticket.’ b. Danish (Indo-European, Germanic; Denmark) [HP Danish, 89] det hele står på billett-en the.NT whole stand:PRES on ticket-DEF.UT ‘All this is (written) on the ticket.’

The word is treated like any other lexical copy from Danish – it receives the typical nativising (stem-)final -i which makes the word a member of a

Romancisation worldwide

11

certain declension class (which hosts the vast majority of nouns of foreign origin). In analogy to what I concluded in the previous section, the presence of billeti in Greenlandic is first of all an instance of Danicisation, i.e., the lexical copy makes Greenlandic resemble Danish as the recipient language now shares a feature with the donor language. This Danicisation is a 1st order process whereas the accompanying Romancisation is only secondary. It is a side-effect of Danicisation and thus counts as a 2nd order phenomenon. Romance features may spread over vast areas as itinerant phenomena. Since Romance languages (and their speakers) are directly involved only when 1st order Romancisation applies, indirect Romancisation has to be kept apart from direct Romancisation. Each time they are copied, the itinerant features have to pass through selection procedures. With each filter through which an itinerant feature passes, the conditions for the selection of the feature change: Greenlandic billeti is a copy of Danish billett and thus reacts to the structural and other rules associated with the word in Danish – and not to the ones the same etymon has in French. The investigation of Romancisation processes is expected to show whether or not there are also categorial differences on the phenomenological level between the two orders of Romanicisation. 2.3. Romance-Romance contacts A word of caution is called for: saying that changes induced by contact with any Romance language count as instances of Romancisation is a simplification which inevitably leads to problems. Consider example (7). (7)

Corsican (Indo-European, East Romance; France) [LPP Corsican III, 15] hè u me avviò be.3SG the my plane ‘It is my plane.’

Corsican of course belongs to the Romance phylum. Its exact de jure status is still somewhat unclear, but de facto it may be considered a regional language of France which is genetically closely related to the Tuscan variety of Italian (Fusina 2003: 103–104). For over two centuries, the basic ItaloRomance component of Corsican has been under constant pressure from the official state language French. Unsurprisingly, French words occur in Corsican discourse where other members of the Italo-Romance dialect cluster make different choices. One example is the word translating English

12 Thomas Stolz plane: standard Italian and the variety of Naples use aeroplano whereas in Northern dialects (Milanese, Bolognese, Venetian) forms of Italian apparecchio ‘machine’ are employed. Corsican avviò is clearly a copy from French avion ‘plane’. Is this a case of Romancisation qua Francisation? The copy of the French term is certainly an instance of Francisation because Corsican becomes in this way lexically more like French while deviating from Italian. However, the Romance character of Corsican has in no way been affected by the process of Francisation: Corsican was already a Romance language and still is a Romance language. Thus, Romancisation would be a misnomer for this kind of Romance-Romance language contact. I therefore suggest that contact-induced changes in one Romance language triggered by another Romance language remain outside the scope of a project on Romancisation. One could of course object because there is a way for Romance languages to become more Romance, namely, via the replacement of non-Romance features by Romance-derived ones. Examples from the history of Rumanian and Moldavian are well-known: originally Slavic features haven fallen victim to purist-minded processes of Re-Romancisation of Rumanian and its closest relatives (Petrucci 1999). We are facing another paradox. While it is true that Re-Romancisation resembles Romancisation, it is different in kind: only Romance languages can be Re-Romancised whereas only non-Romance languages may experience Romancisation. If it makes sense at all to include processes of Re-Romancisation in an investigation of Romancisation, they should be considered as a separate category or 3rd order phenomena. Otherwise the unwanted effect of “etymological secession” is created, i.e., for each Romance language, two distinct components have to be distinguished, namely a proper Romance one and non-Romance one. The former would be largely equivalent to the Latin heritage (in the widest possible sense of the term) – and thus Romancisation would become a synonym of Latinisation. However, I argue that one cannot strip Romance languages of their post-Latin/non-Latin features and then study their behaviour as donor languages in language contact. Moreover, the question of a given item being either +Romance or ¬Romance only seldom determines the borrowing behaviour of speakers of recipient languages.

2.4. Creolisation Three Romance languages are the (major) lexifiers of dozens of Creole languages around the globe. Spanish, Portuguese and French provide the ex-

Romancisation worldwide

13

pression side of the linguistic signs of Creoles from the Caribbean, over mainland South America to West Africa, the Indian Ocean, South and Southeast Asia and Oceania. The three Creole sentences in (8) reflect the high degree of surface “Romanceness” of Romance-based Creoles. In the line immediately below the morpheme glosses, I add the French, Spanish and Portuguese etyma (in boldface) on which the Creole expressions are based. Unsurprisingly, in these Creole sentences, 100% of the words/morphemes have a Romance etymology. For the content words, the semantics of the Creole items are similar to those of their etymological source, whereas the grammatical morphemes – TMA markers, for instance – deviate at least slightly from the functions associated with their donor language equivalents. (8)

Creoles I a. Seselwa (Creole, French-based; Seychelles) [LPP Seselwa I, 10] mon desen pa ti reprezant en sapo my drawing NEG PAST represent a hat mon dessin pas été représente un chapeau ‘My drawing did not show a hat.’ b. Papiamentu (Creole, Spanish-based; Antilles) [LPP Papiamentu I, 10] mi pintura no tabata representá un sombré a hat my drawing NEG PAST.PROG represent mi pintura no estaba-está representar un sombrero ‘My drawing did not show a hat.’ c. Angolar (Creole, Portuguese-based; São Tomé) (Maurer 1995: 51) alê tha masi nôthêntxi bô king be more naive you rei estar más inocente vós ‘The king is more naive than you.’

This looks like Romance indeed. However, in what sense are the Creole examples instances of Romancisation? The principle difficulty lies in the fact that, in prototypical plantation settings, Creoles arise as new languages from a kind of melting pot of many so-called substrate languages. It is often impossible to identify these substrates on the linguistic map of Africa or any other region from where slaves were deported. Which of these sometimes anonymous languages was subject to Romancisation? For multilingual slave populations, Romancisation of a single substrate language is a problematic hypothesis. Since almost the entire lexicon of the Creole is of

14 Thomas Stolz Romance origin, it is also hard to determine the potential matrix language into which Romance words have been integrated. Creolisation of this kind and the above mainstream types of lexical copies cannot be treated as one phenomenon because for Creolisation qua genesis of a new language, it is impossible to name a recipient language. Thus, what we have here is a different kind of Romancisation: in lieu of a particular language, Romancisation affects the global inventory of distinct languages by adding a new one which looks suspiciously Romance. I classify this process as a 4th order phenomenon. If it is true that Creolisation can also apply in binary constellations, i.e. with one (clearly dominant) substrate language, and if it is also true that Creoles can develop gradually without prior Pidginisation and catastrophic language shift, then there are good grounds to subsume these cases under 1st order Romancisation – the only difference to the garden variety Romancisation being the extent to which relexification applies. Of course, Creole languages themselves can be subject to Romancisation in the post-creolisation phase. This is absolutely unproblematic with Creoles whose lexifier is a language which does not belong to the Romance phylum (De Granda 1985a). In the English-based Creole of San Andrés (Colombia), recent lexical copies from Spanish are attested, the same is true of Fa d’Ambû (Equatorial-Guinea) – a Portuguese-based Creole (cf. [9]) (De Granda 1985c). (9)

Creoles II a. Sanandresano (Creole, English-based; Colombia) (O’Flynn de Chaves 1990: 146) im da bazuukero he be basuco_addict ‘He is a basuco addict.’ b. Fa d’Ambû (Creole, Portuguese-based; Equatorial-Guinea) (Barrena 1957: 41) pe bo su cu na mina dos father your be with PL child two ‘Your father has two children.’

Sanandresano is experiencing Romancisation via Hispanisation (bazuukero is a Spanish noun with a Spanish agent noun affix -ero). Likewise, Hispanisation affects Fa d’Ambû (the Spanish numeral dos appears where Portuguese dois would have been expected to appear). However; Fa d’Ambû

Romancisation worldwide

15

already has a Portuguese-derived Romance lexicon predating the Spanish influence. Thus, the situation resembles the one discussed in connection with Romance-Romance contacts (section 2.3): something that is already Romance receives additional impact from something else that happens to be Romance too. Treating Romance-based Creoles in this way however is tantamount to tacitly classifying the Creoles as members of the Romance phylum, on a par with, say, French-influenced Corsican. The secondary Romance influence would thus become an instance of Re-Romancisation. In Creolistics, the problem of the genetic affiliation of Creoles has not been satisfactorily settled yet. Even where the historical background of the lexicon is clearly discernible, phonology and grammar (not to mention other levels) cannot be easily accommodated to our expectations of a Romance language – but on the other hand, at least in traditional historical linguistics, grammatical properties are not normally decisive in the genetic classification of a given language. As too many issues are unclear, copies from Romance languages in Romance-based Creoles should not be given priority in a project on Romancisation. 2.5. Bilingual matters In the above sections, I have reviewed contact scenarios which involve relatively stable copies of features by (ideally) entire speech-communities. In addition to these bona fide instances of contact-induced language change, there is also an array of phenomena which are largely ephemeral and individual. Among these, the best known category is code-switching (together with the related notions of code-mixing, code-alternation and language crossing [Rampton 1995]). Myers-Scotton (1993a) claims that codeswitching and borrowing are two poles on a continuum of more or less similar processes. For the sake of argument, I only take prototypical codeswitching into account. In (10), I quote two examples of code-switches in which French words are interspersed into the utterances of speakers of Moroccan Arabic and Pondicherry Tamil who are bilingual with French. The arguments put forward in the remainder of this section are also valid for cases of Romance-Romance code-switching, code-mixing, code-alternation and language crossing.

16 Thomas Stolz (10) Code-switching a. Moroccan Arabic (Afro-Asiatic, Semitic; Morocco) (Heath 1989: 34) fi-h kayn l-‘id l-kbir ka-n-db -u there_is DEF-holiday DEF-great PRES-1-slaughter-PL in-it les moutons the:PL sheep ‘There is also the Great Feast in which we slaughter sheep.’ b. Pondicherry Tamil (Elamo-Dravidian, Dravidian; India) (Kelkar-Stephan 2005: 159) k&leyil* e##u ma1ikku école p2v*n morning:LOC:EMPH eight hour:DAT school go:FUT.1SG ‘I always go to school at eight o’clock in the morning.’ In both examples, an NP is inserted into otherwise non-Romance utterances. The NPs keep their Romance phonological shape, i.e., they are not nativised. What distinguishes them from lexical copies of the kinds discussed in the previous sections is the fact that they occur only occasionally and their occurrence is not predictable on the basis of Arabic or Tamil. As a matter of fact, Moroccan Arabic has a Semitic word for les moutons which is normally used, namely h wala (Heath 1989: 34). Similarly, there are various words denoting schools and educational institutions in Tamil: kalaicc&lai or kalviy4ri are the primary choices for speakers of Tamil. Les moutons and école appear only in the discourse of speakers who are bilingual with French – other monolingual speakers of Moroccan Arabic and Tamil would not use these expressions. Moreover, these French elements do not occur regularly in the speech of bilinguals but each one only turned up once unexpectedly in two large corpora. Thus, they are not part of a fixed register or sociolect but constitute the occasional overlap of two codes at a given point in time – with a fair chance that none of the French elements will ever be used again in first language discourse by the same (or other) speakers. Since the French lexical intruders do not permanently remain in the native lexicon of the speakers, they definitely do not instantiate the Romancisation of la langue. What they Romancise is la parole – but even this only temporarily. Apart from the fact that it is often difficult to tell which of the languages in bilingual discourse functions as the matrix, the elements which partake in code-switches vary from utterance to utterance and from speaker to speaker, who are free to switch or refrain from switching – a choice that may be governed by social factors (Myers-Scotton 1993b) or cognitive-

Romancisation worldwide

17

pragmatic ones (Matras 1998). Formally, the switches may include single words (or even single morphemes) to entire phrases or longer stretches of uttered material. This high degree of variation and liberty does not allow us to draw up a closed list of those words which define the French-Arabic or French-Tamil “register”. This is the reason why one cannot pinpoint the linguistic matter which constitutes the lexicon of code-switching to and from French. What characterises bilingual discourse is not so much a set of identifiable elements from languages A and B but the potential for switching from A to B and back. Thus, Romancisation, in the case of code-switching, does not change a language but it marks the (mostly sub-conscious) options of a certain intersecting segment of two speech-communities, namely those who know both French and Tamil or Arabic. Code-switches from Romance may thus figure as 5th order phenomena. This order also comprises other cases of code-mixing, code-alternation and language crossing. Whether or not interference in foreign language acquisition should be included in this class cannot be decided here. Suffice it to say that learner varieties or interlanguages usually are of an individual nature and often unstable over time. Nevertheless, it may still be interesting whether there are language-type specific differences between code switches. 2.6. Language planning and related issues Only as an aside, I will also touch briefly upon issues of language planning (Zimmermann 1999), i.e., the conscious attempt of institutions or individuals to shape a given (non-Romance) language in a certain way, by having recourse to Romance features. In many speech-communities worldwide, no autochthonous terminology is available for a large number of specialised semantic fields of modern technology, etc. It comes as no surprise therefore, to find an abundance of lexical copies from Romance languages in these areas wherever the recipient language has some historico-cultural ties with a Romance-speaking country. For the West Atlantic language Ful, for instance, the commission in charge of the creation of languages for special purposes (= MAPE) has suggested adopting French terms and accommodate them to the phonological structure of Ful: French vitamine ‘vitamin’ > Ful witaamine, French moteur ‘motor’ > Ful moteer, French crayon ‘pencil’ > Ful kiriyon, etc. (Sampson 1990: 102). Admittedly, this is not the only way neologisms are created. Very often, the copy of segmental matter is avoided by calquing on the content level – a process which may be motivated by purism. The Romancisation may thus be only covert.

18 Thomas Stolz Superficially, the coining of neologisms appears to be loosely connected to the creation of artificial world-languages such as Esperanto. As sentence (11) demonstrates, Esperanto owes much of its vocabulary to Romance. (11) Esperanto (artificial world language) [LPP Esperanto I, 8] mia disegno ne prezentis 'apelon my drawing NEG represent:PAST hat:ACC ‘My drawing did not show a hat.’ All lexical morphemes in (11) have an association with Romance (the bound morphemes -is for past tense and -n for the accusative case have a different background, though) although it is not always clear exactly which of the Romance languages has served as the model. In a way, the Romance flair of the Esperanto lexicon and of the Romance-based Creoles pose the same problem: there is no recipient language which has copied the Romance features. The absence of an identifiable recipient language distinguishes these cases from the conscious attempts to create new lexemes via copying from a Romance donor language. The latter are instances of Romancisation of languages, albeit a fully planned and achieved Romancisation. Esperanto and similar languages enrich the catalogue of Romance-like languages of the world.

2.7. Beyond major languages The section on code-switching is suggestive of another important aspect of Romancisation. Trivially, Romancisation occurs in those varieties which are actually spoken by the speakers. These varieties are not always identical with the standard or state languages of which the five major Romance languages surveyed so far are representative. Written and spoken registers may differ widely. Dialects, regional and minor languages come into play as well – and it turns out that they are by no means less successful in language contact than the often quoted major languages. The same holds true for the recipient languages: the Romance influence may be much more evident in the non-standardised varieties whereas the standard language itself (artificially or not) remains almost immune to foreign features. Haase (1992) demonstrates that many Romance features in the Labourdin variety of Basque originate from Gascon (the neighbouring variety of Occitan in Southwest France). Many of these Romance features have not been admitted to the unified Basque norm called Batúa. Similarly, Friulan and Venetian –

Romancisation worldwide

19

two regional Romance varieties in northern Italy – have influenced the Slovenian variety spoken in their vicinity in Italian territory, whereas standard Slovenian does not share these Friulanisms and Venetianisms (Steenwijk 1992). The oldest, rather thick Romance layer in the heavily Romancised lexicon of Maltese is of Sicilian origin – copies from standard (Tuscan) Italian are, in a manner of speaking, newcomers (Brincat 2000). The closer a variety of a language is spoken to the territory of a potential donor language, the higher the probability becomes that this variety will copy more features from the donor than the standard variety or varieties which are spoken at a greater distance. In the Bavarian German dialect of the Fersental in northern Italy, we encounter lexical copies from Italian as in (12) which are unknown in standard German and most of the other diatopic varieties of German. (12) Fersental German (Indo-European, Germanic; Italy) (Rowley 2003: 197) s ist importantet ver a khinn der earscht kontatto pet de schual it is important for a child the first contact with the school ‘The first contact with the school is important for a child.’ There is of course a standard German word Kontakt ‘contact’ whose etymology however does not go back to Italian. Thus, kontatto in Fersental German is clearly a direct copy of Italian contatto ‘contact’. In the German diatopic system, the adjective importantet ‘important’ is unique to the Fersental variety which has copied the word from Italian importante ‘important’. Similarly, the copies from Rumanian, which are reported for the local German dialects of Rumania, are restricted to these varieties and do not occur elsewhere in the German diatopic system (Ney 1984). All these cases are perfectly in line with the prototype of 1st order Romancisation and thus have to be taken into account in our search for Romanisms in the languages of the world. 2.8. Definition After having reviewed the various disguises in which overt lexical Romancisation may come, I am now in a position to formulate a preliminary definition of the phenomenon that should become the primary research object. It is likewise possible to take the lexical evidence as our basis for generalisations about all kinds of feature copies – be they overt or covert. For the sake of simplicity, I consider 1st order phenomena to be the prototype of Romancisation which is defined in (13).

20 Thomas Stolz (13) Romancisation – a preliminary definition of the prototype Every process which leads to at least the stable and not individual (full or partial) copying of an overt or covert feature of an ideally identifiable Romance donor language in an identifiable and already existing non-Romance recipient language is an instance of Romancisation regardless of the pre-history of the Romance feature itself. All those cases which fail to fulfil these requirements but nevertheless directly or indirectly involve copies from Romance do not qualify as prototypical instances of Romancisation. Still, they instantiate Romancisation of different orders. 3.

Non-lexical copies

3.1. A survey A language is not coextensive with its lexicon. Phonology and morphosyntax together with other system-like, organised levels beyond the sentence are at least equally important components of a language, although they are notoriously less accessible to the observer’s eye. Contact-induced change also happens in these areas – but normally only after the lexicon itself has been affected by foreign influence. Phonological, morphological and syntactic copies are characteristic of stages of language contact which require a longer and/or more intensive exposure of the recipient language to the donor language, namely stages 3-4 on the borrowing scale (Thomason 2001: 70–71). Romance languages have contributed to the re-shaping of the phonology, morphology and syntax of many non-Romance languages. Space restrictions do not allow me to elaborate on these. In what follows, I content myself with enumerating a selection of pertinent examples to which many more could be added. Note that some of the cases are still highly controversial but, for obvious reasons, I refrain from mentioning the counterarguments. Note also that in some of the cases the Romance influence is perhaps only of a supportive nature, i.e., the impact exerted by a Romance donor added force to an already existing tendency in the recipient language. Another proviso: with a view to making the catalogue as transparent as possible, I restrict the exemplification to those cases which are “positive” in the sense that no simple loss of a category occurred under Romancisation.

Romancisation worldwide

21

a) Phonology – diffusion of the uvular articulation of rhotics (= apico-alveolar flap [
] > dorso-uvular trill []/dorso-uvular fricative []) starting in 17th century France and spreading successively into Germanic, – copy of rounded front vowels /y/, /ø/, /œ/ from French in Breton (Ternes 1992: 431) and /y/ in Laburdin Basque (Haase 1992: 29–31), – copy of nasal vowels from French in Breton (Ternes 1992: 431), – introduction of mid-high vowel phonemes in Philippinian languages (under the impact of Spanish) and Maltese (under the impact of Italian) (Stolz 2003), high-back /u/ in modern varieties of Nahuatl (under Spanish influence) (Farfán 1999: 115), – introduction of consonantal phonemes/phonematisation of allophones: – Nahuatl /g/, /f/ from Spanish (Farfán 1999: 115), – Maltese /p/, /g/, //, //, /v/ from Italian (Krier 1976: 16), – Laburdin Basque //, //, // from French/Gascon (Haase 1992: 31–32), – Chamorro /r/, /l/ from Spanish (Topping 1973: 33–34); – suprasegmentals: copy of ultimate stress site in Chamorro polysyllables from Spanish (Topping 1973: 46) and antepenultimate stress site in Maltese from Italian (Krier 1976: 26–27); – phonotactics: introduction of binary tautosyllabic consonant clusters (PLOSIVE + LIQUID) in Chamorro (Topping 1973: 36–37). b) Morphology – the French infinitive marker -{} has been copied by German and has become productive as a derivational morpheme -{}with which denominal verbs can be formed (Buchstabe ‘letter (of the alphabet)’  buchstabieren ‘to spell’), – the French privative prefix {}- has entered the derivational morphology of Basque and is productively used on non-Romance stems (as in desberdin ‘unequal’ [Haase 1992: 50]), – Spanish feminine suffix -{} is productively – though only optionally – used to mark gender on animate nouns in Basque (i.e. gender is introduced as a grammatical category in otherwise gender-less Basque) as in artzaintsa ‘shepherdess’ (Haase 1992: 49–50), – the Portuguese agent noun morpheme -{} is productively used with the same function in Tetun Dili even in combination with nonRomance stems (as in Tetun Dili dukur ‘sleep’  dukurdór ‘sleepy head’ [Hajek and Williams-van Klinken 2003: 59]); Muysken (2001: 75) reports various Quechua stems which take the Spanish-derived cognate morpheme -{} (as in wañuchidor ‘killer’), Balsas Nahuatl

22 Thomas Stolz has tlawankero ‘drunkard’ with Spanish agentive marker -{} (Farfán 1999: 121); -{} is also productive in Chamorro (Rodríguez-Ponga 1999b: 162); for similar copies of Spanish -{}, -{} and other derivational morphemes in Tagalog, cf. Wolff (2001: 240–243), – Yucatec Maya has copied not only the agent noun suffix -{} from Spanish (as in ts’òonero ‘hunter’) but also the diminutive marker – {} (as in chichanito ‘very small’) – both affixes can be added freely to non-Romance stems (Stolz, Ch. 1998: 183), – the Sicilian deprecative marker -{} has come to be used in Maltese on inherited Semitic stems (as in sakranazz ‘drunkard’ which contains the triconsonantal Semitic root S-K-R associated with [excessive] drinking [Stolz 2003: 285]), – the Sicilian past participle suffix -{} and its allomorphs have become productive in Maltese where it may also combine with Semitic material (as in emmnut ‘believed’). c) Syntax – Maltese has copied the Italian dynamic passive periphrasis – a construction made up of the inflected passive auxiliary venire ‘to come’ and a past participle of the lexical verb (in Maltese ,ew mitluba {come.PERF:3PL} {PTCP:ask:PL} ‘they were asked’ [Krier 1976: 60– 61]); – Breton has developed a verb of having under the influence of French avoir ‘to have’ (Heine and Kuteva 2005), Nahuatl has copied the predicative possessive constructions from Spanish tener ‘to have’ (Farfán 1999: 139); – Italian subject focus constructions have been copied by Cimbrian (Kolmer 2005: 182–183); – comparative constructions in several dozens of Amerindian and Austronesian languages have been re-organised according to the Spanish pattern by copying the degree marker más and/or the relator que (Stolz and Stolz 1995, 2001); – the copulas ser and estar of Spanish have given rise to a similar dichotomy in Balsas Nahuatl (Farfán 1999: 132–133), – the indefinite article of Spanish has been copied by Chamorro in the generalised form of un ‘a(n)’ (Stolz and Sabater Fuentes 2002) This still open list is alone suggestive of a very rich phenomenology which awaits investigation and evaluation. Romance influence is not only visible

Romancisation worldwide

23

in almost every area of grammar, but there are also recurrent patterns whose interpretation requires the attention of general linguists. All the more so as several of the above features would normally be considered marked (at times even highly marked) properties, i.e., their spread over language boundaries is somewhat unexpected. With a view to demonstrating that the observed recurrence is not a marginal phenomenon, I present new data from the copying of function words in Romance language contacts in the subsequent section. 3.2. Function words: contra Research on Hispanisation of Amerindian languages in Mexico marks the starting point of Romancisation research as outlined in this contribution (Zimmermann 2004a: 29–30). In the seminal paper by Zimmermann (1987), the importance of grammatically relevant copies from Spanish in Otomí is highlighted for the first time. In quick succession, numerous follow-up studies, first with a Mesoamerican focus (Stolz and Stolz 1996a; Hekking 1995; Stolz, Ch. 1998; Verbeeck 1998) which was then extended to a PanAmerican one (Hekking and Muysken 1995) and later also embraced Austronesia (Stolz and Stolz 1996b, 1997; Stolz, Th. 1998, 2002) have demonstrated that certain classes of function words are especially prone to being copied in language contact situations where Spanish is the donor language. These function-word classes are mainly discourse particles and conjunctions, i.e., inter-sentential and intra-sentential connectors. There are striking similarities in the inventories of copied items no matter which recipient language is involved in the language contact situation. The same seems to hold true for the donor language: recent studies on Italianisation processes (Stolz 2005, 2007) reveal that the preferred function words copied from Italian are again discourse particles and conjunctions which are equivalent to the ones copied from Spanish elsewhere. A third class of function words that are rather successful in language contact are the prepositions. All Romance languages are prepositional, but not all of their partners in contact belong to the same typological class. Thus, Romance prepositions may become postpositions when copied by a postpositional language. In (14)–(15), I present a selection of examples of the preposition contra ‘against’ which has been copied from Spanish and Italian in a variety of contact constellations.

24 Thomas Stolz (14) Copies from Spanish a. Basque (Lafitte 1998: 171) amaren kontra mintzatzea itsusi da mother:GEN against speak:VN:DEF disgusting be.3SG ‘Talking back to the mother is disgusting.’ b. Algerian Arabic (Afro-Asiatic, Semitic; Algeria) (Heath 1989: 288) xrej kõ ra ‘liya come_out.PERF against against:1SG ‘He is against me.’ c. Chamorro (Austronesian, Philippinian; Marianas) (Topping 1973:128) baba bida-ña kontra i taotao bad do-POR.3 against the people ‘His work is bad against the people.’ d. Mixtec (Amerindian, Otomangue; Mexico) (Alexander 1980: 82) ni candonda i contra gobierno PERF rebel 3SG against government ‘They raised against the government.’ (15) Copies from Italian a. Maltese (Afro-Asiatic, Semitic; Malta) [Tazza 74] l-Urugwaj marru fil-vanta,, kontra l-Bra$il DEF-Uruguay go.PERF:3PL in:DEF-advantage against DEF-Brazil ‘Uruguay defeated Brazil.’ b. Resian Slovenian (Indo-European, Slavic; Italy) (Steenwijk 1992:162) ti si ál kúntra mlé you be.2SG put:PTCPL against me ‘You are against me.’ In spite of the genetic, areal and typological differences on the side of the recipient languages, the outcome of contact with Spanish or Italian is largely the same: the same Romance preposition has been copied and fully integrated into the system of the recipient language. It is an intriguing fact that exactly the same copy occurs in so widely different languages and regions. This homology is the more striking as none of the recipient languages was “in structural need” to copy contra because autochthonous means to express the concept of contrariness were available for all of them. The languages in (14)–(15) are not the only ones which have copied contra (there

Romancisation worldwide

25

are many more cases especially in Mesoamerica and in Italy). Moreover, there are also a handful of other Romance prepositions which behave much like contra when being copied, i.e., they enter the adpositional systems of many different languages in many different regions. It is one of the most important tasks of a project on Romancisation to identify and explain these and other recurrent patterns. 4. Conclusions In a series of recent articles, Zimmermann (2004b, 2005a–b, this volume) develops a neuro-scientifically inspired constructivist theory with a special focus on language contact. For this theory, individual speakers and their individual constructive brains are the basic units. In lieu of a homogeneous inter-individual language, there is a cluster of idiolects which influence each other to the extent of appearing to be identical. The convergence of these idiolects is determined by viability, which determines which features can be copied from one idiolect into the other. This viability-based model is valid for both the shaping of a supposedly common language and the copying of features in language contact. With reference to Hekking and Muysken (1995) who insist that the copies from Spanish in Otomí and Quechua do not yield absolutely identical inventories, Zimmermann (this volume) emphasises that the outcome of language contacts is largely individual, for which he also refers to Stolz and Stolz (2001) who have shown that Austronesian and Amerindian languages copy different elements of the Spanish comparative construction. Stolz (2003) argues that different varieties of a recipient language down to the idiolect, may display different degrees of donor language influence and thus often render it difficult to determine whether a language as such is strongly or only weakly Romancised. Superficially, the variation that occurs in the contact behaviour of the recipient languages lends support to the individualistic model of the constructivist approach. If the individual component is indeed so strong, is generalisation possible at all? As the discussion of the preposition contra in the previous section 3.2 was supposed to suggest, the contact behaviour of recipient languages is by no means idiosyncratic. One and the same choice from the donor language is made by many languages. On the microlevel, the copies are inevitably bound to diverge because languages are individuals (Bechert 1981) in the sense that, with the possible exception of the still unproven case of full metatypy (Ross 2001), each language represents a unique combination of

26 Thomas Stolz features. Thus, the integration of a copy into a recipient language system has to be accommodated to the structural (or other) traits of this particular language. In the case of the Hispanisation of Amerindian and Austronesian comparative constructions, the copies diverge because of the simple fact that the recipient languages employ different types of comparative constructions which favour the copying of different Spanish elements. What these recipient languages have in common though, is the fact that they reorganise their comparative constructions by way of partially copying the Spanish patterns. One may say that on the meso-level, the recipient languages behave in such strikingly similar ways that this must attract our attention. Constructivists may ask themselves why certain features are almost always viable whereas others fail to fulfil the viability criterion. What is the reason for so many individual minds to concur as to what feature is worthwhile copying from another idiolect cluster? Matras’ (1998) model of cognitive-pragmatic prominence of copies-to-be in the mind of bilingual speakers most probably complements Zimmermann’s approach. Those who do not fully subscribe to a constructivist theory of language (contact) have to answer essentially the same question, namely whether there is any languageinternal or language-external explanation for the fact that recipient languages all over the world seem to follow the same path. Before one can set out to weigh the pros and cons of the different approaches and theories, it is absolutely necessary to do a considerable amount of empirical groundwork. The many individual studies mentioned above notwithstanding, our knowledge about Romancisation is practically still in its infancy as there is not even an overall conspectus of the languagecontact phenomena with Romance donors. In the 1970s, Bal (1988: 43) had already projected a typology of loanwords […] a study of the means of penetration and of spread of these, […] a socio-linguistic analysis and […] the drawing up of a ‘dictionary of Portuguese loan-words in the languages of Africa and Asia’. [original inverted commas]

This idea should be extended to cover the entire range of Romancisation processes: all Romance donor languages must be taken account of, all kinds of contact-induced processes – be they overt or covert, lexical or grammatical, etc. – have to be registered. It would also be helpful to determine the degree of Romancisation of recipient languages by a statistic evaluation of parallel corpora, for which the ones I used for the present purpose provide a suitable basis. Thus, Bal’s dictionary of loan-words turns into an encyclopedia of Romancisation. The work on this encyclopedia requires the co-

Romancisation worldwide

27

operation of many encyclopédistes from many disciplines. It is a demanding task which will consume time, energy and other resources. However, once Romancisation is described in this way, language contact studies will make considerable progress and general linguistics will gain important new insights which justify the labour we are facing.

Notes 1.

2.

3.

4.

I am grateful to Alexandra Aikhenvald, Walter Breu, Martin Haspelmath, Bernd Heine, Ewald Hekking, José Antonio Flores Farfán, Miki Makihara, Yaron Matras, Anthony R. Rowley and Klaus Zimmermann for giving me access to their contributions to the linguistics of language contact. Dik Bakker and Georg Bossong kindly commented on a first version of this article. Aziza Nawazir and Tuyet Dobis were helpful in my search for translations of Harry Potter in Southeast Asian languages. A heartfelt word of thanks is also directed to the members of my research team at the University of Bremen: Sonja Kettler, Tamar Khizanishvili, Nataliya Levkovych, Cornelia Stroh, Christel Stolz, and Aina Urdze assisted me in technical matters. Rachel Montague took care of my English. The responsibility for the ideas expressed in this contribution is solely mine. For Romancisation to apply, Romance languages must exist in the first place. It is still a matter of dispute among specialists when exactly the Latin era ends and the Romance era starts and whether or not there was a phase of overlap and co-existence. Contacts may very well have occurred long before this historical date. The Latin heritage in non-Romance languages is not always easy to separate from features copied from early Romance languages. Old Latinisms in Berber seem to be less of a problem, as the tree names below suggest, because the respective Latin words either have not survived into Romance or come in a phonologically too divergent shape to qualify as sources for the Berber expressions (all of which contain the feminine circumfix t-…-t): tifirest ‘pear tree’ < Latin pirus, tulmut ‘elm’ < Latin ulmus, takweruct ‘oak’ < quercus (Basset 1906: 440). Many scholarly articles and monographs – mostly from within Romance Philology – are about this topic. Since I am not concerned with foreign influence on Romance languages here, it suffices to refer the reader to the contributions in the volume edited by Silva-Corvalán (1995) the majority of which discuss the genesis of regional traits of varieties of Spanish as a result of contact with indigenous languages of the Americas, Africa, Asia and the Basque countries. In the sentential examples, boldface identifies those items or structures which are copied from Romance (or are the focus of the accompanying discussion). For examples from primary sources, I am myself responsible for the gram-

28 Thomas Stolz matical analysis. Where my secondary sources already provide a morphemeby-morpheme gloss, I also use it here – unless I deemed it necessary to make changes. For each sample language, I add information (given in brackets immediately following the glossonym) on its genetic affiliation and its present site on the linguistic map of the world. 5. ACC = accusative, AUG = augment, CL = class, DAT = dative, DEF = definite, EMPH = emphatic, EXI = existential, F = feminine, FOC = focus, FUT = future, GEN = genitive, GOAL = goal, IMPERF = imperfective, INDIC = indicative, INF = infinitive, INS = instrumental, LIM = limitative, LOC = locative, NEG = negative, NON.A/S = non agent/subject, NON.FOC = non-focus, NT = neuter, P = PATIENS, PASS = passive, PAST = past tense, PERF = perfective, PL = plural, POR = possessor, PPD = postpositive determinant, PRES = present tense, PROG = progressive, PTCPL = particple, REC = reciprocal, REL = relative, SG = singular, SUB = subordinator, SUP = superlative, TOP = topic, UT = utrum, VIS = visible, VN = verbal noun. 6. Theoretically, Vietnamese tem could also be derived from English stamp but tem corresponds closely to the phonological adaptation patterns of lexical copies from French in Vietnamese (Barker 1969). Uncontroversial cases are Vietnamese ô-tô ‘car’ < French auto, ga-tô ‘cake’ < French gateau, len ‘wool’ < French laine, etc. 7. In Rapanui, acuerda is no longer interpreted as the 3rd person singular of the Spanish reflexive verb acordarse ‘to remember’. Rapanui acuerda is the basic verbal form which is accompanied by all kinds of particles indicating person, number and TMA categories. According to Makihara (2001: 197), “Spanish reflexive verbs […] are usually incorporated in their nonreflexive forms.” 8. In Swahili, initial syllables of loanwords are occasionally re-analysed as class prefixes according to the Bantu system. Thus, the originally monomorphemic Portuguese quintal /k.tal/ was reinterpreted as bimorphemic {/ki/}-{/talu/} whose regular Swahili plural form bears a different prefix {/vi/}- (vitalu ‘gardens’) (Legère 1989: 116). 9. In Tariana, the intervocalic voiced dental stop /d/ often appears as flap [] in allegro style (Aikhenvald 2002: 47). 10. According to Raidt (1983: 64–65), the Portuguese word entered the Afrikaans lexicon via the Portuguese-based Malayan Creole of Dutch East India – and thus would not instantiate a direct copy from Portuguese itself. 11. In the majority of the cases in (1)–(4), code-switching (cf. section 2.5) is precluded by the fact that speakers of the recipient languages – at least nowadays – do not have any competence in the donor language, which is a prerequisite for code-switching: only bilingual or multilingual speakers are able to switch codes (Myers-Scotton 1993a: 3). For Huave, Rapanui and Basque, however, bilingualism with Spanish can be taken for granted. To a lesser extent, this is also true for Tariana and Portuguese, Kabiyé and French, Malagasy and French. Notwithstanding the fact that certain segments of the speech-communities of

Romancisation worldwide

29

these recipient languages also know the donor language sufficiently well, none of the examples fits the description of code-switching. The Romance words never retain the full list of properties of the linguistic sign of the donor language – which is what one would normally expect in code-switching (but see Myers-Scotton 1993a: 163). In Huave and Tariana, the phonological shape of the Romance items (Spanish maestro ‘teacher’ and Portuguese pecado ‘sin’) has been accommodated to the phonotaxis of the recipient language and thus the words have been completely nativised. Additionally, the meaning of Tariana pekaru ‘joke’ deviates too much from the Portuguese etymon to count as a simple intercalation from the donor language. For Rapanui acuerda ‘remember’, code-switching is ruled out as an explanation because the Spanish verb is integrated in a fixed form which is not only valid for all Spanishderived verbs but would also not make sense from the point of view of the donor’s morphosyntax. Moreover, Malagasy biè ‘ticket’ and Basque enpresa ‘firm’ are the only or primary choices for all speakers of these languages to express the associated concepts. For Kabiyé, Lébikaza (1992) states that only Kabiyé speakers with a certain knowledge of French employ the term frère in their discourse. However, they do it consistently with a meaning which deviates slightly from the French frère ‘brother’, namely ‘elder brother’. 12. The descriptive situation for lexical copies overseas is especially good in the case of Portuguese and Spanish. The Afro-Romance bibliographies provided by Bal (1988) clearly show that French influence on the indigenous languages of the officially francophone countries has rarely been the topic of linguistic study. The following overview thus focuses on Ibero-Romance donor languages. Even with this restriction, it does not exhaust the extant literature on the subject of lexical copies. Information on this topic can very often also be gathered from studies which otherwise focus on phonological and grammatical Romancisation or sociolinguistic aspects thereof. Portuguese words in various African languages are the topic of Giese (1954), Martins (1958), Bradshaw (1965), Kiraithe and Baden (1976), Bal (1979a–b), Hair (1980), De Granda (1985a), Legère (1989), Huth (1991), Pasch (1997) among many others. As to the Portuguese impact on languages spoken in Asia, the monumental lexicon of Asian Lusisms compiled by Dalgado (1982, 1988) in the early 20th century is still without serious competitors. In addition, Knowlton (1966, 1976) studies the Portuguese elements in Tamil and Thai. Japanese has also integrated numerous Portuguese words (Kamiya 1994, Bossong 2003). The Spanish lexical impact on colloquial Moroccan Arabic is inventorised in Ibn Azzuz Haquím (1953) and more closely studied (together with the much more frequent French derived words) by Heath (1989). De Granda (1985b) gives an account of the Spanish contribution to the indigenous languages of Equatorial Guinea. Quilis (1973, 1976) and Wolff (1972, 2001) look at the lexical Hispanisms in Tagalog and Cebuano (and other Philippinian languages). Quilis’ disciple, Rodríguez-Ponga (1995, 1997) surveys the Spanish-derived items in

30 Thomas Stolz the lexicon of present-day Chamorro and Pohnpean. The Spanish lexical element in the indigenous languages of the Americas is probably as richly documented as the Portuguese impact in Africa. Boas (1930), Bright and Thiel (1965), Karttunen and Lockhart (1976), Bright (1979), De Pury-Toumi (1979), Hill and Hill (1981) represent only a selection of those authors which have addressed the issue of the lexical Hispanisation of varieties of Nahuatl. Calvo Pérez (2001) is a recent study of lexical Hispanisms in Quechua. Parodi (1998) compares lexical Hispanisms in Amerindian and Austronesian languages in order to reconstruct the phonology of early colonial Spanish. For reasons of space, I refrain from giving the full list of pertinent studies on lexical copies from Spanish, many of which are difficult to access. Several contributions to Stolz et al. (forthcoming a) are concerned with lexical Hispanisms in the Americas – though at times only indirectly. The Italian lexical influence, especially in the neo-Arabic varieties of North Africa, is described by Cifoletti (1986, 2003); for Italianisms in the languages of Northeast Africa, cf. Tosco (this volume). For French, I refer the reader to the contributions by MoyseFaurie (this volume), Rose and Renault-Lescure (this volume), Bakker and Papen (this volume) and the occasional title outside this volume: Arokianathan (1999) is one of these rare items which deal with French influence in South Asian languages. 13. Some of the word-forms in the Croatian example are sub-standard because they reflect the rather uneducated colloquial style attributed to the character Hagrid.

Sources HP Afrikaans 2000 Janie Oosthuysen (trans.) J. K. Rowling. Harry Potter en die towenaar se steen. Johannesburg: Human & Rousseau. HP Albanian 2002 Amik Kasuruho (trans.) J. K. Rowling. Harry Potter dhe guri filozofal. Tirana: Dituria. HP Basque 2000 Iñaki Mendiguren (trans.) J. K. Rowling. Harry Potter eta sorginharria. Donostia: Elkarlanean/Salamandra. HP Croatian 2001 Zlatko Crnkovi (trans.) J. K. Rowling. Harry Potter i kamen mudraca. Zagreb: Algoritam. HP Danish 2000 Hanna Lützen (trans.) J. K. Rowling. Harry Potter og det vises sten. København: Gyldendal.

Romancisation worldwide

31

HP Greenlandic 2002 Stephen Hammeken (trans.) J. K. Rowling. Harry Potter ujarallu inuunartoq. Nuuk: Atuakkiorfik. HP Indonesian 2000 Lisitiana Srisanti (trans.) J. K. Rowling. Harry Potter dan batu bertuah. Jarkarta: Gramedia Pustaka Utama. HP Vietnamese 2000 L Lan (trans.) J. K. Rowling. Harry Potter và hòn dá phù thEy. TP. H Chí Minh: Nhà xut bàn trè. LPP Corsican 1990 Santu Casta (trans.) Antoine de Saint-Exupéry. U Principellu. Ajaccio: Akenaton. LPP Esperanto 1989 Pierre Delaire (trans.). Antoine de Saint-Exupéry. La eta princo. S.L.: Kanada Esperanto-Asocio. LPP Maltese 2000 Toni Aquilina (trans.) Antoine de Saint-Exupéry. I)-(kejken Prin)ep. Msida: Mireva. LPP Papiamentu 1982 Edward de Jongh and Stanley Cras (trans.) Antoine de Saint-Exupéry. E prens chikí. Willemstad: Editoryal Antiyano. LPP Rumanian n.d. Benedict Corlaciu (trans.) Antoine de Saint-Exupéry. Micul prin#. Bucureti: Ion Creang. LPP Seselwa 1985 Sabadin, James J. (trans.) Antoine de Saint-Exupéry. Pti prens. S.L.: Lenstiti Pedagozi Nasyonal. LPP Yiddish 2000 Shloyme Lerman (trans.) Antoine de Saint-Exupéry. Der kleyner prints. Nidderau: Michaela Naumann. Tazza 1994 Manwel Gauci. It-tazza tad-dinja. +rajja ta’ 64 sena. Storja, ri$ultati, ritratti. Pietà: Indipendenza.

References Aikhenvald, Alexandra Y. 2002 Language contact in Amazonia. Oxford: Oxford University Press. Aikhenvald, Alexandra Y. and R.M.W. Dixon (eds.) 2001 Areal diffusion and genetic inheritance: problems in comparative linguistics. Oxford: Oxford University Press.

32 Thomas Stolz Albalá Hernández, Paloma 2000 Americanismos en las Indias del Poniente. Voces de origen indígena americano en las lenguas del Pacífico. Frankfurt a.M.: Vervuert. 2001 Voces indígenas americanas en las lenguas del Pacífico. In Lo propio y lo ajeno en las lenguas austronésicas y amerindias. Procesos interculturales en el contacto de lenguas indígenas con el español en el Pacífico e Hispanoamérica, Klaus Zimmermann and Thomas Stolz (eds.), 279–294. Frankfurt am Main: Vervuert. Alexander, Ruth María 1980 Gramatica mixteca. Mixteco de Atatlahuca. México: Instituto Lingüístico de Verano. Arokianathan, S. 1999 Influence of French in 18th century Pondicherry Tamil dialect. In French in India and Indian nationalism 1700–1963, K. S. Mathew (ed.), 357–363. Pondicherry: Pondicherry University. Bakker, Dik, Jorge Gómez Rendón and Ewald Hekking 2007 Spanish meets Guaraní, Otomí and Quichua: a multilingual confrontation. This volume. Bakker, Peter 1997 A language of our own. The genesis of Michif, the mixed Cree-French language of the Canadian Métis. Oxford: Oxford University Press. Bakker, Peter and Robert A. Papen 2007 French influence on the native languages of Canada and adjacent USA. This volume. Bal, Willy 1974 Portuguese loan-words in Africa and the Orient. Aufsätze zur Portugiesischen Kulturgeschichte 13: 280–300. 1979a O destino de palavras de origem portuguesa num dialecto quicongo. In Afro-Romanica studia, Willy Bal (ed.), 95–146. Albufeira: Poseidon. 1979b À propos de mots d’origine portugaise en Afrique noire.In AfroRomanica studia, Willy Bal (ed.), 147–161. Albufeira: Poseidon. 1979c (ed.) Afro-Romanica studia. Albufeira: Poseidon. 1988 Portuguese loan-words in Africa. In Mélanges Willy Bal. Africana Romanica, Dieter Kremer (ed.), 33–52. Hamburg: Buske. Barker, Milton E. 1969 The phonoogical adaptation of French loanwords in Vietnamese. Mon-Khmer Studies 3: 138–147. Barrena, P. Natalio 1957 Gramatica annobonesa. Madrid: Consejo Superior de Investigaciones Científicas. Basset, Réné 1906 Les mots arabes passés en berbère. In Orientalische Studien. Theodor Nöldeke zum 70. Geburtstag gewidmet, Carl Bezold (ed.), 439–443. Gießen: Töppelmann, Vol. 1.

Romancisation worldwide

33

Bechert, Johannes 1981 Notiz über eine Möglichkeit, die historisch-vergleichende Sprachwissenschaft zu vervollständigen, oder: Lesefrüchte zur Verbesserung Mitteleuropas und anderer Weltgegenden. Papiere zur Linguistik 25: 85–89. Bidese, Ermenegildo et al. (eds.) 2005 Das Zimbrische zwischen Germanisch und Romanisch. Bochum: Brockmeyer. Boas, Franz 1930 Spanish elements in modern Nahuatl. In Todd memorial volume of philological studies, Vol. 1, N.N. (ed.), 87– 89. New York: Columbia University Press. Bolekia Boleká, Justo 1999 Aprender el bubi. Método para principantes. Madrid: Casa de África. Boretzky, Norbert and Birgit Igla 1994a Interferenz und Sprachwandel. In Sprachdynamik. Auf dem Weg zu einer Typologie sprachlichen Wandels. Vol. III: Interferenzlinguistik, Benedict Jeßing (ed.), 7–139. Bochum: Brockmeyer. 1994b Wörterbuch Romani-Deutsch-Englisch mit einer Grammatik der Dialektvarianten. Wiesbaden: Harrassowitz. Bossong, Georg 2003 O elemento português no japonês. In A língua portuguesa em viagem: actas do Colóquio Comemorativo do Cinquentenário do Leitorado de Português da Universidade de Zurique, 20 a 22 de Junho de 1996, Marília Mendes (ed.), 213–236. Frankfurt am Main. Bradshaw, A. T. 1965 Vestiges of Portuguese in the languages of Sierra Leone. Sierra Leone Language Review 4: 5–37. Bright, William 1979 Notes on hispanisms. International Journal of American Linguistics 45: 267–271. Bright, William, and R. W. Thiel 1965 Hispanisms in a modern Nahuatl dialect. Romance Philology 18: 442–452. Brincat, Joseph 2000 Il-Malti- elf sena ta’ storja. Malta: PIN. Brody, Jill 1995 Lending the ‘unborrowable’: Spanish discourse markers in Indigenous American languages. In: Spanish in four continents. Studies in language contact and bilingualism, Carmen Silva-Corvalán (ed.), 132– 47. Washington D.C.: Georgetown University Press. 1998 On hispanisms in elicitation. In Convergencia e individualidad. Las lenguas mayas entre hispanización e indigenismo, Andreas Koechert, and Thomas Stolz (eds.), 61–84. Hannover: Verlag für Ethnologie.

34 Thomas Stolz Calvo Pérez, Julio 2001 Préstamos, calcos y paráfrasis del castellano al quechua: mecanismos para su evaluación y adopción. In Lo propio y lo ajeno en las lenguas austronésicas y amerindias. Procesos interculturales en el contacto de lenguas indígenas con el español en el Pacífico e Hispanoamérica, Klaus Zimmermann and Thomas Stolz (eds.), 83–106. Frankfurt a.M.: Vervuert. Cifoletti, Guido 1986 Prestiti italiani nel dialetto del Cairo. Milano 2003 Note di aggiornamento sugli italianismi nel dialetto del Cairo. Plurilinguismo 10: 113–116. Curnow, T. J. 2001 What language features can be ‘borrowed’? In Areal diffusion and genetic inheritance: problems in comparative linguistics, Alexandra Y. Aikhenvald and R.M.W. Dixon (eds.), 412–436. Oxford: Oxford University Press. Dalgado, Sebastião Rodolfo 1982 Glosário luso-asiatico. 2 vols. Hamburg: Buske. 1988 Portuguese vocables in Asiatic languages. New Dehli: Asian Educational Services. De Granda, Germán 1985a Portuguesismos léxicos en el bubi y en el pidgin english de la isla de Bioko (Fernando Poo). In Estudios de lingüística afro-romanica, Germán de Granda (ed.), 117–126. Valladolid: Universidad de Valldolid. 1985b Préstamos léxicos de aculturación en dos lenguas bantu de Guinea Ecuatorial. In Estudios de lingüística afro-romanica, Germán de Granda (ed.), 127–140. Valladolid: Universidad de Valldolid. 1985c Procesos de aculturación léxica en el criollo portugués de Annobón. In Estudios de lingüística afro-romanica, Germán de Granda (ed.): 181–194. Valladolid: Universidad de Valldolid. 1985d (ed.) Estudios de lingüística afro-romanica. Valladolid: Universidad de Valldolid. De Pury-Toumi, Sybille 1979 La logique de l’emprunt en Mexicano (Nahuatl). Amerindia 4: 63–75. Etxebarria, Maitena 2004 Español y euskera en contacto: resulados lingüísticos. Revista Internacional de Lingüística Iberoamericana 2(4): 131–148. Farfán, José Antonio Flores de 1999 Cuatreros somos y toindioma hablamos. Contactos y conflictos entre el náhuatl y el español en el sur de México. México: CIESAS. França, António d’Oliveira Pinto da 2003 Portuguese influence in Indonesia. Lisboa: Prefácio.

Romancisation worldwide

35

Fusina, Jacques 2003 Le corse. In Les langues de France, Bernard Cerquiglini (ed.), 95– 106. Paris: Presse Universitaires de France. Giese, Wilhelm 1954 Portugiesische Lehnwörter im Mbundu (Angola). Orbis 3: 199–206. Haase, Martin 1992 Sprachkontakt und Sprachwandel im Baskenland. Die Einflüsse des Gaskognischen und Französischen auf das Baskische. Hamburg: Buske. Hair, P. E. H. 1980 Portuguese contacts with the Bantu languages of the Transkei, Natal and Southern Mozambique. African Studies 39: 3–46. Hajek, John and Catharina Williams-van Klinken 2003 Um sufixo românico numa língua austronésia. Revue de Linguistique Romane 66 (265/266): 55–65. Haspelmath, Martin 2001 The European linguistic area: Standard Average European. In Language typology and language universals, Martin Haspelmath, et al. (eds.), 1492–1520. Berlin/New York: Mouton de Gruyter. 2007 Loanword typology: Steps toward a systematic cross-linguistic study of lexical borrowability. This volume. Heath, Jeffrey 1989 From code-switching to borrowing. A case study of Moroccan Arabic. London, New York: Kegan Paul International. Heine, Bernd and Tania Kuteva 2005 Language contact and grammatical change. Cambridge: Cambridge University Press. Hekking, Ewald 1995 El otomí de Santiago Mexquititlán: desplazamiento lingüístico, préstamos y cambios gramaticales. Amsterdam: IFOTT. Hekking, Ewald and Pieter Musyken 1995 Otomí y quechua: una comparación de los elementos gramaticales prestados del aspanol. In Lenguas en contacto en Hispanoamérica. Nuevos enfoques, Klaus Zimmermann (ed), 101–118. Frankfurt a.M.: Vervuert. Hill, Jane H. and Kenneth C. Hill 1981 Regularities in vocabulary replacement in modern Nahuatl. International Journal of American Linguistics 47: 215–226. Hurch, Bernhard 1989 Hispanisierung im Baskischen. In Beiträge zum 5. Essener Kolloquium über ‘Grammatikalisierung: Natürlichkeit und Systemökonomie’ Vol. 1, Norbert Boretzky et al. (eds.), 11–35. Bochum: Brockmeyer.

36 Thomas Stolz Huth, Karin 1991 Portugiesische Lehnwörter im grammatischen System des Kimbundu (Angola). In Studien zum Portugiesischen in Afrika und Asien, Matthias Perl and Axel Schönberger (eds.), 21–38. Frankfurt a.M.: Teo Ferrer de Mesquita. Ibn Azzuz Haquím, Mohammad 1953 Glosario de mil quinientas voces españolas usadas entre los marroquies en el arabe vulgar. Madrid: Instituto de Estudios Africanos. Johanson, Lars 2000 Structural factors in Turkic language contacts. Richmond: Curzon. Kamiya, Taeko 1994 Tuttle new dictionary of loanwords in Japanese. Ruttland: Tuttle. Karttunen, Frances and James Lockhart 1976 Nahuatl in the middle years. Language contact phenomena in texts of the colonial period. Berkeley/Los Angeles/London: University of California Press. Kelkar-Stephan, Leena 2005 Bonjour maa: the French-Tamil language contact situation in India. Aachen: Shaker. Kiraithe, Jacqueline M. and Nancy T. Baden 1976 Portuguese influences in East African languages. African Studies 35: 3–31. Knowlton, Edgar C. 1966 Portuguese-Tamil linguistic contacts. In Proceedings of the 1st International Conference of Tamil Studies, N.N. (ed.), 607–617. Kuala Lumpur. 1976 Portuguese and Thai language contacts. Revista Portuguesa de Filologia 16: 1–11. Koechert, Andreas and Thomas Stolz (eds.) 1998 Convergencia e individualidad. Las lenguas mayas entre hispanización e indigenismo. Hannover: Verlag für Ethnologie. Kolmer, Agnes 2005 Subjektklitika als Kongruenzmarkierer: ein Vergleich zwischen bairischen und alemannischen Sprachinseldialekten in Norditalien (Zimbrisch und Walserdeutsch). In Das Zimbrische zwischen Germanisch und Romanisch, Ermenegildo Bidese et al., 164–192. Bochum: Brockmeyer. Krier, Fernande 1976 Le maltais au contact de l’italien. Hamburg: Buske. Lafitte, Pierre 1998 Grammaire basque. Donostia: Elkarlanean. Lastra, Yolanda and Jorge A. Suárez 1980 La investigación de las interferencias entre las lenguas indígenas y el español. In Perspectivas de la investigación lingüística en Hispano-

Romancisation worldwide

37

américa, Juan M. Lope Blanch (ed.), 31–43. México: Universidad Nacional Autónoma de México. Lébikaza, Kézié Koyenzi 1992 L’interférence des termes de parenté français dans le kabyé des locuteurs natifs en contact avec la langue française. Afrikanistische Arbeitspapiere 32: 65–84. Legère, Karsten 1989 Portugiesische Lehnwörter im Swahili. In Beiträge zur Afrolusitanistik und Kreolistik, Matthias Perl (ed.), 113–123. Bochum: Brockmeyer. Makihara, Miki 2001 Modern Rapanui adaptations of Spanish elements. Oceanic Linguistics 40 (2): 191–223. Martins, M. A. de Morais 1958 Contribução ao estudo da influência do português na língua quicongo. Garcia de Orta 6 (1): 33–51. Matras, Yaron 1998 Utterance modifiers and the universals of grammatical borrowing. Linguistics 36 (2): 281–331. 2002 Romani. A linguistic introduction. Cambridge: Cambridge University Press. Maurer, Philippe 1995 L’angolar. Un créole afro-portugais parlé à São Tomé. Hamburg: Buske. Moyse-Faurie, Claire 2007 Borrowings from Romance languages in Oceanic languages. This volume Muysken, Pieter 1988 Media Lengua and linguistic theory. Canadian Journal of Lingustics 33 (4): 409–422. 2001 Spanish grammatical elements in Bolivian Quechua: the Transcripciones Quechuas Corpus. In Lo propio y lo ajeno en las lenguas austronésicas y amerindias. Procesos interculturales en el contacto de lenguas indígenas con el español en el Pacífico e Hispanoamérica, Klaus Zimmermann and Thomas Stolz (eds.), 59–82. Frankfurt a.M.: Vervuert. Myers-Scotton, Carol 1993a Duelling languages. Grammatical structure in codeswitching. Oxford: Clarendon Press. 1993b Social motivation for codeswitching. Evidence from Africa. Oxford: Oxford University Press. Ney, Karin 1984 Rumänische Transferenzen in vier siebenbürgisch-sächsischen Ortmundarten des Kreises Hermannstadt, Rumänien. Marburg: Elvert.

38 Thomas Stolz O’Flynn de Chaves, Carol 1990 Tiempo, aspecto y modalidad en el criollo sanandresano. Bogotá: Centro Colombiano de Estudios en Lenguas Aborígenes. Parodi, Claudia 1998 Hispanismos léxicos en las lenguas de América y Asia y la pronunciación del Español. In Convergencia e individualidad. Las lenguas mayas entre hispanización e indigenismo, Andreas Koechert and Thomas Stolz (eds.), 85–114. Hannover: Verlag für Ethnologie. Pasch, Helma 1997 Entlehnungen aus dem Portugiesischen in afrikanischen Sprachen. In Afrolusitanistik – eine vergessene Disziplin in Deutschland? Ruth Degenhardt, Thomas Stolz and Hella Ulferts (eds.), 352–384. Bremen: Universität Bremen. Petrucci, Peter R. 1999 Slavic features in the history of Rumanian. München: LINCOM Europa. Quilis, Antonio 1973 Hispanismos en tagalo. Canadian Journal of Romance Linguistics 1: 70–92. 1976 Hispanismos en cebuano. Contribución al estudio de la lengua española en Filipinas. Madrid: Alcalá. Raidt, Edith H. 1983 Einführung in die Geschichte und Struktur des Afrikaans. Darmstadt: Wissenschaftliche Buchgesellschaft. Ramat, Paolo and Thomas Stolz (eds.) 2002 Mediterranean languages. Bochum: Brockmeyer. Ramos, Teresita V. 1971 Tagalog structures. Honolulu: University of Hawai’i Press. Rampton, Ben 1995 Language crossing and the problematisation of ethnicity and socialisation. Pragmatics 5 (4): 485–513. Rasoloson, Janie Noëlle 1997 Lehrbuch der madagassischen Sprache. Hamburg: Buske. Rodríguez-Ponga, Rafael 1995 El elemento español en la lengua chamorra (Islas Marianas). Unpublished PhD-thesis. Madrid: Facultad de Filología, Universidad Complutense. 1997 Elementos españoles en la lengua de Ponapé. In España y el Pacífico, Antonio Garcia-Abásolo (ed.), 295–305. Córdoba: Asociación Española de Estudios del Pacífico. 1999a ¿Qué se hablaba en las islas Marianas a finales del siglo XIX? In 1898: España y el Pacífico. Interpretación del pasado, realidad del presente, Miguel Luque Tavalán et al. (eds.), 521–526. Madrid: Asociación Española de Estudios del Pacífico.

Romancisation worldwide 1999b

39

Neologismos españoles en la lengua chamorra (Islas Marianas). In V Jornadas de metodología y didáctica de la lengua española: el neologismo, José Manuel González Calvo et al. (eds.), 155–163. Cáceres: Universidad de Extremadura. Rose, Françoise and Odile Renault-Lescure 2007 Contact-induced changes in Amerindian languages of French Guiana. This volume. Ross, Malcolm 2001 Contact-induced change in Oceanic languages in North-West Melanesia. In Areal diffusion and genetic inheritance: problems in comparative linguistics, Alexandra Y. Aikhenvald, and R.M.W. Dixon (eds.), 134 –160. Oxford: Oxford University Press. Rowley, Anthony R. 2003 Liacht as de sproch. Grammatik des Deutsch-Fersentalerischen. Lusern: Istituto Culturale Mòcheno-Cimbro. Sakel, Jeanette and Yaron Matras 2007 Modelling contact-induced change in grammar. This volume. Sampson, K. A. 1990 Le role linguistique du francais et de l’anglais dans l’expression des concepts techniques et semi-techniques dans les langues africaines. Afrikanistische Arbeitspapiere 22: 101–114. Sandfeld, Kristian 1968 Linguistique balkanique. Problèmes et résultas. Paris: Klincksieck. Silva-Corvalán, Carmen (ed.) 1995 Spanish in four continents. Studies in language contact and bilingualism. Washington D.C.: Georgetown University Press. Stairs, Glem A. and Emily F. de Stairs 1983 Huave de San Mateo del Mar. México: Colegio de México. Steenwijk, Han 1992 The Slovene dialect of Resia: San Giorgio. Amsterdam: Rodopi. Stolz, Christel 1998 Hispanicisation in modern Yucatec Maya: grammatical borrowing. In Convergencia e individualidad. Las lenguas mayas entre hispanización e indigenismo, Andreas Koechert and Thomas Stolz (eds.), 165–194. Hannover: Verlag für Ethnologie. Stolz, Christel and Thomas Stolz 1995 Spanisch-amerindischer Sprachkontakt: Die ‘Hispanisierung’ mesoamerikanischer Komparationsstrukturen. Iberoamericana 58/59 (2/3): 5–38. 1996a Funktionswortentlehnung in Mesoamerika. Spanisch-amerindischer Sprachkontakt (Hispanoindiana II). Sprachtypologie und Universalienforschung 49 (1): 86–123. 1996b Transpazifische Entlehnungsisoglossen. Hispanismen in Funktionswortinventaren beiderseits der Datumsgrenze. In Areale, Kontakte,

40 Thomas Stolz Dialekte. Sprache und ihre Dynamik in mehrsprachigen Situationen, Norbert Boretzky et al. (eds.), 262–291. Bochum: Brockmeyer. 1997 Universelle Hispanismen? Von Manila über Lima bis Mexiko und zurück: Muster bei der Entlehnung spanischer Funktionswörter in die indigenen Sprachen Amerikas und Austronesiens. Orbis 39: 1–77. 2001 Hispanicised comparative constructions in indigenous languages of Austronesia and the Americas. In Lo propio y lo ajeno en las lenguas austronésicas y amerindias. Procesos interculturales en el contacto de lenguas indígenas con el español en el Pacífico e Hispanoamérica, Klaus Zimmermann and Thomas Stolz (eds.), 35–56. Frankfurt a.M.: Vervuert. Stolz, Thomas 1998 Die Hispanität des Chamorro als sprachwissenschaftliches Problem. Iberoamericana 70 (2): 5–38. 2002 General linguistic aspects of Spanish-indigenous language contacts with special focus on Austronesia. Bulletin of Hispanic Studies 79: 133–158. 2003 Not quite the right mixture. Chamorro and Malti as candidates for the status of mixed language. In The Mixed Language debate. Theoretical and empirical advance, Yaron Matras and Peter Bakker (eds.), 271–315. Berlin /New York: Mouton de Gruyter. 2005 Italianisierung in den alloglotten Sprachen Italiens. In Das Zimbrische zwischen Germanisch und Romanisch, Ermenegildo Bidese, et al. (eds.), 43–70. Bochum: Brockmeyer. 2007 Allora: On the recurrence of function-word borrowing in contact situations with Italian as donor language. In Connectivity in grammar and discourse, Jochen Rehbein, Christiane Hohenstein and Lukas Pietsch (eds.), 75–99. Amsterdam /Philadelphia: John Benjamins. Stolz, Thomas and Anna Sabater Fuentes 2002 El uso del artículo indeterminado en la primera traducción chamorra del nuevo testamento.In Traspasando fronteras: el reto de Asia y el Pacífico (II), Francisco Javier Antón Burgos, and Luis Oscar Ramos Alonso (eds.), 731–739. Valladolid: Centro des Estudios de Asia. Stolz, Thomas, Dik Bakker and Rosa Salas Palomo forthc. Hispanisation. The impact of Spanish on the lexicon and grammar of the indigenous languages of Austronesia and the Americas. Berlin / New York: Mouton de Gruyter Suárez, Jorge A. 1977 La influencia gramatical del español en la estructura gramatical del náhuatl. Anuario de Letras 3: 115–164. Ternes, Elmar 1992 The Breton language. In The Celtic languages, Donald MacAulay (ed.), 371–452. Cambridge: Cambridge University Press.

Romancisation worldwide

41

Thomason, Sarah G. 2001 Language contact. Washington, D.C.: Georgetown University Press. Topping, Donald 1973 Chamorro reference grammar. Honolulu: University of Hawai’i Press. Tosco, Mauro 2007 A case of weak romanisation: Italian in East Africa. This volume. Verbeeck, Lieve 1998 Borrado de la memoria, grabado en la lengua: interferencias españolas e inglesas en el Mopan de Belice. In Convergencia e individualidad. Las lenguas mayas entre hispanización e indigenismo, Andreas Koechert and Thomas Stolz (eds.), 141–164. Hannover: Verlag für Ethnologie. Williams-van Klinken, Catharina, John Hajek and Rachel Nordlinger 2002 A descriptive grammar of Tetun Dili. Canberra: Pacific Linguistics. Wolff, John U. 1972 A dictionary of Cebuano. Ithaca: Cornell University. 2001 The influence of Spanish on Tagalog. In Lo propio y lo ajeno en las lenguas austronésicas y amerindias. Procesos interculturales en el contacto de lenguas indígenas con el español en el Pacífico e Hispanoamérica, Klaus Zimmermann, and Thomas Stolz (eds.), 233–252. Frankfurt a.M.: Vervuert. Zimmermann, Klaus 1987 Préstamos gramaticalmente relevantes del español al otomí: una aportación a la teoría del contacto entre lenguas. Anuario de Lingüística Hispánica 3: 223–253. 1999 Politica del lenguaje y planificación para los pueblos amerindios: ensayos de ecología lingüística. Frankfurt a.M.: Vervuert. 2004a El contacto de las lenguas amerindias con el español en México. Revista Internacional de Lingüística Iberoamericana 2 (4): 19–40. 2004b Die Frage der Sprache hinter dem Sprechen: Was kann die Gehirnforschung dazu beitragen? In Towards a dynamic theory of language. A festschrift for Wolfgang Wildgen on occasion of his 60th birthday, Andrea Graumann, et al. (eds.), 21–57. Bochum: Brockmeyer. 2005a Die Entstehung der Kreolsprachen: ein Versuch aus konstruktivistischer Sicht. In Portugiesische Kreolsprachen: Entstehung, Entwicklung, Ausbau und Verwendung, Annette Endruschat and Axel Schönberger (eds.) 11–43. Frankfurt a.M.: DEE. 2005b Interferenz, Transferenz und Sprachmischung: Prolegomena zu einer konstruktivistischen Theorie des Sprachkontaktes. In Das Zimbrische zwischen Germanisch und Romanisch, Ermenegildo Bidese et al. (eds.), 3–23. Bochum: Brockmeyer. 2007 Constructivist theory of language contact and the Romancisation of indigenous languages. This volume.

42 Thomas Stolz Zimmermann, Klaus and Thomas Stolz (eds.) 2001 Lo propio y lo ajeno en las lenguas austronésicas y amerindias. Procesos interculturales en el contacto de lenguas indígenas con el español en el Pacífico e Hispanoamérica. Frankfurt a.M.: Vervuert.

Loanword typology: Steps toward a systematic cross-linguistic study of lexical borrowability Martin Haspelmath

1. Introduction This paper gives an overview of some of the general issues arising when one studies lexical borrowing across languages. It discusses the motivations and goals (Section 2), kinds of loanwords (Section 3), factors influencing borrowability (Section 4), and factors determining the borrowing behavior of different languages (Section 5). The context is a collaborative crosslinguistic project on loanwords and lexical borrowability coordinated by the Max Planck Institute for Evolutionary Anthropology (2007–2008).1 2. Motivation and goals One of the most important tasks of diachronic linguistics is to establish general constraints on language change. There are two main types of constraints on language change: paths of change, which limit the direction that changes can take (cf. Haspelmath 2004), and rates of change, which give us an idea about the frequency or speed with which certain types of changes occur. Constraints on language change are of interest for at least two reasons: (i) A theoretical reason: Understanding the nature of language change presupposes identifying constraints on language change. If there were no such constraints, if anything goes, then we would have a difficult time understanding how and why change occurs. (ii) An applied/practical reason: Constraints on language change are a prerequisite for reconstructing unattested changes and unattested linguistic situations. An applied area of particular interest is the reconstruction of linguistic family trees. Linguistic family trees are increasingly found relevant by researchers in other disciplines such as archeology and molecular anthropology (e.g. Renfrew et al. 2000; McMahon and McMahon 2003). Whether genealogical classification is based primarily on shared innovations (as in

44 Martin Haspelmath the comparative method) or on shared retentions (as in lexicostatistics), constraints on language change are absolutely crucial for the reconstruction. For instance, historical phonologists routinely make use of information about likely phonological changes. Consider a hypothetical situation as in (1), where A, B, and C stand for three related languages. (1)

A

B

C

reconstructed:

p h i s

p s y 

f h y 

*p *s *y ?*s ?*

Since historical linguists know that p > f, s > h, and y > i are quite likely unconditioned changes, they would reconstruct *p, *s, and *y, regardless of the number of daughter languages that preserve these sounds. Where nothing is known about directionality (as in the case of s and S), we do not know which sound to reconstruct. In phonology, this knowledge seems to be largely impressionistic, even after almost two centuries of research on sound changes in the world’s languages. No handbook of attested phonological changes exists so far.2 Likewise, we have very little systematic information about general tendencies of lexical semantic change, although again a lot of research has been devoted to this topic (e.g. Wilkins 1996; Blank 1997). The only area where the research of the last few decades has been summarized in a handbook-like publication is the area of grammatical semantics (grammaticalization, see Heine and Kuteva 2002). However, while we know quite a bit about paths of change for grammaticalization, we know next to nothing about the rate of change here. In the area of lexical change, a lot of work has been done on the rates of change in the lexicostatistical research tradition. While Swadesh (1955) and Lees (1953) seem to have assumed that the replacement rate for lexical items is identical for all the lexical meanings under consideration, it became soon clear that this assumption is overly optimistic (e.g. Sankoff 1970). Moreover, it was often pointed out that the rate of replacement of words on the Swadesh lists is different in different historical situations (more recently e.g. by Blust 2000). Some proposals have been made for identifying the most stable lexical meanings (e.g. Dyen et al. 1967; Lohr 1999), but none of these has been based on a broad database from a representative sample of the world’s languages. One of the most important confounding factors for any type of approach to genealogical classification is lexical borrowing. Researchers in lexico-

Loanword typology 45

statistics have long recognized this, and at the beginning the hope was apparently that the 200 words on the Swadesh list (or at least the 100 words on the reduced list, Swadesh 1955) are so resistant to borrowing that this confounding factor can be neglected for these word meanings. However, the Swadesh lists do not seem to have been based on any kind of systematic research, but just on Swadesh's intuitive sense of which word meanings would be the most easily identified across languages and at the same time likely to be highly conservative. In any event, subsequent research on language contact showed that borrowing can be quite massive, and that the situation of well-studied Indo-European languages such as German, French or Russian (where loanwords are easy to identify and occur in rather circumscribed domains) may be quite atypical (cf. Thomason and Kaufman 1988). Thus, it is of paramount importance for lexicon-based historical linguistics to get a clearer idea about the differential borrowability of different types of words. It should not be difficult to advance our knowledge beyond what current textbooks have to offer in this regards. Typically they invoke a vague notion of “basic vocabulary” (or “core vocabulary”), e.g. From a purely linguistic perspective, the most important fact is that different spheres of the vocabulary are borrowed more easily, others significantly less easily. For instance, the most successful resistance to borrowing is offered by BASIC VOCABULARY, words referring to the most essential human activities, needs, etc., such as eat, sleep; moon, rain; do, have, be, … (Hock and Joseph 1996: 257) Here it should be noted that the distinction between basic and nonbasic vocabulary is a rough practical distinction, not a well-supported theoretical notion. Several decades ago … Morris Swadesh devised two lists of basic vocabulary items… His goal was to include only items that are unlikely to be borrowed… There was, and is, no theoretical foundation for this notion of universal-and-thus-hard-to-borrow basic vocabulary, and in fact all the items on Swadesh's list can and have been borrowed. Still, the lists are useful, because in most cases are at least less likely to be borrowed than more culture-specific vocabulary… (Thomason 2001: 71–72)

Hock and Joseph give a sketchy definition of “basic vocabulary” and a short list of examples, but neither is particularly helpful. And Thomason makes no attempt to go beyond Swadesh. The idea of the ongoing Loanword Typology project is that it should be possible to get a clearer idea of lexical borrowability by examining the loanwords in a reasonably representative and reasonably large set of languages

46 Martin Haspelmath (say, 30–40 languages), and by making inductive generalizations over the data assembled in this way. The planned outcome of the project will be an edited volume consisting of 30–40 language-particular chapters and a number of more general chapters that explain the methodological choices and discuss the results. Each language-particular chapter will be authored by a specialist of the language who knows enough about neighboring languages and historical-comparative linguistics of the family to identify the loanwords in the language. The project identifies a fixed list of word meanings which are translated into each language. The list consists of 1460 lexical meanings, most of which are taken from Mary Ritchie Key’s Intercontinental Dictionary Series.3 Each chapter consists of a data part and a discussion part: the data is a list of those words on the project list that can be identified as loanwords, plus perhaps other loanwords whose meanings are more specialized and hence do not appear on the project list. In addition, the source of each loanword is identified to the extent that it is known. The discussion part attempts to generalize over the data and puts the loanword into the relevant context (structural, historical, cultural, etc.), trying to explain why these words and no others were borrowed. While the data part is relatively standardized, authors have a lot of freedom in the discussion part. The remainder of this paper will mention a few general issues that will be relevant for any project that studies lexical borrowability in a comparative perspective.

3. Kinds of loanwords Let me start with some terminology. It is now customary to use the terms recipient language for the language that acquires a loanword and donor language for the language that is the source of the loanword. A loanword can be defined as a word that is transferred from a donor language to a recipient language, and it should not necessarily be equated with “borrowed word”, because some linguists define borrowing in a narrow way that excludes the effects of shift-induced interference or substrate (e.g. Thomason and Kaufman 1988: 37ff.). More general terms for contact-induced change are transfer and copying (Johanson 2002). According to Ross (1991), two other kinds of contact situations need to be distinguished, in addition to typical borrowing and typical shift-induced interference. He notes that typical borrowing is created by native speakers who consciously import a word from another language, whereas typical shift-induced interference is created by non-native speakers who uncon-

Loanword typology 47

sciously impose features of their native language to the recipient language. But imposition may happen to native speakers as well, especially when their native language is not their dominant language. In such cases, according to Ross, native speakers may transfer syntactic features from a dominant language to their native language, resulting in metatypy (see also Ross 1997, 2001). Moreover, in addition to importing words from a language spoken by a different group into their language, speakers may also import words from a language of their own group into the majority language, thus creating a new variety of the majority language that expresses the minority group’s cultural identity. An example of this might be the Yiddish words used in English by Jewish Americans. Table 1. Ross’s (1991) four types of contact-induced change agents of change: native speakers non-natives processing ease motivation for transfer:

metatypy

substrate (= shift-induced interference)

“exo-borrowing” (= adoption)

“endo-borrowing” (= retention)

[imposition] “culture” [importation]

One of the best-known taxonomies of borrowed items comes from Haugen (1950), who distinguishes between loanwords (form and meaning are copied completely), loanblends (words consisting of a copied part and a native part), and loanshifts, where only the meaning is copied. Loanshifts fall into two subtypes: loan translations (like Spanish rasca-cielos from sky-scraper) and semantic borrowings, where only the meaning is copied (like German kontrollieren, which originally only meant ‘check’, but is now also used in the sense ‘have control over’, as in English). Myers-Scotton (2002: 239) distinguishes between cultural borrowings and core borrowings, which according to her have very different origins. Cultural borrowings are words for new objects (e.g. espresso) or words for new (non-object) concepts (e.g. zeitgeist), and they usually appear abruptly when influential groups use them. Core borrowings, by contrast, are words that more or less duplicate already existing words (e.g. OK in German, which replaces gut, or einverstanden). Core borrowings “usually begin life in the recipient language when bilinguals introduce them as singly occurring codeswitching forms in the mixed constituents of their codeswitching”.

48 Martin Haspelmath 4.

Factors for differential borrowability of word meanings

4.1. Borrowability scales The most important type of constraint on borrowing that has been discussed in the literature is the borrowability scale (also called borrowing hierarchy).4 For instance, Matras (1998) proposes the scale in (2) for coordinators: (2)

‘but’ > ‘or’ > ‘and’

Similarly, Field (2002: 38) proposes the scale in (3): (3)

content item > function word > agglutinating affix > fusional affix

Such scales can be interpreted in three ways: (i) Temporal: A language borrows elements on the left before it borrows elements further to the right. (ii) Implicational: A language that contains borrowed elements on the right also contains borrowed elements further to the left. (iii) Quantitative: A language borrows more elements belonging to the types on the left than elements belonging to the types further to the right. (iv) Probabilistic: Elements belonging to the types on the left are more likely to be borrowed than elements further to the right. The temporal and implicational interpretations are generally difficult to distinguish, as are the quantitative and probabilistic interpretations. It should be noted, however, that absolute quantities are of little interest. Thus, Haugen (1950: 224) notes that 75.5% of all American Norwegian loanwords in his corpus are nouns, but he does not say what percentage of all the words are nouns. If the entire corpus contains 75.5% nouns, then loanwords show no special behavior and there is no basis for saying that nouns are borrowed preferentially. One of the goals of the Loanword Typology project is to find implicational borrowability scales for (sets of) lexical meanings (see Section 4.5, and Table 6 in Section 6.2). 4.2. Morpheme type It is widely acknowledged that lexical items are more likely to be borrowed than grammatical items, and that words are more likely to be borrowed than bound morphemes (e.g. Moravcsik 1978). Field (2002) adds the claim that

Loanword typology 49

agglutinative affixes are borrowed more easily than fusional affixes (see [3] above). Van Hout and Muysken (1994) cite supporting data on content vs. functions word types from their Quechua corpus (which contains many Spanish borrowings), shown in Table 2. Table 2. Native and borrowed function vs. content words in Quechua

function word content word total

not borrowed

borrowed

total

105 (80.8%) 592 (63.7%) 697

25 (19.2%) 338 (36.3%) 363

130 (100%) 930 (100%) 1060

p = 0.0001174

4.3. Parts of speech It is widely acknowledged that nouns are borrowed more easily than other parts of speech (e.g. Whitney 1881, Moravcsik 1978, Myers-Scotton 2002: 240). Van Hout and Muysken (1994: 42) give the following explanation: A very important factor involves one of the primary motivations for lexical borrowing, that is, to extend the referential potential of a language. Since reference is established primarily through nouns, these are the elements borrowed most easily.

According to Myers-Scotton (2002: 240), nouns are borrowed preferentially “because they receive, not assign, thematic roles”, so “their insertion in another language is less disruptive of predicate-argument structure”. Van Hout and Muysken (1994) again cite data from their Quechua corpus, shown in Table 3. Table 3. Native and borrowed nouns vs. verbs in Quechua Quechua native words

Quechua Spanish loans

total

Types Nouns Verbs

194 214

184 (49%) 81 (27%)

378 (100%) 295 (100%)

Tokens Nouns Verbs

1,101 1,929

823 (42%) 241 (11%)

1,924 (100%) 2,170 (100%)

50 Martin Haspelmath These data show that in Quechua, a much higher percentage of noun types and tokens are loanwords than verb types and tokens. The difficulty of borrowing verbs as verbs has sometimes been addressed in the literature. For instance, it has been said that verbs cannot be borrowed in French because of their elaborate inflection, so that it is difficult to incorporate other languages’ verbs into French (Meillet 1921, cited in Thomason and Kaufman 1988: 348; however, French does have loan verbs such as shooter). Moravcsik (1975, 1978: 111–112) observed that if verbs are borrowed, they seem to be borrowed as if they were nouns: the borrowing language employs its own means of denominal verbalization to turn the borrowed forms into verbs “before” using them as such (see also Moravcsik 2003, Wichmann and Wohlgemuth 2008). Not much is known about adjective borrowing, but adjectives remind us of the fact that when we talk about borrowability of different parts of speech, we have to specify whether we mean donor part of speech or recipient part of speech (cf. Curnow 2001: 415). As is well known, part-of-speech systems differ quite dramatically when it comes to adjectives, so here both the donor and the recipient systems should be taken into account. A concrete example is an observation made by Dik Bakker (p.c.): Quechua borrows considerably more (donor-language) adjectives from Spanish than Otomí does, which may well have to do with the fact that traditional Otomí lacks adjectives (see the corpus counts in Hekking and Muysken 1995 and Hekking and Bakker 1999). 4.4. The role of token frequency One way in which the notions “basic vocabulary” and “core vocabulary” can be interpreted is as the words which are used most frequently. It would not be surprising if they were resistant to borrowing, because it is well known that high-frequency items are resistant to other types of language change such as analogy. Van Hout and Muysken (1994) find some evidence for this in their Quechua corpus. The following table gives the percentage of Spanish loans in 7 frequency classes.

Loanword typology 51 Table 4. Word frequency and borrowing rate token frequency Quechua Quechua Spanish total types % borrowed from Spanish frequency class native types loan types 1 2–3 4–7 8–15 16–31 32–63 > 63

1 2 3 4 5 6 7

185 116 86 72 27 17 14

170 99 46 29 11 7 1

355 215 132 101 38 24 15

47.9 46.0 34.8 28.7 28.9 29.2 6.7

total

Van Hout and Muysken conclude that frequency in the recipient language may operate as an inhibiting factor for borrowing. 4.5. Lexical semantic field One area where little systematic research has been done is the lexical semantic fields that loanwords tend to come from. However, it seems clear that there are many regularities here. For instance, victorious invaders will typically borrow placenames, names for local plant and animal species, and languages of peoples ruled by a foreign invaders will typically adopt military terms (see, e.g., Vennemann’s 2000 inference that the Germanic peoples must once have been dominated by a foreign ruling class, perhaps of northern African origin). It is here that the Loanword Typology project can make an important contribution. 5.

Factors for differential borrowing behavior among different languages

5.1. Intensity of language contact The most obvious sociolinguistic factor favoring borrowing is widespread bilingualism. This is often called “intensity of contact”. Thomason and Kaufman (1988) propose a five-point scale of intensity of contact:

52 Martin Haspelmath [1]: [2]: [3]: [4]: [5]:

casual contact, slightly more intense contact, more intense contact, strong cultural pressure, very strong cultural pressure).

They claim that beginning with stage 3, we also find “nonbasic” vocabulary among the loanwords. Brown (1999) finds that there is significantly more borrowing into Native American languages from Spanish than from English or French. The likely explanation, according to Brown, is that Native Americans have often been bilingual in Spanish because they were integrated into Spanish society much more and earlier than Native Americans in the British and French colonies (and later the U.S. and Canada). An additional factor is probably the prestige of a language, although this of course correlates with widespread bilingualism (people are more likely to learn another language if it is prestigious). However, widespread bilingualism without great prestige does occur (e.g. Spanish and Guaraní in Paraguay, where Guaraní is spoken by many speakers of non-Guaraní origin, but Paraguayan Spanish has very few Guaraní loanwords), and languages may be widely regarded as prestigious but still few people speak it (e.g. French in 19th century Europe, or English in much of the world today). 5.2. Purism It is sometimes claimed that different cultures have different attitudes toward borrowing than others, which would explain differential borrowing behavior (e.g. it is claimed that Icelander are purists and hence their language has very few loanwords, cf. tölva ‘computer’, útvarp ‘radio’, etc.). However, unless there is legislation or language academies with a high degree of social acceptance, it seems to be difficult to find evidence for the exact role of speaker attitudes, and we must be careful to avoid circular reasoning. 5.3. Structural incompatibility Structural incompatibility has often been invoked as explaining resistance to borrowing, although in recent years it has come under attack (especially by Thomason and Kaufman 1988). For grammatical borrowing, it seems

Loanword typology 53

undeniable that it plays a role (e.g. it seems very unlikely that an isolating language like Vietnamese would borrow a case suffix), but it is not clear at present whether it might be relevant for lexical borrowing. 5.4. Genealogical relatedness McMahon (1994: 204) implies that related languages are more likely to borrow from each other, especially if they are so closely related that mutual intelligibility is relatively easy to establish. This would explain why English borrowed “basic vocabulary” items such as skin, sky, get, they, them, their from Old Norse. 6.

Establishing borrowability through language comparison: two examples

I this section I give two examples of the kinds of results that a systematic cross-linguistic study of loanwords can yield. 6.1. Items of acculturation in languages of the Americas Brown (1999) examined words for 77 “items of acculturation” (things/concepts unfamiliar to Native Americans before the European invasion) in 292 Native American languages. This research resulted in tables such as Table 5, where lexical meanings are ranked by borrowability. The number following each lexical meaning is the percentage of languages in which it is a loanword from a European language. Thus, “coffee” is a loanword in 81% of the languages for which Brown has information (not always all the 292 languages), whereas the remaining 19% have native words for “coffee” (compounds, derivatives, or simple words whose meaning was extended or shifted). Among many other things, Brown observed the following tendencies: (1) “Words for natural kinds tend more strongly than those for introduced artifacts to be associated with high borrowability.” (Brown 1999: 56) (2) Within the category of words for introduced living things, terms for animals tend to show greater borrowability scores than words for plants.

54 Martin Haspelmath Brown’s (1999: 66) explanation for the first tendency is that Native Americans encountered living things more often during interaction with Europeans, while artifacts were more often encountered in other contexts. Brown’s study seems to be the only systematic comparative study of loanwords so far in the literature. It is more limited than the Loanword Typology project in that it considers only 77 lexical meanings, but this allows him to take into account a very high number of languages. Table 5. Items of acculturation, ranked by index of borrowability (European loan percentage) coffee coriander cat garlic orange cheese donkey lemon apple cow Saturday pig soldier peach mule goat sugar tea horse rice table soap bottle

81 72 70 69 67 66 64 60 59 58 56 55 55 54 54 53 53 53 52 52 52 51 51

watermelon ribbon cabbage lettuce hour sheep onion key barley turnip button wheat apricot box school scissors Wednesday match nail candle spoon shovel bread

49 48 47 47 45 44 41 41 39 39 38 38 37 36 35 35 35 34 33 32 31 30 29

flour board butter wagon peas cup mile pistol window clock grapes oats book needle paper hundred beets thread chicken town rich money fork

29 28 28 28 28 27 27 27 27 25 25 23 22 21 20 19 18 17 16 15 13 11 10

Loanword typology 55

6.2. Words of the IDS list in 80 Austronesian languages The Comparative Austronesian Dictionary (Tryon 1995) contains lexical information on about 1300 lexical meanings for 80 Austronesian languages, including information on whether a word is known to be a loanword. The 1300 meanings come from the list of the Intercontinental Dictionary Series (see note 3). We went through all four volumes of the dictionary and identified the percentage of languages showing loanwords for each lexical meaning.5 Table 6 shows the average percentages for each lexical field. Table 6. IDS lexical fields, ranked by average percentage of loanwords in the 80 Austronesian languages of Tryon (1995) Chapter 4: Chapter 15: Chapter 2: Chapter 12: Chapter 16:

PARTS OF THE BODY; BODILY FUNCTIONS AND CONDITIONS

Chapter 1: Chapter 13: Chapter 10: Chapter 17: Chapter 19: Chapter 18: Chapter 20: Chapter 8: Chapter 11: Chapter 9:

PHYSICAL WORLD IN ITS LARGER ASPECTS

Chapter 14: Chapter 7: Chapter 21: Chapter 3: Chapter 5: Chapter 22: Chapter 6:

TIME

SENSE PERCEPTION MANKIND: SEX, AGE, FAMILY RELATIONSHIP SPATIAL RELATIONS: PLACE, FORM, SIZE EMOTION (WITH SOME PHYSICAL EXPRESSIONS OF EMOTION); TEMPERAMENTAL, MORAL, AND AESTHETIC NOTIONS QUANTITY AND NUMBER MOTION, LOCOMOTION, TRANSPORTATION, NAVIGATION MIND, THOUGHT TERRITORIAL, SOCIAL, AND POLITICAL DIVISIONS; SOCIAL RELATIONS VOCAL UTTERANCE, SPEECH; MUSIC WARFARE AND HUNTING AGRICULTURE, VEGETATION POSSESSION, PROPERTY, AND COMMERCE MISCELLANEOUS PHYSICAL ACTS AND THOSE PERTAINING TO ARTS AND CRAFTS, WITH SOME IMPLEMENTS, MATERIALS, AND PRODUCTS DWELLING, HOUSE, FURNITURE LAW ANIMALS FOOD AND DRINK; COOKING AND UTENSILS RELIGION AND BELIEFS CLOTHING; PERSONAL ADORNMENT AND CARE

2.33 2.65 3.02 3.85 4.59 5.36 5.36 5.95 9.70 12.04 12.47 13.16 13.34 13.64 14.73 17.28 20.26 21.09 22.99 23.16 28.22 31.83

Table 7 is an excerpt from the list of lexical items. Each lexical item is preceded by its IDS number and is followed by the percentage of Austronesian languages in the dictionary in which it is a loanword.

56 Martin Haspelmath Table 7. Some randomly selected IDS word meanings, ranked by percentage of loanwords in the 80 Austronesian languages of Tryon 1995 01.212 01.222 01.270 01.310 01.323 01.342 01.352 01.353 01.430 01.440 01.530 01.550 01.560 01.570 01.620 01.630 01.640 01.720 01.740 01.750 01.852 01.880 02.210 02.220 … 05.220 07.560 09.110 14.450 15.220 11.820 … 15.440 18.120 18.210 17.440 … 14.760 02.520 01.280

earth =ground, soil cliff, precipice shore water rough (of sea) reef tide lowtide wood stone, rock moon lightning thunder lightning (as striking) darkness shade, shadow dew wind fog rain (noun) burn (vb intrans) firewood man (vs.woman) woman

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

boil (vb) post, pole do, make noon, midday smell (vb trans) sell

3.89 3.89 3.89 3.89 3.89 3.94

sound, noise sing speak, talk suspect

7.5 7.5 7.5 7.57

summer aunt cave

11.62 11.66 11.68

12.540 20.460 02.760 13.440 . 19.110 19.370 20.310 07.470 08.830 22.220 … 14.780 07.580 09.422 06.730 05.370 03.370 … 03.260 06.920 08.691 20.170 05.270 08.240 … 14.530 18.560 03.560 03.410 07.240 05.630 … 05.930 03.780 05.760 03.460 03.470 03.770 05.880 05.890

measure surrender widow three times

11.68 11.76 12 12

country citizen, subject armor (defensive) shelf citrus fruit preach

20 20 20 20.28 20.89 21.21

season arch tool ring (for finger) spoon he-goat

31.14 31.57 31.57 31.74 31.94 32.07

ram brush pipe soldier kettle shovel

55.10 55.31 55.55 56.92 57.37 57.40

clock, timepiece paper goose horse (equine) key sausage

71.42 72 72.5 73.13 73.13 73.91

beer camel grape ass, donkey mule elephant cheese butter

91.30 93.33 94.11 94.59 100 100 100 100

Loanword typology 57

7.

Some choices of the Loanword Typology project

7.1. Which languages? Since the goal of the Loanword Typology project is to discover universals of lexical borrowing, the fundamental requirement is that the languages should be as diverse as possible, not only genealogically and geographically, but also sociolinguistically. There should be national languages with large numbers of speakers and great prestige, and there should be tribal languages with few speakers and little prestige for outsiders. There should be languages with a long written tradition, and unwritten languages. Moreover, the contact situations should be diverse: There should be languages that have undergone extensive lexical enrichment from outside sources, but also languages that have largely gotten by with their own lexical resources. There should be cases where the loanwords came in exclusively through the spoken language, and cases where many loanwords were introduced through writing. (Of course, in practice the choice of languages is to a large extent also determined by the linguists who are willing to collaborate on this project.) The general requirement of genealogical diversity does not exclude the possibility of including pairs of closely related languages. Such pairs may actually be particularly instructive when the two closely related languages (or varieties of the same language) have been associated with very different sociolinguistic circumstances (David Gil, p.c.). Possible pairs of this kind are Hindi/Urdu, Yiddish/German, Iranian Azerbaijani/Azerbaijani of the Republic of Azerbaijan, Riau Indonesian /Standard Indonesian. 7.2. Which lexical meanings? If only a small number of lexical meanings were covered (say a 200-word list), it would be easier to get data for a large number of languages. However, it is one of the main goals of this project to find out which lexical meanings are resistant to borrowing, so by including only those that have been thought to be resistant, we cannot really test Swadesh’s claim that his words are particularly resistant. Thus, a list such as the IDS word list (consisting of 1310 word meanings) is better suited for the Loanword Typology project. The IDS list does not contain any lexical meanings relating to the modern world (“radio”, “truck”, “hospital”, “election”, etc.), and it contains a disproportionate number of lexical meanings that are relevant primarily to

58 Martin Haspelmath the European (and similar) ecoregion. To make the list more balanced in these regards, 150 meanings have been added, so that the list now consists of 1460 lexical meanings. 7.3. What is a loanword? A loanword is defined as a word that at some point came into a language by transfer from another language.6 Thus, not only manga is a loanword in English (from Japanese, first attested in the OED in 1951), but also very (from French, first attested in 1250) and mill (from Latin, first attested in 962, but probably borrowed several centuries earlier). Evidently, identifying early loanwords such as English mill is not possible in languages that do not have a long written history. Thus, in order to make the data more comparable, it might be advisable to limit oneself to more recent loanwords (say, of the last 300-500 years) even in languages where we would be able to identify older loanwords. The Loanword Typology project is taking a more sophisticated approach: For each loanword, authors are asked to record the approximate time at which the word came into the language. This allows us to filter out older loanwords when we want to compare languages with a well-known history with languages about whose history little is known, and in which only young loanwords are likely to be recognizable as such. Notes 1. 2.

3.

4. 5.

See http://www.eva.mpg.de/lingua/files/lwt.html. However, Juliette Blevins (Max Planck Institute for Evolutionary Anthropology) has begun an effort in this direction (“Handbook of Phonological Change”). The Intercontinental Dictionary Series (IDS) is a long-term project founded by Mary Ritchie Key (University of California, Irvine) that aims to publish electronic lexical databases for a large number of languages from around the world. All these databases are based on the IDS Word List (consisting of 1310 items), which is itself an adaptation of the list used by Buck (1949). The IDS is currently being developed further by Bernard Comrie (Max Planck Institute for Evolutionary Anthropology) (see http://lingweb.eva.mpg.de/ids). Since the term “hierarchy”, when used outside language typology, is generally reserved for taxonomic hierarchies, I prefer the synonymous term “scale”. Thanks to my assistants Ulrike Gurt and Jenny Seeg for their invaluable help.

Loanword typology 59 6.

This definition presupposes continued existence of a language with no break in transmission. If creole languages are regarded as languages with a break in transmission and no straightforward genealogical ancestor, then the notion of loanword does not apply to creole languages.

References Blank, Andreas 1997 Prinzipien des lexikalischen Bedeutungswandels am Beispiel der romanischen Sprachen. Tübingen: Niemeyer. Blust, Robert 2000 Why lexicostatics doesn’t work: the ‘universal constant’ hypothesis and the Austronesian languages. In Time depth in historical linguistics, Colin Renfrew, April McMahon and Larry Trask (eds.), 311– 331. Cambridge: McDonald Institute for Archaelogical Research. Brown, Cecil H. 1999 Lexical acculturation in Native American languages. New York: Oxford University Press. Buck, Carl Darling 1949 A dictionary of selected synonyms in the principal Indo-European languages. Chicago: The University of Chicago Press. Burrow, Th. 1946 Loanwords in Sanskrit. Transactions of the Philological Society 1946, 1–30. Cavalli-Sforza, Luca L. and S.-Y. Wang William 1986 Spatial distance and lexical replacement. Language 62: 38–55. Curnow, Timothy Jowan 2001 What language features can be ‘borrowed’? In Areal diffusion and genetic inheritance, Alexandra Y. Aikhenvald, and R.M.W. Dixon (eds.), 412–436. Oxford: Oxford University Press. Dyen, Isidore, A. T. James and J. W. L. Cole 1967 Language divergence and estimated word retention rate. Language 43: 150–171. Field, Fredric W. 2002 Linguistic borrowing in bilingual contexts. Amsterdam: Benjamins. Haspelmath, Martin 2004 On directionality in language change with particular reference to grammaticalization. In Up and down the cline: The nature of grammaticalization, Olga Fischer, Muriel Norde and Harry Perridon (eds.), 17–44. (Typological Studies in Language 59.) Amsterdam: Benjamins.

60 Martin Haspelmath Haugen, Einar 1950 The analysis of linguistic borrowing. Language 26: 210–231. Heine, Bernd and Tania Kuteva 2002 World lexicon of grammaticalization. Cambridge: Cambridge University Press. Hekking, Ewald and Dik Bakker 1999 El Otomí y el español de Santiago Maxquititlán: Dos lenguas en contacto. Foro Hispánico 13: 45–74. Hekking, Ewald and Pieter Muysken 1995 Otomí y Quechua: una comparación de los elementos gramaticales prestados del español. In Lenguas en contacto en Hispanoamérica: Nuevos enfoques, Klaus Zimmermann (ed.), 101–118. Frankfurt a.M.: Vervuert. Hock, Hans Henrich and Brian D. Joseph 1996 Language history, language change, and language relationship. Berlin /New York: Mouton de Gruyter. Hymes, Dell 1960 Lexicostatistics so far. Current Anthropology 1: 3–44. Johanson, Lars 2002 Structural factors in Turkic language contacts. London: Curzon. Kapitan, M.E. 1994 Influence of various system features of Romance words on their survival. Journal of Quantitative Linguistics 1: 237–276. Lees, Robert B. 1953 The basis of glottochronology. Language 29: 113–127. Lohr, Marisa 1999 Methods for the genetic classification of languages. Ph.D. dissertation, University of Cambridge. Matras, Yaron 1998 Utterance modifiers and universals of grammatical borrowing. Linguistics 36: 281–331. McMahon, April 1994 Understanding language change. Cambridge: Cambridge University Press. McMahon, April and Robert McMahon 2003 Finding families: quantitative methods in language classification. Transactions of the Philological Society 101: 7–55. Moravcsik, Edith A. 1975 Verb borrowing. Wiener Linguistische Gazette 8: 3–30. 1978 Universals of language contact. In Universals of human language, volume 1, Method and theory, Joseph H. Greenberg et al. (eds.), 93– 122. Stanford, CA: Stanford University Press. 2003 Borrowed verbs. Ms., University of Wisconsin, Milwaukee.

Loanword typology 61 Myers-Scotton, Carol 2002 Language contact: Bilingual encounters and grammatical outcomes. Oxford: Oxford University Press. Oswalt, Robert Louis 1975 The relative stability of some syntactic and semantic categories. Working Papers on Language Universals (Stanford) 19: 1–19. 1971 Towards the construction of a standard lexicostatistic list. Anthropological Linguistics 13: 421–434. Poplack, Sandra and David Sankoff 1984 Borrowing: the synchrony of integration. Linguistics 22: 99–136. Poplack, Shana, David Sankoff and Christopher Miller 1988 The social correlates and linguistic processes of lexical borrowing and assimilation. Linguistics 26: 47–104. Renfrew, Colin, April McMahon and Larry Trask (eds.) 2000 Time depth in historical linguistics. 2 vols. Cambridge: McDonald Institute for Archaelogical Research. Ross, Malcolm 1991 Refining Guy’s sociolinguistic types of language change. Diachronica 8 (1): 119–129. 1997 Social networks and kinds of speech-community events. In Archaeology and language I: Theoretical and methodological orientations, Roger Blench, and Matthew Spriggs (eds.), 209–261. London: Routledge. 2001 Contact-induced change in Oceanic languages in north-west Melanesia. In Areal diffusion and genetic inheritance, Alexandra Y. Aikhenvald, and R.M.W. Dixon (eds.), 134–166. Oxford: Oxford University Press. Sankoff, David 1970 On the rate of replacement of word-meaning relationships. Language 46: 564–569. Swadesh, Morris 1955 Towards greater accuracy in lexicostatistic dating. International Journal of American Linguistics 21: 121–137. Szemerenyi, Oswald 1974 The origins of the Greek lexicon: Ex oriente lux. Journal of Hellenic Studies 94: 144–157. Thomason, Sarah Grey 2001 Language contact. Washington, D.C.: Georgetown University Press. Thomason, Sarah Grey and Terrence Kaufman 1988 Language contact, creolization, and genetic linguistics. Berkeley: University of California Press. Thomason, Sarah Grey 1997 On the unpredictability of contact effects. Ms., University of Michigan.

62 Martin Haspelmath Tryon, Darrell T. (ed.) 1995 Comparative Austronesian dictionary: an introduction to Austronesian studies. Parts 1–4. Berlin /New York: Mouton de Gruyter. van Hout, Roeland and Pieter Muysken 1994 Modeling lexical borrowability. Language Variation and Change 6: 39–62. Vennemann, Theo 2000 Zur Entstehung des Germanischen. Sprachwissenschaft 25.3: 233– 269. Whitney, William Dwight 1881 On mixture in language. Transactions of the American Philosophical Association 12: 1–26. Wichmann, Søren and Jan Wohlgemuth 2008 Loan verbs in a typological perspective. This volume. Wilkins, David 1996 Natural tendencies of semantic change and the search for cognates. In The comparative method reviewed, Malcolm Ross and Mark Durrie (eds.), 264–304. New York: Oxford University Press.

Modelling contact-induced change in grammar Jeanette Sakel and Yaron Matras

1. Introduction The outcomes of contact induced language change have been dealt with from many perspectives, such as 1. intensity and impact of contact (code switching, language change, etc.), 2. types of contact situations (linguistic areas, one to one borrowing situations, substrate influence, bilingualism, second language acquisition, etc.), 3. types of contact phenomena (lexical borrowings, grammatical borrowings, calques, borrowing of morphological material), and others such as 4. borrowing hierarchies (which elements are more likely to be borrowed than others). Usually, authors focus on one or only a few of these. Our aim in the context of the research project on Language Convergence and Linguistic Areas (supported by a grant from the Arts and Humanities Research Council) is to develop a systematic approach to analyse and compare the results of contact-induced language change. We wish to take into account the factors involved in different outcomes and to test different types of contact situations, such as linguistic areas. In particular, we wish to focus on the categories affected by contact and compare them to factors such as the social context. In the present paper we discuss our methods of describing contact-induced language change. We will draw on two case studies of languages in contact with Romance languages: the indigenous language Mosetén in contact with Spanish in the foothills of the Bolivian Andes, and the Kelderash dialect of Romani spoken in Romania in contact with Romanian. Since our aim is to present a systematic overview of contact-induced change, we must take into account different perspectives of investigating contact. Some definitions of the concepts associated with language contact have been extended or challenged in recent years. These include 1. borrowing vs. code switching, 2. linguistic areas vs. one to one borrowing situations, 3. matter borrowings vs. pattern replication, and 4. borrowing hierarchies. We will briefly discuss each of these. 1. Code switching and long-term borrowing have traditionally been conceived of as unrelated concepts (cf. the overview on code switching and language change given in Backus 2005). In this view, code switching is

64 Jeanette Sakel and Yaron Matras seen as a mechanism in which two languages are used within the same clause and in the same environment. This is a typical trait of bilingual communication. Borrowing, and with it long-term language change, on the other hand, is seen as the integration of elements from a source language into a recipient language (we use Weinreich’s 1953 terminology). These elements are integrated into the recipient language system and this process usually takes time. In this way, the original definitions of borrowing and code switching differed both in the outcome of the contact, i.e. one versus two languages used in one clause, as well as in the length of contact and “integration”. Most authors draw a clear line between the two when studying contact situations that involve both long-term (propagated) structures and bilingual communication; for example Poplack (1980). The reasons for this can be found in earlier studies of language contact, which often make a sharp contrast between on-the-spot contact phenomena (‘speech’) and integration of features (‘language’). For example Rozencvejg (1976) distinguishes between interference (speech) and convergence (language), treating each of them separately in the two major chapters of his book. His view follows Weinreich’s (1953[1966: 11]) influential work, in which he makes the distinction between interference in speech and interference in language: In speech, interference is like sand carried by a stream, in language, it is the sedimented sand deposited on the bottom of a lake. The two phases of interference should be distinguished. In speech, it occurs anew in the utterances of the bilingual speaker as a result of his personal knowledge of the other tongue. In language, we find interference phenomena which, having frequently occurred in the speech of bilinguals, have become habitualized and established. Their use is no longer dependent on bilingualism.

Weinreich’s distinction is less sharp than Rozencvejg’s (1976) and other interpretations of it. This is well-captured by his analogy of sand, for which we can imagine an intermediate state between being carried by a stream and slowly sedimenting on the bottom of a lake. In recent years, many authors have tried to avoid making a clear distinction between the two (cf. work by Myers-Scotton 1993, 2002; Field 2005, etc.); some argue that both concepts can be settled along a continuum of more-or-less integration (cf. Heath 1984; Gardner-Chloros 1995). 2. Other well-established distinctions, of which some have been questioned recently, involve the types of contact situations, such as linguistic areas, one-to one borrowing situations or instances of substrate influence. For example, authors used to draw a clear line between linguistic areas and other contact situations (cf. Thomason 2001, who treats linguistic areas in a

Modelling contact-induced change in grammar

65

separate chapter of her book). In many cases, though, it is difficult to distinguish between areas and borrowing situations. This becomes clear in e.g. Aikhenvald’s (2002) study of borrowing from Tucano (Tucanoan) into Tariana (Arawak) within the linguistic area of the Vaupés, even though many areal traits in Tucano can be attributed to Tariana, or at least Arawak, origins. For this and other reasons, the clear distinction between linguistic areas and one to one borrowing situations has been questioned. Campbell (2006) argues that areas are not fundamentally different from a mere density of shared isoglosses that go back to the diffusion of features between pairs of languages. Similarly, Bakker (2006) shows that within the socalled area of Sri Lanka similar features diffused within pairs of languages, leading to an overall similar profile of some language that have never been in direct contact with each other (cf. also Matras and Sakel 2007). 3. One of the distinctions between types of contact phenomena is between borrowed matter (MAT) and calques, i.e. patterns (PAT). This distinction has to do with the way elements are affected by the contact (cf. Matras and Sakel 2007; Sakel and Matras 2006).1 Weinreich (1953 [1966: 7]) distinguishes different types of interference: transferred elements vs. interference without outright transfer, which can also be seen in his terminology ‘source’ and ‘recipient’ language for the latter, i.e. MAT, and ‘model’ and ‘replica’ language for the former type of interference, i.e. PAT. Similar approaches are Haugen’s (1950) ‘substitution’ in loanshifts (calques) vs. ‘importation’ (outright borrowing), Go b’s (1956, 1959) ‘substance’ versus ‘form’, and Johanson’s (1992) ‘global copying’ and ‘partial copying’. Altogether, these concepts have been addressed by many authors in the past, but often they did not play a major role in classifying language contact situations (an exception to this is Heath 1984). Furthermore, a number of studies do not take this distinction into account at all, or only attribute minor importance to it. In this way, Thomason and Kaufman (1988) do not generally distinguish between MAT and PAT loans. 4. Other concepts that have been dealt with within language contact studies include the study of the categories affected and possible borrowing hierarchies. Most of the discussions on this are based on Thomason and Kaufman’s (1988: 74–76) borrowing scale. Their only factor for the degree and type of borrowing is the intensity of the contact; that is more intensive contact would lead to more categories affected in this model. While the intensity of the contact indeed plays an important role in what is borrowed, other studies have shown that it is not the only factor involved (cf. Matras and Sakel 2007; Sakel and Matras 2006). Furthermore, Thomason and Kaufman (1988) do not take into account the difference between MAT and

66 Jeanette Sakel and Yaron Matras PAT borrowings, which appear to have very different distributions in different contact situations. Other discussions of borrowing hierarchies and related concepts have been presented by Moravcsik (1978), Ross (1988) and Lass (1997). In the same way as Thomason and Kaufman’s (1988) borrowing scale, these discussions are top-down in theorising from case studies in the literature and not based on their own empirical research. Only few approaches compare sets of first-hand data, such as Curnow’s (2001) examination of the categories affected by contact in the contributions to Aikhenvald and Dixon’s (2001) volume on language contact. Still, dealing with restricted data of very different contact situations, he concludes that small hierarchies may indeed exist, but that it is probably impossible to find the constraints on – and factors behind – borrowing. Other studies have shown that small-scale hierarchies indeed can be established, at least for certain categories. In this way, Matras (1998, 2000) shows that there are categoryinternal hierarchies in borrowing (e.g. but > or > and), and a series of hierarchies are illustrated for a sample of Romani dialects by El ík and Matras (2006: 343ff.). It is clear, then, that various factors have to be taken into account when analysing language contact. Classification within strict categories such as ‘area’ or ‘code switching’ is often not possible, nor is it useful in analysing different types of contact situations. A systematic approach, studying firsthand data, could furthermore lead to new generalisations and test existing ones on the mechanisms behind linguistic contact phenomena. 2. Theoretical background, definitions In our systematic study of language contact, we will take into account various different factors. Most of these deal with the outcomes of language contact. To get a more complete picture of why and how these outcomes differ, we have to begin with studying the initial stages of interference through contact. At the onset of language contact, subconscious factors often play a role. Matras (1998, 2000, 2007) has shown that they relate to speakers’ attempt to reduce the load of processing linguistic operations in discourse, eliminating the need to select among two distinct components of a bilingual individual’s structural repertoire. Occasional, on-the-spot switches can at a later stage become fully integrated into the recipient language; that is undergo propagation in Croft’s (2000) terminology. Social factors play an important role in the process of propagation. Thus, in situations where a highly domi-

Modelling contact-induced change in grammar

67

nant colonizer’s language is in contact with an indigenous language we usually find propagation of elements from the colonizer’s language into the indigenous language (cf. also Stolz and Stolz 1996, 1997). In bilingual situations with few speakers, on the other hand, propagation is less likely (cf. examples in Matras and Sakel 2007). The various types of situations and outcomes of language contact can be settled at various stages within the process, between the very onset, such as on-the-spot switches in a learner’s varieties of a language, and propagation, i.e. established loans in a stable contact situation. In this way, we will not adopt a sharp distinction between code switching and borrowing but view the two of them on a continuum (Gardner-Chloros 1995). In the same way, we do not presuppose a clear-cut boundary between linguistic areas and other contact situations, but will analyse both in the same way and test if such a distinction can be justified on the basis of the categories affected. We will match the categories borrowed up to the social context to see if there are any correlations. Furthermore, we will check for all contact phenomena if they are MAT or PAT loans or intermediate stages of these.2 This means that we take a bottom-up approach, i.e. look at the actual language data first and from these draw our conclusions about possible borrowing hierarchies. 3. The procedure In this section we outline the method adopted in the project on Language Convergence and Linguistic Areas. We have developed a database to compare data on contact-induced, structural language change in different contact situations. Here we include the different grammatical categories and sub-categories that can be affected by contact, and distinguish various types of social circumstances. The database3 is designed in the form of a questionnaire for a grammatical description, beginning with general information about the contact situation, followed by information about contact influence in the phonology, typological profile, morphology and different parts of speech, syntactic issues as well as the lexicon. The database consists of ten chapters:4 1. 2. 3. 4. 5.

Info General Phonology Typology Nominal structures

6. 7. 8. 9. 10.

Verbal structures Other parts of speech Constituent order Syntax Lexicon

68 Jeanette Sakel and Yaron Matras We include the different concepts discussed above. In this way, we do not distinguish between areas and other borrowing situations in the database. Rather, we provide the possibility to classify a situation as belonging to an area, a borrowing situation, or both, depending on how it is classified in the literature or what the researcher focusses on.5 We can then later compare this classification with the actual results from the rest of the database and see if there is a match, and in this way test Campbell’s (2006) hypothesis that areas are mere clusters of one-to-one borrowing situations. We can furthermore compare the different outcomes of borrowing situations to the social factors involved and compare the influence of, for example, official status and language use with the categories that are borrowed to see if there are any correlations. Figure 1 shows how we code the sociolinguistic background and Figure 2 shows the encoding of contact languages, both for Mosetén.

Figure 1. Encoding of the sociolinguistic situation (Mosetén).

Modelling contact-induced change in grammar

69

Figure 2. Encoding of contact languages (Mosetén).

In the same way as coding for different social circumstances, we also code for different types of interference. In this way, we ask whether an element is taken over as MAT (matter) or PAT (pattern) from the contact language. This distinction is irrelevant in some cases, such as word order, which will always just involve the pattern. Within the expression of other concepts, such as coordination, however, MAT and PAT are two possible outcomes of borrowing. Within these categories (mostly within the chapters on nominal structures, verbal structures and other parts of speech, as well as in some part of the chapter on syntax), we provide the possibility to mark for both MAT and PAT in the database, cf. Figure 3. This figure also shows that we can deal with structures for which we do not know if they have arisen through contact influence by marking them as ‘unclear’. It is far less straightforward to find out how to code the degree of integration of an element, as well as where on a continuum between borrowing and code switching an element is settled. Adding to this comes the nature of the data, namely that “code switching”, i.e. on-the-spot loans, are much more difficult to quantify than established loans. Rather, most studies investigating these are of a qualitative nature. We also have to consider the different stages of contact, outlined above. Different processes appear at the

70 Jeanette Sakel and Yaron Matras onset compared to the propagation period, and these would have to be captured as well.

Figure 3. Encoding of MAT (matter) and PAT (pattern) in the database:

It is far less straightforward to find out how to code the degree of integration of an element, as well as where on a continuum between borrowing and code switching an element is settled. Adding to this comes the nature of the data, namely that “code switching”, i.e. on-the-spot loans, are much more difficult to quantify than established loans. Rather, most studies investigating these are of a qualitative nature. We also have to consider the different stages of contact, outlined above. Different processes appear at the onset compared to the propagation period, and these would have to be captured as well. Our solution to this problem is a compromise, opting to be as inclusive as possible. We include all loans that are more or less ‘integrated’ into the recipient language, i.e. that appear with a certain regularity and that are used by different speakers. In addition, we can mark instances of on-thespot loans as well. These can be presented as ‘unclear’ cases in the database, followed by an explanation as to why they have been classified this way. This might not always be systematic, and some contributors to the database may classify something as ‘unclear’ which others would include in their list of integrated elements. However, this is a risk worth taking, and the examples and explanations will always provide more information about the structure in question, if this is needed.

Modelling contact-induced change in grammar

4.

71

Case studies: the contact situations of Mosetén/Spanish and Romani/Romanian

The following two case studies on Mosetén/Spanish and Rumanian/Romani are examples of results from contact situations coded in the database. 4.1. The contact situation of Mosetén/Spanish Mosetén belongs to the small, unaffiliated language family Mosetenan, which is spoken in the foothills of the Bolivian Andes and the adjoining lowland areas of the Amazon basin. The current overview is based on Sakel’s fieldwork on Mosetén of Covendo, one of two major dialects of Mosetén spoken by approximately 800 people (cf. also Sakel 2004). Mosetén borrows heavily from Spanish. The contact situation is as follows: The Mosetenes have been in contact with speakers of Spanish for hundreds of years. Contact began when missionaries entered the area in the 17th century and there has been substantial contact after permanent missions were established in the 19th century. During those years, bilingualism was common, though many people were still monolingual in Mosetén. Another wave of Spanish influence was the settlement of migrants, mostly from the Andean highlands, introduced as a strategy by the Bolivian government in the sixties and still an ongoing process. Today, the Mosetenes live in close neighbourhood to speakers of Spanish, who often claim a higher status in society and look down on what they consider to be ‘primitive natives’. This situation has led to an immense change in language use – often accompanied by the denial of being of Mosetén origin, when speaking to foreigners. Today, all Mosetenes speak Spanish. There are no monolinguals in Mosetén, rather, many children exclusively learn Spanish. Spanish is used to communicate with outsiders, and at all official purposes, such as school, church, hospital, and for trade. Mosetén is only used orally – mostly among family members and friends, while all writing is done in Spanish. Only very recently, Mosetén has been established as a written language (Sakel 1999, 2001, 2002; cf. also Sakel 2004: 50–52), but is not generally used, apart from the attempts of few individuals.6 Mosetén is – and has been – also in a number of other contact situations. For example, it is considered to be part of a Mamoré-Guaporé region (Crevels et al. 2006), and it is located in the transition zone between the Andean highlands and the (western) Amazonian region (the latter has been proposed by Payne 1990). The analysis of the structures resulting from this contact has only just started – and many aspects of it are still unclear. The Mosetenes are today also in

72 Jeanette Sakel and Yaron Matras contact with other indigenous languages of Bolivia. They live among speakers of Aymara, Quechua, Trinitario, Yurakare and Tacana, but apart from a few individual speaker’s bilingualism, there is no considerable contact influence from any of these languages. This can be attributed to Spanish as the lingua franca in all types of communication with outsiders. Also the highland settlers, which often belong to Aymara or Quechua speaking communities speak only Spanish with outsiders. In this way, Spanish is the main contact language for the Mosetenes and the source of many loans in the language. There is both substantial on-thespot code switching among speakers of Mosetén, as well as integration of a number of elements into Mosetén. For a more detailed discussion of the contact situation and the affected elements cf. Sakel (2007). Spanish elements integrated into the Mosetén system include coordinating and subordinating conjunctions, discourse markers, temporal and spatial prepositions and markers expressing restrictions (or delimitation markers). All of these elements can function above the level of the proposition, and have gesture-like functions (cf. Matras 1998): They serve to steer the hearer’s processing of propositional content, quite often away from the anticipated course of processing, thus conveying an element of contrast or restriction. Such is the case – generally speaking, and not necessarily in specific connection with Mosetén – with adversative and concessive conjunctions, with phasal adverbs such as ‘already’ and ‘no longer’, with focus particles such as ‘only’, indefinites such as ‘never’, and prepositions such as ‘except’: they all process hearer expectations in respect of a state of affairs, exempting a propositional unit from the set of anticipated units. We shall return to evaluate this tendency below. With reference to the database questionnaire, most contact phenomena in Mosetén are found in the domains of Other Parts of Speech and Syntax, while there are none or only few phenomena in the other chapters. Going through the categories one-by-one, we’ll give chapter numbers in brackets. Chapters (1) and (2) deal with metadata and with general information on the sociolinguistic situation and have been discussed above already. There does not seem to be any Spanish influence on (3.) the phonology of Mosetén, apart from loanwords that have not yet been integrated into the phonology of Mosetén. No Spanish sounds are used in native Mosetén words. In the same way, (4.) the overall typological profile, including categories such as head or dependent marking and alignment, does not seem to have changed through the influence from Spanish. Two constructions in (5.) nominal structures could have arisen through contact with Spanish, both of which involve calquing of the structure (i.e. PAT). The default gender in the

Modelling contact-induced change in grammar

73

system of gender agreement seems to have changed due to Spanish influence. While the default gender for mixed groups used to be feminine, speakers increasingly use the masculine gender in these cases. Along the same lines, plural marking seems to have been adjusted to Spanish patterns. In Mosetén, inanimate objects are usually not marked for plurality but they are increasingly marked as such, as in Spanish. (On the use of Spanish prepositions in Mosetén see below). Contact influence in (6.) verbal structures is limited to the integration of Spanish verbs into the system by native Mosetén verbal markers. There are two markers, only one of which is productive. The productive marker -yi- ‘do’ is used in a similar way in Mosetén as the verbal marker in complex predicates. The unproductive marker -i- is added only to few verbs: viaje-i- ‘to travel’, dewe-i- ‘to owe’, reso-i- ‘to pray’ and fieshta-i- ‘to party’. Most of the Spanish loans in Mosetén are MAT particle-like elements in the chapter on (7.) other parts of speech. Most of these elements borrowed belong to the category particles and discourse markers. In this way, the Spanish coordinating conjunctions i (from Spanish y, adjusted only in the orthography) ‘addition’, o ‘disjunction’ and pero ‘contrast’ can be used to connect clauses and parts of discourse. The disjunction marker o ‘or’ can also combine phrases. All of the conjunctions appear in the same places as in Spanish, i.e. they are borrowed together with their functions and distributions. They replace native marking of disjunction and contrast by clitics at the end of the second clause, as well as juxtaposition of the clauses to express addition. Sometimes the Spanish and the native elements appear together in double constructions (1)7: (1)

tyiñe-tyi’ pero-ki pen’-ki jai’ba-i.8 semi.red-L.M but.E-CO side-CO white-VI.M.S ‘It (the peanut) is semi-red, but one side is white.’ STE:VC

Some minor shifts in the usage of the Spanish markers can be noted in Mosetén as compared to Bolivian Spanish. In this way, o is mostly used in constructions expressing alternatives (as o – o, similar to a minor use pattern in Spanish), while mere disjunction is a minor use pattern (in the sense of Heine and Kuteva 2005) in Mosetén. In the same way, the function of pero seems to be somewhat extended in that it not only expresses contrast, but can also mark a change in topic.9 This extension in function seems to have been motivated by analogy to the native marker -ki, which like pero can be used to express contrast, but which also marks a change in topic in Mosetén.

74 Jeanette Sakel and Yaron Matras In the same way, many subordinating conjunctions are borrowed from Spanish, including the complementizer ki (the vowel is adapted phonologically from Spanish que) and several adverbializers. The latter are the si ‘if’, pajki ‘so that’ and the temporal markers hasta, desde and cuando. Also in these cases, the functions of the markers in Mosetén are slightly different from those in Bolivian Spanish. The marker si is only used in conditional clauses expressing alternatives (2), and thus occupies a special position within conditional clauses in Mosetén. (2)

me’tyi-tyi’ yäe yi-n “si mi rai’s-e-’ jäe’mä DM 1SG say-1SG.O if.E 2SG want-VI-3F.O DM ji’-chhae-yi-ti khäei’-si’ phe-ya-k-dye’ o rai’s-e’ CA-know-VI-RE.M.S RF.S-L.F speak-NO or.E want-VI-3F.O chhi-ban-mi jäe’mä piñ-i-dye’-in jedye’-jedye’ mö’-yä’ cure-VI-NO-P thing-RD 3F-AD know-again-2SG DM wiya’-in kïchï tsä’-ïn Köwë’dö’-wë-ïn. old.man-P go.on.M.S alive-P Covendo-DR-P ‘Thus she said to me “if you want to study our language, or if you want to know about the (native) medicines, there are the old people in Covendo, they are still alive.”’ STE:JH

A problematic case is the marker pajki, which seems to be a phonologically integrated form of the Spanish purpose marker para que ‘so that’. It expresses both reason and purpose in Mosetén – and thus has an extended function compared to Spanish. If this marker indeed is a loan, this extended function could be explained by analogy to native markers that likewise express reason and purpose. In the same way, the functions of the temporal adverbial clause markers ashta (the Standard Spanish orthography is hasta) and kwando (cuando in Standard Spanish) ‘when’ are also extended when appearing in Mosetén. Ashta is much more common than kwando. Both can denote an endpoint of an event (i.e. ‘until) (3), as well as the succession of events (i.e. ‘when, after’) (4): (3)

mi’-khan mi’ bae’-i me’-ki 3M.SG live-VI.M.S so-CO 3M-IN keo’-te-in ashta tyaj-ke-te in. search-VY.3M.O-P until.E find-VK-3M.O-P ‘They searched (for him) where he lived, until they found him.’ WTE:CT

Modelling contact-induced change in grammar

(4)

75

tyashi si-ti Kallawalla-khan-tyi’-in first enter.M.S Callawalla-AD-L.M-P ashta wën-jö-i resya-ya’-in. until.E move-DJ-M.S church-AD-P ‘The Callawallas are the first to enter when they reach the church.’ WTE:CT

In (4) the meaning is ‘when they come to the church’ and not ‘until they come to the church’, i.e. focusing on the succession of events. Desde expresses ‘from, since’ does not seem to be used with an extended function. Many Spanish discourse markers, that is markers of hesitation, tag questions and the like, are used frequently in Mosetén. These are the tag question nowe, from Bolivian Spanish no ves ‘don’t you see’ (5) and the hesitation marker awer, from Spanish aver ‘let’s see’ (6), and osea ‘that means, so’. Other, less common markers are porlomenos ‘at least’, siquiera ‘at least’, pues ‘thus, then, well’, claro ‘sure’, claro pues ‘well, sure’, bueno ‘well, sure’, pues ‘and then’, and eso es ‘that is it!’. (5)

mö’-nä khä Hernan tipi-ti-’ 3F.SG-CO well Hernan measure-VT-F.S mäei’-ya’ jäe’mä Marcelina Duran-tom, nowe? Marcelina Duran-COM right.E first-AD DM ‘And this Hernan, they measured the first time, together with Marcelina Duran, right?’ COE:EC

(6)

Alberto, äjj, awer-nä khä, mi’ jady-i-ti, Alberto EM let’s.see.E-FO well 3M.SG go.and.come.back-VI-DT.M.S mi’-nä khä jäe’mä chhome’ ïtsä-dye-i. also play.game-NO-VI.M.S 3M.SG-FO well DM ‘Alberto, well, let’s see, he came [hereto, performed the action, and went away again], and he was also - uhh - playing games.’ COE:SM

Several Spanish markers of place and temporal deixis are frequently employed in Mosetén as well. These are the prepositions ashta (phonologically adapted from Spanish hasta) ‘until’, desde ‘from, since’, and rarely a ‘to’, en ‘in’, and furthermore the markers nunca ‘never’, siempre ‘always’, kadawes (from Spanish cada vez) ‘every time’ and ai weses (from Spanish a veces). Some of these are parallel to adverbial clause markers. There are furthermore several Spanish markers that express the opposite meaning or delimitation of some kind. These include embesde (from Spanish en vez de) ‘instead’, ni ‘not even’ and sin ‘without’.

76 Jeanette Sakel and Yaron Matras Another construction that could have arisen through PAT interference is the outline of the numeral system, which today is decimal, as in Spanish, but traces in the language point at it originally being a quinary system (cf. Schuller 1917; Sakel 2004: 168). It is much more difficult to analyse the degree to which the (8.) constituent order of Mosetén has been influenced by Spanish, due to the lack of old documents, which could show what the language was like in precolonial times or at least at an early stage of contact. Still, a number of constructions seem to have arisen due to Spanish influence, such as the use of prepositions, both in the form of MAT loans in their constructions, as well as through remodelling of native material to fulfil the purpose. Spanish has also influenced other parts of the (9.) syntax of Mosetén, in particular coordination and subordination strategies, that were borrowed together with their markers that are discussed above: MAT-loans usually appear within their Spanish grammatical construction, i.e. coordinating conjunctions appear between the clauses they combine. Finally, Mosetén has a vast amount of borrowed Spanish (10.) lexicon, in all (lexical) parts of speech. Summing up, the contact situation is that of a one-to-one borrowing scenario in which the two languages have very different status and prestige. Most structural loans are found at the level of discourse organisation, such as coordinating conjunctions, subordinating conjunctions, deictic markers (of time and space), delimitation markers, and discourse markers. Spanish also seems to have had influence on the grammatical pattern of Mosetén, though this is more difficult to establish.

4.2. The contact situation of Kelderash Romani/Romanian Romani is the only Indo-Aryan language spoken in Europe since the Middle Ages. It is assumed that the ancestors of the Roma were so-called commercial nomads – specialising in itinerant trades such as metalwork and other craftsmanship, and entertainment. The name "om is cognate with the Indian caste-name  om, still widespread in India today. Other derivations of the name are found outside India in the self-appellations of the dom ‘Gypsies’ of the Middle East, and of the lom of Armenia and the Caucasus. ‘ProtoRomani’ is assumed to have emerged first in Central India, then to have moved to the Northwest. Consequently, it shows both Central archaisms (e.g. Old Indic consonant clusters, Middle Indic conjugation markers) and Northwestern innovations (such as case markers and the past tense forma-

Modelling contact-induced change in grammar

77

tion) (cf. Matras 2002). ‘Early Romani’ is the name given to the immediate precursor of present-day Romani dialects, and is believed to have been spoken within the Byzantine Empire, in close contact with Greek, from around the 10th or 11th century AD and up until the collapse of Byzantium, in the 14th–15th centuries, at which time immigration of Romani groups to other parts of Europe began, leading ultimately to the formation of numerous isoglosses that separate present-day Romani dialects and dialect groups. One of the rather conservative and coherent divisions within Romani is known as the group of ‘Vlax’ dialects (cf. Boretzky 2003), so named after the selfappellation of some of the groups, which in turn derives from their prolonged stay in the Wallachian regions. Characteristic of Vlax, apart from a diagnostic combination of internal archaisms and innovations, is the heavy Romanian-Romance influence on their structure and vocabulary. One of the well-known Vlax dialects is Kelderash (also Kalderash). As with many of the Romani groups in the Balkans, their name derives from the group’s original trade specialisation (Romanian c!ld!rar ‘kettle-maker’). In various forms, the Kelderash dialect is one of the most widely documented and studied varieties of Romani (cf. Gjerdman and Ljungberg 1963; Boretzky 1994; Matras 1994; Hancock 1995). It belongs to the Northern Vlax dialects, originating in Transylvania, and spoken by a population which in all likelihood was enslaved until the abolition of Gypsy slavery in the Romanian principalities, in the 1860s. Kelderash families began to emigrate from Romania at this point in time, and their communities are found today in urban centres throughout the world. The data considered here, however, is taken from a speaker of Kelderash from southern-central Romania (Piteti).10 Sociolinguistically, Romani is a minority language in contact. All adult speakers of Romani are bilingual (or multilingual), and Romani serves strictly as a language of the extended family and of occasional communication with Roms who are not part of the clan or the community; until very recently, it was strictly an oral language, and not supported by any institutions. Owing to their traditional occupation patterns as a service economy, Gypsies tend to use the majority language in all work-related transactions. The absence of a written or institutional norm in Romani, and the absence of stable communication with other Romani communities, strengthens the dominance of Romanian as the majority language; nonetheless, Romani too has served as a source of lexical vocabulary, primarily for Romanian slang varieties. In phonology (item 3. on the database questionnaire), the principal shared development is the adoption into Romani of the central vowel //, replacing

78 Jeanette Sakel and Yaron Matras in some environments an original Romani /e/: uk l ‘dog’ < ukel (cf. also k ld rarja ‘kettle-makers, Kelderash’). Major typological features (4.) are largely compatible among the two languages; in part, owing to their shared Indo-European origin, and in part owing to the massive adaptation of Early Romani to Byzantine Greek, and a subsequent ‘Balkanization’ of the language (cf. Matras 1994). Contact influence in the domain of nominal structures (5.) includes the replication of the Romanian plural marker with Romanian-derived nouns (sekret-urja ‘secrets’), and its diffusion to both earlier loans (e.g. Greek for-urja ‘towns’) and later ones (Slavic jas-urja ‘hours’), and the incorporation of the analytic comparative/superlative form Romanian mai in adjectives (maj mito ‘better, best’). Prepositions of Romanian origin include frequent use of în loc d ‘instead’. Borrowing in the field of verbal structures (6.) is characterised primarily through the incorporation of several modal verbs (datisar- ‘to be able to’, tr bu- ‘to be necessary’). Romanian loan-verbs are integrated by attaching to the root a loan verb adaptation marker, usually -is-ar-, consisting of a Greek-derived tense/aspect marker -is- and a Romani valency marker -ar- (see discussion in Matras 2002: 128ff.). The bulk of matter-replications appear in the domain of Other Parts of Speech (7.): Alongside phasal adverbs such as ynk ‘still, yet’, dea ‘already’, and i maj ‘no longer’ we find focus particles such as numaj ‘only’ and some time adverbs, such as dmult ‘for a long time’, diminjaca ‘in the morning’, including names of months and days of the week. Among them are also focus particles and connectors: (7)

nii vov i dati-sar-dj-as te putr-el o udar either he NEG can-LOAN-PAST-3SG COMP open-3SG DEF door ‘He couldn’t open the door, either’

(8)

nii vov nii lesk-o phral naj len love destul neither he nor his-M brother is.not them.OBL money enough ‘Neither he nor is brother has enough money’

The negator i (as in i symas ‘I was not’) appears to derive from a generalisation of this focus particle, contaminated by the indefinite negator i ‘nothing’. Note in (9) that Kelderash Romani borrows the Romanian form of the factual complementizer, k (Romanian c!), expressed elsewhere in Romani generally as kaj, but often through a loan:

Modelling contact-induced change in grammar

(9)

79

nas khr dikh-l-em k see-PAST-1SG COMP was.not at.home ‘I saw (that) he was not at home’

Other grammatical function words of Romanian origin include discourse markers and utterance modifiers: (10) hajda te anklj-as avri COMP exit-1P out let’s ‘Let’s go out’ (11) atuni le la po bango vast kaj dujto intersekcja then take it at crooked hand at-DEF second intersection ‘Then take a left turn at the second intersection.’ Also belonging to Other Parts of Speech is the temporal indefinite yntodjauna ‘always’, and the temporal negative indefinite ‘never’, a hybrid expression containing the Romani negative indefinite i, the Romani numeral ek ‘one’, and the Romanian temporal unit expression data ‘time’: (12) i

a-u i ek data kaj pjaca te tin-au man"o go-1SG NEG one time to.DEF market COMP buy-1SG bread ‘I never go to the market to buy bread.’ NEG

The adoption of word forms for indefinite pronouns also includes nite ‘anything’ as well as of the indefinite marker vare- (in combination with Romani interrogatives, e.g. vare-kon ‘someone’, vare-so ‘something’ etc.). In constituent order (8) and the syntax of complex constructions (9), both Romani (generically, that is, Early Romani and the southeastern European dialects of the language) and Romanian share characteristic features of the Balkan linguistic area, such as flexible VS/SV order, finite adverbial and relative clauses introduced by subordinating conjunctions, finite modal complements (instead of infinitives), distinct conjunctions introducing factual and non-factual complements, and more. Finally, a massive amount of lexical material (10) is borrowed from Romanian into Kelderash Romani, including basic nouns such as fejastra ‘window’, marja ‘sea’, vorba ‘word’, lumina ‘light’, animalo ‘animal’, basic verbs such as tril ‘live’, skriisarel ‘write’, jubisarel ‘love’, and basic adjectives such as lungo ‘long’ and skurto ‘short’.

80 Jeanette Sakel and Yaron Matras 5. Conclusions Our goal was to discuss how to model a comparative description of contactrelated change, in order to arrive at generalisations about mechanisms of language contact. We have presented a method which has the following features: 1) Consideration of the sociolinguistic background, and functions of the language in the community 2) Tagging of types of contact: area, one-to-one 3) Review of all possible structures affected by contact, by word class and category; rendering a kind of grammatical description from the contact perspective 4) Tagging of contact influences for different structural types, MAT and PAT 5) Representation of a continuum of spontaneous and established contact influences 6) Employment a database structure for ease and efficiency of comparison The database is currently in the final stages of setup, with invitations to specialists in contact situations to contribute data, it will eventually go online and be accessible to researchers, constituting a first and unique systematic, modifiable cross-linguistic sample of contact influences. We have chosen here to discuss, briefly, two of our sample idioms that are under the influence of Romance languages, Mosetén and Kelderash Romani. Their histories differ considerably: the first is a colonized language, the second a migratory diaspora language. Nonetheless, we find similarities in the domains of use, and in the role of the respective dominant contact language. All speakers are bilingual, the dominant language serves for all contacts outside the community and with institutions, while the group language remains oral, and a symbol of in-group identity. There is therefore motivation to hold on to the community language, and acceptance of bilingualism. The structural profiles of contact are not dissimilar, either (Table 1). Most prominent are loans in the domain which we termed ‘Other parts of speech’, especially discourse markers and connectors, temporal expressions and focus particles – the inventory characterised by Matras (1998) as ‘utterance modifiers’ triggering ‘monitoring-and-directing’ operations in discourse interaction. A hypothesis lends itself, on the basis of just this one comparison, that oral languages under the influence of dominant, institu-

Modelling contact-induced change in grammar

81

tionalised majority idioms will tend to adopt first those uninflected function words that play a major role in structuring the utterance at the level of the discourse interaction. Note the prominence, among the borrowed items in both languages, of discourse markers, temporal indefinites, and complementisers. Mosetén goes further in its adoption of conjunctions, both coordinating and subordinating, presumably due to the need to adjust its clause combining strategy to the Spanish model, while Kelderash Romani enters into contact with Romanian already possessing similar clause combining patterns, and relying on native conjunctions. Nonetheless, the pattern of conjunction borrowing is prevalent in other Romani dialects in contact with other languages (cf. Matras 2002: 201–202), and Kelderash manifests itself in this regard as rather conservative. On the other hand, compared with Mosetén, Kelderash Romani borrows more in the areas of indefinites, phasal adverbs, and focus particles, as well as in the area of modality. Noteworthy are some other striking similarities between the two cases. In prepositions, both borrow the delimitation form ‘instead’; in nominal structures, the representation of plurality is affected; in verbal structures, valency or ‘verb-ness’ markers are used to adapt loan-verbs; and there are no visible contact phenomena in the domain of morphological typology. Equally striking is the absence of borrowing in certain domains, such as personal pronouns, location deixis, and demonstratives. This brings to our attention the supposed proneness of unbound ‘function words’ to borrowing. In fact, in close examination it appears that it is not the structural class as such that attracts borrowing, but only some of its sub-categories, and so borrowing follows a semantic-functional hierarchy, capturing first and foremostly those grammatical devices that process expectations: anticipated links between propositions, time phases, sets, and qualifications of propositions. In an evaluation of the Romani sample, El ík and Matras (2006: 385–386), drawing on Matras (1998), argue that borrowing targets firstly those grammatical devices that are employed to process instances of potential tension between the message conveyed by the speaker, and the hearer’s expectations. Take for instance the preposition ‘instead of’, which exempts an item from a presupposed set, or the contrastive marker ‘but’ or focus particle ‘only’, which convey similar exemptions. In such instances of mental tension surrounding the processing of the utterance, there is an advantage to the bilingual speaker to improve the efficiency of retrieval of the correct expression by eliminating the need to select between sub-components of the linguistic repertoire – the two ‘languages’. The result is ‘fusion’ – the nonseparation of the two linguistic systems around the device in question. It is this which, in diachronic perspective, we call ‘grammatical borrowing’.

82 Jeanette Sakel and Yaron Matras The cross-linguistic study of grammatical (and lexical) borrowing is still in its infancy, despite the vast amount of literature dedicated to the subject. We hope to have outlined a model that will enable to pursue this study in a more systematic and transparent fashion, so that solid cross-linguistic generalisations may be arrived at soon, opening new avenues toward interpreting the role of structure, semantic and pragmatic function, and social constraints in the complex process known as contact-induced change. Table 1. Grammatical contact phenomena in Mosetén and Kelderash Romani Questionnaire chapter

Sub-category

Spanish in Mosetén

Romanian in Kelderash Romani

1. Information

Fieldwork (Sakel)

Fieldwork (Manchester Romani Project; Zatreanu)

2. General

Moseten < Mosetenan (isolate); family and community oral language; only as L1; all speakers bilingual

Kelderash Romani < Vlax Romani < Romani < IndoAryan; family and community oral language; only as L1; all speakers bilingual

3. Phonology

Adoption of central vowel /
/

4. (Morphological) typology 5. Nominal structures

6. Verbal structures

7. Other parts of speech

in loc d
‘instead’

Prepositions: delimitation

embesde ‘instead’, sin ‘without’

Prepositions: temporal

ashta ‘until’, desde ‘from, since’

General:

masculine as default; explicit plurality

plural marker -urja; adjective comparison maj

Verbal structures: loan verb integration

with native verbalising markers

with native valency markers and Greek-derived tense / aspect markers

Verbal structures: other

modal verbs datisar- ‘can’, tr
b- ‘must’

Discourse markers: pues ‘then’ sequential

atuni ‘then’

Discourse markers: awer, nowe, bueno other

hajda

Coordinating conjunctions

i ‘and’, o ‘or’, pero ‘but’

Complementiser

general: ki

Adverbial subordinators

kwando ‘when’, si ‘if’, pajki ‘so that’, ashta ‘until’

Indefinites: temporal

nunca ‘never’, siempre ‘always’

factual: k


i ek data ‘never’, yntodjauna ‘always’

Modelling contact-induced change in grammar

83

Table 1. (cont.) Grammatical contact phenomena in Mosetén and Kelderash Romani Questionnaire chapter

Sub-category

Spanish in Mosetén

Romanian in Kelderash Romani

Indefinites: other

vare-/ or-, nite ‘anything’

Phasal adverbs

ynk
‘still, yet’, i maj ‘no longer’, dea ‘already’

Focus particles

nii … (ni)i ‘(n)either … (n)or’, numaj ‘only’

8. Constituent order

Position of Spanish prepositions

Compatibility of both systems with Balkan ‘type’ (areal convergence)

9. Syntax

Partial adoption of Spanish format in clause combining

Compatibility of both systems with Balkan ‘type’ (areal convergence)

Temporal expressions

kadawes ‘every time’

d
mult ‘for a long time’, diminjaca ‘in the morning’, days of the week, months

Lexicon: other

In most semantic domains

In most semantic domains

10. Lexicon

Notes 1.

2. 3. 4. 5.

6. 7.

MAT loans usually involve both the morphological material of an element and the structure in which it appears (PAT). Only in few cases do MAT loans only involve the actual matter, cf. Stolz and Stolz (1996, 1997), Sakel (2007) on function changes in analogy with native markers. Such intermediate stages include the function change of a native element due to phonological similarity with an element in the contact language. In this case, it is difficult to determine whether the loan in question is MAT or PAT. The database is set up in File Maker Pro. Collaboration with Tadmor and Haspelmath’s ‘loanword typology’ project, looking at the lexical aspects of language contact. We present each contact situation in a separate questionnaire, even though it involves the same language. I.e. Mosetén is both part of a contact situation with Spanish, as well as possibly being part of a proposed Amazonian area (Payne 1990) or Mamoré-Guaporé area (Crevels et al. 2006), having been in contact with other Foothill and Amazonian languages for a long time. Both of these situations are coded in separate questionnaires for Mosetén. An exception is the work by my main consultant, Juan Huasna Bozo, who is teaching Mosetén at the newly established ‘Universidad de Monseñor Jorge Manrique en Palos Blancos’ in the Mosetén region. AD = adessive; CA = causative; CO = contrastive marker; COE = conversation ex.; COM = comitative; COMP = complementizer; CT = Cleto Tahe; DJ = assoc.

84 Jeanette Sakel and Yaron Matras motion -kho-; DEF = definite article; DM = discourse marker; DR = downriver relation, DT = assoc. motion -ti-; E = Spanish loan; EC = Esteban Condo; EM = emphatic marker; F = feminine; FO = focus; IN = inessive; JH = Juan Huasna; L = linker; LOAN = loan verb marker; M = masculine; NEG = negation marker; NO = nominalisation; O = object; OBL = oblique; P = plural; PAST = perfective tense; RD = reduplication; RE = reflexive, reciprocal; RF = logophoric reference; S = subject; SG = singular; SM = Sacarias Misange; STE = spoken text example; VC = Vitoriano Chairique; VI = verbal stem –i-; VK = verbal stem -ki-; VT = verbal stem -ti-; VY = verbal stem -yi-; WTE = written text example; 8. The clitic -ki appears on the first element of the second clause to marker contrastive coordination. Since the conjunction pero is borrowed and added between the clauses, the speaker struggles to identify the first element of the second clause, which can be seen in the marker ki appearing twice. 9. This might also be a feature inherent to Spanish, cf. Sakel (2007). 10. Data collected by Mihaela Ztreanu for the Manchester Romani Project, 2004. For project details and fieldwork methodology see http://www.llc.manchester. ac.uk/Research/Projects/romani/

References Aikhenvald, Alexandra Y. 2002 Language contact in Amazonia. Oxford: Oxford University Press. Aikhenvald, Alexandra Y. and R.M.W. Dixon (eds.) 2001 Areal diffusion and genetic inheritance. Oxford: Oxford University Press. Backus, Ad 2005 Codeswitching and language change: one thing leads to another? International Journal of Bilingualism 9 (3/4): 307–340. Bakker, Peter 2006 The Sri Lanka Sprachbund: the newcomers Portuguese and Malay’. In Linguistic areas: Convergence in historical and typological perspective, Yaron Matras, April McMahon and Nigel Vincent (eds.), 135–159. Basingstoke: Palgrave-Macmillan. Boretzky, Norbert 1994 Romani. Grammatik des Kaldera-Dialekts mit Texten und Glossar. Wiesbaden: Harrassowitz. 2003 Die Vlach-Dialekte des Romani. Strukturen – Sprachgeschichte – Verwandtschaftsverhältnisse – Dialektkarten. Wiesbaden: Harrassowitz. Campbell, Lyle 2006 Areal linguistics: a closer scrutiny. In Linguistic areas: convergence in historical and typological perspective, Yaron Matras, April McMahon and Nigel Vincent (eds), 1–31. Basingstoke: Palgrave-Macmillan.

Modelling contact-induced change in grammar

85

Crevels, Mily, Pieter Muysken and Hein van der Voort 2006 The Mamoré-Guaporé region: a linguistic area?’, paper presented at the workshop on ‘Current issues in areal typology’ at the meeting of the DGfS (Deutsche Gesellschaft für Sprachwissenschaft) in Bielefeld, Germany, 24.2.2006. Croft, William 2000 Explaining language change. An evolutionary approach. Harlow, England: Longman Linguistics Library. Curnow, Timothy J. 2001 What language features can be ‘borrowed’? In Areal diffusion and genetic inheritance, Alexandra Y. Aikhenvald, and R.M.W. Dixon (eds.), 412–436. Oxford: Oxford University Press. Elík, Viktor, and Yaron Matras 2006 Markedness and language change: The Romani sample. Berlin /New York: Mouton de Gruyter. Field, Fredric 2005 Long-term effects of CS: Clues to structural borrowing. International Journal of Bilingualism 9 (3/4): 341–360. Gardner-Chloros, Penelope 1995 Code-switching in community, regional and national repertoires: the myth of the discreteness of linguistic systems. In One speaker, two languages, Lesley Milroy and Pieter Muysken (eds.), 68–111. Cambridge: Cambridge University Press. Gjerdman, Olof and Erik Ljungberg 1963 The language of the Swedish Coppersmith Gipsy Johan Dimitri Taikon. Uppsala: Lundequist. Go b, Zbigniew 1956 The concept of isogrammatism. Buletin Polskiego Towarzystwa Jezykoznawczego 15: 1–12. 1959 Some Arumanian-Macedonian isogrammatisms and the social background of their development. Word 15 (3): 415–435. Hancock, Ian 1995 A handbook of Vlax Romani. Columbus: Slavica. Haugen, Einar 1950 The analysis of linguistic borrowing. Language 26 (2): 210–231. Heath, Jeffrey 1984 Language contact and language change. Annual Review of Anthropology 13: 367-384. Heine, Bernd and Tania Kuteva 2005 Language Contact and grammatical change. Cambridge: Cambridge University Press. Johanson, Lars 1992 Strukturelle Faktoren in türkischen Sprachkontakten. Stuttgart: Franz Steiner Verlag.

86 Jeanette Sakel and Yaron Matras Lass, Roger 1997 Historical linguistics and language change. Cambridge: Cambridge University Press. Matras, Yaron 1994 Untersuchungen zu Grammatik und Diskurs des Romani, Dialekt der Kelderaa / Lovara. Wiesbaden: Harrassowitz. 1998 Utterance modifiers and universals of grammatical borrowing. Linguistics 36 (2): 281–331. 2000 Fusion and the cognitive basis for bilingual discourse markers. International Journal of Bilingualism 4 (4): 505–528. 2002 Romani: a linguistic introduction. Cambridge: Cambridge University Press. 2007 Contact, connectivity and language evolution. In Connectivity in grammar and discourse, Jochen Rehbein, Christiane Hohenstein and Lukas Pietsch (eds.), 51–74. Amsterdam: Benjamins. Matras, Yaron and Jeanette Sakel 2007 Investigating the mechanisms of pattern replication in language convergence. Studies in Language 31 (4): 829–865. Moravcsik, Edith 1978 Language contact. In Universals of human language, Joseph Greenberg, Charles A. Ferguson and Edith Moravcsik (eds.), 93–122. Stanford: Stanford University Press. Myers-Scotton, Carol 1993 Duelling languages. Oxford: Oxford University Press. 2002 Contact linguistics: Bilingual encounters and grammatical outcomes. Oxford: Oxford University Press. Payne, Doris L. 1990 Morphological characteristics of Lowland South American languages. In Amazonian Linguistics – studies in Lowland South American languages, Doris L. Payne (ed.), 213–242. Austin: University of Texas Press. Poplack, Shana 1980 Sometimes I’ll start a sentence in Spanish Y TERMINO EN ESPAÑOL: Toward a typology of codeswitching. Linguistics 18: 47–104. Ross, Malcolm D. 1988 Proto Oceanic and the Austronesian languages of Western Melanesia. Canberra: Pacific Linguistics. Rozencvejg, Viktor Jul’evic 1976 Linguistic interference and convergent change. The Hague: Mouton. Sakel, Jeanette (ed.) 1999 Poromasi’ Pheyak’dye’in [old stories]. Bolivia: Proyecto GRAMO, 2nd revised edition 2002. 2001 Ojtere’ [the rooster]. Bolivia: Proyecto GRAMO. 2002 Tsinsi’ kirjka [our book]. Bolivia: Proyecto GRAMO.

Modelling contact-induced change in grammar

87

Sakel, Jeanette 2004 A grammar of Mosetén. (Mouton Grammar Library.) Berlin /New York: Mouton de Gruyter. 2007 Language contact between Spanish and Mosetén – a study of grammatical integration. International Journal of Bilingualism 11(1): 26– 53. Sakel, Jeanette and Yaron Matras 2006 Prototypical linguistic areas and borrowing hierarchies, toward an integral investigation of linguistic areas. Paper presented at the workshop on ‘Current issues in areal typology’ at the meeting of the DGfS (Deutsche Gesellschaft für Sprachwissenschaft) in Bielefeld, Germany, 22.2.2006. Schuller, Rudolph 1917 Introduction to Benigno Bibolotti’s Moseteno Vocabulary and Treaties from an unpublished manuscript in possession of Northwesterns University Library. Evaston and Chicago: Northwestern University. Stolz, Christel and Thomas Stolz 1996 Funktionswortentlehnung in Mesoamerika, Spanisch-Amerindischer Sprachkontakt. Sprachtypologie und Universalienforschung 49(1): 86–123. 1997 Universelle Hispanismen? Von Manila über Lima bis Mexiko und zurück: Muster bei der Entlehnung spanischer Funktionswörter in die indigenen Sprachen Amerikas und Austronesiens. Orbis 39 (1) [1996– 1997]: 1–77. Thomason, Sarah G. 2001 Language contact: an introduction. Edinburgh: Edinburgh University Press. Thomason, Sarah G. and Terence Kaufman 1988 Language contact, creolization and genetic linguistics. Berkeley: University of California Press. Weinreich, Uriel 1953 Languages in contact: Findings and problems. The Hague: Mouton. [Reprint 1966.]

Loan verbs in a typological perspective Søren Wichmann and Jan Wohlgemuth

1. Introduction In this paper we provide an overview of structural patterns involved in the transfer of a verb from one language to another. The primary aim is to establish a descriptive framework for such loan verbs. As we shall see, some languages borrow verbs by simply inserting a root-like form of the verb into their own morphologies or otherwise treat the loan verb as a native item. Commonly, however, some special derivation process or a light verb like ‘to do’ is required to accommodate the loan verb. In addition, in some rare cases a language may borrow entire inflectional paradigms along with the verb. This paper will present a classification of these major strategies and subtypes thereof. An obvious question that arises is whether it is possible to predict which strategy speakers of a given type of language might use when borrowing verbs. Do features of the source and/or target language determine the structural pattern associated with borrowed verbs? Our answer to this question will be highly tentative. We do think that structural features, especially of the target language, are relevant for the outcome of a given borrowing event, but evidence such as the existence of more than one borrowing pattern in one and the same language suggests that the structural outcome cannot be directly predicted from structures of the languages involved. There are clearly additional factors involved, such as degrees of bilingualism or areal tendencies. Even if it is not directly possible to predict which strategy a language will use, we hypothesize that if a change in the strategy occurs because of increased language contact the change will move in a specific direction along a loan verb integration hierarchy, which we set up in this paper. 2. Previous research Few studies have been devoted to the topic of borrowed verbs. Moravcsik (1975) is an early, pioneering work which is still a useful point of departure for discussion. The paper, however, makes the overgeneralized claim that “[a] lexical item whose meaning is verbal can never be included in the set

90 Søren Wichmann and Jan Wohlgemuth of borrowed properties” (Moravcsik 1978: 111). From the context of the paper as a whole this statement may be interpreted as saying that verbs can never be borrowed as verbs (it does not say that verbs cannot be borrowed at all, as, for instance, Campbell 1993 has interpreted it to mean). Thus, if a verb is transferred from one language to another it will in the first instance be borrowed and treated as a noun, and will require some sort of verbalization in order to be treated as a verb in the target language. One major problem with this generalization is that it requires Moravcsik to posit zero derivational morphemes to save her generalization in cases where there is no overt derivational mechanism present. Moreover, she admits that she does not have a good explanation for why the treatment of borrowed verbs as nouns should be universal in the first place (Moravcsik p.c. 2003). In spite of these problems, Moravcsik’s proposal is not uninteresting. The treatment of loan verbs as nonverbs is, indeed, quite common, and does call for an explanation. In the course of this paper (Section 6) we shall return to this issue and propose an explanation. Since 1975, the literature on language interference and code switching has been rapidly increasing, and several studies provide examples of loan verbs. They are all restricted to individual language cases, however. Pugh (1999) stands out as a broader study focused on structural patterns involving loan verbs, but it is still limited to a particular group of languages, namely Finnic. Similarly outstanding, Mifsud’s (1995) in-depth study is focused on loan verbs borrowed into Maltese. The only major contribution since Moravcsik (1975) is Ch. 7 of Muysken (2000), a rich source of data and discussion. Muysken identifies many of the patterns that we illustrate in this paper, but classifies them somewhat differently. Where we distinguish four major types – the light verb strategy, indirect insertion, direct insertion, and paradigm transfer – Muysken (2000) divides the first type into three subtypes, essentially collapses our next two types, and ignores the fourth. In the course of our paper we shall return to differences between Muysken’s and our approaches. Our own research builds on presentations by Wichmann (2004a, 2004b) at various workshops and Wohlgemuth’s work towards a dissertation on the topic of loan verbs, which has so far resulted in a database design and a still growing collection of data (cf. Wohlgemuth 2005a, 2005b). Both authors have carried out their work in relation to the Loanword Typology Project of the Max Planck Institute for Evolutionary Anthropology. As of May 2007, we have collected a sample with data from over 250 languages in 375 donorrecipient combinations. There are data from all over the world and a wide range of different language families are represented.1

Loan verbs in a typological perspective

3.

91

Terminological issues

3.1. Verb We do not operate with any strict definition of ‘verb’, since such an entity can be difficult to define unequivocally. Even if generally inadequate (Croft 2000: 65), a notional definition according to which verbs denote actions or events will suffice for the present purposes. We count as loan verbs all items that function as verbs in the source language, even if they are treated as nouns (and are subsequently verbalized) in the borrowing language (a common phenomenon). In contrast, we exclude cases where the item functions as a noun in the source language, such as Finnish jobbata ‘to work’, where the infinitive verbal ending -ata has been added to the English noun job (Nau 1995: 72), or Japanese sekkusu suru ‘have sex’, literally meaning ‘to do sex’ (Schmidt 2005). In some cases it is impossible to know whether the borrowed form is to be regarded as a verb or a noun in the source language for reasons of homophony or lability of part-of-speech membership, e.g. French faxer (from Vendelin and Peperkamp 2006: 10), which conceivably could just as well represent a borrowing of the English noun fax (machine) as of the English verb (to) fax. In such cases we choose to be inclusive, and simply treat the form on a par with other loan verbs. 3.2. Light verb The way we use the term ‘light verb’ is also consciously vague. We employ the term for verbs like ‘do’ or ‘make’ or verbs of a similarly broad referential scope, which are used in complex constructions where they have an auxiliary-like function. Our use corresponds to that of the original coiner of the term, Jespersen (1954 VI, 117–118), rather than to generativist literature, into which it was introduced by Grimshaw and Mester (1988). Crosslinguistically, a common form of light verb is ‘to do’, and it has been observed that one of the major functions of ‘do’-periphrasis cross-linguistically is precisely to integrate loan verbs (Jäger 2004; cf. also van der Auwera 1999 for additional background on the typology of ‘do’-periphrasis). Since verbs other than ‘to do’ may be involved in loan verb integrating constructions we use the broad term ‘light verb strategy’ for such cases rather than the otherwise simple and catchy term ‘the do-strategy.’

92 Søren Wichmann and Jan Wohlgemuth 3.3. Borrowing A major problem in defining borrowing is how to distinguish true loans from ‘nonce borrowings’, i.e. words that are introduced into the target language in an ad hoc fashion (Poplack et al. 1988). Good evidence for a true borrowing would be that the item is replacing or has replaced an earlier, synonymous word or that it denotes some kind of object or action which was once new to the culture but which has now become an integral part of it. These may be called ‘cultural borrowings’ (Myers-Scotton 1993). Recorded language history showing the presence of the word over several generations is of course also good evidence. Phonological modification may be indicative of loanword status if there is a contrast with phonologically unmodified nonce borrowings (Heath 1989: 23–25). Other criteria sometimes invoked are the occurrence of a foreign lexeme in the speech of monolinguals, its frequency, or the perception of the speakers themselves regarding its status in the language (Schatz 1989: 132). In many cases, however, none of these criteria can be applied. It is often difficult to apply criteria involving semantics since translation equivalents may often have differences in connotations (Backus 1996: 115–131). The situation becomes particularly difficult in the case of verbs. Since the semantics of verbs is usually more general than that of nouns, it is difficult to establish whether some native verb in the target language is or is not synonymous with the putative loan verb. For similar reasons it is difficult to establish on cultural grounds that a true borrowing must have taken place. And as regards recorded language history, this is often not available. Phonological modification is a criterion that only applies occasionally as explained in the previous paragraph. And, finally, we often do not have information concerning the occurrence of the word in the speech of monolinguals, its frequency within a corpus, or speakers’ perceptions regarding its status. Thus, we admit to often having to make educated guesses as regards what is a true borrowing and what is a nonce borrowing. 4. Observed patterns In the following we review the four major patterns of loan verb integration that we have found: the light verb strategy, indirect insertion, direct insertion and paradigm transfer.

Loan verbs in a typological perspective

93

4.1. The light verb strategy The light verb strategy most often involves a verb meaning ‘to do’ for the integration of loan verbs – so often, in fact, that one would be tempted to simply call it the ‘do-strategy’, were it not for the fact that a few languages employ other verbs, as we show further on in this section. We begin by exemplifying the more common strategy, picking three random examples from our sample. (1)

Manange [Tibeto-Burman] < Nepali [I.E.] (Kristine Hildebrandt, p.c., 2004) 1 hai 1la-pa yawn do-NOM2 ‘to yawn’ [cf. Nep. haii aau-nu ‘yawn come-INF’]

(2)

Texistepec Popoluca (Texistepequeño) [Mixe-Zoquean] < Spanish (Wichmann 1996: 79) I njunu nwyat pensar ñyaka’ap? I njunu ny-wat pensar ny-yaka’-p and how 2.ERG-do think 2.ERG-kill-FUT ‘And how do you intend to kill him?’ [Sp. pensar]

Turkish verbal borrowings from French are nominalized by means of sourcelanguage devices before they enter into the target language’s ‘do’-construction. (3)

Turkish < French (Lewis 1967: 154, via Dan Slobin, p.c.) isole etmek isolated do/make ‘to isolate, insulate’ [Fr. isolé]

Other examples where a phrasal ‘do’-verb is involved in loan verb accommodation may be drawn from the following languages and sources, some of which are cited in Muysken (2000): – American Portuguese [Romance] < English (Pap 1949:114–117) – Amharic [Semitic] < English (Pete Unseth, p.c.) – Anatolian Arabic < Turkish [Turkic], Kurdish [Indo-Iranian] (Vocke and Waldner 1982: XLIV, 215) – Armenian (Eastern) < Russian [East Slavic] (Kozintseva 2003: 222) – Awa Pit [Barbacoan] < English (Curnow 1997: 156)

94 Søren Wichmann and Jan Wohlgemuth – – – – – – – – – – – – – – – – – – –

Bangla [Indo-Aryan] < English (Moravcsik 1975: 14) Basque < French, Spanish [Romance] (Haase 1992: 87, Trask 1997) Greek < U.S. English [West Germanic] (Moravcsik 2003) Hausa [Chadic] < English (Madaki 1983) Hindi [Indo-Aryan] < English (Kachru 1978) Hungarian [Finno-Ugric]< English (Moravcsik 1975: 14) Kaqchikel [Mayan] < Spanish (Stenson 1998: 224) Lebanese Arabic < French (Abou 1962: 65) Malayalam, Kannada, Telugu [Dravidian] < English (Moravcsik 1975: 14) Moroccan Arabic [Semitic] < Dutch (Boumans 1998: 223–269) Navaho [Athapaskan] < English (Canfield 1980: 219) Panjabi [Indo-Aryan] < English (Romaine 1985) Pipil [Uto-Aztecan] < Spanish (Campbell 1985: 144) Popoloca [Otomanguean] < Spanish (Veerman-Leichsenring 1991: 160, 289, 290, 441, 479) Sarnami (Surinam Hindustani) [Indo-Aryan] < Sranan [English-based creole], Dutch [West Germanic], English (Kishna 1979) Tamil [Dravidian] < English (Sankoff et al. 1986, Annamalai 1978) Turkish [Turkic] < Dutch (Backus 1992: 77) Urdu [Indo-Aryan] (Moravcsik 1975: 14) Yakut [Turkic] < Russian (Brigitte Pakendorf, p.c.)

In some languages the light verb forms a compound with the borrowed verb. We do not consider this treatment significantly different from the phrasal construction but rather something that would follow from the way that the target language generally behaves with regard to complex verbs. Below we provide a couple of examples. Korean sometimes uses borrowed verbs to express particular nuances of meaning or to render a message less comprehensible to uninvited listeners. The following example is of a French loan verb, but similar examples of borrowing from English also exist. (4)

Korean [Isolate] < French (Thekla Wiebusch p.c. via Soyoung Roger-Yun, p.c.) Mary-ka John-eul detester.hae-yo Mary-NOM John-ACC hate.do-DECL ‘Mary hates John.’

In Pech, Spanish loan verbs are accommodated by uniting the loan verb, which assumes the shape of the infinitive minus final r, with the Pech verb

Loan verbs in a typological perspective

95

stem í-k- ‘do’. A clash of vowels is resolved by normal morphophonological processes. (5)

Pech (Paya) [Chibchan] < Spanish (Holt 1999: 62) resé-kre a-í -kpray-do ‘to pray’ [Sp. rezar]

In a few languages verbs other than ‘do’, but of a similar degree of semantic generality, are observed in the light verb borrowing strategy. These are used either in addition to, or instead of, ‘do’. The following are cases in point; other examples involve English borrowings into Japanese (Hinds 1986: 28) or Georgian and Avar borrowings into Bezhta (Comrie 2005). (6)

Carib [Carib] < Guianese French Creole (Renault-Lescure 2004: ex. 19) man pentiré poko paint busy.with 3SG.COPULAR ‘He is painting.’ [GFC pentiré]

(7)

Itelmen [Chukotko-Kamchatkan] < Russian (Georg and Volodin 1999: 57) werit e0es believe be ‘to believe’ [Rus. =>@?AB]

Sometimes a motivation for using a light verb other than ‘do’ relates to transitivity. Thus, in the Azoyú variety of Tlapanec, Spanish loan verbs are accommodated with either ‘do’ or ‘make’, the latter being used to express the causative of the Spanish verb. The ‘do’-construction is also used for certain Tlapanec expressions of a verbalizing nature and the ‘make’-construction is also used for creating periphrastic causatives of native verbs, though in the latter case a subordinating particle di2 is involved. (8)

Tlapanec [Otomanguean] < Spanish (Wichmann, field notes 2003) a. nu-ni kompayã IPFV-do.3PL accompany ‘They are accompanying.’ [Sp. acompañar]

96 Søren Wichmann and Jan Wohlgemuth b. mu-
y.=lu
 eskapa FUT.PL.AGENTIVE-make.1PL.INCL>3SG.AN =1PL.INCL escape a
gu ci tu
wahma yu
3 girl who tied.up.3.SG there ‘We’ll make the girl who is tied up there escape.’ [Sp. escapar] Peculiar to languages of northern Australia is a type of complex predicate where, even in the native vocabulary, an open class of uninflecting ‘coverbs’ (sometimes called ‘preverbs’) combines with a closed class of inflecting verbs. Given the presence of such a construction, loan verbs may readily be inserted into the coverb slot, as in the following example: (9)

Gurindji [Pama-Nyungan] < Jaminjung [Jaminjungan] (McConvell 2005: 3) tibart wani-nya jump fall-PAST ‘(S)he jumped.’

A similar example, involving loan verbs from the English-based creole Kriol into Jaminjung, is cited in Schultze-Berndt (2003: 151). Related to this is the kind of verbal compounding seen in Warlpiri borrowings from English, as illustrated in (10a–b). Warlpiri appears to have undergone a change whereby erstwhile constituents in the coverb construction have lost their syntactic and prosodic independence (Nash 1982). The choice of native compounded verb is guided by transitivity, -jarrimi ‘INCHOATIVE’ occurring with intransitives and -mani ‘get/take/affect’ with transitives. (10) Warlpiri [Pama-Nyungan] < English (Bavin and Shopen 1985: 82) a. sliipi-jarrimi sleep-INCH ‘to sleep.’ b. jasi-mani chase-affect ‘to chase.’ Future studies of loan verbs in Australia could provide important new insights into loan verb typology given the structural peculiarities of the languages here as well as the great amount of linguistic diffusion on this continent (e.g., Dixon 2002: 24–30).

Loan verbs in a typological perspective

97

4.2. Indirect insertion In many languages an affix is required to accommodate loan verbs. Once the affix is added the normal inflectional patterns may be applied. Although based on limited data, it seems to be a valid generalization that whenever the affix has a function in addition to that of accommodating loan verbs, this function relates to the flagging of part-of-speech membership. Sometimes the affix is a verbalizer, sometimes a nominalizer, and sometimes it marks a particular class of verb. Often, however, the affix has no other function than that of accommodating loan verbs. Nevertheless, in those cases where we have been able to identify the etymologies of such ‘loan verb affixes’ (henceforth glossed ‘LV’), they have been found to borrowed along with loan verbs from other languages where their functions also relate to the flagging of part-of-speech membership. We return to these cases in section 5 below. In the following we simply illustrate the range of this pattern, which we call the indirect insertion pattern. Meyah is among the languages that employ an LV affix of unknown origin. Apparently the prefix ebe- is needed to accommodate loan verbs not beginning with a vowel, since Meyah verbs normally must be vowel-initial. (11) Meyah [East Bird’s Head] < Indonesian [Austronesian] (Gravelle 2002: 149) di-ebe-belajar 1SG-LV-learn ‘I am learning.’ [Ind. belajar ‘to learn’] Another case is Manange, which employs a suffix -ti to accommodate some loan verbs. As we saw in example (1) above, in other cases a light verb strategy is used. The reason why one or the other strategy is used is not clear (in section 6 below we provide more examples of languages employing more than one strategy). (12) Manange [Tibeto-Burman] < Nepali [Indo-Aryan] (Kristine Hildebrandt, p.c., from field notes) ro bolai-ti 1mi call-LV EVID REP ‘He called (for the frog)’ [Nep. bolai] Another Tibeto-Burman language employing a LV affix of unknown origin is Belhare. Here the affix is -ap (Balthasar Bickel, p.c.).

98 Søren Wichmann and Jan Wohlgemuth Nahuatl has been in intensive contact with Spanish for half a millennium and freely borrows words from all parts of speech. The following example is from the Pastores variant of Eastern Guerrero. Nahuatl verbs may be divided up into different verb classes. One of them is characterized by ending in -oa. Synchronically, this suffix does not have any particular function. All Spanish loan verbs are treated as members of the class of verbs in -oa. As is normal with these verbs, the a of -oa becomes suppressed under further suffixation. (13) Nahuatl (Pastores) [Uto-Aztecan] < Spanish (Wichmann, field notes 1992) se de ihwante ya ki-puntaro-tika-ya now 3OBJ-point-PROG-IPFV one of them ‘Now he had one of them at gunpoint.’ [Sp. apuntar] As a case where the accommodating affix does have a grammatical function in the target language, we may cite Shipibo-Konibo. This language is in intensive contact with Quechua and Spanish, and uses the same verbalizing suffix -n to accommodate verbs from both of these languages, as illustrated in (14). (14) Shipibo-Konibo [Pano-Ucayalina] < Quechua [Quechuan] and Spanish (Valenzuela 2003) justamente la educación r-iki no-n yoi-ti precisely F.ART education EVID-COP 1.PL-ERG say-INF.ABS atipa-n-ke la… el único camino que that can.Q-n-CMPL F.ART M.ART only road no-a cambia-n-ti 1PL-ABS change:3-n-INF ‘Precisely education is, we could say, the … only road that can change us…’ [Q. atipa ‘to be able to’, Sp. cambiar ‘to change’)] As evidenced by data cited by Valenzuela (2004), other Pano-Ucayalina languages such as Wariapano and Kapanawa follow the Shipibo-Konibo pattern of using the -n suffix for the integration of Spanish loan verbs. Many European languages use LV affixes, and, as was the case with PanoUcayalina, the patterns may be shared among closely related languages. Thus, Germanic languages like German, Danish, and Dutch use a cognate suffix to accommodate French and Latin loan verbs (cf. section 5 below). The Finnic languages Ingrian and Votic both require the suffix -tt for Rus-

Loan verbs in a typological perspective

99

sian loan verbs (Pugh 1999: 120). Finally, various Slavic languages like Russian, American Czech, and American Polish affix a verbalizing suffix -ova to borrowed English verbs (Moravcsik 1975: 15). Other cases of languages that use indirect insertion are: Welsh, which borrows many English verbs by adding a deverbalizer -io (Orin Gensler, p.c.); Modern Greek, which uses -%;-, derived from the Italian infinitive -are, to accommodate French loans (Mackridge 1987: 315); the Turkic language Yakut, which uses a verbalizer (either -LA: and very rarely -Iy) for Russian loan verbs (Brigitte Pakendorf, p.c.); and the Chibchan language Rama, which employs the verbalizer -ting for Spanish loan verbs (Grinevald n.d.: 174). 4.3. Direct insertion By ‘direct insertion’ we refer to a process whereby the loan verb is plugged directly into the grammar of the target language with no morphological or syntactic accommodation. We include such cases in this category even if a phonological modification has taken place. The borrowed form may be rootlike, infinitive-like, imperative-like, inflected for third person or nominalized by means of devices in the source language, and probably others. Sometimes it is difficult to establish which of these various source language forms of the loan verb is taken over. In any case, all of the source language instantiations seem to represent a sort of citation form in the analysis of the borrowing speakers. In the following we provide examples of each of these subtypes of direct insertion. Direct insertion of a root-like stem occurs in borrowings by various Germanic (e.g., German and Danish) or Romance languages (e.g. Spanish) from English. It also occurs among languages that belong to many other families, however. The following example is from the northwestern Amazon region, where Tukano has become an increasingly important lingua franca. The target language illustrated is Hup, which is spoken by a minority of predominantly hunter-gatherers. True borrowings into Hup may be distinguished from nonce borrowings by the criterion relating to phonological modification. Borrowed verbs are truncated to one syllable in conformity with Hup phonotactic structure (15a), whereas the appearance of Tucano verbs in Hup speech that results from nonce code-switching is identified by the absence of truncation (15b).

100 Søren Wichmann and Jan Wohlgemuth (15) Hup [Maku] < Tukano [Tukanoan] (Patience Epps, p.c.) a. am-Fn ãh yu-té-h 2SG-ABS 1SG wait-FUT-DECL ‘I’ll wait for you’. [Tuk. yuu ‘to wait’] (Patience Epps, p.c.) b. an pihi-tæn, wetam
-tæD -æD y ãh-ãh ãh n
-
h 1SG.OBJ call-COND help-CNTRFCT-IPFV 1SG-DECL 1SG say-DECL ‘If they (Tukanos) call me, I should help, I say’. [Tuk. piha ‘to help’; wetam
also from Tukano, but exact source form remains to be identified] French loan verbs in the Figuig Berber language of Eastern Morocco provide another example of direct insertion of a root-like stem (16 a–b). According to Maarten Kossmann (p.c.), Moroccan Arabic loan verbs are treated like the French ones. (16) Figuig Berber [Berber] < French (Maarten Kossmann, p.c.) a. i-gõfla 3SG.M-be.swollen/PFV ‘he is swollen up.’ [Fr. gonfler ‘to swell’] b. tt-gõfli-x IPFV-be.swollen-1SG ‘I am swelling up.’ The direct insertion of an infinitive-like stem is found with Spanish borrowings into the Purépecha language of Michoacán, Mexico. From various examples in Chamoreau (2000) it appears that Spanish verbs, slightly modified phonologically by adding an apparently epenthetic i to the infinitive, are plugged directly into the verbal morphology. (17) Purépecha (Tarascan) [Isolate] < Spanish (Chamereau 2000: 142) 'iki ¢'iPku-i-ka-Ø 'Xi pe'Nsari-a-ka 1 think-PROG-ASSERT.1/2 that dead-COP-SUBJ-3 ‘Me, I thought he was dead.’ [Sp. pensar] Other cases of direct insertion of root- or infinitive-like stems may be drawn from the following contact situations: – Bislama [English-based creole] < Samoan [Austronesian] (Crowley 1990: 138), Melanesian Pidgin (Crowley 1990: 110), French (Crowley 1990: 118, 128), English (Crowley 1990: 128), Tok Pisin (Crowley 1990: 134)

Loan verbs in a typological perspective

– – – – – – – – – – – – – – – – – – – – –

101

Bolivian Quechua [Quechuan] < Spanish (Muysken 2000: 63) 16th-century Quechua [Quechuan] < Spanish (Lockhart 1998: 43) Carib [Carib] < Sranan and Dutch (Renault-Lescure 2004) Coptic [Afro-Asiatic, Egyptian] < Greek (Lambdin 1983) Estonian [Finnic] < German (Neetar 1990: 356), Low German (Neetar 1990: 356), Russian (Neetar 1990: 355–356) Evenki [Tungus] < Yakut [Turkic] (Malchukov 2003: 242) Fijian [Austronesian] < English (Schütz 1978: 6, 38, 44, 241; 1985: 142) Finnish < English (Nau 1995: 72) French < English (Vendelin and Peperkamp 2006: 10) Gawwada [Afro-Asiatic, Cushitic] < Amharic [Semitic] (Tosco 2005) Jakarta Indonesian [Austronesian] < English (Chaer 1976: 43, 95), Dutch (Chaer 1976: 235) Jalonke [Niger-Congo, Mande] < Fula (Guinean) [Niger-Congo, Atlantic] (Lüpke 2005) Ket [Yeniseian] < Russian (Minaeva 2003: 48; Werner 2002) Korean [isolate] < English (Kang 2003: 254) Lama [Niger-Congo] < French (Ulrich 1997: 458) Lower Sorbian [Slavic] < German (Bartels 2005) Miskito [Misumalpan] < English (Hale 1994: 270) Mosetén [isolate] < Spanish (Sakel 2005) Nenets [Uralic] < Russian (Malchukov 2003: 239) Pilagá [Mataco-Guaicurú] < English (Vidal 2001: 117) Tok Pisin [English-based creole] < English (Smith 2002: 94, 97, 99, 104, 112, 207)

According to Maarten Kossmann (p.c.), who provided us with the examples in (18), the Northern Songhay language Tasawaq (Niger) takes over Tuareg verbs with the Tuareg 3SG.M person prefix and subsequently treats them as normal verb stems. Sometimes, though, the 3SG.M prefix is absent, possibly for phonological or phonotactic reasons which we have not been able to determine. In (18) the Tuareg loan ëlmëq ‘to swim’ is borrowed in the 3M perfective Form, i-lmäq ‘he swam’. (18) Tasawaq [Nilo-Saharan, Northern Songhay] < Tuareg [Afro-Asiatic, Berber] a. gháy yílmàq 1SG swim.PERF ‘I swam’

102 Søren Wichmann and Jan Wohlgemuth b. ghá bb-ílmàq 1SG IPFV-swim ‘I am swimming’ c. ghá `mm-ílmàq 1SG SUBJ-swim ‘would that I swim’ Tuareg has nouns with ‘verby’ semantics like tusrak ‘sneezing’ or tusut ‘coughing’. Since most Tasawaq verbs can also function as nouns these formally nominal forms can also be treated as verb stems (19). (19) Tasawaq < Tuareg (Maarten Kossmann, p.c.) ghá b-tásrìg 1SG IPFV-sneeze ‘I am sneezing’ [Tua. tusrak ‘sneezing’] This last example is related to the type where the directly inserted item is a form nominalized by means of devices in the source language. An example of this is provided by the following Michif sentence (cited in Muysken 2000: 208). While verbs in the mixed French-Cree language Michif normally come from the Algonquian language Cree, some are from French and English. These take the prefix lï, which is from the French article le, and a suffix -i: from the French infinitive -er. (20) Michif < English (Bakker 1997b: 115–116) lï kat dï yjt gi:-lï-slibre:t-i-na:n DET four of July 1PA-DET-celebrate-INF-IA.1PL ‘We celebrated the Fourth of July’. 4.4. Paradigm transfer In some rare cases the loan verb is not adapted to the recipient language’s morphology at all but is borrowed along with significant parts of the donor language’s verbal morphology which maintains its function. We call this paradigm transfer. This is different from borrowed verbs with ‘fossilized’ donor language morphology like those in (18) and (19) above, where an inflected form is borrowed but where the donor language inflection has no function in the recipient language at all; in cases like (18) or (19), all inflection or derivation applied to the borrowed verb uses morphology native to the recipient language.

Loan verbs in a typological perspective

103

A restricted case of derivational morphology borrowed along with the verb is shown in (21). (21) Mingrelian < Georgian (Lela Zamu ia, p.c., 2005) a-mvid-en-s TV-soothe-PRS-3SG ‘(s)he soothes’ [Geo. a-mvid-eb-s ] In a handful of its verbal borrowings from Georgian, Mingrelian maintains the thematic vowel (TV) a- of Georgian – as opposed to o- which would be normal in Mingrelian. No further morphology was borrowed along with these verbs (Lela Zamu ia, p.c., 2005). In cases of what we designate here as actual paradigm transfer, however, the donor morphology maintains its function within the recipient language to a much greater extent. For instance, the person agreement affixes on borrowed verbs may be those of the donor language, as in the following example where -sun, the Turkish 2nd person marker, occurs with the borrowed verbs: (22) Romani (Agia Varvara) < Turkish (Bakker 2005: 9) jazarsun and o sxoljo ka siklos te okursun ta te in ART school FUT learn.2 COMP read.2SG and COMP write.2SG ‘at school you will learn how to read and write’ [Tur. okumak ‘to read’, yazmak ‘to write’] Turkish loan verbs in Agia Varvara Romani are inflected with their original Turkish suffixes in present and past tense; only the first person plural pasttense suffix deviates from the Turkish paradigm due to analogical leveling (Igla 1996: 214–216). A similar pattern is reported by Newton (1964) for Kormatiki (also called Araviká), an Arabic dialect of Cyprus, which is heavily influenced by Cypriot Greek; “C[ypriot Greek] verbs in K[ormatiki] are conjugated exactly as they are when they occur in C[ypriot Greek]” (Newton 1964: 47). It may at times prove difficult to distinguish such forms of loan verb integration from occasional word-level code-switching. In the case of Agia Varvara Romani, however, code-switching can be ruled out. The inflected non-native words are always and exclusively verbs and there are no similar “switches” involving other word classes (e.g. nouns with case-making), as one would expect with code-switches of fluent bilingual speakers. Furthermore, these unadapted borrowings occur freely in Romani sentences, not just in fossilized idiomatic expressions where they might be considered phrase-

104 Søren Wichmann and Jan Wohlgemuth level switches (cf. Igla 1989, 1996). It may be added that present-day speakers are no longer bilingual in Romani and Turkish, because the speakers of this variety moved from Turkey to Greece several generations ago. Paradigm transfer only occurs in intensive contact situations, involving bilingualism or the extended contact of a mixed language (like Kormatiki) with one of its sources (here: Greek). 4.5. A brief note on Romance languages In a contribution to a volume concerned specifically with Romance languages it makes sense to pause to briefly look at how these behave. We will restrict ourselves to a statistical overview, however. This overview also provides an opportunity to provide some information on the overall statistics of the four different patterns discussed above. In our database there are 24 instantiations of different types of patterns from 21 language pairs involving a Romance recipient language (Romanian, Catalan, French, Italian, Provençal, Portuguese of Portugal, Brazil, and USA). A few of these recipient languages, then, use more than one type of pattern. We have 80 instantiations of patterns from 73 pairs involving a Romance donor-language (Latin, Spanish, French, Italian, Portuguese of Portugal and Brazil). Romance-to-Romance borrowings are represented by two examples only. The distribution of patterns across Romance recipient languages deviates from the global distribution inasmuch as in the overwhelming majority of the Romance cases (20 out of 24 instantiations or 83.3%) the direct insertion strategy is used. Compare this to the global sample, where this strategy is found for about 52% of the instantiations. Indirect insertion and the light verb construction are much less frequent. They are each found twice (8.3% each), which is less than the global distribution of 24% for indirect insertion and 22.9% for the light verb strategy. No cases of paradigm insertion have been found. Globally, this strategy accounts for less than one percent of all instantiations of different patterns. Languages borrowing from Romance, on the other hand, show a more representative distribution of patterns. The light verb strategy accounts for the bulk of cases (49 out of 80, or 61.25%), but indirect insertion (21.25%) and the light verb strategy (17.5%) are much closer to the values from the global sample.

Loan verbs in a typological perspective

105

5. The borrowing of borrowing patterns Bakker (1997a) may be the first study to show that a language may borrow borrowing patterns from another language. The language dealt with is Romani, whose various dialects in fact use various inflectional suffixes from Greek to accommodate borrowings from other languages. These suffixes, which vary somewhat from dialect to dialect, do not mark particular grammatical categories in Romani, but only serve to signal that the lexeme has been borrowed. Of particular interest are the formal patterns of verb borrowing, summarized by Bakker (1997a: 12–13) as follows: Borrowed verbs are only integrated with a loan marker between the borrowed verb and the inflection. These markers find their source in aorist markers. Anatolian Greek dialects use the Turkish aorist/preterit marker -das does the Sepeides Romani dialect. Other Anatolian Greek dialects use the -iz- element which is derived from the Greek sigmatic aorist (Boretzky and Igla 1991: 35). This element is also used in several Romani dialects, sometimes followed by the Romani element -ar-, notably in Vlax dialects. Other Romani dialects use -in-, which is not reported from borrowings into Greek, but which is a common ending in Greek, and of increasing frequency since classical times.

While the author considers the borrowing of a borrowing mechanism “highly unusual if not unique” (Bakker 1997a: 18), it does occur elsewhere. To demonstrate this, we will give a couple of additional examples. One example is provided by Romance verbal borrowings in Nordic languages, especially from Latin and French, which take suffixes that descend from Old Nordic -era. This, in turn, seems to be based on the Middle Low German loan verb adaptation suffix -êren (Simensen 2002: 955). The -era pattern must have been adopted from Middle Low German along with some verbs borrowed from their Romance sources via this language, such as fallera ‘deceive, mislead’, formera ‘to form, shape’, and spazera ‘to walk’, but it was extended to other loan verbs even those taken over directly from Latin, such as disputera ‘to dispute’, komponera ‘to compose’, and traktera ‘to treat, entertain’. Interestingly, an alternative pattern existed, as evidenced by shorter forms in -a, e.g., disputa and kompona (Simensen 2002: 955). Nevertheless, it is the -era pattern which won out, and which was applied to the full range of Latin and Old French loan verbs. Another example comes from the Uto-Aztecan language Yaqui, which borrows Spanish verbs using the verb class marker -oa from Nahuatl, another Uto-Aztecan language. This suffix functions as a loan-verb-accommo-

106 Søren Wichmann and Jan Wohlgemuth dating element in Nahuatl, as was exemplified in (13) above. The pattern is productive in Yaqui, and seems to be applied both to older borrowings and nonce borrowings. Some examples of individual verbs (from Zarina Estrada, p.c.) are: wantaroa ‘to hold’ (Sp. aguantar), kombilaroa ‘to mix’ (Sp. combinar), piaroa ‘to lend’ (Sp. fiar), leiaroa ‘to read’ (Sp. leer), passaroa ‘to pass’ (Sp. pasar), and pensaroa ‘to think’ (Sp. pensar). The following sentence example is from an oral narrative reproduced in Silva Encinas (2004, ex. 91; communicated to us by Zarina Estrada). (23) Yaqui [Uto-Aztecan] < Spanish che’a chúkula into te retratár-oa-wa-k CONJ 1PL portray-oa-PASS-PERF more later ‘And later we were photographed.’ [Sp. retratar] More intensive research would surely reveal additional examples of the borrowing of a borrowing strategy. This phenomenon appears to be common enough that one hypothesis to suggest itself when one encounters a loanverb-accommodating affix without a known origin is that the affix may well derive from another language where it has a similar function.

6. Languages with more than one strategy At least seven of the over 80 languages in the sample show more than one strategy for borrowing verbs. This ratio will probably turn out to be higher when more data become available on a broader spectrum of language combinations. Generally, one has to distinguish two different occurrences of languages with more than one borrowing strategy: a) those where a language has changed strategies in the course of time but did not have more than one productive pattern in use at any one time, and b) those where a language makes use of different productive patterns at the same time. Finnish would be an example for the first scenario. While its general situation of language contact has not changed substantially over the last few centuries, it has employed different patterns at different periods for borrowed verbs of the same origin. The Nordic -era suffix mentioned above also made its way into Finnish where it was joined by the verbal suffix -ta to form the complex -eerata, as in frankeerata ‘to stamp, affix postage’ (Nau 1995: 65). This suffix is apparently not fully productive anymore and has mostly been replaced by -oida, as in maximoida ‘to maximize’ (Nau 1995: 65). The most recent borrowings, however, seem to be accommo-

Loan verbs in a typological perspective

107

dated by direct insertion of a root-like stem, yielding infinitives like e.g. chätätä ‘to chat’ (Hennariikka Kairanneva, p.c.; Florian Siegl, p.c.). When a language shows different borrowing patterns at the same time, this calls for an explanation. While it is tempting to speculate that the structural outcome of a verb-borrowing event may be predicted from structural properties of the target language, perhaps in combination with properties of the source language, such a hypothesis turns out to be problematical. Different strategies for different donor languages is exemplified by English loan verbs in other Germanic languages such as German, Dutch or Danish, which are directly inserted, as opposed to Romance loan verbs, which are indirectly inserted, requiring cognates of the Middle Low German suffix -êren, discussed in section 5 above. Thus, properties of the target language are not sufficient to make predictions. Moreover, there are also examples where one and the same language borrows verbs from one and the same source language using different strategies for different individual verbs. Thus, the Finnic language Karelian uses the Finnic suffix - for some Russian verbs but not for others (Pugh 1999:120). So here both indirect and direct insertion is found. Another case is that of Nepali loan verbs in Manange, where some involve a ‘do’-construction, cf. (1) above, while others take a suffix -ti, cf. (12). Finally, Anthony Grant (p.c.) informs us that both in verb borrowings from Farsi to Urdu and from Hebrew to Yiddish the light verb strategy is common, but that direct insertion exists as well. A similar situation is found with English loan verbs in Spanish and Greek. Here even one and the same verb may be treated by these two different strategies, e.g., English to click (with a computer-mouse) can be borrowed as Spanish hacer clic as well as clicar (the Internet offers many examples of both, and also demonstrates a similar competition in Catalan and Portuguese), and in Greek we similarly find both 86:< 8978 (kano klik) and 89786;< (klikaro). Such examples show that structural properties are not sufficient to make predictions, not even when both the source and the target languages are taken into account. One might attempt to rescue the hypothesis by arguing that among the different patterns there could be a major one and a minor one(s), and that at least the major one might be explained structurally, whereas the minor one(s) would be due to some obscure peculiarities. This, however, is belied by cases where a language only has one major strategy for verb borrowing and where this strategy is different from the one expected from the morphosyntactic resources of the language. For instance, in the Moroccan Arabic spoken in the Netherlands we find that Dutch verbs are accommodated by means of a ‘do’-construction which corresponds neither to Moroccan Arabic

108 Søren Wichmann and Jan Wohlgemuth nor to Dutch patterns (Maarten Kossmann, p.c., in reference to Boumans 1998). A mirror-image of this occurs in Welsh, which freely allows finite verbs to be paraphrased with a construction involving the corresponding verbal noun and (following it) the light ‘do’-verb, but where indirect insertion nevertheless appears to be the preferred pattern of loan verb integration (Orin Gensler, p.c.). 7.

Generalizations

7.1. The loan verb integration hierarchy The kind of evidence offered in the preceding section shows that the choice in a given language of one of the four major loan-verb-accommodation patterns cannot be predicted absolutely from structural properties of the languages involved. We cannot exclude that there are tendencies in this direction, but our material is not yet extensive enough to make sound statistical judgments. In any case, it is necessary to look for additional explanations of the patterns observed in individual languages. One such explanation might be areal tendencies. For instance, the light verb strategy is a very widespread way of treating English loan verbs in languages of India. Again, more material would allow us to make somewhat more precise statements regarding this. On the other hand, it is also quite likely that it will not be possible to make any firmer observations than the tendency of borrowing patterns to show areal distributions. It is unlikely that any particular pattern is concentrated in one part of the world to the exclusion of others, and it is likely that some areas are more conducive to uniformity of borrowing patterns than others. A final explanation, necessitated by the existence of different strategies in one and the same language, relates to degrees of bilingualism. Although much more research has to be done correlating different structural patterns with different sociolinguistic settings associated with the particular borrowing events, we would like to venture the hypothesis that if a language has different patterns, these could correlate with the degrees to which speakers of the target language are exposed to the source language(s). We propose, as an idea to be tested in future research, the following hierarchy: (24) Loan Verb Integration Hierarchy light verb strategy < indirect insertion < direct insertion < paradigm transfer

Loan verbs in a typological perspective

109

The degree to which a loan verb is integrated into the target language may be considered inversely proportional to the amount of formal mechanics expended by the target language on accommodating the loan verb (ignoring, for the present purposes, phonological aspects). From this point of view, the lowest degree of integration is associated with the light verb strategy, which involves a whole extra constituent for the integration. A somewhat higher degree of integration is associated with indirect insertion, where just an affix is required. In the case of direct insertion we have complete integration: here the loan verb is treated as if it were native. It is less straightforward to place paradigm transfer in the hierarchy since, one the one hand, no formal accommodation effort has been expended while, on the other hand, the loan verb is in a sense unintegrated inasmuch as it retains the inflectional morphology of the source language and resembles a code-switch. This accounts for our use of the composite symbol “ < ”, indicating that paradigm transfer might be considered part of the hierarchy, but that it has a special status. Interpreted in the sense just mentioned, as a simple descriptive device for degree of morphological integration, the hierarchy is hardly problematical. As mentioned, however, we would also like to suggest that it could be used to make predictions regarding borrowing behavior. Although we would not in general venture to predict which formal strategy a given language would use for integrating loan verbs, we would like to suggest that if a language already has a strategy and changes this or adds another one, then the new strategy’s placement in the hierarchy relative to the earlier strategy would be determined by the relative degree of bilingualism in the source language or languages. Thus, more bilingualism would mean the choice of a strategy further to the right in the hierarchy and less bilingualism a strategy further to the left. We would predict, for instance, that if language X borrows verbs from language Y by means of the light verb strategy or indirect insertion, but uses direct insertion from language Z, then this would mean that there is more bilingualism in Z than in Y on the part of the mother-tongue speakers of X. In section 5 above we have cited a few cases where one language uses different strategies to integrate loan verbs from different source languages or where different verbs from one and the same source may be treated differently. To resume, the contact situations were the following: – – – –

German, Dutch, Danish < English vs. Romance Karelian < some Russian verbs vs. others Manange < some Nepali verbs vs. others Modern Greek < some French and English verbs vs. other English verbs or even one and the same English verb

110 Søren Wichmann and Jan Wohlgemuth – Spanish < some English verbs vs. others or even one and the same English verb – Urdu < some Farsi verbs vs. others – Yiddish < some Hebrew verbs vs. others At present we do not have detailed studies available of any of these cases. We would, for instance, predict that the Manange loan verbs from Nepali which take -ti date to a period with more bilingualism in Nepali than the loan verbs involving the light verb construction. Predictions based on similar lines of reasoning would apply to the other cases. At present this is of course pure speculation, but we have at least provided a testable hypothesis, which should encourage more detailed research into the outcomes of different contact situations. Boumans (1998) has also reflected on a possible correlation between the structural outcome of verb borrowing and the kind of contact situation involved. He contrasts Moroccan Arabic/Dutch code-switching in the Netherlands with Arabic/French code-switching in Morocco. In the former situation Dutch verbs are integrated by means of a ‘do’-construction, whereas in the latter French verbs appear to be directly inserted – they are “inflected by means of attaching Arabic prefixes and suffixes to the French verb stem” (Boumans 1998: 369). Boumans suggests that the occurrence of a ‘do’-construction in code-switching is “characteristic for migrant bilingualism in modern industrialized societies” (Boumans 1998: 369). We find this suggestion unnecessarily bold and far too specific but are sympathetic to Boumans’ more general hypothesis that sudden and intense contact may lead to relatively unintegrated borrowings of the type represented by the light verb strategy. We stress again, however, that a testable hypothesis requires the existence of alternative patterns in one and the same target language such that structural factors can be excluded and sociolinguistic settings more directly compared. Only in this way can a hypothesis regarding a correlation between structural and sociolinguistic factors be stated in a testable way. 7.2. Moravcsik’s proposals The frequent cases of direct insertion (see the large list in section 4 above) run counter to Moravcsik’s (1975) proposed universal according to which it is impossible to borrow a verb as a verb (cf. our discussion of her position in section 1 above). She argues that languages which use direct insertion nevertheless comply with the generalization since they have morphologies

Loan verbs in a typological perspective

111

that allow a noun root to be treated as a verb. However, since there is no positive evidence that the borrowed roots in question are in fact being treated as nouns, the argument is not strong. In general, it would seem that there must be a clear denominalization procedure involved before one can truly argue that verbs are borrowed as nouns. Admittedly such procedures are attested in a large number of cases. Thus, Moravcsik’s generalization often applies, even if it cannot count as a universal. Why is it, then, that verbs are often treated as nouns when transferred to another language? The explanation that we propose is that in the transfer process verbs may become alienated from the morphosyntactic contexts that define their partof-speech membership and that they thus ‘arrive’ in the target language underspecified for this feature. With regard to the sociolinguistic implications of the loan verb integration hierarchy we previously suggested that direct insertion implies a relatively high degree of bilingualism. This hypothesis also feeds into this discussion of Moravcsik’s generalization. If the treatment of verbs as non-verbs shows these verbs to be underspecified for partof-speech membership, then the treatment of verbs as verbs, i.e. when direct insertion occurs, would conversely show those verbs to have retained their specification for part-of-speech membership. This could not happen without a good command of the larger structures of the donor language(s) on the part of at least some of the borrowers. In sum, we have three reactions to Moravcsik’s proposal. First, we rephrase the generalization that verbs cannot be borrowed as verbs to a generalization that verbs often get borrowed as non-verbs – not necessarily as nouns, but simply underspecified for partof-speech membership. Secondly, we would state this as a widespread tendency rather than an absolute universal. Thirdly, we would argue that the explanation is ultimately of a sociolinguistic nature, hypothesizing that the cases where Moravcsik’s generalization (in its modified form) holds imply a somewhat lower degree of bilingualism than the cases where it does not hold. 8. Conclusions and questions for future research The purpose of this paper was to present a suitable descriptive classification of the patterns of borrowing of verbs found in the world’s languages and to extract generalizations, including possible predictions, from the patterns found. The data investigated largely represent a convenience sample based on the literature on language contact, personal communication, and descriptions

112 Søren Wichmann and Jan Wohlgemuth of individual languages. More data could be drawn from (especially) the last-mentioned type of source, but the evidence at hand seems sufficient to substantiate our claim that cross-linguistically there are four major strategies for borrowing verbs: what we have called the light verb strategy, indirect insertion, direct insertion, and paradigm transfer. These four strategies were defined and exemplified in section 4 above. In some cases not only verbs, but also the strategy for borrowing verbs may be transferred from one language to another. That is, affixes which are used in a particular language to accommodate loan verbs, following the indirect insertion strategy, may be borrowed by another language where they continue to be used to accommodate new loan verbs. Three examples of this ‘borrowing of a borrowing pattern’, first identified in Bakker (1997a), were provided in section 5. The recurrence of the phenomenon provides a guide for the philologist. That is, when one encounters a loan-verb-accommodating affix which cannot be shown to originate in the borrowing language itself (e.g., via a process of grammaticalization), it is possible that it will turn out to have been borrowed from another language where its function is or was similar. Another important observation was the not uncommon existence of more than one borrowing pattern in one and the same language. This is found in around 10% of the languages of our sample (cf. section 6). From our investigation we made the generalization that the early proposal of Moravcsik (1975) according to which it is impossible to borrow a verb as a verb should be rejected. Examples where this happens are, in fact, numerous. Nevertheless, cases that support Moravcsik’s approach are also numerous, and we propose to explain these as cases where the verb is treated in the recipient language as underspecified for part-of-speech class membership as a consequence of its alienation from the morphosyntactic context of the donor language (see section 7.2). Another generalization made in this paper relates to our proposed loan verb integration hierarchy, cf. (24) above. We see this primarily as a means to explain the existence in one and the same language of more than one borrowing strategy. The hierarchy is based on the hypothesis that the degree to which a loan verb is integrated into the target language may be considered inversely proportional to the amount of formal mechanics expended by the target language in accommodating the loan verb. The degree of integration, in turn, may relate to the degree of bilingualism in the donor language. Thus formal differences in the way that one and the same target language borrows verbs may relate to different degrees of intensity of contact with the donor language(s).

Loan verbs in a typological perspective

113

Our study has revealed that loan verb typology is a rich and rewarding area of investigation. The present study is anything but exhaustive. In future research the database should be expanded. At present it is quite hard to come by examples of loan verbs in descriptive grammars, but we hope that this situation will slowly change as awareness of the interest of studying loan verbs increases. Most of the presently available published data involve major European donor languages, and we need more data involving other donor languages. As mentioned in section 4.1, Australia is an example of an area which needs better investigation. Not only does the database need to be broadened, we also need more detailed, in-depth studies of how individual languages have borrowed verbs over time. A major result of the present paper is that if we are to make any predictions whatsoever as regards preferred verb borrowing strategies, we need to look at cases where one language has more than one strategy, and it is crucial that such studies be both diachronically and sociolinguistically oriented. We need to know when a given borrowing occurred and under what social circumstances. The data we have presented here are far from sufficient to sustain the proposed loan verb integration hierarchy. Currently the hierarchy is to be regarded simply as a hypothesis to be tested, and we hope that other scholars will join us in this research.

Notes 1.

2.

We would like to thank the following colleagues for generous sharing of data or comments and discussion: Esben Alfort, Peter Bakker, Balthasar Bickel, Claire Bowern, Thomas Chacko, Mercè Coll-Alfonso, Caroline Cordero-d’Aubuisson, Bernard Comrie, Patience Epps, Mary Espinosa, Fred Field, Orin Gensler, Anthony Grant, Martin Haspelmath, Roland Hemmauer, Kristine Hildebrandt, Robert Hoberman, Gerd Jendraschek, Olesya Khanina, Madzhid Khalilov, Zaira Khalilova, Maarten Kossmann, Yaron Matras, Edith Moravcsik, Brigitte Pakendorf, John Philips, William Poser, Jeanette Sakel, Kim Schulte, Keith Slater, Dan Slobin, Nancy Stenson, Pete Unseth, Edward Vajda, Bernhard Wälchli, Thekla Wiebusch, Lela Zamu ia, and Ghil’ad Zuckermann. Many of the people to whom our thanks go responded to a query posted by Wichmann on LINGUIST List, Vol. 15-1674, May 29, 2004. ABS = absolutive; ACC = accusative; AN = animate; ART = article; ASSERT = assertive; CMPL = completive; CNTRFCT = counterfactive; COMP = complementizer; COND = conditional; CONJ = conjunction; COP = copula; DECL = declarative; DET = determiner; ERG = ergative; EVID = evidential; F = feminine; FUT = future; IA = intransitive action; INCH = inchoative; INCL = inclusive; INF = infini-

114 Søren Wichmann and Jan Wohlgemuth tive; IPFV = imperfective; LV = loan verb affixes; M = masculine; NOM = nominative; OBJ = object; PASS = passive; PAST = past tense; PERF = perfective; PFV = perfective; PL = plural; PROG = progressive; PRS = present; Q = question marker; REP = repetitive; SG = singular; SUBJ = subjunctive; TV = thematic vowel.

References Abou, Selim 1962 Le bilinguisme arabe-français au Liban. Paris: Presses Universitaires de France. Annamalai, E. 1978 The anglicised Indian languages: a case of code mixing. International Journal of Dravidian Linguistics 7 (2): 239–247. Backus, Ad 1992 Patterns of language mixing. A study in Turkish-Dutch bilingualism. Wiesbaden: Harrassowitz. 1996 Two in one. Bilingual speech of Turkish immigrants in the Netherlands. (Studies in Multilingualism 1.) Tilburg: Tilburg University Press. Bakker, Peter 1997a Athematic morphology in Romani: the borrowing of a borrowing pattern. In The typology and dialectology of Romani, Yaron Matras, Peter Bakker and Hristo Kyuchukov (eds.), 1–21. Amsterdam: Benjamins. 1997b A language of our own: the genesis of Michif, the mixed Cree-French language of the Canadian Métis. New York: Oxford University Press. 2005 Intertwining and Michif. Paper presented at the Romanisation Worldwide Conference; Bremen, 5–9 May 2005. Bartels, Hauke 2005 Loan words in Lower Sorbian – Literary language vs. dialects. Paper presented at the Loanword Typology Workshop, Max Planck Institute for Evolutionary Anthropology, Leipzig, 6–7 June 2005. Bavin, Edith and Tim Shopen 1985 Warlpiri and English: Languages in contact. In Australia, Meeting Place of Languages, Michael Clyne (ed.), 81–94. (Pacific Linguistics, C-92.) Canberra: Australian National University. Boretzky, Norbert and Birgit Igla 1991 Morphologische Entlehnung in den Romani-Dialekten. (Arbeitspapiere des Projektes ‘Prinzipien des Sprachwandels’ 4.) Essen: Fachbereich Sprach- und Literaturwissenschaften an der Universität Essen. Boumans, Louis 1998 The syntax of codeswitching: analysing Moroccan Arabic/Dutch conversation. Tilburg: Tilburg University Press.

Loan verbs in a typological perspective

115

Campbell, Lyle 1985 The Pipil language of El Salvador. Berlin /New York: Mouton de Gruyter. 1993 On proposed universals of grammatical borrowing. In Historical Linguistics 1989. Papers from the 9th International Conference on Historical Linguistics, Henk Aertsen and Robert Jeffers (eds.), 91–109. Amsterdam/Philadelphia: John Benjamins. Canfield, Kip 1980 Navajo-English code-mixing. Anthropological Linguistics 22: 218– 220. Chaer, Abdul 1976 Kamus Dialek Melayu Jakarta – Bahasa Indonesia. Jakarta: Nusa Indah. Chamereau, Claudine 2000 Grammaire du purépecha parlé sur des îles du lac de Patzcuaro. Munich: LINCOM EUROPA. Comrie, Bernard 2005 Loan words in Bezhta: Some preliminary thoughts. Paper presented at the Workshop on Grammatical and Lexical Borrowing, Manchester, 29 September – 1 October 2005. Croft, William 2000 Parts of speech as language universals and as language-particular categories. In Approaches to the typology of word classes, Petra M. Vogel and Bernard Comrie (eds.), 65–102. Berlin/New York: Mouton de Gruyter. Crowley, Terry 1990 Beach-la-Mar to Bislama. The emergence of a national language in Vanuatu. Oxford: Clarendon Press. Curnow, Timothy Jowan 1997 A grammar of Awa Pit (Cuaiquer): an indigenous language of SouthWestern Colombia. Ph.D. dissertation, Australian National University. Dixon, R. M.W. 2002 Australian languages. Their nature and development. Cambridge: Cambridge University Press. Georg, Stefan and Alexander P. Volodin 1999 Die itelmenische Sprache. Grammatik und Texte. (Tunguso Sibirica 5.) Wiesbaden: Harrassowitz. Gravelle, Gilles 2002 Morphosyntactic properties of Meyah word classes. In Languages of the Eastern Bird’s Head, Ger P. Reesink (ed.), 109–180. (Pacific Linguistics 524.) Canberra: Australian National University. Grimshaw, Jane and Armin Mester 1988 Light verbs and -marking. Linguistic Inquiry 19: 205–232.

116 Søren Wichmann and Jan Wohlgemuth Grinevald, Colette G. n.d. A grammar of Rama. Université de Lyon. Report to National Science Foundation, BNS 8511156. Manuscript in possession of the authors. Haase, Martin 1992 Sprachkontakt und Sprachwandel im Baskenland. Die Einflüsse des Gaskognischen und Französischen auf das Baskische. Hamburg: Buske. Hale, Ken 1994 Preliminary observations on lexical and semantic primitives in the Misumalpan languages of Nicaragua. In Semantic and lexical universals: theory and empirical findings, Cliff Goddard and Anna Wierzbicka (eds.), 263–283. (Studies in Language Companion Series 25.) Amsterdam /Philadelphia: Benjamins. Heath, Jeffrey 1989 From code-switching to borrowing. A case study of Moroccan Arabic. London /New York: Kegan Paul International. Hinds, John 1986 Japanese. Croom Helm Descriptive Grammars. London: Croom Helm. Holt, Dennis 1999 Pech (Paya). Munich: LINCOM EUROPA. Igla, Birgit 1989 Kontaktinduzierte Sprachwandelphänomene im Romani von Ajia Varvara (Athen). In Vielfalt der Kontakte. Beiträge zum 5. Essener Kolloquium über “Grammatikalisierung: Natürlichkeit und Systemökonomie” von der Universität Essen, Vol. 1, Norbert Boretzky, Werner Enninger and Thomas Stolz (eds.), 67–80. (Bochum-Essener Beiträge zur Sprachwandelforschung 7.) Bochum: Brockmeyer. 1996 Das Romani von Ajia Varvara. Deskriptive und historisch-vergleichende Darstellung eines Zigeunerdialekts. Wiesbaden: Harrassowitz. Jäger, Andreas 2004 The cross-linguistic function of obligatory ‘do’-periphrasis. Paper presented at the 2004 Conference of the Australian Linguistic Society. . Jespersen, Otto 1954 A modern English grammar on historical principles. London: George Allen & Unwin, and Copenhagen: Ejnar Munksgaard. Kachru, B. 1978 Toward structuring of code-mixing: an Indian perspective. International Journal of the Sociology of Language 16: 28–46. Kang, Yoonjung 2003 Perceptual similarity in loanword adaptation: English postvocalic word-final stops in Korean. Phonology 20: 219–273.

Loan verbs in a typological perspective

117

Kishna, Sita 1979 Lexicale interferentie in het Sarnami. M.A. thesis, University of Amsterdam. Kozintseva, Natalia 2003 Armenian and Russian: Grammatical contacts. In Studies in Eurolinguistics, Vol. 1: Convergence and Divergence of European Languages, Sture Ureland (ed.), 219–234. Berlin: Logos. Lambdin, Thomas O. 1983 Introduction to Sahidic Coptic. Macon: Mercer University Press. Lewis, Geoffrey L. 1967 Turkish grammar. Oxford: Clarendon Press. Lockhart, James 1998 Three experiences of culture contact: Nahua, Maya, and Quechua. In Native traditions in the Postconquest World, Elizabeth Hill Boone and Tom Cummins (eds.), 31–53. Washington D.C.: Dumbarton Oaks. Lüpke, Friederike 2005 Contact between Jalonke (Mande) and Pular (Atlantic). Paper presented at the Workshop on grammatical and lexical borrowing; Manchester, 29 September – 1 October 2005. Mackridge, Peter 1987 The modern Greek language. Oxford: Clarendon Press. Madaki, Rufa’i Omar 1983 A linguistic and pragmatic analysis of Hausa-English code-switching. Ph.D. dissertation, University of Michigan. Malchukov, Andrej 2003 Russian interference in Tungusic languages in an areal-typological perspective. In Studies in Eurolinguistics, Vol. 1: Convergence and Divergence of European Languages, Sture Ureland (ed.), 235–251. Berlin: Logos. McConvell, Patrick 2005 Meaning change and stratigraphy in Gurindji loanwords: plugging the loanword typology project into regional contexts and databases. Paper presented at the Loanword Typology Workshop, Max Planck Institute for Evolutionary Anthropology, Leipzig, 6–7 June 2005. Mifsud, Manwel 1995 Loan verbs in Maltese. A descriptive and comparative study. (Studies in Semitic languages and linguistics 21.) Leiden: Brill. Minaeva, Vera 2003 Russian grammatical interference in Ket. STUF 56.1/2: 40–54. Moravcsik, Edith 1975 Borrowed verbs. Wiener Linguistische Gazette 8: 3–30. 1978 Language contact. In Universals of human language 1, Greenberg Joseph (ed.), Charles A. Ferguson, and Edith A. Moravcsik (ass. eds.), 93–122. Stanford: Stanford University Press.

118 Søren Wichmann and Jan Wohlgemuth 2003 Borrowed verbs. Manuscript in possession of the authors. Muysken, Pieter 2000 Bilingual speech. A typology of code-mixing. Cambridge: Cambridge University Press. Myers-Scotton, Carol 1993 Duelling languages: grammatical structure in code-switching. Oxford: Clarendon. Nash, David 1982 Warlpiri verb roots and preverbs. In Papers on Warlpiri grammar, in memory of Lothar Jagst, Stephen Swartz (ed.), 165–216. (Working Papers of SIL-AAB, Series A, Vol. 6.) Berrimah, N.T. Nau, Nicole 1995 Möglichkeiten und Mechanismen kontaktbewegten Sprachwandels unter besonderer Berücksichtigung des Finnischen. Munich /Newcastle: Lincom Europa. Neetar, Helmi 1990 Lehnverben und die deverbale Nominalderivation: insbesondere im Estnischen. In Materialy VI Mezdunarodnogo Kongressa FinnoUgrovedov Vol. 2, Savel’eva E. A. (ed.), 355–357. Moscow: Nauka. Newton, Brian 1964 An Arabic-Greek dialect. Papers in Memory of George C. Pappageot. Supplement to Word 20: 43–52. Pap, Leo 1949 Portuguese American speech: an outline of speech conditions among Portuguese immigrants in New England and elsewhere in the United States. New York, Columbia University: King’s Crown. Poplack, Shana 1980 Sometimes I’ll start a sentence in Spanish Y TERMINO EN ESPAÑOL. Linguistics 18: 581–618. Poplack, Shana, David Sankoff and Christopher Miller 1988 The social correlates and linguistic processes of lexical borrowing and assimilation. Linguistics 26: 47–104. Pugh, Stefan M. 1999 Systems in contact, system in motion. Studia (Uralica Upsaliensia 30.) Uppsala. Renault-Lescure, Odile 2004 Contacts de langues et réanalyse de certaines structures prédicatives en kali’na (langue caribe). Paper presented at the International Joint Conference of the SCL-SCPL-ACBLPE on Caribbean and Creole Languages, Curaçao, 11–15 August 2004. Romaine, Suzanne 1985 The syntax and semantics of the code-mixed compound verb in Panjabi/English bilingual discourse. Paper presented at the Georgetown University Round Table on Languages and Linguistics.

Loan verbs in a typological perspective

119

Sakel, Jeanette 2005 Universals of grammatical borrowing. Paper presented at the Romanisation Worldwide Conference; Bremen, 5–9 May 2005. Sankoff, David, Shana Poplack, and S. Vanniarajan 1986 The case of Nonce loans in Tamil. Centre de Recherches Mathématiques, Technical Report 1348. University of Montreal. Schatz, Henriette F. 1989 Code-switching or borrowing? English elements in the Dutch of Dutch-American immigrants. ITL Review of Applied Linguistics 83: 125–162. Schmidt, Christopher K. 2005 Untitled paper presented at the Loanword Typology Workshop, Max Planck Institute for Evolutionary Anthropology, Leipzig, 6–7 June 2005. Schultze-Berndt, Eva 2003 Preverbs as an open word class in Northern Australian languages: synchronic and diachronic correlates. In Yearbook of Morphology 2003, Geert Booij and Jaap van Marle (eds.), 145–177. Dordrecht: Kluwer Academic Publishers. Schütz, Albert J. 1978 English loanwords in Fijian. In Fijian language studies: borrowing and pidginization, Albert Schütz (ed.), 1–50. (Bulletin of Fiji Museum 4.) Suva, Fiji: Fiji Museum. 1985 Fijian language. Honolulu: University of Hawaii Press. Silva Encinas, Manuel Carlos 2004 La secuencia temporal en el discurso narrativo en la lengua yaqui. M.A. thesis, Universidad de Sonora. Simensen, Erik 2002 The Old Nordic lexicon. In The nordic languages: an international handbook of the history of the North Germanic languages, Oskar Bandle, Kurt Braunmüller, Ernst Håkon Jahr, Allan Karker, HansPeter Naumann and Ulf Teleman (eds.), 951–963. (Handbücher zur Sprach- und Kommunikationswissenschaft 22.1.) Berlin / New York: de Gruyter. Smith, Geoff P. 2002 Growing up with Tok Pisin. Contact, creolization, and change in Papua New Guinea’s national language. London: Battlebridge. Stenson, Nancy 1998 Spanish loanwords in Cakchiquel. In Studies in American Indian languages: description and theory, Leanne Hinton, and Pamela Munro (eds.), 223–233 Berkeley: University of California Press.

120 Søren Wichmann and Jan Wohlgemuth Tosco, Mauro 2005 Gawwada (East Cushitic; South-West Ethiopia). Paper presented at the Workshop on grammatical and lexical borrowing, Manchester, 29 September – 1 October 2005. Trask, R. Larry 1997 The history of Basque. London /New York: Routledge. Ulrich, Charles H. 1997 Loanword adaptation in Lama: testing the TCRS model. Canadian Journal of Linguistics 42.4: 415–463. Valenzuela, Pilar M. 2003 Los préstamos verbales del español en el shipibo-konibo: aportes a un propuesto universal del contacto lingüístico. Paper presented at the 51th International Congress of Americanists, Santiago de Chile, 14–18 July 2003. 2004 Los verbos del español en tres lenguas indígenas de Sudamérica: estudio de las estrategias gramaticales del préstamo verbal. Paper presented at the VII Encuentro Internacional de Lingüística en el Noroeste, Universidad de Sonora, Hermosillo, 17–19 November 2004. van der Auwera, Johan 1999 Periphrastic ‘do’: typological prolegomena. In Thinking English grammar: to honour Xavier Dekeyser, Guy A. J. Tops (ed.), 457– 470. Leuven: Peeters. Veerman-Leichsenring, Annette 1991 Gramática del popoloca de Mezontla (con vocabulario y textos). Amsterdam/Atlanta: Rodopi. Vendelin, Inga and Sharon Peperkamp 2006 The influence of orthography on loanword adaptations. Lingua 116: 996–1007. Vidal, Alejandra 2001 Pilagá grammar (Guaykuruan family, Argentina). Ann Arbor: University of Michigan Dissertation Services, 2001. (UMI Number 3004006). Vocke, Sibylle and Wolfram Waldner 1982 Der Wortschatz des anatolischen Arabisch. Erlangen: Universität Erlangen. Werner, Heinrich 2002 Vergleichendes Wörterbuch der Jennisej-Sprachen. Wiesbaden: Harrassowitz. Wichmann, Søren 1996 Cuentos y colorados en popoluca de Texistepec. Copenhagen: C.A. Reitzel. 2004a Structural patterns of verb borrowing. Paper presented at the Workshop on Loan Word Typology, Max Planck Institute for Evolutionary Anthropology, Leipzig, 1–2 May 2004.

Loan verbs in a typological perspective 2004b

121

Loan verbs in a typological perspective. Paper presented at the Seminar on Contact Linguistics, Linguistic Circle of Copenhagen, 14 December 2004. Wohlgemuth, Jan 2005a Loanword typology: establishing a database of verbal borrowings. Paper presented at the Romanisation Worldwide Conference; Bremen, 5 May 2005 and at the 37 th Studentische Tagung Sprachwissenschaft, Saarbrücken, 7 May 2005. 2005b Towards a typology of verbal borrowings. Paper presented at the Postgraduate Linguistics Conference; Victoria University, Wellington, 13–14 August 2005.

Why we need dynamic models for sociolinguistics and language contact studies Wolfgang Wildgen

1. Introduction: Why new models of language contact? Sociolinguistics and language contact studies have been in search of adequate theoretical models since the work of Labov and Bickerton, to name just two major theoretical figures in the two fields. Due to the dominance of Chomskian linguistics in the 70s and 80s both made compromises with the logical machinery underlying Chomsky’s proposals: – Labov has designed grammars with variable rules more or less adapted to current generative models (e.g., the standard model of Chomsky 1965). This led to major difficulties. First the probabilistic format responding to the application of statistics in sociolinguistics contradicted major postulates (in Chomsky programmatic booklet: Syntactic Structures), secondly the sociological claims and hypotheses of Labov’s sociolinguistics were quickly overridden by the consideration of inner linguistic complexities (e.g., phonological and morphological contexts of rule application). This was partially due to the fore-grounding of grammatical rules and grammar in sociolinguistic studies. In his later work, Labov gave up on the issue of a sociolect-(variation-) grammar and returned to more classical issues of urban dialectology and language change. More mathematically minded sociolinguists like David Sankoff pursued the issue, but had little impact on further development (cf. my own publications on variable rules and linguistic variation in Wildgen 1977a, b; 1994b). – Bickerton started from analyses in Creole linguistics (cf. Bickerton 1975) and formulated his bioprogram hypothesis as an explanation of universal features of Atlantic and Pacific Creoles. In analogy to Chomsky’s UG (Universal Grammar) he postulated a universal species-specific capacity to create new grammars in situations of language loss or in the case of a very imperfect transmission of language between generations. If the parent-generation communicates in a rudimentary Pidgin (loosing or not using their own language) the children-generation may repair this lack by activating the human bioprogram of linguistic creativity, which in

124 Wolfgang Wildgen cases of normal acquisition is quickly overridden or integrated by the grammar inherent in the language of adult care-givers. In his publication on language, species and protolanguages, Bickerton again used selected features of a Chomskian grammar, now called Government and Bindingtheory (Chomsky 1981), and proposed that X-bar-structures and casestructures (T-roles) make up the core of a protolanguage (cf. Wildgen 2004a: chapter 8 for a critical comment). In a more recent book with Calvin (Calvin and Bickerton 2000: 129) this issue is concentrated on three ingredients of a social “calculus”: (1) the ability to distinguish individuals and (2) to distinguish different types of actions, and finally (3) the representation of the roles of participants in actions. The machinery of grammar calculus is given up in favor of more pragmatic dispositions. If we take these two proponents of sociolinguistics and contact linguistics, we can summarize that between 1966 and 2000 the search for models at first followed the Chomskian paradigms, but could not really profit from their major features. The authors had to make an eclectic choice of features, which in many cases were later abandoned by Chomsky himself.1 Most empirical results were obtained without any connection to these models. One can propose two kinds of conclusions: a) These fields are not yet ripe for models formulating central principles and regularities; they should be further developed collecting raw observations and data (using the tools of classification and descriptive statistics). b) The type of models, which were at least initially well adapted to computational, technical linguistics, is not able to capture the insights and questions raised by sociolinguistics and contact linguistics. I guess that many of those who are struggling with problems of fieldwork, data collection and data-evaluation, will support the first conclusion. In the following I will instead follow the line of the second alternative. My major argument is that in the long run any data collection, due to the divergence of aspects, controlled factors and investigated levels, will ask for an instrument which can bring all the partial results together. The necessity to create a clear, systematic, and coherent architecture of the results, will always show up at a certain point. The demand for an appropriate model is unavoidable in any scientific enterprise.

Dynamic models for sociolinguistics and language contact studies

125

2. Why language is a dynamic system rather than a static structure To be clear, one should recognize that although Chomsky’s dissertation (in 1955) had the title: The logical structure of language, which reminds us of the Vienna circle and its American proponents (e.g. Quine), the program of his 1957 booklet Syntactic Structures definitely had a dynamic flavor. In chapter 3 (An Elementary Linguistic Theory) Chomsky describes finite state grammars as “the minimal linguistic theory that merits serious consideration” (Chomsky 1957: 24). Such a grammar describes sentences as state transitions. Although Chomsky is not concerned with its statistical features, this model stands in the tradition of the Russian mathematician Markov (cf. Markov chains with transition probabilities). Such a system produces/analyzes a sentence from left to right, considering choices (bifurcations), loops, transition probabilities and stand- and stop-conditions. The following schema is from Chomsky (1957), I have only added transition probabilities in order to show its affiliation to Markov’s models.

Figure 1. Example of a Markov-chain producing the sentences: The man comes, the men come (cf. Chomsky 1957: 19 for a similar schema).

In a psycholinguistic reading, such a model could describe the planning and uttering of two alternative sentences, i.e., two processes in time. Due to the transition-probabilities p1 … pn an information-value can be computed for the two sentences. I neglect the complications introduced by context-free, context-sensitive and transformational rule schemata. In all cases a dynamic interpretation in which time is either the measure of production or reception/ analysis is possible. It is not favored in Chomsky’s theory who sees grammar Gi just as a recursive definition of the notion “grammatical sentences” in a given language Li. The major features of this specific dynamic system are: – The steps are discrete; i.e., we have a discrete time, hopping from state to state. – The transitions may have a probability measure (as in Markov chains) and a Shannon information value.

126 Wolfgang Wildgen – The rule schema applied allows for the classification of different types of models. The so-called Chomsky-hierarchy is a choice from a larger set of rule-schemata, which are more or less context (memory) dependent. I will take a big step here from sentence production/reception (time in milliseconds) to language learning (time in months, years). In describing the acquisition process Chomsky (1979: 68) uses the same Markov chain model. The child starts with an initial state So for which Chomsky assumes some inborn UG (Universal Grammar). Crossing the learning stages S1 … Sn the child finally proceeds to the knowledge of grammar as a stationary state Ss. He is now a native speaker, the proper object of the analyses of grammarians. If we add transition probabilities to the rules (cf. Klein and Dittmar 1979), we may describe language acquisition as a kind of Markov chain, in which children/learners may show deviant lines though the final outcome is the same. This idea has been elaborated in the research on Pidgin-like immigrant language acquisition in Heidelberg. Thus the immigrants have a starting position V0 and going through the transitory stages: V1 …Vn could finally arrive at Vs, the full competence in the target language.2 There are, however, even in this very simple description major differences which have to be considered: a) Typically, the sequence is halted before the speaker has reached full competence. The competence of immigrant workers normally “freezes” at a certain stage depending on their age and their contacts (the type of work and the linguistic contacts it admits, etc.). The result is a rather rudimentary version of the target, a kind of Pidgin German (cf. Heidelberger Forschungsprojekt 1975 and Klein and Dittmar 1979). b) The targets are themselves multiple. In later studies the research group studied the local target for working class speakers in Heidelberg, i.e., lower class dialect (a variant of “Pfälzisch”). In reality, there are more targets. Beyond the dialect and a local version of the German standard, the immigrant meets the foreigner talk of German workers, and the Pidgin varieties of other immigrants. c) The initial stage, Vo, is variable depending on the native language/dialect of the immigrant and possible other languages he/she has learned, either in school or while she/he worked in other countries (e.g., France, Switzerland, Netherlands, etc.). Therefore, the simple linear model proposed by Chomsky and presupposed in all further elaborations breaks down (cf. Bechert and Wildgen 1991: 118– 121, for further discussion).

Dynamic models for sociolinguistics and language contact studies

127

A more fundamental problem concerns the network of grammatical categories one may presuppose. In the practical work on inter-grammars in the Heidelberg project, it became clear that it is almost impossible to find a general and stable set of such categories (phonological, morphological, lexical, syntactic) which are neutral to the process of acquisition, i.e., the differences between grammars in matching the transition stages Vi … Vn do not just consist of a change in the probabilities of rule-application as the basic set of such rules and the categories they use are themselves affected by the process. If one looks back at Chomsky’s example of primary language acquisition, one wonders if So and Ss, the initial and final states, are really given. The obvious fact of language change tells us that Ss the target, changes over time and that this change is due to language acquisition. So is not given at birth (or in the moment of impregnation), the process rather begins at the end of the first year, when the child begins to utter words, and it continues until the ages of ten to twelve. As the results of Piaget and the Geneva school of genetic psychology have shown, the acquisition of senso-motoricskills and basic symbolic schematization prepare the first utterances at the cognitive level. As these processes are not identical in all children and contexts, So must be variable too. In the transition from So to Ss, the child must make inferences from given inputs. 3 Moreover, the attainment of the target, even if it is homogeneous, is only possible at the limit. As in the case of the immigrants, the child stops this process before it can reach this limit; i.e., at a stage: Ss-i. As a consequence, the competence of one child (even if all targets are identical) is not identical with that of other children. As any society is more or less multilingual, these targets vary. Thus we have three sources of change between the generations: 1) Variations in the starting position of language acquisition after maturation and cognitive development. 2) Variations in the incomplete stop-positions of the acquisition process. 3) Variations in the targets themselves. As generations do not follow in a specific and discrete rhythm valid for a whole population, these variations define a continuum of linguistic change, i.e., language is in a steady process of change based on the accumulation of an imperfect “copying” of the target and of the intrinsic variability of targets in a language community. It is from this perspective that language change may be compared to biological evolution driven by mutation and genetic

128 Wolfgang Wildgen drift. The difficult part in this comparison concerns selection and speciation. In any case, simplistic dynamics of a linear and discrete sequence of states, identically realized by all language learners as suggested by Chomsky, is an illusion and we have to consider non-linear and continuous dynamic systems in our quest for a model adapted to the questions arising from sociolinguistics and contact linguistics.4 3. Language shift and language loss as sudden (catastrophic) events In biological evolution one may observe rather radical anatomical changes which induce far reaching differences in behavior and in the relation of a species to its ecology (to other co-species or to the physical/biological environment). Upright walk, the fabrication of instruments, and the transition between warning calls to a full-fledged lexicon were such transitions in the evolution of modern man. In linguistic change, large scale moves in the manner of the Great Vowel Shift in English or Grimm’s law for plosives (p, t, k) are classical examples. Shifts between morphological types: flectional, agglutinative, analytic, between ergative/nominative case systems, or in the basic word order (SVO, SOV, etc.) of sentences are such macro-changes. I cannot treat this topic in detail here because it only indirectly concerns language contact. It is, however, clear that large scale and very systematic changes are not the product of the infinitely variable usage conditions but depend on the search for optimality of a subsystem (phonological, morphological, syntactic) or (more rarely) the system as a whole. This means that language as a system (of subsystems) has internal restrictions due to selforganization, which allow us to distinguish unstable, transitory and stable (optimal) states. The process in search for stability and optimality is nonlinear, i.e., after a phase of transition with high variability (turbulence) a plateau is reached which functions as an attractor of the system. It can be called an internal goal (not a target for any individual speaker but the “causa finalis” of some “invisible hand”). In Wildgen and Mottron (1987: 103–109) the Great Vowel Shift in the history of English was modeled using catastrophe theory. In the first picture (below) we may observe a rather closed field of long vowels. After the Great Vowel Shift some of them have changed dramatically from the class of prototype vowels to that of glides (from monophthongs to diphthongs), i.e. from a category with one center to one with two centers (and a motion in the articulatory space). The picture below shows the linear sequence of minima on the frequency scale and how some of them reach the frontier line. As a consequence they have to move

Dynamic models for sociolinguistics and language contact studies

129

to a new quality. The dramatic shift from monophthongs to diphthongs is understood as a transition through a frontier (a fold-line) at the left and a jump to a higher type of attractors, which are more like limit-cycles (with two state dimensions in the vowel space). Labov et al. (2006) describe chain shifts in Northern America, e.g. the chain shift of Northern Inland dialects. They show that specific topological laws concerning peripheral and non peripheral vowels apply and they measure the vectors of language change in the vowel space (defined by the formants F1 and F2). These results show that the techniques of vector analysis and optimality laws are relevant for a theory of language change.

Figure 2. The “Great Vowel Shift” as a global shift of an attractor configuration beyond the borderline and the creation of diphthongs as a structural consequence (cf. Wildgen and Mottron 1987: 104f.)

An even more dramatic shift occurs if a population gives up its language in favor of another language or in the case of language death. This case has occurred many times in the process of the colonization of the Americas, either because the population died out physically or because it adopted a Creole language induced by colonial contact and gave up its indigenous language. A rather simple case of language shift may be observed if regional dialects are given up in favor of some dominant prestige language. This was

130 Wolfgang Wildgen the case in Bremen, where the shift from the Bremer Platt (Low German urban dialect) to High German occurred at the end of the 19th and the beginning of the 20th century. Based on biographical interviews with older inhabitants (born around 1900) we could reconstruct a rather quick break down of the dialect in the urban context. If we compare the period from old Saxon to Middle Low German and current Low German dialects which have a time depth of at least 1000 years (ca. 900 to 1900) to the languageshift and -loss which took only two generations (ca. 1870 to 1920), we can conclude that the second change was 200 times quicker than the historical language change. Details of the process have been elucidated during fieldwork done in the years 1982–1988. The general picture is that of a dominance shift in the use of spoken Low German in favor of the written language of school and administration: High German. This shift is first a shift in the attitudes towards the value (market position) of the dialect and then a shift in the preferred choices in a bilingual situation (cf. Wildgen 1986 and 1989 for further details). Figure 3 shows the dominance shift as a change of attractor landscapes with two attractors: A and B at time 0, 1 and 2. Level c describes a linguistic state as a kind of fluid caught by an attractor; at time t=1 it fills both attractors, at time t=2 it has definitely changed to attractor B.

Figure 3. Shift in an attractor field

Figure 4 shows the bifurcation space (with the unfolding parameters: u and v) in a bipolar dynamic field (the surfaces of stable and unstable states with the border lines where the catastrophic jump occurs are shown above). The type of catastrophe which drives a change of the type indicated in Figure 3 is called a cusp-catastrophe.

Dynamic models for sociolinguistics and language contact studies

131

Figure 4. Language shift as a path in a bipolar field called cusp. The zone of unstable states is the area where both languages coexist with different prestige. At the catastrophic jump one of them disappears. Beyond the cusp point a smooth transition is possible.

A more refined synergetic model has also been proposed in Wildgen (1986: 126–135) in analogy to models for migration in towns triggered by in-group and out-group sympathies; i.e. small demic migrations in towns to specific dwelling areas seem to have similar dynamics as symbolic migrations of linguistic varieties (cf. also Wildgen 2005a). In the case of Bremen, the autobiographic novels of Georg Droste (1876–1935) show that the urban context, in which he lived at the end of the 19th century witnessed a dramatic demic (=population) migration, as new streets were built with houses sold to middle and high class citizens, whereas the poor population living in a network of cottages (“Katen”) linked by foot paths (“Gänge”) had to leave the area. After a phase of social and linguistic contacts the population and the language had shifted. Thus the shift of language was parallel to a massive social migration inside the town which was partially triggered by work migration from rural areas into the town and by the migration of middle and upper class people from other towns to Bremen. This shows that the parallel treatment of demic and linguistic migration is not just a fiction of the model; both processes stand in a causal relationship and show similar dynamics.

132 Wolfgang Wildgen 4. The dynamics of interference/transference in language contact The basic problem of traditional contact studies has been the impact of elements, structures of a contact language on the language studied and the possible integration of the effect. Thus, for German, one may analyze how Latin words and patterns were integrated into German Bible translations, and how old and middle French, later modern French and English left traces in the lexicon and grammar of German. The underlying theoretical problem is that of the “blending” of linguistic structures and of the psychological (mental/neural) and sociological conditions of such a process.5 We may start with the technical question, how an item of language L1 is transferred to language L2. In order to answer this question, we first need a proper format for the representation of the items in contact. If we follow the “information paradigm” of Pollard and Sag (1987)6, the phonology (PHON), the syntax (SYN) and the semantics (SEM) of an item can be described by a set of features (phonological in the first, syntactic in the second, and semantic in the third case). The interference or blending is then mainly an operation on the feature sets: PHON and SEM (I shall ignore the syntactic feature set in the following). Basically if PHON = {p1 … pn } and SEM = {s1 … sm }, then the blending may be just a choice function, a filter of two pairs of feature sets: (PHON1, SEM1) and (PHON2, SEM2). This filter may only unite the feature lists or exclude some of the features which are contradictory or incongruent. Thus, one can describe the phonological transfer from lat.: tabula to old High German: tavala /tabala and to New High German: Tafel by a transformation (change and deletion) of the phonological feature matrix, which excludes or modifies phonemes not available in the German system (and via sound changes affecting the result of borrowing). In a similar way, the semantic features can be modified or restricted, such that readings like “panel of wood” only show up in compounds like “Tafelbild” or the reading as “board” only in “Wandtafel”, whereas the actual word for “table” is rather “Tisch” than “Tafel”. In general, possible synonyms tend to specialize; i.e. the reading of a borrowed element tends to diverge from the meaning of the inherited synonym. The fact that the preferred (prototypical) shapes of tables in the context of the kitchen or the dining room can shift independently from the lexical label, points to further onomasiological dynamics in this field. My questions are, however, more fundamental: 1. Are check lists with criteria (or features) an adequate description (in phonology and semantics)?

Dynamic models for sociolinguistics and language contact studies

133

2. Are transformations operating on discrete finite sets (replacement, deletion, possibly addition) the correct model for the dynamics one can observe in language contact situations? The simple example above already shows that rather complex operations in the articulatory/auditory space occur not only in the integration of foreign patterns, but also in the operations of language change which shape them after borrowing. In the semantic or even in the referential domain (cf. the tradition “Wörter und Sachen”) continuous shifts occur, i.e., the blending is not describable by a kind of replacement mechanism characteristic for generative grammars. As continuous and non-linear models are rare in linguistics, no straightforward solution is available for this basic problem of contact linguistics. I will in the following discuss three major epistemological problems of language contact studies: – Language is ontologically inhomogeneous (it concerns different domains). – The dynamics are different (continuous, discrete, fractal) in different domains. – Transfer phenomena occur on all levels. 1. Ontology: The tempting idea of Dretske (1981) and Pollard and Sag (1987) was that we have just one homogeneous space, that of information, which may be further separated into: phonological, semantic, syntactic information. However, Lakoff has shown the fundamental divergence of domains/spaces and the relevance of mappings between them (cf. his theory of cognitive metaphors). A rather specific architecture has been proposed by Brandt (2004: 26) who distinguishes four domains with one center (cf. Fig. 5).

D1 Natural domain

D4 Spiritual domain D3 Mental domain

Ontological network proposed by Brandt (2004)

Figure 5. The panorama of semantic domains (cf. Brandt 2004: 26)

D2 Cultural domain

134 Wolfgang Wildgen The etymological path of a word may show referential changes (D1) which are due either to a mapping change relative to the mental image (D3) or to cultural changes affecting the world it refers to (D1). More abstract blends involve D2 and D4; thus the lexicon of intellectual concepts may be either linked to the thinking/believing of individuals (D4) or to cultural/collective modes/myths of thinking (D2). The blending theory of Fauconnier and Turner (2002) considers the existence of basic schemata in a so-called “generic space”, which may control the selection of a blend. In terms of Brandt’s ontology, they may be rooted in D3 (cognition), which contributes specific measures of simplicity and optimality. The process of blending can be exemplified in the case of metaphors, irony, and wit. However, it remains rather obscure how the semantic spaces on which these processes operate are structured (as default the authors assume some discrete feature matrix). Phonological blends operate on an acoustic/auditory/articulatory space, which is of type D1. Insofar as cultural preferences (D2) and properties of an abstract system of oppositions/contrasts (D3) contribute to its dynamics, it does not only refer to D1.7 As a conclusion one may say that the artificial homogeneity of feature-(information-) models does not correspond to the phenomenological stratification of human sign behavior and the kind of blend/ integration which occurs in language contact cannot be described as a purely informational process resting on features matrices. 2. Dynamics: The natural domain (D1) evidently has a continuous spacetime organization and a type of dynamics corresponding to these spaces, whereas the mental domain D3 entails a preference for quasi-categorical spaces, and thus rather asks for a kind of discrete dynamics. Generative grammars have radicalized this choice and have favored a cognitive interpretation (D3) excluding the individual/spiritual (D4), cultural (D2) and objective (D1) aspects of the ontological network.7 The cultural domain, most visible in sociological, ethnological, cultural studies of language, is characterized by statistical flows, decentralized dynamics and chaotic phases, with the effect of divergence or convergence. Models encompassing the effects of this domain are neither low dimensional and continuous as D1, nor categorical and discrete as D3, but rather stochastic with flows (fashions, trends, invisible hand effects). I dare not go into the details of the cultural domain D4, but myth, religion, literature, and art show the very complicated effects which mix individual creativity and cultural trends (cf. for innovation in art Wildgen, 2004b). 3. Transfer: Linguistic and cultural contacts involve all the basic spaces. In the case of D4 the construction of linguistic identity involves the inven-

Dynamic models for sociolinguistics and language contact studies

135

tion or assembling of myths. Evaluative shifts underlying language shift involve theories/ideologies and future projections, which direct the projective socialization of parents (mothers) in view of the career of their children. Therefore, it would be a mistake to consider transfer-phenomena just on the level of D1 (phonetics) or D3 (neural organization of bilingual minds). Nevertheless, the basic spaces D1 and D3 seem to govern a large part of transfer processes and therefore traditional descriptions of lexical borrowing have made a good strategic choice, when they concentrated on words (D3) and referents in the world (D1) (cf. the school called “Wörter und Sachen”). In general, the splendid simplifications which made life easier for computational linguists in the 70s and 80s are not helpful in the realm of sociolinguistics and language contact studies. 8

5. Conclusion The macro-dynamics of language shift on a historical scale are accessible to methods of dynamic systems theory and can use the already successful models for demic migration and economic dynamics. At the cognitive level, it becomes clear that non-linear and chaotic dynamics are relevant and these insights limit the application of the mechanistic computer metaphor to cognitive linguistics. In the field of contact linguistics proper, the constructions of political and cultural identities and cultural dynamics (including its economic conditions) have been highlighted in studies on Pidgins and Creoles. An integrating model which involves the dynamics of the whole architecture of domains is not yet accessible, but it becomes clear that the ontological “flattening” of the grammatical model which neglects, for reasons of simplicity, the phenomenological multiplicity of the linguistic sign, has been a garden path.

Notes 1.

2.

Such partial and interdisciplinary applications, mostly after the establishment of a new generation of generative models, had almost no impact on the further development of generative linguistics itself. It did not even change the sociolinguistic abstinence of theoreticians in this tradition. Cf. Heidelberger Forschungsprojekt (1975), Klein and Dittmar (1979) and for a short summary Bechert and Wildgen (1991: 118–129).

136 Wolfgang Wildgen 3.

4.

5. 6. 7.

8.

9.

Usage based theories of language acquisition (cf. Tomasello 2003) show the eminent variation of acquisition processes and how they depend on the language addressed to children by adult care givers or siblings and peers. The evolution of human language, mainly the dramatic differences between humans and other animals in this respect, suggests that language is an open rather than a closed system. It has an essential range of flexibility and creativity. These features are not only visible in the historical change of languages but also in language acquisition and situated language use. For a general description of borrowing and transfer cf. Bechert and Wildgen (1991). It is philosophically based on Dretske (1981), and further elaborated in HPSGgrammars. The central role of abstract (mental) relations and processes made structural linguists like Hjelmslev think that the linguistic system is basically an abstract formal machinery like logics. Montague grammars and generative grammars are very close to such a position. It seems to be a basic epistemological trend in linguistics to reduce the ontologically mixed network to one major field of concern and to eliminate all other aspects as less relevant. This is not a Galilean style, as some authors argue, but just a lack of semiotic insight. At the beginning of modern linguistics Ferdinand de Saussure has criticized diachronic linguistics for its incomplete view on the linguistic sign (insofar as it can only have access to the external shape of the sign, its “signifiant”, cf. Saussure 2003). A similar critique can be made to technically minded grammars, as they are semiotically truncated. Actual cognitive models try to overcome this restriction (cf. Wildgen 2008).

References Bechert, Johannes and Wolfgang Wildgen (in cooperation with Christoph Schroeder) 1991 Einführung in die Sprachkontaktforschung. Darmstadt: Wissenschaftliche Buchgesellschaft. Bickerton, Derek 1975 Dynamics of a Creole system. Cambridge, MA: Cambridge University Press. 1981 Roots of language. Ann Arbor: Karoma. Brandt, Per Aage 2004 Spaces, domains, and meaning. Essays in cognitive semiotics. (European Semiotics Series 4.) Bern: Lang. Calvin, William H. and Derek Bickerton 2000 Lingua es Machina. Reconciling Darwin and Chomsky with the human brain. Cambridge, MA: MIT Press.

Dynamic models for sociolinguistics and language contact studies

137

Chomsky, Noam 1957 Syntactic structures. The Hague: Mouton. 1965 Aspects of a theory of syntax. Cambridge, MA: MIT Press. 1979 A propos des structures cognitives et de leur développement: une réponse à Piaget. In Théories du langage, théories de l’apprentissage. Le débat entre Jean Piaget et Noam Chomsky, Massimo Piatelli-Palmarini (ed.), 65–87. Paris: Éditions du Seuil. 1981 Lectures on government and binding. Dordrecht: Foris. Dretske, Fred I. 1981 Knowledge and the flow of information. Cambridge, MA: Bradford / MIT-Press. Fauconnier, Gilles and Mark Turner 2002 The way we think. Conceptual blending and the mind’s hidden complexities. New York: Basic Books. Heidelberger Forschungsprojekt “Pidgin-Deutsch” 1975 Sprache und Kommunikation ausländischer Arbeiter. Analysen, Berichte, Materialien. Kronberg/Ts.: Skriptor Heine Bernd and Tania Kuteva 2005 On the explanatory value of grammaticalization. Leipzig: Max Planck Institute for Evolutionary Anthropology, Paper. Klein, Wolfgang and Norbert Dittmar 1979 Developing grammars. Berlin /New York: Mouton de Gruyter. Labov, William 1972 Contraction, deletion, and inherent variability of the English copula. In Language in the Inner City. Studies in the Black English Vernacular, William Labov, 65–129. Philadelphia: University of Pennsylvania Press. 2001 Principles of linguistic change, vol. 2: Social factors. Oxford: Blackwell. Labov, William, Sharon Ask and Charles Boberg 2006 The phonological atlas of North America. Berlin /New York: Mouton de Gruyter. Lakoff, George and Mark Johnson 1981 Metaphors we live by. Chicago: University of Chicago Press. Pollard, Carl and Ivan A. Sag 1987 An information-based syntax and semantics, Vol. 1: Fundamental. (CSLI Lecture Notes 13.) Stanford: Leland. Sankoff, David 1978 Linguistic variation. Models and methods. New York: Academic Press. Saussure, Ferdinand de 2003 Wissenschaft der Sprache: neue Texte aus dem Nachlaß. Frankfurt am Main: Suhrkamp.

138 Wolfgang Wildgen Tomasello, Michael 2003 Constructing a language. A usage-based theory of language acquisition. Cambridge, MA: Harvard University Press. Wildgen, Wolfgang 1977a Differentielle Linguistik, Entwurf eines Modells zur Beschreibung und Messung semantischer und pragmatischer Variation. (Linguistische Arbeiten 42.) Tübingen: Niemeyer. 1977b Kommunikativer Stil und Sozialisation. Eine empirische Untersuchung. (Linguistische Arbeiten 43.) Tübingen: Niemeyer. 1986 Synergetische Modelle in der Soziolinguistik. Zur Dynamik des Sprachwechsels Niederdeutsch-Hochdeutsch in Bremen um die Jahrhundertwende (1880–1920). Zeitschrift für Sprachwissenschaft 5.1: 105–137. 1989 Bremer Sprachbiographien und die Verdrängung des Niederdeutschen als städtische Umgangssprache in Bremen. In Niederdeutsch und Zweisprachigkeit. Befunde – Vergleiche – Ausblicke, Ulf-Thomas Lesle (ed.), 115–135. Leer: Schuster. 1994a Process, image, and meaning. A realistic model of the meanings of sentences and narrative texts. Amsterdam: Benjamins. 1994b Sprachliche Variation: Ebenen und Prinzipien. Grazer Linguistische Studien 41: 95–109. 1996 Dynamik des Haiti-Kreols: ein Fallbeispiel für die Interaktion von Sprachen und Kulturen. In Togo, Kamerun und Angola im euro-afrikanischen Dialog, Hella Ulfers and Martin Franzbach (eds.), 112–124. Bremen: Universitätsverlag. 2004a The evolution of human languages. Scenarios, principles, and cultural dynamics. Amsterdam: Benjamins. 2004b Conceptual innovation in art. Three case studies on Leonardo da Vinci, William Turner and Henry Moore. In Seduction, community, and speech: a festschrift for Herman Parret, Frank Brisard, Michael Meeuwis and Bart Vandenabeele (eds.), 183–196. Amsterdam: Benjamins. 2005a Migration von Sprachen und Kulturen – Überlegungen zur kulturellen Dynamik von symbolischen Formen (besonders im Alpenraum). In Das Zimbrische zwischen Germanisch und Romanisch, Ermenegildo Bidese, James R. Dow and Thomas Stolz (eds.), 24–42. (Diversitas Linguarum 9.) Bochum: Universitätsverlag Dr. Brockmeyer. 2005b Sprachkontaktforschung. In Soziolinguistics/Soziolinguistik. An international handbook of the science of language and society, Ulrich Ammon et al. (eds.), 1332–1346. Berlin / New York: Mouton de Gruyter. 2008 Kognitive Grammatik. Klassische Paradigmen und neue Perspektiven. Berlin /New York: Mouton de Gruyter.

Dynamic models for sociolinguistics and language contact studies

139

Wildgen, Wolfgang and Laurent Mottron 1987 Dynamische Sprachtheorie. Sprachbeschreibung und Spracherklärung nach den Prinzipien der Selbstorganisation und der Morphogenese. Bochum: Brockmeyer. (Download of the first section: http:// www.fb10.uni-bremen.de/homepages/wildgen/pdf/das_dynamische_ paradigma.pdf )

Constructivist theory of language contact and the Romancisation of indigenous languages Klaus Zimmermann

1. Preface For about a few years, I have been trying cautiously to apply constructivist theory, on the basis of neurophysiologic raisings (Roth 1996, 2003, Glasersfeld 2003, Hirsch 1997, Fabbro 1999, Maturana 1980, Schmidt 1987), to a theory of language as well as to a theory of language contact, including Creolisation (Zimmermann 2004b, 2005a and b). As the topic of this book, the notions of Romancisation, Hispanicisation, Frenchification etc. pose a challenge to my theoretical approaches to language contact concerning the necessity of elucidating the role constructivist theory plays in these “processes”, in the following, I am going to take a closer look at the extent to which the above mentioned processes can be approached from the constructivist perspective and whether this perspective brings something about that can contribute to a better understanding of them. Firstly, I am going to outline my version of a constructivist theory of language contact. I am then going to clarify the relation between the contact “facts” given and what is understood by Romancisation, Hispanicisation, Frenchification, etc. Finally, I am going to state some cases of Hispanicisation that provide evidence on the postulates of a constructivist theory of language contact, assuming that the concepts of Romancisation, Hispanicisation, and Lusitanisation etc. are linguistic-cultural processes in which an asymmetrical result is produced between the respective languages in contact, with a Romance language exerting a strong impact on the nonRomance language, and that this difference can be explained by a neurophysiology-based constructivist language theory. 2. Constructivist theory of language contact The constructivist approach to language contact emanates from the fact that any type of verbal activity is governed by the brain, which is the reason why all the factors (namely sociocultural ones) that may affect the production

142 Klaus Zimmermann and reception of utterances pass perception and cognitive processing. Moreover, the factors that – as is it assumed traditionally – influence and even determine interactive and linguistic behaviour are influential only in the way they are constructed by cognition. These factors are not exerting influence in their “objective form” then, as scientific objectivist theory claims. They behave exclusively in the way they are constructed. This does not signify the negation of their existence. Rather does it underline the need for a theory that explains the influences one language has on another (indicating a certain kind of causality) and elucidates the form these factors are constructed by the cognition of individuals and their relevance to them (which might not be identical to the analysts common construction of these factors), so that an ascertainment of their impact can be undertaken. This implies that factors such as ideologies, myths, “religious beliefs” as well as “mistaken” views on the environment may have influence on language contact, too. Therefore, I argue for an ascertainment of language contact that starts from the individual brain, basing myself on the neurophysiologic assumption that every brain has a feature that is individual, very flexible and dynamic, which we commonly refer to as language, postulating a strong identity between the individuals. With this axiom, I am going to draw upon the discoveries and theoretical raisings made in the field of neurophysiology (which – this has to remain clear – sometimes are necessarily speculative). However, this theory can also relate to previous approaches that are not explicitly based on neurophysiology. In his book, Uriel Weinreich (1953), founder of modern contact linguistics, for example, stresses several times the axiom that says that “the individual is the location where the processing takes place”. Also Eugenio Coseriu, expounding language change and, without doubt, drawing upon W.v. Humboldt, localises the causes of this change in the speech itself (the exchange of verbal messages between individuals), that is the composition of utterances and their units which express and communicate something to someone in a historical context (Coseriu 1974: 176). Pointing this out, I would like to make clear that we must keep in mind the mentioned aspect of the cognitive construction of meanings, whether conceptual (lexical) or operative (grammatical). Thus, my objective of presenting an outline of a constructivist theory of language contact is not novel in its entirety. It is partly the continuation of an existing thought that has not yet been fully accepted, and it is partly an attempt to phrase it in its radical perspective. The attempt of explaining language contact on the neurophysiologyoriented constructivist basis has a few consequences which I would like to sum up briefly: this approach allows a clear and well-founded distinction be-

Constructivist theory of language contact

143

tween four structural types of language contacts in the individual, relying on the criterion of cognitive cerebral activity, which in each case is different.1 a) Elements of L1 in L2, i.e. the perception of L2 and its elements and structures through the brain marked by L1 with the aim to speak an L2. We refer to this process as interference. In this case, the basic interactive language is, in its nature, an L2 language for the speaker. b) Elements of L2 in L1, i.e. the incorporation of L2’s elements and structures in L1. We refer to this process as transference. The basic interactive language in this case is the speaker’s mother tongue. Examples of transferences are: lexical borrowing, foreign words, borrowed meaning, calques, loan-word translation, etc. c) The speaking of two languages without drawing upon one basis. This results in a subjective sole language. We refer to this process as blending of languages or hybrid languages. We can distinguish between two types of blending: structural blending and communicative blending. Structural blending involves the use of structural, grammatical and lexical elements of both languages without differentiating between the languages’ identities (examples: children’s early bilingual speech, the formative stage of Creole languages, language intertwining [after Bakker and Mous 1995]). Communicative blending involves the alternative use of two languages while respecting the structural property of each of the languages (examples: code-switching and, particularly, code-mixing, cf. Auer 1984, 1998). d) It would be wrong to reduce studies on language contact to the detection of elements of one language in the other or in the blending, or to the positive detection of previously existing entities. The language contact as a situation can also provoke the creation of phenomena that did not exist before in either of the two languages in contact. A convincing example of this is the creation of simplifications and reductions in complexity. This phenomenon can be observed in the process of pidginisation as well as with other types of contact. Another example is the existence of neologisms whose elements do not come from two languages but from one only, emerging in some contact situation and creating a new specific identity, e.g. an immigrants’ identity. The famous verlan spoken by adolescent Maghreb and other immigrants in French suburbs is a modus operandi that is a well-known procedure within the French language since centuries, but out of productive use, until the actual reinvention, then however, by a group of French-speaking people of a different ethnic background (with bilingual parents), creating a new

144 Klaus Zimmermann non-standard jargon. The German-based Kanak-Sprak created by turkish youngsters is another example (cf. Auer 2003). Moreover, neurophysiologic Constructivism also shows us that languages, their identities and their alterity (otherness) in comparison to other languages are cognitive constructions too. Linguists have known this fact for some time. However, we have not yet drawn the relevant conclusions from that: in the fields of dialectology and sociolinguistics, it is pointed out that there are no discrete and objective borders between languages and varieties (Hudson 1980: 38–44) but that we construct our language concepts on the basis of certain criteria. These criteria are constructed and variable: something that is constructed as an important criterion at some historical moment and in some cultural context is not going to be constructed the same way by speakers that find themselves in a different era or a different political context. This can be retraced by looking at the cognitive construction of the differences between Romance languages (the difference between Galician and Portuguese; the difference between French and Franco-Provençal; and also the difference between Spanish (Castilian) and Catalan and by comparing these constructed differences to those of Bavarian, Swabian or SwissGerman which are considered as dialects and not proper languages, and even to the construction of Niederdeutsch (Low German) as a dialect during a certain period of time, not as an error of linguists but as a politically motivated, social construction. Other evidence for this is provided by the construction or non-construction of separate languages in many areas that we consider as areas of language transition (e.g. in the continuum between Spanish and Portuguese in the so-called fronterizo [borderland language in Uruguay]). On the basis of this political and social construction, it is also stated that speakers or professional actors often intervene in the structure and the status of their (or other) languages (by reason of language planning, for example) in order to transform them so that they fit with the constructed concept. Purism2, certainly, is one of the measures that aim at the re-establishment of language concepts whose validities were threatened in some way by transferences and blends. The concept of language (with the either explicitly or not explicitly constructed attribute of purity) and the concept of the ‘good’ L2-speaker (with the explicitly or not explicitly constructed attribute of perfection and nearnative-competence) that an individual has who wants to or has to speak an L2 have therefore a crucial impact on his/her cognitive handling of the contact. It is exclusively the desire to speak an L2 the same way as the native speakers do which evokes the effort to eliminate interferences of L1 in L2.3

Constructivist theory of language contact

145

If a speaker or a speaker’s community do not attach importance to the “structural and authentic integrity” of their own language or the other (which is not a natural aim but a social construction) then this enables every kind of interference, transference and blends. In fact, there are cases of language contact where the production of interferences, transferences and blends is not only not stigmatised but in fact cultivated. That is, they convert into an intentional object and into a linguistic behavioural habit (e.g. certain professional and generational groups in Germany at present do so with English, but also speakers of indigenous languages in Latin America who integrate a foreign vocabulary from Spanish and perform code-switching in order to acquire prestige and esteem within their native community, cf. Hamel 1988; Schrader-Kniffki 2003). The processing of the human brain bears a universal aspect concerning human cognitive capacity and an individual aspect considering the fact that every individual has a different individual brain (this is valid at the organic level, as well as concerning its specific cognitive formation through life) such as a different body which has an individual configuration too4. In the first step of analysis, we can state that the biological differences, the specific linguistic biography and the inevitably individual activity (autopoiesis) of every individual human produce a different construction in the brain of what we commonly refer to as ‘language’ in all its aspects: the construction of conceptual and functional meanings, the construction of acoustic images, the construction of strategies for interactive and situational use etc. (Zimmermann 2004a). In the second step of analysis, we can identify the existence of the desire to reduce the difference between the constructs. To a certain degree, speakers (the speakers’ brains) reach a partial uniformity by means of a cognitive activity in the very moment of trying to communicate, which we refer to as communicative viabilisation. This activity consists in the following process: the idiosyncratic linguistic constructions, which refer to the construction of both the physical and the meaning part of signs (conceptual and functional ones), and which every individual brain inevitably has to produce for the individual’s own sake, and where the origin of the idiolectal version of languages resides, are constantly compared to those of other individuals by listening to their messages and adjusting one’s own signs to those of the others reciprocally in the communicative practice.5 We can also say that every individual develops his/her own strategies for managing the language-contact situation they are involved in as well as for perceiving and constructing formal and meaningful (conceptual and functional) elements of L2. However, as they share the contact situation with other individuals, they also perceive the strategies and the strategies’ results

146 Klaus Zimmermann established by others. Because of the communicative viabilisation, a resemblance of strategies and results can emerge, which we conceive through abstraction, observing very often only the last moment of the process as a supra-individual strategy, be it by a group or by an entire community. This description of the process refers to the individual creative processing. It is understood that, in reality, something inverse can evolve. The individual has already been infiltrated by the strategies developed by others (by his/her parents, for example) and he/she is content to imitate them. But, as we have seen, before imitating them, he/she must have reconstructed them as an individual cognitive process of appropriation. Once decided about the strategy of how react to the existence of an L2, the concrete process of appropriating L2 begins – a psycholinguistic process. The individuals, who already have an L1 at their disposal and thus a linguistic-cognitive network of connections established by and in the brain, now make use of precisely this established network. These networks are consistent with the characteristic features of each language. Hence, the perception and the construction of elements of L2a differ between speakers of different varieties of L1b and speakers of an L1c. We therefore have two types of cognitive-cerebral processing: a) strategies to manage the contact, and b) glottocentric cognitive processing.

2.

Romancisation /Hispanisation /Lusitanisation etc.

2.1. Romancisation as concept of comparative studies Our neurophysiologic constructivist approach to linguistic contact focuses – as we have seen – on the aspect of cerebral processing and the aspect of the existing cognitive receptor system, and it assumes that this processing is individual and therefore, to some extent, different in every brain. Hence, it looks at the initial phase of contact at the individual level, explaining the social and homogenised processes within the contact via the communicative viabilisation of strategies and the processing’s results. There is nothing that keeps us from defining a process like Hispanisation firstly as an individual, and secondly as a parallel and/or subsequent process in many individuals. From the constructivist point of view, every individual of a community A that undergoes Hispanisation has to substitute their prior cognitive-linguistic system for the system of Spanish or of one of its variants (or for parts of the system). The compilation of these individual processes results in what is commonly understood by Hispanisation, a collective process by which is

Constructivist theory of language contact

147

affected a community of speakers. However, we must not simplify Hispanisation by regarding it merely as the sum of individual processes. Being a social process, it bears any kind of socio-political strategies, such as imitations, conformist decisions, pressure, the creation of circumstances that favour the substitution as well as they evidently favour resistance, imperfect substitutions, changes of strategies and returns, etc. Firstly, we have to clarify two aspects: the active and the receptive aspects. The semantics of words allows us to understand that the term ‘Hispanisation’ derives from the transitive verb to hispanicise somebody, i.e. bringing one’s language and culture to and sometimes imposing them on other people(s), as well as from the reflexive verb to hispanicise oneself in the sense that Hispanisation is also an active process of integrating something into one’s own system. That means that elements of the Spanish language are borrowed or one’s native language (or perhaps one’s adopted language by prior substitution) is substituted for Spanish. It is the latter derivation which is of importance for a constructivist theory of Hispanisation. Both processes (borrowing and substituting), in fact, interact with each other – very often in a conflicting way. Moreover, we have to distinguish between the notions of Hispanisation, Lusitanisation, Frenchification etc. and the concept of Romancisation. The latter is a concept that intends to describe a phenomenon starting from the concept of Romance languages, i.e. not from any kind of adoption of a language (I am simplifying the issue here for the sake of clarity) but from a family of languages coming from the same root, which is Latin. Therefore, I raise the question of what could be a more appropriate definition of Romancisation and ask what relevance, what scientific interest this kind of question brings about. It has to be taken into account that this approach is of comparative nature. Hispanisation, as a supposedly “real”, “concrete” process, is understood as being the substitution of one linguistic-cognitive system for another. If, however, we are constructing the process of Romancisation of the last five centuries, we do this because we want to compare the processes of Hispanisation, Lusitanisation, Italianisation, etc. But we also have to clarify the motives and the interests of the comparison, that is, the hypotheses that underlie the comparison. To begin with, we shall recall the distinction made between Erste and Zweite Romanisierung (First and Second Romanisation), which were introduced by Alwin Kuhn (1951)6. “First Romanisation” refers to the imposition of Rome’s culture and language on, as well as their adaptation by the peoples conquered during the dominion. This case, in fact, concerns the

148 Klaus Zimmermann expansion of – roughly speaking – one single language. One might call it better Latinisation for the sake of distinction. However, in this article and book, we deal with what Kuhn termed “Second Romanisation”, which we better shall call Romancisation in order to make the terminological distinction clear; because the first is quite different from the second, the latter concerning the expansion of different languages in different situations, regions and time periods and also with different cultural and political protagonists.7 What they do have in common is the shared origin of their languages and the general aim to expand colonially and exploit economically. However, these multiple expansions are quite different in time, region, type and cultural-linguistic provenience. They constitute a diversity of processes that we, as linguists, categorise from the outside under a common heading, drawing upon the criterion that there is a relationship between the donor languages. It is an important aspect that must be kept in mind in order not to mistake Romancisation for a single process. Romancisation rather concerns a multitude, a variety of processes (Hispanisation, Lusitanisation, Frenchification and Italianisation) linguistically independent from one another, to a certain degree, at the moment of their execution (sometimes conflicting among the colonising protagonists). Categorising it today as a multitude of European expansions emanates from the comparison’s aim, a category a posteriori. Moreover, considering their objectives and many other factors, these processes do generally not differ from other processes such as Anglicisation, Germanisation, ‘Dutchisation’ (e.g. in Surinam and the Netherlands Antilles), ‘Russianisation’ (e.g. in Siberia). That is, at this level of consideration, they do not differ from other cultural-linguistic processes within the framework of the European colonial expansion from the 15th century onwards.8 Taking on a comparative position, it seems appropriate to ask ourselves whether, as a directive hypothesis, if the partial processes of Hispanisation, Lusitanisation, Frenchification, etc. might share a common characteristic on the linguistic and psycholinguistic level. We would have to negate this hypothesis, if we took on the receivers’ point of view: in fact, the variety of appropriation processes is huge, considering the diversity of languages (Spanish, Portuguese, French, Italian…) and their different receiving communities (different languages of different families and types in different regions, different historical situations and different cultural characteristics). We could assume with more probability that there is a common or similar process within Hispanisation because of the existing common language politics during the colonial phase in the regions dominated by Spain, such as Spanish America, Philippines and some other parts of Austronesia (Sueiro Justel 2002). After becoming independent, in the early and late 19th century,

Constructivist theory of language contact

149

though, linguistic politics changed and countries like the Philippines underwent a very different political and linguistic history, which gradually produced differences between the languages affected in the different countries. 2.2. Listing of the phenomena of Romancisation As a second step, we shall argue which of the contact processes are accepted as phenomena of Hispanisation, Frenchification etc. I would briefly like to mention the following phenomena: a) Lexical borrowings (neologisms as well as relexifications, including complete relexifications in some languages), semantic borrowings, calques, etc. b) Textual and pragmatic borrowings c) Appropriation and integration of grammatical elements or structures (morpheme borrowing or imitation of the positioning of speech elements, etc.) d) Language intertwining, i.e. the systematic recourse to two grammatical systems e) Pidginisation f) Creolisation g) Using two languages as a conversational style: code-switching and codemixing h) Displacement of one language by another (the process that constitutes the most radical impact). I would like to remark that it is to this latter process that we assign for example Latinisation, i.e. the substitution of certain languages for Latin. Acknowledging all the above-mentioned processes, we can also, retroactively, speak of a Latinisation of Old German, bearing in mind the multitude of (lexical, grammatical and textual) borrowings from Latin in Althochdeutsch. Note that in Spanish America, displacement parallels other types of Hispanisation, having some people of the same ethnic group undergoing a process of displacement and others continuing to speak their native languages with the above mentioned “influences”. I suggest the recognition of all the mentioned processes – each one individually – as partial manifestations of Latinisation, Hispanisation, Lusitanisation, etc. It is not out of place to ask whether the explanation that the processes of Hispanisation constitute independent processes invalidates the concept of

150 Klaus Zimmermann Romancisation itself. I do not think it does. But we have to define properly the epistemic interest of this scientific construction of Romancisation. Claiming that what we refer to as Hispanisation is a multitude of cognitive processes that are processed independently of one another in the individual brain, although they do have a common political cause (the colonialism by Spain), we can justifiably ask whether these processes, despite being independent, produce equal results in individuals and in corresponding languages. If so, what are they due to? Are they due to the structure of their donor languages, to the structure of the receiving languages, or to appropriation strategies for foreign/second languages, which supposedly are universal for human beings? From a constructivist point of view, the similitude of the processes and results within a community is accounted for by viabilisation.9 A transindividual identity or similarity of these processes ought to be interpreted as the casual coincidence of the cognitive construction and elaboration of contact strategies or by copying one or both of them. If, in contrast, there is no substantial similitude in the results of Hispanisation in, for example Nahuatl, Quechua, Mapudungun, Otomi, Aymara, etc., which would be the conclusion of a detailed analysis and consistent with the proposals of constructivist theory, is this then due to the individual or glottocentric processing of those affected by it? I would like to point to the fact that the initial enthusiasm at having detected an identity of structures between the different Creole languages became somewhat diminished when the analyses of these languages advanced: “Those earlier creolists with their cavalier claims (as well as the more sober role models) have given way to a younger generation whose knowledge of the relevant facts is often broader and deeper” (Holm 2004: 64). Another approach can be one that is neither interested in the contact process nor in its determining social factors but rather in its result, e.g. its typological result. Although we are dealing here with a multitude of processes that are independent of each other, it has to remain clear that they might produce similar or identical results that would finish at a structural convergence of previously unrelated languages, solely because of the fact that they are exposed to the same (dominant, colonial) donor language. The latter approach seems to be justifiable with regard to the concepts of Hispanisation, Lusitanisation and Frenchification. The concept of Romancisation, however, does not seem very promising for this hypothesis, for Romance languages despite their genetic relationship differ today in various typological characteristics (French, at least to Iberoromance languages). Such differences would be e.g. the obligatory personal pronoun and the position of the negating particle or the verbal inflection in spoken French. Nonethe-

Constructivist theory of language contact

151

less, it seems appropriate to include in the comparative view the linguistic attitudes of the “giving” and receiving communities and their practice of communicating with the speakers that are subject to Hispanisation, Frenchification etc., which can be very different from each other. Even more important is the observation that the results of the different processes seem to be equal or similar. However, a closer look reveals that they are not. Stolz and Stolz (2001), for example, demonstrated that there are influences of Spanish in the system of comparison as well as Spanish borrowings of prepositions and conjunctions in many Native American and Austronesian languages. Still, these superficial similarities contrast with some striking differences: the authors also prove that, when undertaking the comparison, the form of the results produced by the influences on the different languages are very diverse. The case of preposition borrowings, too, reveals a considerable diversity in quantity and effect in the languages examined.10 I take this as a clear indication of the glottocentrism in contact strategies. Hekking and Muysken (1995) show that the amount and the degree of the integration of borrowings of Spanish functional words in Otomí (indigenous language in Mexico) and in Quechua (indigenous Andean language) are very different. According to them, this is accounted for by the structural difference between these two languages. Zimmermann (1987), in contrast, sees the borrowing motivated by social and communicative reasons. So I would argue that the difference demonstrated by Hekking and Muysken might be explained by the way speakers of Quechua and Otomí adopt a different individual glottocentric strategy and construct a different prestige of the Spanish language among each other. Due to its nature, we can determine the impact of gathering structural contact results and the method for comparing them. They are indispensable and imperative as gathering the data and generating hypotheses about the possible cognitive processes, but they cannot provide explanations for the results. From the theoretical perspective of constructivism, we can consider differences in contact results as an indication validating its postulate of individual and glottocentric cognitive processing. However, the similarities of the Spanish influences do not rise from an intrinsic characteristic of the structure of the Spanish language but from the simple theoretical situation that the language of the conquerors and colonisers was Spanish and none other. The comparison of the contact phenomena in various affected languages has to consider another important methodological aspect, though. The majority of examples of the affected Amerindian languages, Austronesian lan-

152 Klaus Zimmermann guages, African languages, etc. come from fragmentary studies and from corpus documents that are not representative. Thus, little or nothing is known about the extent to which the phenomenon is integrated as being part of the system of every speaker. In fact, in the contact situation, we will normally detect individually variable language uses of Hispanicisation. In contact theories, this variation has been partially recognised and modelled in continuum constructions (cf. de Camp 1971; Hill and Hill 1986; Zimmermann 1992: 208). But it is clear that these continuum models of contact varieties (e.g. Hispanisms in Amerindian languages and the Spanish ethnolects spoken by indigenous people – cf. Zimmermann 1992: 227–237) hide the individual variation too, establishing new constructions of contact varieties but at an intermediate level. If, however, we take a closer look at the data, we can detect individual patterns and styles of using transferences, interferences, code-switching etc., which demonstrates that Hispanicisation is a long and very individual process, becoming, after years and at a collective level, consolidated in an outcome that might be shared by a great number of speakers due not to the character of the donor language but to viabilisation. This viabilisation can be detected in the fact that of regional, social or situational variants of borrowing. This latter case arose between speakers of Otomí (Hamel 1988: 617, Zimmermann 1992: 208–209) and speakers of Zapotec (Zimmermann 2008 and Schrader-Kniffki forthcoming). 3. Problems a constructivist theory of language contact has to address As stated above, an investigation that aims at proving the postulates of a constructivist theory of contact would have to draw upon empirical comparative studies on structures the way Stolz and Stolz (1997, 2001) conducted them. The fundamental value of studies like these lies in its capacity for recounting the apparent traces left by the cognitive-constructive contact process. Constructivist theory presupposes the existence of at least two steps: a) the individual processing of the L2 by every speaker or of an initial speaker generating the appropriation of L2, and b) the process of communicative viabilisation and socialisation (diffusion) within the speakers’ community. If we, for the purpose of the analysis, isolated the genesis of the contact phenomena found in every affected speaker, I argue, we would obtain an individually different outcome and if we took on a comparative view between various speakers, we would obtain a variation of results. This variation could be detected in the process of viabilisation too, where the quantity of variants gradually decreases: some speakers affected by the con-

Constructivist theory of language contact

153

tact adopt the strategy used by others (assuming it to be a more appropriate one) until, at a (temporarily) defined “final” stage, there is one result, with the other alternatives being eliminated (for example, a lexical borrowing instead of a borrowing of meaning in the Hispanicised or Lusitanised language or contrary). There can be yet other factors that might direct this process within the processes of viabilisation and diffusion throughout the community: it is possible that the language use of people with certain social prestige has a greater impact, since they might impose their vision and attitudes, or they are accepted as leaders whose pattern concerning their contact strategy is more convincing11. In many communities we can observe too that there is a metacommunicative discourse about the different contact results, which might result in the appreciation of a deliberate ‘contactological practice’ (e.g. purist discourse). And there might also be interventions by political and administrative authorities that favour or prohibit one of the existing variants, or even try to eliminate it in everybody’s language use (language planning agencies). 4.

Some examples of Hispanisation (and Frenchification) from the constructivist point of view

4.1. The effects caused by language contact and their individual and subgroup variation As mentioned above, we observe a variation in the utterances made by individuals that are affected by the same contact combination and the same factors. On a macro-sociolinguistic level, we have individuals of the same group or the same community that either yield to the pressure of substituting their mother tongue for the language imposed – the latter being socially and politically more powerful – and others, having the same social, generational, sexual, professional, etc. characteristics, who resist this pressure and continue speaking their native tongue. For example, very often it is considered that indigenous people temporarily migratory in Spanish America are more likely to accept the substitution, whereas those who stay in their place of birth tend to retain their language. At the quantitative level, this observation might find certain corroboration. However, we usually content ourselves with the assumed quantitative logic. The majority respond to a deterministic law or rule. What we should do, however, is to ask about the reasons for the different behaviours. Why are some people influenced by

154 Klaus Zimmermann the same factors that others are not affected by? We can conclude, at least on a theoretical level, that what we presuppose as the deterministic force does not work hundred per cent. And I assume that this divergence resides in a different mental construction of these same factors. On the one hand, one can submit to the factors that trigger the substitution (constructing a picture of reality in which the affected person is without power and has to resign, or a reality in which the situation is evaluated and the act of submission/substitution is perceived as beneficial), or, on the other hand, the person is regarded as a subject of the situation and can evaluate it in a different way, namely as an unjust situation (which is a construction too), and consequently resist or even fight against the pressurising factors of the substitution. We can observe this in Paraguayan society, which at present is divided into two camps: those in favour of the official and educational use of Guaraní and those who favour the non-use. The same counts for smaller bilingual Amerindian communities such as villages. 4.2. Hispanicisation as an asymmetrical contact It was stated in the past that prototypically, language contacts can produce effects on both languages concerned. Spanish and the Amerindian languages have been in contact for more or less five centuries, and we can verify a clear asymmetry in the effects brought about by the contact, comparing the two language areas. On a sociolinguistic level, Spanish is expanding, whereas the Amerindian languages, with some special nuances like e.g. in Guaraní, they find themselves in retrocession, i.e. they are threatened with extinction (Zimmermann 1999; Crystal 2000). Generally, this is explained by the power and the political, economic and cultural pressure exerted by the Hispanic-speaking community on the Amerindian communities, forcing them to adapt personally and politically. We have also stated that the effect and the degree of interpenetration of one language into the other had been asymmetrical.12 There are far more Spanish borrowings (lexical ones as well as morpho-syntactic ones) in the Amerindian languages than vice versa. This impact was noticeable during the colonial period already: Los indios usan ya de muchos vocablos tomados de la lengua española, o porque no los ay en la suya, o porque se les han pegado con el trato de los (Bertonio, SJ [1612] 1984: A 7 v°– A 8 r°) españoles, como Candelero;13

Constructivist theory of language contact

155

los yndios ladinos […] españolizan lo que hablan […] y yerranlo tanto que ni bien hablan su lengua, ni bien imitan la nuestra.14 (González Holguín, SJ [1607] 1975: f.119r)

In order to explain these two asymmetries, one usually reverts to the extralinguistic factors borne within the relationship of power between the two affected speaker groups: the members of the powerful group impose their language through political power and cultural hegemony; for the same reasons, the members of the dominated group find themselves forced to learn the imposed language. With the progression of this process, the latter group borrows more elements of the dominating language than vice versa, also because of the other culture being more appealing. It happens irrespective of the fact that they face the imposition of a new political and religious system with their corresponding terminologies. Without negating these relations of power, it seems appropriate to point out that the language of the dominating group is not always imposed, nor must this kind of asymmetrical distribution of the borrowings necessarily evolve and that, for this reason, we can not identify it as a law. The Greek did not adopt Latin as a mother tongue during the Roman Empire; nor did the Franks or the Visigoths impose their Germanic language in neither Latinised Gaul nor the Iberian Peninsula. Despite their military success, both of these ethnic people adapted to the conquered culture and left few traces of their own languages in the adopted Romance language; nevertheless they maintained their role as the dominating political group. I therefore argue that the extralinguistic factors do not provoke a determined linguistic effect merely because of their existence and that they do not behave in a sort of active manner. It is rather that the linguistic outcome depends on whether or not the speakers perceive and construct these factors as relevant ones. 4.3. The case of blending (Media Lengua, Michif) During the past years, some language blends have been “discovered” which classic theory on language contact not only has been unable to explain but also would have qualified as ‘impossible’. I am referring here to the cases of the Media Lengua (Muysken 1979) in the Ibero-American world and Michif (Bakker 1995) in the Franco-American world. There are more cases to be read about in Bakker and Martens’ (1995) book on language intertwining. The Jopará in Paraguay, which still lacks a valid structural description, could

156 Klaus Zimmermann actually be considered under the heading of mixed languages. These cases of language blends provide evidence for the fact that in the past, people in this region used two languages freely, without being concerned either about the boundaries that existed between the languages or about grammatical categories, thus giving way to the actual mixed languages. They did not integrate the elements of one language in the other but spoke using two languages at the same time. The results that have become established themselves differ from each other: the case of total Spanish relexification while the grammar of the mother tongue (Quechua) is maintained (case of Media Lengua), and the case of the blending in which a certain grammatical pattern (today seeming to be regular) is the combination of the French nominal system and the Cree verbal system. These well described two cases (and I think Jopará and the occurrence of Pidgin languages too) demonstrate that language contact and the strategy for handling the speaking of the two languages available to an individual or to a group, do not follow a universal pattern. Some individuals or some groups choose to reject the blending (Purist orientation), whereas others make use of the two languages without heeding the (scientifically constructed) provenience of the linguistic recourses and without separating these two languages, thus producing what we would then refer to as blending.

4.4. Attitudes towards Hispanisation: purism vs. non-purism Zimmermann and Stolz (2001: 9) compared existing attitudes towards the influences Spanish has on very diverse languages, such as Chamorro in Austronesia and Mesoamerican languages. They pointed out that there are Spanish borrowings of the same kind in both language areas. However, they verified an important difference: Whereas in Austronesia, especially the speakers of Chamorro consider the borrowings Spanish as elements that form part of their Chamorran language and culture (where, conversely, English borrowings are regarded as foreign), in the respective Mesoamerican languages, those elements are considered to be distinct from their own culture; they are even looked upon as being an indication of the initiation of the extinction of the affected Mesoamerican languages. Even without drawing upon constructivist theory, this example reveals that the linguistic elements coming from one language or the other do not function as such in the brain; i.e. it is not their autochthonous or non-autochthonous “reality” (as often identified by scientific etymology) that determines their usage, but it is the speakers’ way of perceiving and constructing them as native or non-

Constructivist theory of language contact

157

native elements (which might or might not coincide with the findings of diachronic linguistics). Another example would be our studies on functional words borrowed from Spanish language in Otomí, conducted in the 1980s. Native speakers were presented with two sets of sentences in Otomí, one of which contained borrowings of Spanish functional words while the other one contained their autochthonous equivalences without borrowings at all (Zimmermann 1987). Being asked to give their opinion on the native character of the sentences, it turned out that both sets of examples were considered to be native by more or less half of the subjects. The differences were not statistically remarkable.15 I would like to conclude that the concept of the purity of language itself is a construction that has evolved historically in some language communities, not in all. Only on the basis of this construction have they been able to work out what we refer to as Purism, i.e. a cognitive system and a corresponding undertaking that identifies (more or less professionally) as well as it excludes the elements and structures that are considered as foreign to a language. Arguing that Purism and its construction of language elements as autochthonous vs. foreign is a matter of construction does not implicate that it is invalid or false16, as all concepts inevitably are cognitive constructions. We cannot live without constructions. Otherwise, this would imply that we could think and live without concepts. However, recognising the concepts’ character, we realise that the history of human thinking has been a constant struggle for a subjective approximatively adequate construction, which has improved more and more but will never completely conform to the real world.

5. Conclusion The examination of Romancisation is an undertaking that involves the comparison of individual processes of Hispanicisation, Lusitanisation, Frenchification, Italianisation, etc. In the course of the analysis, we not only must aim at the comparison of the results of these contact processes by drawing upon on a system-linguistic approach, but we also have to deal theoretically and methodologically with the postulate of the cognitive individual, regional and culturally caused variation of strategies for and attitudes towards the handling of the contact situation as well as with the language structures established before the contact in the cognitive system of the speakers and users of an L2. External factors in this process do not behave in a universal manner but diverge in a constructive way and must be elucidated in their cultural

158 Klaus Zimmermann specific relevance. The processes of Romancisation involve both the appropriation of entire languages, which then, because of substrate influences (interferences), can often result in the emergence of a new variety, and but the borrowing and integration of elements and structures into the own language. The empirical results of the contact manifestation can only be interpreted expediently, if they are based on an adequate theory that covers the cognitive and attitudinal (and also emotional) factors that have become relevant for the speaker and through which the socio-cultural frame factors become effective. The cognitive construction processes themselves cannot be observed; however, by means of diverse empirical methods applied in sociolinguistics, we can ’capture’ instants of these construction processes. The detection of contact results in performance utterances must be completed by a process oriented search for the way how viabilisation operates in language contact. Notes 1.

2. 3.

4.

5.

There are several divergent definitions of interference and transference in the linguistic literature. My clarification is based on the central criteria of cognitive processing. Cf. the collection of articles on aspects of Purism in various cultural contexts: Brincat et al. (eds.) (2003). Moreover, this attitude is often related to the motivation or the lack of motivation to be integrated in a foreign community, cf. the famous differentiation between integrative and instrumental motivation in L2-Learning, Lambert (1967). Precisely this individual configuration represents the origin of what we refer to as individual endowment. The differences are minimal in comparison with other species, but they are highly significant within the species at the social level and at the level of genetic selection. W. v. Humboldt had acknowledged the role of speaking in uniformisation. If we take the concept of “Objectivität” (Humboldt means it in terms of objectivation) as the goal of viabilisation, then “…ist das Sprechen eine notwendige Bedingung des Denkens des Einzelnen in abgeschlossener Einsamkeit. In der Erscheinung entwickelt sich jedoch die Sprache nur gesellschaftlich, und der Mensch versteht sich selbst nur, indem er die Verstehbarkeit seiner Worte an Andren versuchend geprüft hat. Denn die Objectivität wird gesteigert, wenn das selbstgebildete Wort aus fremdem Munde wiedertönt“ (Humboldt 1963 VII: 55-56, accentuation by K.Z.). [„… speaking will be the necessary condition for the thinking of every individual in seclusion. In its manifestation though, language only evolves in the community; and the human being only comprehends him-/herself by verifying experimentally the comprehensibility of his/her words with other human be-

Constructivist theory of language contact

6.

7.

8. 9. 10. 11.

12.

13.

14.

159

ings, as the objectivation is increased when the word formed on one’s own is echoed by a foreign mouth.] Kuhn (1951: 409) also points out that we are dealing in the second process with some sort of immense ‘linguistic laboratory’, preceding in this way the purpose of the conference from which this book results. Cf. Zimmermann (1992: 18). In fact, the First Romanisation-Latinisation cannot be regarded as a homogeneous process either, as it extends over various centuries under different political and social conditions in Rome, involving very different migratory people. In comparison, though, it is not at all as diverse as the Romancisation. Evidently we might not forget other colonial expansions like Arabisation in North Africa and Spain. For more details about this activity cf. Zimmermann (2004a). See for example the synopsis of the borrowing of functional words in Indomexican languages Suárez (1983: 136). What Labov (2001) describes as a diffusion pattern that reveals linguistic features and that induces a change generalised in a variety or a language may also work with the diffusion of strategies and results in language contact situations. According to Mendieta’s testimony, it seems that this was different during the early colonial period. “Y lo mesmo pasa acá de nuestra lengua española, que la tenemos medio corrupta con vocablos que a los nuestros se les pegaron en las islas cuando se conquistaron, y otros que acá se han tomado de la lengua mexicana. Y así podemos decir, que las lenguas y costumbres y personas de diversas naciones, se ha hecho en esa tierra una mixtura o quimera […]” (Mendieta, Book IV, chap.44). [And the same happens here with our Spanish language. We have it half corrupted with words that were added to ours on the islands when they had been conquered and those that they took from the Mexican language. And so we can say that, from the languages and customs and people from diverse nations, a mixture or a chimera has evolved here in this land.] In constructivist terms, I would say that during the colonial period, the constructions of language concepts and the constructions of the involved languages within the colony gradually changed, and due to this, the attitude toward the blending, the borrowing, etc. changed, too. Initially, the “mixture” did not seem to be scandalous to the speakers. Later, though, they assumed a more segregating attitude, as one can infer from Mendieta’s quotation, i.e. a more “conscious” and more purist attitude. [Those Indians already use many words taken from the Spanish language, either because they do not have them in their own language or because the words were imposed on them by the treatment of the Spanish, like Candelero.] [Those assimilated Indians make everything they say sound Spanish and they adulterate it in such a way that they neither speak their own language well nor imitate our properly.]

160 Klaus Zimmermann 15. Cf. the studies of attitudes towards language mixing of Hill and Hill (1980) and Hill (1988). 16. For a view of purism in language planning in the context of modernisation of Amerindian languages, cf. Zimmermann (2003, 2006).

References Auer, J. Peter C. 1984 Bilingual conversation. Amsterdam/Philadelphia: Benjamins. 1998 Introduction: ‘Bilingual conversation’ revisited. In Code-switching in conversation: language, interaction and identity, Peter Auer (ed.), 1–24. London: Routledge. 2003 “Türkenslang”. Ein jugendsprachlicher Ethnolekt des Deutschen und seine Transformationen. In Spracherwerb und Lebensalter, Annelies Häcki Buhofer (ed.), 255–264. Tübingen: Francke. Bakker, Peter 1995 Michif, the Cree-French mixed language of the Métis buffalo hunters in Canada. In Mixed languages: 15 case studies in language intertwining, Peter Bakker and Maarten Mous (eds.), 13–33. Amsterdam: IFOTT. Bakker, Peter and Maarten Mous 1995 Introduction. In Mixed languages: 15 case studies in language intertwining, Peter Bakker and Maarten Mous (eds.), 1–11. Amsterdam: IFOTT. Bechert, Johannes and Wolfgang Wildgen 1991 Einführung in die Sprachkontaktforschung. Darmstadt: Wissenschaftliche Buchgesellschaft. Bertonio, Ludovico SJ 1612 [Reprint 1984] Vocabulario de la lengua aymara. Cochabamba: Ediciones Ceres. Brincat, Joseph, Winfried Boeder and Thomas Stolz (eds.) 2003 Purism in minor languages, endangered languages, regional languages, mixed languages. Bochum: Brockmeyer. Coseriu, Eugenio 1974 Sincronía, diacronía e historia: El problema del cambio lingüístico. Montevideo. Crystal, David 2000 Language death. Cambridge: Cambridge University Press. de Camp, David 1971 Towards a generative analysis of post-creole speech continuum. In Pidginization and creolization of languages, Dell Hymes (ed.), 349– 370. Cambridge: Cambridge University Press.

Constructivist theory of language contact

161

Fabbro, Franco 1999 The neurolinguistics of bilingualism: an introduction. London: Psychology Press. Fasoli-Wörmann, Daniela 2002 Sprachkontakt und Sprachkonflikt in Paraguay. Frankfurt a.M. et al.: Lang. Glasersfeld, Ernst von 1987 Siegener Gespräche über Radikalen Konstruktivismus. In Der Diskurs des Radikalen Konstruktivismus, Siegfried J. Schmidt (ed.), 401–440. Frankfurt a.M.: Suhrkamp. 2003 Konstruktion der Wirklichkeit und des Begriffs der Objektivität. In Einführung in den Konstruktivismus, Heinz von Foerster et al. (eds.), 9–39. München /Zürich: Piper. González Holguín, D. (SJ) 1607 [Reprint 1975] Gramática y arte nueva de la lengva general de todo el Peru, llamada lengua Qquichua, o lengua del Inca. Vaduz-Georgetown: Cabildo. Hamel, Rainer E. 1988 Sprachenkonflikt und Sprachverdrängung in der verbalen Interaktion: die zweisprachige Kommunikationspraxis der Otomí-Indianer. Frankfurt a.M.: Lang. Hekking, Ewald and Pieter Muysken 1995 Otomí y quechua: una comparación de los elementos gramaticales prestados del español. In Lenguas en contacto en Hispanoamérica: Nuevos enfoques, Klaus Zimmermann (ed.), 101–118. Frankfurt a. M./ Madrid: Vervuert/Iberoamericana. Hill, Jane 1988 Ambivalent language attitudes in modern Nahuatl. In Sociolingüística latinoamericana, Rainer E. Hamel, Yolanda Lastra and Héctor Muñoz Cruz (eds.), 77–99. México D.F.: UNAM. Hill, Jane and Kenneth Hill 1980 Mixed grammar, purist grammar, and language attitudes in modern Nahuatl. Language in Society 9: 321–348. 1986 Speaking Mexicano: dynamics of syncretic language in Central México. Tucson, Az.: University of Arizona Press. Hirsch, Joy 1997 Distinct cortical areas associated with native and second languages. Nature 388: 171–174. Holm, John 2004 Pidgin and Creole studies. In Sociolinguistics: an international handbook of the science of language and society. Vol. 1, Ulrich Ammon, Norbert Dittmar, Klaus Mattheier and Peter Trudgill (eds.), 58–67. Berlin /New York: Mouton de Gruyter.

162 Klaus Zimmermann Hudson, Richard A. 1980 Sociolinguistics. Cambridge: Cambridge University Press. Humboldt, Wilhelm von 1963 Werke in fünf Bänden.Vol. 3, edited by Andreas Flitner and Klaus Giel. Darmstadt: Wissenschaftliche Buchgesellschaft, 3th edition. Kuhn, Alwin 1951 Die Romanischen Sprachen. Bern: Francke. Labov, William 2001 Principles of linguistic change. Vol. 2. Social Factors. Malden, MA/ Oxford: Blackwell. Lambert, Wallace E. 1967 A social psychology of bilingualism. Journal of Social Issues 23: 91– 108. Maturana, Humberto R. 1980 Biology of cognition. In Autopoesis and cognition: the realization of the living, H. R. Maturana & F. J. Varela (eds.), 5–58. Dordrecht/Boston: Reidel. Melià, Bartomeu 1974 [Reprint 1982] Hacia una ‘tercera lengua’ en el Paraguay. In Sociedad y lengua: bilinguismo en el Paraguay, vol. 1, Graziella Corvalán and Germán de Granda (eds.), 107–168. Asunción: Centro de Estudios Sociológicos. Mendieta, Jerónimo de 1596 [Reprint 1870] Historia eclesiástica Indiana. Ed. by Joaquín Gracía Icazbalceta, Mexico. Mühlhäusler, Peter 1986 Pidgin and creole linguistics. Oxford: Blackwell. Muysken, Pieter 1979 La mezcla de quechua y castellano: El caso de la ‘media lengua’ en el Ecuador. Lexis 3: 41–56. Rosenblat, Ángel 1964 La hispanización de América: el castellano y las lenguas indígenas desde 1492. In Presente y futuro de la lengua española: Actas de la Asamblea de Filología del I Congreso de Instituciones Hispánicas, [without editor], 189–216. Madrid: [without publisher]. Roth, Gerhard 1987 Erkenntnis und Realität: Das reale Gehirn und seine Wirklichkeit. In Der Diskurs des Radikalen Konstruktivismus, Siegfried J. Schmidt (ed.), 229–255. Frankfurt a.M.: Suhrkamp 1996 Das Gehirn und seine Wirklichkeit: kognitive Neurobiologie und ihre philosophischen Konsequenzen. Frankfurt a.M.: Suhrkamp [5th revised edition]. 2003 Fühlen, Denken, Handeln: Wie das Gehirn unser Verhalten steuert. Frankfurt a.M.: Suhrkamp [new, completely revised edition].

Constructivist theory of language contact

163

Schmidt, Siegfried J. (ed.) 1987 Der Diskurs des Radikalen Konstruktivismus. Frankfurt a.M.: Suhrkamp. Schrader-Kniffki, Martina 2003 Spanisch-zapotekische Bitt- und Dankeshandlungen: Sprachkontakt und Höflichkeit in einer amerindischen Kultur Mexikos. Frankfurt a.M./Madrid: Vervuert/Iberoamericana. forthc. El español en contacto con el zapoteco (Oaxaca/México): Un estudio sociolingüístico. Stolz, Christel and Thomas Stolz 1995 Spanisch-amerindischer Sprachkontakt: die Hispanisierung mesoamerikanischer Komparationsstrukturen. Iberoamericana 58/59: 5–42. 1997 Universelle Hispanismen? Von Manila über Lima bis Mexiko und zurück: Muster bei der Entlehnung spanischer Funktionswörter in die indigenen Sprachen Amerikas und Austronesiens. Orbis 39: 1–77. 2001 Hispanicised comparative constructions in indigenous languages of Austronesia and the Americas. In Lo propio y lo ajeno en las lenguas austronésicas y amerindias, Klaus Zimmermann and Thomas Stolz (eds.), 35–36. Madrid: Iberoamericana/Frankfurt a.M.: Vervuert. Suárez, Jorge A. 1983 The Mesoamerican Indian Languages. Cambridge: Cambridge University Press. Sueiro Justel, Joaquín 2002 La política lingüística española en América y Filipinas (siglos XVIXIX). Lugo: Tristram Thomason, Sarah G. and Terrence Kaufman 1988 Language contact, creolization and genetic linguistics. Berkeley: The University of California Press. Weinreich, Uriel 1953 Languages in contact. New York: Linguistic Circle of New York. Zimmermann, Klaus 1986a El español de los Otomíes del Valle del Mezquital (México), un dialecto étnico. In Actas del 2° Congreso internacional sobre el español de América (Ciudad de México, 27–31 de enero de 1986, José G. Moreno de Alba (ed.), 234–240. Mexico City: UNAM. 1986b Eine Episode der Sprachplanung in Mexiko: Die ‘Comisión para la Defensa del Idioma Español’. Neue Romania 4: 105–126. 1987 Préstamos gramaticalmente relevantes del español al otomí: una aportación a la teoría del contacto entre lenguas. Anuario de Lingüística Hispánica 3: 223–253. 1992 Sprachkontakt, ethnische Identität und Identitätsbeschädigung. Aspekte der Assimilation der Otomí-Indianer an die hispanophone mexikanische Kultur. Frankfurt a.M.: Vervuert.

164 Klaus Zimmermann 1999

Política del lenguaje y planificación para los pueblos amerindios: ensayos de ecología lingüística. Frankfurt a.M./Madrid: Vervuert/ Iberoamericana. 2003 Fremdeinflüsse, Sprachpurismus und Sprachplanung in amerindischen Sprachen (am Beispiel des Otomí und des Guaraní). In Purism in minor languages, endangered languages, regional languages, mixed languages, Joseph Brincat, Winfried Boeder and Thomas Stolz (eds.), 315–347. Bochum: Brockmeyer. 2004a Die Frage der Sprache hinter dem Sprechen: Was kann die Gehirnforschung dazu beitragen? In Towards a dynamic theory of language: a festschrift for Wolfgang Wildgen on occasion of his 60th birthday, Andrea Graumann, Peter Holz and Martina Plümacher (eds.), 21–57. Bochum: Brockmeyer. 2004b Ecología lingüística y planificación lingüística. In Lluís i Vidal-Folch, Ariadna/Palacios, Azucena (eds.), 93–109. (Lenguas vivas en América Latina/ Llengües vives a l’Amèrica Llatina.) Barcelona: Institut Català de Cooperació Iberoamericana. 2005a Die Entstehung der Kreolsprachen: ein Versuch aus konstruktivistischer Sicht. In Portugiesische Kreolsprachen: Entstehung, Entwicklung, Ausbau und Verwendung, Annette Endruschat and Axel Schönberger (eds.), 11–43. Frankfurt a.M.: DEE. 2005b Interferenz, Transferenz und Sprachmischung: Prolegomena zu einer konstruktivistischen Theorie des Sprachkontaktes. In Das Zimbrische zwischen Germanisch und Romanisch, Ermenegildo Bidese, James R. Dow and Thomas Stolz (eds.), 3–23. Bochum: Brockmeyer. 2006 El problema del purismo en la modernización de las lenguas amerindias. In Los retos de la planificación del lenguaje en el siglo XXI. Roland Terborg and Laura García Landa (eds.), vol. 2, 501–524. México: Universidad Nacional Autónoma de México-Centro de Enseñanza de Lenguas Extranjeras 2008 Diglosia y otros usos diferenciados de lenguas y variedades en el México del siglo XX: entre el desplazamiento y la revitalización de las lenguas indomexicanas. In Historia sociolingüística de México, Rebeca Barriga and Pedro Martín Butrageño (eds.). Mexico: El Colegio de México (forthcoming). Zimmermann, Klaus and Thomas Stolz 2001 Procesos interculturales e hispanización en el contacto de lenguas en el Pacífico e Hispanoamérica. In Lo propio y lo ajeno en las lenguas austronésicas y amerindias, Klaus Zimmermann and Thomas Stolz (eds.), 7–13. Frankfurt a.M./Madrid: Vervuert/Iberoamericana.

Spanish meets Guaraní, Otomí and Quichua: A multilingual confrontation Dik Bakker, Jorge Gomez Rendon and Ewald Hekking

1. Introduction To date, there are around 7000 languages spoken in the world’s 200 countries. According to the survey in Thomason (2001: 250f.), by far most of these countries have selected one language for their official tongue. The reality, however is that an average of 35 languages are spoken in each of them. For some of the larger countries, the number may run well into the hundreds. Although figures for actual multilingualism are difficult to provide, for most people in the world, multilingualism is simply a fact of social and private life. Probably the most famous example from the literature is the case study by Gumperz and Wilson (1971) of the Indian village of Kupwar. It boasts four languages, two Dravidian (Kannada, Telugu) and two Indo-Aryan (Marathi, Urdu), which are all used actively in daily life by an important part of the population. As a result of this multilingual situation, all local variants of these languages have taken over structural properties from each other to the extent that a word-by-word translation between any two of the languages yields an understandable utterance. An example from Latin America is the area of the Vaupés River in the northwest Amazonian region where multilingualism is obligatory and marrying someone who belongs to the same language group is considered akin to incest. (Jackson 1974; Aikhenvald 2002) In general terms, the majority of the around fifty million strong indigenous population in Latin America are bilingual in their native language and either Spanish or Portuguese (Gordon 2005). It is no wonder that especially this area has seen a host of publications1 about both the social and linguistic aspects of bilingualism. A number of international conferences have been organized on language contact in Latin America, more recently in Amsterdam (2004) and Bremen (2005). A lot of descriptive work has been done especially in the area of loanwords, and to a much lesser extent in the area of morphosyntactic and phonological change. Research has typically been concentrated on the effects of the colonial languages Spanish

166 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking and Portuguese on a specific Amerindian language, such as Quechua (cf. Hekking and Muysken 1995) and Otomí (Hekking and Bakker 1998; 1999; 2006). Less frequently, the effects of Amerindian languages on the colonial languages have been studied, as in Haboud (1998) for Quechua and Spanish, and in studies on substrate-induced differentiation of regional varieties of Latin American Spanish (cf. Rosenblat 1964). Rarer still are studies which take a comparative, multi-contact perspective, where the colonial language is the single source language, i.e. the language from which the borrowings stem, and a number of languages in America, Asia and Europe are the target languages, i.e. the languages which borrow words and other linguistic elements. An example of this is the work of Stolz and Stolz (1996, 1997, 2001). Apart from work on the purely lexical and grammatical aspects of language contact, we find many studies which concentrate on the social aspects of bilingualism, both from the perspective of the individual speakers and the society at large. Especially the sociolinguistic situation in Paraguay, where both Spanish and Guaraní are the official languages has generated a lot of studies (cf. Corvalan and Granda 1982; Morínigo 1990). The bulk of the work in this area is couched in terms of traditional contact linguistics, with a predilection for the borrowing effects on the target (indigenous) language rather than the interference effects on the source language. The main object of interest is the lexicon. Most studies are descriptive-typological and do not work within a specific theoretical framework or from a coherent set of hypotheses concerning the phenomena to expect or not to expect. Also, it is not always clear on what data the observations are based, and what the status of the borrowings is. For instance, it is often not clear from the discussions whether the observed contact phenomena concern real borrowings or mere code switches. Often, no information is available about the level of bilingualism of the speakers who provided the data, and whether we are dealing with spontaneous spoken language or a written variety. This makes it sometimes difficult to interpret the results and compare them to results from other studies. Consequently, it is not always clear to what extent such studies may really deepen our insight into the process of language change through language contact. The current article reports on an ongoing study in language contact in the Spanish Americas, which aims at repairing several of the shortcomings mentioned above. It is organized as follows. In Section 2 we will discuss the major principles and strategies we employ, meant to repair some of the methodological pitfalls. In Section 3 we introduce some hypotheses concerning borrowing that may be derived from the principles discussed in Section 2. The effects of language contact may become visible only in a large

Spanish meets Guaraní, Otomí and Quichua

167

corpus of linguistic data from a variety of sources. The computer is an important tool for the systematic exploration of these data. Section 4 sketches a notational system and a program we developed for the processing of digitalized bilingual data. In the rest of the article, we will illustrate our methodology on the basis of three case studies in language contact, all with Spanish as the source language of the borrowing process. It concerns three typologically rather different target languages. Section 5 deals with Guaraní, a Tupi language spoken in Paraguay. Section 6 discusses contact phenomena found in different Quichua dialects of Ecuador.2 Thirdly, Section 7 studies data from Otomí, an Otomanguean language from Central Mexico. What is possible and not possible in borrowing is very much a matter of tendencies rather than absolute universals. Tendencies can only be borne out by comparison between languages. Therefore, in Section 8, we will compare the results from the three languages, and see to what extent any more general conclusions may be drawn.

2.

General outlines

As argued above, the methodology of language contact research has been left relatively underdeveloped, and mainly at the discretion of the individual researcher. Therefore, much of the work that has been done in the field so far seems to be rather incomparable. As a result, the amount of generalization over these results is rather limited and tentative. In order to help repair this situation, we have formulated several points of departure to direct our own investigations in the field. We think they are of a general nature, and may be easily transferred to other language contact situations.

2.1. Borrowing versus shift-induced interference First, and foremost, this study is about borrowing as opposed to interference caused by shift to another language, in the sense of Thomason (2001: 66f.). This means that we are interested only in cases where the source of some element is more or less clearly a second language for the speakers whose first language we are studying. Cases of imperfect learning will be left out of consideration as much as possible. Of course, the notion of imperfect learning is not unproblematic in itself. Any version of a target language containing some clear instance of borrowing from some source language not present in the variety of the earlier generations of speakers may be seen as

168 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking the result of imperfect learning on the side of the speaker in question. This is especially the case if the element borrowed replaces an existing element of the target language, and it may be maintained that the new variety is no longer ‘pure’. The distinction between borrowing and interference is crucial, however, since the predictions for the two phenomena are known to be quite different in terms of what may be found in the target language and what not, and the order in which changes may take place. We have tried to cater for the problems involved by requiring that the Amerindian language serves as the dominant instrument for everyday interaction in the speech community under investigation. For the individual speakers who provide our data we require that the Amerindian language is their first language, and that they use it on an everyday basis. In this way we hope to exclude cases of interference, which typically will also be attested in the same communities. 2.2. One source language Within this study, the source language of the borrowing process is kept constant, in our case Spanish. By taking this language as the only one from which borrowed material is studied, there is as little linguistic variation on the input side as possible. Arguably, for the structural domains of grammar (morphology and syntax) variation between the several varieties of Spanish in Latin America is relatively modest, or even insignificant. As for more superficial variation, i.e. lexical and phonological, potential dialectal differences in the areal varieties of Spanish are controlled by not taking some standard version of Spanish as a point of departure, but rather the Spanish as spoken by the monolinguals in the respective areas. In that sense we fully agree with van Hout and Muysken (1994), inasmuch as we also include the Spanish of the bilingual speakers of the area whose first language is the Amerindian language under consideration. Our assumption is, however, that there are no fundamental typological differences between the respective varieties of Spanish that would have substantial bearing on the comparison of the respective loan processes. 2.3. Typological variety among target languages In contrast to the lack of typological variety in the source material of the borrowing process, the target languages should be selected such that, mutually, they manifest considerable typological distance. Only under such con-

Spanish meets Guaraní, Otomí and Quichua

169

ditions typological conclusions of some depth and width concerning the borrowing process may be drawn. This may be done by comparing the differences between the respective sets of borrowed features, to the extent that these are manifest in the first place. The three languages chosen for this article, i.e. Guaraní, Otomí and Quechua seem to fulfil this typological requirement, as will be argued in the respective sections below.3 2.4. Culture kept constant In most language contact studies it is accepted that cultural aspects play an important role in what will and what will not be borrowed. Some authors even assume that ‘anything goes’ provided that the cultural pressure is strong enough (cf. Thomason 2001). Thus, in order to make the results of two cases of language contact comparable, cultural parameters should be controlled as much as possible. There is no way to do this in any type of realistic experimental setting. By concentrating on Spanish America, we think that a certain amount of cultural unity however is guaranteed. Of course, there are vast differences between the many indigenous cultures of America, and the ways in which they interacted and still interact today with the Spanish-speaking world. These will undoubtedly be reflected in the specific loanwords that may be found as such, and the semantic fields from which they stem. However, with the areal restriction we take into consideration for our data collection, we think that the influence of culture is minimalized. As a result, the differences that we may find between the respective Amerindian languages with respect to borrowing from Spanish may be ascribed to a large extent to typological differences between the borrowing languages. 2.5. Target language: spoken data from a representative group Language contact is not the contact between two abstract entities such as languages. It is a dynamic process that takes place in the interaction between the speakers of two language communities whose lects are sufficiently different to create a communicative gap. In order for such a situation to arise, the communities involved have to be exocentric in terms of Andersen (1988), i.e. a reasonable proportion of the speakers of one community has a regular contact with those of the other community, and some become more or less bilingual in the process. For a variety of reasons, these bilingual speakers will introduce elements of their second language into their first. Over time,

170 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking and to the extent that these innovators have a certain amount of authority inside and outside their community, monolingual speakers will copy some of the innovations which will then propagate throughout the community, including learners. This process may or may not be speeded up by an increase in bilingualism within the community both in breadth and depth. As a result, new varieties of the local (target) language may arise, which include aspects of the external (source) language. Apart from so-called catastrophic changes (Ross 2003: 177f.), when sudden events, such as massive deportation, lead to a complete restructuring of a speech community and its language, language change is a gradual and very complex process, often stretching over several generations. Therefore, the best way to study the impact of language contact on the languages involved is not by making two snapshots of the target language at two stages far enough apart to reveal obvious differences. Rather, it should be analyzed on the basis of data produced by a diverse group of individual speakers who find themselves at different stages of the borrowing process.4 The closest one can get to observing such a process is through longitudinal sampling and the in-depth study of stretches of discourse between a variety of relevant speakers, some reappearing in later samples and some new. Since such a complex setup is typically out of the question in most cases, next best is the construction of a corpus of spontaneously spoken samples of the target language by a range of speakers from different age groups, thus simulating time to some extent. Furthermore, the set of speakers should be stratified as best as possible according to a number of sociolinguistically relevant factors, such as gender, level of education, profession, and contact with the source language community, among others. A sample that is minimally representative in terms of such parameters would amount to between 50 and 80 people. With the typical average text length of around 1000 tokens this would give us a minimum corpus size of 50,000–80,000 tokens. Consistent with this view, when we use metaphors such as ‘language T borrows element E from language S’ this should be read as ‘a significant and increasing number of speakers of T uses E, which is originally an element of S only’. E could be a lexical item, a bound morpheme, a specific construction, and so forth. As in most of the literature on borrowing we will use the notions ‘significant’ and ‘increasing’ in a rather impressionistic way here. They are in need, however, of a proper definition.

Spanish meets Guaraní, Otomí and Quichua

171

2.6. Source language: from the community itself Although today the role of external sources – television, film, internet, written press – on language is considerable, we think that in most cases the language used in the local communities themselves is the greatest determining factor for language change. Therefore, we collected a corpus of spoken Spanish from the same communities from which we collected the target language samples. Part of the data has been collected from monolingual speakers of Spanish, typically mestizos. However, we also asked some of the native speakers who provided us with the data to contribute a spoken text in Spanish as well. This gives us an instrument to measure their level of bilingualism. Thus, we can relate the proficiency in the mother tongue and the amount of borrowing from Spanish to the proficiency in Spanish, and the diversions from the norm found in it. 2.7. Diachrony Language change as a result of contact may be a matter of only a short period of time. Media Lengua (Muysken 1994; Gómez Rendón 2005), a mix of Spanish and Quichua, is an example of a contact language that probably has been created in the span of only two or three generations. In that sense, a well chosen sample which adheres to our sociolinguistic criteria under 2.5 would be good enough to trace rapid changes across generations. However, many other instances of contact-induced language change take more time. Therefore, we think that an adequate corpus should also contain samples from earlier stages of the target language, preferably also from before the contact era. In most cases, such data is not available, not even in the form of written varieties. In the case of Latin America, however, there exist secondary sources on some of the indigenous languages, such as dictionaries and grammars written by missionaries (cf. Urbano 1990 [1605]; Ruiz de Montoya 1993 [1640]). When available, we try to employ such sources by lack of primary ones, assuming that they may be considered reliable. 2.8. Regularities and hypotheses concerning borrowing Not much theory has been developed to date with respect to what may be borrowed and what not. Moravcsik (1978) is a classical proposal with respect to the universals of borrowing. However, most of the universals suggested by her, and others found elsewhere in the literature are shown by

172 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking Campbell (1989) to have counterexamples. This seems to suggest that such universals may be tendencies at best. A second point of orientation is the borrowing scale proposed by Thomason and Kaufmann (1988: 74f.), and its revised version in Thomason (2001: 70f.). This scale measures the extent to which a language has changed as a result of borrowing from another language on the basis of the types of lexical and grammatical elements that are supposed to be or not be borrowed at different stages. Also this scale is not absolute. It provides tendencies which may be influenced by typological differences and which may be overruled by social factors. With the necessary caution, however, some predictions may be derived from this scale as to what types may and may not be found among the borrowed elements in a language. This makes the scale falsifiable, with the proviso of the overriding social and typological factors. Thirdly, language typology makes certain predictions as to which combinations of features may and may not be found in languages. These so-called universals of language often go back to the original work of Greenberg (1963) on implications between formal features in syntax, morphology and phonology. Most of them were presented as tendencies from the start, have been shown to be not absolute since, or were refined by authors such as Hawkins (1983) in order to make them more reliable. If we take them as more than contingencies, and assume that they represent strong tendencies towards what may or may not be found in a grammar, then certain predictions may be derived from them as far as borrowing is concerned. A final constraint on borrowing, provided by the field of language acquisition, is one on markedness: elements are less likely to be borrowed to the extent that they are marked. This is typically to be taken in the literal sense of (morphological) complexity rather than the metaphorical sense of ‘remarkable, unlikely’. Also this claim may be falsified, at least in principle.

2.9. Explanation: the role of theories Although linguistic theories typically aim at describing linguistic competence (and performance) in the light of the grammar of the first language only, and are not about borrowing, they should be able to make predictions about the nature of the language borrowing process, at least in principle. For syntactic theories of the generative type, borrowing, both by adult speakers and first language learners, appears to be a real challenge with respect to universal grammar, especially as far as it concerns the structural aspects of the source and the target language. For cognitively and function-

Spanish meets Guaraní, Otomí and Quichua

173

ally oriented approaches to grammar, such as Construction Grammar (cf. Goldberg 1995; Croft 2001) and Functional Grammar (Dik 1997) borrowing seems to be less of a problem vis-à-vis the major principles. According to such theories, constraints that might be predicted for borrowing at the formal and semantic levels might be overruled by pragmatic or social factors, rendering any absolute prediction untenable. Nevertheless, also in these cases, predictions might be derived, be it that they concern the relative rather than the absolute amount of (several types of) borrowing. In that respect, they more or less naturally coincide with the Thomason scale and with predictions from language typology. More specifically, theories on parts of speech provide a framework from which testable predictions about borrowing may be derived. We will discuss the one introduced in Hengeveld (1993), which is syntactically oriented and embedded in Functional Grammar. As another example of a testable framework, Myers-Scotton’s (2002) Matrix Language Theory makes predictions about the types of elements one may find in code switches but also in borrowing and other scenarios of language change. Although the current study is about borrowing and not about code switching, the borderline between the two is not always clear. It could be claimed that definitive borrowing of some element is often preceded by its incidental use in code switches. Studying the correspondences and differences between borrowing and code switching may in fact lead to better definitions of the two related phenomena. Finally, language change takes place as a result of partial or full bilingualism of the speakers of the languages concerned. Studies in language contact should, therefore, take heed of what is known about first and second language acquisition, and of what is known about the bilingual mind. Although not presented explicitly as a theory on borrowing and shift, Thomason (2001: 129f.) presents seven what she calls mechanisms according to which contact-induced language change operates. She mentions passive familiarity, first and second language acquisition, code alternation and code switch, negotiation and deliberate decisions. More specific mechanisms mentioned are correspondence rules (phenomena from the source language are projected onto existing phenomena in the target language in a one to one fashion) and simplification (distinctions made in the source language are ignored when a form is introduced in the target language). These are all linguistic environments or practices enhancing language change, and which may throw a light on what kind of changes may take place as a result of language contact and with what probabilities. They are definitely not motivations for change. In principal, all motivating factors are assumed by

174 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking Thomason to be social and non-linguistic. However, a few changes are mentioned which might be interpreted as motivated by purely grammatical reasons.5 As an example may serve the borrowing of the English conjunction and by bilingual speakers of Maori. Arguably, its use leads to less complex structures than the Maori system which is based on the preposition me ‘with’. The latter form also seems to undergo reanalysis leading to its use as a conjunction (Thomason 2001: 136). In typological work it is sometimes argued that a language has copied a specific grammatical strategy or feature from another language. Evidence is typically based on the fact that languages genetically related to the target language do not have such a feature nor is there any indication of internal diachronic development. Comrie (1990) gives an example from Turkish, an Altaic language with mainly infinite subordinate structures, which has borrowed the conjunction ki from Persian and uses it to form finite subordinate clauses such as the complement clause in (1) below. (1)

Turkish (Comrie 1990: 203)6 herkes bilir ki dünya yuvarlak-tir everyone know CONJ earth round-COP ‘Everyone knows that the earth is round.’

Another example stems from Estonian. Unlike related Finnish, this language has a passive construction that is arguably borrowed from Russian or German through contact. As a final example, according to Dench (1998: 37) person clitics in the Pama Nyungan languages Yingkarta and Wajarri are an innovation with respect to the genetically related Australian languages from the north and may be borrowed from unrelated contact languages in the east. Some take this as an example of a purely formal reason for borrowing, i.e. the optimalization of the grammatical system. In earlier work (Hekking and Bakker 1998; 1999; 2006) some of the principles of borrowing alluded to above were tested on the basis of part of the data then available for Otomí. In Section 3 they will be reformulated and extended. 3. Hypotheses on borrowing Here we will put forward some hypotheses on borrowing which are in part inspired by work in functional theories of grammar and in part by more general work on contact induced change.

Spanish meets Guaraní, Otomí and Quichua

175

A central issue in the study of language change in general is the extent to which changes are internally motivated or caused by external factors. On the basis of de Saussure’s (1916) distinction between langue (the ‘inner’ language, the system) and parole (the ‘outer’ language, its use), linguistic theory has sought to base explanations on the former rather than the latter, in the case of Generative Grammar even exclusively. Functionally oriented theories, such as Functional Grammar have steered away from this tendency, and placed more emphasis on language use and the role of discourse. We will follow the latter route when looking for factors that might explain the change of languages in the presence of other languages. The functionally oriented hierarchy under (2), proposed earlier in Bakker and Hekking (1999) will serve as a basis. (2)

Social > Communicative > Cognitive > Formal factors

We may assume that a language that is functioning satisfactorily as a means of communication in some community for a considerable period of time, is a complete system, with its own mechanisms of adaptation, variation and change. Therefore, there seems to be no inherent reason for it to copy any aspect, either form or structure of another language with which it gets into contact at a certain stage. However, languages do, and the most fundamental reason for this seems to be the desire of the speakers of both languages for mutual understanding, be it that this desire is not necessarily the same for both groups, and their individual members. In catastrophic cases, when contact between two language communities increases in a gradual fashion, and the two communities may merge into one, there will arise a certain amount of (partial or full) bilingualism among the speakers involved. This process will be more or less bidirectional in case the position of the original communities is more or less equal. This may lead to the development of a Sprachbund, where the languages share a restricted set of features, but remain mutually non-understandable and do not change typologically in any fundamental way. Or we may witness complete bilingualism (or multilingualism) for a large group of speakers, as in the examples of Kupwar and Vaupés mentioned in the introduction, in the long run with rather dramatic results for the languages involved. However, if one community has the edge over the other, economically, socially, politically, as is the case of the languages studied here, bilingualism will generally be mainly unidirectional, and affect only one language – the indigenous one – in a serious way. The results may run from the borrowing of a restricted number of lexical items from specific semantic domains to massive lexical and grammatical borrow-

176 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking ing, typological change, relexification, language shift, loss and finally language death.7 With social factors forming the basis of language contact, and communication and mutual understanding being the major goals of the speakers concerned, we may reformulate the hierarchy of (2) as in (3), where we have translated the more general terminology into the corresponding components of the grammar. (3)

Pragmatics > Semantics > [Syntax-Morphology-Phonology]

Pragmatics covers the communicative aspects of language, such as the types of speech act and finer tuned discourse markers. Semantics deals with the content of the message speakers want to get across. It is precisely with respect to these two central, somewhat more abstract aspects that are crucial for successful communication that languages overlap most. This leads us to the expectation that much of the ‘early’ borrowing must be found precisely in these two areas of linguistic description. Individual lexical items that fill obvious semantic gaps are easiest to borrow since they enter an open class, for which extension is a routine process in any language, also for speakers above the language acquisition age. Secondly, we expect that languages may quite easily borrow certain discourse markers that are characteristic of the source language, such as connectors, pause fillers etcetera. In the latter case, gap filling is not a very likely motivation in the strictly functional sense. Rather, the use of such often highly frequent and syntactically peripheral and therefore outstanding markers give the target language some of the flavour of the prestige language, and suggests bilingualism.8 At the right hand side of the hierarchy in (3), in brackets, we find the formal components of grammar. In our view of the language system, these components form the (dependent) vehicle of communication rather than its (independent) content. With few exceptions, formal aspects of the source language will typically not be the object of borrowing as such but will rather be potential obstacles to it. Within the formal component of the grammar we distinguish a further subhierarchy going from syntax via morphology to phonology. This subhierarchy runs more or less parallel to the degree to which the functions of the respective subcomponents of the grammar are transparent to the (partially) bilingual speaker of the target language. For instance, we expect that a syntactic feature such as constituent order will turn out to be more transparent and therefore easier to copy than morphological features such as plural or agreement marking. With respect to the phonological component, suprasegmental aspects will be more

Spanish meets Guaraní, Otomí and Quichua

177

accessible and therefore borrowable than individual phonemes, which often remain among the last ‘fossils’ even for semi-bilinguals. The motivation-versus-constraint interpretation of the hierarchy in (3) operates roughly as follows. Bilingual speakers may be motivated to express a certain functional (i.e. pragmatic or semantic) feature F of source language S, represented by a formal element ES (a morpheme, a word, a construction) in an utterance of target language T. Typically, F is not the only function of ES, but the most outstanding in the bilingual situation. It may even get an interpretation which is not necessarily the one given to it by native speakers of the source language. If the target language T lacks F, the result is an enrichment of the utterance and, if copied by others, eventually also of the lexicon or the grammar of T as such. If T already possesses an element expressing F, then the result of its use may be a clearer way of expressing that function, because of its unique interpretation, or because element E stands out more than the original form which codes F. Borrowing is easier to the extent that F is a real functional extension of the target language and ES fits easier into the grammatical structure of T. The more similar both languages are in a typological sense, the higher the probability that ES will find its way into an utterance of T, and the grammatical system of T. This typological similarity between S and T should be taken not only in the obvious synchronic sense but also in the diachronic sense of T being in a state where it might develop an ES-like element internally.9 The above is resumed in the borrowing hypothesis below. This hypothesis is restricted to non-catastrophic, unbalanced language contact situations with a more or less clear distinction between a source and a target language. Borrowing Hypothesis: H0. Any element of source language S may be borrowed by target language T. The probability of some element ES of S to be borrowed by T is dependent on the following linguistic factors: a) its place on hierarchy (3) b) its frequency of use in S, and later T c) the typological distance between S and T on the relevant parameter(s) d) the (maximum) level of bilingualism in the speech community From this general hypothesis, several more specific hypotheses may be derived. Note that, since they are framed in terms of probabilities rather than absolute yes/no’s, they should be tested by comparing languages of both the same and of a different type. Below, the > sign should be read as ‘easier to borrow than’.

178 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking H1. pragmatically outstanding elements > non-outstanding elements H1.1 discourse marker > topic marker H2. lexical elements > grammatical elements H2.1 open class > half open class > closed class H2.1.1 Noun > Verb > Adjective > Adverb > Adposition H2.1.2 Adposition > … > Auxiliary > … > Article H2.2 free form > clitic > bound form H2.2.1 Adposition > Case affix H3. syntactically simple elements > structures H3.1 free lexical element > fixed collocation > noun phrase > construction As for the status of lexical items, we follow the view of Functional Grammar which acknowledges only nouns, verbs, adjectives and adverbs as part of the lexicon, with their own concrete lexical meaning. All other elements have abstract meanings and are assumed to be part of the grammar. However, we consider adpositions as an intermediate category in the sense that they may be either lexical or grammatical, depending on their status in the relevant grammar.10 As for the typological aspects of the parts of speech a number of proposals have been made in the literature. Some authors take a fundamentally cognitive-semantic position towards their definition. A clear example of this is Croft (2001). On the other end of the scale we find those who define the parts of speech on purely formal grounds, i.e. their role in morphosyntax. Hengeveld (1993) takes this position. Yet others, such as Evans and Osada (2005), take an intermediate stance. In this contribution we will put to the test only one approach, viz. Hengeveld (1993), since some tentative predications may be derived from it with relation to borrowing. Hengeveld’s theory only considers the major parts of speech, nouns (N), verbs (V), adjectives (A) and manner adverbs (MAdv). Nothing is said about other adverbs, nor about grammatical elements. The four types distinguished are defined on the basis of their prototypical syntactic behaviour. Thus, V is the head of a predicate phrase, N the head of a referential phrase, while Adv and MAdv typically take the modifying position in predicate and referential phrases, respectively. Languages are ordered on a seven-point scale running from maximum flexibility (type 1) to maximum rigidity (type 7). The 7 language types are defined as in Table 1 below.

Spanish meets Guaraní, Otomí and Quichua

179

Table 1. Typology of parts of speech systems (Hengeveld 1993) Type

Lexical Categories

1 2 3 4 5 6 7

Contentive V, Non-verb V, N, Modifier V, N, A, MAdv V, N, A V, N V

Type 1 has only one overarching part of speech, which covers the whole range from V to MAdv. In other words, many forms may be used in any of the four syntactic positions, without any extra marking that is indicative of their derivational status. Type 2 languages have an independent category V and a generalized category for the rest. Type 4 shows maximum diversification: all four basic lexical types are attested. Type 5 lacks basic manner adverbs, and type 6 basic adjectives and adverbs. In terms of this typology, Spanish is a type-4 language, i.e. its lexical entities typically specialize for one of the four parts of speech: e.g. its nouns cannot be used as modifiers, as shown in (4a/b); and virtually all adjectives need morphological adaptation in order to be used as adverbs (5). (4)

a. *una casa piedra ART house stone ‘a house of stone’

b. una casa de piedra ART house of tone ‘a house of stone’

(5)

di-me la verdad sinceramente/*sincero honestly/honest tell-1SG.DAT ART truth ‘Tell me the truth honestly’

On the basis of this theory, the following general predictions seem to be warranted. We expect a target language T of type 1–4, which has all four syntactic positions available, to borrow all four types N, V, A and MAdv from a source language S without much constraint. When T is more flexible than S, there are two possibilities: functional adaptation or functional specialization. According to the first, more liberal hypothesis, borrowed elements will be treated as if they belonged to the lexicon of T: e.g., in a type2 language, borrowed adjectives from a type 3–5 language may be used as heads of referential phrases apart from being used as modifiers. According

180 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking to the second, less liberal hypothesis, borrowed elements will figure only in their original function. If T is less flexible than S then we only expect specialization among the borrowed elements in the relevant area, e.g. in a type-3 language some [Verb, Non-verb] elements borrowed from a type-2 source language may be used exclusively as heads and others exclusively as modifiers of referential phrases. On the rigid side of the scale, i.e. T languages of types 5–7, we expect to find very low numbers of elements from an S language which have an ‘unknown’ part of speech, and specialization for elements which are borrowed into one of their original classes. E.g. a type-6 language will in principle not borrow a [A, MAdv] element from a type-3 language, and it will borrow [V, Non-verb] elements from a type-2 language only in the function of heads of predicate and referential phrases, respectively. All these predictions may be seen as further specifications of hypothesis H2.1.1. There is one striking difference between H2.1.1 on the one hand and Hengeveld’s typology on the other hand, i.e. the order of V and N on the hierarchy. Nouns being the most open class, we predict that they will be the first type to be borrowed, and that they will also have the highest frequency among the parts of speech in the list of borrowings. Although no prediction concerning relative frequencies might be derived from Hengeveld’s typology, it seems to suggest a very central position for verbs. They are a class of their own in all but type-1 languages. Apart from parts-of-speech oriented hypotheses, we will also have a look at some of the more structural aspects of language contact and change, as studied by language typology.11 Some of the well-established parameters that have been discussed in the typological literature over the last 40 years seem to be relatively deeply entrenched in the grammatical system of a language, and take a long time to change, typically many hundreds of years. This leads us to the following general hypothesis on borrowing. H4. There is a positive correlation between the time a typological parameter takes to change under neutral circumstances, i.e. without a strong external pressure, and the time it takes for it to change in a bilingual situation. From this, the following more specific hypotheses may be derived. H4.1 Borrowed elements fall in line with T morpho-syntax, and are easier borrowed when their basic syntactic position in terms of Head-Modifier relations in T is the same as in S. E.g. adpositions will be borrowed in their original syntactic position, and only if it is available in the T syntax. E.g. a postpositional T language may borrow postpositions from S but no prepositions.

Spanish meets Guaraní, Otomí and Quichua

181

H4.2 The frequency of the existing constituent order patterns in T may change in the direction of orders frequently attested in S. However, this will typically not lead to the introduction of new orders, not attested before the contact period. E.g. a VSO language in contact with a SVO language may change its basic order to SVO given that this order is typically among the alternative orders anyway. It will not easily adopt SOV, which is rather rare as an alternative in VSO languages.12 H4.3 Languages may borrow elements which express some already existing function more analytically than the original strategy. E.g. when a language marks a possessive relation with case marking, it may borrow an adposition expressing possession in S. In actual utterances, both strategies may be combined leading to so called ‘doubling’. The opposite, i.e. borrow a more grammaticalized form, is unlikely to happen. These hypotheses will be tested in the Sections 5 through 7, where we discuss the data we collected for the three languages which concern us here. First, in Section 4, we will briefly discuss some of the tools we developed for analyzing the data. 4. Notational system and computational tools As argued above, the empirical basis for any research into contact-related language change should be a relatively large body of spoken language, typically tens of thousands of tokens per language pair, produced by a number of different native speakers. The data are typically collected in recorded fieldwork sessions, by inviting subjects to tell a story either on a freely chosen or a specific topic. The latter can be directed by pictures, as the well-known Pear and Bee stories.13 The recordings are then transcribed and put into computer-readable form. We chose a solution where the transcriptions are only very partially phonological. For the source language – always Spanish – only the most striking differences from the standard spelling are coded in the protocols. Some examples may be found in (6) below, stemming from speakers of Otomí. (6)

a. albañil ‘bricklayer’ albañi albañil albeni albini

182 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking b. vecino ‘neighbour’

besino besinu bisino bisinu

In general, the Spanish tokens are rendered in standard spelling. For the native tokens, a standardized spelling per language is used throughout. For the languages relevant for this article, more details on the collection and the coding of the data will be given in the respective sections below. The recorded texts are entered into the computer via a standard word processor. The structure of the resulting file is as follows. IDENT OF INFORMANT: AGE: LOCATION: PROFESSION: EDUCATION: < further social parameters > TARGET LANGUAGE: SOURCE LANGUAGE: RECORDED BY: TOPIC: $ < text >

Figure 1. Structure of data file

In the actual text, the loanwords are marked by slashes, as in example (7a) from Otomí, glossed in (7b). (7)

a. bí 'mui jar /skwela/ /entre/ 'naha ne hñu ya jeya. b. bí

'mui j-ar skwela entre 'naha ne hñu be LOC-DEF.SG school between one and three ya jeya. DEF.PL year ‘I went to school for between one and three years’ PST3

On the basis of a text thus coded, a computer program compiles frequency tables of all types – i.e. different forms – found for the target language and the source language. Figure 2 is part of the table compiled for the Otomí informant known under the initials AEL, who provided us with a text with a length of 267 tokens, 16 of which were Spanish words of 11 different types. The table corresponds to the top of the alphabetically ordered list of types per language.14

Spanish meets Guaraní, Otomí and Quichua

183

Frequencies for informant: AEL TARGET (Otomí) ar 7 bi 2 1 bongu boni 1 2 boxu bí 31 bötsi 13 da 2 ... Types 71 (86.59%) Tokens 251 (94.01%) TTR 3.54

SOURCE (Spanish) bentanä 1 boi 2 despwes 1 frasko 1 i 2 kong 2 nsusya 1 panal 3

informants (35) 26

11(13.41%) 16(5.99%) 1.45

7 21 3 82 267 3.26

Figure 2. Lexical overview for one informant (fragment)

For both languages, the type-token ratios (TTR) are given.15 All borrowed types found in the input text are incrementally added to a dictionary, which also keeps track of the overall number of informants that have used a certain type. These totals are given for each of the forms in the list at the righthand side. So, of the 35 informants in the corresponding corpus, 26 informants have uttered the form bentana one or more times. Apart from containing the overall borrowing lexicon, the dictionary file may be used in several different ways. Firstly, any unmarked form found in a new text that matches a dictionary type may be automatically assigned slashes, marking it as a loanword.16 Secondly, it may be used to solve both problems created by alternative spellings of source tokens and for lemmatizing inflected forms. The dictionary should then be prespecified as in (8). (8)

escuela escuelas > escuela skwela > escuela

If these entries are present in the dictionary, then the forms escuelas (plural of Spanish escuela ‘school’) and skwela (Otomí pronunciation of escuela) will be assigned to the dictionary type escuela. Optionally, the input text may be enriched with parts of speech information for the source language tokens. This is done by extending the form with a one or two letter code for the part of speech in Spanish, plus a code or the syntactic position in which it is found in the target language.17 Thus enriched, the input sentence of (7) may look as follows. (7’) bí 'mui jar /skwelaNHR/ /entrePRL/ 'naha ne hñu ya jeya.

184 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking Thus, Spanish skwela ‘school’ is coded as a noun (N) and is functioning as a head of a referential phrase (HR) in this Otomí text. And the Spanish preposition (P) entre ‘between’ is coded as a syntactic relator (RL) in Otomí. Such codes are also represented in the lexicon, and may therefore be assigned automatically by default, i.e. unless explicit coding in the text overrides such a default. A full-blown lexical entry for the form skwela may therefore look as in (8’) below. Such an entry provides the (canonical) form, the part of speech in Spanish (here N, i.e. noun) and the syntactic positions in which it has been found in the text (here both head and modifier of a referential phrase). (8’) escuela N (HR, MR) The program produces totals per part of speech, both for individual texts and for the whole corpus of texts for a particular target language. The following figure gives a fragment of a part of speech survey for the Spanish loan words found in the complete Otomí corpus. PoS ADJ ADJ ADJ ADJ ADJ

SFnc -

TYP

TOK

MR 52 HP 13 HP MR 9 HP MP MR 1 TOTAL 75

188 23 52 4 257

Figure 3. Total for parts of speech (fragment)

Thus, of the 75 Spanish adjectives found in the collective Otomí texts, 52 were used in their prototypical function of modifiers of the head noun of a noun phrase (MR); 13 were used only as the main predicate of a sentence (head of a predicate phrase, HP); 9 were found both as main predicate and modifier of a head noun; and one was found to perform three different functions. The coding system makes it possible to isolate borrowed lexical elements and bound morphemes in the text. There is also a notational convention for isolating code switches. Any stretch that is analyzed as such may be included in [ ]. By default, these are excluded from the analysis. Optionally, however, they may be added to the totals of the source language. In the analyses below, the code switches are left out of the consideration. So far for the tools we developed. We will now look at the actual data that we collected and analyzed for our three languages. We will start with Guaraní, then discuss Quichua and finally Otomí.

Spanish meets Guaraní, Otomí and Quichua

5.

185

The case of Guaraní

Guaraní is a Tupi language spoken by 3,946,904 (2002) people in Paraguay. Most speakers of Guaraní speak also Spanish with different levels of proficiency. Although Paraguayans take pride in being the only bilingual country in Latin America, bilingualism in Paraguay is neither stable nor symmetrical and the Indian language carries numberless traces of its century-long struggle with Spanish. On the basis of the investigation of a large corpus of spontaneous speech collected during a three-month period of fieldwork in Paraguay (2004–2005), this section deals with Spanish lexical borrowing in Paraguayan Guaraní (PG). We will first make the reader familiar with some historical and sociolinguistic issues of Guaraní-Spanish contact in Paraguay. The second step is to address the typological classification of PG according to parts of speech and other parameters. In the third section we identify the expected outcomes of Spanish borrowing in PG according to its typological features. Finally we will present the statistical results concerning the borrowing of lexical classes and analyze them in the light of the hypotheses described in Section 3. Some conclusions are presented in the last section. 5.1. Historic context of the contact between Paraguayan Guaraní and Spanish Juan Díaz de Solís (1516) and Alejo García (1524) explored the territory of present-day Paraguay as part of their individual – equally unsuccessful – enterprises of finding easier routes to greater riches. Paraguay itself, however, did not attract the newcomers as other densely populated and economically promising areas of the new empire did. A scattered native population did not offer any important workforce to the Conquerors.18 Therefore, comparatively few Spaniards settled in the area and were easily outnumbered by the Indian population. Thirty-eight years after its foundation in 1536, Asunción counted one Spaniard on every ten Indians. In the context of this demographic unbalance, intermarriage became the best strategy for Spaniards and Indians to create long-lasting bonds that encourage a peaceful coexistence. Spaniards used to marry several Indian women at a time and polygamy became a common practice in Paraguay. The resulting racial mixture spread rapidly. Over the years Guaraní speakers – mainly mestizos – absorbed most Spanish enclaves in the area.19 Though Spanish continued to be used for all official transactions, Crown officials protested that Guaraní was displacing Spanish in the area to the point that

186 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking even the few remaining descendants of Spaniards preferred to speak the native language with each other (cf. Morínigo 1982). But however limited its use, Spanish continued to be associated with the elite’s political power throughout the history of Paraguay. Concurrently with these developments, the Guaraní peoples who had remained outside the Spanish influence were object of individual evangelization enterprises by Franciscan and Jesuit missionaries. In particular, Jesuits (1609–1768) developed a unique social organization in their so-called reducciones where Indians were encouraged to use only their native language for oral and written communication in their daily lives. Differences between the standardized Guaraní spoken by the Indians in the reducciones and the Guaraní spoken by the mestizos and Spanish settlers in the towns began to emerge. After the Jesuit missionaries were expelled from Spanish America in 1768, the Indians of the thirty existing reducciones either fled to live in the wilderness or integrated into the mainstream mestizo society and thus made their contribution to the effervescent language pool of the Paraguayan society. The specific relations and influences between the Guaraní developed in the Jesuit reducciones and the Guaraní spoken in the urban areas remain unknown to date (Dietrich 1995: 204). Over time, the Guaraní dialect spoken in urban areas became strongly influenced by Spanish both in the lexicon and the grammar. Although many advocates of purism consider this Guaraní (also called Jopará ‘mixture’) a mere corruption, it has extended to all sectors of the Paraguayan society and a number of literary pieces have been written in it since the second half of the nineteenth century.20 The debate about the status of this sometimescalled ‘third language’ is far from finished. For the purpose of this article, we consider Jopará and Paraguayan Guaraní as synonyms. When independence from Spain was declared in 1811, Paraguay was an isolated district of the Viceroyalty of Rio de la Plata. However, the new establishment did not imply a new status for Guaraní. Throughout their independent history Paraguayans have shown ambivalent attitudes towards their language: they hold Guaraní in high esteem as symbol of their heritage and past but at the same time associate the language to backwardness and primitivism. This split attitude helps explain why political leaders often took contradictory stands and made divergent decisions on the issue of language use. Gaspar Rodriguez de Francia, who ruled Paraguay from 1814 until his death in 1840, promoted monolingual Spanish schooling but used Guaraní for all administrative and political issues. His successor Carlos Antonio López, a European-grown progressist and blatant detractor of Guaraní, launched a

Spanish meets Guaraní, Otomí and Quichua

187

campaign in 1848 to replace all native family names with Spanish names. With López’ death in 1862, his son Francisco Solano López became president of Paraguay and led his country to the bloody war with Argentina, Brazil and Uruguay. However tragic the war was in demographic and economic terms, it helped promote Guaraní as the highest landmark of Paraguayan identity. Guaraní was used from the battlefield to the trenches either as a secret code or as the language of folk songs. Francisco López realized the agglutinating potential of Guaraní and organized during the war a Congreso de Grafía (i.e. a spelling congress) in which new orthographic rules were set for the language. The six-year war left a decimated nation at the mercy of the winning powers. And the winners were not interested in preserving any trace whatsoever of the Guaraní culture. A new puppet regime was appointed on 15 August 1869 with a mass celebrated in the Cathedral by the General Vicar of the Argentinean army, who preached about “the need to regenerate the Paraguayan people in order to promote their development” and the need to dispel Guaraní from their territory for being “a dreadful creation of ignorance and backwardness” (Trinidad Sanabria 1997). Shortly afterwards, the Argentinean political writer and activist Domingo F. Sarmiento was appointed to conduct an educational reform in Paraguay along the lines of his ideological tenant of the new American civilization, in which a “savage language” had no place (Trinidad Sanabria 1997). The linguistic policy of favouring Spanish monolingualism at the cost of the native language was implemented in one way or another for the next half century. Again, in the 1930’s the tragic developments of war showed Guaraní as a centripetal force. The Chaco War between Paraguay and Bolivia motivated the recognition of Guaraní at all spheres of the administration. Politicians, military and religious leaders gave their speeches and harangues in Guaraní while poets composed popular songs in the language. And yet, the liberal, revolutionary and dictatorial administrations that followed did little or nothing to give official status to Guaraní. The first winds of change came on 15 August 1967 when a new Constitution granted Guaraní the status of ‘national language’. But the new status did not have any practical consequences during the long dictatorship of Alfredo Stroessner. After Stroessner’s downfall in 1989 and the establishment by the 1992 Constitution of the official status of Guaraní vis-à-vis Spanish, the Ministry of Education began to implement a model of Bilingual Bicultural Education according to which every child must be taught in his/her own mother tongue. This implementation has not escaped controversy. For one group of the population the bilingual program is perpetuating the same

188 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking structures of oppression on the native language by giving too much space to Spanish borrowings. For others the histories of Guaraní and Spanish are indeed one and it is inevitable that the native language carries all kinds of traces from Spanish (just like Paraguayan Spanish carries traces from Guaraní). At the turn of the new millennium the arena of languages has become the arena of political and cultural issues. Paraguay is a unique sociolinguistic case in the context of Latin America. But this uniqueness is founded less on its claimed bilingualism than on the fact that Guaraní is the only Indian language in Latin America spoken by non-Indian citizens as their mother tongue. According to the 2002 census,21 Guaraní monolinguals (27%) were significantly more numerous than Spanish monolinguals (6.56%), particularly in rural areas. Interestingly enough, the percentage of bilinguals was only 59%, that is, less than two thirds of the country’s population. These figures show that Paraguay is far from being a model bilingual society – if by bilingualism we mean the use of two languages on an equal basis by the whole linguistic community – but might become one in the future if we consider the developments of the last decade.22 5.2. Typological classification of Guaraní Guaraní is a flexible type-2 language in Hengeveld’s classification of partsof-speech systems. The language has only two lexical classes: verbs and non-verbs. The class of verbs is clearly identified by the existence of two morphological paradigms as shown in the following examples. (9)

a. (che) a-guata 1SG 1SG-go ‘I walk’

(10) a. (che) ai-pota 1SG 1SG-want ‘I want’

b. (nde) re-mba’apo 2SG 2SG-work ‘You work’ b. (nde) rei-pota 2SG 2SG-want ‘You want’

On the other hand, non-verbs may occupy any of the following syntactic positions without any derivational morphology: head of a referential phrase, modifier of referential phrase, and modifier of predicate phrase. Some examples illustrate this: (11) a. ko

karai tuja b. che tuva tuja man old 1SG father old ‘That old man’/‘that man is old’ ‘The oldness of my father’ DEM

Spanish meets Guaraní, Otomí and Quichua

(12) a. che ro-hayhu asy 1SG 2OBJ-love intense ‘I love you intensely’

189

b. nde rayhu asy 2SG love intense ‘your intense love’

The same lexeme tuja is the modifier of a referential phrase in (11a) and the head of a referential phrase in (11b). Likewise, asy is the modifier of a predicate phrase in (12a) and the modifier of a referential phrase in (12b). The above classification, however, does not show a typical characteristic of Guaraní: the possibility for most lexemes in the language to be used predicatively. This feature is perhaps more evident in the case of so-called quality-attributive verbs (Gregores and Suárez 1967: 138). These lexemes may be used as heads both of predicate and referential phrases as shown in (13a/b): (13) a. che-vy’á ne-recha-rehe 1SG-happiness 2.ACC-see-by ‘my happiness of seeing you’

b. a-vy’á ne-recha-vo 1SG-happiness 2.ACC-see-when ‘I am happy to see you’

In similar terms, a predicative reading of (11a) understands tuja not as ‘old’ but ‘to be old’. The predicative use of nouns, adjectives and manner adverbs is further illustrated in the following examples: (14) a. pe

kyse puku DEM knife red ‘that red knife’

b. che che-kyse 1SG 1SG-knife ‘I have a knife’

(15) a. a-jahe’o pochy-rehe 1SG-cry anger-by ‘I cry from anger’

b. che che-pochy 1SG 1SG-anger ‘I am angry’

(16) a. o-mbohovai mbarete 3-react strong ‘He reacts strongly’

b. o-mo mbarete 3-CAUS strong ‘He strengthens [it]’

In spite of this evidence, we have not classified Guaraní as a type-1 language (the most flexible as for parts of speech, without a distinction between nouns and verbs). The main reason for our choice is that there clearly exists in Guaraní one class of verbs identified on the basis of morphological distribution (cf. Nordhoff 2004) while no such conclusion may be drawn for any of the other major parts of speech. So far for the classification in terms of parts of speech. We will briefly sketch PG in terms of some other typological parameters. From a morpho-

190 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking logical point of view PG is an agglutinative language. Some authors also characterize PG as a ‘typical polysynthetic language’ (Trinidad Sanabria 1998). Though this statement remains to be proved in terms of ‘typicality’, it is certain that the high degree of synthesis shown by pre-Hispanic Guaraní suffered considerably from the contact with Spanish. Morínigo (1982) offers solid evidence of this and other contact-induced changes. Also, PG is an active language, i.e. there are two types of intransitive verbs, active and stative. Active verbs mark their subjects with the marker of the transitive subject; stative verbs get the marker of the transitive object (cf. VelazquezCastillo 2002). Constituent order in Guaraní is relatively free, with SVO arguably as the unmarked order in the clause (Gregores and Suarez 1967: 182). Within the noun phrase, the order also varies: head-modifier in attributive constructions, modifier-head in possessive constructions. PG lacks gender and number distinctions. Pre-Hispanic Guaraní did not have articles to express definiteness but PG makes an extensive use of Spanish definite article la (Gregores and Suarez 1967: 144). However, the functional status of the borrowed article in PG is not the same as that of the article in Spanish. Accordingly, we classify Spanish-derived la in PG within the overall category of deictics, because it marks not only definiteness but also reference and cohesion (see below Section 5.4.2). As additional features worth mentioning here, PG is a pro-drop and postpositional language and shows frequent noun incorporation (Velazquez-Castillo 1995). 5.3. Predictions about Spanish borrowings in Paraguayan Guaraní Having established that Guaraní is a type-2 language and Spanish type 4 in terms of Hengeveld’s part of speech typology discussed in Section 3, the following predictions can be made for linguistic borrowing. We expect that Guaraní borrows nouns, verbs, adjectives and (manner) adverbs in this order of frequency. If functional adaptation applies, we would find Spanish nouns (N), adjectives (A) and manner adverbs (MAdv) in all relevant functions, i.e. as head (HR) and modifier (MR) of referential phrases and as modifiers (MP) of predicate phrases. If functional specialization applies, we would find N, A and MAdv only as HR, MR and MP, respectively. As regards other categories, it is expected that PG does not borrow articles and prepositions from Spanish, as there is no syntactic position for this element in the language. Similarly, from the development of PG towards hypotaxis, it is expected that PG tends to borrow Spanish subordinators and

Spanish meets Guaraní, Otomí and Quichua

191

discourse markers, provided they fit in the typological profile of the language. Finally, one expects that the preference of PG for the predicative use of non-predicative elements is reflected in a similar use of originally nonpredicative borrowings.

5.4. Spanish borrowing in Guaraní: the data In this section we will present our results for borrowing from Spanish in Guaraní and test our findings on the basis of the relevant borrowing hypotheses. Our corpus consists of spoken texts provided by 38 speakers, with a total length of 57,828 tokens. Table 2 contains the figures for Guaraní and for Spanish borrowings in terms of tokens, types and type-token ratio. Table 2. Spanish borrowings in PG N=38

Target Language (PG)

Source Language (Sp)

Total

Tokens Types TTR

47772 (82.6%) 9220 (77.0%) 0.19

10056 (17.4%) 2760 (23.0%) 0.27

57828 11980 0.21

It is important to note that in the Guaraní texts we found a considerable amount of code switches, more so than in the corpora of the two other languages. This is probably caused by the fact that a great many speakers of PG are in fact perfectly bilingual. This may have its effect on the amount and type of borrowing. The influence of Spanish on the PG lexicon becomes evident from the overall figures in the table. Both in terms of tokens and types, the amount of Spanish items represents around one fifth of the corpus. This is just an average, however. The figures for individual speakers range from a mere 5.7% to a maximum of 28.5%. Especially the speakers from urban areas where Jopará has its stronghold show a relatively high score. Nevertheless, the variation is much less related to age than to geographical area and the division between rural and urban areas is not clear-cut, as it has always been maintained (cf. Garvin and Mathiot 1982: 29). Spanish borrowing in PG is not restricted to the lexicon, however. A parallel development is attested in the morphology and the syntax of the language due to the long-term contact with Spanish (Gomez Rendón forthcoming a). One illustrative example is today’s preference in Guaraní for more analytical constructions that use Spanish subordinators and deviate

192 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking drastically from the polysynthetic, much less hypotactic model of classical Guaraní. Another contact-induced change – one that was not considered for statistical purposes but certainly exerts influence on the shape of the present language – is the prolific use of code switching strategies. In order to analyze the borrowing profile in more detail, we classified borrowings according to the parts of speech they belong to in the source language. The figures from Table 3 allow us to make some statements about the respective contributions of Spanish lexicon and grammar to PG. Table 3. Spanish borrowings in PG: parts of speech

Tokens Types

Verbs

Nouns

Adjectives

Manner Advs.

Others

1842 (18.3 %) 578 (21.5 %)

3738 (37.2 %) 1223 (45.5 %)

747 (7.4 %) 313 (11.9 %)

95 (0.9 %) 45 (1.7 %)

3634 (36.1 %) 518 (19.3 %)

Table 3 shows a hierarchical ordering of lexical classes in the borrowing process, both in terms of tokens and types, according to the pattern N > V > A > MAdj. This is in fact what hypothesis H2.1.1 predicts. Furthermore, the contribution of the overall category ‘Others’ is significant and nearly equivalent to that of borrowed nouns in terms of tokens, and verbs in terms of types. It is precisely here that we find the contribution of Spanish grammatical borrowings to PG. These will be discussed in more detail in Section 5.4.2. First we will have a look at the major parts of speech. 5.4.1. Major parts of speech Let us consider now the use of Spanish borrowings in the corpus in order to test the hypotheses of lexical adaptation and specialization. Loanwords were analyzed for their function in PG in order to know which syntactic slots they occupy in the target language. The following are the results for the main lexical classes.

Spanish meets Guaraní, Otomí and Quichua

193

Table 4. Spanish borrowings in PG: parts of speech versus function Syntactic Function HP HR MP MR HP HR MP MR

Verbs (V)

Nouns (N)

Adjectives (A)

Manner Advs. (MAdv)

(Tokens)

(Tokens)

(Tokens)

(Tokens)

1842

8 3675

10 199 35 498

95

54 (Types)

(Types)

(Types)

(Types)

578

7 1214

7 91 18 242

45

37

A first impression of these figures sets aside adjectives and nouns from verbs and adverbs. While V and MAdv are used only in their prototypical positions of head (HP) and modifier (MP) of predicate phrases, respectively, N and A are used almost in all syntactic positions. For N this is only marginally so: 98.3% of the tokens are found in the prototypical HR function; for 93.6% of the types this is the only function. Only 10 Spanish nouns function as MR and 34 as HR plus one other function. They are never found in MP position. For A the distribution is somewhat better. Prototypical use prevails here as well: 63.6% of the Spanish adjectives have MR as the only function while 13.7% have MR combined with one or more functions. However, 22.7% are not used in MR function at all, but mainly as HR (85.9%) or MP (11.3%). As for the borrowing hypotheses, A seems to be the only category that gives some support to our expectations of flexibility and functional adaptation. Of the Spanish nouns, only some 3.6% are used in functions other than the prototypical one. And some of the MR uses might in fact be interpreted as possessive constructions, as in (17) below. The (rare) use of a Spanish noun as a HP is found in (18), where lado ‘side’ is inflected like a verb. (17) ha upépe katu oi-kove ha ña-hendu Kirito ñe’ and then sometimes 3-live and 1PL-listen Christ word ‘and sometimes they live and listen to Christ’s word’

194 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking (18) la

che gente-kuéra che rú-gui o-lado DEM 1.POSS family-PL 1.POSS father-ABL 3.side ‘my family sides with my father’

So, there are strong arguments for functional specialization rather than adaptation of Spanish nouns in PG. Possibly, this is due to semantic factors, but we will leave the analysis of the individual word meanings out of consideration here. As we have already seen above, borrowed Spanish adverbs are completely inflexible in their use in Guaraní utterances. This lack of flexibility is by no means restricted to manner adverbs. Adverbs of time and place show the same distributional restriction. The other syntactic function of adverbs is the modification of modifiers. Since this use is restricted to a closed class of adverbs that do not show manner, place or time features, we have not included them here. The inflexible behaviour of adverbs might be caused, at least in the case of manner adverbs, by their morphological marking in Spanish, where manner is usually expressed by suffixing -mente to the respective adjectives, as with English -ly. Thus, for manner adverbs, the hypothesis of functional specialization seems to hold without exception. But even for adjectives matters are not so clear-cut. When we look at some of them in more detail, the adaptation hypothesis finds even less support. Let us first look at the use of Spanish adjectives with HP function in PG. In such cases, there is always some kind of verbal marking present. In example (19) the Spanish adjective fanatico ‘fanatic’ is marked for third person. And in (20) provechoso ‘useful’ is marked for person and tense. (19) la

mbo’ehára Guaraní i-fanático teacher Guaraní 3-fanatic ‘That Guaraní teacher is a fanatic’ DEM

(20) I-provechoso-va’erã pe i-vida diaria-pe DEM 3.POSS-life daily-LOC 3-useful-FUT ‘That will be useful in their daily life’ In the much more frequent use of A as HR, the forms are morphologically unmarked. This is shown in (21) and (22). (21) o avei-j-apyté-pe colorado ho’a va’ekue preso fall PST imprisoned 3.be too-3-middle-LOC red ‘Among them was a red one who also was put in jail’

Spanish meets Guaraní, Otomí and Quichua

195

(22) umía

kampesino no-entendé-i PRO.DEM peasant NEG-understand-NEG ‘Those peasants don’t understand’

All three Spanish forms appearing in these sentences, as well as the majority of the other ones attested in the corpus, although of adjectival origin, are frequently used as nouns also in Spanish syntactic contexts, without any further formal adaptation. They may well be borrowed as N rather than A. Therefore, even for A, the case for functional adaptation is not very strong. 5.4.2. Other parts of speech Apart from the rest of the adverbs the category labelled ‘Others’ in Table 3 contains the following grammatical categories: articles (19.1% of the Spanish tokens), conjunctions (7.5%), numerals (1.7%), discourse markers (0.8%), adpositions (0.5%) and pronouns (0.2%). The prolific use of Spanish articles and conjunctions in PG deserves some attention. Classical Guaraní lacked the category of article and expressed definiteness by other means if necessary, for instance through the use of deictics. The borrowing of the Spanish article in PG has not resulted however in the creation of a new grammatical category. PG uses the Spanish article especially as an independent anaphoric (23), cataphoric (24) or elliptical (25) element for establishing cohesion in the discourse (cf. Gómez Rendón forthcoming a). The only two forms used are la (feminine; 91.8% of the cases) and lo (nonfeminine; 8.2%). (23) alguno-ko noñe’eiete la kastellano, some-DEM NEG.speak.NEG.very DEM Spanish(x) oi-ke-rõ eskuela-pe-nte la ñ-aprende-pa 3-come-when school-ALL-only DEM (x) 1PL-learn-ALL ‘Some don’t speak Spanish at all, only when we go to school, we learn it well’ (24) la

o-ñembo’é-re, la kastellano DEM (x) 3-learn-for DEM Spanish(x) ‘For them to learn it, Spanish’

(25) o gente-kuera la no-ñe’ -se-i-va there.is people-PL DEM(x) NEG-speak-want-NEG-that ‘There are people who don’t want to speak it [Spanish]’

196 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking The extensive borrowing of the Spanish article in PG, which is second only after N and before V in terms of tokens, is possibly enhanced by the existence of a set of PG demonstratives that are used much in the same way. The Spanish articles are either used as a demonstrative with [+distal, –visible] meaning, a shade not present in the PG demonstratives. They are also used as independent elements of an anaphoric character. In turn, these developments are closely related to the incorporation of conjunctions, the second most frequent grammatical class in borrowing. Classical Guaraní (ClG) also lacked the category of conjunctions and showed strong preference for juxtaposition over coordination and subordination. Today, however, PG makes extensive use of Spanish coordinators and subordinators. The heavy borrowing of Spanish articles and conjunctions, with the discourse functions assigned to them in PG, gives support to our hypothesis H1, which favours the borrowing of pragmatically outstanding elements. It goes counter, however, to hypothesis H2, which disfavours grammatical elements, and particularly to subhypothesis H2.1.2, which puts articles at the bottom end of the borrowing hierarchy. Spanish numerals are relatively frequently used. Although ClG had a fivevalue number system that is still used in PG, values above 5 are usually borrowings from Spanish. In our data, all values between 2 and 10 are from Spanish (around 41.1% of the borrowed numerals in terms of tokens) and so are another 65 higher values. The evident limitations of the vernacular numeral system resulted in its almost total replacement by Spanish for all practical purposes. Although adpositions are predicted to be the most likely candidate for borrowing from the category of grammatical items, their number is relatively low in PG. The number of different prepositions used is 12, with an average frequency of 4 per preposition. Since ClG is a postpositional language, we assume that hypothesis H4.1 is relevant here. This presents syntax as a potential constraint to borrowing. Finally, since PG is a pro-drop language, one would not expect the frequent use of personal pronouns in the first place. Borrowing would be further disfavoured by the infrequent use of pronouns in Spanish, which is also a pro drop language. In fact, all 6 types found in the corpus are not personal pronouns but indefinites of the type ‘someone’.

Spanish meets Guaraní, Otomí and Quichua

197

5.5. Conclusion By way of conclusion we can say that the borrowing pattern for the major parts of speech exhibited by Guaraní supports both hypotheses H1 and H2 in terms of the expected type and token frequencies. Although predictions based on parts-of-speech types suggest functional adaptation rather than specialization, this is not confirmed for nouns and manner adverbs, and only marginally for adjectives. As far as grammatical categories are concerned, we have seen that there are an unexpectedly high number of Spanish articles. This goes counter to hypothesis H2.1.2, which puts articles towards the bottom of the borrowing hierarchy. Also, since the category of article is absent from PG grammar at all, there might be in fact no syntactic position for articles in the language, according to hypothesis H4.1. However, we also saw that most of the articles were not used in their prototypical function, as definiteness markers in a noun phrase, but as freestanding discourse connectors. This is indicative of the fact that borrowing probabilities should be based not only on the categories of elements in the source language but also on the potential and actual functions they get in the target language. More evidence for this will be found in the case of the other two languages we have studied. Considering the evidence above, the present profile of the language is not one of extensive borrowing only (i.e. stage 2 on Thomason’s scale) but rather one of strong convergence towards Spanish. From this perspective it is not difficult to side with those who see in today’s Paraguayan Guaraní a ‘third language’ that is neither Spanish nor Guaraní (cf. Melià 1974).

6.

The case of Quichua

Ecuadorian Quechua is a language of the Quechua family spoken in Ecuador. Like other Quechua languages, it has been in contact with Spanish since the European conquest. The long and intensive contact between Quechua and Spanish has resulted in a pervasive reciprocal influence. The present section deals with Spanish lexical borrowing in Ecuadorian Quechua, or Quichua (EQ). It explores the statistical results from the investigation of a corpus of spontaneous speech collected during fieldwork in Ecuador (2003– 2004). In this section, we follow the pattern of the previous section on Guaraní. Section 6.1 sketches some historical and sociolinguistic aspects of Spanish-Quichua contact in the Ecuadorian Andes. Section 6.2 addresses the classification of EQ according to its parts of speech system and other

198 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking parameters. Section 6.3 presents our expectations around borrowings from Spanish in EQ according to the theoretical assumptions of Section 6.2. Section 6.4 tests these hypotheses on the basis of the corpus. Some conclusions follow in Section 6.5. 6.1. Historic and sociolinguistic context of contact between Quechua and Spanish Language contact is part and parcel of conquest and colonization settings such as those that characterize the Andean history before and after the European invasion. The Northern Andes were subject to two successive conquests by the Incas and the Spaniards in less than sixty years. Crossroads for the traffic of goods, languages and people, the Northern Andes featured a great cultural variety by the time the first Europeans arrived. Long-distance traders from Chincha in central Peru introduced Quechua in the present territory of Ecuador around the first centuries of the second millennium (Torero 2002: 93). Chincha Quechua became the source of the present Ecuadorian dialects, with a significant contribution of pre-Inca languages and other Quechua varieties brought by soldiers, officers and immigrants from several areas of the Inca Empire.23 Quechua along with ten other native languages were spoken in the Ecuadorian Andes by the time of the Spanish invasion. One hundred years after the foundation of the first Spanish cities in the Northern Andes, most native languages had disappeared or were on the path of extinction. Quechua was the exception. The evangelization of native peoples speaking different languages seemed feasible only to the extent that one language of widespread distribution could be used as lengua general, especially if that language was already in use by native peoples. Quechua met these requirements. The Third Lima Council (1583) ordered the translation of Catholic texts into a standardized variety of Quechua resembling very much the Ecuadorian dialects (Adelaar 2004: 183).24 In the following decades a number of grammars and dictionaries were prepared to help missionary work in the Royal Audience of Quito and other cities of the Viceroyalty of Lima. The flowering time of Quechua came to a halt by the end of the eighteenth century in the frame of the Bourbon reformation, when a series of failed Indian uprisings called the attention of the Spanish Crown to the dangers of Quechua as a uniting factor and instrument of revolt. With the birth of the Andean republics in the nineteenth century Spanish remained the only official language supported by the state. Until the late 1970s Quechua

Spanish meets Guaraní, Otomí and Quichua

199

was not taught in schools, and Indian pupils used to carry the heavy load of learning subjects alien to their culture in a language they scarcely heard at home. Nowadays the situation of (Spanish) monolingualism imposed on multilingual and multicultural societies prevails throughout the Andes, with the result of an ever-increasing pressure for the Hispanisation of native communities. Another side of language contact in the Andes is socio-cultural. From the inception of the colonial regime, the Crown furthered the separation of Indians from Spaniards into what came to be known as ‘the two republics’. The late seventeenth century witnessed the emergence of the hacienda system, based on the appropriation by the ruling elites of vast areas of land at the cost of Indian communities. This establishment contributed to isolating Quechua-speaking peoples from education and other public institutions, except for the Church. The republics founded after the Independence wars of the 1820’s did not effect any changes in the situation and Indians remained to a certain extent non-integrated in the nascent nations. Despite the varied forms of non-integration, Indians always had some type of contact with the Spanish-speaking world. In permanent need of labour for public works and private service, Spaniards and mestizos used – and continue to use – the Indian workforce in urban and rural areas. Indian labour migration became an everyday issue in the life of mestizo towns. The foundation of the Andean republics in the early nineteenth century did not change things either. Indian migration from the countryside to the cities was even stronger in the twentieth century. In this context, the emergence of Quechua varieties heavily influenced by Spanish becomes a matter of fact. 6.2. Typological classification of Quichua Just like Guaraní, Quichua is an example of a verb-nonverb (type 2) language, in which verbal lexemes occupy the syntactic slot assigned to predicate phrase heads, and non-verbal lexemes occupy the other three slots. Examples (26)–(27) from Schachter (1985: 17) provide the evidence. (26) a. rika-sha-ka: hatun-ta see-PST-1SG big-ACC ‘I saw the big one’

b. chay hatun runa DEM big man ‘that big man.’

(27) a. rika-sha-ka: alkalde-ta see-PST-1SG mayor-ACC ‘I saw the mayor’

b. chay alkalde runa DEM mayor man ‘that man who is mayor’

200 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking Hatun ‘big’ is both a referential phrase modifier in (26b) and a referential phrase head in (26a). Similarly, alkalde ‘mayor’ is used both as head noun in (27a) and modifier of the head noun in (27b). Evidence against this classification has been recently presented by Beck (2002: 144ff.) and will be discussed in the following. According to Beck, the lack of distinction between nouns and adjectives is not complete because only the latter can be modified by adverbs like maymi ‘very’ as shown in (28): (28) chay warmi maymi sumak-mi DEM woman very pretty-FOC ‘That woman is very pretty’ Interestingly, Cole gives an example of (ungrammatical) noun modification precisely with maymi (Cole 1985: 99–100), which was elicited by one of us in site as fully grammatical: (29) chay warmi maymi duktur-mi DEM woman very doctor-FOC ‘That woman is a real doctor’ Beck’s argument is meant to point out a semantic distinction between property concepts and entity concepts. Hengeveld’s theory of parts of speech does not exclude such a distinction but argues for the existence of one nonspecialized lexical class that embraces both concepts. As Beck himself admits: “The existence of a semantic distinction of this type is in itself not enough to establish that there is a parts-of-speech distinction between nouns and adjectives in the lexicon” (2002: 144). Beck’s second argument states that noun-noun constructions are better treated as compounds on the basis that nouns acting as attributes of other nouns cannot occur more than once in the same noun phrase – as opposed to adjectival modifiers that may be stacked indefinitely. As additional evidence Beck mentions that noun-noun constructions themselves may be attributives of other nouns, as in (30), from Cerrón-Palomino (1987: 300). (30) hara chakra rumi corn field stone ‘stone of the cornfield’ In (30) hara and chakra as a whole modify the noun head rumi. Readings such as ‘field stone of corn’ or ‘corn field of stone’ are not possible. The

Spanish meets Guaraní, Otomí and Quichua

201

only possible reading considers the first two nouns as forming an attributive compound just like its English equivalent ‘cornfield’. A phonological test would consist in analyzing the stress pattern to know whether hara chakra is realized as a single word (with stress only on the penultimate syllable) or two individual words (with double stress). Unfortunately, this phonological information is not available at the moment. Beck’s third argument against the typological classification of EQ as a type-2 language is that property concept words being used apparently as heads of referential phrases as in (31) below are indeed adjectives standing for deleted heads in elliptical constructions. Conclusive proof of this interpretation is, according to Beck, “their reliance on context to supply the identity of a nominal head”. Therefore, sentences like (31) would be ungrammatical if out of context. (31) puka-ta ri-ka red-ACC see-PST ‘he sees the red one’ The claim that contextual reference is always required for the correct interpretation of sentences like (31) is not conclusive either. Colour terms are universally associated to objects and do not exist independently, being to this extent context-dependent in any human language and irrelevant for a noun-adjective distinction.25 What really matters here is not the contextual dependency of lexemes like puka in (31) but the fact that they occupy the syntactic slot of referential heads without any further measures and take nominal morphology, in this case the accusative marker. Additional evidence for our classification of EQ as a type-2 language comes from the fact that both nouns and adjectives can be used as predicate phrase modifiers. Consider the following examples. In (32) yanka ‘useless’ occupies the position of referential phrase modifier and, without any derivation, the position of predicate phrase modifier in (33). (32) kai-ka yanka yura ka-n-mi that-TOP useless plant be-3-AFF ‘that is a useless plant’ (33) kaina chaupi tuta-kaman yanka shuya-ku-rka-ni yesterday middle night-up.to useless wait-DUR-PST-1SG ‘yesterday I waited until midnight to no avail’ Similarly, utka ‘speed’ functions as head of a referential phrase in (34) while the same word modifies the predicate shamui ‘come’ in (35).

202 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking (34) utka-ka rura-shpa alli-mi ka-n speed-TOP work-GER good-AFF be.3 ‘quickness is good in working’ (35) utka huasi-man shamu-i speed house-ALL come-IMP ‘come home quickly’ On the basis of the above we conclude that the evidence against classifying EQ as a type-2 language is insufficient. Therefore, we will assume in the following that Quichua is a language that makes no distinction between nouns, adjectives and adverbs. From a morphological point of view, EQ is an agglutinative language though its morphological profile has suffered from two changes involving simplification with respect to other Quechua varieties: the loss of verbobject agreement and the loss of possessive nominal suffixes. Cole (1985: 6) presents the examples in (36a/b) for second-person object agreement in (Peruvian) San Martin Quechua and Ecuadorian Quechua. (36) San Martín Quechua a. ñuka-ka maka-yki 1SG-TOP hit-2OBJ ‘I hit you’

Ecuadorian Quechua b. ñuka-ka kan-ta maka-ni 1SG-TOP 2SG-ACC hit-1SG ‘I hit you’

As for the loss of possessive nominal suffixes, consider the following examples from Cerrón-Palomino (1987: 200): (37) Junin Quechua a. maki-yki hand-2SG.POSS ‘your hand’

Ecuadorian Quechua b. kanpak maki 2SG.GEN hand ‘your hand’

The extensive use of pronouns in EQ as compared to other Quechua varieties has, in turn, resulted in the obligatory use of pronouns not only in subordinate but also in main clauses where other varieties use them only for emphasis. To this extent we can claim that EQ developed from a pro-drop to a non pro-drop language. A parallel development in subordinate clauses in EQ is the increasing tendency to prefer coordination and relativization to the traditional strategy of nominalization, with the result of independent clauses linked by connectors instead of one main clause with embedded noun phrases.

Spanish meets Guaraní, Otomí and Quichua

203

Constituent order is rather fixed in EQ and displays an almost invariable SOV pattern. Within the noun phrase, the order is also fixed most of the time: modifier-head in attributive and possessive constructions. EQ lacks grammatical gender and the lexical category of article to express definiteness. Finally, all adpositions in EQ are postpositions, which many authors classify as members of a large set of inflexional suffixes (Cerrón Palomino 1987: 270ff.). 6.3. Predictions about Spanish borrowings in Quichua Since Quichua was classified as a type-2 language, just like Guaraní, we can repeat the predictions for borrowing here. Thus, Quichua will borrow N, V, A and MAdv in that order of frequency. Again, if functional adaptation applies – it did not for Guaraní – we will find Spanish nouns, adjectives and manner adverbs in all relevant functions, i.e. as heads and modifiers of referential phrases and as modifiers of predicate phrases. If functional specialization applies – as it did more or less for Guaraní – we will find N, A and MAdv only as HR, MR and MP, respectively. In relation to other categories, it may be hypothesized that Spanish borrowings in EQ will not include prepositions and articles, as there is no syntactic position for them in the language. On the other hand, pronouns might be borrowed on the basis that EQ has become a non pro-drop language though this tendency may be disfavoured in the first place by the fact that Spanish is a pro-drop language itself. Finally, if we consider the recent developments in EQ discussed in Section 6.2, coordinators and subordinators are also expected in borrowing. 6.4. Spanish borrowing in Quichua: the data Here we will present our observations about borrowing from Spanish in Quichua and again test our findings on the basis of the relevant borrowing hypotheses. The Quichua corpus consists of spoken texts provided by 25 speakers, with a total length of 79,718 tokens. Table 5 contains the figures for Quichua and for Spanish borrowings in terms of tokens, types and typetoken ratio.

204 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking Table 5. Spanish borrowings in EQ N=25 Tokens Types TTR

Target Language (EQ)

Source Language (Sp)

Total

64620 (81.1%) 19023 (83.5 %) 0.29

15098 (18.9 %) 3768 (16.5 %) 0.25

79718 22791 0.28

Looking at these global figures, the influence of Spanish on EQ lexicon is close to 20% of the corpus in the case of tokens. The percentage is somewhat less, however, for types. We will come back to a comparison between the languages in Section 8. The figures for the individual speakers range from only 4.0% of the tokens to a stunning maximum of 49.1%. While this percentage is evidence of a strong lexical influence from Spanish, it is by no means uncommon to each of the Ecuadorian dialects of Quechua. Arguably there exists a correlation between this percentage and some changes in EQ morphosyntax along the borrowing scale proposed by Thomason (2001). EQ uses a few Spanish morphemes productively and shows at the same time syntactic deviations such as inverted possessive constructions and verbs in medial position. For the respective parts of speech, we found the following numbers. Table 6. Spanish borrowings in EQ: parts of speech

Tokens Types

Verbs

Nouns

Adjectives

Manner Advs.

Others

2672 (17.7%) 600 (15.9%)

8220 (54.0%) 2037 (54.0%)

1276 (8.5%) 514 (13.6%)

88 (0.6%) 37 (1.0%)

2842 (18.8%) 580 (15.4%)

Again as predicted by hypothesis H2.1.1, there is a hierarchical ordering of lexical classes in the borrowing process, both in terms of tokens and types, according to the pattern N > V > A > MAdj. However, the position of N is much more outstanding than in the case of PG, mainly at the cost of the category ‘Others’. The lexical and grammatical borrowings will be discussed in Sections 6.4.1 and 6.4.2 respectively. 6.4.1. Major parts of speech Analysis of the borrowed items in terms of Spanish parts of speech and functions in Quichua syntax gave the overall results in Table 7 below.

Spanish meets Guaraní, Otomí and Quichua

205

Table 7. Spanish borrowings in EQ: parts of speech versus function Syntactic Function HP HR MP MR HP HR MP MR

Verbs (V)

Nouns (N)

Adjectives (A)

Manner Advs. (MAdv)

(Tokens)

(Tokens)

(Tokens)

(Tokens)

2668 3

58 7618 141 392

16 287 131 841

88

(Types)

(Types)

(Types)

(Types)

596 3

20 1767 29 215

10 128 49 327

37

1

1

When we analyze the functional aspects of Spanish borrowings in EQ for the respective parts of speech, we find the following. Verbs are typically used in their HP function only. For adjectives and nouns the situation is quite different. Although the prototypical functions prevail here as well, with 87% (N) and 64% (A) respectively, there are much more than marginal frequencies for the other three functions. Almost 11% of the Spanish nouns are used as nominal modifiers while 24.9% of Spanish adjectives function as nominal heads. A substantial 9.5% of Spanish adjectives operate as verbal modifiers. Let us first look at some examples of the use in EQ of Spanish nouns. (38) chayka chicha-ka aidante-pak-ka then chicha-TOP assistant-ALL-TOP jala-shka shina barro manga-pi earth pot-LOC carry-PERF so ‘the chicha was then carried in an earthen pot to the assistant’s house’ (39) ashtawan juirza baila-k ka-rka more strength dance-DUR be-PST ‘they were dancing more intensely’ Example (38) shows the use of Spanish noun aidante (from ayudante ‘assistant’) as HR and barro ‘mud’ as MR of the Quichua head manga ‘pot’. And (39) illustrates the use of the Spanish noun juirza (from fuerza ‘strength’) as a manner adverb modifying the predicate head bailak ‘dance’.

206 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking The next examples illustrate the different uses to which Spanish adjectives are put in EQ other than their prototypical syntactic function of referential phrase modifiers. Thus, the Spanish adjectives antiwo ‘old’ and pariju ‘equal’ are used as head of a referential phrase (40) and modifier of a predicate phrase (41), respectively. (40) chasna antiwo-ka ka-rka so old-TOP be-PST ‘the old (use) was like that’ (41) Sankuan-ta-ka pay-wan pariju baila-nchik-mi Sanjuan-ADV-ka 3SG-INST equal dance-1PL-FOC ‘at the Saint John’s Festival we dance all together’ Spanish antiwo in (40) stands for ‘that which is old’ and refers to a distant past when some implicit referent was in use. In (41) the adjective pariju (from parejo, ‘equal’, ‘from both sides’) occurs in the function of a predicate phrase modifier, pointing out the manner in which people danced during Saint John’s festivals, i.e. men and women danced together on that occasion. While Spanish nouns and adjectives are rather flexible in relation to their syntactic functions in EQ, this is again not the case for manner adverbs. There is not a single example of a manner adverb in a syntactic position different from the canonical MP. As suggested already for PG, this might be due to the morphological markedness of Spanish adverbs with the derivational suffix -mente. The majority of the borrowed manner adverbs are of this type (29 out of 37 types). The fact that these overtly marked adverbs are used only as predicate phrase modifiers might suggest that bilinguals are aware of the function of the adverbializer -mente in the source language and use all corresponding lexemes exclusively in the syntactic slot of predicate modifiers. If this is true, then bilingual processing seems to play a crucial role in the borrowing process and may overrule other factors. While ‘bare’ manner adverbs are relatively few in the corpus (88 appearances), EQ shows preference for complex Spanish constructions in order to indicate manner. Thus, we find prepositional phrases like de repente ‘suddenly’, a propósito ‘on purpose’, por ley ‘obligatorily’, a la carrera ‘rapidly’ and many others acting as adverbials (cf. example 42 below). Counted as single lexical entities, which is probably the way they should be interpreted syntactically, these constructions cover around 1.0% of the borrowed tokens. Apart from that we find quite a few Spanish adjectives used as predicate phrase modifiers without further measures being taken: 49 types with 131 appearances.

Spanish meets Guaraní, Otomí and Quichua

207

6.4.2. Other parts of speech Apart from a small set of adverbs (around 2.5% in terms of types and tokens), the category labelled ‘Others’ in Table 6 contains the following grammatical categories: conjunctions (6.9% of the tokens), interjections (1.4%), numerals (0.9%), discourse markers (0.6%), and prepositions (0.5%). Pronouns and articles (0.1%) are very rarely borrowed, two categories which were very frequently borrowed by PG. In other respects, the picture is roughly the same. There are a total of 17 different Spanish prepositions almost always used as part of fixed adverbial expressions but not as prepositions themselves. Example (42) illustrates this: (42) ñuka-ka por-gusto ri-ni ufia-ngapa 1SG-TOP by-pleasure go-1SG drink-PURP ‘I gladly go to drink’ Thus, syntax seems to provide a constraint, not on borrowing as such, but on employing the element in an ‘impossible’ syntactic slot. The same motivation as for PG could be given for the presence in the corpus of around 70 different types of numerals from Spanish, i.e. the pragmatics of negotiation and the fact that, in financial transactions, Spanish is often the language of communication. The use of conjunctions can be easily linked to certain syntactic developments in EQ where nominalization strategies are being replaced by relativization (hence also the introduction of some pronouns, mainly relative and interrogative) as the preferred mechanism of clause linking (cf. Gomez Rendón forthcoming b). Consider the use of the Spanish coordinator y ‘and’ in (43) where traditional Quichua uses simple juxtaposition or the additive suffix on both coordinated elements. Likewise, observe the use of the Spanish relative conjunction que ‘that’ in (44) which connects the otherwise nominalized subordinate clause to the predicate nini ‘I say’: (43) ñuka ayllu kanchis-pura kan, pichca churi ñuka y ñuka warmi 1SG family seven-among be.3 five child 1SG CONJ 1SG wife ‘my family is composed of seven people: five children, me and my wife’

208 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking (44) ñuka ñuka japi-shka-manta ni-ni que ñukanchik 1SG 1SG understand-PERF-ABL say.1 that 1PL llakta-kuna-pi yachai-kuna-ka tiya-n exist-3 community-PL-LOC idea-PL-TOP ‘from what I understood I say that there are ideas in our communities’

6.5. Conclusion Also Quichua gives support to hypotheses H1 and H2 in terms of type and token frequencies. In contrast to Guaraní, however, data support the functional adaptation strategy more than the specialization strategy, be it only for nouns and adjectives. Manner adverbs behave inflexible in all cases. The way Spanish prepositions are used in Quichua syntax is further support for the hypothesis that the part of speech of a borrowed element in the source language and the function assigned to it in the target language should be taken into consideration when analyzing borrowing effects. From the social and historical background in Section 6.1 and the linguistic data we collected, it is safe to identify EQ with the third stage of Thomason’s borrowing scale: an average degree of bilingualism among the linguistic community (Büttner 1993: 48–49) and lexical borrowing of nonbasic and basic vocabulary. Contact-induced changes in the lexicon include the borrowing of a significant percentage (> 20%) of vocabulary, including Spanish conjunctions and numerals. Contact-induced changes in the structure include Spanish-derived SVO word order instead of typical Quechua SOV; subordinate constructions headed by Spanish conjunctions and SVO clauses, instead of typical Quichua participial constructions; and a few Spanish inflectional suffixes of gender and number.

7.

The case of Otomí

Otomí is a language from Central Mexico, which since the European conquest has been in contact with Spanish and has undergone a pervasive influence from that language. Below we describe the Spanish lexical and grammatical borrowings in two Otomí dialects spoken in the State of Querétaro, viz. the dialect of Santiago Mexquititlán and the dialect of San Miguel de Tolimán. Both belong to the branch of north-western Otomí, one of the larger variants of Otomí with around 33,000 speakers. Santiago Mexquititlán is located in the southern part of Querétaro, in the mountains

Spanish meets Guaraní, Otomí and Quichua

209

of the neovolcanic axis of Mexico, and is a town with a population of around 15,000 inhabitants. The Otomí spoken in Santiago Mexquititlán is similar to the Otomí dialect of the villages in the north of the State of Mexico. San Miguel de Tolimán is situated in the northern part of Querétaro, in the semidesert of the Sierra Madre Oriental, and is a village with a population of around 700 inhabitants. The Otomí spoken here is similar to the Otomí dialect of the Valle del Mezquital in the State of Hidalgo. In both villages the vast majority of the population are Otomís. For our research we base ourselves on a corpus that consists mainly of spontaneous speech collected during our fieldwork between 1993 and 2004. In Section 7.1 we describe the historic and sociolinguistic context of the contact between Otomí and Spanish. In Section 7.2 we give a typological classification of Otomí according to its word order, lexical parts-of-speech system, syntactic relations and numerical system. On the basis of that classification Section 7.3 presents some predictions about what changes may be expected in Otomí as a result of its contact with Spanish. Section 7.4 will discuss what has been attested in our corpus. And in 7.5 we will draw some conclusions.

7.1. Historic context of the contact between Otomí and Spanish Otomí is a central Amerindian language, currently spoken by around 310,000 mostly bilingual speakers on the highlands around Mexico City in the States of Mexico, Hidalgo, Querétaro, Puebla, Guanajuato, Tlaxcala, Veracruz and Michoacán. Together with Mazahua, Ocuilteca, Matlatzinca, Pame and Chichimeca it constitutes the Otopame group within the Otomanguean family. It was the native language of the old inhabitants of the Valley of Mexico and its surrounding valleys. Throughout history its speakers had to confront the Aztecs, Spaniards and Mestizos, speakers of Nahuatl and Spanish, both belonging to other language families. Since the Otomís had to surrender to the Nahuas from the 15th century onwards, there has been a very close contact between the Otomí and Nahuatl languages. During that contact the Nahuas developed a very negative image of the Otomís, which later was passed on to the colonial chroniclers, such as Sahagún (1982 [1557]), whose Nahuatl speaking informants considered the Otomís “toscos e inhábiles” (coarse and unskillful). The very fact that the word Otomí probably is a derivation of the Nahuatl word totomítl ‘bird hunter’ is an example of the negative image imposed by the Nahuas (Jiménez Moreno 1939).26 The Otomís like to call themselves Ñäñho, Ñöñhö or

210 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking Ñähñu, i.e. ‘he who speaks well’ and their language Hñäñho, Ñhöñhö or Hñähñu (Hekking 1995). Since the Otomís were the second most numerous group after the Nahuas on the highlands, the Spaniards were strongly interested in their conversion to catholicism. Despite the fact that the language was considered very difficult because of its complex vowel and consonant system, a spelling system for Otomí was developed and grammars, vocabularies, catechisms and legal documents were written in it. Especially missionary friars of the Franciscan order have studied the Otomí language, above all Fray Alonso de Rangel, who arrived in Mexico in 1529 and died in 1547. He published the first Arte i Doctrina Cristiana en lengua Otomí. Another Franciscan who deserves mentioning is Fray Pedro de Cárceres, who wrote his Arte de la lengua Otomí in Querétaro around 1580 following the spelling developed by Fray Alonso de Rangel. These spelling rules were improved by Fray Alonso Urbano, who wrote the linguistically important trilingual Arte breve de la lengua Otomí y vocabulario trilingüe, españolOtomí-náhuatl at the beginning of the 16th century. Sociolinguistically very interesting is the Códice Martín del Toro from the second part of 17th century. This document aims to show that Pedro Martín del Toro, the leading figure of the text, belongs to a lineage of Otomí nobles who militarily had formed an alliance with the Spaniards. In it there is a constant code switching between Otomí and Spanish (Guerrero 2002). It should be mentioned that the colonial documents written in Otomí are not easily readable since the authors not always distinguish the Otomí phonemes that do not exist in Spanish from those that do, specifically some vowels. Only trained linguists with a good command of Otomí are able to read them, often with difficulty as in the case of the Códice Martín del Toro. After the Independence of Mexico in 1813 major changes occurred in the indigenous community. Officially the indigenous groups were no longer recognized. As a result of the loss of their special status, many Otomís could not afford their education any more and Otomí was no longer written by the civil authorities, only by a handful of scholars. It was in the nineteenth century that a process of language shift started. The Mexican Revolution (1911–1917) did not result in social change for the Otomí population, nor did it foster more recognition of their language, or stopped the language shift. On the contrary, after a long history in which they have been degraded socially little by little, the Otomís now belong to the lowest social levels of the Mexican society, as is the case for many other indigenous groups in Mexico. They dwell in the most remote and less fertile places on the highlands living on an agriculture of subsistence, reason why many of them have chosen to emigrate to the bigger towns, such as

Spanish meets Guaraní, Otomí and Quichua

211

Mexico City, Guadalajara and Monterrey. In the 20th century several attempts have been made to integrate the indigenous communities in the national community by means of the officially called Educación Bilingüe given by indigenous teachers. These, however, often have a very negative attitude towards their own roots and are ignorant about bilingual education. Consequently, most Otomís are illiterate in their first language and very often have insufficient command of the standard variety of Mexican Spanish. Otomí is only spoken inside informal domains such as the family. During the last 20 years, because of the construction of roads and schools, the growing influence of the media and the increasing trade and emigration, contact between the relegated Otomís and the Spanish speaking Mestizos has dramatically increased. As a result, a rapid increase in contact phenomena from Spanish is observed in the respective varieties of the Otomí language (Hekking 1995, 2001, 2002; Hekking and Bakker 1998, 1999, 2006). As discussed extensively in Zimmermann (1992), at this stage in its history Otomí has become a stigmatized language, only spoken by poor and traditional people. As a result of this, many Otomís do not want to teach the indigenous language to their children any more. In our view, this makes Otomí an endangered language, despite its current high number of speakers. 7.2.

Typological classification of Otomí

Otomí has many lexical elements which specialize as verbs or nouns. Although nouns may be more or less freely used as the main predicate of a clause, and then may be marked for tense, aspect, and person and number of the subject, just like verbs, only they may be nominal heads of noun phrases and are then obligatorily preceded by an article or a possessive. On the other hand, only verbs may be marked for object, for a beneficiary, a locative or by a reflexivity marker. This is exemplified in (45). ar nguu ar majä hyonguu bi hyom-bi bricklayer PST.3=build-BEN DEF.SG house DEF.SG priest ‘The bricklayer built the house for the priest’

(45) ar

DEF.SG

So, Otomí is a rigid language in terms of Hengeveld (1993). The major question is whether Otomí has adjectives. Many words that would translate to adjectives in Spanish and English, are nouns in Otomí, since they are preceded by an article. An example is ‘ra’yo ‘new’. Others such as dathi ‘ill’, txutx'ulo ‘small’ and johya ‘happy’ are of a verbal nature since they

212 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking are obligatorily preceded by verbal proclitics such as the continuative aspect marker (Lastra de Suárez 1997: 30f.). However, yet another group with typical adjectival meanings such as ‘thin’, ‘thick’, ‘bitter’, ‘sweet’, ‘cold’, ‘warm’, ‘beautiful’, ‘ugly’, ‘good’, ‘high’, ‘low’ and colour words are found in bare form in an adnominal position, and are therefore considered to be adjectives by a number of linguists (cf. Ecker 1952; Hekking and Andrés de Jesús 1984; Voigtlander and Echegoyen 1985; Lastra de Suárez 1992, 1994, 1997; Andrews 1993; Bartholomew 2004). On the other hand Palancar (2006) claims on the basis of his data from the Otomí spoken in San Ildefonso, in the state of Querétaro, that all lexemes which denote property concepts in Otomí are encoded as verbs and nouns and not as adjectives. Although it is generally acknowledged that there are less adjectives in Otomí than for example in Spanish or English, there seems to be enough reason to assume that there is at least some – possibly contact induced – specialization for the A category. This is exemplified in (46) below (from Hekking, Andrés de Jesús and Santiago de Quintanar, forthcoming). ‘behñä! (46) a. ¡Maje, nzatho goodbye handsome woman ‘Goodbye, handsome woman’

b. ar

’ñu thuhme DEF.SG sweet bread ‘The sweet bread’

We also find some elements which operate as manner adverbs in Otomí. Examples are found in (47a/b). ‘behñä ma ‘nitho ndi metu-we ár leave quickly 3 tell-1.INC POSS.3SG wife nä’ä xki thogi what PSTPRF.3 happen ‘He left quickly in order to tell his wife what had happened’ b. gi ñuni njante, gi jänt’i hingi ñähi gi ñuni komo-ngu 2 eat slowly 2 swallow NEG.2 chew 2 eat like-like gar mi’ño 3 coyote ‘Eat slowly, you eat without chewing, like a coyote’

(47) a. bi

PST.3

However, they are about the only examples of such use. Typically, a verbal construction is found where English or Spanish would use an adverb, as in (48). ‘bexui da hingi handi xí hño ja-r eye NEG see 3PRF good LOC-DEF.SG darkness ‘I can’t see very well in the darkness’

(48) ma

POSS.1

Spanish meets Guaraní, Otomí and Quichua

213

So far, this leads us to the categorization of Otomí as a type 5 language with the extra stipulation that the class of adjectives is only semi-open in the sense that it does not allow easy extension. Just like the other Otomangue languages, Otomí is a tone language. It distinguishes three tones: a high, a low and a rising tone. The basic constituent order is VOS, a common trait of Otomangue in general (Suárez 1983; Yasugi 1995). From a morphological point of view, Otomí has a rather complicated synthetic structure on the lower syntactic levels, in particular the noun phrase and the verbal complex. But at the same time Otomí has an analytical structure at the sentence level and is a language with asyndetic compounding or juxtaposition of constituents. It has very few explicit markers of the semantic or syntactic relations between constituents, such as prepositions, coordinators, subordinators and relatives, as can be seen in example (49). (49) nu’bya di ne ga tsoni Nxuni today 1 want FUT.1 arrive Morelia ‘Today I want to arrive at Morelia’ Some relations are marked on the verb, such as the beneficiary in (45) above, and the suffix -wi in (50) which marks accompaniment. (50) ar

Xuwa mi ñä-wi ár to DEF.SG Juan IMPRF.3 speak-ACMP POSS.3SG mother-in-law ‘Juan spoke with his mother-in-law’

Otomí does not have a full class of adpositions. However, the particles dige, ja, ir nge and ngu mark a variety of relations between noun phrases and other parts of the sentence. Their exact interpretation depends heavily on the context. See the use of dige in (51a/b): pa Maxei (51) a. di ñä-he dige ma boni-he 1 speak-1PL.EXC about POSS.1 trip-1PL.EXC to Querétaro ‘We speak about our trip to Querétaro’ b. 'nar jä’i pwede da du dige-r t’ete IND.SG person may FUT.3 die through-DEF.SG witchcraft ‘A human being may die through witchcraft’ Otomí has several coordinators (ne ‘and’, 'nehe ‘also’) and subordinators (wa ‘or’, jange ‘therefore’, ngetho ‘because’, ngu ‘as’, and nu'bu ‘if’, ‘when’) at its disposal. However, the prevailing form for both coordination and subordination of clauses is asyndetic juxtaposition.

214 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking In Classical Otomí a relative clause, which is always in a postnominal position, generally is attached to a main clause according to the gapping strategy without any form of connection (cf. Comrie 1990: 147f.). It is possible to relativize on any part of the relative clause, that is to say not only on the Subject, the Direct or Indirect Object, and the possessor in a possessive construction, but also on the accompaniment, the beneficiary, the instrumental and the locative. Like many Mesoamerican languages Otomí uses a vigesimal system for its numerals, a system that according to Bartholomew (2000) was adopted by the Nahuas, but not by the Spaniards. Otomí numerals often function as verbs or as nouns. When used as a noun they are preceded by the nominal clitic ar and when used as a verb, as shown in (52), they are accompanied by verbal proclitics and suffixes. (52) di hñu-he 1 three-PL.EXC ‘We are three’ 7.3. Predictions about Spanish borrowings in Otomí With Otomí a type 5 language and Spanish a type 4, we expect easy borrowing for classes V and N under functional specialization, i.e. V in HP function and N in HR function. Adjectives will probably be borrowed, only as MR. It remains to be seen whether this will happen at a large scale, since the class of adjectives seems to be only semi-open in Otomí. Spanish manner adverbs will not be borrowed since the language lacks the category and a real MP slot. For the other categories, it may be expected that prepositions will be borrowed, since a slot for them seems to be present in the form of particles with a corresponding function. The same goes for coordinators and subordinators. Furthermore, we expect that many speakers will use Spanish numerals for values higher than 10 since the vigesimal system is unpractical in the Hispanic world. Finally, we assume that we will quite frequently find Spanish discourse markers, typically as first elements of a clause.

Spanish meets Guaraní, Otomí and Quichua

215

7.4. Spanish borrowing in Otomí: the data For Otomí we collected data from native speakers of two rather different dialects, 31 from Santiago Mexquititlán (OS) and 28 from San Miguel de Tolimán (OT). In both cases, all speakers were interviewed on the basis of a series of questions about the sociolinguistic and cultural situation in the two communities. Furthermore, the same informants were asked to make a translation of 88 Spanish sentences into Otomí. In all we collected a corpus of 110,541 tokens. Table 8 gives an overview for the two dialects. In these totals, no distinction is made for the two types of elicitation. Table 8. Spanish borrowings in OS, OT and O N=31 Tokens Types TTR N=28 Tokens Types TTR Total (N=59) Tokens Types TTR

Target Language (OS)

Source Language (Sp)

Total

61693 (87.5%) 5344 (81.0%) 0.09

8841 (12.5%) 1255 (19.0%) 0.14

70534 6599 0.09

Target Language (OT)

Source Language (Sp)

Total

33277 (83.2%) 2672 (72.2%) 0.08

6730 (16.8%) 1029 (27.8%) 0.15

40007 3701 0.09

Target Language (O)

Source Language (Sp)

Total

94970 (85.9%) 8016 (77.8%) 0.08

15571 (14.1%) 2284 (22.2%) 0.15

110541 10300 0.09

There is an overall borrowing rate of around 14% in terms of tokens, with the OS variant somewhat lower (12.5%) and the OT variant slightly higher (16.8%). In both cases, the percentage for the types is considerably higher. This is probably due to some amount of variation in the spelling of the Spanish loanwords, not fully catered for by the lexicon we constructed. For the individual speakers the figures range from 6.7% (OS) and 10.7% (OT) to 20.1% (OS) and 26.0% (OT). The differences between the two dialects may be explained on the basis of the fact that San Miguel de Tolimán is a very small community, and has very intensive contact with the Spanish speaking world. When we look at the two data sources, we find that the amount of borrowing is less in the spontaneous data than in the translation data. This is shown in Table 9. We restrict ourselves to numbers of tokens, and make no differentiation for the two dialects.

216 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking Table 9. Spanish borrowings in Otomí for two types of elicitation N=59

Target Language (O)

Interviews Translation task Total

52447 42523 94970 (85.9%)

Source Language (Sp)

Total

7623 (12.7%) 7948 (15.8%) 15571 (14.1%)

60070 50471 110541

In Otomí not only content words are borrowed, but also a high amount of function words. The adoption of some of these function words makes that some synthetic constructions in the Otomí grammar are changing into analytical constructions. For example, the Spanish loanword ko (from con ‘with’) is replacing the Otomí suffixes -hu, -wi, -he and -’be (Hekking 1995). Furthermore it has been observed that basic constituent order VOS is frequently replaced by SVO, the basic order of Spanish (Lastra de Suárez 1994, 1997; Hekking 1995). Tone does not seem to have been affected so far. Looking in more detail at the parts of speech of the lexical borrowings, we find the following totals. Table 10. Spanish borrowings in O: parts of speech Verbs

Nouns

Adjectives Manner Advs.

Others

Santiago Tokens Types

400 (4.5%) 118 (9.4%)

3478 (39.3%) 158 (1.8%) 583 (46.5%) 43 (3.4%)

28 (0.3%) 10 (0.8%)

4777 (54.0%) 501 (39.9%)

344 (5.1%) 2856 (42.4%) 137 (2.0%) 115 (11.2%) 505 (49.1%) 53 (5.2%)

5 (0.1%) 4 (0.4%)

3388 (50.3%) 352 (34.2%)

Tolimán Tokens Types

As predicted by hypothesis H2.1.1, also for Otomí borrowing takes place according to the pattern N > V > A > MAdj. The contribution of the overall category ‘Others’ is very significant indeed. In terms of tokens it surpasses even the total of the four major categories. In terms of types only nouns have a higher percentage. As can be seen from table 10, there is no striking difference between the two dialects. For most analyses we will therefore aggregate the data of all 59 respondents. Apart from separate lexical entities, we have found a huge amount of Spanish phrases, which are not to be considered as code switches but as composite or frozen borrowings, because they appear in our corpora several times and for several informants. Most of them are noun phrases such as:

Spanish meets Guaraní, Otomí and Quichua

217

barryo kinto ‘neighbourhood five’, el beintisinko de julyo ‘the twenty fifth of July’, la mera verdad ‘the pure truth’ and rweda de fortunä, ‘wheel of fortune’. We will discuss some details in terms of lexical and grammatical borrowings in Sections 7.4.1 and 7.4.2, respectively. 7.4.1. Major parts of speech For the borrowings belonging to the four major parts of speech in Spanish we found the functional distributions given in Table 11. Table 11. Spanish borrowings in O: parts of speech versus function Syntactic Function N=59 HP HR MP MR

Verbs (V)

Nouns (N)

Adjectives (A)

Manner Advs. (MAdv)

(Tokens)

(Tokens)

(Tokens)

(Tokens)

725

37 6275

120 11 1 162

7

HP HR MP MR

(Types)

(Types)

(Types)

191

24 880

53 4 1 25

3

31 (Types)

10

As for the other languages, verbs are borrowed as the second largest of the four lexical classes, and exclusively in the prototypical function of head of a predication. Spanish verbs are always borrowed in the third person singular present (the stem form) and may then be accompanied by any Otomí verbal proclitic or suffix. Both the bare stem form is found, as in example (53a) and the diphthongued version some verbs get in Spanish when the corresponding syllable gets stress (53b). (53) a. bi

ma bi konta-wi yá kompañera nu-’u go PST.3 tell-1.INC POSS.3 friend DEM-3PL ot’-ar wela do-DEF.SG grandmother ‘She visited her girlfriends to tell them what her grandmother had done’ PST3

218 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking b. nu-ge bwelo-ga mi kwenta ke hä DEM-DEM grandfather-1 IMPRF.3 tell that yes mi ting-ar oro ’ne-r plata IMPRF.3 find-DEF.SG gold also-DEF.SG silver ‘My grandfather used to tell that he used to find gold and also silver’ Nouns are borrowed in high numbers, and overwhelmingly as heads of referential phrases. This is in full compliance with the rigid type assigned to Otomí, with its specialized noun class. Spanish nouns are always adopted by Otomí in the singular, and may be accompanied by any Otomí nominal proclitic or suffix. That we find Spanish adjectives, though not in large numbers, is also expected on the basis of the class 5 categorization of the language. What is striking, however, is that Spanish adjectives, unlike V and N do not specialize for their prototypical function, i.e. MR. Quite a large number (40% of the tokens, 64% of the types) function as predicates. They are accompanied by verbal (54a) or nominal (54b) morphology, often according to the treatment the corresponding vernacular word would get. (54) a. nu'bya dige-'bya ya bi m-bibo … now with.respect.to-now already PST.3=PST.3-smart ‘… nowadays they became smart …’ b. ya

na pobre i hinti mi pets'i nada poor and nothing IMPRF.3=have nothing ‘They were very poor and had nothing’ DEF.PL INT

When used as a nominal modifier in Otomí, Spanish adjectives appear in their bare form, typically with the masculine suffix -o.28 Interestingly, they may be found in prenominal position, following Otomí syntax, but also postnominally, following Spanish syntax. Finally, the borrowing of several manner adverbs, though not very frequent, goes counter to our predictions. The borrowing of adverbs is definitely not uncommon in Otomí. A further 663 Spanish adverbs (tokens; 61 types) of other categories (Time, Place) are found in our corpus.

Spanish meets Guaraní, Otomí and Quichua

219

7.4.2. Other parts of speech If we subtract the lexical elements found under Other in Table 10 above, we are left with a large number of grammatical borrowings: 7502 of 15,571 borrowed tokens, i.e. 48.2%. By far the most frequent category among these are Spanish prepositions. We found 3305 tokens for this category, or 21.2%, placing it second after the nouns. The majority of these are used in their function of relating a noun phrase to another constituent. Around a third of the prepositions are markers of a subordinate clause. In fact, these may be derived not from the Spanish prepositions as such but from the complex Spanish subordinators consisting of a preposition followed by the general subordinator que ‘that’, as in porque ‘because’ and para que ‘so that’. An example of this use is found in (55). 'yode xi hño ya jä'i (55) ñä nts'edi, pa da speak strong for FUT.3 hear 3 good DEF.PL person ‘Speak loud, so that the people hear you well’ After the prepositions follow coordinators such as i ‘and’, o ‘or’ and pero ‘but’ (7.5%), and discourse markers such as pues ‘well’ and este, a demonstrative used as a continuation marker (6.5%). The fourth category are subordinators like aunque ‘although’, porque ‘because’ and de que ‘so that’ (6.1%, 40 types). Relatively few pronouns are found, of four different subtypes: personal, impersonal, interrogative and relative (0.5%). Finally, we found 42 different numerals, totalling to 0.9% of the Spanish tokens. All values under 10 are among them. Coordinators, subordinators, discourse markers, numerals and pronouns are all used mainly or exclusively in their prototypical functions. Spanish articles, certainly a category in Otomí, are not attested in our corpus. To sum up, most predictions concerning grammatical borrowing got support from the actual data, the use of lower numerals being an exception. 7.5.

Conclusion

As in the case of both Guaraní and Quichua, borrowing hypotheses H1 and H2 find support in terms of the relative token and type frequencies, be it to a somewhat lesser extent than the other two languages. This is mainly due to the high frequency of borrowed prepositions, not only in terms of tokens but also in terms of types. An explanation may be that Otomí has several particles which function like prepositions, and has in fact a syntactic slot

220 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking for that category. This is in line with hypothesis H4.1. Given that Otomí leaves many relationships between noun phrases and other constituents unmarked, especially at the clause level, the wide variety of Spanish prepositions may provide an extra resource in this respect. The same may be said about the category of manner adverbs. An interesting phenomenon is the fact that borrowed adjectives are used not only as modifiers in noun phrases, but also as heads of predications. In terms of types this is even the prevalent use. Per word type, however, the function is almost always unique. Only three types are used both as MR and HP. These are blanco ‘white’, viejo ‘old’ and rico ‘rich’. Like other Spanish adjectives, they are regularly found in ‘headless’ noun phrases in Spanish, and may therefore be analyzed as nouns rather than adjectives, or as homonyms for that matter. So we may conclude that by far most adjectives which are borrowed specialize, be it not necessarily for their prototypical function, as the part of speech theory would predict in this case. The basic constituent order, VOS in Classical Otomí, is being replaced by SVO, which is the basic order of Spanish, a marked though existing order of the classical language. Finally, given the huge number of Spanish function words that are borrowed, such as prepositions, coordinators and subordinators, with the concomitant restructuring of subordinate clauses, and the ongoing change in basic clause order, we estimate that Otomí is situated somewhere between point 2 and 3 on the borrowing scale of Thomason (2001). 8. Comparing the results In the previous sections we analyzed the contact data we collected for three individual, typologically quite different languages, viz. Guaraní (PG), Quichua (EQ) and Otomí (O). As observed earlier, much of what happens in language contact situations follows the same more or less general patterns. The differences are quantitative rather than qualitative. Thus, comparison is an absolute necessity. Therefore, in this section we will compare the results of the preceding sections in the light of the hypotheses of section 4, and see to what extent the typological differences between the languages throw any light on the differences we may find. Let us first have a look at the overall number of borrowings. Table 12 displays the totals of tokens borrowed from Spanish found in the corpora of the three languages.

Spanish meets Guaraní, Otomí and Quichua

221

Table 12. Comparison between borrowed tokens

Source Target

Quichua (EQ) N=25

Guaraní (PG) N=38

Otomí (O) N=59

15098 (18.9 %) 64620 (81.1 %)

10056 (17.4 %) 47772 (82.6 %)

15571 (14.1 %) 94970 (85.9 %)

Although the differences seem to be slight on first view, they turn out to be significant on a 2 test at the 0.5% level. This would imply that in terms of language use, the presence of Spanish is greatest in EQ, somewhat less in PG and least in O. It is interesting now to compare the figures for the speakers with the lowest and the highest percentages of Spanish tokens for the three target languages. This is done in Table 13; some outliers have been taken out. Table 13. Minimum and maximum Spanish tokens Quichua (EQ)

Guaraní (PG)

Otomí (O)

4.0 % 27.0 % 18.9 % 8.92%

5.7 % 28.5 % 17.4 % 6.42%

6.7 % 26.0 % 14.1 % 3.97%

Mimimum Maximum Mean SD

So we see that, although the borrowing percentages are slightly different for the three languages, the minimum and maximum percentages do not differ much. The values for the standard deviation suggest that in the case of Otomí the borrowing percentages are somewhat closer to the mean than for the other two languages. Now let us look at the figures for the individual parts of speech. In Table 14 below we put the figures for the categories which make a more than marginal contribution to the total of the borrowings for at least one of the languages. Note that under the category Adv we have aggregated all adverbs, including MAdv.29 Outlying values are in bold. Table 14. Comparison of most frequent parts of speech

Art Adpos V DsMrk A

Quichua (EQ)

Guaraní (PG)

Otomí (O)

OS – OT

0.0% 0.5% 17.7% 0.6% 8.5%

19.4% 0.5% 18.3% 0.8% 7.4%

0.0% 21.2% 4.8% 6.5% 1.9%

0.0 – 0.0% 17.8 – 25.8 4.5 – 5.1% 7.2 – 5.7% 1.8 – 2.0%

222 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking

N Subj Conj Adv Complex REST

Quichua (EQ)

Guaraní (PG)

Otomí (O)

OS – OT

54.4% 1.6% 6.9% 3.4% 1.0% 6.3%

37.2% 4.6% 4.4% 2.4% 0.1% 5.0%

40.7% 6.1% 7.5% 4.5% 0.0% 6.8%

39.3 – 42.4% 6.8 – 5.3% 9.1 – 5.3% 4.6 – 4.3% 0.0 – 0.0% 8.9 – 4.1%

What is immediately clear from this table is that there are vast differences between the contributions of the respective parts of speech to the total. The real outlying language is Otomí; the differences between Quichua and Guaraní, though considerable, are less striking. This is in accordance with the typological correspondences and differences between the three languages. That the borrowing behaviour of Otomí speakers is not due to purely local processes gets support from the fact that the figures for the Santiago (OS) and Tolimán (OT) dialects, as given in the rightmost column, converge to a high extent. The only significant differences between the two are the extremely high number of prepositions in OT and the high number of conjunctions in OS. We think therefore that it is safe to assume that the differences are largely due to the differences in the grammatical make up of the target languages. The most extreme case are the articles borrowed by Guaraní, and already mentioned in Section 5.4.2. They may fill a gap in the sense that Classical Guaraní does not have articles. However, it is unlikely that this is in fact the reason for their presence. The same can be said of Quichua, for which we find not a single example for this category. It turns out that they are mainly used as a freestanding anaphor. In case of their use within a noun phrase, they may often be interpreted pragmatically, as demonstratives indicating stress or contrast. Only in a restricted number of cases they merely code definiteness. Second are the adpositions, which are borrowed abundantly by Otomí, but hardly by EQ and PG. They fall in step with the syntax of O, which has some adverbs with the function and in the syntactic position of prepositions. So Spanish prepositions may fill a functional gap in a language that on the whole leaves nominal relations unmarked and implicit. In EQ and PG, both postpositional, the borrowing of prepositions is arguably hampered for structural reasons. The third most striking difference is the very low number of verbs borrowed by Otomí. It is not clear to us why this is the case. All three languages, like Spanish have a specialized category for verbs. And the part of

Spanish meets Guaraní, Otomí and Quichua

223

speech type of O (high specialization) is in many respects closer to that of Spanish (maximum specialization) than the other two languages (both highly flexible). As we will see below, there seems to be no ‘compensation’ for this lack of borrowing from the other lexical categories, N and A. The next most remarkable difference between the categories in Table 15 is the use, in Otomí, of Spanish discourse markers, mainly the particle pues and the demonstrative este. Both are found very frequently in all varieties of spoken Mexican Spanish as well. They appear only very infrequently in the EQ and PG corpora, but they are much less characteristic of the colloquial Spanish of both Ecuador and Paraguay. The following exception is again for Otomí. The language borrows a very reduced number of adjectives in comparison to the other languages. EQ and PG have a flexible class for the typical nominal and adjectival functions, and might therefore be flexible with the acceptation of both N and A. Otomí specializes for these categories, but there is only a restricted, semi-open class of adjectives in Otomí itself. Most of the meanings for which there would be an adjective in Spanish, are expressed by a verb or a noun in Otomí. This may explain why there are apparently restrictions on borrowing them. One of the categories for which Quichua is an outlier is the high percentage of nouns that it borrows. Well over half of the total loans belong to this category, while both PG and O have around 40%. A high percentage of nominal borrowings may be indicative of an early stage of language change through contact. However, the overall amount of Spanish borrowings of both lexical and grammatical nature, which is the highest of the three languages under investigation, is strongly indicative of a later stage. Below we will contrast the amount of nouns borrowed only with two other lexical types, V and A, and see whether there may be an explanation for the relatively high number of Spanish nouns in Quichua. Finally, Quichua has a rather low percentage of subordinators among the borrowed tokens. The number of different types, nine, is the same as for Guaraní, though. All these are rather frequent in colloquial Spanish. On the prefinal row in Table 14 we find the total for complex borrowings. Only for Quechua do we find a number of them. By far the most of these are of the type found in (56) below, i.e. Spanish prepositional phrases with a clearly fixed meaning.

224 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking (56) a lo mejor ‘at best’ a tiempo ‘in time’ a ver ‘let us see’ de repente ‘suddenly’ por ejemplo ‘for example’

a lo primero ‘firstly’ a veces ‘sometimes’ de acuerdo ‘okay’ de todas maneras ‘anyhow’ por lo menos ‘at least’

If we look at the three major parts of speech V, N and A in isolation, and assume for the sake of argument that these are in ‘competition’ for the same semantic space in all languages, then we see a remarkable difference between the three languages again. The columns in the middle of Table 15 below show that the relative proportions of coverage for Quichua and Guaraní are relatively close to each other. This relativizes a bit, though not entirely the high percentage of nouns for Quichua in Table 14 above. However, for Otomí the figures are vastly different, with a strong overrepresentation of N at the cost of V and A. On the basis of these figures we might tentatively conclude that Otomí attracts referential material while Quichua and above all Guaraní tend to interpret borrowings more predicatively. The columns on the righthand side make it clear that for Quichua and Guaraní [V, N, A] represent 70–80% of overall borrowing, while in the case of Otomí they cover less than 50% of the total. So, also from this perspective EQ and PG differ strongly from O. Table 15. Proportion major parts of speech Part of Speech N V A TOTAL

Relative Percentages EQ 67.6% 22.0% 10.5% 100.0%

PG 59.1% 29.1% 11.8% 100.0%

O 85.9% 10.1% 4.0% 100.0%

Absolute Percentages EQ 54.4% 17.7% 8.5% 80.6%

PG 37.2% 18.3% 7.4% 72.7%

O 40.7% 4.8% 1.9% 47.4%

Now let us aggregate the figures for the individual parts of speech to totals for lexical and grammatical borrowings. We use the term ‘lexical’ for the classes V, N, A and MAdv plus all other types of adverbs that we found. The rest of the elements will be subsumed under ‘grammatical’. Table 16 provides an overview, again in terms of tokens, and rounded off to whole percentages.

Spanish meets Guaraní, Otomí and Quichua

225

Table 16. Comparison in terms of lexical versus grammatical

Lexical Grammatical

Quichua (EQ)

Guaraní (PG)

Otomí (O)

84% 16%

65% 35%

52% 48%

These differences are even more remarkable than the individual figures of Table 14 and 15. They reflect the mirror image of the order that we get when we rank these languages in terms of number of tokens borrowed. As indicated above the major cause of the high percentage for Otomí is the high number of Spanish prepositions. It could be argued that these elements are on the borderline between lexical and grammatical, or that some are more lexical and some more grammatical. When we add, for the sake of argument the prepositions to the lexical categories the figures for grammatical borrowing would drop to 15.5% (EQ), 34.2% (PG) and 26.9% (O), respectively. As we have already seen in Table 14, the high number of grammatical borrowings in Guaraní is mainly caused by the Spanish articles la and lo. They make up 55.9% of the total of this category. Quichua and Otomí do not borrow articles at all. When we would take out articles altogether from the calculations the total percentages for grammatical borrowing would be 15.5% (EQ), 18.4% (PG) and 26.9% (O). These two manipulations would still leave the languages in the reverse order of the one reflecting the total number of borrowings in Table 12. It seems to be obvious that these two measures – overall amount of borrowing and percentage of grammatical elements – are not necessarily positively correlated. There may even be a negative correlation in the sense that, at some stage of borrowing, the part of loans stemming from closed classes reaches a limit, whereafter the borrowing from the open classes, and above all nouns goes on, extending its percentage further. Whatever may be the precise meaning of this, the figures presented in the tables in this section must lead to a reconsideration of several of the hypotheses presented in Section 4. We think that these observations have implications for general borrowing scales such as the one in Thomason (2001). Possibly, factors such as level of bilingualism, and amount of code switching, briefly mentioned in the section on Guaraní, should be included as factors. To round of this section, we will compare the functions in which the lexical categories are used in the respective languages. As has already been pointed out in the sections on the individual languages, borrowed V and MAdv are virtually completely fixed, in all three languages, to their prototypical functions of HP and MP, respectively. Below, in Table 17, we find the relative uses for the categories N and A.

226 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking Table 17. Relative distribution over functions Quichua (EQ)

Guaraní (PG)

Otomí (O)

0.7 % 92.8 % 1.7 % 4.8 %

0.2 % 98.3 %

0.6 % 99.3 %

1.4 %

0.1 %

1.3 % 22.5 % 10.3 % 66.0 %

1.4 % 26.8 % 4.7 % 67.1 %

40.8 % 3.7 % 0.3 % 55.1 %

NOUN HP HR MP MR ADJ HP HR MP MR

For the nouns the differences are rather small. As to be expected, Otomí is the least flexible language of the three, and close to total rigidity in its application of Spanish nouns. But one would have expected more flexibility indeed for the other two languages, which do not have a specialized class of nouns. An explanation may be that nouns are borrowed with the more specific nominal meaning they have in Spanish rather than the more flexible type of meaning that non-verbal lexemes might have in EQ and PG. For the adjectives, the picture is as one would expect for EQ and PG: a rather high flexibility. The second most frequent use is as the head of a referential phrase, in line with non-verbal elements of these languages themselves. The picture for the – relatively few – Spanish adjectives borrowed by Otomí does not seem to support the assumption made above that this language has a tendency to interpret its borrowings referentially rather than predicationally. Around 41% of the borrowed adjectives are used as predicates. A final remark should be devoted to code switching. As indicated above we have distinguished between code switches and ‘real’ borrowings. The former have been left out of the calculations. However, there are considerable differences between the numbers of code switches in the three corpora. This is shown in Table 18. Table 18. Number of tokens in code switches

Tokens Percentage of real borrowings

Quichua (EQ)

Guaraní (PG)

Otomí (O)

3103 20.6 %

10500 95.8 %

1717 11.0 %

Spanish meets Guaraní, Otomí and Quichua

227

For Otomí the figure is rather low. However, in the case of Guaraní adding the code switches to the borrowings would virtually double the figures. This is strongly indicative of the bilinguality of the Paraguayan language community. 9. Conclusions In this contribution we have compared the borrowing phenomena in three typologically different languages as a result of contact with the same source language. On the basis of data collected according to a more or less uniform strategy we tested several hypotheses concerning borrowing which were derived from a theory on parts of speech and from the recent literature on language contact. Hypothesis 1, about pragmatically more or less outstanding markers, has received strong support above all from Otomí. Also in the other languages, we found only discourse markers at the extra-sentential level. No intrasentential topic or focus marker was found in anywhere. Hypothesis 2, which predicts that more lexical than grammatical elements should be found, was supported by Quechua and Guaraní, but only marginally by Otomí, unless we consider adpositions as lexical rather than grammatical elements. When we look at the respective subhypotheses, we observe the following. There is a strong tendency for open classes to be borrowed more easily than closed classes. This is the case for both lexical (H2.1.1) and non-lexical (H2.1.2) classes, with two notable exceptions. If borrowed at all, as for Otomí, adpositions rank at the top of the grammatical categories, as predicted, and definitely not at the bottom of the lexical categories. In that sense they fare better as grammatical categories. And articles, which are expected to rank at the lower end of the grammatical hierarchy, if borrowed, as in the case of Guaraní score higher than any of the other grammatical categories. But the hypotheses are applied under the assumption of ‘all things being equal’. However, things are not equal: we have seen that constraints originating from typological differences between the languages may explain several of the discrepancies observed, be it not all of them. Particularly the heavy borrowing of the Spanish article in Guaraní is puzzling. Possibly, the function a borrowed element gets in the target language plays a crucial role in the explanation of such cases rather than its original function in the source language. Hypothesis H2.2, i.e. that the amount of boundedness of source elements plays a role in their being borrowed, has not been considered here at all. Although we have quite some information on this from our corpus, which suggests that especially Quichua

228 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking has relatively many mixed complex forms, space prevents us from going into details on this here. Hypothesis H3 predicts that we would find many more separate lexical elements than complex structures. This is indeed the case, overwhelmingly so, especially when we disregard code switches. Only in Quichua do we find a fair number of construction-like borrowings. For most of them it may be assumed that they are borrowed as unanalyzable entities. As far as the hypotheses under H4 are concerned, we can say the following. H4.1, about head-modifier relationships finds support from the large amount of prepositions borrowed by Otomí as opposed to the very low numbers of Spanish prepositions which made it into both postpositional languages. As far as main clause orders are concerned (H4.2), we observed that Otomí shifts from VOS towards SVO order, both the basic order in Spanish, and a variant of Otomí. Also Quichua, rather rigidly SOV, shifts towards SVO, a rare though not an ungrammatical order in pre-Spanish Quechua. This is more or less in line with the hypothesis. The basic order for adjectives in Spanish is postnominal, with prenominal order rather infrequent and marked for most adjectives. Still, this does not prevent Quichua, with prenominal adjective ordering, to borrow relatively more adjectives than Guaraní. In this case, the hypothesis finds no support from typology. Finally, H4.3 finds support from many ‘doublings’ in especially Otomí, where the borrowed form is typically higher on the grammaticalization cline than the native element. The predictions made on the basis of Hengeveld’s part of speech theory found support in a more or less general sense. The borrowing characteristics in terms of the four central types V, N, A and MAdv is rather similar for Guaraní and Quechua, both type 2, and quite different for Otomí, type 5. However, when we look in more detail, quite a few discrepancies remain, especially the borrowing of relatively many nouns and adverbs and the low number of verbs by Otomí and the extremely high number of nouns by Quichua. Our major conclusion is that, although most of the hypotheses get support to some extent, the formal conditions have to be specified in much more detail to be of real value. Since they are mainly based on hypotheses from the literature on borrowing, our observations on these three examples of borrowing must have repercussions for some generally accepted ideas. Obviously, in order to confirm these insights, and derive a robust theory on borrowing, a more refined analysis of the data than we have made here is called for. In that future work, more attention should be paid to the role of code switching and the level of bilingualism of both the language communities and the individual informants.

Spanish meets Guaraní, Otomí and Quichua

229

Notes 1. 2.

3.

4. 5.

6.

7.

8.

For a recent overview of bilingualism in Latin America, see Escobar (2004). We follow here the convention of calling Quichua the Ecuadorian dialects that form a branch of the Quechua language family. Ecuadorian Quichua is classified as part of Quechua II (Torero 1964) or Quechua A (Parker 1963). The dialects of this subfamily share a number of traits with other varieties from Southern Peru, Bolivia and Argentina (Adelaar 2004: 185ff.). Obviously, for reasons of comparison there should also be languages with more or less the same typological characteristics. The scope of the current study prevents us from including these. For one of the languages, however, we will briefly look at the situation in two different dialects. Cf. Milroy (2003) for extensive argumentation on this view. One might claim that any change in a language and its grammar is eventually motivated by the fact that it is meant to improve its communicative function. This position would deny any autonomy to grammar. Croft (1991) has coined the term extreme functionalism for such a view, and argued convincingly against such a position for monolingual models of grammar. We would like to leave some room for grammar internal explanations of change as well, at least as a theoretical possibility. 1 = first person, 2 = second person, 3 = third person, ABL = ablative, ACC = accusative, ACMP = accompaniment, ADIT = additive, ADV = adverb, AFF = affirmative, ALL = allative, ART = article, BEN = benefactive, CAUS = causative, COM = comitative, CONJ = conjunction, COP = copula, DAT = dative, DEF = definite, DEM = demonstrative, DIM = diminutive, DIREV = direct evidence, DUR = durative, EXC = exclusive, FOC = focus, FUT = future, GEN = genitive, GER = gerund, HP = head of a predicate phrase, HR = head of a referential phrase, IMP = imperative, IMPRF = imperfective, INC = inclusive, IND = indefinite, INFER = inference, INST = instrumental, INT = intensifier, INTERP = interpelative, LIM = limitative, LOC = locative, MP = modifier of a predicate phrase, MR = modifier of a referential phrase, NEG = negative, OBJ = (in)direct object, PERF = perfect, PL = plural, POSS = possessive, PRES = present, PRO = pronoun, PROL = prolative, PST = past, PSTPRF = past perfect, PURP = purposive, REP = reportative, SG = singular, SUB = subordinator, SUPL = superlative, TOP = topicalizer. Cf. Sasse (1992: 19f.) for a complete scenario leading to language death. All these stages of bilingualism are reported for languages in Spanish America in Gordon (2005). Ruhlen (1991) lists 479 extinct languages from the area. This is not to say that all have been driven into extinction as a result of contact with Spanish. Similar views on borrowing are expressed in the work of Stolz and Stolz (1997, 2001). However, Stolz and Stolz (1996) do not consider functional aptness (or gap-filling) a major factor in borrowing at all. Another analysis that has been given to the borrowing of discourse markers, notably by Matras (1998), is that it is indicative of the restructuring of the discourse of the target

230 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking

9.

10. 11. 12.

13.

14.

15.

16.

17.

18.

language in the direction of the source language. Others, like Zimmermann (1992: 299–305), explain Spanish borrowing as mainly based on the prestige that this language has with respect to the target language. Thomason (2001: 93) rightly warns for jumping to conclusions about borrowing too easily in such cases. In the same vein, Comrie (1990: 3) talks about the ‘quite widespread myth’ concerning the typological stability of languages both diachronically and in the light of language contact that ‘must be explored’. See Bakker and Siewierska (2002) for an analysis of adpositions within the framework of Functional Grammar. Cf. Song (2001) for a recent introduction to language typology. Compare universal U3 in Greenberg (1963). According to Siewierska (1998: 493), from a sample of 171 languages, 63% of the VSO languages have SVO as an alternative order, and only 13% (also) SOV. We have collected these for part of the languages discussed here. However, in order to keep the three corpora comparable, we have left them out of our analyses below. Note that the notion type is used in its most simplistic way, i.e. the pure form, disregarding morphological complexity or homonymy. In other words, no lemmatization has taken place. This has a great influence on the interpretation of this measure, especially when comparing languages of different morphological types. Below, we will mainly be concerned with numbers of tokens. The program gives several other statistics, e.g. TTR2, a type-token ratio where the number of types is divided by the square root of the number of tokens. This figure is somewhat more stable and less dependent on the actual text length (cf. van Hout and Muysken 1994). Both devices are optional, and should be applied with caution. Due to the inherent ambiguity of lexical material no device as simple as the ones discussed here might ever take such decisions with near 100% certainty. All results should be checked on hindsight. E.g. the pattern match procedure scores around 90–95% security for Otomí, but considerably lower for Quichua. This may be due to the amount of overlap between the two phonological systems involved, to the extent that they are reflected in the spelling systems. Rather than providing a straightforward part of speech code for the target language as well, we decided to code its function in the target language syntax. As will be explained below, in Section 4, it is not always the case that certain parts of speech will be relevant for the lexicon of the target language involved. Three Guaraní-speaking groups originally inhabited the territory of present Paraguay: Kario, Tovat and Guarambaré Indians were the first Guaraní to contact the Spaniards to be finally absorbed in the mixing process. Itat, Tapé and Paranaguá Indians had their first contact with Europeans only through the Jesuit missionaries and were perhaps the ones who most suffered from Spanish exploitation, as they were reluctant to mixing. Finally, the Mbya group had

Spanish meets Guaraní, Otomí and Quichua

19. 20.

21. 22.

23.

24.

25.

26.

27. 28.

29.

231

only occasional contact with Europeans and thus preserved its culture and language to a great extent well into the twentieth century (cf. Trinidad Sanabria 1997; Meliá 1992). The absence of a caste system in Paraguay’s emerging colonial society left ample space for interracial and intercultural practices. Some of the first literary works written in Jopará date from 1867 when the satirical journals Cabichu’í and Cacique Lambaré were printed during the war against Argentina, Brazil and Uruguay. For full data on both censuses, visit the website www.dgeec.gov.py. With respect to the 1992 census, Guaraní-Spanish bilingualism shows an increase of over 10% and there is a corresponding decrease in Guaraní monolingualism of 12%. This mosaic of languages in the Northern Andes might have influenced on the configuration of Ecuadorian Quechua as a koineized variety and could shed light on the tendency of some Ecuadorian dialects to show higher degrees of Spanish lexical borrowing. Borrowing in some dialects can be massive and lead to the origin of mixed languages such as Media Lengua, reported only in Ecuador (Muysken 1985; Gomez Rendón 2005). The fact that Quechua was initially standardized with evangelization purposes and so disseminated by non-native speakers (Spanish clergymen) may be an important factor when it comes to explaining the presence of Spanish elements in all the varieties of Quechua. For a discussion of this and other relevant factors that may have led to the rapid spread of Quechua in the territory of present Ecuador, see Muysken (2000). Typically, dictionaries define colours with reference to objects known in the community’s sociocultural context. See, for example, the Webster dictionary’s definition for ‘white’, which resort to metaphors such as intensity of light, racial groups and the like. The hypothesis that Otomí is a derivation of the two Otomí words otho ‘without’ and mui ‘house’ is another example of the contempt in which the Otomís were held. Even the Spanish adjectival quantifier cada ‘each’ which does not inflect for gender is found several times in the ‘masculine’ form cado. There does not seem to be any correlation between the numbers of manner adverbs on the one hand and other adverbs on the other hand. Cf. Q: 88 MAdv + 432 other; G: 95 MAdv + 142 other; O: 33 MAdv + 663 other. It could be argued that in comparisons such as the ones in Tables 14 and 15 only categories should be compared which are indeed in competition for some semantic or grammatical space. However, certain categories may imply each other morpho-syntactically, such as articles and nouns, and therefore may have a positive or negative influence on their overall distributions among borrowings.

232 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking References Adelaar, Willem (in collaboration with Pieter Muysken) 2004 Languages of the Andes. Cambridge: Cambridge University Press. Aikhenvald, Alexandra 2002 Language contact in Amazonia. Oxford: Oxford University Press. Aitchison, Jean 1981 Language change: progress or decay. London: Fortuna. Andersen, H. 1988 Centre and periphery: adoption, diffusion and spread. In Historical dialectology, Jacek Fisiak (ed.), 39–85. Berlin /New York: Mouton de Gruyter. Andrews, Henrietta 1993 The function of verb prefixes in Southwestern Otomí. (The Summer Institute of Linguistics and the University of Texas at Arlington Publications in Linguistics 115.). Dallas: SIL. Bakker, Dik and Ewald Hekking 1999 A functional approach to linguistic change through language contact: the case of Spanish and Otomí. Working Papers in Functional Grammar 71: 1–32. Bakker, Dik and Anna Siewierska 2002 Adpositions, the lexicon and expression rules. In New perspectives on argument structure in functional grammar, Ricardo Mairal Usón and María Jesús Pérez Quintero (eds.), 125–178. Berlin/NewYork: Mouton de Gruyter. Bartholomew, Doris 2000 Intercambio lingüístico entre Otomí y Náhuatl. In Estudios de cultura otopame 2, Yolanda Lastra de Suárez and Noemí Quezada (eds.), 189–201. Mexico: Universidad Nacional Autónoma de México, Instituto de Investigaciones Antropológicas. 2004 Notas sobre la gramática. In Diccionario del hñähñu (Otomí) del Valle del Mezquital (Hidalgo), Luis Hernández Cruz et al. (eds.), (Vocabularios Indígenas 45.) México, D.F.: Instituto Lingüístico de Verano. Beck, David 2002 The typology of parts of speech systems: the markedness of adjectives. New York: Routledge. Büttner, Thomas 1993 Uso del quichua y el castellano en la Sierra ecuatoriana. Quito: Ediciones Abya Yala. Campbell, Lyle 1989 On proposed universals of grammatical borrowing. In Papers from the 9th international conference on historical linguistics, Henk Aertsen

Spanish meets Guaraní, Otomí and Quichua

233

and Robert J. Jeffers (eds), 91–109. Amsterdam /Philadelphia: John Benjamins. Cerrón-Palomino, Rodolfo 1987 Lingüística Quechua. (Biblioteca de la Tradición Andina.) Cusco: Centro de Estudios Rurales Andinos Bartolomé de las Casas. Cole, Peter 1985 Imbabura Quechua. London: Croom Helm. Comrie, Bernard 1990 Language universals and linguistic typology. Oxford: Blackwell. Corvalan G. and Germán de Granda (eds.) 1982 Sociedad y lengua: bilinguismo en el Paraguay. Asunción: Centro Paraguayo de Estudios Sociológicos. Croft, William 1991 Syntactic categories and grammatical relations: the cognitive organization of information. Chicago: Chicago University Press. 2001 Radical construction grammar: syntactic theory in typological perspective. Oxford: Oxford University Press. Dench, Alan Charles 1998 Yinkarta. München: Lincom Europa. Dietrich, Wolf 1995 El español del Paraguay en contacto con el Guaraní. Ejemplos seleccionados de nuevas grabaciones lingüísticas’. In Lenguas en contacto en Hispanoamérica: Nuevos enfoques, Klaus Zimmermann (ed.), 203– 216. Frankfurt a.M.: Vervuert. Dik, Simon C. 1997 The theory of functional grammar. Berlin /New York: Mouton de Gruyter. Ecker, Lawrence 1952 Compendio de gramática Otomí: Introducción a un diccionario Otomí-español. México, D.F.: Instituto Nacional de Antropología e Historia, Secretaría de Educación Pública. Escobar, A. M. 2004 Bilingualism in Latin America. In The handbook of bilingualism, Tej K. Bhatia and William C. Ritchie (eds), 642–661. Oxford: Blackwell. Evans, Nicholas and Toshiki Osada 2005 Mundari: The myth of a language without word classes. Linguistic Typology 9 (3): 351–390. Garvin, Paul and Madeleine Mathiot 1982 La urbanización del idioma Guaraní. Problema de lengua y cultura. In Sociedad y Lengua: Bilingüismo en el Paraguay. Vol 1., G. Corvalán and Germán de Granda (eds.), 27–41. Asunción: CEPAG. Goldberg, Adele E. 1995 Constructions: a construction grammar approach to argument structure. Chicago: Chicago University Press.

234 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking Gómez Rendón, Jorge 2005 La media lengua de Imbabura. In Lenguas en contacto en los Andes, Pieter Muysken and Hella Olbertz (eds), 39–57. Berlin: Vervuert Hispanoamericana. forthc. a Spanish grammatical borrowing in Paraguayan Guaraní: an overview. Manchester Project on Grammatical Borrowing. forthc. b Spanish grammatical borrowing in Imbabura Quechua: an overview. Manchester Project on Grammatical Borrowing. Gordon, Raymond G. (ed.) 2005 Ethnologue: languages of the world. 15th edition. Dallas: SIL. Online version: http://www.ethnologue.com/. Greenberg, Joseph Harald 1963 Universals of language: report of a conference held at Dobbs Ferry, N.Y. April 13 –15, 1961. Cambridge: MIT Press. Gregores, Emma and Jorge Alberto Suárez Savini 1967 A description of colloquial Guaraní. The Hague/Paris: Mouton. Guerrero 2002 El Códice Martín del Toro. De la oralidad y la escritura, una perspectiva Otomí. Siglos XV–XVII. Tesis de Licenciatura en Etnohistoria. ENAH. Gumperz, John Joseph and R. Wilson 1971 Convergence and creolization: a case from the Indo-Aryan/Dravidian border’. In Pidginization and creolization of languages, Dell Hymes (ed.), 151–167. Cambridge: Cambridge University Press. Haboud, Marleen 1998 Quichua y Castellano en los Andes Ecuatorianos: los efectos de un contacto prolongado. Quito: Ediciones Abya Yala. Hawkins, John A. 1983 Word order universals. Cambridge: Cambridge University Press. Hekking, Ewald 1995 El Otomí de Santiago M: desplazamiento linguïstico, préstamos y cambios grammaticales. Amsterdam: IFOTT. 2001 Cambios gramaticales por el contacto entre el Otomí y el español. In Lo propio y lo ajeno en las lenguas austronésicas y amerindias: procesos interculturales en el contacto de lenguas indígenas con el español en el Pacífico e Hispanoamérica, Klaus Zimmermann and Thomas Stolz (eds), 127–151. Frankfurt: Vervuert-Iberoamericana. 2002 Desplazamiento, pérdida y perspectivas para la revitalización del hñäñho’. Estudios de Cultura Otopame 3.3: 221–248. Hekking, Ewald and Dik Bakker 1998 Language shift and Spanish content and function words in Otomí. In Actes du 16e Congres International des Linguistes, Bernard Caron (ed.). Oxford: Elsevier Sciences.

Spanish meets Guaraní, Otomí and Quichua 1999

235

El Otomí y el español de Santiago Maxquititlán: Dos lenguas en contacto. Foro Hispánico 13: 45–74. 2006 Problemas en la adquisición de una segunda lengua: El Otomí frente al Español. In Dinámica lingüística de las lenguas en contacto, Claudine Chamoreau and Yolanda Lastra (eds). Hermosillo: Universidad de Sonora. Hekking, Ewald and Severiano Andrés de Jesús 1984 Gramática Otomí. Querétaro (México): Universidad Autónoma de Querétaro. Hekking, Ewald, Severiano Andrés de Jesús and Paula Santiago de Quintanar forthc. Diccionario bilingüe, explicativo e ilustrado Otomí-Español del Estado de Querétaro. México D.F./Querétaro: Instituto Nacional de Lenguas Indigenas (INALI) / Universidad Autónoma de Querétaro. Hekking, Ewald and Pieter Muysken 1995 Otomí y Quechua: una comparación de los elementos prestados del español. In Lenguas en contacto en Hispanoamérica: nuevos enfoques, Klaus Zimmermann (ed.), Frankfurt a.M.: Vervuert. Hengeveld, Kees 1993 Non-verbal predication: theory, typology and diachrony. Berlin /New York: Mouton de Gruyter. Hickey, Raymond (ed.) 2003 Motives for language change. Cambridge: Cambridge University Press. Jackson, J. 1974 Language identity of the Colombian Vaupes Indians. In Explorations in the ethnography of speaking, Richard Bauman and Joel Sherzer (eds), 50–64. New York: Cambridge University Press. Jiménez Moreno, Wigberto 1939 Origen y significación del nombre Otomí. Revista Mexicana de Estudios Antropológicos III: 62–68. Lastra de Suárez, Yolanda 1992 El Otomí de Toluca. México, D.F.: Instituto de Investigaciones Antropológicas, UNAM. 1994 Préstamos y alternancias de código en Otomí y en español. In Investigaciones Lingüísticas en Mesoamérica Carolyn MacKay and Verónica Vázquez (eds), México: Universidad Autónoma de México. 1997 El Otomí de Ixtenco. México, D.F.: Instituto de Investigaciones Antropológicas, UNAM. Matras, Yaron 1998 Utterance modifiers and universals of grammatical borrowing. Linguistics 36 (2): 281–331. Milroy, James 2003 On the role of the speaker in language change. In Motives for language change, Raymond Hickey (ed.), 143–157. Cambridge: Cambridge University Press.

236 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking Morínigo, Marcos Augusto 1982 Impacto del español sobre el Guaraní. In Sociedad y lengua: bilingüismo en el Paraguay, Graziella Corvalán and Germán de Granda (eds.), Asunción: Centro Paraguayo de Estudios Sociológicos. 1990 Raíz y destino del Guaraní. Asunción: Universidad Católica. Moravcsik, Edith 1978 Universals of language contact. In Universals of language, Vol. I, Method and theory, Joseph Greenberg (ed.), 95–122. USA: Stanford University Press. Muysken, Pieter 1985 Contactos entre Quichua y Castellano en el Ecuador. In Memorias del Primer Simposio Europeo sobre Antropología del Ecuador, Yanez Segundo Moreno (ed.), 377–472. Quito: Abya Yala. 1994 Media Lengua. In Mixed Languages: 15 case studies in language intertwining, Peter Bakker and Maarten Mous (eds.), 207–211. Amsterdam: IFOTT. 2000 Semantic transparency in Lowland Ecuadorian Quechua morphosyntax. Linguistics 39 (5): 973–988. Myers-Scotton, Carol 2002 Contact linguistics: bilingual encounters and grammatical outcomes. New York: Oxford University Press. Nordhoff, Sebastian 2004 Nomen/Verb Distinktion in Guaraní. MA thesis Universität Köln. Palancar, Enrique L. 2006 Property concepts in Otomí: a language with no adjectives. International Journal of American Linguistics 2 (3): 325–366. Parker, Gary 1963 La clasificación genética de los dialectos quechuas. Revista del Museo Nacional 32: 241–252. Rosenblat, Ángel 1964 La hispanización de América: el castellano y las lenguas indígenas desde 1492. In Presente y futuro de la lengua española: Actas de la Asamblea de Filología del I Congreso de Instituciones Hispánicas, 189–216. Madrid: [without publisher]. Ross, M. 2003 Diagnosing prehistoric language contact. In Motives for language change, Raymond Hickey (ed.), 174–198. Cambridge: Cambridge University Press. Ruhlen, Merrit 1991 A guide to the world’s languages. London: Edward Arnold. Ruiz de Montoya, Antonio 1993 [1640] Arte de la lengua Guaraní. Asunción: Centro de Estudios Paraguayos “Antonio Guasch”.

Spanish meets Guaraní, Otomí and Quichua

237

Sahagún, Bernardino de 1982 [1557] Historia general de las cosas de Nueva España, 5th ed. México: Porrúa. Saussure, Ferdinand de 1916 Cours de linguistique générale. Paris. Sasse, Hans-Jürgen 1992 Theory of language death. In Language death, Matthias Brenzinger (ed.), 7–30. Berlin /New York: Mouton de Gruyter. Schachter, Paul 1985 Parts of speech systems. In Language typology and syntactic description 1, Timothy Shopen (ed.), 3–61. Cambridge: Cambridge University Press. Siewierska, Anna 1998 Variation in major constituent order. In Constituent order in the languages of Europe, Anna Siewierska (ed.), 475–552. Berlin /New York: Mouton de Gruyter. Song, Jae Jung 2001 Linguistic typology: morphology and syntax. London: Longman. Stolz, Christel and Thomas Stolz 1996 Funktionswortentlehnung in Mesoamerika. Spanisch-amerinindischer Sprachkontakt. Sprachtypologie und Universalienforschung 49(1): 86–123. 1997 Universelle Hispanismen? Von Manila über Lima bis Mexiko und zurück: Muster bei der Entlehnung spanischer Funktionswörter in die indigenen Sprachen Amerikas und Austronesiens. Orbis 39: 1–77. 2001 Hispanicised comparative constructions in indigenous languages of Austronesia and the Americas. In Lo propio y lo ajeno en las lenguas austronésicas y amerindias: procesos interculturales en el contacto de lenguas indígenas con el español en el Pacífico e Hispanoamérica, Klaus Zimmermann and Thomas Stolz (eds), 35–56. Frankfurt a.M.: Vervuert. Suárez, Jorge Alberto 1983 The Mesoamerican Indian languages. Cambridge: Cambridge University Press. Thomason, Sarah G. 2001 Language contact. An introduction. Edinburgh: Edinburgh University Press. Thomason, Sarah G. and Terrence Kaufmann 1988 Language contact creolization, and genetic linguistics. Berkeley: University of California Press. Torero, Alfredo 1964 Los dialectos quechuas. Anales Científicos de la Universidad Agaria 2: 446–478.

238 Dik Bakker, Jorge Gómez Rendón and Ewald Hekking 2002 Idiomas de los Andes: lingüística e historia. Lima: Ed. Horizonte. Trinidad Sanabria, Lino 1997 La evolución histórica del Guaraní. www.datamex.com.py/Guaraní/ mrandeko/trinidad_evolucion_historica_del_ idioma_Guaraní.html. 1998 Polisíntesis Guaraní. Contribución para el conocimiento tipológico de esta lengua amerindia. Asunción: Intercontinental Editora. Urbano, Alonso 1990 [1605] Arte breve de la lengua Otomí y vocabulario trilingüe. R. Acuña (ed.). Ciudad de México: Universidad Nacional Autónoma. van Hout, Roeland and Pieter Muysken 1994 Modelling lexical borrowability. Language Variation and Change 6 (1): 39–62. Velazquez-Castillo, Maura 1995 Noun incorporation and object placement in discourse: the case of Guaraní. In Word order in discourse, Pamela Downing and Michael Noonan (eds), 555–579. New York: John Benjamins. 2002 Grammatical relations in active systems: the case of Guaraní. Functions of Language 9 (2): 133–167. Voigtlander, Katherine and Artemisa Echegoyen 1985 Luces contemporáneas del Otomí: Gramática del Otomí de la sierra. México, D.F.: Instituto Lingüístico de Verano. Yasugi, Yoshiho 1995 Native American languages. An areal-typological perspective. Osaka: National Museum of Ethnology. Zimmermann, Klaus 1992 Sprachkontakt, ethnische Identität und Identitätsbeschädigung: Aspekte der Assimilation der Otomí-Indianer an die hispanophone mexikanische Kultur. Frankfurt am Main: Vervuert.

French influence on the native languages of Canada and adjacent USA Peter Bakker and Robert A. Papen

1. Introduction At the Romancisation world-wide conference held in May, 2005 at the University of Bremen, Germany, the authors of this article read individual papers: “Michif and the case of intertwining” (Bakker) and “French influence on the Native languages of Canada” (Papen). Because of the considerable overlap in contents, the authors realized that a joint paper would be substantially better with regards to both subjects. In this paper we limit ourselves to lexical and some morphosyntactic borrowing from French by the Aboriginal languages of Canada and the adjacent parts of the USA. Our main focus will be on lexical borrowing but we will also deal with other results of contact between French and the indigenous languages of North America. We will not deal with place names and proper names – several Amerindian groups borrowed both family names and first names from French. We will discuss code-switching between French and various Algonquian languages in currently French-Algonquian bilingual communities in the Canadian province of Quebec. Both historical aspects (established older borrowings from French) and contemporary language use will be dealt with. The geographical area we deal with is mostly the Northern part of North America, i.e. Alaska, Canada and the northern part of the United States. We are well aware that there are also French speakers in Louisiana, and that French influence on Amerindian languages also took place in southeastern USA and along the river networks between Louisiana and the northern parts – in particular along the Mississippi and Missouri rivers – but that is beyond the limits of this paper. This paper is structured as follows. In section 2, we review the relevant aspects of the history of the French colonization of the territory, first in Acadia and New France, and later the Great Lakes area, then into the Great Plains and finally into the Northwest (present-day Northwest Territories). We will then attempt to give a few details on the various Native groups which

240 Peter Bakker and Robert A. Papen came into contact with the French, in both eastern and western Canada and conclude briefly on the current knowledge of French by the First Nations peoples of Canada. We then describe the demographic situation for Native languages in Canada, drawing specific attention to the existing languages, numbers of speakers, and the phenomena of language shift and language death. In section 3, we describe lexical borrowing, in a combination of geographical and linguistic overviews of the impact of French, mostly on the basis of the various available dictionaries of the pertinent languages. In section 4, we discuss more recent borrowing strategies, particularly as they pertain to the situation prevalent in bilingual Native communities in the province of Quebec. In section 5, we broach the various language mixing strategies that have been used by a number of language groups, such as codeswitching (Montagnais and Attikamek), mixing or fusion (Michif), pidginization (Chinook Jargon), jargonization (Slavey Jargon), and heavy lexical borrowing (French-Cree). We finally offer a few concluding remarks. 2.

A brief history of the French presence in Canada

Portuguese and Basque fishermen were the first Europeans to visit the Canadian coast. The first official European “discovery” of what is now Canada is attributed to John Cabot who sailed along the coast of present-day Newfoundland in 1497, followed by Verrazano in 1524 and finally by Jacques Cartier, who in 1534 sailed up the St. Lawrence River into the heart of what is now Quebec, but it took some 70 years for the first French colonists to arrive. 2.1. French colonization in the 17th and 18th centuries In 1605, Pierre du Guast, Sieur des Monts, established the colony of PortRoyal on Ste. Croix Island, on the river of the same name, which now separates the Canadian province of New-Brunswick and the American state of Maine, an area where Maliseet-Passamaquoddy (two closely related dialects), of the Algonquian language stock, is spoken (see maps 1 and 2). Of course, it is now well-known that Basque fishermen had traded with the local Indian tribes of the Gulf of St. Lawrence area throughout the latter part of the 16th century and Bakker (1989a) has shown that there existed a Basque-Algonquian pidgin during this period. The colony of Port-Royal was totally disbanded after only one year and moved further inland (pre-

French influence on the native languages of Canada and adjacent USA

241

sent-day Annapolis Royal), in what is now Nova-Scotia, where Maliseet and Mik’maq (also Algonquian languages) were spoken. So it can be asserted that the first Native languages to come into contact with French were Maliseet-Passamaquoddy and Mik’maq. In 1608, Samuel de Champlain established a colony at Stadacona (now Quebec City). At that time, the principal tribes living in the St. Lawrence River valley, at least from Stadacona westward, were the Huron-Wendat, who became staunch allies of the French, particularly in their incessant wars against their enemies, the Iroquois. Both these groups spoke a variety of Iroquoian languages, such as Huron (extinct) and Mohawk. In 1642, Paul Chomedey, Sieur de Maisonneuve, established the community of VilleMarie, which was eventually to become the city of Montreal. In time, other tribes living in the St. Lawrence River valley or the Saguenay River basin also came into close contact with the French: the Montagnais (Innu) on the north shore of the St. Lawrence and on the Saguenay, the Maliseet and the Abenaki on the south shore, and the Mik’maq in the Gaspé Peninsula; inland, the French met the Attikamek in the St. Maurice River basin, all of these languages being of the Algonquian stock (see Map 1). What drove the French ever westward was their insatiable demand for furs. This led them first to the Ottawa River valley, where they encountered other Algonquian-speaking tribes such as the Algonquin. Quite rapidly, they reached the Great Lakes basin, where they met a great number of linguistic groups, either of Iroquoian stock, such as the Mohawk, Oneida, Onondaga, Cayuga, Seneca etc., or of Algonquian stock, such as the Ojibwa, Odawa, Menomini, Fox, Illinois, Potawatomi, etc. By the 1670s, the whole area of the Great Lakes had been explored and a number of trading posts had been established (at Detroit, Green Bay, Chicago, etc.). At the same time, however, the English established a number of trading posts in Hudson Bay, to the north. In 1686, d’Iberville defeated the English navy in Hudson Bay and all of the Northwest became French territory. In 1710, Port-Royal in Acadia was captured by the British and in 1713, by the Treaty of Utrecht, France was forced to cede Newfoundland, Acadia and the Northwest; it kept Cape Breton Island and all of its inland colonies. In 1735, Pierre Gaultier de Varennes, Sieur de la Vérendrye, and his sons reached and explored the Plains, west of Lake Superior, eventually traveling as far west as the Rocky Mountains. As they traveled westward, the French explorer-traders came into contact with numerous other tribes, many speaking Algonquian languages, such as Saulteaux, a dialect of Ojibwa and various dialects of Cree in what is now Ontario, Assiniboine and Dakota of the Siouan family, in what is now Manitoba. Further west,

242 Peter Bakker and Robert A. Papen they encountered the Blackfoot, and most probably the Atsina, the Arapaho and the Cheyenne, all of Algonquian stock. In what is now Alberta, they also came into contact with the Sarsi, who speak an Athapaskan language and as they began to explore the vast regions to the northwest (present-day Northwest Territories) they encountered a variety of tribes, speaking Athapaskan languages such as Chipewyan, Slavey, Dogrib, Carrier, Beaver, etc. (see Map 1). Quite rapidly, the French men who remained in the West to trade with the Indians tribes took local brides. Thus numerous children were born of mixed parentage, who came to be known as Métis (also spelled Metis – with the final -s pronounced or not: may-tiss or may-tee). The members of this new population were generally bilingual, speaking both their father’s language, French, and their mother’s Indian language. In 1759, the French forces at Quebec were defeated by the British at the Battle of the Plains of Abraham. Four years later, France ceded all of its colonies in North America, except for two minuscule islands off the coast of Newfoundland (Saint-Pierre and Miquelon). All of Louisiana, as well as Martinique, Guadeloupe and their dependencies in the Caribbean, were ceded to the British. In 1779, so as to guarantee a steady supply of furs from the West, a number of Scots businessmen from Montreal created the fur-trading company that was to be called the North-West Company. This company, whose upper echelon were always English-speaking, mostly hired French-speaking Canadians and later on French-speaking Métis to paddle the canoes, haul the merchandise and generally do the heavy work. The result, of course, was that French rapidly became the lingua franca for much of the fur trade in most of the northern part of North America. 2.2. The spread of French in the 18th and 19th centuries From the very beginning of the French presence in the New World, the desire for trade with the local Amerindians was not the only force motivating men and women alike. The early settlers also had a fervent desire to Christianize the local population. Thus numerous French-speaking missionaries rapidly founded missions all over the Great Lakes basin, and later, in the West and in the Northwest and finally in the Arctic. Often, they proselytized in French, although it is known that these missionaries usually quickly learned the local languages. As we shall discuss in section 3.7, the influence these missionaries had on some Aboriginal languages is quite important.

French influence on the native languages of Canada and adjacent USA

243

2.3. The spread and decline of French in the 20th century During the early part of the 20th century, the Catholic Church was quite active in developing missions among the local Indian and Inuit tribes, throughout Canada, establishing itself on or near many of the Indian reservations that had been and were still being created. In many cases, the missionary priests, brothers or nuns built not only churches but also schools, orphanages and hospitals. During the first half of the century, the federal government decided on a campaign to “de-Indianize” the various First Nations groups and created local boarding schools for Indian children, where they could “learn the White people’s ways”. Of course, the Native languages in these schools were forcibly repressed. Not surprisingly, many of the missionaries working for the Lord in the mission churches, orphanages, hospitals and boarding schools were French-speaking, and came either from Europe or from Quebec. The religious and medical services as well as the education given were most often conducted in the French language. Many thousands of Amerindians and Métis in Canada thus were obliged to learn French as a second language. Of course, since this French was mostly learned in classrooms, the variety of French they were exposed to was a rather “standard” (either Canadian or European) variety. In order to distinguish between this variety of French, which many First Nations persons still speak, or at least understand, and other varieties of French, we will refer to it as “Mission French”, for lack of a better term. In Canada (except for Quebec), the use of French by Native peoples is steadily declining. Table 1 details the knowledge of English and/or French for speakers of Algonquian languages, as well as for those of the Athapaskan, Iroquoian, Salishan and Siouan language families. The table shows that, according to the 2001 census, First Nations people living outside Quebec rarely know French. Not surprisingly, the majority of speakers of Algonquin, Attikamek, and Montagnais – who reside in Quebec – also know French. Some 22% of Cree speakers profess to have knowledge of French. Of course, more than 13,000 Cree live on the eastern shore of James Bay, in Quebec territory, where they have access to education in French (at least partially). Exceedingly few speakers of Athapaskan, Iroquoian, Salishan or Siouan languages have any knowledge of French.

244 Peter Bakker and Robert A. Papen Table 1. Knowledge of French and/or English by First Nations populations (Statistics Canada, 2001) Languages Aboriginal languages Algonquian languages Algonquin Attikamek Blackfoot Cree Malecite Mik’maq Montagnais/ Naskapi Ojibwa Oji-Cree Athapascan languages Iroquoian languages Salishan languages Dakota/Sioux

English Neither English and French nor French

Total

English only

French only

187,670

154,250

11,805

6,540

15,165

130,630

106,585

10,765

445

8,295

1,445 4,615 2,745 72,880 760 7,230 9,705

285 0 2,700 65,280 740 6,845 1,605

655 3,465 0 190 10 0 6,445

445 115 30 3,255 15 285 625

65 1,035 15 4,165 10 95 1,035

21,005 9,725 16,955

19,975 8,725 15,890

0 0 20

145 15 115

880 990 925

475

425

10

45

0

2,600

2,575

0

10

10

3,905

3,785

0

35

85

2.4. Amerindian groups currently in contact with French Canada is officially a bilingual country. The only officially bilingual province is New Brunswick, and Quebec is officially monolingual French. There are communities of French speakers in the other provinces and in the northeastern part of the United States, and of course in Louisiana, but only in Quebec is there currently a real impact from the French-speaking community on Aboriginal groups. Outside Quebec, there may be individuals knowing both French and Aboriginal languages, but not on a massive level. The only exception being the more or less complete linguistic assimilation of the Houma Indians to the French-speaking Cajun population in Lafourche Parish in Louisiana (Papen and Rottet 1997; Rottet 2001).

French influence on the native languages of Canada and adjacent USA

245

In Canada, around 3% of the population identifies as Aboriginal (approximately 1 million persons). Of these, 185,125 people reported to speak an Aboriginal language in the 1991 census. In 1996 that number was 190,165, of whom 138,105 used the Aboriginal language at home. As the home is the most natural location for the use of Aboriginal language, it appears that only around 14% of Aboriginal people speak an Aboriginal language at home, and that this number is declining. There are between 53 and 70 Aboriginal languages spoken in Canada (depending on different subdivisions and the dialect/language distinction), and only three of them are reported not to be endangered (Cree, Ojibwa and Inuktitut). Even though French is the only official language in the province of Quebec, L1 English speakers represent some 20% of the population. In addition, there are speakers of Amerindian languages and a plethora of immigrant languages. Historically, some of the native groups in Quebec have allied with English speakers (in particular the Mohawk, the Cree, the Naskapi and the Inuit), which means that they speak English as a second language, or are shifting towards English rather than French. Three Amerindian language families are represented in Quebec: (A) The Eskimo-Aleut family, which covers a range of dialects and languages spoken from Greenland to Siberia. In Northern Quebec, the Eastern Canadian Inuktitut variety is spoken. In 1996, 26,700 of the 41,578 Inuit population (64%) of Canada reported to be speakers of Inuktitut, with a low of 10% in Labrador and a high of 92% in Quebec. (B) The Iroquoian family, mainly represented by Mohawk, is spoken in the southern area, mostly in the St. Lawrence River valley, near Montreal. Other Iroquoian languages are also spoken in Ontario. The second language (for many, the first language) of the Mohawk in Quebec is English and knowledge of French is not widespread. The Huron language was once widespread, but it died out in the early 20th century; today the Huron-Wendat speak French as an L1. There are attempts to revive the language as reconstructed from historical documents. (C) Algonquian languages are by far the most numerous in Quebec. The family is widespread geographically also beyond Quebec, and is represented in Quebec and New Brunswick by Mik’maq, Maliseet-Passamaquoddy, Western Abenaki (all of the Eastern Algonquian branch), Algonquin (an Ojibwa dialect) and several dialects of the Cree-Montagnais-Naskapi continuum: Attikamek, Southern and Northern East Cree, Montagnais and Naskapi. Of these groupings, only the Montagnais, the Attikamek, the Algonquin and some of the Mik’maq are today generally speakers of both French and an Aboriginal language.

246 Peter Bakker and Robert A. Papen Table 2. Aboriginal language families in Canada, with declared number of L1 speakers (based on Norris 1998) Algonquian family

146,635

Cree Ojibwa Montagnais-Naskapi Oji-Cree Mik’maq Blackfoot Attikamek Algonquin Maliseet Inuktitut Athapaskan family Dene/Athapascan* Chipewyan Carrier Dogrib Chilcotin South Slavey North Slavey (Hare) Babine Gwich’in(Loucheux) Kaska Tutchone Beaver Sarsi Sekani Tahltan Han Tagish Iroquoian family Mohawk Oneida Onondaga

87,555 25,885 9,070 5,400 7,310 4,145 3,995 350 655 27,780 c. 21,420 10,310 1,455 2,190 2,085 705 2,620 290 500 430 400 400 300 50 c.40 35 7– 8 2 c. 950 585 200 50 –100

Cayuga Seneca Tuscarora Salishan family Thomson Shuswap Okanagan Comox Lillooet Halkomelem Sechelt Bella Coola Straits Salish Squamish Siouan family (Dakota) Tsimshian family Tsimshian Nisga’a Gitxsan Wakashan family Heiltsuk Nootka Kwakw’ala Nitinat Haisla Tlingit isolate Haida isolate Kutenai isolate

40–60 25 7– 8 c. 2,390 595 500 400 400 200 200 40 20 20 15 4,295 c. 1,850 750 700 400 c. 745 300 200 190 30 25 145 240 120

* These figures are based on the 1996 Canada Census where a number of respondants gave Dene or Athapascan as their L1, rather than a specific Athapascan language. Most Chipewyan refer to their language as Dene rather than Chipewyan.

French influence on the native languages of Canada and adjacent USA

247

Many Algonquin and Mik’maq have shifted to French (sometimes English), and the Huron, Abenaki and Quebec Maliseet have all shifted to French. All Canadian Aboriginal languages are endangered, even the three languages deemed “safe” in some reports. Although there are still children learning Cree, Ojibwa, Montagnais, Attikamek, Naskapi and Inuktitut, there are probably many more Cree and Ojibwa children who learn English as their L1. Also, even though most Montagnais or Attikamek children learn their L1 at home, all formal learning at school is done in French. This implies that the great majority of the Montagnais and the Attikamek are (relatively) fluent bilinguals. Languages spoken outside Quebec belong to five additional families: the Athapaskan, Siouan, Salishan, Wakashan and Tshimshian groupings, and three isolates, Kutenai, Haida and Tlingit. Numbers of Canadian speakers can be found in Table 2. 3. Loanwords from French in Amerindian languages1 In this section we will discuss lexical loans. In quoting forms, we will preserve the spelling from the sources, unless specified otherwise. Map 2 shows the number of French loanwords for those languages for which we have data. The actual number will be higher. The French loans are quoted as they appear in the sources, and in some cases, when discussing phonetic detail, in IPA. Brown (1999) is a study of loanwords from European languages into Amerindian languages. He took a list of 77 more or less typical European words or concepts as his point of departure, and investigated which of those were expressed by loanwords in Amerindian languages, and from which languages. He only took into account those languages for which he could find translation equivalents for at least 31 of the 77 terms (Brown, 1999: 17). The terms cover imported food (apricot, barley) and animals (cat, sheep), new institutions (school, store), new technology (candle, clock/watch, key) and new measures (hour, mile, Wednesday). We limit our discussion to North America. Of the 11 Arctic languages investigated, only one (“Inuit Eskimo”, as spoken in Quebec-Labrador) has French loans. Of the 19 languages of the Subarctic, 13 have French borrowings. Of the 26 languages of the northeast of North America (covering also the Great Lakes area), only three are free of French loans. All of the 13 Northwest Coast languages and the two Plateau languages contain probable or certain French borrowings, whereas only one (plus three doubtful) of the

248 Peter Bakker and Robert A. Papen 16 Plains languages has French loans. Nine of the 11 languages of the Southeast have French loans, but none from the Great Basin, the Southwest or California. The regional patterns are clear: most French loans are found in areas where the French settled (northeast, southeast), and the areas where French fur traders and missionaries were active (Subarctic) also have their share. At first sight, the Northwest Coast seems to be odd, as the French were only marginally present there. Many of the French loans in that area may have entered these languages from the Chinook Jargon (CJ) interethnic language. For Carrier (interior British Columbia) however, CJ played no major role in its adoption of French words, as shown by Prunet (1990). Most loans are found in Algonquian and Athapaskan languages. The Algonquian family covers several dozens of languages, spoken from the east coast to the Rocky Mountains (see maps 1 and 2). The languages belong to several groupings and are usually not mutually intelligible. The earliest mention of the use of French words dates from the early 1600s in the Maritimes. The French visitor Lescarbot reported that local Natives were unable to pronounce certain French words (Bakker 2005). It cannot be established whether these words actually became part of the language at that time. The first integrated loans are reported from an area where French was not a contact language. Geary (1945) argued that nasaump in Natick/Narragansett of New England is a borrowing from French la soupe ‘soup’. The word is also found in 19th century sources of Algonquin, Mik’maq and Montagnais, spoken much more to the north. Also, some other words like napopi and nsôbân are argued to be French (la boubaisse (?), la soupane ‘corn mush’). Before discussing patterns of lexical borrowing, a word on Canadian French is in order. We will then deal with different geographical areas where French has had some lexical influence, moving roughly from east to west. 3.1. Clarification on the term “Canadian French” In the literature, French as spoken or used in Canada is typically referred to as “Canadian French”, just as “Belgian” or “Swiss” French is used to refer to the French spoken in Belgium or in Switzerland. However, for Canada, at least, this obfuscates a much more complex situation. First, historically there have always been two distinct vernacular varieties of French in Canada: the French developed in Acadia, referred to as Acadian French (AF), and the French developed in what was then New France. This latter French

French influence on the native languages of Canada and adjacent USA

249

was eventually exported westward to what was to become Ontario and the Prairie provinces as well as to the New England and Midwestern states of the USA. Many linguists today refer to this French variety as “Laurentian French” (LF); of course, in Quebec, the language is referred to as “français québécois”. Moreover, as in many language situations, a local “standard” (spoken and written) form of the language has evolved. In Canada, this standard is quite close to Standard European French (SEF), even though the lexicon is, as expected, different in many respects, as are some aspects of pronunciation. Even English loans tend to be different from those typically used in France (France footing vs. Canada jogging; France foot(ball) vs Canada soccer). We refer to this variety as Standard Canadian French (SCF), since it is used everywhere in Canada (for example, it is the variety used by the national media). Most often, what is described as “Canadian French” or even “Québécois French” (e.g. Orkin 1971; Walker 1984; Dumas 1987; Hewson 2000, etc.) is in reality the spoken vernacular variety spoken in Quebec and its diaspora; we shall refer to this variety as Popular Laurentian French (PLF). Grammatically, PLF is similar to European Popular French in many respects and it is specifically in pronunciation that the differences are to be found: – – – – – – – – –

high vowels are laxed in stressed checked syllables (except /v, z, , v/); vite, rude and bouche are pronounced as [vt], [ d] and b] respectively; long and lengthened open stressed vowels, both oral and nasal, are diphthongized: père [pa], pâte [p
 t], peur [pœ], pente [pãu¯t], etc.; high vowels can be voiceless or even deleted in unstressed syllables: participer [pat(i)s(i)pe]; /
/ in final position is most often realized as []: Canada [kanad
]; the digraph oi is pronounced [we] or [w] in forms such as moi, toi, soi, as [w] in forms such as mois, bois, trois and as [w
 ] or [wa ] in words such as soir, boire, etc.; in rural areas, // is regularly opened to [æ] or even to [a] before a /rC/ cluster: verte [væt]; dental stops become alveolar affricates before high front vowels: tu dis [tsydzi]; in some rural areas, the voiced velar stop /g/ is palatalized to [j] in some words such as gueule [jœl] and / / is velarized or even glottalized: jamais [xam] or [am]; final consonant clusters are usually simplified (typical also of vernacular EF): table [tab], poste [ps], arbre [ab], etc.

250 Peter Bakker and Robert A. Papen –

–

[l] is regularly inserted before vowel-initial verbs if the word preceding is vowel-final: ça l’a pas de bon sens ‘it makes no sense’. In fact, this intrusive ‘l’ is also found in a number of compounds: lampe à l’huile ‘oil lamp’, etc.; initial /l/ of definite articles and pronoun clitics la, les, lui, as well as final /l/ of il, elle and final /R/ of the preposition sur ‘on’ typically delete. This gives rise to various vowel sequences (hiatus) which French tends to avoid. The solution in PLF is to fuse the vowel sequences into long vowels: sur la table ‘on the table’ becomes [satab] which contrasts with [satab] sa table ‘his/her table (Papen 1998a).

For obvious reasons, it is this variety of French rather than SCF that has had the most impact on the various Amerindian languages of Canada.

3.2. Lexical borrowing in Mik’maq and Maliseet. Mik’maq speakers have been in contact with French since the early 1600s. The oldest European loans are half a dozen Basque loans (Bakker 1989b; 2005). Today, English is the main contact language of the Mik’maq, even for some residing in Quebec. There are some 80 lexical loans from French that we identified in DeBlois and Metallic (1984). In addition, many personal names go back to French (Piel < Pierre). The loans cover a number of semantic domains: religion (patliàs < patriarch ‘priest’; lapaltnewei < le baptême ‘holy water’; guljiewei < croix ‘cross’, if not from Basque gurutze); European cultural items (leglans/laGalans < la grange ‘barn’, magasan < magasin ‘store’, mulin < moulin ‘mill’, lapulusan < la prison ‘prison’), including imported animals (tèsipow < des chevaux ‘horse’, lattòllaw < le taureau ‘bull, stud horse’, pelgwet < perroquet ‘parrot’), card playing terms (pig < pique ‘spade’, sapatis < Jean-Baptiste ‘jack’); measurements (latìm < la dime ‘tithe’, galgie < quart ‘quarter’, greetings (adiu ‘adieu’, pusù < bonjour ‘hello’, punàne < bonne année ‘Happy New Year’). Mik’maq is also one of the very few languages displaying borrowed verbs (albeit with Mik’maq verbal flexional endings): sàsèwit p/t/k, v/f > p) and were often borrowed without the typical determiner: kalu < carreau ‘diamonds (card suite)’, talep < trèfle ‘clubs (card suite)’, pik < pique ‘spades (card suite)’, puti < bouteille ‘bottle’, pinekan < vinaigre ‘vinegar’, anisheliu < ange ‘angel’, akalishau < Anglais ‘English(man)’, kape/kawpi < café ‘coffee’, etc. However, the great majority of nouns were later borrowed with a French article, definite or indefinite (rarer), singular or plural. The borrowed determiner loses all of its grammatical and semantic values, as is the case for Mik’maq or Maliseet. Table 3 gives a number of examples: Table 3. Montagnais nouns borrowed from French (adapted from Drapeau 1980) Montagnais loan

Gloss

French form

‘pepper’ ‘cabbage’ ‘pancake, disc’

poivre chou crêpe

‘beer’ ‘potato’ ‘cellar’

bière patate cave

With plural indef. article ‘des’ tapuepal teshu tekalep With fem. sing. def. article ‘la’ lapiel lapatat lakap

French influence on the native languages of Canada and adjacent USA Montagnais loan

253

Gloss

French form

‘plate’ ‘bleach’ ‘referee’

assiette eau de javel arbite

‘can’ ‘tomato’ ‘Band council’ ‘cement’ ‘unemployment’

canisse tomate conseil de bande ciment chômage

With sing. pre-V. article ‘l’ lashiet loshapel larbit With sing. def. article ‘le’ lekaniss letomat lekaussei lessima leshumash

Drapeau (1980) also shows that all more recent borrowings, including all spontaneous loans, always begin with le- and are usually not phonologically adapted. In fact, there now exist a number of doublets, where older speakers provide a phonologically adapted borrowed form (with or without an article reflex) while younger speakers produce an equivalent loan always with initial le- and without phonological adaptation: older lakalabat versus more recent lekravat < la cravatte ‘necktie’. As well, older Montagnais speakers will use a Montagnais form where younger speakers use spontaneous loan equivalents, always with le- and rarely phonologically adapted: kalukunau vs. legato < gateau ‘cake’, ushkatiapi vs. lekarot < carotte ‘carrot’, tsistemau vs. lesigaret < cigarette ‘cigarette’, kamukuat ushkatiapi vs. lebet < bette ‘beet’, etc. In other words, it is now possible to recognize any recent borrowing from French in Montagnais since it will always begin with le- and it will generally not be phonologically adapted. To our knowledge, no other North American Amerindian language has used this (conscious?) strategy as a marker of “borrowing”. Of course, as we shall see in section 5.1, since most younger Montagnais are now bilingual, they currently have recourse to another “borrowing” strategy: code-switching, with a huge impact on the language. 3.4. Lexical borrowings in Ojibwa, Menomini, Attikamek and Cree Despite the fact that Ojibwa and Cree were both, like French, lingua francas of the fur trade and the transportation industry, sometimes in the same territories, French loans are relatively uncommon in Ojibwa and Cree, with the exception of the Algonquin dialect of Ojibwa, spoken in the southern

254 Peter Bakker and Robert A. Papen Quebec-Ontario border area. Cuoq’s (1886) dictionary of Algonquin contains 36 French loans, 14 of which are reported for modern Golden Lake Algonquin (Aubin 1981, 1999). Other Ojibwa dialects have fewer French loans, and most of these seem to form a subset of the Algonquin loans. Also the forms of the loans seem to indicate that the loanwords traveled from east to west. There are some remarkable similarities between Algonquin loans and those of Mik’maq or Maliseet; for instance Algonquin anjeni < ange ‘angel’, Mik’maq ansalèwit, Ojibwa anjeni, Montagnais anasheliu, Atikamek aaeri. Most borrowings are nouns, and cover the same semantic domains as the other languages: religion and cultural innovations in the domains of food, clothing, and implements. Attikamek, spoken in central Quebec, has some 45 integrated loanwords from French according to Béland (1978). Cooper (1945) had reported briefly on his fieldwork on Attikamek in the mid-1920s, but did not report on French influence. Béland also lists less than a handful of English loans. These are all integrated loans, adjusted phonologically and morphologically to Attikamek. Some are also part of compounds, such as pataki-manco: ‘ladybug’, from French pataque ‘potato’+ Att. manco: ‘insect’ (cf. French [bête à] patate), and tekarep-kitotta:kan ‘record player’ (lit. ‘instrument that makes pancakes make noise’, < des crêpes ‘pancakes, record’). A couple of features set Attikamek apart from other languages that borrowed from French. First, there are two borrowed verbs a:ka:eh- < engager ‘to hire someone’ and ka:ro:ke < carreau ‘diamonds’ ‘to play cards’. Second, there is a great diversity in the way words are integrated: two words retain a reflex of des, three retain one of du, six retain les/le, three appear with na (< la), four with the reflex of à la and eight have been borrowed in their bare forms. Four nouns have an integrated French locative preposition + article: aripa < à la barge ‘barge’, arapo < à la poche ‘pocket’, araap < à la chambre ‘bedroom’, and arames- < à la messe ‘mass’ (only in compounds).With na we find namina:ss < la mélasse ‘molasses’(only in compounds), na:pa:ne- < la farine ‘flour’ (only in compounds), napyer < la bière ‘beer’. The l > n change is unexpected since from an Attikamek perspective one would expect l > r, as in the examples above. This makes it very likely that these words were in fact borrowed from Algonquin, where the /n/ reflex is indeed found in the same words. The following nouns seem to be derived with a plural article: rikka:rat < les carottes ‘carrot’, rikkirak < les claques ‘overshoe, snow-boot’, riccik < les tuques ‘toque’, ripot < les bottes ‘boot’, ritta:tt- < les tartes ‘pie’ (only in a compound). Two of these nouns are usually plural (carrots, overshoes/

French influence on the native languages of Canada and adjacent USA

255

boots) but the other two are not necessarily plural (toque, pie), making a French les reflex problematic. The /ri/ form could reflect the difficulty in realizing the distinction between singular /l´/ le and plural /le/ les or, less likely, it could be a compromise form indicating borrowing, as in Montagnais (3.3). The latter argument would leave unexplained the fact that eight of the unambiguous French roots appear without any article at all, or that some dozen words retain article reflexes other than les/le. Some are probably recent borrowings such as pa:na:n < banane ‘banana’, but pwete < boîte ‘box’, to:ro:w < taureau ‘bull’ and patak < pataque ‘potato’ are probably quite old. Finally, there is a medial (roughly, an incorporated noun) -komess-, ‘friend’ which seems to be too close to French commis ‘clerk’ to be due to chance, though the semantics are somewhat problematic. Starting from Algonquin, and even in a few cases from Mik’maq, there are chains of loanwords that have spread from east to west. The word for ‘Englishman’ is borrowed from French ‘Anglais’ in Mik’maq agalasiew (Newfoundland angalsheou; Pierronet 1800), Montagnais akalishau/agaleshou, Algonquin aganeca, Attikamek ekarea:w, Moose Cree âkalâsiw, Plains Cree akayas-, Menomini saakanaas etc. Valentine (1994: 492) studied the spread of some lexical items through the Ojibwa dialect area. There is “a substantial amount of post-contact vocabulary that is quite specific to the area of Algonquin and Nipissing”. Some of these are from French: omoodayaabik, obooday, moodaabik < la bouteille ‘bottle’, naapaane, gaapaane, aapaane < la farine ‘flour’, transoo < trente sou ‘quarter’, dibweban < du poivre ‘pepper’, badak < pataque ‘potato’. The words mooniyans < Montréal ‘whiteman’ and nitii < le thé ‘tea’ are only found in Saulteaux, and shared with Cree. Valentine (1994: 500) suggests for the latter “maybe a southern route of trade?”. The Menomini, who today live in Michigan and Wisconsin, were hardly in direct contact with the French, yet the language displays at least a dozen French loans and a few from English. As well, Menomini borrowed and calqued extensively from neighboring Algonquian languages, notably from Sauk-Fox, Potawatomi and Ojibwa. Apart from the card playing terms (Haas 1968) piik < pique ‘spades’, kaanoow < carreau ‘diamonds’, tans < trèfle ‘clubs’, natuap < un atout ‘trump’, we find koohkoos < cochon ‘pig’, koohkoop < concombre ‘cucumber’, snepaan < ruban ‘ribbon, silk, satin’, suuniyan < sou ‘coin’, moswn < mouchoir ‘shawl’, anaamehaaw < à la messe ‘s/he practices the Christian religion’ – all of these also found in Ojibwa/Algonquin (Bloomfield 1962: 23, 33–34, 36). It is likely that some of these loans are also found in other languages of the Great Lakes area.

256 Peter Bakker and Robert A. Papen The number of French loans in Cree is relatively low. Pentland (1982) identifies some thirteen French borrowings, covering technical innovations: (latwel < la toile ‘canvas’) and domestical animals (torôw < taureau ‘bull’). Most are shared with Chipewyan, others with Algonquin, some are unique: lamwêl < la moelle ‘marrow, lepôs < le pouce ‘inch’. 3.5. Lexical borrowings in Salishan, Wakashan, Sahaptian languages and Kutenai To our knowledge, there is no study of French loans in Salishan languages, though a few French loans do exist in these languages. Jan van Eijk (p.c.) informs us that most French loanwords came into the languages via Chinook Jargon, and the semantic domains covered are imported animals such as cow, pig and sheep, further religious terms such as priest, apostle, bishop and some other innovations such as table and salt. Chinook Jargon is also responsible for a number of French loans in Wakashan (British Columbia) and Sahaptian languages (Oregon). In the latter, we find words such as lecok < le coq ‘hen, rooster’, lépois < les pois ‘peas’, lesac < le sac ‘sack’ chalamet < calumet ‘pipe’, lakamine < la gamelle ‘soup’. There are also a number of French loans in Kutenai (an isolate), and French names are or were used as well (Garvin 1947, 1948). The loans are limited to religious vocabulary. French mythology and fairy tales also spread to some of the Salishan groups (Boas 1925), and also to other groups, notably Algonquian (Lang, 1991). Remarkably, even the name of the French-Canadian trickster Ti-Jean < Petit Jean ‘Little John’ was borrowed into some Salishan languages, e.g. as Buchetsá < (Jean)-Baptiste ‘John (the Baptist)’. Boas (1925) and Lang (1991) assume that the voyageurs, the canoemen working for the fur trade companies, were responsible for the spread of French folklore through North America. 3.6. French loans in Akwesasne Mohawk Mohawk is an Iroquoian language. Akwesasne Mohawk is the only dialect of Mohawk, and perhaps also the only Iroquoian language, that has directly borrowed from French. The Akwesasne Mohawk live on a reservation straddling the borders of the provinces of Quebec and Ontario and the state of New York. They moved there from coastal New York at the instigation of French-speaking Jesuit priests in 1669 and hence were in contact with

French influence on the native languages of Canada and adjacent USA

257

French language and culture for an extended period. Later, English influence became dominant. Bonvillain (1978: 32) remarks that “loans are restricted in Mohawk”. Several dozens of Mohawk personal names are loans from French, and in addition, words from three semantic domains: food, implements and names of countries. This may be an exhaustive list: lasós < la sauce ‘sauce’, lapahpÚt < la barbotte ‘catfish’, teló < taureau ‘bull’, páloet < brouette ‘wheelbarrow’, atenopíl < automobil (or English?) ‘car’, jolamø < Allemand ‘German’, kanakyø < Canadien ‘French Canadian’, pápa < papa ’father’, sotál < soldat ‘soldier’, shikalét < cigarette (or English?) ‘cigarette’. Bonvillain believes most of the loanwords to be of relatively recent origin because of the presence of labial consonants in some of them. Mohawk does not have labial consonants such as /p b m f v/, and these are usually replaced with velars. Generally, French loans undergo more adjustment to Mohawk phonology than do English loans. These French loans are only found in Akwesasne, not in the other Mohawk communities. For example, Huot (1948) studied lexical acculturation at the Six Nations Reservation in Ontario, and reported no French loanwords. As early as the 19th century, some Mohawk traveled west with the fur trade and settled in Alberta and British Columbia (Kelly Lake). Their descendants have shifted to Cree and/or English, and it cannot be established whether their language was influenced by French in earlier generations.

3.7. Ile-a-la-Crosse and surroundings: French words in a Métis Cree variety Ile-a-la-Crosse is one of the oldest non-First Nations settlements in the Canadian West. It was established in 1776. It became a local meeting place, centered around a trading post, a church and later a regional hospital. Until the late 20th century, however, many families stayed “in the bush”, and lived off the land, supplemented with some supplies from the local store and other institutions. Families increasingly moved into town in the late 1900s, in order to be closer to the school and the church. The church, the hospital and the school were run by Catholic, French-speaking nuns and priests. This has had a great impact on the local language. The community is a very strong Métis community, with a minority of non-Métis, often teachers, priests and hospital personnel. There are also Métis communities farther north (Buffalo Narrows) and south (Green Lake, Beauval and others) where varieties of the same language as in Ile-a-la-Crosse are spoken. The linguistic situation in Ile-a-la-Crosse is quite complex due to strong variation and changes in the community. The information in this section is

258 Peter Bakker and Robert A. Papen based on Bakker’s fieldwork in the late 1980s, Ahenekew’s (1997) dictionary, a video (Gabriel Dumont Institute 2002) and the community website (www.michif.dev.kcdc.ca), all providing different kinds of data. Locally the language was and is called nîhiyawîwin, the Northern Cree word for ‘Cree’, or in English Cree. There were also French speakers (both Eastern Canadian and European missionaries), as well as speakers of the particular Métis variety of French (Papen 1998b, 2004). Not surprisingly, a range of French words, both from the mission and from Métis French, penetrated the local Cree language. Ahenekew’s dictionary, made with the help of local elders, marks all French-derived words with a special symbol. Roughly 650 entries, out of an estimated 10,000 entries listed in the dictionary, are marked as derived from French. Almost all of the French-derived words in the dictionary also have Cree alternatives. However, the 650 entries can be reduced to around 320 French roots. There is an increasing tendency to replace French nouns with English nouns among all generations. The video mentioned above is in Cree, interspersed with many English words but next to nothing from French. A few elderly speakers use no French (or English) words at all – claiming that this was due to their lack of formal education – whereas most elderly people do use French words, but they are also able to avoid them, replacing them with Cree or English. In the last decade, people started to call their language Michif. The Cree and French elements differ significantly from the Michif variety discussed in 5.2, strongly suggesting an independent genesis. The current denominator Michif stresses the common bond with the Métis in southern Saskatchewan and Manitoba. The impression one gets is that Ile-a-la-Crosse probably had some speakers of Michif, Mission French and Northern Cree, combined to different degrees. Cree, and later English, started to dominate. For younger people, the situation is a little different. Most families shifted from Cree to English in the late 1900s, whereas Cree had been the dominant language in the families. Today there are no more children dominant in Cree, and English is the dominant language of the community. English loans are frequent, but few are listed in the Ahenekew (1997) dictionary. 3.8. French borrowings in the Northern Athapaskan languages It is sometimes said that Athapaskan languages are less prone to borrowing than other languages (e.g. Haas 1968: 170). This is a myth that perhaps goes back to Sapir (1921: 196):

French influence on the native languages of Canada and adjacent USA

259

The Athabascan languages of America are spoken by peoples who had astonishingly varied cultural contacts, yet nowhere do we find that an Athabaskan dialect has borrowed at all freely from a neighbouring language.

However, there are indeed borrowings from European languages in most Athapascan languages. Alaskan and Canadian Athapaskan languages have borrowed from French and Russian and more recently also from English. Tanaina (Alaska) may be the most extreme example, with 20 borrowings from Eskimo and as many as 300 from Russian (Krauss 1983, 1996), and less than a handful from French. Some Athapaskan languages are quite rich in French loanwords. Not all languages have been documented to the same extent, and we did not have access to all the relevant data. Much of this discussion is based on Krauss (1983), supplemented by Rice (1989) for Slavey, Prunet (1990) for Carrier, Dremeaux (2003) for Gwich’in, Haas (1968) and Pentland (1982) for Chipewyan. The total number of French word-types in Northern Athapaskan languages is ca. 130 (see also Map 2). There may be many more, however; Bill Poser (p.c.) identified some 175 distinct French loans in one variety of Carrier alone. The five languages with the highest number of French borrowings are Carrier (80–175), Slavey (58), Hare (40), Chipewyan (23) and Gwich’in (21). Almost all of these are nouns and fall into the semantic domains of religious terms, trade goods, tools, household activities and food. There is a striking overlap between the words that all these languages borrowed. Only Carrier, Hare and Slavey have some unique borrowings that are not attested in the other languages. This suggests mostly indirect borrowing between indigenous languages, rather than direct borrowing from French into these Athapascan languages. There are several additional arguments for neighboring languages (or a third source; see below) as the sources for these loans, mostly based on phonological developments. First, almost all nouns are borrowed with the definite article (e.g., lëgëshuu in Han, lago~hshroo in Gwich’in, lígoshõ in Hare < le cochon ‘pig’), whereas others are consistently borrowed without the article. Two nouns (meaning ‘sugar’ and ‘dish towel’) attested in two or more languages are always borrowed without the article. In only two of the French words documented in more than one language, is there variation: sOldat < soldat ‘soldier’ in Carrier but lísaldá in Hare and Slavey, and she~gwëläh in Han but leshekolé in Hare and Fort Good Hope Slavey. The sequence in the French words roi ‘king’ (Carrier), poivre ‘pepper’ (Han, Gwich’in), la soie ‘silk’ (Carrier, Hare, Bearlake, Slavey,

260 Peter Bakker and Robert A. Papen Dogrib) always has the reflex /we/, suggesting a PLF or Métis French source, since it is pronounced /wa/ in SEF and SCF. The word for ‘spoon’ shows palatalized reflexes of /kw/ in all three languages, suggesting the same source, ultimately /tSwiyr/ rather than /kwiyr/. Both French and Athapaskan languages have nasal vowels, but the French nasal vowels have consistently become oral in several words (l’argent ‘money’, le cochon ‘pig’, torchon ‘dish towel’), and the word for ‘key’ (la clé) is nasal in the Yukon River languages Gwich’in, Han, Northern Tutchone. A further indication is that French words are adjusted in similar ways as Athapaskan words: speakers know that certain phonemes or processes characterize the differences between their own language and that of their neighbors, and the borrowed forms display these conceived differences. For example, Dogrib displays final consonant loss compared to Chipewyan (Chipewyan bes, Dogrib be ‘knife’), and this is also the case in the French loanwords: lames < la messe becomes lame in Dogrib ‘church’. Furthermore, both the form and the content of the loans in some cases show an overlap with those in Algonquian languages. In short, these French words are generally not borrowed directly from French in the different Northern Athapaskan languages, but rather from other Athapaskan languages, or from a third source. These data render vacuous any statement claiming lesser proneness to borrowing in Athapaskan languages. Do the semantic domains reveal in what context these French words entered the languages? Some of them undoubtedly came via the Catholic Church. This is most clearly so for Carrier, where 24 of the 80 published loans are of a religious nature. All other languages, however, show just two or three religious words (‘pope’, ‘candle’ and ‘mass’), and hence the Church is less important as a source. Krauss (1983) and Dremeaux (2003) suggest that trade around the forts was responsible for most loans. This, however, is not the impression one gets upon closer scrutiny of the semantic domains. Household words (‘dust cloth’, ‘towel’, ‘spoon’, ‘broom’, ‘frying pan’, etc.) predominate and suggest domestic chores, perhaps in the context of an acculturating institution for “civilizing” the First Nations, such as residential schools or Catholic institutions. This is suggested by the existence of loans for agricultural activities, cooking, needlework, house maintenance and social activities. There are also items that could have played a role in a trading context (‘thread’, ‘shawl’, ‘carbine’, ‘bullet’, etc.), but not a single one exclusively. In fact, all of the French terms would play a major role in household chores and social contexts. A domestic context is therefore much more likely responsible for the transmission of these French words between languages than trade or religion.

French influence on the native languages of Canada and adjacent USA

261

The Carrier language, spoken in the central part of British Columbia, probably has the highest number of French loanwords. No fewer than 80 words were still in use in the late 1980s (Prunet 1990) plus a number of personal names, but the actual number may be as high as 175 (B. Poser, p.c.). Some 60 words have agglutinated determiners – all definite articles – and 20 words appear without determiners, in most of which a definite article would be unnatural or impossible in French. In addition, Carrier has (direct or indirect?) borrowings from Cree, even for some animal names (‘cat’, ‘pig’). Chinook Jargon, the trade language of the Northwest, was widely spoken in the coastal areas of Oregon and British Columbia (Grant 1996), but according to Prunet, the words were not borrowed from Chinook Jargon. On phonological grounds, and with arguments based on the fact that Carrier contains many French loans not found in Chinook Jargon, Prunet (1990) argues convincingly that the bulk of the French vocabulary was borrowed directly and not via Chinook Jargon. For seventy of the eighty loans, a Chinook Jargon source appears impossible. The source of the borrowings “must have been employees of the Hudson’s Bay Company or Catholic missionaries” (Prunet 1990: 490). Prunet suggests that there was influence both from “Canadian French” and “Standard French”. He does not mention Métis French, and he seemed to be unaware of the fact that the Métis spoke and still speak their own distinct variety of French. It is very well possible that the “Canadian French” words were actually borrowed from Métis French speakers. Words like l bdag ‘potato’, liled ‘milk’, fizigab ‘name of first rifle’ seem to be from Métis French la patak, li lt (rather than a spelling pronunciation as suggested by Prunet), and fizi. Furthermore, the possible reflexes of Métis French plural article /li/ < les in words like liz anyo ‘onions’, li ga ‘gloves’, etc., suggests the possibility of Métis French as a source. A change of /le/ to /li/ is unlikely in Carrier, as Carrier has both /e/ and // as phonemes. A religious term like lezwif ‘Jew(s)’ shows a different plural element, suggesting a missionary origin. A third argument is the reflexes of dental stops in front of /i/. As noted by Prunet (1990: 490–491), in Canadian French (both in SCF and PLF) these dental stops change into alveolar affricates (/ts, dz/), but in Carrier they do not do so in the French names of the days of the week, the name ‘Baptiste’ and the word for ‘communion’, lkarisdi < l’Eucharistie (see also Poser 2004 for a discussion of palatalization in Carrier and possible Métis influence). Most other words probably did not necessarily enter Carrier by way of Métis French speakers.

262 Peter Bakker and Robert A. Papen 3.9. French in Athapaskan-based pidgins: Slavey Jargon and Jargon Loucheux Slavey Jargon and Jargon Loucheux are two pidgins of the far Northwest. Neither of them has been well documented before they became out of use in the early 20th century. Slavey Jargon was used in the MacKenzie River district, Jargon Loucheux along the Yukon and Peel Rivers, both tributaries of the MacKenzie river. Travelers and missionaries regularly mention it, but seldom quote it or document it. Bakker (1996a) and Bakker and Grant (1996) summarize what was known about these pidgins. In the meantime, more data have been discovered with regards to Slavey Jargon, but they have only partly been analyzed (Dremeaux 2003; Mishler, in press). Emile Petitot (1889: 292–293) summarized the distinction between the two pidgins as follows (our translation from French): The Loucheux Pidgin [le jargon loucheux], used in the Yukon and among the Dindjié [Loucheux, Gwich’in] of Peel River, is made up of scraps of French, English, Chipewyan, Slave, Dindjié and even Cree vocabularies. It is not used in the MacKenzie, where broken Slavey [le jargon esclave] predominates. The latter comprises only French, Cree and Dènè Slavey elements.

The question to discuss here is of course: which French words, and in which forms were they used in the two pidgins (if they can be kept distinct at all)? We may assume three sets of sources that could provide information. First, linguists’ fieldnotes on the pidgins. But there are only a few lines available (see below). Second, there are a limited number of quotes in missionary and travelers’ accounts – notably Petitot’s work. Third, there is a set of loanwords in Athapaskan languages that could be attributed to Slavey Jargon (Krauss 1983). The historical quotes give an impression of either a simplified form of Slavey, or a form of spontaneous mixture between (mainly) French and Slavey. Some short quotes can be found in Bakker (1996a), for example, (the Slavey elements are underlined): (1)

Oh! foolish boy! foolish boy! Néni t’as perdu la cabosse, hein? Néni t’as peur, nétcha? ‘Oh! foolish boy! foolish boy! You have lost your head, hey? You have great fear?’ (Petitot 1889: 295)

The word néni means ‘you’ and nétcha is ‘big’ (K. Rice, p.c.). Additional quotes in Slavey Jargon (from Petitot 1891: 20, 24) do not contain any

French influence on the native languages of Canada and adjacent USA

263

French. More research is necessary to find out how French elements function in Slavey Jargon and in Jargon Loucheux. Krauss (1983) provides an overview of French words in northwest Athapaskan languages, and attributes these to Slavey Jargon. He lists up to 60 French words borrowed in different languages (see 3.8), and in fact there are more. However, Krauss presents no evidence for a connection between the pidgin and the French loans. There is only one source available for Slavey jargon based on actual fieldwork, rather than circumstantial evidence such as loanwords from French, or indirect evidence such as quotes in memoirs. Dremeaux (2003) quotes Slavey Jargon material collected by C. Mishler from Sandy Roberts in 1973, all in all some ten brief sentences, some 25 distinct words. One word is English (yuw ‘you’), seven words are French: kombrah < comprend ‘understand’, labahrt < la barque ‘boat’ or barge, demos < dimanche ‘Sunday’, fini < fini ‘finish’, salahbri < saloperie ‘trash’, sahgriy mowjiy! < sacré maudit! ‘lit. sacred damn’ – a swear word whose phonetic form suggests a Métis French origin. The other words all seem to be Athapaskan. She also quotes from another source, in which one French word is used, a verb: l’aime ‘to love, to like’. It is striking that most of these French words from Slavey Jargon utterances are not found in the lists of French borrowings into Athapaskan language. Furthermore, since only three forms are nouns, this represents a proportionally very high number of verbal loans. In the Athapaskan languages, all identified loans are nouns (Krauss 1983). Brown (1999) has argued that (at least in the American context) words are seldom borrowed across language family boundaries, but instead between related languages or from an “economical” language, i.e. a lingua franca, pidginized or not. When we compare the data with what little is known about interethnic languages in the Athapaskan area, this seems to be confirmed. Close to the area to the south of the Athapaskan area, there was another pidgin spoken, called Chinook Jargon or Chinook Wawa. This pidgin also had a significant French component, up to 200 words (Lang 1995).

3.10. French words in Chinook Jargon The proportion of French words in Chinook Jargon, the pidgin of the Northwest coast, has been between 11 and 19% of the documented vocabulary, or in absolute numbers between 34 (1841) and 153 (1894).

264 Peter Bakker and Robert A. Papen Table 4. Evolution of the lexicon of Chinook Jargon (based on Grant 1996: 1192– 1193)

Chinook Nuuchahnulth English French others/unknown

1841 (252 words)

1865 (497 words)

1894 (1394 words)

1924 (547 words)

111 (44 %) 18 (7 %) 41 (16 %) 34 (13 %) 48 (19 %)

205 (41 %) 24 (5 %) 67 (13 %) 94 (19 %) 107 (22 %)

214 (15 %) 25 (2 %) 570 (41 %) 153 (11 %) 432 (31 %)

169 (31 %) 24 (4 %) 250+ (46 %) 87 (16 %) 17 (3 %)

Lang (1995) has studied the French component of Chinook Jargon of 1846, by which time some 124 words were documented for the pidgin. Some are articles of trade and travel (lagóm < la gomme ‘pitch, glue’, lahash < la hache ‘axe’), others are food items brought from the east (lasuk < le sucre ‘sugar’, lawen < l’avoine ‘oats’, lipá < le pain ‘bread’), local wild food (larasín < la racine ‘roots’, lúkutchi < les coquilles ‘shellfish’), smoking (lapíp < la pipe ‘pipe’, lahb < l’herbe ‘surrogate tobacco’), horse terminology (lablid < la bride ‘bridle’, liwhet < le fouet ‘whip’, sandele < cendré ‘roan’), local and imported animals (lilú < le loup ‘wolf’, limuto < le mouton ‘sheep’, latig < le tigre ‘lynx’). In contrast to the other languages, Chinook Jargon also contains very basic vocabulary from French, such as body part terminology (latét < la tête ‘head’, lima < la main ‘hand’) and a number of exclamations, such as abá < eh bien ‘well then’, mashie < merci ‘thank you’. There is also a significant religious vocabulary – not surprising, taking into account the fact that CJ was also used as a language in the mission – even in print: esepek < l’(arch)évêque ‘bishop’, sesu kli < Jésus Christ ‘Jesus Christ’, lesapot < les Apôtres ‘apostle’. When we compare the forms and the semantic domains of the French words in Chinook Jargon and those in the Algonquian languages further east, we can only conclude that they derive from different sources. There is not much overlap between the two areas, and sometimes different terms are used for the same item (CJ and Athapaskan languages cosho < cochon ‘pig’, elsewhere kookoos; lepan/lipá < le pain ‘bread’, elsewhere karakona < Iroquois, or lagalet < la galette). Sometimes the form diverges: CJ labooti < la bouteille ‘bottle’, in the other languages without agglutinated article: Mik’maq pútai, Algonquin/Ojibwa obotei, Montagnais puti, Ile-a-la-Crosse botay. Sometimes CJ has a form without an agglutinative article where other languages have one: CJ melass < melasse ‘molasses’, Attikamek naminaass, Algonquian naminas, Montagnais lamilash, Cree lamênâs, lamêlas

French influence on the native languages of Canada and adjacent USA

265

(but Mik’maq mêlasis-). The French body part terms and equestrian terminology are only found in Chinook Jargon. Of course, some words are the same, but it is difficult to prove a common origin if the form does not deviate in a particular way in two sources. One specific CJ term lúkutchi < les coquilles suggests a possible Métis French origin, because of the presence of the palatal affricate for the French velar stop, which is typical of Métis French (Papen 1998b, 2004). Nevertheless, these facts suggest that the Chinook Jargon network was largely populated by different people than those in the network of the Prairies or the Great Lakes – and different again from the networks further north in Athapaskan-speaking territory.

4. Discussion: the spread of French loans through North America With regards to the geographical presence of French borrowings in North America, we can make a number of observations. Languages in direct contact with French show impact from that language – which is not surprising. What is more surprising is the fact that languages that have never been in systematic contact with French also show French borrowings. Strikingly, often the same borrowings reoccur, sometimes with a particular form; the unavoidable conclusion is that these items were diffused from one Native group to another, without direct contact with French. We have discussed some examples in the preceding section. A case in point here is the word for ‘flour, bread’, derived from French la farine. Some of the attested forms from Quebec-Ontario are: napané (Mississauga Ojibwa 1801; Chamberlain 1891), napanin (Algonkin 1886; Cuoq 1886), naapaaneenuk (Eastern Ojibwa 1891; Chamberlain 1891), naapaanenak ‘flour’ (Golden Lake Algonquin; Aubin 1999). Blackfoot, on the western Prairies, has napayin(i) ‘bread’, suggesting a direct French source (Baldwin 1994). The word is not attested in Carrier, but it is found all the way around the Pacific coast. Chinook Jargon has sapolil of unclear origin, Upper Chehalis Salish has saplil, Yakima saplil and similar forms exist in Salishan, Wakashan and Penutian languages (Thomason, on Chinook List, 23 October 1999). If this is from la farine, the sa- element remains unexplained. Changes from /n/ > /l/ and /r/ > /l/ are conceivable in the context of the Northwest, where several languages changed /n/ > /l/ or /d/ in the past 200 years (Kinkade 1985). Tlingit apparently has a cognate sakwnein. If this set of words is indeed from la farine, the word traveled from the Great Lakes to the Pacific, and the change from naapaanin to the

266 Peter Bakker and Robert A. Papen form sakwnein (p > kw, l > n are regular adaptations in Tlingit) shows a connection with la farine which is remotely possible, but only traceable after multiple changes. Brown (1999: 127) shows a similar list with words for ‘tea’ in Ahtna, Northern Tutchone, Gwich’in, Slavey, Chipewyan, Beaver, Carrier, Yakima and Kalispel. Brown surmises that two pidgins were responsible for the spread of these and other French words: Chinook Jargon and Slavey Jargon. He shows that 11 words from French are used in Chinook Jargon with the definite articles. Subsets of these found their way into no fewer than 22 languages of British Columbia, Northern Oregon, Southern Alaska – roughly coinciding with the spread of the pidgin (Grant 1996). However, only three of the languages with leti < le thé as their word for ‘tea’ were part of the Chinook Jargon area, and for the rest, Slavey Jargon must be held responsible. The most extreme example of an unusual form is the word for ‘ribbon’, found in the same idiosyncratic form from the Great Lakes to the Pacific Northwest (Brown 1999: 125). This word is attested, from east to west, as sénipEn in Mississauga Ojibwa, seenipa in Odawa Ojibwa, snepaan in Menomini, senibaan in Cree, Chipewyan sounibanen, Beaver soónipan and Carrier sunîmpal. If Brown’s etymology of ses ribans (sic, for ses rubans) ‘his/her ribbons’ is correct, with an unusual unique possessive element, this word could only have been transmitted through North America without direct interference from French speakers. Other languages show other forms: Algonquin deniband, Attikamek tiripaan, Ile-a-la-Crosse rribâñ, Chinook Jargon leloba.

5.

Bilingualism: codeswitching and mixed languages

In this section we discuss two extreme results of language contact: intensive code switching between Montagnais and French in Northern Quebec, and the mixed language Michif that emerged among bilingual Cree-French Métis in Western Canada.

5.1. Code switching and lexical erosion in Montagnais In a series of papers, Drapeau (1991, 1993, 1994, 1995) discusses the current situation in the Montagnais community of Betsiamites. As mentioned earlier, the very young and the very elderly are monolingual in Montagnais (the latter representing 8% of the total adult population), everyone else is bilin-

French influence on the native languages of Canada and adjacent USA

267

gual (at various levels of competence) in both Montagnais and French. Montagnais remains the language of everyday interactions while French is the main language of instruction in the community schools. Because Montagnais is used in just about every possible situation, the language has had to adapt to new topics and new situations. Instead of neology (i.e. creating a new term or adapting an existing one to a new meaning), the majority of Montagnais speakers have developed strategies that are typical of bilingual communities; specifically, code switching. Analysis of natural speech data collected by Drapeau shows that “mixing French into Montagnais is the dominant discourse mode among fluent bilinguals” (Drapeau 1995: 158). According to her data and analyses, intersentential or interclausal switching accounts for less than 16% of the instances of code switching. Most of the cases are therefore intrasentential and are determined by the syntax of the matrix language, which is always Montagnais. The code-switched elements are usually “phonologically and morphologically unintegrated French phrases, mostly NPs and PPs, which account for approximately 64% of code-switches” (Drapeau 1995: 158). On average, these switches are less than three words in length. They are often simply nouns with their associated determiner. The other 20% of switches are single-word switches, mostly greetings, days of the week, names of the month, time expressions, expressions of quantity and measures, adverbs, conjunctions, discourse markers and various interjections, as well as verbs and adjectives. It is well known that languages often borrow names of the days of the week, names of the month, greetings and so on. Therefore some of these forms could potentially represent true borrowings rather than “code switches”. On the other hand, as was discussed in Section 3.3, borrowed nouns in Montagnais typically take on the agglutinated reflex of the French article le. On the other hand, greetings, names of the month, measures, etc. do not take definite determiners in French and it is unlikely that bilingual Montagnais speakers would insert a reflex of a French determiner where this would be impossible in the donor language. This strongly argues for a “code switching” strategy rather than a “borrowing” strategy for these forms. Verbs are never integrated morphologically to Montagnais; they are inserted bare (usually an infinitive) and they always co-occur with the Montagnais support verb tuutam ‘to do’, as in the following example, taken from Drapeau (1991), where the support verb is in bold, the French constituents are in italics: (2)

Mukw ekue peikwun engager nitutakuti utehe, tshissina. Nanaa tshashtaian le CÉGEP, tshia? Ekue nutepalipan miam un professeur, auen ne kanutepalt, remplacer ne nitutuau pour un an. Ekue engager

268 Peter Bakker and Robert A. Papen mipua tutakuian. Ekuan, depuis ce temps-là, tshissina, nanitam réengager ekue tutakuian, tshia… ‘But then, they hired me anyway, you know. When I finished CÉGEP (junior college), you see? They were one teacher short at the time; the person who was lacking, I replaced him/her for one year. And then, they hired me. So, since then, you know, they’ve always re-hired me, you see…’ This type of code-switching represents what Myers-Scotton (2002) calls Embedded Language Island insertions in her Matrix Frame Model, where the Matrix language (here Montagnais) determines the morphosyntactic frame, into which lexical material from French is inserted. Drapeau (1995: 160) observes that Montagnais adults “make alternate use of the mixed and non-mixed codes depending on the age and the background of the addressee, as well as on topic and setting”, just as do fluent bilinguals in many other communities. Intersentential switches are totally absent in monitored speech and even intrasentential switches are much less prevalent. Speech addressed to elders (especially those who are monolingual) contains fewer code switches. On the other hand, normal unmonitored conversations between young or middle-aged bilinguals are replete with switches of the type described above. Furthermore, Drapeau points out that many of these switches involve French lexical items for which simple Montagnais equivalents are easily available. Drapeau (1993) briefly discusses the results of analyses of speech data gathered over a number of years. She particularly looked at code switching behavior in a three-hour hotline in Montagnais on the community’s local radio station (recorded in 1987) involving 18 different community members (aged 24–71) who called in to voice their opinion (198,418 words); a random sample of free conversations between a Montagnais observer-participant and 10 adults aged between 18 and 57 (recorded in 1991; 36,791 words) and a series of recordings conducted by a community member of 5 bilingual speakers (aged 17–34) (recorded in 1980; 24,069 words). Drapeau calculated for each sample the total number of noun types and the proportion of noun types in each language (MN = Montagnais noun, FN = French noun). The results are shown in Table 5 (taken from Drapeau 1993):

French influence on the native languages of Canada and adjacent USA

269

Table 5. Analysis of noun types Sample

Total number of N types

Proportion of FN types

Proportion of MN types

461 271 502

61.4 % 53.5 % 60.0 %

38.6 % 46.5 % 40.0 %

1980 1987 1991

In every case, the proportion of French noun types is higher than 50 %. Drapeau also calculated the proportion of French nouns for which no existing equivalent is easily available in Montagnais, allowing her “to assess the extent of the communicative needs of the speakers” (Drapeau 1993: 143). The results are as follows: Table 6. Proportion of French types with no Montagnais equivalent 1980 sample 43.8 % (N=124)

1987 sample 64 % (N=93)

1991 sample 31.2 % (N=94)

These figures show quite clearly that the need to borrow is not the only motivating drive for switching since more than half of the code-switches in the 1980 and 1991 samples and one-third in the 1987 hotline sample had Montagnais equivalents available. Nevertheless, the need to borrow does account for a substantial proportion of switches to French, since the native language does not seem to be able to meet the communicative needs of the speakers, particularly when dealing with contemporary topics. Drapeau points out that in the hotline data, in order to avoid borrowing, 93 different words would have had to be coined in Montagnais! This tendency of regularly inserting French lexical material into Montagnais sentences, particularly in every-day speech, has had a number of consequences. The most interesting one is that caretakers most often tend to use the mixed code even when talking to monolingual children. Children are therefore intensively exposed to this mixed speech and “preschool-age children reciprocate by inserting French phrases into Montagnais and occasionally using French verbs accompanied by a Montagnais support verb” (Drapeau 1995: 160), as in the following examples (taken from Drapeau 1995: 161): (3)

Apu takuat nil mon coussin! ‘I don’t have my pillow!’

270 Peter Bakker and Robert A. Papen (4)

Nasht rasé nipa ishinakutan nipishkueuna ute ‘I would have my hair shaved completely here (pointing to the nape of her neck).’

As noted above, these lexical items have perfectly good and easily available equivalents in Montagnais, since they can be considered as “core vocabulary”, but Montagnais children rarely hear the Montagnais equivalent, due to their regular exposure to this mixed input and therefore there has been a serious decline in the level of Montagnais lexical skills in the younger generations. A lexical naming experiment was undertaken in 1991, where subjects had to identify (by naming either in Montagnais or in French) over 100 drawings representing core vocabulary lexical items, culturally important objects for the Montagnais as well as items of every-day life (excluding more recent inventions). All of these items have a Montagnais lexical equivalent. The test was given to elementary school children, middle-school students, high school students and adults aged from 20 to 59 years (Drapeau 1994). The results show that adults named 92% of the items in Montagnais, high school adolescents named 72% of the items in Montagnais, middle school children gave Montagnais equivalents 62% of the time and 4th grade elementary school children identified only 48% of the items in Montagnais. There is thus a gradual erosion of Montagnais core vocabulary in Betsiamites. Results show, for example, that names for common body parts are generally well maintained (tooth, nose, mouth, ear, etc.) but less common body parts (neck, knee, heel, elbow, etc.) are much less well-known, even by the adults. Only the most common terms referring to nature (snow, sun, sand, flower, tree) in Montagnais are known by all speakers, but less usual ones (waterfall, moon, cloud, mountain, lake, island, wave) are known only in French by the 4th grade children. Moreover, in numerous cases, when children attempted to give a Montagnais term, they often erred, giving a wrong form (often the hyperonym); for example, giving the form nipi ‘water’ for sketches showing a lake or a waterfall. Similarly, they gave the Montagnais equivalent of ‘insect’ for a drawing showing an ant. In Montagnais there are distinct words for ‘foot’ and ‘leg’ and for ‘hand’ and ‘arm’, but the naming test shows that Montagnais children are now using the Montagnais word for ‘leg’ to refer to both ‘leg’ and ‘foot’ and the Montagnais word for ‘hand’ for both ‘hand’ and ‘arm’. A similar naming test was given to a number of 4-year old preschool children. Results show that they spontaneously gave answers in French for many of the items, such as ‘pillow’, ‘house’, ‘ear’, ‘horse’, etc. For some

French influence on the native languages of Canada and adjacent USA

271

items such as ‘milk’, ‘apple’ and ‘doll’, none of the children used the correct Montagnais word. Instead they used French words exclusively (Drapeau 1994: 367). This has had the following consequence: A proficiency continuum has thus arisen in the community. While bilingual adults are competent in Montagnais, French-Montagnais, and French, young people tend to be proficient exclusively in the mixed code and French. Young children, at the time they reach school age, speak and understand very little French but use a type of mixed Montagnais that integrates a great quantity of French phrases. Predictably, elders complain bitterly about the failure of young people to speak and understand Montagnais correctly. (Drapeau 1995: 162)

As mentioned in section 3.3, the Montagnais community of Betsiamites has opted to maintain its ancestral tongue rather than switch allegiance and adopt French as its L1. Nevertheless, Drapeau (1993:144) argues that the ever-growing communicative needs of any community resisting linguistic assimilation are such that there is simply a single option available: massive borrowing and code switching, since: [T]he need for lexical elaboration is so high in persistent communities confronted with the communicative demands of the modern day world, that there is virtually no way that small linguistic enclaves could ever cope with this problem without importing massively. Spontaneous coining at the community level is no longer possible as speakers are overburdened by the number of items to create. […] The alternative is either to shift to the majority language or to transfer massively from the majority language.

What has just been said for the Montagnais case applies mutatis mutandis to the Attikamek communities, who are in a highly similar situation. Drapeau did a preliminary study of Attikamek code switching in two hours of recorded conversations and the results are highly similar to those found for Betsiamites Montagnais. According to Drapeau and Bakker (1994), around 80% of the switches are intersentential and among these, approximately 75% are multi-word switches, mostly NPs and PPs. Less than 5% of the single-word switches are bare nouns. According to Bakker (1997: 186) “These figures indicate that similar social situations with typologically almost identical language pairs lead to very similar types of mixtures.”

272 Peter Bakker and Robert A. Papen 5.2. Michif: Cree verbs, French nouns Michif is a special case. It has been called the “nec plus ultra” of mixed languages (Papen 1987) because basically all of the noun phrases are from French, whereas virtually all the verbs are from Cree (Rhodes 1977). On the basis of a quick 10-page count of the number of French word-types in the Michif dictionary (Laverdure and Allard 1983), one arrives at an estimated 10,000 French words in this language. Of course this estimate is based on a rather small percentage of pages (first 10 pages out of some 350), so 5,000 French words seems like a more reasonable estimate of the number of French words in Michif (see also Niederehe 1996). In view of this large number, one cannot reasonably call this “borrowing” and some other explanation must be found. The French elements ultimately go back to PLF (see section 3.1), but they have undergone a number of sound changes. The most conspicuous ones are the merger of /e/ and /i/ and of /o/ and /u/; the palatal affrication of dentals and some velars before front vowels, and the harmonization of sibilants. Some typical Michif words are lii suyii < les souliers ‘shoes’, shu < chaud, choux, ‘warm, cabbage’ sasoer < chasseur ‘hunter’, chwizinn < cuisine ‘kitchen’, joel < gueule ‘mouth’. All these deviations from PLF in Michif are also found in Métis French, which is thus the real source of the French component of Michif. There are Cree dialects where /e/ and /i/ have merged, but not the Southern Plains Cree dialect which forms the other component of Michif. There is a systematic combination of parts of speech from French (nouns, definite and indefinite articles, numerals, equative and identificational copulas, adjectives), from Cree (verbs, personal pronouns, demonstratives, question words, existential/locative copulas), or from both languages (adpositions, adverbs, the numeral for ‘one’, discourse markers, negation, etc.). The Michif verb preserves virtually all the complexities of the Plains Cree verb (with sometimes six or seven consecutive bound morphemes), and the nouns maintain French gender, although they also take on Cree animate/ inanimate gender in order to guarantee correct Cree verb agreement. Michif is spoken by hundreds, perhaps thousands of Métis in scattered locations mainly in Manitoba, Saskatchewan, and North Dakota. The Métis are descendants of Amerindian women (mostly Ojibwa) and French-speaking fur traders. They became a separate population group around 1800, and are now constitutionally recognized in Canada as one of the three Aboriginal groups (along with the Inuit and the First Nations). Biologically and culturally, the Métis are usually characterized as a hybrid population, with Amer-

French influence on the native languages of Canada and adjacent USA

273

indian and European ancestry, and a culture that chooses the best of both worlds. Traditional foods, clothing, means of transportation and music constitute a mixture of First Nations and French-Canadian (also Scottish-Canadian) elements. Also, the Métis have traditionally been multilingual, and Michif is a mixed language with an estimated 50-50 distribution of French and Cree elements. As we have focused mostly on the lexicon elsewhere, we will focus on phonology and grammar here. There is clearly an Algonquian influence in the French-derived NP. This, however, seems to predate the intertwining with Cree, since most or all of the Algonquian features are also found in French spoken by Métis. Everything typical for the French part of Michif is also found in Métis French and Métis French existed before the creation of the intertwined language we have been calling ‘Michif’. Indeed, Métis French is the primary source of the French element in Michif. 5.2.1. Phonology The number of phonemes of Michif is notoriously difficult to establish. Part of the problem is caused by the fact that the Métis French component and the Plains Cree component have preserved their own phonological systems, but is a French-derived /t/ the same as a Cree-derived /t/? Even if one were to count these as the same (which methodologically would be questionable since there are different allophones for these two phonemes), one would get a number of between 24 and 31 consonants, and some 14 vowels (Bakker 1997; Papen 2003a, 2005) – which is cross-linguistically high, but not exceptional. In general, French lexical items maintain typical French phonology and Cree lexical items maintain typical Cree phonology. The two phonologies are therefore stratified, although there is some “leakage” between the two (Papen 2003a). 5.2.2. Morphology The Michif verb has inherited all the complexities of the Cree verb, with two sets of inflectional person affixes, tense/mood/aspect suffixes and prefixes, voice, valency suffixes – sometimes several in a row. Even the verb stems are compounds of two or more root-like elements. Each verb stem can probably yield thousands of different derived and inflected forms (Bakker 2004, 2006).

274 Peter Bakker and Robert A. Papen Michif nouns from French are devoid of inflection, but are relatively rich in derivation. As mentioned, they maintain French gender but because of the Cree demonstratives and the Cree verb system, which requires gender agreement with subject and object, French nouns also take on Cree (animate/inanimate) gender (Papen 2003b). Michif thus combines the morphologically most complex parts of the two languages. 5.2.3. Stratification Michif is a highly stratified language. Cree nouns are inflected in Michif just as in Cree, e.g. possessive: o-maamaa-wa ‘her/his mother(s)’, where o- marks third person possessive and -wa indicates that there are two third persons involved. Cree nominalization strategies, with the suffixes -iwin forming abstract nouns and -ikan forming instruments, are available, creating deverbal Cree nouns: kimutiw ‘(s)he steals’ > kimutiwin ‘theft’, potatikeew ‘(s)he blows’ > potatikan ‘horn’. These nouns always have French masculine gender and Cree inanimate gender. The French part uses a number of French-derived derivational processes, and suffixes like -euse (female actor) and -asse/-esse (abstract noun) have created forms not found in other French varieties such as begeuz ‘female beggar’ (< Eng. beg + Fr. -euse) and pruvas ‘proof’ (< Fr. prouv(er) + asse) and zholiyes ‘beauty’ (< Fr. joli + esse). In addition, some phrases have lexicalized to become expressions: aen fe-li-saen ‘a hypocrite’ (< faire le saint), enn fe-la-bonn ‘a prude’ (< faire la bonne) (Bakker and Papen 1997). 5.2.4. Mixture within words Words in Michif are almost always either completely Cree or completely French. There are a few exceptions however: the Cree suffix -ipan ‘deceased’ and a few other Cree suffixes can be used with French nouns (mon granper-ipan ‘my deceased grandfather’, not solda-inaan ‘our soldiers’) and perhaps a dozen French verbs are used with Cree: (5)

ee

-li -brodii-yaan -the- embroider-1SG ‘that I embroider’ COMP

2

French influence on the native languages of Canada and adjacent USA

275

5.2.5. Constituent order As in Cree, sentential order is relatively free, allowing all combinations of SVO/SOV/VSO/VOS/OVS/OSV in transitive sentences. The order in French noun phrases is almost always as in French, with three general exceptions. Possession is expressed differently: where French has les chevaux de la fille or more marginally les chevaux à la fille ‘the girl’s horses’, Métis French has la fiy sii zhvuu, much as in Ojibwa and Cree iskweesis o-teem-a. Whereas all other varieties of French express possession as: possessed – preposition – possessor, Michif French uses a construction possessor – poss.pron. – possessed, which is similar to the Algonquian construction. Some Algonquian processes, such as the floating of a quantifier to an earlier position in the sentence, are also present in Michif: (6)

mihcheet kiinipaheewak lii zhvuu Many they-killed-them the horses ‘They killed many horses’

Also, quantifiers co-occur with articles, which is not allowed in French: peeyak en zhwal ‘one (a) horse’, trwa lii faam ‘three (the) women’ (vs. French: un cheval, trois femmes). Again, these constructions are typical of Métis French. Neither Cree nor French makes case distinctions. In Cree, the semantic roles of subject and object are distinguished in the verb, where referential affixes agree with subject, object and/or indirect object. When two or more third persons are involved, one also gets a marker on the noun, marking (roughly) non-topic, called the “obviative”. This nominal ending is also optionally used in Michif, and is added to French noun phrases: (7)

sa primyeer faam kii-wanih-eew (…) Kii-wiikim-eew wife PAST-lose-3>3 PAST-marry-3>3 his-F first oonhiin la faam-a that the woman-OBV ‘He had lost his first wife (…) He married that woman’

Only the second occurrence of faam is marked here for obviative. Both pre-nominal and post-nominal French adjectives are used. In French, adjectives agree in gender (and number) with the noun. In Michif only prenominal adjectives agree with the noun, not post-nominal ones: en bo shyen ‘a-M beautiful-M dog’, enn bel vwacheur ‘a-F beautiful-F cariole’, but aen paen blan ‘a-M white-M bread’, la maenzon blan (*blansh) ‘the-F white-M house’.

276 Peter Bakker and Robert A. Papen A language with such an intricate mixture of two languages cannot be an ordinary case of borrowing. It has been proposed (Bakker 1997) that the language came into being through a process called “language intertwining”, in which two languages combine into one, by using the grammatical system of one language and the lexicon of another. There are perhaps two dozens such languages known in the world, and usually these languages look differently in that they mix at the word level, and combine syntax, morphology and phonology from one language with the stems from another – adjusted phonologically and morphologically. An example from a French mixed Romani variety (Calvet 1981): (8)

6.

Dikav la rakli comme elle est ukar look.at the girl how she is pretty Fr. Regarde la fille comme elle est belle ‘Look at the girl how pretty she is!’ (Romani dikhav “I see”, rakli “(non-Romani) girl”, ukar “pretty”)

Conclusions

We have generally covered most instances of French influence on the Native languages of Canada. We now offer a few conclusions.

6.1. Types of French First, there is an impact from at least three distinct types of French in North America. The least significant is European French. European French words were introduced into Native languages probably exclusively through the presence of European Catholic missionaries in Native communities, initially Jesuits and Recollets, later usually Oblates and a great variety of nun orders. Some of the French loans, especially those with a religious character, show a distinctly European origin. These borrowings may also be more localized than some of the others. Second, there is an impact from Canadian (specifically Laurentian) French varieties (see 3.1). This set of varieties had the most impact, directly or indirectly, on the French loans in North America. The third variety is a specific variety of PLF, namely that spoken by the Métis. The Métis were traders and middlemen in the West. Their variety is characterized by the use of /t/ and /d / for CF /ts, dz/, and the loss of a

French influence on the native languages of Canada and adjacent USA

277

phonemic distinction between /e, i/ and /o, u/ (see also Section 5.2). In addition, some velars are palatalized (EF /kœr/, MF /tœr/) and some vowels become falling diphthongs: O > wa, as in /koSõ/ vs. /kwa õ/ ‘pig’, korney vs. kwarney ‘crow’ (Papen 1998b, 2004).

6.2. Typological and sociolinguistic aspects It appears that structural or typological features of the borrowing languages have little of no impact on the susceptibility to borrowing. Most Amerindian languages in North America are polysynthetic (see Mithun 1988; Baker 1996; and the papers in Evans and Sasse 2002). As polysynthetic languages (which by no means form a homogenous class) are supposed to be the least prone to borrowing (Field 2002), it may appear surprising that all major families and isolates of North America have borrowed from French and other European languages. Brown (1999: 90) lists the following mean loanword percentages (based on a fixed set of 77 imported objects or concepts): Salishan 28, Eskimo 17, others (mostly isolates) 15, Athapaskan 12, Algonquian 12, Iroquoian 9, Siouan 3. Brown (1999: 91) concludes that “if language structure factors affect lexical borrowing, they do so only minimally”. Brown (1999) believes that the closer interaction of the Russians and Spanish with the Natives (compared to the English and French), was responsible for a greater number of loanwords from those languages than from French and English. Sociolinguistic features of language communities, such as attitudes, are much more important than the typological features of the borrowing language (but cf. Field 2002).

6.3. The role of “economical” languages Brown (1999) emphasizes that “economical” languages, i.e. languages used in interethnic communication, played a major role in the spread of loanwords. For the area under consideration, he especially mentions Chinook Jargon and Slavey Jargon, both pidginized languages. On closer scrutiny, there appears to be a whole network of trade languages in the northern part of the continent, or parts of the continent: French in the east and along the fur trade routes in most of Canada; several Ojibwa dialects (Odawa, Algonquin) east and south of the Great Lakes (Rhodes 1982; Nichols 1995 for a pidginized version), and Cree north of the Great Lakes and on the prairies and southern woodlands (Rhodes 1982; Bakker and Grant 1996; Bakker

278 Peter Bakker and Robert A. Papen 1996b), Slavey Jargon and Jargon Loucheux (subarctic woodlands), Plains Sign Language (Prairies into southern Canada), Plateau Sign Language in interior British Columbia (Boas and Hale 1890), Chinook Jargon along the Northwest coast. These pidginized and non-pidginized lingua francas cover almost all of Canada, except the High Arctic. At first sight, Brown’s claim that economical languages are mostly responsible for the spread of loans is partly confirmed. Most of the French loans at least are found in areas where French was the language of the settlers, and areas where spoken pidgins were in use (see Map 2). The two Metis varieties, Ile-a-la-Crosse Cree and Michif, are exceptions and the result of exceptional circumstances. Notes 1. 2.

An appendix with French loanwords in Amerindian languages is accessible at: http://www.aal.au.dk/en/linguist/frenchloanwords COMP = complementizer, F = feminine, M = masculine, OBV = obviative, SG = singular

Maps

Map 1. Language families, provinces, geographical names

Map drawn by Iris Bakker

Chinook Jargon or Wawa (pidgin) (ca. 1800 –1930)

Ojibwa/Ojibwe as lingua franca (full language and pidginized) (–1900)

Broken Slavey and Jargon Loucheux (pidgins) (ca.1800 –1930)

Cree as lingua franca (–ca.1900

This map shows approximate ranges of Amerindian languages. Two types of numbers are indicated for selected languages. The numbers before the slash indicate the number of French loans on Brown’s (1999) closed list of terms for cultural innovations. The numbers after the slash indicate the total number of French loans thus far identified. Blanks indicate that no loans are known. In most cases this means that they are absent.

French influence on the native languages of Canada and adjacent USA

279

Map 2. Numbers of documented French loanwords in Amerindian languages, and ranges of contact languages

280 Peter Bakker and Robert A. Papen References Ahenekew, Vince 1997 Michif/Cree dictionary. Nêhiyawêwin Masinahi an. Saskatoon: Saskatchewan Indian Cultural Centre. Aubin, George F. 1981 Remarks on Golden Lake Algonquin. In Papers of the 12th Algonquian Conference, William Cowan (ed.), 39–46. Ottawa: Carleton University. 1999 Golden Lake Algonquin: A look at some interesting data. In Papers of the Thirtieth Algonquian Conference, David H. Pentland (ed.), 1–11. Winnipeg: University of Manitoba Press. Baker, Mark 1996 The polysynthesis parameter. New York: Oxford University Press. Bakker, Peter 1989a The language of the coast tribes is half Basque: a Basque-Amerindian Pidgin in use between Europeans and Native Americans in North America, ca.1540 – ca.1640. Anthropological Linguistics 31 (3–4): 117–147. 1989b Two Basque loanwords in Micmac. International Journal of American Linguistics 55 (2): 258–261. 1996a Broken Slavey and Jargon Loucheux: a first exploration. In Language contact in the Arctic: Northern Pidgins and contact languages, Ingvild Broch and Ernst Håkon Jahr (eds.), 317–320. Berlin /New York: Mouton de Gruyter. 1996b Hudson Bay traders’ Cree: A Cree pidgin? In Nikotwâsik Iskwâhtêm, Pâskihtêpahih! Studies in Honour of H. C. Wolfart, John D. Nichols and Arden C. Ogg (eds.), 1–34. Winnipeg: Algonquian and Iroquoian Linguistics Memoir 13. 1997 A language of our own. The genesis of Michif – The mixed CreeFrench language of the Canadian Métis. New York: Oxford University Press. 2004 The verb in Michif. In La Lawng: Michif Peekishkwewin. The heritage language of the Canadian Metis. Vol 2: Language theory, Lawrence Barkwell (ed.), 63–80. Winnipeg: Pemmican Publications/ Manitoba Metis Federation Michif Language Program. 2005 Preface to the 2005 edition. In A vocabulary of the Souriquois jargon, Peter Bakker and Kristina German Johansen (eds.), 1–17. Bristol, PA: Evolution Publishing. 2006 Algonquian verb structure: Plains Cree. In What’s in a verb?. (LOT Occasional Series 5), Grazyna Rowicka and Eithne Carlin (eds.), 3– 27. Utrecht: LOT.

French influence on the native languages of Canada and adjacent USA

281

Bakker, Peter and Anthony P. Grant 1996 Interethnic communication in Canada, Alaska and adjacent areas. In Atlas of languages of inter-cultural communication in the Pacific, Asia and the Americas, Stephen A. Wurm, Peter Mühlhäusler and Darrell T. Tryon (eds.), 1107–1169. Berlin /New York: Mouton de Gruyter. Bakker, Peter and Robert A. Papen 1997 Michif: A mixed language based on French and Cree. In Contact languages. A wider perspective, S. Thomason (ed.), 295–363. Amsterdam: John Benjamins. Baldwin, Stuart J. 1994 Blackfoot neologisms. International Journal of American Linguistics 60 (1): 69–72. Béland, Jean-Pierre 1978 Atikamekw morphology and lexicon. Ph.D. diss, University of California, Berkeley. Bloomfield, Leonard 1962 The Menomini language. Charles Hockett (ed.). New Haven, CT: Yale University Press. Boas, Franz 1925 Romance folklore among American Indians. The Romance Review 16: 199–207. [Reprinted 1940 in Race, language and culture, F. Boas, 517–524. New York: MacMillan.] Boas, Franz and Horatio Hale 1890 Sign language. In The Sixth Report on the North-Western Tribes of Canada, 87–88. London: The Association. Bonvillain, Nancy 1978 Linguistic change in Akwesasne Mohawk: French and English influence. International Journal of American Linguistics 44: 31–40. Brown, Cecil 1999 Lexical acculturation in native American languages. New York: Oxford University Press. Calvet, Georges 1981 La langue. In Les populations tsiganes en France, Jean-Pierre Liégeois (ed.), 42–43. Paris: Ministère de l’Éducation Nationale, Direction des Écoles/ Centre de Recherches Tsiganes. Chamberlain, Alexander F. 1891 The Aryan element in Indian dialects. Canadian Indian, February 1891: 3–8. Also published separately. 8 pp. Cooper, John M. 1945 Tête-de-Boule Cree. International Journal of American Linguistics 11 (1): 36–44.

282 Peter Bakker and Robert A. Papen Cuoq, Jean-André 1886 Lexique de la langue Algonquine. Montreal: J. Chapleau. DeBlois, Albert and Alphonse Metallic 1984 Micmac lexicon. Ottawa: National Museum of Man, Mercury Series. Drapeau, Lynn 1980 Les emprunts au français en Montagnais. In Inuktitut et les Langues Amérindiennes au Québec, Special issue of Cahiers de Linguistique, Judith McNaulty (ed.), 29–49. Québec: Presses de l’Université du Québec. 1991 Dictionnaire Montagnais-Français. Sillery, QC: Presses de l’Université du Québec. 1993 Language birth: An alternative to language death. In Proceedings of the XVth International Congress of Linguists, vol. 4, André Crochetière, Jean-Claude Bélanger and Conrad Ouellon (eds.), 141–144. Quebec: Presses de l’Université Laval. 1994 Bilinguisme et érosion lexicale dans une communauté montagnaise. In Langues et Sociétés en Contact, Pierre Martel and Jacques Maurais (eds.), 363–376. Tübingen: Max Niemeyer Verlag. 1995 Code switching in caretaker speech and bilingual competence in a native village of Northern Quebec. International Journal of the Sociology of Language 113: 157–164. Drapeau, Lynn and Peter Bakker 1994 Typological constraints on code-mixing: The case of three Algonquian languages in contact with French. Paper presented at the World Congress of Sociology, Session on Code-Switching, Bielefeld, Germany, July 1994. Dremeaux, Lillie 2003 Slavey Jargon and the presence of French loanwords in Northern Athabascan. Senior thesis, Swarthmore College. Dumas, Denis 1987 Nos façons de parler. Les prononciations en Français Québécois. Sillery, QC: Les Presses de l’Université du Québec. Evans, Nicholas and Hans-Jürgen Sasse 2002 Problems of polysynthesis. (Studia Typologica 4.) Berlin: Akademie Verlag. Field, Fredric W. 2002 Linguistic borrowing in bilingual contexts. Amsterdam/Philadelphia: John Benjamins. Gabriel Dumont Institute 2002 Kitaskinaw i pi Kiskinohamakoya. The land gives us our knowledge. Saskatoon: Gabriel Dumont Institute. Garvin, Paul L. 1947 Christian names in Kutenai. International Journal of American Linguistics 13: 69 –77.

French influence on the native languages of Canada and adjacent USA

283

Garvin, Paul L. 1948 Kutenai lexical innovations. Word 4: 120 –126. Geary, James A. 1945 Algonquian nasaump and napôpi: French loanwords? Language 20 (1): 40–45. Grant, Anthony P. 1996 Chinook Jargon and its distribution in the Pacific Northwest and beyond. In Atlas of languages of inter-cultural communication in the Pacific, Asia and the Americas, Stephen A.Wurm, Peter Mühlhäusler and Darrell T. Tryon (eds.), 1184–1208. Berlin /New York: Mouton de Gruyter. Haas, Mary 1968 Notes on a Chipewyan dialect. International Journal of American Linguistics 34 (3): 165–175. Heine, Bernd and Tanja Kuteva 2005 Language contact and grammatical change. Cambridge: Cambridge University Press. Hewson, John 2000 The French language in Canada. Munich: Lincom Europa. Huot, Martha Champion 1948 Some Mohawk terms of acculturation. International Journal of American Linguistics 14: 150–154. Ile-a-la-Crosse n.d. website: www.michif.dev.kcdc.ca Kinkade, Dale M. 1985 More on nasal loss on the Northwest coast. International Journal of American Linguistics 51: 478–480. Krauss, Michael 1983 Slavey Jargon: Diffusion of French in Northern Athabaskan. Ms. Fairbanks: Alaska Native Language Center. 1996 The Russian language in Alaska and in Alaska native languages. In Atlas of languages of intercultural communication in the Pacific, Asia and the Americas, Stephen A.Wurm, Peter Mühlhäusler and Darrell T. Tryon (eds.), 1209 –1212. Berlin /New York: Mouton de Gruyter. Lang, George 1991 Voyageur discourse and the absence of a fur trade pidgin. Canadian Literature 131: 51–63. 1995 Le lexique du français des voyageurs dans le jargon tchinouk d’avant 1846. Revue Québécoise de Linguistique Théorique et Appliquée 12: 247–264. Laverdure, Patline and Ida Rose Allard 1983 The Michif dictionary. Turtle Mountain Chippewa Cree. John C. Crawford (ed.). Winnipeg: Pemmican Publications.

284 Peter Bakker and Robert A. Papen Leavitt, Robert and David Francis 1985 Lolusuwakonol. Passamaquoddy-Maliseet Dictionary. Fredericton: Micmac-Maliseet Institute. Mishler, Craig in print “That’s a rubbaboo”: Slavey Jargon, a nineteenth century Subartic speech community. To appear in Journal of Pidgin and Creole Languages. Mithun, Marianne 1988 System-defining structural properties in polysynthetic languages. Zeitschrift für Phonetik, Sprachwissenschaft und Kommunikation 41 (4): 442–452. Myers-Scotton, Carol 2002 Contact linguistics. Bilingual encounters and grammatical outcomes. Oxford: Oxford University Press. Nichols, John D. 1995 The Ojibwe verb in “Broken Oghibbeway”. Amsterdam Creole Studies 12: 1–18. Niederehe, Hans-Josef 1996 Eine Zukunft für das Mestizen “Französische”: Mitchif. Zeitschrift der Gesellschaft für Kanada Studien 30 (2): 47–58. Norris, Mary Jane 1998 Les langues autochtones du Canada. Le Quotidien: Tendances sociales canadiennes, no. 11-008, Statistique Canada: 8–17. Orkin, Mark 1971 Speaking Canadian French. Revised ed. Toronto: General Publishing Co. Papen, Robert A. 1987 Le métif: Le nec plus ultra des grammaires en contact. Revue Québécoise de Linguistique Théorique et Appliquée 6 (2): 57–70. 1997 Le contact des langues au Canada: Le cas du français et des langues autochtones. In Canada et bilinguisme, Marta Dvorak (ed.), 111–125. Rennes: Presses universitaires de Rennes. 1998a French: Canadian varieties. In Language in Canada, J. Edwards (ed.), 160–176. Cambridge: Cambridge University Press. 1998b Le parler français des Métis de l’Ouest canadien. In Français d’Amérique: Variation, Créolisation, Normalisation, Patrice Brasseur (ed.), 147–161. Avignon: CECAV, Université d’Avignon. 2003a Michif: One phonology or two? In Proceedings of WSCLA VIII, University of British Columbia Working Papers in Linguistics, vol. 12, Yunhee Chung, Carrie Gillon and Rachel Wodjak (eds.), 47–58. Vancouver: University of British Columbia. 2003b Le mitchif: Une question de genre. In Francophonie et langue dans un monde divers en évolution: Contacts interlinguistiques et socio-

French influence on the native languages of Canada and adjacent USA

285

culturels, Robert A. Stebbins, Claude Romney and Micheline Ouellet (eds.), 119–142. Winnipeg: Presses universitaires de Saint-Boniface. 2004 Sur quelques aspects structuraux du français des Métis de l’Ouest canadien. In Variation et Francophonie, Aidan Coveney, MarieAnne Hintze and Carol Saunders (eds.), 105–129. Paris: L’Harmattan. 2005 Le mitchif: Langue franco-crie des Plaines. In Le Français en Amérique du Nord: État Présent, Albert Valdman, Julie Auger and Deborah Piston-Hatlen (eds.), 327–348. Saint-Nicolas, QC: Les Presses de l’Université Laval. Papen, Robert A. and Kevin J. Rottet 1997 A structural sketch of the Cajun French spoken in Lafourche and Terrebonne parishes. In French and Creole in Louisiana, Albert Valdman (ed.), 71–108. New-York: Plenum Press. Pentland, David H. 1982 French loanwords in Cree. Kansas Working Papers in Linguistics 7: 105–117. Petitot, Emile 1889 Quinze ans sous le Cercle Polaire. Mackenzie, Anderson, Youkon. Paris: E. Dentu. 1891 Autour du Grand Lac des Esclaves. Paris: A. Savine. Pierronet, Thomas 1800 Specimen of the Mountaineer, or Sheshatapooshshoish, Scoffie, and Micmac Languages. Collections of the Massachussetts Historical Society, Ser. 1, Vol. 6: 16 –33. Boston. Poser, William J. 2004 Dating velar palatalization in Carrier. [Internet version, 11 May 2004] Prunet, Jean-François 1990 The origin and interpretation of French loans in Carrier. International Journal of American Linguistics 56 (4): 484–502. Rhodes, Richard 1977 French Cree – A case of borrowing. In Actes du Huitième Congrès des Algonquinistes, William Cowan (ed.), 6 –25. Ottawa: Carleton University. 1982 Algonquian trade languages. In Papers of the Thirteenth Algonquian Conference, William Cowan (ed.), 1–10. Ottawa: Carleton University. Rice, Keren 1989 A grammar of Slave. Berlin /New York: Mouton de Gruyter. Rottet, Kevin 2001 Language shift in the Coastal Marshes of Louisiana. New York: Peter Lang. Sapir, Edward 1921 Language: an introduction to the study of speech. New York: Harcourt.

286 Peter Bakker and Robert A. Papen Valentine, Randolph J. 1994 Ojibwe dialect relationships. Ph.D. diss., University of Texas at Austin. Walker, Douglas 1984 The pronunciation of Canadian French. Ottawa: University of Ottawa Press.

Portuguese influence on Kulina1 Stefan Dienst

1. Beginnings of Portuguese settlement in northern Brazil The arrival of the Portuguese in South America dates back more than five hundred years, but the direct influence of the Portuguese language on the indigenous languages of the Amazonian hinterland began only much later. The first Portuguese ship reached the Brazilian coast in 1500 and the colonisation of the country began in 1532. But it was only at the beginning of the 17th century, after driving the French out of the area, that the Portuguese started to settle on the coast of northern Brazil. 2. Nheengatu2 The mostly male Portuguese settlers in northern Brazil intermarried with the local Tupinambás, speakers of a language of the Tupi-Guarani branch of the Tupian language family. The children resulting from those relationships learned the language of their indigenous mothers as their first language. In Brazil, this language, a form of Tupinambá influenced by Portuguese, is today called língua geral amazônica ‘Amazonian General Language’ or nheengatu. It was the predominant language of the Neobrazilian society in northern Brazil until it was outlawed and the use of Portuguese prescribed in 1757. The language has since been in constant decline, but it is still spoken in the border area of Brazil, Colombia and Venezuela, with speakers living in all three countries. The language is known as yeral in Colombia and Venezuela. 3. Exploration of the Amazon and Catholic mission The first Portuguese expedition up the Amazon river started in 1637. From then on the Catholic church carried out missionary work in the Amazonian hinterland, always following the major rivers as the only possible travel routes. The language the priests taught the indigenous people who settled in

288 Stefan Dienst their missions was Nheengatu. With speakers of different languages settling in one mission, that could lead to the demise of the indigenous languages and the result were Nheengatu-speaking communities far away from the coastal homeland of the language. Many of the people who continued to live in their traditional ethnic communities also acquired some knowledge of Nheengatu and for them it became the language used for contacts with the Neobrazilian society. Nheengatu retained this function (to an ever decreasing extent) until the beginning of the 20th century. 4. The spread of Portuguese Nheengatu is now forgotten in most places where it was once spoken, replaced by Brazil’s national language, Portuguese. Most of the languages whose speakers came into contact with the Portuguese or with Neobrazilians centuries ago are also long gone and little or nothing can be said about the influence Portuguese may have had on them before they were replaced by it or before their speakers fell victim to genocide or contagious diseases. Even in remote areas of Brazil, where numerous indigenous languages are still spoken, many are seriously endangered or moribund. But there are also others, like Kulina, which are still in much better shape. 5. Kulina3 Kulina belongs to the Arawan language family of south-western Amazonia (Figure 1). The major Kulina groups are found on the Purus river in Peru and Brazil (in the state of Acre) and on the Juruá river in Brazil (in the south of the state of Amazonas). All other languages of the Arawan family are or were spoken in Brazil, in the state of Amazonas. The Kulinas in Acre are unique among the indigenous peoples of that state in their degree of language preservation. Many of the older Kulina men and a few older women know Portuguese well enough to talk about issues which are relevant for everyday life in the jungle, but surprisingly younger men often have a more limited command of Portuguese than their fathers. Most women and all children are monolingual in Kulina. The present state of Acre was settled by Brazilian rubber tappers at the end of the 19th century, when it was a part of Bolivia. Due to the demands of the Brazilian settlers, Brazil pressured Bolivia into ceding the area, which became a part of Brazil in 1903.

289

Portuguese influence on Kulina

Arawan

Madihá

Kulina

Western Jamamadi

Deni

Eastern Jamamadi

Madi

Sorowahá

Banawá

Jarawara

Paumari

†

Arawá

Figure 1. The Arawan language family

5.1. Spanish influence It is not known when the Kulinas first came into contact with non-indigenous people on the Peruvian side of their territory. But there is some linguistic evidence that it was before they had contact with Brazilians. While the Kulinas living in Peru have various Spanish loans in their dialect, only two words of Spanish origin are used in the Brazilian varieties, matshito ‘machete’ and kotshiro ‘knife’ from cuchillo. These metal tools are among the Western products which have had the deepest impact on the daily lives of the Kulinas and they must have been among the first they obtained. That just their names are of Spanish origin indicates that the first Western influence came from the Peruvian/Bolivian side but lasted only a short time before Brazilian influence became dominant. 5.2. Nheengatu influence It is difficult to tell when members of the ethnic group whose language is discussed here first came into contact with Europeans or Neobrazilians since in the history of Amazonia the same ethnonym has often been used for unrelated groups and in many cases it is not possible to identify groups named in early sources. An indigenous people named ‘Curina’, living on the Amazon river, was mentioned as early as the seventeenth century (de

290 Stefan Dienst Figueroa et al. 1660–1684 [1986]: 73), but this was probably an unrelated group. The name ‘Curina, Culina, Culino, etc.’ then travelled along the Amazon and up the Juruá river, a tributary of the Amazon, where it was used for new ethnic groups the explorers encountered. The first account of ‘Culinos’ who can clearly be identified as Kulinas of the Arawan language family is from the British explorer William Chandless, who travelled up the Juruá river in 1867 (Chandless 1869). Even if the first contact of the Kulinas with non-indigenous Brazilians was as late as the 19th century, it was still at a time when Nheengatu was used as a lingua franca (although it was already in decline). For this reason, the first words that entered the Kulina language due to contact with Brazilians came from Nheengatu, not from Portuguese. The following Nheengatu loans are found in Kulina. Kulina dzara kari(w)a modobi mokawa nana warowa

‘non-indigenous person’ (obs.) ‘non-indigenous person’ ‘peanut’ ‘shotgun’ ‘pineapple’ ‘mirror’

Nheengatu (Stradelli 1929) iára ‘owner, master’ caryua ‘master’ mandubí ‘peanut’ mucáua ‘shotgun’ naná ‘pineapple’ uaruá ‘mirror’

The more conservative form kariwa for ‘non-indigenous person’ is used in the Juruá dialect of Kulina. In the Purus dialect, the /w/ has been dropped and the word has the form karia. The words modobi, nana and warowa are used in the Purus dialect of Kulina, but not in the Juruá dialect. 5.3. Portuguese influence There is no Nheengatu influence on Kulina any longer and in Brazil there is no Spanish influence any more, either. The influence of Portuguese, however, continues and can be expected to have an increasingly greater impact in the future. But at present, it is rather modest. Although the Kulinas are in constant contact with health workers, government officials, missionaries and researchers from other parts of Brazil and from abroad, the variety of Portuguese that influences their language the most is the one spoken by their non-indigenous neighbours, i.e. the Popular Brazilian Portuguese of rural Acre and southern Amazonas state.

Portuguese influence on Kulina

291

5.3.1. Vocabulary Portuguese influence is most noticeable in the Kulina vocabulary. The words discussed in this section are used by monolingual Kulina-speakers and can be considered established lexemes whose form has been adapted to the phonology of the language. Loans which have not (yet) been fully adapted to Kulina phonology are discussed in the following section. NOUNS Unsurprisingly, a lot of nouns relating to Western culture have been borrowed, although Kulina neologisms are also frequent, e.g. wawahina ‘syringe’, derived by reduplication from the verb wahina ‘to harpoon, to skewer’ and kokoro ‘fish-hook’, formed in the same way from the verb koro ‘to throw’. The following is a list of Portuguese loans for non-traditional items. Kulina dzinero haidzo itsada kahaniro katsatsa keneko koidze maratsia noparina panera pereko raraidza tahapa tsao

Portuguese dinheiro rádio enxada carneiro cachaça caneco colher melancia lamparina panela prego laranja tarrafa sal

‘money’ ‘radio’ ‘hoe’ ‘sheep’ ‘sugar-cane brandy’ ‘mug’ ‘spoon’ ‘watermelon’ ‘oil lamp’ ‘metal cooking pot’ ‘nail’ ‘orange’ ‘casting-net’ ‘salt’

Portuguese loans are not limited to words for recently introduced items. The names of some local animals, such as kamariáo ‘iguana’ (Port. camaleão), and pato ‘duck’ (Port. pato) have also been borrowed. In the case of some words which have been borrowed from Nheengatu into Portuguese, it is not clear if they entered Kulina from one language or the other, e.g. dzaboro ‘jabiru (a stork species)’ (Port. jaburu, Nheengatu jaburú).

292 Stefan Dienst VERBS There are also a small number of Portuguese verbs which have been borrowed into Kulina. Kulina verbs belong to two different morphological types, inflecting and non-inflecting. Inflecting verbs take prefixes and suffixes, whereas non-inflecting verbs are followed by an auxiliary which takes the affixes. (1)

inflecting verb o-wada-de4 1SG-sleep-PAST ‘I slept.’

(2) non-inflecting verb kona o-na-de swim 1SG-AUX-PAST ‘I swam.’

When Portuguese verbs are borrowed into Kulina, they always become non-inflecting verbs. (3)

hina, tarabaia i-na-na! come.on work 1NSG-AUX-HORT ‘Come on, let’s work!’ (tarabaia from Port. trabalhar ‘work’)

Occasionally, a Portuguese verb and its object together become a Kulina verb. (4)

ahidza ti-madi-hi, pakaheda ta-dza here.F 2-live-IMP.F pay.rent 2.AUX-NFIN ‘You will live here, paying rent!’ (pakaheda from Port. pagar renda ‘pay rent’)

NUMERALS Kulina has only two basic numbers, ohari- ‘(to be) one’ and pama- ‘(to be) two’, which are inflecting verbs. In the past, compound forms were used for three and four. Nowadays, Portuguese numerals are used from three onwards. Unlike the native numbers, they are not verbs. When they are used as noun phrase modifiers, they follow the head noun, like the native numbers and most other Kulina noun phrase modifiers, while in Portuguese modifying numerals precede the head. The examples below show the noun phrase ‘three weeks’ in Kulina, Popular Brazilian Portuguese and Standard Portuguese. While both words are Portuguese loans in Kulina, the constituent order is different from Portuguese.

293

Portuguese influence on Kulina

Kulina: semána tres week three ‘three weeks’

Popular Brazil. Portug.: três semana three week ‘three weeks’

Standard Portuguese: três semana-s three week-PL ‘three weeks’

The borrowing of European numerals is a typical phenomenon of Amazonian languages of Brazil. Many Amazonian languages have only a small set of native lexical numbers (Dixon and Aikhenvald 1999: 9). But many of the indigenous languages of the Peruvian and Bolivian lowlands have been under Quechua influence and they have often borrowed higher numerals from Quechua rather than Spanish.

5.3.2. Phonology Kulina has a simple phonology with 16 consonants, 4 vowels, word-final stress and mainly (C)V syllable structure with only a few instances of diphthongs. Table 1. Consonant phonemes

voiced obstruents voiceless unaspir. obstruents voiceless aspir. obstruents nasals flap approximant

bilabial

dental

alveolar

b p ph m

d t th

dz ts tsh n r

velar

glottal

k kh

h

w

SYLLABLE-FINAL /s/ There are no syllable-final consonants in Kulina. In Brazilian Portuguese, only a few consonant phonemes can occur syllable-finally and the only one found in a number of recent loans in Kulina is /s/, e.g. in bis.koi.ta from Portuguese biscoito ‘biscuit, cookie’. But it is not certain that this coda consonant will be preserved for a long time. The word for ‘school’, Portuguese escola, is pronounced iskola or iskora by people who know Portuguese, but from monolinguals tsikora can be heard, a form completely adapted to Kulina phonology. An alternative pronunciation for biskoita, which can be heard from children, is bitsikoita, with an epenthetic /i/ and a change of the

294 Stefan Dienst pronunciation of the consonant from [s] to [ts]. This and similar forms of other loans may eventually prevail, eliminating /s/ in coda position in established loans. But since new words are freely borrowed, the phenomenon is unlikely to disappear, just moving on from older loans to newer ones. DIPHTHONGS Another unusual phonological feature found in biskoita is the diphthong /oi/. In the inherited vocabulary of Kulina, only a few instances of the diphthong /ai/ are found, e.g. in maitha ‘yesterday’. The diphthong /oi/ is found in the word koiza (the name of a fermented beverage and the festival at which it is consumed). This word is probably a loan from another indigenous language, as similar words are found in unrelated neighbouring languages. The word may have entered Kulina before the first contact with Portuguese. While the existence of diphthongs cannot be regarded to be a result of Portuguese influence, borrowing from Portuguese has led to a significant increase of Kulina words containing diphthongs. The diphthongs /ao/ and /eo/, which occur in loans such as kamariáo ‘iguana’ from Portuguese camaleão and papéo ‘book, paper’ from Portuguese papel ‘paper’, are otherwise only attested in onomatopoeic words. PHONEME /ts/ The phoneme /ts/ is a recent innovation in Kulina phonology (Dienst 2005). It entered Kulina due to borrowing from neighbouring indigenous languages before the contact with Portuguese. But the number of loans containing this phoneme remained quite limited until the language started to borrow from Portuguese. With the continual borrowing from Portuguese, the frequency of /ts/ has significantly increased. The phoneme is used in Portuguese loanwords in place of the Portuguese phonemes /s/ and /
/ as well as the allophone [] of the phoneme /t/, which occurs before /i/ in most Brazilian varieties of Portuguese, including those of Acre and southern Amazonas. Kulina boratsa ‘biscuit, cookie’ katsado ‘hunter’ paretsi ‘Indian other than Kulina’

Portuguese bolacha [bo'la
a] ‘dry biscuit’ caçador [kasa'do()] ‘hunter’ parente [pa'i] ‘relative’

Portuguese influence on Kulina

295

STRESS Stress always falls on the last syllable of Kulina words, but in recent loans, the Portuguese stress can be preserved, e.g. in iskóra ‘school’ from escola. But, as mentioned above, this word can be completely assimilated and in the pronunciation tsikora, the stress is word-final. CONCLUSIONS Due to Portuguese loans, two formerly marginal phenomena of Kulina phonology, diphthongs and the phoneme /ts/ have become considerably more frequent. The phonological features which are innovations due to Portuguese influence, syllable-final /s/ and phonemic stress, are restricted to recent Portuguese loans. If it wasn’t for the continued influence of Portuguese, those features would probably soon be lost due to the gradual complete phonological assimilation of loans. 5.3.3. Other influences Portuguese does not seem to have had any influence on the grammar of the largely monolingual speech community in Acre. A Portuguese feature which could be observed in the usage of a bilingual speaker living in the town of Eirunepé in Amazonas is the contrastive use of two demonstratives. (5)

a-hari towi o-ne-hera-ni, a-kha-hari towi DEM-M GOAL 1SG-AUX-NEG.F-DECL.F DEM-REM-M GOAL o-na-haro 1SG-AUX-NAR.F ‘I don’t want this one, I want that one.’

In this example, the remote demonstrative akhahari, referring to something further away from the speaker, contrasts with the demonstrative ahari, without the remoteness-suffix -kha. The latter demonstrative refers to something closer to the speaker. This reflects the use of demonstratives in Portuguese. Though Kulina has a variety of demonstratives, which can be used to express relative distance from the speaker, speakers in Acre never use them contrastively, as in the example from Amazonas above. They would use, for example, ‘ahari… ahari…’, corresponding to English ‘this (one)…that (one)…’.

296 Stefan Dienst The speaker from whom example (5) comes has spoken Portuguese since her childhood, which is unheard of among the Kulina in Acre and also highly exceptional among the Kulina in Amazonas. There is no indication that the structural influence of Portuguese seen in the example is found among the Kulina of Amazonas in general. 5.4. Conclusions The Kulinas have been in contact with Portuguese-speakers for more than a century, but most of them remain monolingual. The main impact of Portuguese has so far been an enrichment of the vocabulary. In some cases Portuguese loans have replaced older words. There has been some influence on the phonology, but none on the grammar of the overwhelming majority of speakers who continue to live in indigenous villages. Given the present healthy state of the language, it is likely to survive for much longer than most other Amazonian languages which are still spoken today. The contact with Portuguese-speakers will probably become more intense over time and the language is bound to show a stronger Portuguese influence in the future. Notes 1.

2. 3.

4.

The earliest traces of Western influence found in Kulina, an indigenous language of Brazil and Peru, are loans from Spanish and Nheengatu, the old lingua franca of northern Brazil. The more recent, but also more enduring influence of Portuguese has left its mark on Kulina phonology as well as the vocabulary. But there is as yet no perceptible Portuguese influence on the grammar of most speakers. The information on Nheengatu is based on Rodrigues (1996). The Kulina data are based on the author’s fieldwork in the Brazilian state of Acre in 2002/2003, 2004 and 2005 and a short visit to the state of Amazonas in 2004. aspir. = aspirated, AUX = auxiliary, DECL = declarative, DEM = demonstrative, F = feminine, GOAL = goal, HORT = hortative, IMP = imperative, M = masculine, NAR = narrative, NEG = negation, NFIN = non-finite, NSG = non-singular, obs = obsolete, PAST = past tense, PL = plural, Port. = Portuguese, REM = remote, SG = singular, unaspir. = unaspirated.

Portuguese influence on Kulina

297

References Chandless, William 1869 Notes on a journey up the River Juruá. Journal of the Royal Geographical Society 39: 296 –311. de Figueroa, Franciso, Cristobal de Acuña et al. 1660–84 [1986] Informes de Jesuitas en el Amazonas. Iquitos: Centro de Estudios Teológicos de la Amazonía. Dienst, Stefan 2005 The innovation of s in Kulina and Deni. Anthropological Linguistics 47: 424–441. Dixon, R. M.W. and Alexandra Y. Aikhenvald 1999 Introduction. In The Amazonian languages, R.M.W. Dixon and Alexandra Y. Aikhenvald (eds.), 1–21. Cambridge: Cambridge University Press. Rodrigues, Aryon Dall’Igna 1996 As línguas gerais sul-americanas. Papia 4 (2), 6 –18. (Also available online: http://orbita.starmedia.com/~i.n.d.i.o.s/textos/txt009lg.htm) Stradelli, Ermanno 1929 Vocabularios da Lingua Geral, Portuguez – Nheêngatú e Nheêngatú – Portuguez. Rio de Janeiro: Livraria J. Leite.

Creolization and the fate of inflections1 John Holm

1. Introduction Until recently, most linguists studying creole languages assumed that an inherent part of the creolization process (whose exact mechanisms are still not well understood) was the loss of any inflectional morphemes in the lexical donor language in any pidgin or creole languages that grew out of contact. For example, Bickerton (1999: 69) asserts that With few if any exceptions, bound morphemes are either lost completely in the process of pidginization that immediately precedes creolization or (less often) assimilated by lexical items.

(One example of such assimilation of a superstrate inflection is the São Tomé Creole Portuguese word for ‘wave’: zonda from Portuguese as ondas ‘the waves.’) Even more recently, McWhorter (2005: 156) refers to the creole prototype as having “no inflectional affixes” (although he immediately cites an apparent exception, to be discussed below). However, this characterization of creolization is no longer tenable. As creolists from Stolz (1989) to Bakker (2002) have demonstrated, pidgins and basilectal creoles can indeed have inflections. Clements (1996) established that the Indo-Portuguese of Korlai has not only katad ‘sung’ (cf. Portuguese [P] cantado), but also katan ‘singing’ (P cantando) and kato ‘sang’ (P cantou). Since the Portuguese retreated from Korlai in 1740 and there has been no direct contact between Korlai CP and Portuguese since that time, the creole’s inflections cannot credibly be attributed to decreolization and must, therefore, have existed since the language’s genesis. What this implies is no less than a paradigm shift in creole linguistics. How could such a basic misunderstanding have occurred? To answer this question, this paper will examine what happened to superstrate inflectional morphemes in those creoles lexically based on Portuguese by systematically comparing the fate of inflections in the Portuguese noun phrase (section 2) and verb phrase (section 3) in five geographical groups of creoles which also reflect typologically distinct substrate languages:

300 John Holm 1) The Upper Guinea creoles include those of the Cape Verde Islands (CV) and Guiné-Bissau (GB); their structure reflects a substrate (and, for GB, an adstrate) of West Atlantic and Mande languages, groups within the Niger-Conger family. 2) The Gulf of Guinea creoles include four varieties found on three islands: São Tomé (São Tomense [ST] and Angolar); Príncipe (Principense); and Annobón (Fa d’Ambo). Their substrate languages also belong to the Niger-Congo family, but to different branches: Benue-Congo and Bantu. 3) The Indo-Portuguese creoles, spoken in India and Sri Lanka, can be divided into those with Indic substrates (e.g. Marathi, Gujarati, and Sinhala) and those with Dravidian substrates (e.g. Malayalam, Kannada and Tamil). 4) The Malayo-Portuguese creoles include Papia Kristang (spoken in the city of Malacca in Malaysia) and Batavian Creole Portuguese (once spoken in what is now Jakarta, with other varieties once spoken elsewhere in Indonesia). The substrate of both is Bazaar Malay, a simplified form of contact Malay, as well as Malay itself and Hokkien Chinese. 5) The separate status of the moribund Sino-Portuguese varieties is debated. Once spoken in Macao and Hong Kong, their substrate was assumed to be Cantonese, whereas it may in fact have been Malayo-Portuguese due to settlement patterns. The conclusions drawn in section 4 will relate the substrate-influenced typological differences among these groups of creoles to the assumptions creolists have made about the loss of inflections being an inherent part of the creolization process.

2.

Noun phrase inflections

2.1. Number inflections Portuguese nouns and most of their modifiers take a plural inflection -s: (1)

Portuguese este livro this book

estes livros these books

Creolization and the fate of inflections 301

2.1.1. Upper Guinea creoles Kihm (1994: 131) notes that “Plural marking is…the only inflectional morphology there is in [Guiné-Bissau] Kriyol.” Based (at least in part) on the Portuguese plural -s, nouns denoting humans (or humanized animals) can take a plural -s after vowels (2) or -is after consonants (3): (2)

GB CP (Kihm 1994: 132) omi ‘man’ omis ‘men’

(3)

GB CP (Kihm 1994: 132) minjer ‘woman’ minjeris ‘women’

However, unlike Portuguese, GB CP only marks the plurality of such nouns when it is relevant and not already implied by a numeral or quantifier. Moreover, there is no number agreement with adjectives or other modifiers in basilectal varieties, although this can occur in more decreolized lects, as in (4): (4)

GB CP (Kihm 1994: 132) sapatus altus ‘high [-heeled] shoes’

While this is likely to be a set phrase, it is worth noting that in (4) above sapatus is not [+human] either, and one might assume a general breakdown of rules in decreolizing varieties that could account for the very existence of GB CP plural -s in the first place. However, Intumbo (p. c.) argues against this being the case, given the inflection’s regular occurrence in basilectal varieties. Support for the pluralizer’s early presence in GB CP can be found in the inflectional marking of plurality in its West Atlantic substrate/adstrate languages such as Balanta, which make a morphological distinction between the singular and plural forms of class-marking prefixes on nouns (Intumbo 2005). The same pluralizing inflection with similar grammatical constraints occurs in Cape Verdean CP, both in the relatively basilectal Sotovento varieties (Baptista 2002: 38) and in the more acrolectal Barlavento varieties (Holm and Swolkien 2004). Baptista (2003: 328) notes Comrie’s observation that

302 John Holm It is not surprising, given the universal that plural distinction is more likely higher up the animacy hierarchy, to learn that this marker is first made obli(Comrie 1981: 223) gatory [in Tok Pisin] with human nouns.

The assumption that the presence of this inflection is the result of decreolization (an assumption echoed in the study of many other Atlantic creoles) does not stand up to a close examination of the sociolinguistic facts in the case of the Barlavento island of São Vicente (Holm and Swolkien 2004) and is further undermined by its existence in basilectal varieties of GB CP, which are historically closely related to CV CP. 2.1.2. Gulf of Guinea creoles Ivens Ferraz (1979: 60–61) notes that in São Tomé CP the same form of a noun is used for both singular and plural, but to stress plurality the third person plural pronoun (in~e or n~e; ~ represents the nasalization of the following vowel) can be placed before the noun: (5)

ST CP (Ivens Ferraz 1979: 60–61) n~e mwala 3PL2 woman ‘the women’

He further notes a parallel construction in Angolar, Principense, and various African languages, concluding that São Tome CP has a “virtual absence of inflexion” (Ivens Ferraz 1979: 60–61). 2.1.3. Indo-Portuguese Clements (1996) gives a number of examples of Korlai CP in which the base form of the noun is used in plural contexts, i.e. with no plural inflection: (6)

Korlai CP (Clements 1996: 164–167) d kadz me in house itself Pedru su doy kadz Pedru GEN two houses

Creolization and the fate of inflections 303

“There is no plural marking in Korlai…Daman CP…has grammaticalized tud [‘all’] as a plural marker for nouns” (Clements, p.c.). Cardoso (p.c.) points out that this seems to be the case in the Indo-Portuguese of Diu as well, although tud is not used consistently: when a numeral or quantifier cooccurs with the noun, it never takes plural marking. He also encountered the marking of plurality via reduplication (e.g. moir-moir ‘Muslims’ from P moiro/mouro idem.), but this seems to be a fossilized case rather than a productive device. Dalgado (1998 [1900]: 81) noted that in the Indo-Portuguese of Ceylon (modern Sri Lanka), Em regra, os substantivos formam o plural com o acrescentamento de um s: animals, reptils, pãos, coraçãos, pecadors. [As a rule nouns form the plural by adding an s…]

He went on to list a number of exceptions, including a word meaning ‘sons and daughters’, fifes or fifi (Dalgado 1998: 81), noting that in the indigenous languages there are words equivalent to mother father for ‘parents’ and to daughter son for ‘children of both sexes.’ Although Tamil does not use reduplication to pluralize nouns as does Malay (section 2.1.4), Ivens Ferraz (1987: 352) pointed out that the existence of such indigenous forms “could assist in incorporation of other reduplicated forms”, noting that such plural forms could be found in the Indo-Portuguese of Diu, Mangalore and Nagappattinam as well. Albring and Lourenço (2004: 30) attribute their presence to the influence of Malayo-Portuguese on the Indian varieties due to a presença…de malaios em Sri Lanka e na Índia, canalizados, por exemplo, para o serviço militar, sobretudo pelos Holandeses. [the presence…of Malays in Sri Lanka and India, brought in, for example, for military service, especially by the Dutch]

In the modern Indo-Portuguese of the Sri Lankan community of Batticaloa, Smith (1979b: 197) notes how colloquial Batticaloa Portuguese (BP) has converged with colloquial Batticaloa Tamil (BT) in its use of suffixes for marking both number (PL plural) and case (DAT dative): (7)

BP/BT Smith (1979b: 197) a. BP ba:k b. BP ba:k -s BT ma:Tu BT ma:Tu -ka(L) cow (NOM) PL cow (NOM)

304 John Holm c. BP ba:k -p BT ma:TT -ukku DAT cow

d. BP ba:ka -s -p BT ma:Tu -kaL -ukku cow PL DAT

Smith reports that Batticaloa CP contains a great number of morphological and syntactic features with Tamil (i.e. Dravidian) and Sinhala (Indic) counterparts, such as case inflections on nouns and verbal inflections (see sections 3.2.3 and 3.3.3), which are not to be found in the Sri Lanka CP described by Dalgado (1900). All of the CP speakers in Batticaloa also speak Tamil – often better than CP – so it is possible that the creole’s morphology is the result of recent wholesale borrowing associated with language death. However, Smith found similar features in the CP used in a Sinhala-speaking region. Since the speakers of the latter variety have not been in contact with the Batticaloa community, he hypothesizes that the morphological features date at least from the 19th century lingua franca and perhaps earlier. Smith (1984: 298) concludes that It is not clear whether the literary language is representative of an earlier spoken variety or whether it was simply an invention of educated churchmen based on the variety found in songs, the literary representations of other varieties of Indo-Portuguese and some elements of standard Portuguese.

In a recent reevaluation of the data, Smith (2005) concludes that the “literary” variety that Dalgado studied was largely an invention of Wesleyan missionaries. 2.1.4. Malayo-Portuguese Baxter (1988: 49) points out that the difference between count and noncount nouns in Papia Kristang is that the former reduplicate: (8)

Papia Kristang krensa ‘child’

krensa-krensa ‘children’

While adjectives are reduplicated for intensification and verbs are reduplicated for iteration, with nouns reduplication is a pluralizing device in a language without inflections.

Creolization and the fate of inflections 305

2.1.5. Sino-Portuguese Batalha (1974: 34) notes that Macau CP also uses reduplication to indicate the plural: (9)

Macau CP quiança-quiança ‘as crianças’ [‘the children’]

However, she points out that the semantics of reduplication in Cantonese differs, being closer to ‘all the children’, suggesting that Malay influenced this creole feature more (Batalha 1974: 34). 2.2.

Gender inflections

In Portuguese, most masculine nouns and their modifiers end in -o while most feminine nouns and their modifiers end in -a: (10) Portuguese o carro novo the car new ‘the new car’

a bicicleta nova the bicycle new ‘the new bicycle’

2.2.1. Upper Guinea Kihm (1994: 126) refers to Guiné-Bissau CP’s “conspicuous absence of Gender; even third person pronouns do not distinguish it.” Moreover, “agreement is entirely absent” (Kihm 1994: 126). In Sotovento varieties of Cape Verdean CP, there are gender-linked pairs of nouns like the following: (11) CV CP (Baptista 2002: 42) fidju ‘son’ vs. fidja ‘daughter’ noibu ‘groom’ vs. noiba ‘bride’ These contrasts, confined to [+animate] and especially [+human] nouns, appear to follow those of Portuguese (cf. filho versus filha; noivo versus noiva), but the CV CP forms could well be single morphemes like their English equivalents, i.e. not have gender inflections (see section 2.2.4 below).

306 John Holm While CV CP adjectives, as a rule, do not agree in number with the nouns they modify, what appears to be gender agreement does occur with both singular and plural head nouns if these are [+ human]: (12) CV CP (Baptista 2002: 66) un minina bonita a girl beautiful-FEM ‘a beautiful girl’ uns minina bonita some girl beautiful-FEM ‘some beautiful girls’ However, adjectival gender agreement is optional, and with [–animate] nouns non-agreement is preferable in some varieties (Baptista 2002: 67). In the Barlavento variety of São Vicente, however, instances of such agreement occur: (13) CV CP (Holm and Swolkien 2004) madera torta wood curved-FEM ‘curved wood’ It cannot be ruled out that the above is something of a set phrase, in which case the retention of Portuguese gender agreement would be more likely, as in the retention of number agreement in (4) above. In general, apparent inflections of grammatical gender seem more likely to be a result of decreolization than do occurrences of the noun pluralizer -s, given the latter’s presence in basilectal varieties of both CV CP and GB CP. 2.2.2. Gulf of Guinea Ivens Ferraz (1979: 60) claims that in São Tomé CP “a few words inflect for masculine and feminine”: (14) São Tomé CP sglu ‘son-in-law; father-in-law’ sgla ‘daughter-in-law; mother-in-law’ by analogy with Portuguese sogro ‘father-in-law’ and sogra ‘mother-in-law.’ However, see section 2.2.4 below.

Creolization and the fate of inflections 307

2.2.3. Indo-Portuguese Clements (1996: 117) points out that “Although all the nominal paradigms of both Marathi and Portuguese exhibit gender (two/three), person and number distinctions, no significant vestige of this has survived in KP [Korlai Portuguese].” Cardoso (p.c.) notes that it is the same in Diu CP: Crucially, numerals do not exhibit any gender marking; otherwise we might think we just don’t ‘hear’ the marker in the noun (since in Diu CP words tend to end with the consonant of the last stressed syllable in the lexifier). There are some gender pairs, such as sior ‘mister’/siar ‘lady’ (e.g. no ‘siar d Fat ‘Our Lady of Fátima’), but very rare.

Gender marking appears to be equally uncharacteristic of authentic Sri Lanka CP since Smith (1979b: 214) decided that “For the purpose of the present discussion, texts exhibiting grammatical gender distinctions or SP [Standard Portuguese] verb morphology will be ignored.” Smith notes that “SLP has the same lexicalized gendered nouns as in other varieties: fiiyu ‘son’/fiiya ‘daughter’… But there is no grammatical gender: uN kadeera ‘one chair’, boom kondisaam ‘good condition’…” (p.c.) 2.2.4. Malayo-Portuguese According to Tomás (2004: 139–140) the existence of pairs based on Portuguese etyma such as Portuguese sogro ‘father-in-law’ and sogra ‘mother-inlaw’ are not proof of gender inflections in Papia Kristang – or other creoles: Não encontramos em kristang a categoria morfológica género. Encontramos, contudo, uma distinção sexual nalguns nomes…caracterizados pelo traço semântico [+HUMANO], e pertencendo na sua maioria à nomenclatura do sistema de parentesco, conservam uma distinção com raizes etimológicas no português…sogra/sogru. [‘We do not find in Papia Kristang the morphological category gender. However, we do find a distinction of natural sex in some nouns … characterized by the semantic feature [+HUMAN] and mostly belonging to a system of naming family relationships; they retain this distinction through their Portuguese etymological roots…[e.g.] sogra/sogru ‘mother-in-law; father-in-law.’ (my translation, JH)]

Papia Kristang, like Bazaar Malay, is an isolating language without inflections. It would seem incompatible with this fact to analyze these words as having inflections marking grammatical gender. However, Baxter (p.c.)

308 John Holm notes that this could still be considered a kind of narrow gender morphology, limited to agentives (e.g. kuzinyeru/kuzinyera ‘cook’, other [+human] nouns (e.g. biuba/biubu ‘widow/widower’) and certain adjectives (e.g. dodu/ doda ‘crazy’). Maurer (p.c.) notes that this concept of “narrow gender marking” holds for the Creole Portuguese of Batavia and Tugu, where there are also distinct agentive forms ending in -dor and -dera, and contrasting forms such as bivu ‘widower’ (cf. P viuvo idem) and biva ‘widow’ (P viuva). Tomás (p.c.) further notes that in these two varieties, “The [N + omi/femi] strategy for sex differentiation seems to have included such [+ human] nouns as irmang ommie/irmang moleer ‘brother/sister’.” 2.2.5. Sino-Portuguese According to the literature, Macau CP does not mark nouns or their modifiers for grammatical gender (Baltalha 1974: 25, 34). However, Baxter (p.c.) again takes issue, noting that Macau CP does “show partial and variable gender… concord.” 3.

Verb phrase inflections

3.1. Tense inflections Portuguese verbs have many inflections for person, tense and mood, e.g. (15) Portuguese cantar ‘to sing’ cantava ‘s/he was singing’ cantou ‘s/he sang’ 3.1.1. Upper Guinea At first sight, the Guiné-Bissau CP past marker ba looks like the Portuguese inflection -va, used to mark the imperfect tense of first conjugation verbs: (16) GB CP (Kihm 1994: 99) Mandingas kontinwa nega ba rasa. Mandings continue refuse PAST pray ‘The Mandings kept on refusing to pray.’

Creolization and the fate of inflections 309

However, GB CP ba is clearly not a bound morpheme: pronouns can intervene between it and the verb: (17) GB CP (Kihm 1994: 99) N konta u ba… 1s tell 2s PAST ‘I told you…’ And it can even occur without a verb: (18) GB CP (Kihm 1994: 99) Kil omi ba i kin that man PAST it who ‘Who was that man?’ Kihm suggests conflation with GB CP kaba ‘finish’ (cf. P acabar idem) and substrate forms: “We may suppose that both /-va/ and (a)cabar, the latter reinforced by the phonetically similar Manjaku, Diola-Fogny, and perhaps still other languages, ba, had to enter into the process” (Kihm 1994: 103). In Cape Verdean CP, however, Baptista (2002: 201) finds that “…-ba is a verbal inflection found exclusively bound to verb stems…whereas ba is a non-inflectional (unbound) Tense marker in GBC [Guiné-Bissau Creole].” In fact, the inflectional status of CV CP -ba is of some theoretical importance since “the presence of a single inflectional affix on the verb stem (-ba) is a sufficient cue to trigger verb movement” (Baptista 2002: 266). 3.1.2. Gulf of Guinea São Tomé CP has an etymologically related past marker tava (cf. P estar ‘be’, estava ‘was’): (19) São Tomé CP (Ivens Ferraz 1979: 83) e tava ka ngungunu ku zõ 3s PAST PROG grumble with John ‘He was grumbling at John’ However, it is a preverbal marker of tense rather than an inflection.

310 John Holm 3.1.3. Indo-Portuguese Clements (1996: 110–112) notes that the highly inflected verbal paradigm of Marathi has forms marked for three genders, as well as person and number, for many tenses, including the simple past. The Portuguese verbal paradigm has forms marked for person and number for many tenses, including the simple past, as in (15) above. Korlai CP has forms, not marked for person or number, for many tenses, including the simple past: (20) Korlai CP (Clements 1996: 112) kato ‘sang’ He comments further that the retention of verbal suffixes in KP is a clear example of a marked feature being incorporated into KP due to mutual linguistic accommodation. Given that the contact situation involves only two languages, it is possible that the homogeneity of the situation may have fostered their retention, even though formally the suffixal markers in Portuguese and Marathi do not resemble each other. (Clements 1996: 118)

Elsewhere Clements argues that “the retention of kato is due to the fact that Marathi also has suffixes and the KP suffix -o could thus be easily interpreted as a past morpheme” (Clements 1996: 229). Cardoso (p.c.) notes that “In Diu CP there is only one form for each tense, and it seems to be invariably based on the third person singular in Portuguese. There is a form for the Present (Es é rezão é ‘That is the reason’) [and] one for the Past (Que ficô? ‘What happened?).” Regarding Sri Lankan Indo-Portuguese, Smith notes that it is “interesting that Sinhala and Tamil both have a clear distinction between present and past, but none of the tense affixes of Portuguese survive in SLP” (p.c.). 3.1.4. Malayo-Portuguese Papia Kristang has no inflections and uses preverbal markers to indicate tense and aspect: (21) Papia Kristang CP (Baxter 1988: 180) Patrick ja falá ku eli bai kaza Patrick PF tell R 3s go house ‘Patrick told him to go home.’

Creolization and the fate of inflections 311

Tomás (p.c.) notes that in Batavian Malayo-Portuguese (Schuchardt 1890) “there are two forms for the verb ‘to make’: faay and fajie [fadzi]. The latter is presented (p. 116) as a past form: Noos faay ‘we make’ Nos fajie ‘we made’ Nos ja faay ‘we have made’… But no other verb shows anything like this.” She concludes that this variation is unlikely to indicate tense marking. 3.1.5. Sino-Portuguese Macau CP is structurally very similar to Papia Kristang, also using preverbal markers rather than inflections to indicate tense and aspect: (22) Macau CP (Batalha 1974: 35) já vai ‘(he) went’ 3.2.

Present participle inflections

Portuguese verbs form the present participle by dropping the -r of the infinitive and adding -ndo: (23) Portuguese cantar cantando ‘singing’

ferver fervendo ‘boiling’

3.2.1. Upper Guinea Kihm (1994: 251ff.) suggests that these participial forms may have been reanalyzed due to their chance formal similarity to a derivational morpheme in local languages’ causative verbs: (24) Mandinka fáa ‘be full’ bèng ‘meet’ dómori ‘eat’

fandi ‘fill’ bendi ‘bring together’ dómorindi ‘feed’

He speculates that speakers of these languages reanalyzed Portuguese pairs like the following as having the same derivational morpheme forming causatives:

312 John Holm (25) Portuguese ferver ‘boil’ mamar ‘suck’

aferventar ‘boil’ [transitive] amamentar ‘suckle; give suck’

This gave rise to the derivational marking of causativity in some GB CP verbs (although the rule no longer seems to be productive): (26) GB CP (Kihm 1994: 251) firbi ‘boil’ firbinti ‘boil’ [transitive] kume ‘eat’ kumente ‘make eat’ Although the present participial ending in Portuguese is an inflectional morpheme, the GB CP morpheme is derivational. Causative verbs with such derivational morphemes are apparently unknown in the other Upper Guinea creole, Cape Verdean. 3.2.2. Gulf of Guinea The Gulf of Guinea varieties of Creole Portuguese seem to be strictly isolating languages; none of them has preserved the Portuguese present participial inflection as such to the present author’s knowledge. 3.2.3. Indo-Portuguese Clements (1996: 125) notes the survival of the Portuguese gerund inflection (e.g. cantando): (27) Korlai CP (Clements 1996: 111) katan (The absence of the final Portuguese -do in its Korlai reflex is because all syllables after the tonic accent in Portuguese are lost in Indo-Portuguese; another feature of Korlai CP phonology is the denasalization of vowels.) This suffix is used in the formation of several imperfective tenses, e.g. (28) Korlai CP (Clements 1996: 125) a. Luiz kume-n b. Luiz ti kume-n Luiz PAST eat-IMPF Luiz eat-IMPF ‘Luiz is eating’ ‘Luiz was eating.’

Creolization and the fate of inflections 313

Clements notes that the creole form with -n is also used in the following construction: (29) Korlai CP (Clements 1996: 125) el vi-n, n fon l hedze. S/he coming, we phone will do ‘When s/he gets here, we’ll call.’ He suggests that “This development seems to have been influenced by the use of a Marathi absolutive form in -n” in parallel constructions (Clements 1996: 125). As for Sri Lanka CP, Smith (1979a: 194) discusses an imperfect participle suffix -t m which marks a present progressive with tm ‘is’ (30) and a past progressive with tiñ ‘was’: (30) Sri Lanka CP (Smith 1979a: 194) eli kum-tm tm he eat-IMP-PPL is ‘He is eating.’ Smith (p.c.) notes that “taam also occurs as a conjunctive/emphatic particle, clearly derived from também [P ‘also’]; I suspect that progressive taam is from the same source, but the semantic motivation seems lacking.” 3.2.4. Malayo-Portuguese Portuguese gerunds were apparently not normally found in the MalayoPortuguese of Batavia. Maurer (p.c.) came across a single gerund in the extensive texts published by Schuchardt (1890): passando ‘passing’, which seems to be used as in Portuguese. 3.2.5. Sino-Portuguese Macau CP does not have forms based on the Portuguese present participle, to the best of the present author’s knowledge.

314 John Holm 3.3.

Past participle inflections

Portuguese verbs form the past participle with the inflection -do: (31) Portuguese cantar ‘to sing’ cantado ‘sung’ comer ‘eat’ comido ‘eaten’ 3.3.1. Upper Guinea Kihm (1994: 243) notes that in Guiné-Bissau CP “Passive is an entirely productive formation affecting all transitive verbs and materialized as a /-du/ suffix. This morpheme is obviously descended from the Portuguese past participle suffix /-do/. Regularization has been thorough, however, so that there is no trace in Kriyol of the so-called “irregular” forms such as escrito ‘written’ (Kriyol skirbidu ‘be written’), feito ‘done’ (Kriyol fasidu ‘be done’), and so forth.” This form is a verb with passive meaning, which can take as its subject either the direct object (32) or (unlike passive constructions in Portuguese) the indirect object (33) of the corresponding active verb: (32) GB CP (Kihm 1994: 245) Dinyeru pistadu el. money be-lent him. ‘Money was lent to him.’ (33) GB CP (Kihm 1994: 245) I pistadu dinyeru. he be-lent money ‘He was lent money.’ These constructions are characteristic of basilectal varieties; such passive verbs behave just like their active counterparts regarding tense, mood and aspect markers, and there is no question of their having a “zero copula”, although constructions with yera ‘be’ can occur in the acrolect. Kihm characterized the formation of passive verbs in GB CP as a lexical process, i.e. one involving a derivational suffix rather than an inflection (Kihm 1994: 243).

Creolization and the fate of inflections 315

There are similar verbs ending in -du with passive meaning in Cape Verdean CP (Baptista 2002: 112–113), although the status of the suffix as an inflection is unclear. Baptista refers to “the use of the -du inflection on the verb stem” (Baptista 2002: 112), although elsewhere she says that -ba “is the single inflectional affix on the verb stem” (Baptista 2002: 266). In the Barlavento variety of São Vicente, the Portuguese passive construction is found, complete with the irregular forms of both the auxiliary verb ser and the past participle: (34) São Vicente CV CP (Dominika Swolkien, p.c.) Trabói foi fet work AUX PPL ‘The work was done.’ 3.3.2. Gulf of Guinea One of the first morphemes considered as possible counterevidence to the view that creoles have no inflections was -du in Príncipe CP (Holm 1988, 1989). Pidgins and creoles are sometimes claimed to be languages without any inflectional morphology whatsoever. While this seems to be true of most Atlantic varieties that are not decreolizing, there are some ambiguous cases. The Portuguese-based creoles of West Africa have what appears to be an inflectional marking of the past participle, e.g. Príncipe CP fá ‘speak’, fádu ‘spoken’ corresponding to P falar and falado respectively (Günther 1973). Forms ending in -du can be used actively (e.g. n táva fádu ‘I had spoken’) but such construction are rare; they are primarily used to give verbs a passive meaning (e.g. Ótu samádu pédu ‘The other was called Peter’). It is possible that this creole -du plays a semantic and syntactic role that is distinct from that of its Portuguese model. (Holm 1988: 95–96)

In other words, the Gulf of Guinea creoles’ -du (or -ru in some varieties) seems to be parallel to the Upper Guinea creoles’ -du as described by Kihm (3.3.1). McWhorter (2005: 156) notes that “Fa D’Ambu, with its occasional use of the suffix -du to form participial adjectives (xaba ‘to finish,’ xabadu ‘finished’ [Post 1995: 195–196]) depart[s] slightly from the prototype” of creoles without inflections. Maurer (1995: 91) analyzes Angolar -ru as a suffix. Although he doesn’t stipulate whether he considers it derivational or inflectional, he calls the result a past participle:

316 John Holm Le participe passé est formé à l’aide du suffixe -ru ~ -du (dérivé du suffixe portugais -do), ajouté à la racine verbal. Le participe passé peut modifier un substantif. [‘The past participle is formed with the help of the suffix -ru ~ -du (derived from the Portuguese suffix -do) added to the root of the verb. The past participle can modify a noun.’ (my translation, JH)]

(35) Angolar (Maurer 1995: 91) kikiê thagaru poisson saler-PP ‘du poisson salé’ Lorenzino (1998a, 1998b) points out that the differences in usage between Angolar CP -ru and São Tomé CP -du can be attributed to the closer contact which the latter had with Portuguese on plantations in the 16th century, as opposed to the relative isolation of the maroons who developed Angolar. Both Angolar Ta malaru (T represents a voiceless interdental fricative) and ST CP sa maladu developed from the Portuguese passive construction está amarrado ‘be tied/moored.’ However, the ST affix is restricted to Portuguese-derived verbs in Portuguese-like participial constructions like the preceding, while the Angolar affix has been extended to non-verbs (e.g. siaru ‘full [emphatic]’ from sia ‘full’; cf. P cheio idem.) and non-Portuguese words (e.g. kutaru ‘dry’ from Kimbundu kukuta ‘to dry’). 3.3.3. Indo-Portuguese Clements (1996: 109ff.) notes that Portuguese, Marathi and Korlai CP all form a number of tenses with the past participle: (36) Portuguese Present perfect Past perfect

tem cantado tinha cantado

‘he has sung’ ‘he had sung’

(37) Marathi Present perfect Past perfect

gayla ahe gayla hota

‘he has sung’ ‘he had sung’

(38) Korlai CP Present perfect Past perfect

katad ti katad

‘(he) has sung’ ‘(he) had sung’

Creolization and the fate of inflections 317

Again, the fact that Korlai CP retained (part of) the Portuguese past participle suffix seems to be due to the superstrate and substrate sharing parallel composite tenses for the same grammatical category. Cardoso (p.c.) notes a similar form in Diu CP: ja tin dized. As for Sri Lankan CP, Smith (1979a: 195) notes that the suffix “-tu, ‘perfect participle’, is actually a post-clitic whose preferred position is attached to the verb”: (39) Sri Lankan CP eli kum-tu j -k -nd

he eat-PFC:PPL PAS-PFC-go ‘He ate and went away.’ (Lit. ‘He went away having eaten.’) Smith notes that “the BT suffix is phonetically [ittu] which could easily be the source of BP -tu” (p.c.). (40) Sri Lankan CP (Smith 1979b: 200) BP (ja:)-prend-tu BT paTi-cc-i´r-´r-u PAS-study-PFC study-PAS-PFC-PAS-PPL both: ‘having studied [and finished]’ Smith (p.c.) further notes that BP also has a passive construction using the participle in -du: (41) Sri Lankan CP pooy faya tEEmpu taam nuka fikaa fay-eedu EMPH NEG AUX do-PPL can do time ‘The time that we could do it [build a house], it wasn’t done.’ He also notes that this variety uses -du forms as derived adjectives and that these constructions seems to be fairly productive: (42) Sri Lankan CP Oorta inciidu kaaza ‘The compounds are full of houses.’ 3.3.4.–5. Malayo-Portuguese and Sino-Portuguese Baxter (p.c.) reports that both Kristang and Maquista have “deverbal adjectives based on the participle…These can mostly be modified by ja.” Maurer

318 John Holm (p.c.) notes that in Batavia and Tugu CP “there are some examples of past participles, used attributively and predicatively, but I haven’t found examples marked by ja.” Tomás adds that “Schuchardt [1890] mentions (p. 207) instances of Portuguese past participles, which in Tugu (19th century) function as nouns: iskrebedu ‘writing’, kontadu ‘account, count’…” 4. Conclusions Table 1. Portuguese inflections in the Portuguese-based creoles

Upper Guinea Gulf of Guinea Indo-Portuguese Malayo-Portuguese Sino-Portuguese

2.1. Plural -s

2.2. Gender -o/-a

3.1. Past -va/-ou

3.2. Pres.Part. -ndo

3.3. Past Part. -do

+ – +/– – –

+/– – – – –

+/– – + – –

(+) – + – –

(+) (+) + – –

+ = survival as inflection; +/– = survival mixed, depending on variety; (+) = survival, but not as inflection; – = loss

Table 1 provides an overview of the data presented in sections 2 and 3, although it runs the risk of distortion due to space constraints. The symbol + indicates that the Portuguese inflection survived as such in the creole, presumably from its very genesis rather than being introduced during recent decreolization. The symbol +/– indicates that the Portuguese inflection survived (or possibly re-emerged through decreolization) in some varieties within a group but not in others (e.g. Upper Guinea ba, an inflection in Cape Verdean, but a free morpheme in Guiné-Bissau). The symbol (+) indicates that the Portuguese inflection has survived in the creole, but not as an inflection (e.g. having become a derivational morpheme such as P -ndo or -do in Guiné-Bissau CP). In the case of plural marking, +/– indicates that the plural inflection did not survive in the Indo-Portuguese of Korlai, but it did in Sri Lankan CP, although in the latter case the historical question of when such morphology emerged in the creole is so unclear that it would seem preferable to reserve judgment on these. In this context, Smith points out: I think you need to distinguish between plural marking and case inflection. The former must have existed in the very earliest SLP, since (a) it clearly

Creolization and the fate of inflections 319 comes directly from a Portuguese inflection, and (b) is not found as a free form, so once lost could not be reinstated. Case inflections, on the other hand, are clearly not modelled on Portuguese case inflections and they all derive from free forms which have been further grammaticalized through encliticization and reduction, so there can be arguments about exactly when the grammaticalization took place. (p.c.)

While a similar case could be made for the attrition of Korlai CP, one would have to be able to explain where the needed Portuguese inflections could have come from to serve as models for the Korlai inflections if the latter are a recent addition to the language. Some of the borderline cases presented above (e.g. the fate of the Portuguese past participle -do in the creoles) confirm Kihm’s assertion that “the distinction between inflection and derivation is clearly a matter of degree” (Kihm 2002: 334). In the end there are a few clear-cut cases in which Portuguese inflections survived as such in creoles: (1) the case of plural -s in the Upper Guinea creoles, (2) the case of -va in Cape Verdean, (3) the case of the verbal inflections in Indo-Portuguese. In the first case, it seems credible that this inflection emerged early on in the creole’s development given its regular occurrence in basilectal varieties and the fact that there is a partly parallel system of marking of plurality in Guiné-Bissau CP’s West Atlantic substrate/adstrate languages such as Balanta, which make a morphological distinction between the singular and plural forms of classmarking prefixes on nouns. In the second case, it cannot be ruled out that CV CP -ba was at one time a free morpheme like GB CP ba, but became an inflection under influence of Portuguese -va. In the third case, I agree with Clement’s analysis (1996: 118, 229): the Portuguese inflections survived as such due to the fact that Marathi is also an inflected Indo-European language with partly parallel suffixes. A knowledge of these surely facilitated its speakers’ interpretation of these Portuguese verbal inflections and the grammatical categories they represent. The reason for the widespread belief that creolization entails the loss of all inflectional morphology grew out of the general absence of inflectional morphemes as such in the group of creoles that has been most studied, the Atlantic creoles of the Caribbean region and coastal West Africa. All of these creoles have the Western European lexifiers that were the languages of empires (Portuguese, Spanish, French, Dutch and English) and NigerCongo or West African substrates, i.e. at least partially inflected superstrates and largely non-inflected, isolating substrate languages. And while the coming together of such languages under the conditions that produce

320 John Holm creoles was indeed unlikely to produce creoles that retain any inflections, the complete loss of inflectional morphology is not an inherent part of the process of creolization. All that was missing to disprove this assumption was a careful study of a creole that grew out of a superstrate and a substrate that are both inflected. The understanding that pidgins and creoles can have inflectional morphology, as first asserted by Stolz (1989) and reiterated by Bakker (2002), has important implications for defining the characteristics of these two kinds of contact languages. There has indeed been a paradigm shift in contact linguistics. Notes 1. I would like to thank Alan Baxter, Hugo Cardoso, Carlos Fontes, Tjerk Hagemeijer, Incanha Intumbo, Liliana Inverno, Gerardo Lorenzino, Philippe Maurer, Ian Smith, Dominika Swolkien and Isabel Tomás for their very helpful comments on an earlier draft of this paper. And I am especially grateful to J. Clancy Clements, whom I consulted in the course of writing this paper and who generously shared with me some of his yet unpublished work: Clements to appear (based on Clements 2003) and Clements forthcoming. To our surprise, we had reached similar conclusions based on analyses that we had developed independently dealing with some of the same data. I would like to emphasize the priority of his 2003 talk, and the central importance of the data and discussion in Clements 1996 in pointing the way to the conclusions reached here. However, responsibility for any shortcomings that may remain in this paper is mine alone. 2. AUX = auxiliary, DAT = dative, EMPH = emphatic, FEM = feminine, GEN = genitive, IMP = imperfect, IMPF = imperfective, NEG = negation, NOM = nominative, PAST = past tense, PF = perfective aspect particle, PFC = perfective, PL = plural, PP = past participle, PPL = participle, PROG = progressive, R = recipient relator.

Creolization and the fate of inflections 321

References Albring, Nikola and Eugénia Lourenço 2004 A hipótese de Dalgado de recíproca transfusão parcial: avaliando a sua utilidade pela comparação do crioulo português de Sri Lanka com o Papia Kristang. In Los criollos de base ibérica: ACBLPE 2003, Mauro Fernández et al. (eds.), 19–31. Madrid /Frankfurt a.M.: Iberoamericano /Vervuert. Bakker, Peter 2002 Pidgin inflectional morphology and its implications for Creole morphology. In Yearbook of Morphology, Geert Booij and Jaap van Marle (eds.), 3–33. Dordrecht: Kluwer. Baptista, Marlyse 2002 The syntax of Cape Verdean Creole: the Sotovento varieties. Amsterdam /Philadelphia: John Benjamins. 2003 Inflectional plural marking in pidgins and creoles: a comparative study. In Phonology and Morphology of Creole Languages, Ingo Plag (ed.), 315–332. (Linguistische Arbeiten 478.) Tübingen: Niemeyer. Batalha, Graciete Nogueira 1974 Língua de Macau: o que foi e o que é. Macau: Centro de Informção e Turismo. Baxter, Alan N. 1988 A grammar of Kristang (Malacca Creole Portuguese). Canberra: The Australian National University, Research School of Pacific Studies (Department of Linguistics). Bickerton, Derek 1999 How to acquire language without positive evidence: what acquisitionists can learn from creoles. In Language creation and language change: creolization, diachrony, and developmen, Michel DeGraff (ed.), 49–74. Cambridge, MA/London: MIT Press. Braun, Maria and Ingo Plag 2002 How transparent is creole morphology? A study of Early Sranan word formation. In Yearbook of Morphology, Geert Booij and Jaap van Marle (eds.), 81–104. Dordrecht: Kluwer. Clements, J. Clancy 1996 The genesis of a language: the formation and development of Korlai Portuguese. Amsterdam /Philadelphia: John Benjamins. 2003 Portuguese-based non-standard language varieties in Africa and Asia. Invited presentation. Annual meeting of the Berkeley Linguistic Society (29), Berkeley CA. to appear The Portuguese-based creoles in Africa and Asia. Berkeley Linguistic Society 29, University of California, Berkeley.

322 John Holm forthc.

Chapter 8: Portuguese- /Spanish-lexified creole languages. The linguistic legacy of Spanish and Portuguese: Colonial expansion and language change. Cambridge: Cambridge University Press. Comrie, Bernard 1981 Language universals and linguistic typology: syntax and morphology. Oxford: Blackwell. Dalgado, Sebastião Rodolpho 1900 Dialecto Indo-Português de Ceilão. Lisbon: Imprensa Nacional. [Reprint 1998, Lisbon: Comissão Nacional para as Comemorações dos Descobrimentos Portugueses.] DeGraff, Michel 2001 Morphology in creole genesis: linguistics and ideology In Ken Hale: a Life in language, Michael J. Kenstowicz (ed.), 53–121. Cambridge, MA: MIT Press. Günther, Wilfried 1973 Das portugiesische Kreolisch der Ilha do Príncipe. Marburg: Marburger Studien zur Afrika- und Asienkunde. Holm, John 1988–89 Pidgins and Creoles. 2 vols. Cambridge: Cambridge University Press. 1989 Do creoles have inflectional morphology? Invited paper presented at the Workshop on Creole Morphology, University of Amsterdam. Holm, John and Swolkien Dominika 2004 A expansão do crioulo cabo-verdiano para São Vicente: Factores sócio-históricos na difusão. Paper presented at the Colóquio Internacional: Cabo Verde – origens da sua sociedade e do seu crioulo. Friedrich-Alexander-Universität, Erlangen-Nürnberg (Germany), September 23–25. Intumbo, Incanha 2005 Guiné-Bissau Creole Portuguese, its superstrate and a substrate: a comparison of the structure of the noun phrase. Paper presented at the Creole Conference, Max Planck Institute, Leipzig, June 3–5. Ivens Ferraz, Luiz 1979 The Creole of São Tomé. Johannesburg: Witwatersrand University Press. 1987 Portuguese Creoles of West Africa and Asia. In Pidgin and Creole languages: essays in memory of John E. Reinecke, Glenn G. Gilbert (ed.), 337–360. Honolulu: University of Hawaii Press. Kihm, Alain 1994 Kriyol syntax: the Portuguese-based Creole language of GuineaBissau. Amsterdam /Philadelphia: John Benjamins. 2002 Inflectional categories in creole languages. In Phonology and morphology of Creole languages, Ingo Plag (ed.), 333–363. Tübingen: Niemeyer.

Creolization and the fate of inflections 323 Lefebvre, Claire 2002 The emergence of productive morphology in Creole languages: the case of Haitian Creole. In Yearbook of Morphology, Geert Booij and Jaap van Marle (eds.), 35–80. Dordrecht: Kluwer. Lorenzino, Gerardo A. 1998a The Angolar Creole Portuguese of São Tomé: its grammar and sociolinguistic history. Munich: Lincom Europa. 1998b The diachronic relevance of affixation in two Afro-Portuguese Creoles. Paper presented to the joint meeting of the Society for Pidgin and Creole Linguistics and the Society for Caribbean Linguistics, New York City. Maurer, Philippe 1995 L’angolar: un créole afro-portugais parlé à São Tomé; notes de grammaire, textes, vocabulaires. Hamburg: Buske McWhorter, John H. 2005 Defining Creole. Oxford, New York: Oxford University Press. Plag, Ingo (ed.) 2003 Phonology and morphology of Creole languages. Tübingen: Niemeyer. Post, Marike 1995 Fa D’Ambu. In Pidgins and creoles: an introduction, Jacques Arends, Pieter Muysken and Norval Smith (eds.), 191–204. Amsterdam: John Benjamins. Reinecke, John Ernest, S. M. Tsuzaki, David DeCamp, I. F. Hancock and R. E. Wood (eds.) 1975 A bibliography of pidgin and creole languages. Honolulu: University Press of Hawaii. Schuchardt, Hugo 1890 Kreolische Studien. IX. Über das Malaio-Portugiesische von Batavia und Tugu. Sitzungsberichte der kaiserlichen Akademie der Wissenschaften zu Wien 122: 1–256. Smith, Ian 1979a Substrata vs. universals in the formation of Sri Lanka Portuguese. Papers in pidgin and creole linguistics 2: 183–200. 1979b Convergence in South Asia: a creole example. Lingua 48: 193–222. 1984 The development of morphosyntax in Sri Lanka Portuguese. York Papers in Linguistics, 291–301. (=Papers from the York Creole Conference, September 24–27, 1983, M. Sebba and L. Todd [eds.]) Dept. of Linguistics, University of York. 2005 Les missionaires wesleyens et la conversion du créole portugais de Sri Lanka. Paper presented to the Associação: Crioulos de Base Lexical Portuguesa e Espanhola, Orléans.

324 John Holm Stolz, Thomas 1989 Kreolische Morphologie. Zeitschrift für Phonetik, Sprachwissenschaft und Kommunikationsforschung 42 (1): 48–55. Tomás, Maria Isabel 1992 Os crioulos portugueses do Oriente: uma bibliografia. Macau: Instituto Cultural de Macau. 2004 O Kristang de Malaca: processos linguísticos e contextos sociais na obsolescência das línguas. Ph.D. dissertation, Universidade Nova de Lisboa.

Borrowings from Romance languages in Oceanic languages Claire Moyse-Faurie

1. A short summary of European first contacts The recent history of Oceania – starting from the end of the 16th century – is full of events that had considerable impact on the linguistic situation. The discovery of the Pacific area and its inhabitants by Europeans took about three centuries. In the sixteenth and well into the seventeenth century, the European exploration of the South Pacific was almost entirely the domain of the Spanish and of the Portuguese: Alvaro de Mendaña, first in 1567–69, and then in 1595 with Pedro Fernandez de Queirós travelled through the South Seas and discovered the Solomon and the Santa Cruz archipelagos successively; Luis Vaéz de Torres (1605–06) along with Queirós reached Espiritu Santo in Vanuatu; Queirós went as far as Tahiti which he named Sagittaria. Above all, these navigators were misioneros whose main task was to convert inhabitants and teach them prayers in Latin and Spanish. This situation changed, however, in the early years of the seventeenth century when the Dutch ventured into the South Pacific in search of new markets and the Great Southland: Jacob Le Maire and Willem Schouten (1615–17) discovered Tuamotu and Futuna (Horn island). Abel Tasman (1642–43) reached the Land of Van Diemen (known nowadays as Tasmania), New Zealand, Tonga and Fiji. During the 18th century, the rival French and British powers began their great expeditions, trying to take control of still unexplored countries. On the French side: Louis Antoine de Bougainville reached Tahiti in 1768, then Samoa and New Britain. In 1792, Antoine Raymond d’Entrecasteaux made a stop in New Caledonia, 18 years after its discovery in 1774 by James Cook and the Forster. The French expedition conducted by Jean-François de Galaud de Lapérouse ended in 1788 in a wreck on the reefs of Vanikoro, one of the Eastern Solomon Islands. Forty years later, in 1826, an Irish captain, Peter Dillon, found wreckage of one of the Lapérouse expedition ships in a narrow pass in the reef. Dumont d’Urville, a French navigator also in search of Lapérouse, joined him in Vanikoro. On this island, it is said that an edible

326 Claire Moyse-Faurie plant is called cassoulet, which has no equivalent in neighbouring languages. In April 2005, a hundred researchers and members of the French National Marine headed to Vanikoro in order to pursue their investigations and to try to finally solve the “mystère Lapérouse”. Oral traditions have been recorded which could provide reliable information as to what happened to the sailors rescued from the French expedition, and could also help to find out if the cassoulet plant really is a souvenir of their stay in Vanikoro… Whalers, sandalwooders and blackbirders in the 18th century, followed by Protestant clergymen, are responsible for the introduction of many English words into all of the Pacific languages. Then the need for labour in the new Fijian, Samoan and Australian plantations (along with what was called the blackbirding), gave rise to English-based pidgins. Catholic missionaries are responsible for Latin borrowings. Later, in what became the French overseas territories, colonial political, administrative and educational control also had great influence on Oceanic languages. Finally, in 1856 the opening of a Marist Mission not far from Numea, where families from various areas of the mainland of New Caledonia were assembled to work on the Mission plantations, and their children attended French schools, along with repressive colonial constraints and population transfer following the Kanak insurrections of 1878, gave birth to a plantation French-based Creole called Tayo. The main part of this article concerns French and Latin borrowings in languages spoken in the French overseas territories (New Caledonia, French Polynesia and Wallis and Futuna), considering mainly lexical and phonological aspects, as well as a few morphological ones. I will also mention the Romance etymons found in Bislama (the English-based Pidgin of Vanuatu), the specifics of the French varieties spoken in the Pacific, and two extreme cases: the French Creole Tayo and the Kayafou language of the young people living in the Numea suburbs. Many other contact phenomena took place in Oceania; for example, contact between Austronesian and Papuan languages (see for instance Ross 2001), or Polynesian borrowings in Melanesian and Micronesian languages due to Polynesian trade and conquests, some of which ended in Polynesian settlements, giving rise to what are called Polynesian Outliers, which are Polynesian languages which endured considerable phonological and grammatical modification through Melanesian or Micronesian contact (see for example Ozanne-Rivierre 1994). Several articles have been published on Dutch loans in Polynesia (Geraghty and Tent 1997; 2004). Given the fact that in Oceanic languages, the dominant language as far as lexical borrowings are concerned was undoubtedly English; a result of the massive introduction of English ‘by Bible

Borrowings from Romance languages in Oceanic languages

327

and bottle’ (Hollyman 1962, quoting Carr 1951), English borrowings have been quite extensively studied (Tent and Geraghty 2004, for instance). German had considerable influence in the nineteenth and early twentieth centuries in many Oceanic countries (Samoa, New Guinea, Nauru, Belau, Marshall islands, etc.), but very little linguistic influence, except on the vocabulary of early Tok Pisin. Only six Samoan words borrowed from German were still in use in the 1980s (Lynch 2004: 198), whereas Samoan has borrowed hundreds of English words since the first missionaries came to Samoa in 1830. Very few studies have been carried out on Romance borrowings in the Oceanic languages, except for the Spanish and Portuguese influence, as has been the case for Chamorro (see Palomo 2008). According to Topping (1973: 6), the most notable influence on the Chamorro language and culture came from Spanish: “There was a wholesale borrowing of Spanish words and phrases in Chamorro, and there was even some borrowing from the Spanish sound system”. However, there was apparently no effect on Chamorro grammar. There are also several studies on the Portuguese influence, as in Central Maluku, Eastern Indonesia (Grimes 1994): “The Portuguese arrived in Maluku in 1512. For almost a century, the Portuguese were the only significant European presence in Central Maluku”. Then the Dutch arrived and stayed until after World War II, when the Spice Islands became part of the Republic of Indonesia. When the Portuguese arrived, Malay was already used as a lingua franca in a region of many languages, and they used Malay in both trade and the propagation of Christianity. The Dutch tried to teach Dutch in schools at first, but soon gave up and also turned to Malay. This situation gave birth to Ambonese Malay, a mixed language of Central Maluku which reflects the social history of this region. Only a few lexical borrowings from Portuguese are used nowadays in this Creole, such as pandela ‘flag’ < Port. ‘bandera’; losi buini ‘container used to store fruit of the Areca palm’ < Port. ‘dos’. Borrowings from Portuguese are found in other varieties of Malay such as Manado Malay (North Sulawesi): greja ‘church’, roda ‘cart’ < Port. ‘roda’; sapewo ‘hat < Port. ‘chapeu’; leto ‘handkerchief’ < Port. ‘lenço’; ki:ntali ‘yard, garden’ < Port. ‘quintal’, etc. Fischer (2001: 314) mentions a few French loanwords which are due to the arrival of four French missionaries in Rapanui (Easter Island) in 1866, for example: anio < ‘agneau’ (‘lamb’) and kevare < ‘cheval’ (‘horse’), which were later replaced by the Spanish equivalents. However, Latin borrowings have remained in the Catholic liturgy.

328 Claire Moyse-Faurie We are now going to focus on the French and Latin influence on the Polynesian and Melanesian languages spoken in the French overseas territories, starting with the phonemic integration of loanwords. 2.

Integration of loanwords

2.1. Conforming to the phonology The first borrowings, whatever their origin, have been totally integrated into the syllabic structure and the phonemic system of the Polynesian languages. Consequently, most of the loanwords are not recognized as such by Oceanic speakers. With the exception of the Outliers, Polynesian languages have a syllabic structure of the type (C)V(C)V, with no consonant clusters and no final consonant. Loanwords conform to the syllabic structure by inserting an epenthetic vowel into the cluster and by either adding a final vowel or eliminating the final consonant. For instance, East Uvean (Wallis island) allows no consonant clusters and no final consonant: tribunal > telepinale plastique > palasitike angle > ‘ gele brouette > pulueti

limonade > limonati service > selevisi accident > ‘akisit

programme > polokalama

téléphone > telefoni mazout > masuti Engl. stamp > sitapa Engl. captain > kapiteni

The final vowel added in the loanwords shows no gender assignment. Vowels [u], [e], [i] and [a] in loanwords are maintained but central rounded vowels [œ] and [Ø] as well as [] have been integrated as [e], while vowel [y] (‘u’ in the French spelling system) became either [u] or [i]: Fr. ‘carburateur’ > kapulatea, ‘député’ > tepit or tput. Nasal vowels turn to oral, usually long vowels: [
] > [o], [a
] > [a], [
] > [e] Fr. ‘camion’ > kami ‘truck’; ‘maman’ > mam . Non existing consonants are replaced by the nearest phonetic ones: [b] > [p] Fr. ‘basilique’ > pasilika, ‘brouette’ > puluete; Engl. ‘basket’ > p sikete. [d] > [t] Fr. ‘délégué’ > teleke; Engl.‘Monday’ > monit. [g] > [k] Fr. ‘tergal’ > telekale, ‘glaçon’ > kal s; Engl. ‘gum’ > kamu. [r] > [l] Fr. ‘bière’ > piele, ‘territoire’ > telituale; Latin ‘altare’ > aletale, ‘gratia’ > kalasia; Engl. ‘rice’ > l isi. [], [], [z] > [s] Fr. ‘mouchoir’ > musuala; ‘machine’ > masini; ‘Zoé’ > Soe; ‘gendarme’ > satalamu > salatamu; ‘général’ > senelale. [t], [d] > [t] Engl. ‘jam’ > tiamu, ‘match’ > mati.

Borrowings from Romance languages in Oceanic languages

329

Tahitian has eleven consonant phonemes: oral and nasal stops /p/, /t/, //, /m/, /n/, sibilants /f/, /v/, /h/, the /r/ and two semi-vowels /w/ and /j/. It allows no consonant clusters and no final consonant. Loanwords with /d/, /k/, /g/, /s/, /z/, //, // have been integrated in Tahitian with a /t/; /l/ became /r/ and /b/ became /p/. French: ‘chocolat’ > ttr ; ‘kilo’ > tiro; ‘bébé’ > ppe; ‘pacifique’ > p tifita, later on by metathesis > p titif , etc. English: ‘president’ > peretiteni; ‘governor’ > tavana; ‘locker’ > rota; ‘coffee’ > taofe; ‘club’ > tarapu; ‘book’ > puta, etc. Latin: ‘diluvium’ > t ruvi [ti:rui] ‘deluge’; ‘baptismo’ > bapetizo [papetito] ‘baptism’, etc. In the last few decades, the Tahitian Academy has tried to put an end to the borrowing process and has created many neologisms. In Mangarevan (Rensch 1994), there have been some French and Latin borrowings, the first of which occurred under the influence of three missionaries from the Congrégation du Sacré-Coeur, who arrived in the Gambier islands in 1834. They converted the entire population to Catholicism within three years. French and Latin borrowings were transliterated in the same way as the earlier borrowings from English had been, with voiced stop > voiceless stop; sibilants > voiceless dental stop: b > p; s, d > t; g > k; l > r. Latin: ‘dominica’ > tomenika ‘Sunday’; ‘discipulus’ > titipuro ‘disciple’; ‘ecclesia’ > ekeretia ‘church’; ‘aeternitas’ > eterenite ‘eternity’; ‘catechismus’ > katekimo ‘catechism’, etc. In earlier French and Latin borrowings f > p (but in more recent Tahitian borrowings f > v): Latin: ‘fica’ > pika ‘fig’;‘fenestra’ > penetera ‘window’; ‘farina’ > parina ‘bread’. French: ‘café’ > kape ‘coffee’. According to Rensch, “Although there was no competition from the Protestants, Mangareva being 100% Catholic”, the French missionaries replaced earlier English borrowings with French or Latin transliterations or with a Mangarevan paraphrase: English

Early Engl. borrowing

French /Latin substitutes

watch hammer rice vinegar spoon stockings

uati hamara raiti vineka punu totini

motara marato ri vinekere kuiere pui vavae

(< Fr. ‘montre’) (< Fr. ‘marteau’) (< Fr. ‘riz’) (< Fr. ‘vinaigre’) (< Fr. ‘cuiller’) (Mangarevan compound)

We will see other instances of French-English competition in borrowings below.

330 Claire Moyse-Faurie According to Hollyman (1962), Polynesia had no tradition of bilingualism, and the Polynesian languages have not borrowed a single phoneme from any European language1, although they have fewer than half as many (around ten) consonant phonemes as English. Even if in its spelling Tahitian retains letters that represent European phonemes in borrowed words, the pronunciation is different: written Tahitian bapetizo ‘baptism’ is pronounced [papetito]. On the other hand, bilingualism, even multilingualism, is a traditional characteristic of Melanesia that has been accentuated by European contact. In fact, New Caledonia is well-known for what Haudricourt called ‘egalitarian bilingualism’, between groups with intermarriage relations and no vehicular language. In spite of this particular situation, languages of the mainland of New Caledonia, just as Polynesian languages, most often have integrated loanwords into their phonemic system. For instance, in Nyêlayu (Ozanne-Rivierre 1998), a language spoken in the northern part of the New Caledonian mainland, French consonants /f/, // and /s/ were integrated as aspirated stops: f > ph; , s > ch: ‘fête’ > pher ‘feast’; ‘café’ > kaphe ‘coffee’; ‘vache’ (‘bétail’) > vaachi ‘cow, cattle’; ‘sac’ > chak ‘bag’; ‘sucre’ > chuk ‘sugar’; ‘savon’ > chaavô ‘soap’. As in all mainland Kanak languages, Nyêlayu only has prenasalized voiced stops, and so the following French borrowings are realised as: ‘bonjour’ > bocuur [mbocu:r] ‘hello’; ‘gâteau’ > gato [ gato] ‘cake’. Only voiceless oral and nasal stops are found in the final position: Fr. ‘table’ > taap; Fr. ‘école, collège’ > koleec ‘(high) school’; Fr. ‘graisse’ > gereec ‘grease’; Fr. ‘orange’ > orââc. Finally, since there is no ‘r’ in the initial position, it is preceded in loanwords by an epenthesis: Fr. ‘riz’ > harii ‘rice’; ‘Robert’ > Orober. In Paicî (Rivierre 1983), loanwords with [f] or [v], whatever their origin, are integrated as [p]: Fr. ‘verre’ > per ‘glass’; Fr. ‘café’ > kape ‘coffee’; Engl. ‘flour’ > pwöloa/pölöwa. Since Paicî is a tonal language, loanwords also take tones, and ‘coffee’ is in fact pronounced with a mid tone followed by a high tone: k fé. Rivierre (1994) describes how loanwords in Cèmuhî (another tonal Kanak language) in many cases, “are assigned low tone which is the least frequently used (and therefore the most ‘marked’) of the three tone levels this language possesses”, except when the borrowed word has a word-initial fricative or aspirate; it is then integrated into Cèmuhî with a high tone: Fr. ‘ciseaux’ [sizo] > Cèmuhî cíco ‘scissors’. The Kanak languages spoken in the Loyalty islands have no nasal vowels; loanwords comprising such vowels are integrated with the corresponding oral ones: Nengone sima < Fr. ‘ciment’ [sima
]; avio < Fr. ‘avion’

Borrowings from Romance languages in Oceanic languages

331

[avj
]; magaza < Fr. ‘magasin’ [magaz
], etc. In Drehu, French [y] is integrated as [i]: ‘bureau’ > biro ‘office’; ‘culotte’ > kilot ‘pants’. Most often, the integration of loanwords, especially earlier cases, conforms to the sound system of the Kanak languages that is borrowing them, just as is the case for the Polynesian languages. However, some cases of rephonologisation have taken place. 2.2. Rephonologisation Ajië, a language spoken in the south of the mainland of New Caledonia, is said to have borrowed /f/, /s/ and /l/, but there are only four words beginning with an ‘f’ and none with an ‘s’ in the recently published Claude Lercari’s dictionary (Lercari 2002). In fact, French consonants /s/, /z/, //, // were integrated first as [ç] then as [j] (written ‘y’): ‘seau’ > yio ‘bucket’; ‘cheval’ > yovari ‘horse’; ‘graisse’ > gereyi ‘grease’; ‘ingénieur’ > âyényëë ‘engineer’; ‘acier’ > layié ‘steel’; ‘soie’ > layöa ‘silk’; ‘Céleste’ > Yèlèèyi. /f/ was integrated as /v/: Félix > Vèliiyi; Ferdinand > Vèrèdina; Florence > Völörââyi; Zoé > Yoé; Georges > Yööyi; Charles > Yaalé. However, there are a few loanwords with an optional /s/ instead of /j/, as in the words used by the Protestants for ‘Easter’ Paséka/Payéka and ‘cross’ satauro/yatauro. Phonemic borrowing has been recorded in areas traditionally open to contact influence, as is the case in Drehu (Loyalty Islands). Before European contact, Drehu had twenty-nine phonemes and has borrowed at least three phonemes through the introduction of loanwords: – /b/ in beletr < Engl. ‘belt’; boliko < Fr. ‘bourricot’ (‘donkey’); buruet(e) < Fr. ‘brouette’ (‘wheelbarrow’); butei < Fr. ‘bouteille’ (‘bottle’), etc. – /v/ in valis < Fr. ‘valise’ (‘suitcase’); vizit(e) < Fr. ‘visite’, velo < Fr. ‘vélo’ (‘bicycle’), varad(e) < Fr. ‘véranda’, etc. – /r/ in rais (variants lais, laes) < Engl. ‘rice’; rouz(i) < Engl. ‘rose’, etc. Other cases of rephonologising via borrowing are found in some Kanak languages. For instance, languages from the far north of the mainland lost the distinction for both oral and nasal consonants between aspirates and nonaspirates in intervocalic position. However, according to Rivierre (1994), “the phonemes which had been barred from certain positions reappear by way of loanwords or lexical creations”. Among French borrowings, we find Caac kaphar < Fr. ‘cafard’ (‘cockroach’), kapheye < Fr. ‘caféier’ (‘coffeeplantation’), whereas /-ph-/ is never found in intervocalic position in the inherited lexical stock.

332 Claire Moyse-Faurie In Bwatoo (Rivierre and Ehrhart 2006), /s/ is a new phoneme, only attested in loanwords such as salad < Fr. ‘salade’, semiis < Fr. ‘chemise’ (‘shirt’), sokola < Fr. ‘chocolat’ or bosu < Fr. ‘bonjour’. Also, recent French loanwords are borrowed with an initial consonant cluster, a feature normally only possible between two morphemes: Fr. ‘claquettes’ > klaket ‘tongs’. In Nengone (Loyalty islands), /f/ and /v/ are only attested in a few English, French or Latin loanwords: Fr. ‘avion’ > avio ‘plane’; Engl. ‘flour’ > falawa. There are indeed more cases of phoneme borrowing found in Kanak languages than in Polynesian languages, and it looks like the more phonemes you have, the more phonemes you borrow. However, a few cases of phonemic changes are also found in Polynesian languages. In East Uvean, the sequence /t+i/ [ti] is not found in the inherited lexical stock and [s] regularly appears as an allomorph for /t/ before [i]: sio ‘see’, sisi ‘belt’, ga’asi ‘shell’, words showing regular correspondences with East Futunan tio, titi and ga’ati. /s/ as a phoneme was only found in a few words, probably ancient borrowings from neighbouring languages (such as Fijian or East Futunan), and most of the /s/ occurrences are now found in loanwords from Latin, French or English: Latin: sagato < ‘sanctus’; sel < ‘caelum’; s kilifisio < ‘sacrificium’; kolesio < ‘collegium’. English: hsi < ‘horse’; pusi < ‘pussy(cat)’; snipa < ‘chamber’; suka < ‘sugar’; uasi < ‘watch’; kalasini < ‘kerosene’. French: televisio < ‘télévision’; s pi < ‘champion’; sen or senat < ‘sénateur’. What is striking is that in the more recent borrowings, the sequence ‘t+i’ is reintroduced as [ti] such as in t ‘tea’, limonati ‘limonade’, Soane Patita ‘Jean-Baptiste’, fakalatina ‘Latin’, heletiko ‘heretic’ and t polo ‘lemon’ (< probably English tipple). There can be no doubt that these words were borrowed after the change of pronunciation from [t] to [s] in front of [i], a change which must have occurred at the time of the first European contact. 2.3.

Morphological integration

2.3.1. Article integration In Oceanic languages that are or have been in contact with French, some words have been borrowed with the feminine (la) or the elided article (l’) as part of the base. Such cases of article integration are found in Kanak languages (examples below in Bwatoo, Ajië and Nengone) as well as in Polynesian languages:

Borrowings from Romance languages in Oceanic languages

333

East Uvean: lafeti ‘fête’ (‘feast’), l keleve ‘grève’ (‘strike’), l tuani ‘douane’ (‘customs’), laleke ‘règle (du maçon)’ (‘measuring rod’), lal ‘arrêt’ (‘bus stop’), lpitali ‘hôpital’; l palasi ‘place’; lsini ‘usine’ (‘factory’). Ajië: lapèè ‘pelle’ (‘shovel’), lâpo ‘impôt’ (‘income taxe’), lapööyi ‘poste’; laremé ‘armée’, lalékooli ‘alcool’, lagalaayi ‘glace’ (‘ice cream’)’, laloa ‘loi’ (‘law’), lapoali ‘poêle’ (‘frying pan’). (Lercari 2002) Nengone: lagrip ‘grippe’ (‘flue’), lalen ‘laine’ (‘wool’). Bwatoo: loto ‘auto’, lasos ‘sauce’, lamin ‘mine’. (Rivierre and Ehrhart 2006) The masculine article ‘le’ [l ] was almost never integrated into the base. This could have been due to the fact that it was analyzed as the article – note that there is no gender differentiation in Oceanic languages, or to the fact that the unattested sound [ ] prevented it from being borrowed. A few cases of integration of the French definite partitive article ‘du’ [dy] are also found: Bwatoo: dile ‘lait’ (‘milk’), divâ ‘vin’ (‘wine’); East Uvean t pea ‘beurre’ (‘butter’); Ajië dibëë ‘beurre’, divââ ‘vin’. These morphological integrations are well-known in French-based Creoles, but it is worth noticing that they are not specific to creolisation; they are better analysed as cases of long distance convergence, probably due to the fact that the unstressed French article is perceived as being part of the base. Note that in the West Uvean Polynesian Outlier spoken in the island of Uvea (Loyalty islands), by contrast, the integration of the article did not happen, as shown in loanwords such as: are ‘arrêt’ (‘bus stop’) or oras ‘orange’. Neither did it integrate into Drehu, as this language has a set of preposed deictic articles (lo, la, lai) that led to a short reanalysis: pital < Fr. ‘hôpital’.

2.3.2. Derivation and compounding from loanwords In general, the syntax of the Oceanic languages has not been influenced by contact. According to Hollyman (1962: 319), what is noticeable is the influence of Polynesian or Melanesian syntactical patterns on the French spoken by Polynesians or Melanesians: the influence does not seem to be working the other way.

334 Claire Moyse-Faurie Indeed, borrowed verbs are integrated into the Kanak languages affixal system: Nyêlayu chuuk [chu:k] < Engl. ‘sugar’ is transitivized in chuxee [chuxe:] ‘to sugar’; Drehu vizit < Engl. ‘visit’ is transitivized in vizitën ‘to visit’. Fagauvea (West Uvean) attaches -ina, its most productive transitive suffix, to the borrowed verb, whether it comes from English: saatr ‘salt’ > saatr-ina ‘to salt’, salem-ina ‘to sell’; from Iaai (the neighbouring Melanesian language): huliwa ‘work’ > huliwa-ina ‘work at sth., repare’; or from French: repase-ina ‘iron’, anose-ina ‘announce’, etc. In Drehu, the causative circumfix is also productive in loanwords: asatauro-n ‘to crucify’ (satauro < Latin ‘cross’); as is the collective prefix i-: i-buutr ‘shoes’ (< Engl. ‘boot’). In Nengone and Drehu, classifiers can be attached to loanwords when they fit into the semantic category: the fruit and small rounded object classifiers – Nengone wa- (wanu ‘coconut’), Drehu wene- (weneön ‘breadfruit’) – form new compounds with loanwords either from English or from French: Nengone wa-laim < Engl. ‘lemon’, wa-piese < Fr. ‘pièce’ (‘coin’); Drehu wene-kas < Fr. ‘cachet’ (‘pill’). Nengone cede ‘recipient’ also forms compounds with loanwords: cede-kaf ‘bowl of coffee’ (kaf < Fr. ‘café’). The male/female adjuncts apply to loanwords referring to animals: Engl. ‘goat’ > goutre hmenew, lit. goat female. Most of the Polynesian languages make a distinction between two types of possession, using two possessive markers: a for the alienable/agentive possession, and o for the inalienable/patientive possession. East Uvean loanwords referring to clothes, personal objects or objects carried on oneself were first integrated as o-possessed nouns: malia ‘religious medal’; slie ‘shoes’ < Fr. ‘soulier’; musuala ‘handkerchief’ < Fr. ‘mouchoir’; tauveli < Engl. ‘towel’; topa < Engl. ‘soap’; but more recent borrowings such as uati < Engl. ‘watch’ are a-possessed. Very few cases of grammatical borrowings are found in Kanak and Polynesian languages. One of the few examples encountered is the use of the French alternative coordinator ‘ou’ [u] (‘or’) in Xârâcùù (South Mainland of New Caledonia) instead of the inherited form tööi. Let us turn now to the semantic domains of borrowings.

3.

Semantic areas of loanwords

3.1. East Uvean and East Futunan borrowings English whalers, traders and sailors often introduced the first English borrowings via a pidginized form. English borrowings were reactivated by the

Borrowings from Romance languages in Oceanic languages

335

presence of American ‘marines’ during World War II in Wallis and in New Caledonia. Most of the English loanwords are technical terms or products: – food, drinks: ‘aisi < ‘ice’, keke < ‘cake’, l isi < ‘rice’, lole < ‘lolly(pop)’, onioni < ‘onion’, p nati < ‘peanut’, pipi < ‘beef’, putete < ‘potato’, siamu < ‘jam’, suka < ‘sugar’, t < ‘tea’, uisik < ‘whisky’, etc. – tools, cuttlery: h male < ‘hammer’, kapa < ‘copper’, loka < ‘locker’, meli < ‘mail’, mtok < ‘motorcar’, motopai < ‘motorbike’, nusipepa < ‘newspaper’, paipa < ‘pipe’, pepa < ‘paper’, p kete < ‘bucket’, p sikete < ‘basket’, peni < ‘pen’, pila < ‘pillow’, puna < ‘spoon’, silika < ‘silk’, siliva < ‘silver’, tini < ‘tin’, uasi < ‘watch’, sitima < ‘steamer’, sitapa < ‘stamp’, tauveli < ‘towel’, etc. – sports and games: fitipolo < ‘football’, kilikiti < ‘cricket’, pate < ‘bat (cricket)’, kolo < ‘goal’, kolonea < ‘corner (foot)’; hata < ‘heart (card)’; kalapu < ‘club’, taimani or taimane < ‘diamond’, peti < ‘spade’, etc. – jobs: kvan < ‘governor’, kuka < ‘cook’, ‘fis < ‘officer’, slia < ‘soldier’, tket < ‘doctor’, etc. – concepts: f mili < ‘family’, fiva < ‘fever’, h fekasi < ‘half-cast’, kuata < ‘quarter’, lao < ‘law’, meta < ‘meter’, mnite < ‘Monday’, pilisoni < ‘prison’, etc. – animals: hsi < ‘horse’, p < ‘bee’, pusi < ‘(pussy) cat’, tia > ‘deer’, etc. With the arrival of French missionaries in 1837, East Uvean and East Futunan borrowed words from Latin. The loanwords were adopted in the singular ablative form, ending with a vowel, conforming to the Polynesian syllabic structure. For example, vite ‘vineyard’ is the ablative form of vitis just as vino ‘wine’ is the ablative form of vinum. Besides religious vocabulary, loanwords from Latin belong to the semantic domains of temporal notions (hour, months), some products (key, bread, wine), a few animals quoted in the Bible (lamb, veal, etc.) and all Christian first names: – religious vocabulary: ‘aletale < ‘altare’, ‘ selo < ‘angelo’, ‘kelesia < ‘ecclesia’, ‘etelen < ‘aeterno’, kalasia < ‘gratia’, k tiko < ‘canticum’, kilisitiano < ‘christiano’, kollia < ‘gloria’, koluse < ‘cruce’, misa < ‘missa’, mseniolo < ‘monsignoro’, p tele < ‘patre’, sagata < ‘sancta’, sagato < ‘sancto’, ssuni < ‘jejunio’, etc. – temporal notions: felia < ‘feria’, hola < ‘hora’, skul < ‘saeculo’, temi < ‘tempi’; and all the month names: s nualio < ‘januario’, fepualio < ‘feb-

336 Claire Moyse-Faurie ruario’, malesio < ‘martio’, ‘apelili < ‘aprili’, maio < ‘majo’, snio < ‘junio’, slio < ‘julio’, ‘aukusit < ‘augusto’, sepetepeli < ‘septembri’, ‘okotopeli < ‘octobri’, novepeli < ‘novembri’, tesepeli < ‘decembri’. – products, food: ‘aulo < ‘auro’, falena < ‘farina’, kalav < ‘clavis’, n tul

< ‘natura’, ‘loto < ‘horto’, pane < ‘pane’, etc. – christian first names: ‘Alefeleto, Malia, Kilisitina, Soane, Paulo, Ptel, Malekalita, Falakika, Mikaele, Ssefo, etc. – animals (in the Bible): ‘akeno < ‘agno’, ‘ kuila < ‘aquila’, sino < ‘asino’, kamelo < ‘camelo’, v tulo < ‘vitulo’, etc. Wallis and Futuna became a French territory in 1961 and from then on, borrowings from French became more and more numerous. Attending school became compulsory, teaching was done only in French; radio and television were almost entirely in French. Loanwords from French mainly belong to the political, administrative, educational and technological domains, and their number is constantly increasing: ksei ‘conseiller’ (‘counsellor’); pilo ‘bureau’ (‘office’); politike ‘politique’; selevisi ‘service’; telituale ‘territoire’; vote ‘voter’. A few French borrowings are older, such as l sio ‘ration (militaire)’, which now means ‘chance’! 3.2. New Caledonia In New Caledonia, the various missions have had a major influence on the origin of borrowings: the London Missionary Society arrived in 1841 on the Loyalty islands; the Société de Marie settled in Balade in 1843, ten years before French took control of the territory. In Nyêlayu (Ozanne-Rivierre 1998) spoken in the Balade area, borrowings from French are older and more numerous than in the Protestant parts. In a few cases however, borrowings seem to follow a cultural logic. For example, ‘tea’ is [ti:] everywhere in New Caledonia; but ‘coffee’ is borrowed either from French or from English. The word ‘wine’ was first introduced from English, but nowadays the French word is taking over – l’honneur est sauf –, as in Bwatoo waen > divâ. Borrowings from French found in most of the Kanak languages are listed below: – ‘balai’ (‘broom’), ‘voiture’ (‘car’), ‘avion’ (‘plane’), ‘table’, ‘chemise’ (‘shirt’), ‘lait’ (‘milk’), ‘magasin’ (‘strore’), ‘gendarme’ (‘polieceman’), ‘kermesse’ (‘village fair’), ‘fête’ (‘feast’), ‘bouquet’ (‘bunch of flowers’),

Borrowings from Romance languages in Oceanic languages

– – – –

337

‘citron’ (‘lemon’), ‘chocolat’, ‘salade’, ‘piment’, ‘chaise’, ‘verre’ (‘glass’), ‘savon’ (‘sopa’), ‘sauvage (adj.)’, ‘haricot’ (‘bean’), ‘bourricot’ (‘donkey’), ‘brouette’ (‘wheelbarrow’), ‘se changer/déshabiller’ (‘change one’s clothes’), ‘ciseaux’ (‘scissors’) and of course the frog ‘grenouille’ (Nêlêmwa goronu; Ajië, Bwatoo, Cèmuhî gerenu, etc.) forms of address ‘papa’, ‘maman’, as in Nemi papa ne-ek ‘his father’; Bwatoo mama ne Pol ‘Pol’s mother’. ‘bonjour’: Cèmuhî bwöcu, Bwatoo and Fagauvea bosu, Nêlêmwa bôshu, Ajië böyu, Xârâgurè bosoo, etc. ‘merci’: Xârâcùù mêêsi, Bwatoo maasi, Fagauvea mési. a word for ‘civil marriage’ has been introduced in addition to the inherited one for ‘traditionnal wedding’: Xârâgurè mariasé/xöyö, Bwatoo marie/xhatoon.

And from Latin peccato ‘sin, fault’: Xârâgurè bèkaato, Fagauvea pekato (homophonous with peka to which means ‘bat’s fall’ and happens to be the common Polynesian metaphor for the sex of the woman…) In the Loyalty islands (discovered in 1793 by a British captain), English loanwords for a long time remained much more frequent than French ones. English Protestant missionaries arrived in the 1850’s. The French Roman Catholic missionaries were less successful and according to Darrell Tryon (Tryon 1970), no French borrowings in Drehu (Lifu island) have been found for certain for the pre-1920 period. After the Second World War, French became the language of education and many French borrowings have since been introduced, some of which have replaced the English ones: loto ‘voiture’ (‘car’), avio ‘avion’ (‘plane’), buruet(e) ‘brouette’ (‘wheelbarrow’), bozu ‘bonjour’, farin ‘farine’, gato ‘gâteau’ (‘cake’), fara ‘franc’ (but also faraig < Engl.), saaze ‘se changer’, amad(e) ‘amende’ (‘fine, penalty’), ariko ‘haricot’ (‘bean’), pima ‘piment’, boliko < ‘bourricot’ (‘donkey’), and without any doubt, a new concept: zalu ‘jaloux’ (‘jealous’)!! 4. French versus English loanwords The distribution of French and English loanwords would provide helpful information on the history of Kanak villages. Unfortunately, no systematic research has yet been undertaken in this domain, and I can only give a small idea of the complexity of the data, with a comparison of a few loanwords in four Kanak languages. Loanwords from French are in bold characters, loanwords from English are in italics:

338 Claire Moyse-Faurie French

English

Xârâcùù

Ajië

Nêlêmwa

Fagauvea

fourchette clé beurre cuiller balai savon

fork key butter spoon broom soap

fùrùsète kii dibèè kwiriörö barèè söpö

wuuryèet kii dibëë kùyèè baléé yööpö

fook kile

café cheval

coffee horse

köfi oosi

kövi yevari

kafe hooc

fok kip baca sipun burum sop (but also savon Marseille!) kafe hoos

shivun bulum shaawô

If we compare loanwords in Nêlêmwa and in Ajië, we see that in Nêlêmwa, ‘fork’, ‘spoon’ and ‘broom’ come from English, whereas ‘soap’, ‘coffee’ and ‘key’ are from French. The exact opposite is found in Ajië, where ‘key’, ‘coffee’ and ‘soap’ are English borrowings, and ‘fork’, ‘spoon’ and ‘broom’ come from French! ‘Horse’ and ‘bread’ also have different origins depending on the Kanak language: Fr. ‘cheval’ in Nyêlayu chovan, Fwâi sovan, Ajië yevari; but Engl. ‘horse’ is found in Xârâcùù oosi, Nêlêmwa hooc, Fagauvea and Drehu hoos. English ‘flour’ has been borrowed in Cèmuhî pwaloa, Fagauvea and Drehu falawa, now meaning ‘bread’; whereas French ‘pain’ is the origin of Xârâgurè pêê. A few languages preferred neologisms to borrowings. For example, Cèmuhî borrowed ‘key’ kilé and ‘broom’ balé from French, but balé is in competition with bé-taulimwa (lit. instrument to clean the house), ‘fork’ is bé-wölinaado (instrument for pricking) and ‘pencil’ bé-tii (instrument for carving), bé- being the instrumental prefix used to form various custom presents and names for tools. Even in one language doublets coexist, resulting from successive borrowings, as is the case in East Uvean: Engl. borrowings

English

Fr. borrowings

French

nusipepa siamu

newspaper jam

sulunale kofitula

journal confiture

‘Butter’ is said potilo < Lat. ‘butyrum’ or t pea < Fr. ‘du beurre’; ttini < Engl. ‘stocking’ coexists with the neologism tagava’e (lit. bag foot). In East Futunan, some products have different names depending on which of the two kingdoms one is in, as for instance pomo < Fr. ‘pomme’ in the Sigave kingdom and apo < Engl. ‘apple’ in the Alo kingdom. Such doublets are also

Borrowings from Romance languages in Oceanic languages

339

found in Kanak languages: Bwatoo chopa < Fr. ‘cheval’ and hoosi < Engl. ‘horse’; balo < Fr. ‘ballon’ and bool < Engl. ‘ball’ (however, bool tends to specialize for the cricket ball). In Nêlêmwa (Bril 2000), depending of the dialect: le/melek ‘lait/milk’; pâ/fwalaa ‘pain/flour’, and in the names for months: faraide/vâdredi ‘Friday/vendredi’; wenesde/merkredi ‘Wednesday/ mercredi’; shadede/samdi ‘Saturday/samedi’, etc. 5. Bislama (English based Pidgin spoken in Vanuatu) In this former French-English condominium (independent since 1980), French today faces hard competition from English and Bislama, the Englishbased Pidgin born in the plantations. A few Portuguese etymons are found in Bislama such as pikinini ‘child’ < pequenho, pequenha ‘small’ and kalabus < calabuso [kalábus] ‘jail’. The word bislama itself comes from the Portugese bicho do mar (lit. worm of the sea) ‘holothurian, sea-cucumber’, an animal well-known for its aphrodisiac value among the Chinese. According to Charpentier (1979: 42), the pidginized form of the Portuguese term has been relexified in New Caledonia by French into bichelamar, written nowadays bislama, and designates both the sea-cucumber and the Vanuatu vernacular language Bislama. Different hypotheses have been given for the origin of the Bislama word save. Commonly pronounced sae it means ‘to know, understand’ (mi no save ‘I don’t know’) as well as ‘knowledge, expertise, awareness’ and is also used for ability or permission: mi no save kam tumora ‘I can’t come tomorrow’, mi save karem foto blong yu? ‘May I take your photograph?’, hem i save toktok Franis ‘He speaks French’. According to Charpentier, Bislama save does not come from French savez (2nd person plural form of ‘savoir’) or from English ‘to savvy’, but it probably comes from the Portuguese saber. The word save, found in many other creoles and pidgins, must have belonged to Portuguese sabirs at the time of conquests, influenced by the Spanish spoken in the harbours, hence save. French etymons found in Bislama only represent 2 to 6% of the lexical stock, compared to the 90% of English etymons (plus 4% of Melanesian etymons and a few from various origins, such as Portuguese or Polynesian). Charpentier (1979: 331–336) distinguishes two large semantic domains for the French etymons: (Cf. Crowley 1990 for more complete French and English etymons):

340 Claire Moyse-Faurie – etymons in connection with plantations and trading: arier < ‘en arrière, reculer’ (‘move back’), bebet < ‘bébête’ (‘insect’), legim < ‘légume’ (‘vegetables’), obesin < ‘aubergine’ (‘egg-plant’), rato < ‘râteau’ (‘rake’), busi < ‘bougie’ (‘candle’), bramin < ‘barre à mine’ (digging tool), ariko < ‘haricot’ (‘bean’), pistas < ‘pistache’ (‘peanut’), masut < ‘mazout’ (‘diesel’), kale < ‘caler’ (‘chock, wedge, support’); kamiong [kamio ] < ‘camion’ (‘truck’) offers a good example of unpacking: nasal vowel > oral vowel + nasal consonant. – etymons describing activities or concepts specific to the French community, including cooking: lekol < ‘école’ (‘school’), sondam < ‘gendarme’ (‘policeman’), delegei < ‘délégué’ (‘delegate’), lames < ‘messe’, lafet ‘fête’ (‘feast’); gato < ‘gâteau’ (‘cake’), lae < ‘aïl’, lasup < ‘soupe’, neskafe < ‘café soluble’, pimang < ‘piment’, avoka < ‘avocat’, salad < ‘salade’, seris < ‘cerise’ (‘local cherry’), sitrong < ‘citron’ (‘lime’), pomfrit < ‘pommes frites’ (‘chips’), ragu < ragoût’ (‘stew’), kwasong < ‘croissant’, siro < ‘sirop’ (‘sweet drink mixture’). Kao ‘flat, burst’ comes from French pronunciation of English k.o. – exclamations or set expressions: ale < ‘allez’ (‘and then’) expression for continuing the interaction; saye < ‘ça y est’ (‘that’s it!’); si as an answer to an interrogative-negative question: yu no kam? ‘You don’t come? Si!’. In Bislama, some etymons are potentially derivable from either English or French (Crowley 1990: 109): frang ‘franc’, kilo, bang ‘bank/banque’, sigaret, koniak, plastik, adres, futbol. There are also some French/English doublets: glas/mira ‘mirror/miroir’; duan/kastom ‘douane/customs’; kokot/suspen ‘cocotte/sauce pan’; pomdeter/poteto ‘pomme de terre/potatoe’; sobe/aeskrim ‘sorbet/ice cream’; savat/sus ‘savatte/shoes’; sinwa/ saenis ‘chinois/chinese’; klaket/slipa ‘claquettes/tongs’. Also in Bislama, as we have seen in Oceanic languages, words of French origin may appear with the definite article reanalysed as part of the root: Fr. ‘école’ > lekol ‘school’, ‘messe’ > lames ‘mess’, ‘colle’ > lakol ‘glue’, ‘ampoule’ > lapul ‘light bulb’, etc.

Borrowings from Romance languages in Oceanic languages

6.

341

Caledonian French: ‘lôngage à nouzautes!’ 2

6.1. Specificities French is spoken to different degrees by almost all of the inhabitants in the three French overseas territories. It is the only official language, although all the Oceanic languages spoken in these territories have recently been recognized as ‘langues de France’. In New Caledonia (the Latin name of Scotland!), French is also the vehicular language between the various ethnic groups, and between the Kanaks themselves when they speak different Kanak languages. French is the first language of the French descendants of the settlers and convicts (the ‘Caldoches’) in New Caledonia, and of all the metropolitan French people. In addition, French tends to become the primary language of children from mixed marriages or children living in suburban areas. In French Polynesia, Tahitian has been taught in primary schools since 1982 and is the vehicular language, along with French (the first language of the Popa’a, originating from France), between the various archipelagos. In Wallis and in Futuna, more than 90% of the population speaks East Futunan or East Uvean as a first language, and there is a non-balanced bilingual situation. People of European origin are called Pap lagi. The pronunciation of the French spoken in these territories has been influenced by the vernacular languages as in, for instance, the prenasalisation of voiced stops in New Caledonia: bateau ‘ship’ is pronounced [mbato], the non-differentiation between French nasal vowels /
/ et /a
/: [l
ga] instead of [la
ga] for ‘langage’, [
t
sj
] instead of [ata
sj
] for ‘attention’; or in Polynesia, the trill of the ‘r’. Several lexicographic studies have been done on Caledonian French: see for instance Hollyman’s Observatoire du français dans le Pacifique (1983– 1992), a PhD (Pauleau 1995) and a dictionary Mille et un mots calédoniens (1983). The lexicon shows various specificities. It comprises words introduced by the French colonisers during the 18th century (margouillat ‘gecko’, aiguillette ‘garfish’) some of which are now obsolete in Metropolitan French, such as pochon ‘plastic or paper bag’. There are loanwords from the other French colonies: West Indies (barbadine ‘passionfruit’, bécune ‘barracuda’), Indian Ocean (badamier ‘sea almond’, brède ‘blacknightshade’). They may be named differently: épine ‘thornbush’, lagune ‘lagoon’ and pieuvre ‘octopus’ in Metropolitan French have been replaced by piquants, lagon and poulpe.

342 Claire Moyse-Faurie There are also a few specific lexical terms, as in the following, excerpt from Pauleau (1995): – coutume, faire la coutume: set of all rituals and ceremonies carried out on the occasion of an unexpected event in the community; – chemin coutumier: hierarchies to be observed in order to reach certain goals in a Kanak environment. – chef: in New Caledonia, this term was used by the colonial administration to designate those among the Kanak people who had a certain authority and who were chosen as interlocutors, whether or not they had a real superior status: grand chef at district level, petit chef at tribal level. – roussette: similar to a big bat (Pteropus sp.), whose meat is well appreciated. – monnaie: traditional money, made of shells, the backs and the hair of bats, exchanged during the great traditional ceremonies. – brousse (< Provençal ‘brousso’) ‘vegetation’: in relation to Numea, designates the rest of the Grande Terre (‘Mainland’); broussard is somebody who lives outside of Numea; aller en brousse means to leave Numea; and also feux de brousse, école de brousse, centre de brousse, etc. – boucan, emboucanner: originally a fire fed by wood, fill with smoke; in New Caledonia, it means poison, witchcraft, cast a spell on, bewitch. – fin is the superlative corresponding to Standard French ‘très’ (‘very’): fin valable ‘very good’, fin nul ‘completely worthless’; il est fin gras ‘he is very fat’. – claquette: ‘tong, sandal’. – curios (< Engl. ‘curiosity’): ‘souvenir shop’. Some words that would be found offensive in Standard French are commonly employed in New Caledonia, and may even be honorific, as is the use of vieille ‘old woman’ to designate one’s wife or a woman who has at least one child and the use of vieux ‘old man’ for the husband. In New Caledonia, non-autochthonous inhabitants are named Zoreille, Zozo or Métro ‘from Metropole’, Taipouète ‘Tahitian’, Caldoche ‘Caledonian European’, Wawa ‘East Uvean’, Kiwi ‘New Zealander’, Poken ‘Australian’. The use of compounds whose first element functions as a classifier is one of the most striking lexical innovations. Hollyman (1991) called them ‘binomials’:

Borrowings from Romance languages in Oceanic languages

343

– faux- in various compounds designating trees unknown in France but which recall some Metropolitan ones: faux-mimosa, faux-châtaignier, faux-tabac, faux-poivrier, etc. – pomme-: in various compounds designating fruits which recall the apple: pomme liane, pomme canaque, pomme cannelle, pomme citerne, pomme liane, pomme rose, etc. – poisson-: in various compounds designating different kinds of fish: poisson-couteau, poisson-flûte, poisson-lézard, poisson-trompette, poissonventouse, poisson-beurre, etc. – bois ‘wood’ is also frequently found in compounds designating different species of trees: bois-bouchon, bois bleu, bois de fer, bois de rose, bois noir, bois-carotte, bois pétrole, bois-tabou, etc. The few vernacular borrowings found in Oceanic varieties of French are essentially in the fields of flora, fauna and native life and customs. They have undergone phonemic adaptation to the French system. From Kanak languages: dawa ‘fish, Naso unicornis’, manou ‘loin-cloth’, yossi the more common exclamation, pilou ‘traditional dance’, niaouli ‘tree’, Melaleuca leucondendron’, cagou ‘flightless bird, Phynocetos jubatus’, houp ‘very tall tree, Montrouziera cauliflora’, bounia ‘food cooked in earth oven’, etc. From Polynesian languages: fiu ‘be tired of’, faré ‘house’, tapa ‘bark cloth’, kaikai or kakai ‘meal, eat’, popinée (< fafine) ’Melanesian woman’; tabou ‘taboo’, tamanou ‘tree, Calophyllum neo-ebudicum’, etc. 6.2. Interference, language mixing and Kayafou Young Kanaks whose first language is not French encounter a lot of difficulties during their schooling, which is done completely in French. In fact, they have neither a good ability in French, nor a good knowledge of their first language. This situation leads to many interference processes, such as being unable to use anaphors or difficulties in the choice and position of prepositions. More specifically due to interference with Kanak languages: the use of postposed pronominal possessor, introduced by the French preposition à: le parler à nous ‘our way of speaking’ instead of Standard French ‘notre parler’, la maison à eux ‘their house’ instead of ‘leur maison’; systematic use of the dual: les deux clans des deux mariés; inclusive coordination constructions: avec sa mère elles portent un panier. ‘She and her mother are carrying a basket’.

344 Claire Moyse-Faurie So far language mixing only concerns the lexicon, as shown in this excerpt from a conversation in Drehu during a wedding near Numea (transcribed and translated by Fabrice Wacalie). French words are in bold characters: (1)

Nemen la délire i bad e goo? be what this delirium of group at up there ‘What is the group doing up there?’

(délirer is used for ‘to do’ in Kayafou)

(2)

Pegö, bad a nango ketri balo. nothing group TAM a little touch ball ‘Nothing, they are playing football.’

(3)

Eh! Justement, nyibici gabe kola cinyi Cemus elany? excl. precisely truth that TAM shave Cemus tomorrow ‘Eh! Is it really true that we are going to shave Cemus tomorrow?’

(4)

Ka

moië be qene iosi dagörö bala la enaj i viö? way only language excl. dangerous still this beard of old man ‘That’s the way, did you notice the bad [length] of the guy’s beard?’ TAM

The marginalisation of the young Kanaks living in suburban areas has led them to create their own language, which they call Kayafou, a contraction of ‘casse pas la tête’ (‘don’t bother’) and ‘on s’en fout’ (‘we don’t care’). This is no longer a case of interference or language mixing, but a kind of linguistic reappropriation for those facing school failure and the impossibility of acquiring Standard Metropolitan French, which is considered as foreign and as depriving the young of their cultural identity. Barnèche (2004) summarizes the sociolinguistic use of Kayafou among young Kanaks: A Nouméa, les jeunes entre eux parlent presque exclusivement le français. Mais ils ne parlent pas n’importe quel français, ils parlent leur français. Les jeunes identifient ainsi leur usage du français comme une langue spécifique désignée tantôt comme français à nous ou comme français kaya, français débarrassé, opposé au français français, le français standard. L’usage du français kaya permet dès lors aux jeunes de se revaloriser au travers d’un langage qui leur est propre et dans lequel ils se reconnaissent, et de se défaire ainsi de l’image dévalorisante que leur renvoie la constante correction dont ils font l’objet à l’école.

Kayafou is indeed the way young Kanak found to define a new ethnocultural identity.

Borrowings from Romance languages in Oceanic languages

345

7. Tayo, the French-based creole According to Corne (1989), Tayo, the lexically-French Creole language spoken in Saint-Louis, near Numea, passed through different stages: (i) pidginisation, (ii) pidginisation and creolisation leading to a Creole/French continuum, (iii) rapid development of a homogenous creole, (iv) a break in the continuum such that Tayo and French function as distinct languages. First used as a trade language between French missionaries and the Kanak from different parts of the Mainland, Tayo became the first language of children living in Saint-Louis beginning around 1910. The word ‘Tayo’ probably has a Tahitian origin, taio means ‘friend’/’I am your friend’, a kind of welcome salutation. Tayo then refered to a Kanak man (the woman being called popinée), and in the 1960s, according to Ehrhart (1993), it came to refer to the language of Saint-Louis, known also as the ‘patois de SaintLouis’. Interference processes found in the French spoken by Kanaks have become characteristics of the Tayo language. For instance, possession is expressed with the preposition pu < Fr. ‘pour’ (‘for’): pye pu mwa lit. foot/leg for me, ‘my leg, my foot’. Just as in Kanak loanwords from French, Tayo words have incorporated the French definite article (Ehrhart 1993: 110–112): lakisin < ‘cuisine’ (‘kitchen’), lamer < ‘mer’ (‘sea’), lapli < ‘pluie’ (‘rain’); leglis < ‘église’ (‘church’), or the French partitive article ‘du’: dite < ‘du thé’ (‘tea’), diri < ‘du riz’ (‘rice’), dile < ‘du lait’ (‘milk’), etc. A few words have an s- initial, from the sounding of the final consonant of the French plural article ‘les’ when the following word begins with a vowel: sepol ‘épaule’ (‘shoulder’), sorej ‘oreille’ (‘ear’), sje ‘oeil, yeux’ (‘eye’); setwal ‘étoile’ (‘star’). Tayo tends to extend to the far-south of the mainland, endangering the Kanak languages spoken in that part of New Caledonia.

8. Conclusion Tayo was created from the need for communication; Kayafou for the sake of cultural identification. These two situations are of course completely different from the borrowings described at the beginning of this article; borrowings from Romance languages into Oceanic languages improve communication, and allow business to be done or goods to be exchanged; they considerably enlarge the lexicon and sometimes the phonemic system, but do not alter the syntax of a language.

346 Claire Moyse-Faurie In contrast, borrowings from Oceanic languages into French are less numerous, but they do modify the syntax and the pronunciation, giving rise to new varieties of French. Notes 1. Even in the Polynesian Outliers where borrowings were massive, we find borrowed phonemes from Melanesian or Polynesian languages, but not from European ones, although they are very receptive to borrowing. 2. Lit. ‘language of ours’; nouzautes < Fr. ‘nous autres’ (‘our’) stands for the 1st person plural inclusive.

References Barnèche, Sophie 2004 L’identité linguistique et culturelle des jeunes de Nouméa: Plurilinguisme et pluriethnicité en contexte urbain. Thèse de doctorat. Université de Rouen. Bril, Isabelle 2000 Dictionnaire nêlêmwa-nixumwak-français-anglais. Paris: PeetersSelaf (LCP 14). Charpentier, Jean-Michel 1979 Le pidgin bislama(n) et le multilinguisme aux Nouvelles-Hébrides. Paris: Peeters-Selaf. Corne, Chris 1989 Un créole à base lexicale française en Nouvelle-Calédonie: le tayo ou patois de Saint-Louis. Études créoles 12 (2): 29–42. 1990 L’agencement temporel des événements démographiques dans la création d’une langue créole et le tayo de Saint-Louis en NouvelleCalédonie. In L’Homme et le temps, F. Tolron (ed.), 11–27. Nouméa: Corail. 1995 Pour une évaluation de la contribution des langues mélanésiennes dans la formation du tayo. In Parole, communication et symbole, F. Angleviel (sous la dir. de), 167–203. Paris: L’Harmattan. Crowley, Terry 1990 Beach-la-Mar to Bislama. The emergence of a national language in Vanuatu. Oxford: Clarendon Press. 2003 A new Bislama dictionary. Suva, Fiji: Institute of Pacific Studies. Ehrhart, Sabine 1993 Le créole français de St-Louis (le tayo) en Nouvelle-Calédonie. Hamburg: Helmut Buske Verlag.

Borrowings from Romance languages in Oceanic languages

347

Fischer, Steven Roger 2001 Hispanicization in the Rapanui language of Easter Island. In Lo propio y lo ajeno en las lenguas austronésicas y amerindias, Klaus Zimmermann, and Thomas Stolz (eds.), 313–332. Madrid: VervuertIberoamericana. Geraghty, Paul and Jan Tent 1997 Early Dutch loanwords in Polynesia. Journal of the Polynesian Society 106: 131–160; 395–408. 2004 From lowlands to islands: Dutch loans in Polynesia. In Borrowing: a Pacific perspective, Jan Tent, and Paul Geraghty (eds.). Canberra, A.C.T: Pacific Linguistics, Research School of Pacific and Asian Studies, Australian National University. Grimes, Barbara Dix 1994 Cloves and Nutmeg, traders and wars: language contact in the Spice Islands. In Language contact and change in the Austronesian world, Tom Dutton and Darrell T. Tryon (eds.), 251–274. Berlin /New York: Mouton de Gruyter. Hollyman, K. Jim 1962 The lizard and the axe. Journal of the Polynesian Society 71 (3): 310–327. 1991 What’s in a nomial? Folk classification in New Caledonia. In Man and a half. Essays in Pacific anthropology and ethnobiology in honour of Ralph Bulmer, Andrew Pawley (ed.), 114–118. Auckland: The Polynesian Society. Lercari, Claude 2002 Dictionnaire ajië-français à l’usage des étudiants. Nouméa: CDPNC. Ozanne-Rivierre, Françoise 1994 Iaai loanwords and phonemic changes in Fagauvea. In Language contact and change in the Austronesian world, Tom Dutton and Darrell T. Tryon (eds.), 523–549. (Trends in Linguistics, Studies & Monographs 77.) Berlin /New York: Mouton de Gruyter. 1998 Le nyêlayu de Balade (Nouvelle-Calédonie). (LCP 12.) Paris: PeetersSelaf. 2000 Terminologie de parenté proto-océanienne: continuité et changement dans les langues kanak. In En pays kanak, Alban Bensa, and Isabelle Leblic (eds.), 69–100. Paris: Editions de la Maison des Sciences de l’Homme. Palomo, Rosa Salas and Thomas Stolz 2008 Pro or contra Hispanisms: Attitudes of native speakers of modern Chamoru. In Hispanisation, Thomas Stolz, Dik Bakker and Rosa Salas Palomo (eds.), 237–268. Berlin /New York: Mouton de Gruyter.

348 Claire Moyse-Faurie Pauleau, Christine 1995 Le français de Nouvelle-Calédonie. Contribution à un inventaire des particularités lexicales. Edicef/Aupelf. Perini, Antoine 1987 La phonologie du tahitien et l’intégration des emprunts. La Linguistique 23 (2): 131–141. Rensch, Karl 1994 Early European influence on the languages of Polynesia: the Gambier Islands. In Language contact and change in the Austronesian world, Tom Dutton and Darrell T. Tryon (eds.), 477–495. Berlin /New York: Mouton de Gruyter. Rivierre, Jean-Claude 1983 Dictionnaire paicî-français (Nouvelle-Calédonie). Paris: Peeters-Selaf (LCP 4). 1994 Contact-induced phonological complexification in New Caledonia. In Language contact and change in the Austronesian world, Tom Dutton and Darrell T. Tryon (eds.), 497–522. Berlin /New York: Mouton de Gruyter. Rivierre, Jean-Claude and Sabine Ehrhart 2006 Le bwatoo et les dialectes de la région de Koné (Nouvelle-Calédonie). (LCP 17.) Paris: Peeters-Selaf. Ross, Malcolm 2001 Contact-induced change in Oceanic languages in North-West Melanesia. In Areal diffusion and gentic inheritance, Alexandra Y. Aikhenvald and R.M.W. Dixon (eds.), 134–166. Oxford: Oxford University Press. Tent, Jan and Paul Geraghty 2004 Borrowing: a Pacific perspective. Canberra: Pacific Linguistics. Topping Donald T. 1973 Chamorro reference grammar. The University Press of Hawaii. Tryon Darrell T. 1970 Loanwords in Dehu pre-1920. In Pacific Linguistic Studies in honour of Arthur Capell, Stephen Adolphe Wurm and Donald C. Laycock (eds.), 429–437. (Pacific Linguistic Series C-13.) Canberra: Linguistic Circle of Canberra.

Contact-induced changes in Amerindian languages of French Guiana Françoise Rose and Odile Renault-Lescure

1. Introduction French Guiana is an ideal place for the study of language contact, with its six Amerindian languages, various French based and English based Creoles, immigrant languages like Chinese or Hmong and, what is of interest for this volume, Romance languages like French, Portuguese and Spanish (Launey et al. 2003). This paper will focus on contact between on the one hand, two Amerindian languages, Kali’na and Emérillon, and on the other hand Romance languages: French, Portuguese and Spanish. This contact will regularly be compared with the contact of the same Amerindian languages with Creole languages. Theoretically speaking, this case study is particularly interesting in that it deals with on the one hand two typologically similar languages (the Amerindian languages Kali’na and Emérillon) and on the other hand languages that are typologically distant from each other (Romance languages and Creoles) and likewise distant from Kali’na and Emérillon. Section 2 will give as preliminaries a presentation of the Kali’na and Emérillon languages, with a short history of their contact situations. Section 3 and 4 will respectively deal with the phonological and morphosyntactic contact-induced changes in both languages. 2. Kali’na, Emérillon and their contact situations Kali’na1 is the Cariban language which covers the largest geographical area, spreading from the north-eastern savannahs of Venezuela to the north of the Brazilian state of Amapa, through all three Guyanas. In French Guiana, the Kali’na population (around 3500 people, but a lower number of speakers) is spread out into different villages to the west of Cayenne, as far as the Maroni river (cf. Map 1). Because of its coastal distribution, Kali’na has been in contact with a variety of other populations.

350 Françoise Rose and Odile Renault-Lescure Emérillon is the northernmost member of the geographically widespread Tupi-Guaraní family, present in French Guiana since the late 15th century. The 400 members of the group (whose autodenomination is Teko), are all fluent speakers of the language, and live exclusively in French Guiana, in two areas of the rainforest, one next to the border of Surinam, and the other on the Brazilian border (cf. Map 1). Because of its peripheral and more isolated situation, Emérillon has looser contacts with other non Amerindian populations. These two languages display a different history of contacts. This contact history is basically constituted of three phases: a) First, was the arrival of the Europeans and the merchandise trade. The Kali’na people, living on the Guianese coast, referred to contact goods with the words of the first Europeans they met, namely Spanish and Portuguese, and less commonly English, Dutch and French. The Kali’na lexicon took on a stock of borrowings which were then diffused all along the coast of the Guyanas, from the mouth of the Orinoco to the Approuague, in eastern Guiana. It was mainly diffused by means of a Carib based pidgin named “langue générale galibi” used between the different Amerindian populations, among them, the Emérillon population, who was meanwhile located in the hinterland, between the Inini, the Approuague and the Oyapock rivers. This first phase corresponds to direct but occasional contacts for Kali’na people (called “casual contacts” by Thomason 2001: 70), such as the use of interpreters with traders or missionaries. For the Emérillon people, those contacts were indirect, through the use of the Galibi Pidgin. b) In a second phase, contacts with Romance languages decreased, being taken over by contacts with vehicular languages developed in the colonies with the development of slave trade. Those languages are: – Sranan Tongo, the Creole of the plantations of Surinam, born in the second half of the 17th century. Its lexicon is essentially based on English, with some contribution from Dutch, Portuguese and Kikongo. It has been hypothesized that a great number of its grammatical structures is based on those of the African Gbe languages. – The Guianese Creole, a French based Creole, with possibly some grammatical structures of the Fon language. A certain degree of bilingualism of the Kali’na speakers with Sranan lasted until the end of the 20th century, due to their history as refugees in the Dutch colony in the 17th and 19th century, and to strong commercial relations across the border. Meanwhile, the Emérillon population has just

Contact-induced changes in Amerindian languages of French Guiana 351

Map 1. Map of regional languages of French Guiana (Goury 2001)

352 Françoise Rose and Odile Renault-Lescure been in contact with the English based Creole Aluku in the late 18th century in the mid-Maroni region. The Kali’na population established intense social relations with the Guianese Creoles later on, as they went back to French Guiana and some of them very likely became bilingual. The Emérillon people also established some commercial contacts with Creole populations, but stayed isolated in the southern part of the colony. c) In a third phase, Guiana underwent “francization”, the unification of its administrative system as a French department, and the settlement of French institutions such as administration representations and schools (The first Kali’na children went to school in 1945, and the first Emérillon children in 1956). As a consequence, contact with French became more intense, especially for the Kali’na people, with French tending to substitute Guianese Creole as a vehicular language nowadays. Contacts with Creoles decrease, and speakers attitudes towards those languages change too. Mastering French is more or less seen as a key for social success (for work, studies, and implication in the political and administrative structures). Today, bilingualism with French is more widespread for Kali’na speakers than for Emérillon speakers. It is worthwhile to note that nowadays, those language contacts take place in a context of wider plurilingualism. Migrations of the late 20th century triggered a rise of Surinamese Creoles near the western border of the department and of Brazilian Portuguese near its eastern border. Table 1. Kali’na and Emérillon histories of contacts Historical times

Main contact languages

Type of contact for Kali’na (Kal.)

Type of contact for Emérillon (Em.)

1. first contacts with Romance languages

Spanish (Sp.) Portuguese (Port.)

direct and occasional

indirect

2. rise of the Creoles

Creoles: Sranan Tongo (Sr.) Guianese Creole (Cr.)

some bilingualism

little contact

3. “francization”

French (Fr.)

intense (widespread bilingualism)

quite intense

To summarize, both Kali’na and Emérillon have successively had contacts with European languages, Creoles and finally French, each period being

Contact-induced changes in Amerindian languages of French Guiana 353

characterized by a stronger intensity of contacts. In each period, Emérillon contact situation is somewhat less intense than that of Kali’na. On the whole, contact-induced changes in Kali’na and Emérillon consist essentially of lexical borrowings, regardless of whether the source language is a Romance language or a Creole. A few syntactic changes will be presented further on. However, since lexical borrowing may gradually lead to phonological and structural changes in the borrowing language, we will describe both how borrowings get adapted to the receiving systems, and how the systems adjust to the borrowings. Section 3 will present the integration of the borrowings at the phonological level, and section 4 at the morphosyntactic level. Our main interest will be to compare the integration of borrowings from different languages, in two different but comparable languages. Let us add three caveats. First, it is not always a simple task when studying a particular phenomenon of language interference to determinate whether code-switching or borrowing is concerned. As a consequence, this paper is based only on linguistic facts that are unambiguously borrowings (for a discussion of code-switching, cf. Auer 1999). Second, the source language is not always easy to determine, especially within the following pairs of possible source languages: French and Guianese Creole, Guianese and French West Indies Creoles, and also Sranan Tongo and Aluku, especially when the word is quasi identical in both of the possible source languages. Third, although beyond the scope of this paper, it is important to keep in mind that interferences among Amerindian languages are also attested. For instance, Emérillon has borrowed a Cariban plural marker -kom, possibly from Kali’na. 3. Similarities and differences in the phonological integration of borrowings On the phonological level, besides phone substitution that we will not describe in this paper, many other processes of adaptation of the borrowed items can be found. In 3.1, we will focus on one very specific process: nasalization/denasalization in Emérillon. However, borrowings do not always completely adapt to the system2, and eventually it is sometimes the system itself that adjusts to the borrowed words and therefore undergoes remarkable changes. In 3.2, we will show how the Kali’na phonological system evolved in a substantial way on account of lexical borrowing.

354 Françoise Rose and Odile Renault-Lescure Eventually, and this point is particularly interesting from a theoretical standpoint, we will show in 3.3 how the same phonological constraint yields two different processes of integration for words of the same origin in Kali’na and Emérillon. 3.1. A specific integration process: (de)nasalization in Emérillon One of the peculiarities of Emérillon is its suprasegmental nasality. The /~/ feature is assigned at the lexical level and applies across a given morpheme to specify the [nasal] value of its phonemes. Only vowels and voiced consonants can be specified as [nasal]. Other phonemes are transparent and opaque to nasalization: they are not affected by nasalization and do not block its spreading. Table 2. Examples of oral and nasal morphemes in Emérillon Oral morphemes ba e ~ mba e tap
d o-bo-aku 3-CAUS3-hot

thing house he heats it

Nasal morphemes mã  tãm
 o-mõ-ãtã 3-CAUS-hard

REL

grandfather he hardens it

Foreign items follow this constraint, and are therefore integrated as either oral or nasal morphemes, probably according to the nasal or oral value of the last phoneme, since nasality seems to apply from right to left. Accordingly, phenomena of denasalization and nasalization are observed in the borrowing process:

denasalization

Table 3. Examples of denasalization in the borrowing process Source language

Emérillon

Meaning

ma cocotte [makokt] (Fr.) Mon Père (Fr.) [mõp ]

bakoko| [bakokt] bope| [bopt]

pot priest

nasalization


Table 4. Examples of nasalization in the borrowing process Source language

Emérillon

Meaning

dipen (Cr.) [dipe
] farine (Fr.) ~ farin (Cr.) [fa in] zoranj (Cr.) [zo a ]

n pe
pãn n zona

bread flour orange (fruit)

Contact-induced changes in Amerindian languages of French Guiana 355

3.2.

Contact-induced system-altering changes in Kali’na

The first borrowings from the Romance languages into Kali’na at the time of colonization do not seem to have induced any structural change on Kali’na’s phonological system. This is not true, however, for the subsequent borrowings from Sranan Tongo. These borrowings induced tendencies towards certain phonological changes that were later reinforced by more recent borrowings from French and Guyanese French based Creole. Below are presented the introduction of a new phoneme and the transfer of a voice opposition. 3.2.1. Introduction of a new phoneme The Kali’na phonological system displays eleven consonants classified in Table 5 according to manner and place of articulation. Table 5. Kali’na consonant system Labial

Apical

OBSTRUENTS

stops fricatives

p (f)

t s

SONORANTS

nasals glides liquid

m w

n

Palatal

Velar

Glottal

k

 h

y

l

In the first historical phase of contact, among other regular phoneme substitutions in borrowed words, /p/ regularly substituted for /f/ in borrowings, by virtue of being the only native obstruent at the same place of articulation (Renault-Lescure 1985). This is the case regardless of the source language. (1)

francês (Port.) > palansi

(Kal.) ‘Frenchman’

(2)

swafroe (Sr.)

(Kal.) ‘matches’

> suwapulu

Later, however, variable realizations were tolerated, such as: (3)

fensre

(Sr.)

> pesele ~ fensele (Kal.) ‘window’

After this period of instability, the regular replacement of /f/ by a /p/ stop in borrowings was brought to an end, the fricative sound being finally maintained in borrowings.

356 Françoise Rose and Odile Renault-Lescure (4)

frigi

(Sr.)

> filiki

(Kal.) ‘kite’

This tendency was reinforced afterwards through borrowings from French and French based Creole. (5)

lafinèt (Cr.)

> lafinet[
] (Kal.) ‘window’

(6)

suflèt

(Cr.)

> suflet[
]

(7)

fil

(Cr. ~ Fr.) > fil[
]

(Kal.) ‘sewing thread’

(8)

foto

(Cr. ~ Fr.) > foto

(Kal.) ‘photo’

(Kal.) ‘whistle’

Our hypothesis, in keeping with Weinreich ([1953]1970: 18), is that the empty fricative slot in the labial consonant inventory of Kali’na was a structural factor favoring the introduction of a new phoneme filling a gap in the system. 3.2.2. Transfer of a voice opposition Although there is a the lack of opposition between p/b, t/d, and k/g in Kali’na, the stops /p/, /t/ and /k/ are sometimes realized as voiced stops [b], [d] and [g] when they are not word initial. This voicing of stops is difficult to explain. A study in progress (Renault-Lescure and Gomez 2005) shows a link between some of these realizations and the prosodic and syllabic structures, but does not yet explain all of these realizations. Among the hypotheses, a possible explanation relies on contact with languages in which the voicing opposition is relevant for stops, more specifically through lexical borrowings. With the introduction of loanwords maintaining a voiced stop word-initially, a new opposition is indeed emerging. The oldest loanwords retain the pattern of allophonic distribution, by which voiced stops are borrowed as voiceless word-initially. (9)

bandera (Sp.) > pantila [pandi’la] (Kal.) ‘flag’

(10) barque

(Fr.)

> paliki [paali’gi]

(Kal.) ‘bark’

(11) grasi

(Sr.)

> kalasi [kalaa’çi]

(Kal.) ‘glass’

More recent loanwords, from the Creoles or from French, maintain a voiced realization word-initially, which has led to the introduction of a new opposition:

Contact-induced changes in Amerindian languages of French Guiana 357

(12) pali

‘barrage’

(13) panki ‘skirt’




bali

‘barrel’

(< Sr.)




banki ‘bank (seat)’ (< Sr.)

It is worth noting that Cariban languages do not usually display a voicing opposition, but that in certain of those languages, its emergence has been recorded and presented as a likely consequence of contact (Gildea 1998). 3.3.

Differences in the processing of consonant clusters in borrowings

Both Kali’na and Emérillon share a phonological constraint that restricts consonant clusters. Emérillon syllables are all open, with the exception of final syllables which may be closed by a single consonant. Consequently, no consonant cluster within the domain of the morpheme is allowed, and morphophonemic rules extend this domain to the word level. Table 6. Canonical syllabic pattern of the Emérillon word

(C)V-…- (C)V- (C)V(C) In Kali’na, the syllabic structure is (C)V1(V2,C). Accordingly, consonant clusters are possible, but their number is restricted by the particular distribution of consonants. All eleven consonants, with the exception of the glottal stop, can occur in the onset position. The coda consonants are either nasals (word-internally and word-finally) or the glottal stop (word-internally only). To summarize, the only sequences of consonants that are possible are word internal, C1 being necessarily a nasal or a glottal stop, and C2 being any consonant but the glottal stop. Now words borrowed from Romance languages often contain consonant clusters that are not allowed by the two recipient languages of our study. However, these clusters get integrated differently into the Kali’na system and into that of Emérillon. 3.3.1. In Kali’na Consonant clusters that violate the syllable constraints are readjusted by the insertion of a vowel between two consonants (in bold in Table 7). These processes are observed regardless of the source language. The second, third and fifth lines of Table 7 also show the insertion of a final vowel.

358 Françoise Rose and Odile Renault-Lescure Table 7. Examples of vowel insertion to break up a consonant cluster Consonant clusters

Source language

Kali’na

Meaning

pl fr br sk sp kl

plata (Sp.) francês (Port.) brande-wijn (Dutch) biscuit (Fr.) spoen (Sr.) lakle (Cr.)

plata palansi[si] palantuwini pisukuwi sipunu lakele

money Frenchman rhum crackers spoon key



pVl pVl pVl sVk sVp kVl

The quality of epenthetic vowels is determined by progressive or regressive assimilation, or is by default a vowel prone to devoice (i, ). These rules do not apply to the most recent borrowings (from Creole or French) that conserve consonant clusters. (14) garden

(Sr.)

>

kalden

(Kal.)

‘mosquito-net’

(15) dilwil

(Cr.)

>

dilwil

(Kal.)

‘oil’

(16) taxi

(Fr.)

>

taxi [taksi] (Kal.)

‘taxi’

3.3.2. In Emérillon Two processes occur with borrowed words to maintain the syllable constraints: either consonant deletion simplifies the cluster (as in Table 8) or vowel epenthesis breaks up the consonant cluster (as in Table 9). By and large, we can posit that deletion takes place when the first consonant of the cluster is a liquid4 (and possibly also when the cluster is in final position), and vowel epenthesis takes place between any other two consonants. Table 8. Simplification of consonant clusters Simplification of consonant clusters through deletion

Source language

Emérillon Meaning

rm m rt t ld d final position final position

gendarme [ãda m] (Fr.) marteau [ma to] (Fr.) ~ marto (Cr.) soldat [slda] (Fr.) ~ sòlda, soda (Cr.) (la) piste (Fr.) (la) table (Fr.) ~ tab (Cr.)

ãdam bato soda lapis latab

policeman hammer soldier airstrip table

Contact-induced changes in Amerindian languages of French Guiana 359 Table 9. Examples of vowel insertion to break up a cluster of consonants

tr tr tr sk rt rk

ton tu| ta| sik |et |ak

Source language

Emérillon

Meaning

citron (Fr.) ~ sitron (Cr.) citrouille (Fr.) travail (Fr.) ~ travay (Cr.) biscuit (Fr.) ~ biskwi (Cr.) carta (Sp.) arcabuz (Sp.)

sitono situ|u ta|awa bisiku ka|eta a|akapusa

lemon pumpkin work crackers paper, book, notebook gun

The quality of the epenthetic vowel is usually determined by assimilation to the quality of the following vowel. In certain cases (like ka|eta and a|akapusa), an epenthetic vowel is found in cases where the cluster starts with a liquid, but these words were borrowed indirectly from Spanish via Kali’na. In conclusion, the comparison between Kali’na and Emérillon shows that two languages with a similar constraint on consonant clusters may react differently to accommodate borrowed items. We can however hypothesize that Kali’na does not use the simplification process because it has a tendency to accept (and sometimes even favor) polysyllabic stems and words. Please note that as far as phonology is concerned, the processes induced by interference of Kali’na and Emérillon with other languages are similar regardless of the phonological system of the source language, whether it is a French or English based Creole or a Romance language. 4. Similarities and differences in the morphosyntactic integration of borrowings Once again, our data is particularly enlightening in that both Amerindian languages are typologically comparable, while they clearly differ in structure from French and Creoles, those latter being themselves typologically distant. Many linguistic constraints on contact-induced changes are indeed based on typological similarity and distance. Kali’na and Emérillon are typologically very similar: both tend to be agglutinating and polysynthetic. Accordingly, the predicate is necessarily made up of a root and a person marker, but can also take numerous prefixes and suffixes, as well as clitics and an incorporated noun. Relations between the arguments and the predicate are marked on the predicate with a person index, following a hierarchical system (Renault-Lescure 2002; Rose 2002 and

360 Françoise Rose and Odile Renault-Lescure 2003a), and full NPs are facultative and actually rare in discourse. The syntactic importance of the predicate and the richness of its morphology clearly contrasts with their French or Creole counterparts. In this respect, Kali’na and Emérillon are clearly different from the fusional Romance languages and the isolating Creoles. French verbs are characterized by personal clitics and a fusional conjugation system. Creole verbs are very poor morphologically. On one side of the debate on linguistic constraints on interference (cf. for instance Thomason and Kaufman 1988, chapter 2) are beliefs like Field’s Principles of System Compatibility, which asserts that the borrowing language’s morphological typology (as isolating, agglutinating or fusional) will constrain the possibility of borrowing from another language (Field 2002). In our case, Kali’na and Emérillon, being agglutinating languages, could borrow agglutinating but not fusional morphology. There are indeed no instances of borrowed morphology from Romance languages, which are fusional languages. As a consequence, only roots are borrowed. Another of Field’s assertions is that all languages can borrow instances of isolating morphology (Field 2002). The isolated morphemes of Creoles are indeed easily borrowed into Kali’na and Emérillon, whereas the Romance items that are borrowed are either simple roots, or roots with morphology (i.e NPs, or part of NPs) reanalyzed as simple roots. In 4.1, we will first describe the process through which borrowed items are ascribed to a category in both receiving languages, before showing how they integrate in the morphosyntax of their host systems in 4.2. Finally, section 4.3 will suggest a few syntactic changes induced by borrowings. 4.1. Class assignment In this section, we will focus on the attribution of a word category to borrowings. We will limit the discussion to the four predicative categories Hengeveld (1992, chapter 4) uses to classify languages according to their parts of speech systems. Whereas the Romance languages distinguish clearly these four categories, respectively nouns, verbs, adjectives and adverbs, the Creole languages offer examples of multifunctionality (Bruyn 2002). Some words function as noun, verb or adjective without any change in word class being morphologically marked. This ability seems to be quite common in Sranan Tongo, especially with forms used as a verb and as a noun, and less common in Guianese Creole.

Contact-induced changes in Amerindian languages of French Guiana 361

Kali’na differentiates only three major word classes: nouns, verbs, and adverbs (playing the role of qualifiers once nominalized). All these words shift easily from one class to the other through derivational processes. Emérillon presents four classes of roots, verbs, nouns, adverbs, and a small class of roots with “descriptive/adjectival” meaning that cannot be analyzed as nouns, verbs, nor adjectives (cf. Rose 2003b). 4.1.1. In Kali’na We shall now examine the various strategies found in Kali'na for assigning a word category to borrowings. Borrowings without category change are illustrated in Table 10. Table 10. Borrowings without category changes Category in the source language

Source language

Category in Kali’na

Kali’na

Meaning

noun

zapato (Sp.) kerki (Sr.) dilèt (Cr.) auto (Fr.)

noun (1)

sapato keleke dilet oto

shoes church milk car

noun

perro (Sp.) poespoesi (Sr.)

noun (2)

pelo pusipusi

dog cat

adverb

pannantan (Cr.) tijou (Cr.)

adverb

pannantan tijou

meanwhile always

Borrowed nouns, regardless of their source language and the period of borrowing, are integrated into the class of Kali’na nouns. The vast majority of them fall into the sub-class of alienable possessed nouns – category (1) of Table 10 – while only a few fall into the sub-class of non-possessed nouns which have a suppletive form in the possessive construction – category (2) of Table 10: (17) paila ‘bow’, alakaposa ‘gun’ (< Sp.)

ø-lapal

‘my weapon, my bow, my gun’

(18) kulewako ‘parrot’ pelo ‘dog’ (< Sp.)

y-ek

‘my pet, my parrot, my dog, my cat’

362 Françoise Rose and Odile Renault-Lescure (19) nimoku ‘hammock’ lit (< Cr.) ‘bed’

a-pat

‘your sleeping place, your hammock, your bed’

There are no instances of words being transferred into the sub-class of inalienably possessed nouns (kin terms, body parts, parts of a whole). Borrowed adverbs integrate into the category of Kali’na adverbs. Borrowings with category changes are now illustrated. Tables 11 to 13 show examples of items classified as nouns, verbs, or adjectives in the source languages and that are reanalyzed as nouns in Kali’na. These nouns fall into 3 sub-classes that are specific to borrowed items and show a defective behavior. Table 11. Borrowings of verb/noun bases as nouns Category in Sranan Tongo

Source language

Category in Kali’na

Kali’na

Meaning

verb/noun

begi (Sr.)

noun (3)

begi-

prayer

Table 11 shows that borrowings to the Sranan Tongo verb/noun category fall in a specific sub-class of nouns in the recipient language – nouns (3) – and are always treated as uninflected nominal stems to which a verbalizing suffix is attached to form transitive verbs, carrying a person prefix and a tense suffix on the model of a regular verbalizing process in Kali’na. (20) Tense suffix tamusi si-begi-ma-e God 1A-prayer-VERB-PRES ‘I
pray
God’
(21) Kali’na (Courtz 1997) kasili s-aiku-ma-e beer 1A-liquid-VERB-PRES ‘I make cassava beer’ This strategy involves a creative adaptation process and conforms fully to Kali’na patterns of derivation. A similar result of adaptation is mentioned for Japanese by Loveday 1996 (cited in Winford 2003: 50). Thus Kali’na borrowed the Sranan verb/noun forms as nouns, but conserved their verbal meaning by using them in a verbalizing construction.

Contact-induced changes in Amerindian languages of French Guiana 363

Why is the Sranan verb, unspecified for valency, always used in a transitive construction? All the authors converge in the idea that morphological adaptation and class assignment may be hindered when the recipient language has complex rules. It is the case with the verbal system of Kali’na that displays a sub-system of split intransitivity. One may think that this strategy permits to avoid the assignment of the borrowed verb to one of the verbal intransitive sub-classes, “active” or “stative”. Table 12 illustrates items that were verbs in the source language (the invariable verbal form from Guyanese Creole and the infinitive form from French) and that, once borrowed in Kali’na, are reanalyzed as nouns. They form a new sub-class of nouns – nouns (4), which are characterized by the absence of flexion and their necessary integration into a postpositional group, where the postposition is always poko ‘busy with’. Table 12. Borrowings of verbs as nouns Category in the source language Verb

Source language pentiré (Cr.) nétwayé (Cr.) ~ nettoyer (Fr.) comprendre (Fr.)

Category in Kali’na noun (4)

Kali’na

Meaning

pentiré nétwayé ~ nettoyer comprendre

painting cleaning understanding

The postpositional group functions in a single-participant copular construction (22) or in a two-participant construction with the verb ‘to put’ (23). (22) pentiré poko man paint busy.with 3S.COP.PRES ‘He is painting’ (23) woto nettoyer poko s--ya fish cleaning busy.with 1A-put-PRES ‘I am cleaning the fish’ (lit. ‘I am putting it out for cleaning’) Table 13 shows adjectives from Guyanese Creole and French, or multifunctional forms from Sranan, that are reanalyzed as nominal roots, constituting sub-class (5).

364 Françoise Rose and Odile Renault-Lescure Table 13. Borrowings of adjectives or verb/nouns as nouns Category in the source language

Source language

Category in Kali’na

Kali’na

Meaning

Verb/noun Adjective

pina (Sr.) pur (Fr.)

noun (5)

pina pur

(be) miserable pure person

These nominal roots are invariable and take a predicative suffix to be used as a predicate in a very productive structure with a copula on the Kali’na model. (24) Copula pur-me man pure-PRED 3.COP.PRES ‘He is pure’ (25) Model pitani-me man child-PRED 3.COP.PRES ‘He is a child’ To summarize, multifunctional forms, verbs and adjectives undergo a category change when borrowed into Kali’na. They are systematically reanalyzed as nouns and therefore, in order to be used as predicate, need to be embedded in predicative structures. It is noteworthy that those nouns form sub-classes characterized by a defective behavior. In line with Wichmann and Wohlgemuth’s typology of loan verb embedding patterns (in this volume), two strategies can be identified: the strategy of indirect insertion with affixation of a verbalizer as in example (20), and the light verb strategy as in (22), (23) and (24). This propension to favor the borrowing of multifunctional items supports Thomason’s claim that “less tightly structured features are easier to borrow than features that fit into tightly integrated closed structures” (Thomason 2001: 69). 4.1.2. In Emérillon In contrast, most of the words borrowed from Romance languages into Emérillon are assigned to the equivalent word class. The great majority of those borrowings are nouns, with only one verb and one adverb borrowed as such.

Contact-induced changes in Amerindian languages of French Guiana 365 Table 14. Examples of borrowing with no category change Word category

Source language

Emérillon

Meaning

Noun

camisa (Sp.)

kamita

Verb/noun

travailler (Fr.) ~ travay (Cr.) vite (Fr.)

ta|awa

fabric, traditional skirt (piece of fabric tied to the waist) to work/work

wi| [wit]

quickly

Adverb

The verb ta|awa ‘to work’ originates either in the French verb (or noun) root or in the Creole verb. In Emérillon, it is used with the usual verb morphology (person prefix, tense…). (26) si-ta|awa -ta| 1INCL-work-FUT ‘We will work’ On the other hand, some borrowings are the result of a reanalysis of several morphemes as one root. The most obvious process is the fusion of the French article (or part of it) with the noun it determines. Most of such cases may be indirect borrowings through Creole, and then the fusion probably took place in the borrowing process from French into Creole. Table 15. Fusion of several morphemes into one Emérillon root French

Creole

Emérillon

Meaning

article+noun

l’école [lekol]

lekol

lekol

school

part of the article+noun

orange (PL: des oranges [dezo ã])

zoranj

zona

orange (fruit)

Although rare, the most interesting case concerns the few roots that are borrowed in a different category from the one they belonged to in the source language. Field (2002: 44) defines the Principle of Reanalysis. As far as equivalence of word class is concerned, semantic characteristics appear to be more relevant than word classes themselves. Thus it is possible that a form belonging to one semantic sub-type in a language may need to be reanalyzed as belonging more properly to a corresponding semantic sub-type that correlates to an entirely different word class in the recipient language. This is the case for Romance adjectives borrowed in Emérillon. Emérillon possesses two classes of words with “adjectival meaning”. One expresses qualities concerning human beings, such as physical sensations or mental

366 Françoise Rose and Odile Renault-Lescure phenomena, and is morphosyntactically a sub-class of nouns. The other expresses properties usually applied to objects, like size, color, value, and consists of descriptive roots that can be analyzed neither as nouns, verbs, nor adjectives. Our data show five cases of borrowing of French adjectives. Interestingly, they seem to fall into the two categories described above (i.e. the subclass of nouns expressing human qualities, and the class of descriptive roots expressing object qualities) on semantic grounds. Four of them are reanalyzed as nouns ( i ‘richness’, en ‘youth’, alu ‘jealousy’ and t adisjonel ‘tradition’) and refer mainly to human qualities. Only one enters the descriptive root category, and is definitely an object quality (sale ‘salted’). As Field proposed, this reanalysis seems to be due to the semantic sub-type of the items (Field 2002). Therefore, examples of category change include four examples of French adjectives borrowed as nouns and also one example of a French verb borrowed as an Emérillon noun (Table 16). The French verb toucher ‘to touch’ is used as an Emérillon noun to refer to “social allowances”, on account of the collocation of both items in the French phrase toucher les allocations ‘to receive social allowances’. Table 16. Examples of borrowings with a category change Form in French

Category in French

Meaning in French

Form in Emérillon

Category in Emérillon

Meaning

toucher [tue]

verb

to touch (to receive, for allowances) traditional

zo-tue

possessed noun

the allowances

traditionnel adjective [t adisjonl] jeune adjective [œn]

young

nõdepossessed t adisionl noun i-en possessed noun

our traditions his-youth ~ he is young

Manifestly, whereas French or Creole nouns are always borrowed as facultatively possessed nouns, French verbs and adjectives seem always to be borrowed as obligatorily possessed nouns. We hypothesize that this fact is tightly linked with the high predicatibility of possessed nouns in Emérillon. In fact, every noun with a personal prefix can constitute a possessive predicate, as illustrated in (27). Nouns expressing qualities are most often used with that function (28).

Contact-induced changes in Amerindian languages of French Guiana 367

(27) e-men 1SG-husband ‘I have a husband’ (28) e-kaneõ 1SG-fatigue ‘I am tired’
As a consequence, French adjectives borrowed as obligatorily possessed nouns maintain the possibility to be used as predicates, one of their main functions. (29) ie-i-en 5 RED -3-youth ‘They are young.’ Both Emérillon and Kali’na data seem to confirm Field’s Hierarchies of Borrowability according to which nouns are more easily borrowed than adjectives and verbs (Field 2002). This is observable both in terms of the quantity of items borrowed in each category and in terms of the need for a morphological reanalysis to allow the integration of the borrowed item. Now focusing more specifically on verb borrowing, Weinreich ([1953] 1970) asserted that verbs are hard to borrow. A usual explanation lies in the fact that inflected forms are harder to borrow. Our data illustrate perfectly this issue, since verbs of both Amerindian languages and French take obligatory flexion. The precise nature of verbal borrowings is very enlightening: from Sranan, Kali’na borrowed verb/nouns, from Creole, verbs that are anyway uninflected in the source language, and from French, some non-inflected infinitive verbal forms. We have already seen that all of them are turned into nouns in Kali’na. Now the only French (or Creole) verb that Emérillon borrowed as a verb is actually also a noun in both possible source languages. It is in fact also borrowed as a noun in the recipient language: travail ‘work’/ travailler ‘to work’ gave rise to ta|awa

‘work, to work’. zo-a| -a-pe (30) ta|awa
-am za-bae-oka| work-TRANSF INDET-make-CAUS INDET-son-a-for ‘It would provide our sons some work’ In brief, most verbs that are borrowed in Kali’na and Emerillon display some nominal properties, and all of them are actually borrowed under a noun form (with ta|awa
also displaying a verbal form).

368 Françoise Rose and Odile Renault-Lescure 4.2. Full grammatical integration Now that we have described how borrowings get ascribed to a word category, let us show how, in Kali’na (Renault-Lescure 2005) and Emérillon, borrowed words become fully integrated into the morphosyntactic system of the language. On the one hand, they become compatible with the very rich morphology of their receiving language, and on the other hand, they assume syntactic roles just like those of the other members of the categories they now belong to, with the exception of sub-class (4) and (5) of Kali'na nouns, the defective behavior of which has been described in 4.1.1. The series of examples (31) to (36) show how borrowed nouns and verbs in Emérillon are compatible with the complete morphological apparatus of the language: possessive affixes (31), plural (31), demonstrative (32), derivative suffix (33), second position discourse particles (34), and also with the reduplication process (35), person indexes and tense markers on verbs (36). (31)

o-i u o-sapato-kom 3SG-put 3COREF-shoe-PL ‘He puts on his shoes’

(32) dati a

magas
store ‘This store wasn’t here’

COP DEM

< zapato (Sp.)

< magasin (Fr.) ~ magazen (Cr.)

(33) a|akapusa-uhu o-mõdu|-o bal gun-big 3-send-PL bullets ‘Big guns were sending bullets’

< arcabuz (Sp. via Kal.)

(34) mama-n e|e-
ika Mom-CONTRAST 2SG-kill ‘It is Mom that you killed!’