Grammatical Borrowing in Cross-Linguistic Perspective
≥
Empirical Approaches to Language Typology 38
Editors Georg Bossong Bernard Comrie Yaron Matras
Mouton de Gruyter Berlin · New York
Grammatical Borrowing in Cross-Linguistic Perspective
Edited by Yaron Matras Jeanette Sakel
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 Grammatical borrowing in cross-linguistic perspective / edited by Yaron Matras, Jeanette Sakel. p. cm. ⫺ (Empirical approaches to language typology ; 38) Includes bibliographical references and index. ISBN 978-3-11-019628-3 (cloth : alk. paper) 1. Language and languages ⫺ Foreign elements. 2. Grammar, Comparative and general. I. Matras, Yaron, 1963⫺ II. Sakel, Jeanette, 1973⫺ P324.G73 2007 410⫺dc22 2007042917
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 978-3-11-019628-3 ISSN 0933-761X © Copyright 2007 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.
Contents List of contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Yaron Matras and Jeanette Sakel Types of loan: Matter and pattern . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Jeanette Sakel The borrowability of structural categories . . . . . . . . . . . . . . . . . . . . . . . . . . 31 Yaron Matras Grammatical borrowing in Tasawaq . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 Maarten Kossmann Grammatical borrowing in K’abeena . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 Joachim Crass Grammatical borrowing in Likpe (Sɛkpɛlé) . . . . . . . . . . . . . . . . . . . . . . . 107 Felix K. Ameka Grammatical borrowing in Katanga Swahili. . . . . . . . . . . . . . . . . . . . . . . 123 Vincent A. de Rooij Grammatical borrowing in Khuzistani Arabic . . . . . . . . . . . . . . . . . . . . . 137 Yaron Matras and Maryam Shabibi Grammatical borrowing in Domari. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151 Yaron Matras Grammatical borrowing in Kurdish (Northern Group) . . . . . . . . . . . . . . . 165 Geoffrey Haig Arabic grammatical borrowing in Western Neo-Aramaic. . . . . . . . . . . . . 185 Werner Arnold
vi
Contents
Grammatical borrowing in North-eastern Neo-Aramaic . . . . . . . . . . . . . 197 Geoffrey Khan Grammatical borrowing in Macedonian Turkish . . . . . . . . . . . . . . . . . . . 215 Yaron Matras and Şirin Tufan Grammatical borrowing in Kildin Saami . . . . . . . . . . . . . . . . . . . . . . . . . 229 Michael Rießler Grammatical borrowing in Yiddish . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245 Gertrud Reershemius Grammatical borrowing in Hungarian Rumungro . . . . . . . . . . . . . . . . . . 261 Viktor Elšík Grammatical borrowing in Manange . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283 Kristine A. Hildebrandt Grammatical borrowing in Indonesian . . . . . . . . . . . . . . . . . . . . . . . . . . . 301 Uri Tadmor Grammatical borrowing in Biak . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329 Wilco van den Heuvel Sino-Vietnamese grammatical borrowing: An overview. . . . . . . . . . . . . . 343 Mark J. Alves Recent grammatical borrowing into an Australian Aboriginal language: The case of Jaminjung and Kriol. . . . . . . . . . . . . . . 363 Eva Schultze-Berndt Grammatical borrowing in Rapanui . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 387 Steven Roger Fischer Grammatical borrowing in Nahuatl. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403 Una Canger and Anne Jensen Grammatical borrowing in Yaqui . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 419 Zarina Estrada Fernández and Lilián Guerrero
Contents
vii
The case of Otomi: A contribution to grammatical borrowing in cross-linguistic perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 435 Ewald Hekking and Dik Bakker Grammatical borrowing in Purepecha. . . . . . . . . . . . . . . . . . . . . . . . . . . . 465 Claudine Chamoreau Grammatical borrowing in Imbabura Quichua (Ecuador) . . . . . . . . . . . . 481 Jorge Gómez-Rendón Grammatical borrowing in Paraguayan Guaraní. . . . . . . . . . . . . . . . . . . . 523 Jorge Gómez-Rendón Grammatical borrowing in Hup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 551 Patience Epps Mosetén borrowing from Spanish. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 567 Jeanette Sakel Index of subjects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 581 Index of authors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 594
List of contributors
Mark Alves Montgomery College
Wilco van den Heuvel University of Manchester
Felix Ameka Leiden University
Kristine Hildebrandt University of Manchester
Werner Arnold University of Heidelberg
Anne Jensen University of Copenhagen
Dik Bakker University of Lancaster
Geoffrey Khan University of Cambridge
Una Canger University of Copenhagen
Maarten Kossmann Leiden University
Claudine Chamoreau CELIA (CNRS-IRD-INALCOPARIS VII)/CIESAS-Mexico
Yaron Matras University of Manchester
Joachim Crass University of Mainz Viktor Elšík Charles University, Prague Patience Epps University of Texas at Austin Zarina Estrada Fernández Universidad de Sonora Steven Roger Fischer Auckland, NZ Jorge Gómez-Rendón University of Amsterdam Lilián Guerrero Universidad de Sonora
Gertrud Reershemius Aston University, Birmingham Michael Rießler Humboldt-Universität zu Berlin Vincent de Rooij Amsterdam School for Social Science Research, University of Amsterdam Jeanette Sakel University of the West of England Eva Schultze-Berndt University of Manchester Maryam Shabibi University of Manchester
Geoffrey Haig University of Kiel
Uri Tadmor MPI for Evolutionary Anthropology/ Jakarta field station
Ewald Hekking Universidad de Querétaro
Şirin Tufan University of Manchester
Introduction Yaron Matras and Jeanette Sakel
1. Borrowing in cross-linguistic perspective1 Like any metaphor, the term “borrowing” has its drawbacks. We have decided to ignore possible reservations about the term, both in the title of this collection and in the advice on the use of terminology which we have given to the contributors. Whether the “borrowed” substance is perceived as belonging or as alien, whether its source is described as a “donor” and the language into which it is integrated as the “recipient”, “copier”, or “replica”, seems immaterial as long as clarity prevails as to the kind of phenomena that we are addressing when talking about contact-induced change. We use the term “borrowing” as a cover-term for the adoption of a structural feature into a language as a result of some level of bilingualism in the history of the relevant speech community. This collection is about the structural effects of language contact. We have asked each contributor (or pair of contributors) to focus on the diachronic impact that language contact has had on the structure of a particular language. Accompanying these descriptions are comments on societal multilingualism, the roles that are assigned to various languages in the community, patterns of language mixing, and issues of language policy and language education, which are dealt with in relation to each case study in the introductory sections of each chapter. The purpose of the compilation is to be able to compare the effects of different kinds of contact on different kinds of languages, and so to help forward our understanding of universal effects of language contact.
2. Sampling in contact linguistics Linguistic typology tries to make generalizations about human languages. For this purpose, typologists rely on sampling methods. Language samples make it possible to make generalizations without studying each and every individual language, which would be a costly and time-consuming endeavour. Since Greenberg (1966) it has been accepted that samples should try and reflect at least the present-day diversity of languages in order to be truly
2
Yaron Matras and Jeanette Sakel
representative of human language. Most researchers have therefore made an effort to avoid areal or genetic biases when compiling a sample, though basic typological parameters and extralinguistic factors have played less of a role (cf. Comrie 1981, Stassen 1985, Dryer 1989; Rijkhoff et al. 1993). At times samples have been used by a group of studies in a coordinated fashion, to study the distribution of several different phenomena across the same set of languages; this was partly the case in the EUROTYP project; and in the World Atlas of Linguistic Structures – see Comrie, Dryer, Gil and Haspelmath 2005). Nevertheless, it is fair to say that on the whole linguistic samples have been used in order to study a particular structural phenomenon or structural category. The cross-linguistic study of structural borrowing is a challenge at a different level. Firstly, borrowing can affect many different categories. A comprehensive, comparative study of borrowing must therefore take into account both the “horizontal” diversity of sample languages, and the “vertical” diversity of structural categories on which contact can have an impact. This already makes the sampling of borrowing phenomena a much more complex task than the comparative study of any particular domain of linguistic structure. Second, borrowing is a historical dimension, which can only be identified and assessed if diachronic information on the relevant language(s) is available. This factor seriously disadvantages the consideration of entire areas of the world from which we lack secure and reliable information on linguistic diachrony, and so in effect it counteracts the need to maintain areal diversity in a representative sample. Finally, there is general agreement that the outcomes of language contact (or, to be more precise: of widespread bilingualism in a community) depend not just on structural factors, but to a great extent on extralinguistic factors. The duration and intensity of cultural contact, the roles and status of the participating languages, the degree of institutional support awarded to languages in various stages of their history (e.g. the presence of literacy or use in the public, acrolectal domain), and speakers’ attitudes toward their own and their neighbours’ forms of speech – all these play a vital role in determining the direction of change and so in shaping the structural outcome of language contact constellations. In order to investigate the universal possibilities of contact-induced change, one needs to take sociolinguistic factors into account. The ideal sample for the investigation of contact is therefore one that is, like other samples, stratified to take into account various language-genetic groupings, structural types, and regions of the world; but in addition it must also be informed about diachronic depth and allow the author or user to control factors that are external to language.
Introduction
3
Given these difficulties of sampling it is not surprising that most attempts to make generalizations about contact-induced change have been based on casual observations, rather than on systematic comparative studies. This is true of Moravcsik’s (1978) discussion of borrowing universals even within the context of the Greenbergian project, as well as of Thomason and Kaufman’s (1988) frequently-cited borrowing scale. Some generalizations about borrowing have been proposed with reference to a case study of just one single contact situation (cf. Haugen 1950, van Hout and Muysken 1994, Ross 2001, Field 2002), while some have concentrated on identifying counter-examples to generalizations proposed by others (cf. Cambpell 1993, cf. also Thomason 2001). To the extent that samples have been used in contact linguistics, they have tended to control one of the key factors in the contact situation, such as the donor or the recipient language, or even the type of category affected. Stolz and Stolz (1996, 1998), for example, discuss the borrowing of Spanish function words into a diverse set of languages in Central America and the Pacific. Johanson (2002) discusses the contact behaviour of Turkic languages, and Matras (2002) and Elšík and Matras (2006) evaluate structural borrowing from a diverse set of European contact languages in the dialects of Romani. It is noteworthy that in these samples the extra-linguistic parameters are also kept constant. Thus, Spanish is the colonial language in the Pacific and in Central America, Romani is an oral language of dispersed, socially marginalized, bilingual ethnic minorites, and Persian has played a similar role in the history of various Turkic languages. A wealth of data for comparison from various contact situations can be found in a number of collections devoted to case studies of language contact (e.g. Gilbers, Nerbonne, and Schaeken 2000, Aikhenvald and Dixon 2001, Matras, McMahon and Vincent 2005). Aikhenvald and Dixon (2006) especially contains contributions that cover a wide range of languages, regions, and contact phenomena. These are accompanied by important summary observations on general factors and constraints that operate in language contact situations (cf. Aikhenvald 2006). Put together, these and other excellent contributions to the study of language contact have taken us a significant step forward toward a typology of contact-induced language change. Still missing, however, from the body of work produced in recent years is an attempt at a systematic comparison of the behaviour of grammatical categories across a sample of languages in contact. Of interest is the question whether some grammatical categories are universally more susceptible to contact-induced change than others. A further question is whether there is any recurring correlation between the borrowing of structures in one cat-
4
Yaron Matras and Jeanette Sakel
egory, and those belonging to another. Both these issues can be expressed in terms of hierarchies of borrowing. These in turn may contain either implicational statements (if X is borrowed, then Y is also borrowed), or just plain frequency statements (X is borrowed more frequently in the sample than Y, and hence it can be said to be more prone to borrowing than Y). Equally of interest is the correlation between a category and the type of contact-related change that is more likely to affect it: a shift in meaning or in the distribution of existing structures (which we term “pattern replication” below), or the actual adoption of a structure from another language for circulation in the recipient system (“matter replication”). Finally, we are interested in the interaction between the contact behaviour of a category, and other factors that condition the nature of the contact situation, including both language internal features (such as the typological parameters of the languages involved) and extra-linguistic features (such as the type of bilingualism and the roles played by the respective languages in various domains of communicative interaction). The purpose of this collection is to facilitate a discussion of questions of this kind, and to provide information on the basis of which these questions can be addressed.
3. The data compilation tool A major difficulty in sampling for the purposes of contact linguistic studies is the accessibility of relevant information. Unlike synchronic structural facts about language, the description of borrowing requires diachronic information. Even if such information is available in principle for some languages, it is not always the case that it is included into grammatical descriptions. Grammar books do not generally tend to highlight borrowings at all. Identifying borrowed structures requires a high degree of expertise and specialization in the language and its history, and familiarity, at the very least, with the languages with which it has been in contact. Relevant extralinguistic information is often missing from grammatical descriptions, too. For these reasons, it is hardly feasible for a lone researcher to survey published descriptions of various languages in order to compile a representative sample corpus of grammatical borrowing. Sampling in this field is best achieved through team effort, with experts contributing first-hand information on contact-induced phenomena. Underlying the team effort on which this volume is based is a uniform questionnaire, formatted as a user-friendly database (in FilemakerPro 67).
Introduction
5
The aim of the questionnaire was to obtain a representative and comparable sample of data on contact-induced change in a variety of languages. For this purpose it was distributed to the contributors as a detailed reference grid. The questionnaire can of course continue to constitute a description standard for language contact phenomena, serving as a checklist for information to be covered in an exhaustive description of borrowing into any given language. The questionnaire opens with information on relevant metadata (source of information, affiliation to sub-samples, date of input, and so on), and continues to cover extralinguistic information about the language and the speech community. The remaining chapters cover all principal domains of structure: Phonology, Typology (a characterization of principal typological traits), Nominal structures, Verbal structures, Other parts of speech (e.g. quantifiers, indefinites, phasal adverbs, discourse markers and connectors), Constituent order, Syntax (clause combining), and Lexicon (general information on the presence of lexical loans in various semantic domains, as well as specific
Figure 1. Information page of the Language Convergence database (entry: Domari) showing open chapter menu in the top left corner
6
Yaron Matras and Jeanette Sakel
Figure 2. Encoding the sociolinguistic situation (Mosetén)
questions on expressions of time and space). Using the “Layout” function in FilemakerPro, each chapter is displayed on a separate page, accessible through a menu box (Figure 1). Individual records, each representing a language in contact, can be tagged for different kinds of contact constellations or sub-samples. We can distinguish, for instance, borrowing situations where just two languages are in contact, from observations on a more widespread regional distribution of shared phenomena, or “linguistic areas”. Special attention is given to the coding of a series of extra-linguistic indicators (Figure 2), allowing the user to assess the correlations between sociolinguistic factors and the contact behaviour of a language. Depth of contact is taken into account by distinguishing, where applicable, several different layers of contact (see also Matras 1998): The Current contact language (i.e. the object of widespread bilingualism), a Recent contact language (that may still be spoken by an older generation of speakers), and an Old contact language that has made an impact on the
Introduction
7
Figure 3. Coding of contact languages (Manange)
language in the past, but has little or no contemporary role in the speech community (Figure 3). In order to be able to investigate the precise effects of contact on structural compositions, a distinction is maintained throughout the questionnaire between the replication of linguistic matter (MAT) consisting of actual phonological segments, and the replication of patterns (PAT), which pertains to the semantic and grammatical meaning and the distribution of a construction or structure (see Matras and Sakel 2007). This distinction is encoded alongside every relevant description of a contact phenomenon in an individual category (Figure 4). The advantages of working with the questionnaire database are obvious: While the checklist ensures uniform and comprehensive coverage of the same phenomena, and so comparability throughout the sample, the database allows to filter and to query the results, to retrieve examples of the various kinds of contact phenomena, and to view correlations among the data sets (see e.g. Figure 5).
8
Yaron Matras and Jeanette Sakel
Figure 4. Encoding MAT (matter) and PAT (pattern) replications (Domari)
A sample comparison – piloting just two languages, Kelderash Romani and Mosetén – was already presented, based on the database questionnaire, in Sakel and Matras (2007). The present collection features a case-by-case assessment of borrowing. A preliminary assessment of some salient, common patterns is provided in the two evaluation chapters, by Matras and by Sakel. Our intention at this point is to continue to expand the data sample, and eventually to make the data accessible to users online.
4. Coverage of phenomena and languages In assessing the diachronic impact of contact, many of the contributors faced the dilemma of how to tell apart the ongoing effects of current bilingualism
Introduction
9
within a speech community, and structural changes that may be regarded as permanent. The distinction is often referred to as one between “borrowing” and “code-switching”, but since there is widespread agreement that alternational code-switching is unlikely to fossilize into a permanent pattern (see Backus 2003, 2005), it is in fact the distinction between insertional switches and borrowing that requires clarification. Our recommendation to the contributors was generally to regard as borrowings those cases where a speaker’s selection of a particular structure is not facilitated by a choice between two alternatives, and is not triggered by the speech situation or by the topic of conversation, nor by a need for a special conversation effect. We thus consider as borrowings all structures that appear with a certain regularity and are used by different speakers in a range of different situations. Problem cases are otherwise highlighted in the individual chapters.
Figure 5. Comparing results on the presence/absence of contact phenomena for the following categories: Connectors (O_PA_co), Focus particles (O_PA_fo), Numerals and quantifiers (O_nu), and Personal pronouns (O_P_pp)
10
Yaron Matras and Jeanette Sakel
Even when drawing on first-hand data and the expertise of field linguists, gaps in the coverage are inevitable; in our case this is due mainly to limitations of space, resources, and the constraints of the production schedule. The sample nevertheless grants representation to most areas of the world: Saharan and sub-Saharan Africa (Tasawaq, K’abeena, Likpe, Katanga Swahili), the Middle East (Khuzistani Arabic, Domari, Kurmanji-Kurdish, Western and North-eastern Neo-Aramaic), the Balkans (Macedonian Turkish), Europe (Kildin Saami, Yiddish, Hungarian Rumungro), the Himalaya (Manange), South Asia (Indonesian, Biak), East Asia (Vietnamese), Australia (Jaminjung), the Pacific (Rapanui), Central America (Nahuatl, Yaqui, Otomi, Purepecha), and South America (Imbabura Quichua, Guaraní, Hup, Mosetén). It contains languages with a tradition of native literacy (Vietnamese, Indonesian, Arabic, Turkish, Swahili, and arguably also Neo-Aramaic) and others without one; languages of ethnic minorities, those that are or were majority languages, and regional languages in post-colonial settings; languages with a single contemporary contact language as well as those spoken in either a multilingual setting or a linguistic area. Lacking representation in our sample are languages of North America, Central Asia, and Siberia. Pidgins, creoles, and mixed languages were not considered either, since their borrowing characteristics are potentially different in principle than those of other languages that are not themselves the product of recent contacts (but see Thomason 1997, Matras and Bakker 2003). The present volume appears alongside a two-part publication devoted to lexical borrowing, which is the result of the Loanword Typology project based at the Max Planck Institute for Evolutionary Anthropology in Leipzig. We hope that together, both publications will help shed new light on an ancient process that will have been as instrumental in shaping the development of the world’s languages as human contacts has been in shaping general human cultural experience. Note 1. We gratefully acknowledge support from the Arts and Humanities Research Council for a three-year research project on “Language Convergence and Linguistic Areas” (award number RG/AN4725/APN16320), during which the questionnaire and database tools used for this collection were developed, as well as support from the School of Languages, Linguistics and Cultures for a workshop on “Grammatical Borrowing in Cross-Linguistic Perspective” in Manchester in September 2005 and for the production of the present volume. We also wish to
Introduction
11
thank Georg Bossong for helpful comments, Peter Kahrel for typesetting and copy-editing the manuscript, and Ursula Kleinhenz for her support during the production process.
References Aikhenvald, Alexandra Y., and R. M. W. Dixon (eds.) 2001 Areal Diffusion and Genetic inheritance. Oxford: Oxford University Press. 2006 Grammars in Contact: A Cross-Linguistic Typology. Oxford: Oxford University Press. Aikhenvald, Alexandra Y. 2006 Grammars in contact: A cross-linguistic perspective. In: Alexandra Y. Aikhenvald and R. M. W. Dixon (eds.), Grammars in Contact: A Cross-Linguistic Typology, 166. Oxford: Oxford University Press. Backus, Ad 2003 Can a mixed language be conventionalized alternational codeswitching? In: Y. Matras and P. Bakker (eds.), The Mixed Language Debate: Theoretical and Empirical Advances, 237270. Berlin: Mouton de Gruyter. 2005 Codeswitching and language change: One thing leads to another? International Journal of Bilingualism 9 (3/4): 307340. Campbell, Lyle 1993 On proposed universals of grammatical borrowing. In: H. Aertsen and R. Jeffers (eds), Historical Linguistics 1989: Papers from the 9th International Conference on Historical Linguistics, 91109. Amsterdam: John Benjamins. Comrie, Bernard 1989 Language Universals and Linguistic Typology: Syntax and Morphology. Oxford: Blackwell. Dryer, Matthew S. 1989 Large linguistic areas and language sampling. Studies in Language 13: 257292. Elšík, Viktor, and Yaron Matras 2006 Markedness and Language Change: The Romani Sample. Berlin: Mouton de Gruyter. Field, Fredric 2002 Linguistic Borrowing in Bilingual Contexts. Amsterdam: Benjamins. Gilbers, D. G., J. Nerbonne, and J. Schaeken (eds.) 2000 Languages in Contact. Amsterdam: Rodopi. Greenberg, Joseph H. (ed.) 1966 Universals of Language. Cambridge, MA: MIT Press.
12
Yaron Matras and Jeanette Sakel
Haugen, Einar 1950 The analysis of linguistic borrowing. Language 26 (2): 210231. Haspelmath, Martin, Matthew Dryer, David Gil, and Bernard Comrie (eds.) 2005 The World Atlas of Language Structures. Oxford: Oxford University Press. Johanson, Lars 2002 Structural Factors in Turkic Language Contacts. London: Curzon. Matras, Yaron 1998 Utterance modifiers and universals of grammatical borrowing. Linguistics 36 (2): 281331. 2002 Romani: A Linguistic Introduction. Cambridge: Cambridge University Press. Matras Yaron, and Peter Bakker (eds.) 2003 The Mixed Language Debate. Berlin: Mouton de Gruyter. Matras, Yaron, and Jeanette Sakel 2007 Investigating the mechanisms of pattern replication in language convergence. Studies in Language 31 (4): 829865. Matras, Yaron, April McMahon, and Nigel Vincent (eds.) 2006 Linguistic Areas: Convergence in Historical and Typological Perspective. Houndmills: Palgrave. Moravcsik, Edith 1978 Language contact. In: Joseph H. Greenberg, Charles A. Ferguson, and Edith A. Moravscik (eds.), Universals of Human Language, Volume 1, 93122. Stanford: Stanford University Press. Rijkhoff, Jan, Dik Bakker, Kees Hengeveld, and Peter Kahrel 1993 A method of language sampling. Studies in Language 17: 169203. Ross, Malcolm 2001 Contact-induced change in Oceanic languages in north-west Melanesia. In: Alexandra Aikhenvald and R. M. W. Dixon (eds.), Areal Diffusion and Genetic Inheritance: Problems in Comparative Linguistics, 134166. Oxford: Oxford University Press. Sakel, Jeanette, and Yaron Matras 2007 Modelling contact-induced change in grammar. In: Thomas Stolz et al. (eds.), Aspects of Language Contact: New Theoretical, Methodological and Empirical Findings With Special Focus on Romanisation Processes. Berlin: Mouton de Gruyter. Stassen, Leon 1985 Comparison and Universal Grammar. Oxford: Blackwell. Stolz, Christel, and Thomas Stolz 1996 Funktionswortentlehnung in Mesoamerika, Spanisch–Amerindischer Sprachkontakt. In Sprachtypologie und Universalienforschung 49 (1): 86123.
Introduction 1997
13
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]: 177. Thomason, Sarah G. (ed.) 1997 Contact Languages: A Wider Perspective. Amsterdam: Benjamins. Thomason, Sarah G. 2001 Language Contact: An Introduction. Edinburgh: Edinburgh University Press. Thomason, Sarah, and Terrence Kaufman 1988 Language Contact, Creolization and Genetic Linguistics. Berkeley: University of California Press. van Hout, Roeland, and Pieter Muysken 1994 Modelling lexical borrowability. Language Variation and Change 6: 3962. Weinreich, Uriel 1966 Languages in Contact: Findings and Problems. The Hague: Mouton (first publ. 1953).
Types of loan: Matter and pattern Jeanette Sakel
1. Introduction1 A central concern of contact linguistics has long been to categorize the ways in which elements are borrowed from one language into another. For this purpose Matras and Sakel (2004) introduced the terms matter (MAT) and pattern (PAT) in the questionnaire on which the sample of contact situations in this book is based (cf. also subsequent publications on the issue, such as Matras and Sakel 2007). In the present chapter I will re-visit the definition of MAT and PAT, as well as address what this distinction could mean in phonology. I will furthermore give an overview of the overall distribution of MAT/PAT in the languages of the sample in order to address the validity of a MAT/PAT distinction in the categorization of contact situations.
2. Definitions MAT and PAT denote the two basic ways in which elements can be borrowed from one language into another. We speak of MAT-borrowing when morphological material and its phonological shape from one language is replicated in another language. PAT describes the case where only the patterns of the other language are replicated, i.e. the organization, distribution and mapping of grammatical or semantic meaning, while the form itself is not borrowed. In many cases of MAT-borrowing, also the function of the borrowed element is taken over, that is MAT and PAT are combined.2 In other instances, MAT and/ or PAT are borrowed, but deviate considerably in their form or function from their original source. In some categories, making a distinction between MAT and PAT does not make much sense. For example, word-order changes will invariably be PAT. In other areas, such as phonology, the MAT/PAT distinction applies only in a restricted way, as MAT and PAT are primarily defined as functioning above the morpheme level. The concept behind MAT and PAT is well-grounded in the literature, but only rarely figures in the categorization of contact situations. One exception is Heath (1984: 367), who bases his approach to language contact on this opposition, distinguishing between
16
Jeanette Sakel
“direct transfer of forms from the other language” and “structural convergence”. Other approaches to language contact mention similar distinctions, often with very different terminology, such as Haugen’s (1950) “importation” for outright borrowing and “substitution” for loanshifts or calques. Likewise, Weinreich ([1953] 1966: 7) speaks about “transfer of elements” and “interference without outright transfer”, which re-appears in Weinreich’s distinction between “source” and “recipient language” for MAT-borrowings, and “model” and “replica language” for PAT-borrowings. The distinction is mentioned in many subsequent approaches to language contact, such as Gołąb’s (1956, 1959) “form” versus “substance”, Johanson’s (1992) “global copying” and “partial copying”, Nau’s (1995) “material borrowing” and “loanmeaning, loan-translation” and Treffers-Daller and Mougeon’s (2005: 95) “borrowing, code-switching” versus “transfer”. Many other studies of language contact pay only minor attention to this matter, such as Thomason and Kaufman (1988), who do not include this classification in their influential borrowing scale (1988: 74). Some approaches to contact focus almost exclusively on either MAT or PAT. In this manner, approaches to substrate influence and contact-induced grammaticalization (Siegel 1997; Keesing 1991; Heine and Kuteva 2005) focus mainly on PAT-type loans. Likewise, linguistic areas have often been described as zones in which PAT-borrowing appears. On the other hand, much of the early literature on code-switching primarily addresses contact phenomena of the MAT type, while more recently PAT has been integrated into frameworks of code-switching (cf. Savić 1995, Bolonyai 1998 and MyersScotton 2002: 2122). There are reasons why some approaches seem to favor either MAT or PAT: when studying linguistic areas or substrate influence, focus is often on PAT because a major part of the loans in these situations are of this type and indeed areas and substrate influence are often defined as displaying mainly pattern-loans. On the other hand, when stuyding situations of code-switching, MAT is often very prominent.
3. Integration of MAT and PAT loans Let us now look at how grammatical MAT/PAT-loans are integrated into the recipient language. In many cases of MAT or PAT-borrowing, not the entire function or form is taken over, but the borrowed elements differ from their original source. Take for example the way in which Domari copies Arabic aspect marking (Matras, this volume); Domari borrows Arabic auxiliaries as
Types of loan: Matter and pattern
17
MAT, but it does not simultaneously make use of a subjunctive verb form as in Arabic because it has its own subjunctive. Hence, not the entire construction of aspect marking is taken over, but only parts of it. In Otomi the functions of a borrowed form are extended: the shortened form ko from Spanish como is used in a number of constructions, also expressing ‘made of’, a function it does not have in Spanish (Hekking and Bakker, this volume). Hence, when MAT-elements are borrowed, their functions are not necessarily the same as in the source language; sometimes only parts of the function are borrowed, sometimes the functions are extended and the loans are rarely mere copies of those of their counterparts in the source language. Also the forms of loans are frequently adjusted, for example by phonological integration of MAT-loans into the recipient language, which in some cases makes them difficult to identify as loans without a careful analysis. In the same way, PAT-loans inherently involve a process of grammaticalization, often leading to different patterns as those in the source language (cf. Heine and Kuteva 2005). PAT-borrowing is facilitated by a pivot common to both languages (Matras and Sakel 2007). Such cases of PAT-adjustment in the sample include the shift in grammatical meaning of native Yiddish elements to correspond to Slavic aspectual markers (Reershemius, this volume).
4. MAT and PAT in phonology How could the distinction between MAT and PAT-loans be employed in phonology? In most contact situations, MAT-loans involve phonological changes that can go in two directions: (1) MAT-borrowed element is phonologically integrated into the recipient language; e.g. Mosetén ishkweera for Spanish escuela ‘school’. (2) MAT-borrowed element is not integrated and may introduce new phonemes into the recipient language; e.g. in Jaminjung, where loan-phonemes have risen to phoneme status within the language and are now used with native words (Schultze-Berndt, this volume). Only the latter strategy, (2), involves phonological borrowing in which elements from the source language are transferred into the recipient language. One could argue that loans in (2) are MAT-loans if they introduce a new phone, and PAT-loans if they introduce a new phoneme. The loss of certain phonological distinctions would possibly be a case of PAT, since no new material
18
Jeanette Sakel
is introduced, and a change in the stress patterns could be counted as PAT as well, since it involves the overall patterning rather than actual forms. A change in tone, on the other hand, could be MAT-borrowing if it occurred in isolated loanwords and PAT if it affected the whole language. A cautious attempt to classify the phonological contact phenomena found in the sample is the following: – Borrowing of individual phonemes that are also used in native elements is found in Vietnamese, Indonesian, Jaminjung and Paraguayan Guaraní. This type of loan would probably best be classified as MAT, since it involves borrowing of phonological material without a major disruption of the phonological system. – Borrowing of a series of distinctions is common. K’abeena, Rumungro and Kildin Saami have borrowed a palatal series. K’abeena has furthermore borrowed ejectives, while Tasawaq gained pharyngealization and Vietnamese had a retroflex series introduced by contact. These would be MAT-loans in the sense that elements are borrowed and PAT-loans in the sense that a whole series of phonemes within the phonological system is affected by the contact. Adaptation of stress, syllable structure, prosody or tone systems would classify as PAT-borrowing in phonology since they affect the system, rather than individual elements. Stress patterns are adjusted in Western Neo-Aramaic, Yiddish, Hup, Manange, North-eastern Neo-Aramaic and Paraguayan Guaraní. The syllable structure is adapted due to contact in Yiddish and Indonesian, allowing consonant clusters. Prosody is affected in Hup and Jaminjung, while tone has undergone contact-induced changes in Vietnamese and Manange. Hence, MAT and PAT could be used to classify contact phenomena in the phonology, even though the strength in having this MAT/PAT distinction makes most sense in loans from the morpheme-level upwards.3
5. The MAT/PAT distinction in grammar: findings from the sample We will discuss the data from the sample by looking at the following cases: (1) situations with overall MAT-borrowing; (2) situations with overall PATborrowing; and (3) hierarchical relations between contact languages. When talking of “overall” contact phenomena, we deliberately do not quantify loans
Types of loan: Matter and pattern
19
in contact situations to avoid sampling problems, but rather only look at cases where the overwhelming majority of the loans are of either type.
5.1. Situations with overall MAT-borrowing In the sample, Jaminjung, Biak and Vietnamese stand out in having predominantly MAT-loans. Jaminjung is in contact with Kriol, a creole which is likely to have been influenced by the native languages of the area (cf. Schultze-Berndt, this volume). The patterns of Kriol are in many ways similar to those of Jaminjung possibly due to substrate influence from the languages of the area on Kriol during its development. For this reason, Schultze-Berndt chose not to include PAT-loans in her overview to avoid circularity and in this particular case the explanation for the predominance of MAT-loans is hence merely methodological. Biak, on the other hand seems to have a preference for MAT-loans. Biak word order is highly fixed, which means that in many cases MAT-loans are easier to integrate than a change in word order or re-modelling of existing material (cf. van den Heuvel, this volume). For example due to Indonesian pressure negation is expressed by a sentence-initial adverb. A native element in Biak could have been adjusted as a PAT-loan, but the pressure against the change in word order this would involve was higher than that of borrowing new material. Vietnamese has had substantial contact influence from Chinese. While most other situations involve bilingualism and language contact in the spoken language, Chinese influence on Vietnamese has predominantly been through written materials. The types of loans encountered are mostly MAT. This can be attributed to the type of contact situation, as without oral bilingualism pattern copying is difficult, and MAT-loans prevail. This rather trivial fact makes sense in the current discussion since it is directly related to the types of loans encountered. MAT-loans can appear even in cases of monolingualism, such as in the Muran language Pirahã (not discussed further in this volume), which has exclusively MAT-loans from Portuguese due to very rudimentary bilingualism. Concluding, there are both structural reasons for MAT-loans, as in Biak, where PAT is not preferred as it would lead to changes of a highly fixed word order, as well as reasons found in the type of contact situation linked to the degree of oral bilingualism, as in the case of Vietnamese.
20
Jeanette Sakel
5.2. Situations with overall PAT-borrowing Likewise, a number of languages in the sample show an overall majority of PAT-borrowing. These are K’abeena, Hup, Macedonian Turkish and Khuzistani Arabic, most of which are also part of well-established linguistic areas. K’abeena is discussed for its participation in the Gurage linguistic area (cf. Crass, this volume) and exhibits areal patterns of varying sources. Not surprisingly, most of the contact phenomena found in K’abeena are PATloans. K’abeena has a number of MAT-loans as well, part from lexical elements, it has borrowed a few markers of temporal and causal clauses, discourse markers and interjections. These belong to the category of function words, comprising conjunctions, discourse markers and other elements detached from the main proposition of the clause (cf. Matras 1998). Hup is part of the Vaupés are in the North-Western Amazon. Epps (this volume) focuses on the contact situation between two of the languages from this area, Hup and Tukano. Hup borrows a vast number of PAT-features from Tukano, while there are only few MAT-loans. The reasons for this are strong cultural restrictions in the Vaupés region against MAT-loans. These can nonetheless be overridden, as is the case in MAT-loans from Tukano, including an adverbial particle ‘until’, a negative emphasis marker and a disjunction marker. Again, these belong to the category of function words and the only MAT-loans outside this category in Hup are numerals.4 Macedonian Turkish, which is spoken in the Balkan linguistic area and displays a large number of areal features that can be established when comparing the language to its neighbourng Balkan languages and Standard Turkish. While most loans are PAT, some of the discourse markers, as well as the phrasal conjunctions ama ‘or’ and i ‘and’ are MAT-loans from Macedonian. Matras and Tufan (this volume) draw the conclusions for Macedonian Turkish that MAT structures are probably new additions since the change in hierarchical structures between the languages, while PAT-borrowing must have occurred earlier as a means of “utterance-organizing strategies”. A similar picture emerges in Khuzistani Arabic, which has a majority of PAT-loans, and again the only MAT-loans are found in categories such as discourse markers, fillers, tags and focus particles where they have an interaction-qualifying function (cf. Matras and Shabibi, this volume). The reason why Khuzistani Arabic has many PAT-loans could be its status in the contact situation, which is similar to that of Macedonian Turkish in that both languages are spoken across the border from areas where they are majority languages. The variants discussed here are dominated, but the speakers are
Types of loan: Matter and pattern
21
frequently exposed to the standard languages, for example through television and visits. Speakers may have a wish to keep the two languages apart as they attribute equally high status to them in the different contexts of use. This restricts MAT-loans and leads to mainly PAT-loans, while in the category of function words the two systems collide, very similar to what we found in Hup and K’abeena. The fact MAT-loans of function words are frequent in these languages appears to confirm Matras’s (1998) hypothesis that bilingual speakers will have greater difficulties keeping their linguistic systems apart around such markers than for other domains of functions.
5.3. Relations between languages and MAT/PAT borrowing Let us now look at the data from another perspective, namely whether the relations between the languages in contact have an impact on the types of loans encountered. In the sample, most borrowing appears from hierarchically higher – or dominant – languages into lower, dominated languages. Dominance is here broadly defined: a language is dominant when used for administration, as a lingua franca, and when it has to be learnt by the speakers of the dominated language, which in return is usually not used for any of the above or which is used in less official environments. A language can be dominant in one contact situation, while dominated in another. For example, Katanga Swahili is dominant in being the lingua franca of the region but dominated by French, which is the official and administrative language of the country. Apart from one-to-one borrowing situations, some of the languages in the sample belong to well-established linguistic areas. Table 1 summarizes the contact situations treated by authors of the chapters. Many of the languages in the sample are also part of linguistic areas, but where this is not immediately discussed in the chapters I left it out of the current discussion. X marks the main focus of the contribution to this volume, Y marks another contact situation which is not the main focus of the chapter. We can usually see layers of different types of contact in languages that belong to more than one contact situation. This is visible in the first group of languages in table one, i.e. languages that are dominant in some situations but dominated in others. In this way, Katanga Swahili has only few MAT-loans, most of which are of French numbers and discourse markers. The loans from the substrate languages are PAT, apart from three MAT noun-class markers. If we would expect substrate-influence only leading to
22
Jeanette Sakel
Table 1. Types of contact situation and hierarchical status in contact situation dominated 1 Indonesian, Katanga Swahili 2 K’abeena 3 Biak, Domari, Imbabura Quichua, Khuzistani Arabic, Kurmanji Kurdish, Manange, Mosetén, Northeastern Neo-Aramaic, Otomi, Paraguayan Guaraní, Purepecha, Rapanui, Saami, Tasawaq, Western Neo-Aramaic, Vietnamese, Yaqui, Yiddish 4 Hup, Jaminjung, Likpe, Nahuatl, Rumungro, Macedonian Turkish
areas (treated here)
X
dominant Y
X X
X
Y
PAT-borrowing, this would be exceptional. We can see reasons for why these markers have been borrowed in this way by looking at the system. De Rooij (this volume) shows that many of the Swahili noun class markers correspond in form and function to those of the substrate languages. The differences – in particular the three markers in question – were borrowed to assimilate Katanga Swahili’s system of noun classification to that of the substrate languages. We are therefore not dealing with a mere MAT-loan of three markers in isolation, but with a general adjustment of two systems that are already largely identical. In the same way we find MAT-loans from current substrate languages in Indonesian. Indonesian is the dominant language in contact with other languages spoken in Indonesia, such as Javanese but it has also been dominated by a range of other languages, in particular Sanskrit, Chinese and some European languages. The loans from the dominant languages are both MAT and PAT and have, for example, led to a number of changes in the sound system of Indonesian. The influence from substrate languages is astonishingly likewise MAT and PAT. MAT-loans from Javanese include collective particles, a third-person plural pronoun, some interrogative markers, a focus marker and modal particles. If Javanese was a substrate language at the time of borrowing, these MAT-loans would be exceptional in the light of general preconceptions that substrate-influence is mainly PAT. However, we can explain there MAT-loans from Javanese: being a substrate language
Types of loan: Matter and pattern
23
today, it used to have a different status in the past, described by Tadmor as a “quasi-symbiosis” between Indonesian and Javanese. The MAT-loans may have entered Indonesian during this time. The only language in the second group, K’abeena, has been discussed in the previous section for its prevalence of PAT-loans. When looking at the few attested MAT-loans, these are also from languages that take part in the area: (Ethio-) Semitic lexical elements are used as adverbial clause markers (time, reason, cause), an adjectival suffix is borrowed form Amharic and a number of interjections have the same forms as in other languages of the area. Apart from the adjective suffix these fall under the category of function words. Even though this is only one example, it shows that MAT-loans in the category of function words do appear in linguistic areas.5 The majority of the languages in the sample are in a contact situation with a dominant language. These are listed as group 3 in Table 1. In these languages both MAT and PAT loans are common in most categories, while there is heavy borrowing of function words, which are overwhelmingly borrowed as MAT. The contact phenomena encountered in these languages are in general very similar, independent of whether the languages are spoken by small minorities or whether they are major national languages. In this way, the major South American indigenous languages Paraguayan Guaraní and Imbabura Quichua (cf. Rendon, this volume) borrow heavily from Spanish and have comparable contact phenomena to the small indigenous languages Mosetén (cf. Sakel, this volume) and Purepecha (cf. Chamoreau, this volume). Stolz and Stolz (1996, 1997) come to similar conclusions comparing a range of languages in contact with Spanish. They argue that the similarities in contact phenomena are found within the same categories. Take the following example: most languages in contact with Spanish borrow an element out of the category of temporal adverbial clause markers, such as hasta ‘until’, desde ‘from’, cuando ‘when’, but which ones of these markers are borrowed, and which functions they assume in the recipient language, is language-specific (cf. Stolz and Stolz 1996, cf. also Sakel’s 2007 description for Mosetén). Similarities in contact phenomena have also been attested for languages in contact with Russian as the dominant language. Rießler (this volume) discusses how Russian contact phenomena in Kildin Saami resemble those in other contact situations with Russian as the dominant language, referring to Majtinskaja (19789). The reasons for this have to do with universals of borrowing, rather than individual languages. Comparing the contact phenomena in the sample, we find that similar principles apply for all contact situations of the same type, independent of the source language. Indeed, when compar-
24
Jeanette Sakel
ing the situations where Spanish is dominant to those where Russian or other languages are dominant, elements within the same categories are taken over, and often these are MAT-loans of function words. This also includes dominant languages that are not Indo-European, such as Arabic in contact with Western Neo-Aramaic. These situations are similar due to general principles of contact, rather than language-specific ones.6 The built-up of the system and availability of certain structures in Spanish and Russian may nonetheless play a role in a few cases, since they have similar typological profiles. The final group of languages in Table 1 comprises Hup, Jaminjung, Likpe, Nahuatl, Otomi, Rumungro and Macedonian Turkish. For these languages the authors of the individual chapters discuss different layers of contact in both one-to-one borrowing situations and linguistic areas. The general results are that the languages display PAT-loans from the areas they are in, as well as MAT/PAT loans from their particular one-to-one borrowing situations. In some cases area and borrowing situations overlap, as in the case of Hup in contact with Tukano, both of which are part of the Vaupés area. In other cases, there are clear differences between the contact influence from the area and the borrowing situation, as in Nahuatl, which is part of the Meso-American area (as proposed by Campbell et al. 1986), as well as being part of a one-to-one borrowing situation with Spanish. Comparing the contact phenomena in the latter case between a one-to-one borrowing situation and an area, these are very different: the areal phenomena are predominantly PAT, while Spanish loans are very similar to other situations in which indigenous languages of the Americas are in contact with Spanish as the dominant language and include many MAT-loans. Concluding, MAT-loans appear in dominated languages in one-to-one borrowing situations, but also PAT-loans are frequent in this cases. While linguistic areas and situations of substrate influence display mainly PAT-loans, MAT-loans are very frequent in the category of function words and indeed appear in all languages of the sample (cf. also Matras’s chapter on borrowability of categories, this volume). This suggests that function words are borrowed easily and relatively early on in contact situations. We can see this in layers of contact, for example, in the variant of North-eastern Neo-Aramaic (cf. Khan, this volume) whose speakers have all immigrated to Israel. The original contact situation with Kurdish as the dominant language has been replaced by the new situation with Hebrew as the new dominant language. Indeed, the loans from Hebrew are primarily found in the lexicon and function words. More evidence for this comes from Romani (Elšík, this volume, cf. also Elšík and Matras 2006). Also other sample languages show early borrowing of function
Types of loan: Matter and pattern
25
words: all of the sample languages show some degree of MAT-borrowing, even in those languages where all or most other loans are PAT, such as Hup and K’abeena.7 In many cases MAT-loans of function words have been on a long journey, being borrowed from one language into another. For example, some Kurdish discourse markers and connectors are of Arabic origin and have entered the language through contact with Turkish (Haig, this volume). Likewise, some Hup function words of Portuguese origin have most likely entered the language via Tukano (Epps, this volume).
6. Conclusion We have shown that the distinction between MAT and PAT is very useful for classifying contact phenomena. For contact theory in general this means that the distinction between MAT and PAT-borrowing should be included in attempts to categorize contact situations. We can conclude the following for the regularities behind MAT/PAT borrowing: We have found that hierarchical relations between languages have an impact on the types of loans encountered. For example Arabic is a dominant language in some contact situations and a dominated language in others: it is dominant in contact with Domari and Western Neo-Aramaic, while Khuzistani-Arabic is dominated by Persian. In the first case, the dominant languages incorporate many MAT-loans from Arabic, while in the second case, MATloans are from Persian.8 Hence, the direction of types of contact phenomena (in this case MAT-loans) depends on the hierarchy between the languages. Furthermore, when a dominant language has high status, MAT-loans into a dominated language are often easily accepted. For example many MAT-loans from Spanish appear in Purepecha, Nahuatl and Yaqui, for which Spanish is a highly dominant and high-status language. In the same way, MAT-loans can be shunned from contact situations, such as in areas with social constraints against pattern-replication in the Vaupés, as discussed for Hup above, and also in other areas such as parts of Papua New Guinea (Ross 1996). Also the degree of bilingualism plays a role in the way elements are borrowed. Without bilingualism, patterns are usually not copied and MAT is only borrowed in a restricted sense. The type of contact influence is likewise closely related to degree of (oral) bilingualism. For example Vietnamese has experienced massive contact influence from Chinese, though this was mainly through written materials and rarely through oral transmission involving bilingualism, leading to a majority of MAT-loans.
26
Jeanette Sakel
All languages of the sample display MAT-loans of function words, even those that have a majority of PAT-loans. This suggests that reasons for borrowing function words as MAT are stronger than other constraints in contact situations. Such reasons include the detached nature of function words, which makes them easily borrowable, as well as their function as discourse structuring devices (cf. Matras 1998). It could furthermore point at a general change of contact situations across the world towards situations with one dominant language that is used in administration, television and general communication in a highly mobile, globalized world. This does not mean that dominant languages did not exist in the past, cf. the influence of Chinese on Vietnamese or, for instance, Sanskrit influence on Indonesian. In those cases, however, contact was prevailing in some communities – among the elite, for example, while in other communities or at other times contact was less intensive. In today’s globalized world, however, some languages, often former colonizers’ languages, have impact on entire speech communities. Television, schooling, trade, mobility, indigenous organizations, health, developmental organizations and other ways of communicating with the “outside” world have lead to the rise of these already highly dominant, major languages through increased bilingualism. Many contact situations in the sample show this tendency. For example in Hup, area-internal restrictions such as taboos against MAT-borrowing have been overridden by overwhelming influence of Tukano and eventually Portuguese as administrative languages (cf. also discussion by Aikhenvald 2002).
Notes 1. I would like to thank Yaron Matras and Kristine Hildebrandt for comments on earlier versions of this chapter. 2. MAT-borrowing without any PAT will not be discussed in this chapter as it is very rare and mainly occurs in the lexicon; i.e. usually MAT is taken over with at least part of its original PAT. An example of MAT-only borrowing in the lexicon is the noun handy ‘mobile phone’ in German, which does not have this meaning in the source language English. 3. Yaron Matras made the useful comment that just like the shape of morphemes can combine with a certain meaning and appear in a certain organization pattern, so can a phone acquire meaningfulness as a phoneme, and combine with certain patterns of prosody, tone, or permissible combinations of sounds. Rather than define a perfect match for MAT and PAT in phonology, we can simply remain conscious of the layered structure of phonological representation and the
Types of loan: Matter and pattern
27
fact that contact-related change may affect one level without affecting another. 4. Numerals are very frequently borrowed as MAT in many contact situations. 5. One could speculate that the reasons for this could be temporal dominance of the source language. 6. These principle are anchored in the role of categories in processing discourse, and the way that operating in a bilingual setting influences language processing; see contribution by Matras. 7. We find similar loans in Indonesian, where some function words are taken from dominant languages in the history, e.g. Sanskrit, Arabic, Creole Portuguese and Persian. Other function words in Indonesian were borrowed as content words and then grammaticalized, such as ‘and’ from the Sanskrit word for ‘company’. 8. There are mainly PAT-loans in Khuzistani Arabic for reasons discussed above. Also, Arabic is still a dominant language in some religious contexts in Iran, which skews the picture somewhat.
References Aikhenvald, Alexandra Y. 2002 Language contact in Amazonia. Oxford: Oxford University Press. Bolonyai, Agnes 1998 In-between languages: Language shift/maintenance in childhood bilingualism. International Journal of Bilingualism 2 (1): 2143. Campbell, Lyle, Terrence Kaufman, and Thomas C. Smith-Stark 1986 Meso-America as a linguistic area. Language 62 (3): 530570. Elšík, Viktor, and Yaron Matras 2006 Markedness and Language Change: The Romani sample. Berlin: Mouton de Gruyter. Gołąb, Zbigniew 1956 The concept of isogrammatism. Buletin Polskiego Towarzystwa Jezykoznawczego 15: 112. 1959 Some Arumanian–Macedonian isogrammatisms and the social background of their development. Word 15 (3): 415435. Haugen, Einar 1950 The analysis of linguistic borrowing. Language 26 (2): 210231. Heath, Jeffrey 1984 Language contact and language change. Annual Review of Anthropology 13: 367384. Heine, Bernd, and Tania Kuteva 2005 Language Contact and Grammatical Change. Cambridge: Cambridge University Press.
28
Jeanette Sakel
Johanson, Lars 1992 Strukturelle Faktoren in türkischen Sprachkontakten. Stuttgart: Franz Steiner Verlag. Keesing, Roger M. 1991 Substrates, calquing and grammaticalization in Melanesian Pidgin. In: Elizabeth C. Traugott and Bernd Heine (eds.), Approaches to Grammaticalization Vol. 1: Focus on theoretical and methodological issues, 315342. Amsterdam: John Benjamins. Majtinskaja, K. E. 19789 Zaimstvovannye elementy, ispol’zuemye v finno-ugorskih jazykah pri obrazovanii form naklonenij [Borrowed elements, used in inflectional forms in Finno-Ugric languages]. Études Finno-Ougriennes 15: 227– 231. Matras, Yaron 1998 Utterance modifiers and universals of grammatical borrowing. Linguistics 36 (2): 281331. 2000 Fusion and the cognitive basis for bilingual discourse markers. International Journal of Bilingualism 4 (4): 505528. Matras, Yaron, and Jeanette Sakel 2007 Investigating the mechanisms of pattern replication in language convergence. Studies in Language 31 (4): 829865. Myers-Scotton, Carol 2002 Contact Linguistics: Bilingual Encounters and Grammatical Outcomes. Oxford: Oxford University Press. Nau, Nicole 1995 Möglichkeiten und Mechanismen kontaktbewegten Sprachwandels – unter besonderer Berücksichtigung des Finnischen. Munich: Lincom. Ross, Malcolm D. 1988 Proto Oceanic and the Austronesian Languages of Western Melanesia. Canberra: Pacific Linguistics. Sakel, Jeanette, Yaron Matras 2004 Database of Convergence and Borrowing. Manchester: University of Manchester. 2007 Language contact between Spanish and Mosetén: A study of grammatical integration. International Journal of Bilingualism 11 (1): 2553. Savić, Jelena M. 1995 Structural convergence and language change: Evidence from Serbian– English code-switching. Language in Society 24: 475492. Siegel, Jeff 1997 Mixing, leveling, and pidgin/creole development. In Arthur K. Spears and Donald Winford (eds.), The Structure and Status of Pidgins and Creoles, 111149. Amsterdam: John Benjamins.
Types of loan: Matter and pattern
29
Stolz, Christel, and Thomas Stolz 1996 Funktionswortentlehnung in Mesoamerika, Spanisch–Amerindischer Sprachkontakt. Sprachtypologie und Universalienforschung 49 (1): 86123. 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]: 177. Thomason, Sarah, and Terrence Kaufman 1988 Language Contact, Creolization and Genetic Linguistics. Berkeley: University of California Press. Treffers-Daller, Jeanine, and Raymond Mougeon 2005 The role of transfer in language variation and change: Evidence from contact varieties of French. Bilingualism: Language and Cognition 8: 9398. Weinreich, Uriel 1966 Languages in Contact: Findings and Problems. The Hague: Mouton (first publ. 1953).
The borrowability of structural categories Yaron Matras
1. Introduction The question of the “borrowability” of categories has often been equated with the presence or absence of constraints that rule out the borrowing of certain kinds of structures (cf. Campbell 1993, Thomason 2001 and elsewhere). I use the term here in a different sense. Borrowability is taken to mean the likelihood of a structural category to be affected by contact-induced change of some kind or other (whether matter- or pattern-replication; see Matras and Sakel 2007). From a strictly structure-oriented point of view, one might interpret this as the “ease” with which a category can be re-shaped through contact. I am not quite happy with this formulation, either, since it leaves open the source of the process and its motivation. Nor is the issue resolved by re-stating the obvious, namely by claiming that there is a link between the sociolinguistic norms of a speech community, the intensity of cultural contacts, and the outcomes of structural processes of change (cf. Thomason 2001, following Thomason and Kaufman 1988); for such a statement does not account for the fact that the borrowing of some categories requires more intensive contact than that of others. In other words, it fails to explain the hierarchical relationship between individual positions on the borrowing continuum. When we speak of “ease” of borrowing, we are referring implicitly at least to the communicative behaviour of speakers in a bilingual setting and to changes in that behaviour that have a long-lasting effect on the shape of the language that they use. What interests us in this connection is the likelihood that, in respect of a particular structure which serves a particular function in language processing, speakers might give up the separation of two sub-components within their linguistic repertoire – the two “languages” – and begin to employ the structure in question regardless of the choice of language. Bilingual speakers of English and German, for example, take for granted that the concepts computer, download, and internet are common to both sets of communicative interactions in which they normally engage: those where the chosen language of conversation is English, and those where it is German. Bilingual speakers of Domari and Arabic (see Matras,
32
Yaron Matras
this volume) are fully at ease with the fact that the entire system of clause combining and connectors is shared by their two languages; a shift in the interaction setting will lead them to switch into another “language”,1 and this will affect the selection of various structures – vocabulary, inflections, anaphora and deixis, and so on – but it will not affect strategies of clausecombining, which remain the same in all settings (i.e. for both languages). And for speakers of Macedonian and the local Turkish dialects spoken in that country (see Matras and Tufan, this volume) the way of organizing information in copula sentences is identical regardless of the language that is being spoken. This, in essence, is the core of the diachronic process that we call “borrowing”. With “borrowability”, then, we mean the likelihood that speakers will give up the separation between their “languages” – the mental demarcation line that divides their overall repertoire of linguistic structures – in respect of a particular function-bearing structure (a “category”).2
2. Borrowing hierarchies Essentially two kinds of generalizations have been proposed concerning the borrowing of grammatical categories. Those of the first kind relate to the frequency with which a category may be affected by contact-induced change. Generalizations of the second kind suggest an implicational relationship between the borrowing of individual categories: the borrowing of one category is understood to be a pre-condition for the borrowing of another. The majority of observations on grammatical borrowing belong to the first group (cf. Haugen 1950, Heath 1984, Thomason and Kaufman 1988, van Hout and Muysken 1994, Stolz and Stolz 1996 and 1997, Winford 2003, Aikhenvald 2006). Some statements are based on casual impressions only, while others report the results of counting exercises performed on a corpus. An issue that merits attention is the distinction between the counting of tokens and the counting of types (cf. also van Hout and Muysken 1994: 4243): can we consider nouns to be more borrowable than, for instance, conditional particles, simply because nominal tokens occur in a corpus more frequently than conditional particles? Surely, token frequency will tell us how often a borrowed form is used, but it will not necessarily reveal how likely it is to be borrowed? Counting types, in turn, raises problems of its own: Can we conclude on the basis of type-frequency that adjectives, for instance, are easier to borrow than conditional particles, considering that the first constitute an open class of nearly unlimited types (for the purposes of any practical
The borrowability of structural categories
33
comparison), while a language is likely to have just a very restricted inventory of conditional particles (if at all more than just one)? Such issues make it difficult to compare frequency-based hierarchies drawn from conversational corpora.3 This kind of dilemma does not present itself when comparing the grammatical (and lexical) systems of different languages in a sample (whether a structured one, or a casual one). Here, we are interested in the number of languages within the sample in which category X has been re-shaped as a result of contact. The more languages show borrowing affecting a certain category, the higher the frequency of borrowing for that particular category in the sample. We might then say that this category is “more likely” to be borrowed, relative to other categories. Implicational hierarchies entail frequency hierarchies,4 but go a significant step further in suggesting a constraint on the occurrence of borrowing with any lower-ranking category. The usual format of the statement is: Y is not borrowed unless X is borrowed as well (cf. Moravcsik 1978, Stolz 1996, Matras 1998 and 2002, Field 2002, Elšík and Matras 2006). The postulation of implicational borrowing hierarchies thus goes beyond the assumption that categories have different susceptibility to contact-induced change. It suggests that the process of contact-induced change follows, to some extent at least, a predictable pathway, with one stage leading as a pre-requisite to another. Moravcsik (1978) had attempted in this way to link the borrowing of non-nouns to nouns (“no non-nouns are borrowed unless nouns are also borrowed”), inflectional to derivational morphology (“no inflectional morphology is borrowed unless derivational morphology is also borrowed”), and so on, resulting in a web of inter-dependencies among various structural types (cf. Field 2002 for a statement on agglutinative > inflectional morphology). Matras (1998, 2002) and Elšík and Matras (2006) concentrate their observations on inter-dependencies of what they consider “values” of the same category, i.e. members of a shared structural paradigm. Matras (1998) postulates a borrowing hierarchy ‘but’ > ‘or’ > ‘and’ with respect to coordinating conjunctions, and Matras (2002) and Elšík and Matras (2006) suggest ‘necessity’ > ‘ability’ > ‘volition’ with respect to expressions of modality, and many more.5 Frequency-based hierarchies and implicational hierarchies may complement one another. Stolz and Stolz (1996) and Ross (2001), both relying on frequency-oriented observations rather than strict implicational hierarchies, conclude that contact-induced change begins at the level of the organization of discourse, proceeds to the organization of the paragraph, utterance, and
34
Yaron Matras
sentence levels, and only then reaches the levels of the phrase and word. The postulation of this kind of implicational hierarchy rests in such cases on accumulated observations of a series of frequency hierarchies. Borrowing hierarchies thus provide us with an opportunity to gain insights into the factors that prompt speakers to allow their language systems to converge around a particular structure. Explanations of borrowing generally take one of three directions: (1) The degree of borrowing is related to the intensity of exposure to the contact language, (2) The outcome of language contact is a product of the structural similarities and differences (congruence) among the languages concerned, and (3) Borrowability is a product of inherent semantic-pragmatic or structural properties of the affected categories. Issues such as prestige and domain-specialization of the languages typically fall under 1, while conjectures about functional “gaps” as motivating factors fall under type 2. Our interest in the present context is in explanations of type 3. This interest derives from the realization that structures and paradigm values often behave in an asymmetric manner when it comes to contact-related change. Under the “prestige” or “intensity of contact” effect there is no a priori reason why ‘but’ should be more vulnerable and prone to borrowing than ‘and’; in many cases structural congruence does not provide an answer to this hierarchical relationship, either (cf. Matras 1998). Where category or paradigm values consistently show unequal or asymmetrical behaviour in contact situations, factors promoting borrowing must be sought in the inherent properties that they possess. In trying to explain the borrowability of categories we must therefore return to our initial assumption that ease of borrowing reflects the ease with which speakers are willing to give up the separation of two “language systems” and allow them to converge or to fuse around a particular linguistic function. The question that we ask is therefore: What is it that makes one category (or category value) a more attractive candidate for “system conflation” than another? Elšík and Matras (2006: 370ff.), following Matras (1998), argue that borrowing is motivated by cognitive pressure on the speaker to reduce the mental processing load by allowing the structural manifestation of certain mental processing operations in the two languages to merge. The need to do so arises especially around operations that gauge the presentation of propositional content to hearer expectations, for example connectivity and modality. In these domains, merger of the structures targets in the first instance those conceptual domains where the speaker’s epistemic authority is in question, and the potential for tension at the interaction level is therefore greatest. This
The borrowability of structural categories
35
occurs for example around the expression of condition, contrast, participantexternal force, or other, more general conceptual complexity. This accounts for the direction of the hierarchy, which prioritizes category values such as “contrast” and “external force”. We should at this stage clarify the notion of “category”. One of the factors impeding straightforward comparability among hierarchies postulated in the literature is the vagueness with which category labels are used. Some researchers have, for example, identified a category of “function words”; others speak of “particles”, and others still make reference to a class of “pronouns”. As “function words” we can classify anything from interjections and fillers on to definite articles and demonstratives – categories that show enormous variation in respect of their contact behaviour. “Particles” can include markers of modality, connectivity, aspect, and more, while “pronouns” are used with reference to such functionally diverse entities as anaphoric or thirdperson pronouns, indefinites, and participant deixis. If we suspect that there is a link between category status and borrowing, then we must assume that “category” represents a functional notion, rather than just a constituent slot or a wholesale cover-term. Categories are understood here to be operational devices that trigger mental processing activities in communicative interaction: nouns name objects, interjections direct attention to emotive evaluations of the speech situation, connectors establish links between the processing of individual propositions, word order serves as a map to organize information at the utterance level, and so on. Their concrete representation in a given language is through a structural form, which may or may not have cross-linguistic equivalents. Our agenda is to accommodate the borrowing behaviour of categories within an explanatory model, one that accounts for the link between the processing function which the category triggers, and the degree to which speakers allow its structural representation to converge or fuse among the two (or more) components of their linguistic repertoire. To be sure, different explanatory models may be appropriate for different structural components of language. There is no doubt that the borrowing of institutional terminology from a language that is dominant in the public or acrolectal domain is not a result of mere tension at the level of processing the speech interaction, but rather an attempt to extend the referential world of the “dominant” language into interactions in which the “minority” or “weaker” language is used. The borrowing of phonemes may, in turn, be simply instrumental in serving the authentic integration of loanwords without “distorting” them, by adjusting the phoneme system to accommodate them. Nevertheless,
36
Yaron Matras
these too are conversation-functional factors that motivate speakers to allow their “linguistic systems” to conflate around certain structures and categories. Finally, we must briefly comment on the relevance of exceptions to postulated hierarchies. In a discussion that focuses on absolute constraints on borrowing and sets out to test their validity, the discovery of counterexamples can have a sensational effect in dismantling earlier claims. We aim here at taking what Aikhenvald (2006: 26) describes as the more “positive” route: understanding a variety of factors and preferences that facilitate structural diffusion among languages. In this spirit, it would be naïve and counterproductive to ignore tendencies that are followed by a substantial group of languages within the sample only because they are not followed by all, or indeed because they might be contradicted in one or two instances. Where there appears to be a motivation behind trends, one that is beyond pure coincidence, then these trends deserve our attention. Quite often, it is the counterexample that can be explained as resulting from a local, language-particular constraint that impedes the realization of common patterns in a particular instance. The following sections mirror the organization of the language-oriented chapters in this collection, and are devoted to an evaluative overview of selected patterns arising from the discussion of the 27 sample languages.
3. Phonology Phonology in particular is an area in which borrowings are traditionally considered to fill so-called “structural gaps”, facilitated especially when borrowing does not entail changes to the actual phonemic system but merely to allophonic distribution (cf. Winford 2003: 5556). The notion of “gap” is vague, given that languages have long been considered in descriptive linguistics to constitute autonomous, functional systems. We should therefore perhaps amend the definition to focus on bilingual speakers’ quest for harmony among the two (or more) systems that constitute their linguistic repertoire; absence of harmony as a result of absence of a phoneme in one of those systems is presumably what is meant by a “gap”. There is a functional motivation favouring consistency in the types and points of articulation as well as the distribution rules of allophonic variation, regardless of the speech situation in which language users find themselves, and hence pressure toward convergence of the two phonological “systems”. At the same time, social norms and
The borrowability of structural categories
37
awareness of identity and loyalty toward the group associated with the home language will counteract levelling. The process of phonological borrowing is the outcome of compromises between these two pressures. Our sample shows three different types of change: (1) Incorporation of phonemes from a contact language in loanwords, (2) adjustment in the articulation of a phoneme following the model of the contact language, and (3) incorporation of a borrowed phoneme into the system of inherited words (substituting an inherited phoneme in some words, though not necessarily in the system as a whole). Changes of the second type may lead to simplification of the system, or to its enrichment through new distinctions, or they may simply alter the nature of certain phonemes, leaving the complexity of the system as a whole intact. In the sample, changes of the first type typically add to the phoneme inventory, as do in most cases changes of the third type. On the whole, then, our observation is that language contact in the cases under scrutiny here typically leads to an enrichment of the phonological system. Another general observation is that contact-related change is more likely to affect consonants than vowels; indeed, we may even be able to postulate an implicational hierarchy of contact-related change: (1)
adoption of new consonants > adoption of new vowels
The reason behind this hierarchy is, however, likely to be rather trivial: It is a product of the fact that consonant inventories are generally larger, and so the potential for lack of overlap between consonant systems in contact is higher, resulting in greater pressure to adjust the consonant system. In fact, the hierarchy under (1) need not at all suggest that contact induced change in phonology begins with consonants, and it is not impossible that vowels are as prone, or even more prone to change in situations where there is no significant difference among the languages in the inventory of consonants. Almost all languages in the sample incorporate loanwords along with at least some of their original phonemes, which are new to the recipient system. Examples are Macedonian Turkish /ts/ with Macedonian loans, the Vietnamese sounds /ʆ/, /f/. /v/, and /z/ with Chinese loans, the Domari pharyngeals /ħ/ and /ʕ/ used in Arabic loans, and Imbabura Quichua /b/, /d/, /g/, /ʋ/, /ʒ/, and vowels /e/ and /o/ in Spanish loans. This indicates that for speakers of the languages in question, the integration of lexical loans in an “authentic” manner, i.e. one that closely replicates their original use in the contact language, takes precedence over the preservation of the coherent phoneme structures of the recipient language. The system of the recipient language is adjusted in order
38
Yaron Matras
to accommodate loans in an unmodified form. Of the sample languages, only Mosetén and Biak appear not to extend their phoneme system to accommodate loanwords, while the overlap between the phoneme systems of Kriol and Jaminjung does not require Jaminjung speakers to make any special effort in order to accommodate Kriol loans. The second type of process is the convergence of articulation modes and positions, which is often, as Winford (2006) suggests, a process affecting allophonic variation. There are numerous cases involving the introduction of allophonic variation, among them the interchange of /dʒ/ to /ʒ/ in Domari based on the Arabic model, of /q/ and /ɣ/ in Khuzistani Arabic based on the Persian model, and of /l/ and /r/ in certain positions in North-eastern Neo-Aramaic, based on the Kurdish model. In some cases, variation leads to a shift in articulation, as in the weakening of /h/ in Macedonian Turkish or the replacement of /h/ by /ɦ/ in some varieties of Yiddish, of pretone /o/ by /a/ in others, and a more general shift in this language from dental /l/ to velarized /ɫ/. From a language user’s perspective, all these are instances of harmonization of articulatory patterns, aiming to ease the burden of having to maintain complete separation of two distinct systems in different settings of conversational interaction. In some cases, phonemic distinctions in the recipient language are even given up in order to enable harmonization. Some varieties of Purepecha for instance lose the opposition between retroflex /ɽ/ and flap /r/, as well as between central /ï/ and front /i/, resulting in a vowel system that matches that of Spanish. In other cases the system becomes more complex, as with the introduction of pharyngealized consonants into Domari from Arabic and into Tasawaq from Tuareg, of long vowels into Rumungro from Hungarian, and of palatalization of stops into Kildin Saami and Yiddish from Slavic contact languages. Informal observations lead us to believe that prosody is a domain of phonology that is particularly prone to contact. This can be the result of two interconnected factors. The first is the peripheral role that prosody has in conveying meaning, and the fact that it is a form of expression of emotive modes, operating at the speech act and utterance level, rather than the word level. This allows speakers to mentally disconnect prosody more easily from the matter or shape of words associated with a particular language, making it prone to change and modification in contact situations. The second factor may be the proven neurophysiological separation between prosody and other aspects of speech production, making prosody more difficult to control. Both factors may contribute to the fact that foreign “accents” are most persistent in the area of prosody.
The borrowability of structural categories
39
We have in our sample little data on prosody, however. Several pairs of languages are reported to share prosodic features with their current contact languages, confirming the above hypothesis: They include Domari, Nahuatl, Rapanui, Rumungro, and Indonesian, and to some extent at least also Kurdish, some varieties of Yiddish, Hup, and Kildin Saami. This is already more than can be said about any other area of phonology, where we find borrowings and convergent tendencies, but no wholesale convergence.6 Hence, we might carefully postulate the following frequency-based hierarchy of likelihood of complete convergence in the phonological system: (2)
prosodic features > segmental phonological features
Once again we need to emphasize that this hierarchy does not suggest that segmental phonological features are unlikely to be borrowed unless prosodic features are also borrowed; it merely reflects the tendencies toward full-scale convergence of the systems. In fact, it does not seem possible at this stage to point to any position within the phonological system (e.g. certain articulatory modes or positions, marked features, etc.) as being particularly prone to contact-induced change. It seems that the details of phonological change are entirely a product of the relations among the two systems – or congruence – and any statistics of change are likely to simply reflect the mere likelihood of the two phoneme systems in contact to share certain phonemes, and to differ with respect of others. The one additional generalization that we can make is that the borrowing of phonemes begins with the borrowing of lexical items that contain them: (3)
phonological features in loanwords > independent phonological features
Concluding this section, it seems that there are two alternative strategies that multilinguals can pursue in respect of phonology, taking for granted that language contact will lead at least to a transfer of lexical items from one language to another. The first is to maintain the complete integrity of the recipient language system by adjusting the phonology of any borrowed word to match that of the recipient system. It would appear that this strategy would be facilitated by widespread monolingualism in the recipient language, and the confinement of bilingualism to just a small or peripheral group of intermediaries. It is also possible that this strategy can be maintained for a while in situations where widespread bilingualism is a relatively new phenomenon,
40
Yaron Matras
or where speakers of the recipient language have no need to appear to have native-like or even good command of the contact (donor) language. In our sample, only Mosetén and Biak appear to adopt this kind of strategy. The alternative, which is the route taken by most of our sample languages, is to maintain the authenticity of donor language items by adjusting the phonological system of the recipient language to accommodate phonological features of the donor language. This would seem to be facilitated by widespread bilingualism and the need for speakers of the recipient language to gain the approval of the donor language community. Authenticity in the pronunciation of loanwords is a token of the social value attributed to the donor language, and is emblematic of the social immersion into the donor language culture. The result is the incorporation of phonological features from the donor language into the recipient language. These will at first accompany loanwords from the donor language. With increased multilingualism and the need to operate regularly in two linguistic environments, it is advantageous for speakers of the recipient language – we assume that this is normally the language that occupies the weaker socio-political position, or a minority language – to allow major components of their phonological system to converge with that of the dominant, donor language, and so to rid themselves of the burden to maintain a separation of their two speech modes. This tendency of course competes with the sense of loyalty toward the group-language, which may favour maintaining an old system and limit innovation. The outcome is often a compromise in the form of an adjustment of certain aspects of the phonological system in favour of common patterns, or the adoption of some features but not others from the donor language, as we saw above.
4. Typology A number of languages show signs of movement between morphological types: There are changes from polysynthetic to less polysynthetic structuring in Nahuatl and Imbabura Quichua, Otomi, and Guaraní, from an agglutinative type to a more isolating type in Indonesian, and from an agglutinative to a more analytic type in Purepecha. An increase in reliance on reduplication is found in (agglutinative) Likpe under the influence of (isolating) Ewe. On the other hand there is some acquisition of agglutination in Hup, and in some traits perhaps also in Kurmanji Kurdish as well as in Rumungro: In Kurmanji, the agglutination of case markers benefits from the presence of
The borrowability of structural categories
41
inherited enclitic case markers, which historically form part of a circumposition.7 In Rumungro, the adoption of agglutinative prefixes is a by-product of the almost wholesale adoption of indefinite markers, superlative markers, and a few other morpheme classes that are high on the “relevance” scale and so easily borrowable on functional grounds. It is noteworthy that none of these developments seems to follow any predictable structural path, and the only common denominator is an accommodation to the patterns of a socially dominant contact language. In all cases, the drift begins in individual constructions such as adjective comparison or case marking, and it is yet to be seen whether it will continue to spread. Frequencies and the evaluation of general trends are not applicable to our sample in the domain of typology, as some language pairs happen to belong to similar types (consider Domari and Arabic, Vietnamese and Chinese, Yiddish and Slavic languages), while others, among them those named above, show typological clashes with their contact languages. It is interesting that Macedonian Turkish maintains strong morphological agglutination despite considerable re-structuring in the domain of clause organization; indeed, the loss of converbal morphology in favour of grammaticalized conjunctions might be considered a small, yet not insignificant step in the direction of a typological drift. On this basis, we might conclude that typological drift begins at the clause level. In Khuzistani Arabic, the reinterpretation of the construct state ending and definite article as equivalents of the Persian ezāfe attribution marker does not constitute a drift in morphological type as such, but it does expose the path taken toward morphological re-analysis, here too in a possessive construction at the phrase level. Perhaps one of the more outstanding typological shifts reported in the sample is the change in alignment in North-eastern Neo-Aramaic, modelled on the Kurdish ergative construction. An interesting aspect of the construction is that it does not mirror the complete ergative formation – in the absence of, for example, nominal case in Aramaic – but chooses instead only a number of pivotal features which it reconstructs with inherited means (see also Matras and Sakel 2007). Alignment is contact-sensitive elsewhere, too: We find an expansion of ergativity in Urban Manange in contact with Nepali, and the incipient loss of ergativity in Kurmanji in contact with Turkish. In conclusion, although we do not have cases in our sample that display far-reaching changes in overall morphological typology, it is very evident that morphological type is certainly not immune from contact-induced change. It would seem fair to state, at least cautiously, that there is by and large an opportunist motivation for typological drift: It is subordinate to pressure to-
42
Yaron Matras
ward convergence in a particular salient construction, such as the possessive construction, clause linking, or indefiniteness marking, or else it is triggered by accidental similarities in the shape and position of markers with similar meaning.
5. Nominal structures This is a diverse and complex domain, containing many different sub-categories, and it is perhaps not a surprise that only two languages in the sample, Jamingjung and Biak, are reported to show no contact influence at all under this heading. A prominent sub-domain of nominal structure is case, but here it is noteworthy that no borrowing of bound case markers is attested in the sample. The closest evidence of contact influence on case markers is the reliance in some contact varieties of Kurmanji on postposed markers as enclitics rather than as components of circumpositions. There are also indications of meaning extension of case markers, such as the ablative becoming genitive under Dutch influence in Indonesian, or the loss of a distinction between comitative and instrumental in Imbabura Quichua. Adpositions on the other hand do show matter-replication; attested cases include Indonesian sama ‘with’ and guna ‘for the purpose of’ from Sanskrit, Spanish de in Guaraní, numerous Arabic prepositions in Domari, Spanish por and para in Purepecha, and Tuareg ámmàs ‘inside’ and àláqqàm ‘behind’ in Tasawaq. The preposition indicating ‘between’ is the most frequently borrowed, examples being Indonesian antara from Sanskrit, Spanish entre in Guaraní and other languages, and Arabic bēn in Domari. This gives some vague evidence in support of the hierarchy proposed by Elšík and Matras (2006) for the borrowing of local relations expressions in Romani dialects: (4)
peripheral local relations > core local relations
“Core” relations (‘in’, ‘at’, ‘on’) are borrowed less frequently than “peripheral” relations (‘between’, ‘around’, ‘opposite’), and this finds some support in the appearance of ‘between’ as the most frequent borrowing in the sample. Developments affecting gender marking include a shift in unmarked gender from feminine to masculine in Mosetén, the loss of neuter gender in NE Yiddish (as in the contact languages Lithuanian and Latvian), and the incipient system of nominal classifiers in Hup (classifying inanimates by shape, and animates by gender), adopted from Tukano. Definitely the most extensive
The borrowability of structural categories
43
development in this domain is the borrowing of Chinese classifiers into Vietnamese. There are, in addition, some marginal phenomena such as the loss of gender in pronouns (in Rumungro as well as in North-eastern Neo-Aramaic). Our sample gives us the impression that gender in the narrow sense (of a two- or three-gender system) is more stable in contact situations than more differentiated classifier systems, where influence might be more extensive. Nominal possession is a domain in which contact phenomena are fairly widespread. The most common change to possessive constructions is a modification of word order, often drawing on existing flexibility and enhancing the frequency of a more peripheral pattern to match that of the contact language. Examples of contact phenomena in possession are found in Domari, Macedonian Turkish, Rumungro, Khuzistani Arabic, Guaraní, and North-eastern Neo-Aramaic, concerning mainly the order, and in some cases the distribution of morphs and their meaning; incipient cliticization of the postposed pronominal possessor in Kurmanji; frequent postpositive possessors in Rapanui; use of a preposition bearing the same meaning as in the contact language in Indonesian; and use of a borrowed preposition in Guaraní. Definiteness is known for its areal diffusion, but in the sample we have only few cases of contact developments in this domain. Rapanui illustrates that re-arrangement of definiteness rules may occur when definite markers exist in both systems prior to contact. Khuzistani Arabic shows selective retreat of definiteness marking in some constructions, in contact with a language with no overt marking of definiteness. Interestingly, North-eastern Neo-Aramaic borrows the Kurdish definite article -ak-. Nominal morphology is most frequently replicated in the case of plural markers, which are often maintained in loan nouns, either as productive markers of plurality, as in Yiddish, Rumungro, and Tasawaq, or in conjunction with a native expression of plurality, as in Domari and Quichua. In Likpe we find replication of a specific plurality marking pattern. Plural markers may be said to occupy a position in between derivation and inflection markers. On the one hand, they are potentially linked to the expression of plural agreement elsewhere in the sentence, and so they operate at the level of the sentence rather than just the word. On the other hand, they indicate clear semantic opposition to singulars at the word level. Morphological plural marking thus meets the criteria for semantic transparency which is so often noted as a factor facilitating morphological borrowing (Moravcsik 1978, Matras 1998, Field 2002, Winford 2003). The direct borrowing of derivational morphemes is attested throughout the sample. Macedonian Turkish, Yiddish, Quichua, Purepecha, and Rumungro all borrow diminutive suffixes (with Kildin
44
Yaron Matras
Saami re-organizing its diminutive derivation based on a Russian model), and Quichua and Yiddish borrow agentive suffixes (Quichua -dur from Spanish, Yiddish -nik from Russian). The overall pattern leaves us with a picture that is not incongruent with that reported on in contact linguistics so far. The most widespread changes are in the possessive construction. They affect the nominal phrase at the syntactic or morphosyntactic level, having to do primarily with the position of the possessor and possessed object, and partly with the arrangement of possessive morphology. This is in line with predictions that phrase-level borrowing will be more intense than word-level borrowing. Borrowing of bound markers favours in particular plural markers, diminutive and agentive derivational markers, and classifiers (but not gender markers), confirming that semantic transparency facilitates borrowing. Adpositions are more borrowable than bound case markers (borrowing of which is not attested in the sample), with between being the most borrowable in our sample, confirming the tendency of borrowing to favour peripheral relations, and so for the process of convergence to begin with remote, cognitively less accessible or conceptually more complex domains. In other domains, such as the distribution of case, definiteness, or gender assignment, languages may develop similarities, often by extending or limiting distributional rules. However, bound case and gender markers remain on the whole among the most stable features in the nominal domain, resisting especially direct replication of matter.
6. Verbal structures Little attention has been granted in the literature to borrowing of features belonging to the domain of verbs (on the borrowing of lexical verbs see below); reports on the borrowing of TMA markers are quite rare. It is useful to consider the categories one by one. In the domain of tense, we see contact-induced similarities in the organization of the future tense in several languages: It is lacking in Domari and Arabic, it is suffixed in Hup and Tukanoan, and it shows a similar periphrastic structure in Kildin Saami and Russian. To this we might add the similar organization of the prospective aspect in K’abeena and Amharic. Contact phenomena appear to be somewhat more frequent in aspect and aktionsart, where we find matter replication as well as shared patterns. Domari uses an Arabic habitual auxiliary kān, and Nahuatl introduces a progressive based on the Spanish model. We find aspectual use of the borrowed comple-
The borrowability of structural categories
45
tive ya from Spanish in Guaraní, and sudah from Sanskrit in Indonesian, as well as similar expressions of experiential perfect in K’abeena and Amharic, while Likpe adopts a periphrastic present progressive similar to Ewe, and North-eastern Neo-Aramaic adopts a present progressive particle-turnedprefix based on a Kurdish model.8 Yiddish re-organizes its verbal prefixes to replicate so-called Slavic Aspect distinctions (essentially, grammaticalized aktionsart), and Rumungro shows iteratives that are calqued on Hungarian. Mood is similarly somewhat richer in contact developments, though differences in the structural organization of mood and modality make an exhaustive comparison somewhat difficult. Noteworthy are similarities in the use of the subjunctive in North-eastern Neo-Aramaic and Kurdish, and Domari and Arabic (cf. Matras and Sakel 2007: 843). Vietnamese prohibitive and conditional markers are borrowed from Chinese, while Kurmanji borrows the Turkish conditional marker -ise. Modality shows the most widespread contact phenomena, especially as regards matter replication. Almost half of the sample languages show matter replication of modality markers, such as Turkish gerek ‘must’ in Kurmanji, Arabic lāzim ‘must’ in Domari, Spanish tiene que ‘must’ in Rapanui, Arabic mungkin ‘can’ in Indonesian, Spanish pudi ‘can’ in Quichua, Malay harus ‘must’ in Biak, and more. The most common are markers of obligation (i.e. expressing external forces), followed in turn by necessity, possibility, ability, and desire. This hierarchy is almost always implicational, the only exception being Domari (which borrows markers for all meanings except ability):9 (5)
obligation > necessity > possibility > ability > desire
The hierarchy proceeds from the most intensive external force, to the most participant-internal dimension. It is identical to the hierarchy identified by Elšík and Matras (2006) for the borrowing of modality markers in Romani dialects: necessity > ability > (inability) > volition. The more abstract theme in this hierarchy might be described as the degree of “speaker control”, low speaker control correlating with high borrowability. As far as Domari is concerned, its minor deviation from the implicational hierarchy can be explained by the fact that all the modality markers that it borrows from Arabic are impersonal expressions, or non-verbs. Even bidd‘want’, is a nominal, and its person inflection in Arabic follows the paradigm reserved for nominals, i.e “my-wish” etc. (and this inflection is carried over into Domari as well). The Arabic expression for ability, however, ‘a-qdar-, is an inflected verb, and although Arabic verbs are borrowable in Domari,
46
Yaron Matras
it competes with an inherited Domari verb sak-, which prevails. Thus, it is the formal inconsistency in the system of the donor language which in this case imposes a constraint that breaks the hierarchy. Returning now to a general view of TMA and modality, we have seen the high density of (matter) borrowing in the domain of modality, in some cases also in mood, frequent matter and pattern replications in the area of aspect and aktionsart, and few cases of pattern replication in tense, all involving the future. This picture lends itself to an interpretation in terms of the hierarchy in (6), which depicts the likelihood of the respective categories to be affected by contact: (6)
modality > aspect/aktiosnart > future tense > (other tenses)
By and large, this hierarchy reflects both frequency, and implicational relationships. There is one case in the sample – Kildin Saami – where there appears to be contact influence in the arrangement of tense, but not in modality. Yiddish might be considered a case for contact influence on aktionsart, but it similarly lacks borrowed modal verbs from its Slavic contact languages, though to some extent this might be explained by the presence in this domain of Hebrew loans acquired through literary tradition. The internal rationale of the hierarchy – which, once again resembles the findings for Romani dialects (cf. Matras 2002) – leads us to postulate again that external circumstances that limit the degree of speaker control – mood and modality in general – are the most contact-sensitive. They are followed by a qualification of the internal structure of the event – aspect and aktionsart – these too being beyond the immediate control of the speaker. Only then do we find contact influence in tense, the most intimate relationship between the event and the speaker’s own perspective, though it is noteworthy that in our sample this is limited to the future tense, which identifies the event as being least stable and secure from the speaker’s perspective. The overall theme is therefore once again the speaker’s epistemic authority; its absence or weakening correlates with high borrowability. Existential and possession verbs are affected by contact in several of the sample languages. In Domari, the Arabic copula is adopted in its function as a periphrastic expression of the habitual aspect, and it co-exists with the Domari enclitic copula. But Rapanui uses Tahitan and Spanish forms as copula, and both Indonesian and North-eastern Neo-Aramaic are reported to have developed copula forms through imitation of forms in their contact languages. The borrowing of ‘have’ is on the whole marginal. Spanish tengo is
The borrowability of structural categories
47
employed in Rapanui, while in some other languages, such as Hup, we find calquing of constructions. Contact phenomena in the area of voice and valency are almost exclusively pattern-oriented, and usually involve an increase in frequency distribution of an existing option: analytic reflexives in Quichua, the periphrastic passive in Purepecha, a causative marker derived from the verb ‘to do’ in Manange (copying the function of the Nepali affix -aau), and productive use of the inherited morphological causative in Rumungro. Vietnamese is exceptional in directly borrowing reflexive and passive markers from Chinese. Recent contact-induced grammaticalizations lead to the emergence of causative, passive, and reflexive markers in Hup, and to a reflexive in Likpe. It is difficult to further sub-divide this domain into category components, or to draw clearcut connections with other components of the verb, other than to say that derivation generally appears much more contact-susceptible than tense, and perhaps at the same level as aspect. Both are in a sense statements about the internal organization of events, not directly connected to the speaker’s position, but with no explicit evaluative statement concerning the truth-relevance or factuality of the event (as in mood or modality), either. Moravcsik (1975) had drawn attention to the frequent use of incorporation strategies to accommodate borrowed verbs, and this typological discussion of verb borrowability has recently been revived (cf. Wichmann and Wohlgemuth, forthc.). There appears to be a near-consensus view that the borrowing of verbs is not, of course, impossible, but made more cumbersome in some languages due to the widespread tendency of verbs to be morphologically more complex (see Winford 2003: 52). In our sample, direct borrowing of verbs, without any formal adaptation, is found in Vietnamese, where there is no morphology in either the recipient or the donor language Chinese; in Likpe, where isolating Ewe contributes verb roots into an agglutinative structure; but also in many other languages, including Tasawaq, Quichua, Otomi, Guaraní, Hup, and K’abeena. There is thus obviously no universal constraint on the integration of borrowed verbs.10 Nonetheless, several languages in the sample prefer to apply an overt, morphological accommodation strategy when incorporating verbs of foreign origin into the lexicon. A favourite strategy is the use of so-called “light” verbs. Macedonian Turkish, Kurmanji, Domari, and Khuzistani Arabic belong to a larger isogloss covering the Caucasian–Mideastern–South Asian area, where mainly two light verbs are used, each combining with a root of nominal form of the borrowed verb. The distinction between the two light verbs is usually one of valency, and they usually derive from or are identical
48
Yaron Matras
to the lexical verbs for ‘to make/ to do’ and ‘to be/ to become’. In Domari, the carrier verbs are semi-grammaticalized, and full forms -kar- ‘to do’ and ‘-hr‘to become’ co-exist with the abbreviated integration markers -k- and -Ø-. A somewhat similar strategy is found in Mosetén, where one of the verb adaptation markers is also a valency-augment, in Yiddish, where Hebrew-derived verbs are accompanied by zayn ‘to be’, and in Jaminjung, where loans have coverb status and are always used in combination with a native inflected verb. A series of other languages use a verbalizing augment which is otherwise employed to derive verbs from non-verbs as an integration marker: Mosetén, Nahuatl, Indonesian, Guaraní, Biak, Purepecha, Manange, Rumungro, and Yaqui. The borrowed verb itself usually appears in either the root form, an infinitive form, or an unmarked inflected form, quite often – for Spanish verbs, especially – the third-person singular present. While no constraint on the borrowability of verbs can be upheld, it is nevertheless evident that a large number of languages require greater grammatical effort in integrating verbs than for the integration of nouns. The borrowing hierarchy (7)
nouns > verbs
expresses the grammatical “ease” or simplicity with which elements belonging to these two word classes can be integrated. Why is it that verbs require greater morphological integration effort, and what does this greater effort represent? On the one hand we find a pair of morphologically isolating languages like Vietnamese and Chinese, with no morphological complexity surrounding verbs in either the recipient or the donor language, and where verbs are integrated in a straightforward manner, just like nouns and other parts of speech. It is difficult, however, to attribute the need for explicit loan-verb adaptation markers even in other languages to the morphological complexity of verbs alone. Recall that most languages tend to integrate simple forms of the verb, such as the root, the infinitive, or an unmarked form. There does not seem to be, in those languages, any difficulty in stripping the target verb to its bare lexical essentials, before transposing it into the host morphology.11 I suggest instead that the difficulty lies in the conceptual complexity of the verb, and the fact that when borrowed and integrated, the verb is expected to perform two operations: The first is to serve as a referential lexical item – a content word, not dissimilar to a noun, adjective, or descriptive adverb. The second is to initiate the predication and so to serve as the principal anchor point for the entire proposition of the utterance. This latter function consti-
The borrowability of structural categories
49
tutes its “verbness”. It appears that borrowing of verbs is motivated by a similar need for modifying the inventory of lexical-referential expressions as the borrowing of nouns (and no doubt various specific semantic motivations could be postulated for groups of lexical content words). Speakers thus allow the lexical component of the verb to “cross” the mental demarcation boundary between languages, i.e. they license themselves to employ the same action/event signifier in any speech interaction. The bare lexical stem, however, is not always sufficient in order to assume the role of predication-initiator. A great number of languages therefore require this additional, crucial function to be explicitly marked out in the verbal expression; in other words, they need to transform the strictly “lexical” depiction of an action/event into a predicate. This is achieved through explicit marking of its “verbness”. A more detailed study is required in order to ascertain the conditions under which languages require some form of loan-verb integration. A first, banal observation is that a pre-requisite for the employment of loan-verb adaptation markers is the availability in the recipient language of a morphological procedure to derive verbs from non-verbs. Whereas an isolating language like Vietnamese may rely to a considerable extent on the pragmatics of morpheme juxtaposition as a way of (indirectly) marking out word classes, flectional languages will often require an additional means of identifying derived verbs. This, of course, only explains a part of the story. It is clear that the similarities in loan-verb adaptation strategies found among the languages of the Caucasian–Mideastern–South Asian area are as much areally motivated (i.e. through contact and imitation among the languages) as they are functionally motivated by the respective morphological structures of these diverse languages. Moreover, not all languages that possess verb derivation strategies employ them with loan-verbs. And finally, languages are known to have changed their loan-verb adaptation strategy over time, without adopting any significant changes to their morphological typology. Thus, Romani appears to have shared the “light verb”, valency-marking strategy of loan-verb adaptation during its early, Byzantine period, with languages of the Caucasus–Anatolia–South Asia “area” (see Matras 2002). It then transferred the function of marking out loan-verbs to a set of Greek-derived aspectual markers. Finally, in some contemporary dialects of Romani, such as Sinti, Kaale, and Vlax (spoken in or around Germany, Finland, and Romania/Hungary, respectively), loan-verb integration markers are being reduced altogether. For the time being, our principal conclusion can be that the hierarchy depicted under (7) applies for those cases where integration of a lexical item requires morphological support through derivational means.
50
Yaron Matras
7. Other parts of speech The present section gives an overview of those elements often summarized together as “function words” or “unbound grammatical lexemes”. Lumping them together in one section is a matter of convenience, and follows, as did the previous sections, the structure of the Language Convergence and Borrowing questionnaire, which forms the basis of the descriptive chapter. By discussing them under a shared heading, I am not suggesting at this stage that they share properties that motivate borrowing, nor that their behaviour in contact situations should, for any reason, be uniform or even similar at all.
7.1. Numerals Numerals, in fact, are often considered low on the borrowing scale. This may derive from an assumption that all languages have some form of quantification. Although it is now known that not all languages possess systems for counting discrete entities, it is not necessarily the clash of systems of quantification that provides the motivation for the borrowing of numerals. Several types of borrowing involving numerals can be identified. Pattern replication appears in some languages: A decimal system is reported to have been adopted as a result of contact in Mosetén, Indonesian adopts a Javanese tag-lexeme indicating “teens” and re-organizes its earlier system of numeral juxtaposition above 10 accordingly, Hup adopts the Tukano quintenary system for numerals between 5 and 20, and a combination lexemes (‘ten-and-one’) replace single lexemes (‘eleven’) in some varieties of Kurmanji, replicating the Turkish arrangement. This rather small group suggests the following implicational hierarchy representing the likelihood of pattern-replication in numerals, which is yet to be confirmed by a larger sample: (8) over 10 > below 10 More than two-thirds of the sample languages show some form of direct matter-replication of numerals. This includes most of the languages that reorganize their pattern of numerals, and which often employ borrowed numerals alongside the re-modelled “internal” or inherited system. In some cases, numeral replication is subject to sociolinguistic constraints, with contact-language numerals used as the preferred system for formal purposes such as citing dates and addresses and performing even simple mathematical tasks such
The borrowability of structural categories
51
as counting (as opposed to the casual use of numerals as attributes), in transactions involving money, in names for coins of banknotes, or in some cases in the citation of grades. Such contextual splits are described for the use of Turkish numerals in Kurmanji, for the use of Hokkien numerals in colloquial Indonesian, for dates in Rumungro, and for Chinese numerals in Vietnamese, leading us to postulate a sociolinguistic hierarchy for the likelihood of use of borrowed numerals: (9) more formal contexts > less formal contexts This hierarchy reflects the fact that numerals enter languages through the dominance of the second language in formal and business transactions, and through education and other forms of institutional discourse. In many of the sample languages, especially those in post-colonial contexts, knowledge of the indigenous system of numerals is reported to be in decline, and the younger generation shows a clear preference for borrowed numerals. The adoption of borrowings in such situations clearly favours higher numerals over lower numerals, allowing us to postulate the following implicational hierarchy for the borrowing of cardinal numerals: (10) higher numerals 1000, 100 > above 20 > above 10 > above 5 > below 5 This hierarchy appears related to some of the hierarchies postulated above, where borrowing is facilitated around conceptual complexity and inaccessibility. At the same time, higher frequency in casual language use of lower numerals is clearly a factor supporting the retention of native forms. Languages for which speakers are reported to be using native forms for lower numerals under 5 – either primarily, or alongside borrowings – but mainly or exclusively borrowings for numerals above 5 include Domari, some Kurdish speakers, Jaminjung (which has no native numerals above 3), Tasawaq, Otomi, Guaraní, Purepecha, Yaqui, Kildin Saami, and we might add Rumungro, where 79 are Greek loans into Early Romani, and 6 has been argued to be an early loan from Dardic (Indo-Iranian frontier languages). For some languages, such as Mosetén, Quichua, Nahuatl, and Biak, the cut-off point tends to be 10, while Hup shows a split at 20. Higher numerals show an independent susceptibility to borrowing. Tasawaq for instance borrows its lower numerals from Arabic, but its word for 100 from Tuareg; Rumungro generally shows Greek borrowings, but 1000 is Hungarian; and Vietnamese uses a Chinese word for 10,000.
52
Yaron Matras
Somewhat paradoxically, “0” ranks closer to the higher numerals 100, 1000, and so on: the only K’abeena borrowing is zeeruta, from Italian via Amharic, while in Rumungro it derives, like the higher numerals, from Hungarian. This is not surprising, and shows that cognitive complexity in the counting system operates in respect of the ability to easily identify and appreciate a quantity. This is hindered the greater the quantity, but it is similarly hindered in the absence of any quantity at all. An additional factor that no doubt plays a role is the relative formality of the term “zero”, which is associated with mathematical and other formal notations and transactions, but not with everyday, casual expression of “nothingness”. Sample languages that do not borrow numerals are Macedonian Turkish, Khuzistani Arabic, Yiddish, and Manange (though incipient influence of Nepali on the numeral system is reported). In all but the latter, we can attribute the stability of numerals to a firm tradition of native-language education, media, and literacy, if not widespread among all speakers, then at least firmly anchored in the community and its history. This confirms once again that the borrowing of numerals is motivated not necessarily by “gaps” in the system of counting, but by a much more general accommodation to the language of formal institutions and the public domain in the way of conceptualizing and expressing formal transactions surrounding quantification. The hierarchies presented in (9) and (10) are fully in line with the observations described for Romani dialects by Matras (2002) and by Elšík and Matras (2006), which lends support to their validity as universal indicators. A tentative case can be made for the following hierarchy of the likelihood of borrowing of ordinal numbers: (11) lower ordinals > higher ordinals This hierarchy is presented by Elšík and Matras for Romani. In the present sample it is confirmed by Kildin Saami and Rumungro, which use borrowed ordinals for ‘first’,12 and Western Neo-Aramaic, which uses Arabic ordinals for ‘110’, while Domari, Otomi and Purepecha generally rely on borrowed ordinals. Note that English (not part of the sample) is an exception to the hierarchy, having borrowed second but not first from Romance. The ordinal ‘first’ is often a separate word, quite often suppletive to the rest of the ordinal paradigm. In some languages, this is also true of ‘second’. This structural conspicuousness could be a factor promoting borrowing. In the assessment of Elšík and Matras (2006) the high borrowability of lower ordinals is a direct factor of this universal tendency to prioritize the ordinal ‘first’
The borrowability of structural categories
53
through lexical suppletion, which in turn is an expression of its cognitive saliency. Borrowing therefore simply follows the same path; in other words, the search for a renewable (=suppletive) item as an outstanding marker of the pragmatic saliency of “firstness” exploits the bilingual situation in recruiting an item from the contact language. We might therefore, in respect of ordinal numbers, postulate the following hierarchy of borrowability (12) exclusivity > inclusivity where “exclusivity” is taken to mean the separation of a single concept, entity, topic or state of affairs from a larger set – here in relation to the order of prominence, such as temporal sequence, or the attention granted to the object. Conceptually, this hierarchy is well in line with the contact-susceptibility of such properties as condition (see above), privative ‘without, instead of, except for’ (cf. Elšík and Matras 2006), contrast, phasal change (‘already’), restriction (‘only’), and the superlative (see below), all of which denote a broken chain of expectations, singling out our delimiting one entity from a presuppositional set.
7.2. Pronominal forms The borrowing behaviour of so-called “pronouns” illustrates how limited a wholesale structural approach to category borrowing can be, and how it is the functionality of categories that motivates borrowing. Only Indonesian shows borrowing of personal pronouns, from Sanskrit and Javanese, into a system of highly differentiated, lexicalized terms of address and reference. Other contact-developments in pronouns are limited to the organization of the system of reference: Imbabura Quecha is reported to have developed a polite form of the second-person pronoun kikin on the basis of Spanish Usted, and in some heavily Hispanicized varieties of Guaraní, the inherited distinction between inclusive and exclusive is dissolved. Borrowing of other deictic and anaphoric forms is limited. Spanish la is used as an anaphor in Guaraní, the Arabic resumptive pronoun iyyā- is used in Domari in relative clauses, and Rumungro borrows the Hungarian deictic prefixes am- and ugyan- which are combined with Romani deictic stems. Reflexive pronouns are borrowed in Tasawaq, Western Neo-Aramaic, and Rumungro. Yiddish shows an extension of reflexivity based on a Slavic model, and Indonesian calques a reflexive apparently on a Sanskrit model.
54
Yaron Matras
Reciprocal pronouns are borrowed in Domari and Western Neo-Aramaic, and Rumungro calques a Hungarian model. What motivates the borrowing of reflexive and reciprocal forms? Unlike straightforward deictic and anaphoric reference devices, reflexives and reciprocals may be said to constitute an extension of the derivational system of the verb, contributing to the layout of actors involved in and affected by an event. They are thus part of a construction that revolves around the verb’s “actionality”. Interrogatives are borrowed into several languages. Those that stand out as more highly borrowable are the interrogatives for quantity (‘how much’), borrowed into Domari, Otomi, and Manange, and time (‘when’), borrowed into Domari, Indonesian, Quichua, and Rumungro. Borrowing also affects indefinites. Domari, Otomi, and Rumungro borrow all or almost all of their indefinite expressions, while Tasawaq and Purepecha borrow time indefinities, Guaraní the indefinites for person and thing, and Yiddish borrows indefinite markers. Although there is no direct, predictable link to other categories, borrowing in the domains of interrogatives and indefinites appears to be a more “advanced” stage of borrowing among Other Parts of Speech.
7.3. Connectors/conjunctions The grammatical category that is by far the most susceptible to borrowing is that of connectors (see already Matras 1998). All languages in the sample borrow connectors, and the general picture confirms the implicational hierarchy postulated as universal in Matras (1998), and confirmed by Elšík and Matras (2006) for Romani (and recently by Stolz 2007 for a number of languages in contact with Italian): (13) but > or > and Sample languages that borrow all three connectors include Domari, Mosetén, Nahuatl, Kurmanji, Rapanui, Indonesian, Quichua, Otomi, Guaraní, Kildin Saami, and Western Neo-Aramaic; languages that borrow only ‘but’ and ‘or’ are Tasawaq, Purepecha, Vietnamese, Rumungro, K’abeena, and Likpe. No languages borrow ‘and’ without also borrowing ‘but’ and ‘or’. There are, however, a few languages that deviate slightly from the expected pattern. Macedonian Turkish borrows Macedonian i ‘and’ as well as ili ‘or’ and a ‘or, whereas’, but retains Turkish ama ‘but’; however, the latter is identical to Macedonian ama, which is a Turkish borrowing (cf. Matras 2004).
The borrowability of structural categories
55
Jaminjung uses the borrowed contrastive marker Kriol ani ‘only’ alongside its native bugu, while only borrowed forms are used for addition and disjunction; but this is due to the absence of any native connectors for addition or disjunction to compete with the borrowings. In Manange, Nepali ani ‘and then’ can also be used for clause coordination (cf. Stolz’s 2007 discussion of Italian allora). Neither of these cases necessarily contradicts the hierarchy in (13). Biak, however, is reported to use the Indonesian disjunction marker atau ‘or’ and less frequently the addition marker dan ‘and’, but no mention is made of a borrowed contrastive marker. Hup borrows ou ‘or’ from Tukano. The source is ultimately Portuguese, and is reported to have diffused widely in the area. Aikhenvald (2002), too, reports on Tariana ou ‘or’, with no borrowing of other Portuguese connectors. Similarly, Yaqui appears to borrow only Spanish o ‘or’. The fact that counterexamples can be found does not invalidate the overall observation that contrast is a semantic-pragmatic feature that facilitates borrowing, nor of course that clause-combining is an operational domain that is prone to contact-related change. Most likely, certain constraints of a structural and perhaps also a cultural nature (conventions on structuring discourse and expressing overt contrast) override the universal tendency in some cases. Noteworthy is the cluster of Amazonian languages within which Portuguese ou diffuses, often via secondary sources only. Borrowing in the domain of coordinating conjunctions is missing only in Yiddish and Khuzistani Arabic. Subordinating conjunctions are similarly a frequent target of borrowings. Among the complementizers, borrowing is almost entirely restricted to those that introduce factual clauses, which are borrowed in Domari, Khuzistani Arabic, Rumungro, Western Neo-Aramaic, and Likpe. Although modality has been shown to be contact-prone, at the level of the organization of the complex sentence factual complements show greater event independence and so greater effort is needed in order to process the connection between the two clauses (see already Givón 1990, Dixon 1995, 2006). Factuality at this level is thus quite in line, as a factor promoting borrowing, with contrast, limitation, exemption and other properties that we have so far encountered at the top of borrowability hiererachies. Also well in line with these tendencies is the high presence, among borrowed conjunctions that introduce adverbial clauses, of those that mark concessive relations (borrowed in Yiddish, Tasawaq, Indonesian, Quechua, Guaraní, Domari, and Likpe), causal relations (Domari, Mosetén, Nahuatl, Kurmanji, Rapanui, Jaminjung, Tasawaq, and numerous others that calque causal subordinators), and purpose (e.g. Domari, Mosotén, Nahuatl). High on the borrowing scale are also conditional subordinators
56
Yaron Matras
(borrowed in Domari, Mosetén, Indonesian, Quichua, and Guaraní), while the borrowing of temporal subordinators is often linked to that of conjunctions expressing purpose and cause. We can therefore postulate the following tentative borrowing hierarchies: (14) concessive, conditional, causal, purpose > other subordinators (15) factual complementizers > non-factual complementizers Concession, of course, is tightly linked to contrast and unexpectedness. Condition is an expression of modality. Cause and purpose are both efforts to link independent events, as are factual complementizers, while cause constitutes in addition an explicit argumentation effort at the interactional level. The hierarchies (14)–(15) thus supply us with a series of semantic-pragmatic properties that are borrowing-prone.
7.4. Particles Not many languages in the sample borrow phasal adverbs, but those that do show a clear implicational hierarchy (16) yet, already > still > (no longer) confirming that observed in Matras (1998) for Romani, as well as in van der Auwera (1998) for a sample of European languages. While Rumungro and Domari show borrowings in all positions, Jaminjung borrows ‘yet’, Guaraní only borrows ‘already’, while Otomi also has ‘still’. The semantic opposition involved is one of change vs. continuation, in the first instance. While ‘no longer’ essentially expresses change, its position on the hierarchy is partly influenced by its tendency to be composed of several structural elements. It is therefore the first two positions on the (left of the) hierarchy that are the most meaningful, and which continue the theme of contact-susceptibility of contrast and discontinuity of pre-suppositional expectations. Another particle that shows frequent borrowing is ‘again’ (Domari, Mosetén, Kurdish, Jaminjung, Indonesian, Otomi), expressing an unexpected repetition of events. Half the languages in the sample borrow focus particles, giving the implicational hierarchy (17) only > too > (even)
The borrowability of structural categories
57
once again in line with observations on Romani and other languages in Matras (1998). In fact, the particles ‘only’ and ‘too’ usually go together, but ‘only’ can be considered higher on the hierarchy, since Indonesian and Western Neo-Aramaic have borrowed ‘only’ but not ‘too’. Khuzistani Arabic borrows Persian ‘too’ (hem), but since Persian ‘only’ is itself an Arabic loan (faqat) it is not identifiable as a borrowing in Khuzistani Arabic. The hierarchy in (17) indicates that restriction facilitates borrowing, while the proneness of focus particles (and indeed of phasal adverbs and repetition adverbs) to borrowing shows the vulnerability of the system of processing states of affairs and attitudes that are high on the relevance scale and that assess information in direct relation to existing hearer-sided presuppositions and expectations. Fully consistent with this observation is the overwhelming tendency of the languages in the sample to borrow discourse markers (once again, cf. Matras 1998). There are only two languages that do not show borrowing of discourse markers: Biak and Vietnamese. There is no obvious explanation for the absence of borrowed discourse markers in these languages, except perhaps the fact that using native intonation and modal particles is considered a central characteristic of talking Biak and so an important identity marker,13 and that Chinese influence on Vietnamese was transmitted to a considerable extent via the formal and literary language, rather than via oral discourse. We also find less extensive borrowing of discourse markers in languages with a tradition of native literacy: Yiddish, Khuzistani Arabic, Macedonian Turkish, and Indonesian. Limited borrowing in this domain is also typical of Manange and Hup, while Domari, Mosetén, Jaminjung, Guaraní, and Purepecha show the most extensive use of borrowed discourse markers. It is noteworthy that Tasawaq has its fillers and discourse markers from Hausa, a contemporary “pragmatically dominant” language (cf. Matras 1998). On the whole, the following hierarchy (from Matras 1998) of borrowability, both frequency-based and at least in most cases implicational, could be upheld: (18) discourse markers > other particles Question and answer particles must also be considered in this connection. The former are not a universal phenomenon anyway, and it is not surprising that they are limited, in our case, to Macedonian Turkish, which borrows its question particle from Albanian, interestingly replacing a native Turkish question particle. The borrowing of the positive answer particle ‘yes’ is more common. It is often employed as a turn-taking particle rather than just as a signal of agreement with content, and it perhaps for this reason that it is more
58
Yaron Matras
vulnerable than its negative counterpart ‘no’. Rumungro and Mosetén borrow ‘yes’ (from Hungarian and Spanish respectively), while both ‘yes’ and ‘no’ are borrowed from Tuareg in Tasawaq, from Arabic in Domari, and from Spanish in Guaraní. The small sub-sample suggests an implicational hierarchy for the borrowing of answer particles which agrees with that postulated for Romani dialects by Elšík and Matras (2006: 343): (19) positive > negative
7.5. Grammatical vocabulary A notable gap in the borrowing inventory surrounds expressions of place deixis, which in our sample appears fully resistant to borrowing. Borrowed expressions of place are largely limited to the place indefinites ‘nowhere’ and ‘anywhere’, in which the borrowed component may either be the entire expression or just the indefinite marker (Domari, Otomi, Rumungro). The Yiddish presentative ot from Slavic is used in pointing only, and although a deixis of sorts, it is arguably more a verbalized gesture and hence more closely related to discourse markers than a member of a deictic paradigm. Borrowed time expressions encompass both indefinites (‘always’, ‘never’) and deixis (‘now’, ‘then’). Here, both classes are subject to borrowing and their behaviour appears to be linked in the following implicational hierarchy: (20) always > never > now, then In line with the high frequency of borrowing around indefinites, ‘always’ is the most frequently borrowed, followed by ‘never’. The relevance properties of indefinites – as operators that process hearer-sided presuppositions – explains their higher position on the hierarchy (at least, this is well in line with other hierarchies discussed so far). The deictic expressions ‘now’ and ‘then’ are usually linked, and therefore occupy a single position on the hierarchy; occasionally they are borrowed independently of one another: Tasawaq borrows just ‘now’ from Tuareg, while Rumungro borrows only ‘then’ from Hungarian. Note that the Romani sample (Matras 2002) shows a clear hierarchy which favours the borrowing of ‘then’, beginning most often in its sequential rather than remote-deictic function, while ‘now’ is rarely borrowed.
The borrowability of structural categories
59
Terms for days of the week are generally borrowed from the language of education, or the formal-official language: We find borrowings in Nahuatl, Jaminjung, Indonesian (ultimately from Arabic), Quichua, Otomi Guaraní, Biak, Purepecha, Hup (from Portuguese), Rumungro, K’abeena, Kildin Saami, and Western Neo-Aramaic. Borrowing of expressions for times of day (‘morning’, ‘noon’, etc.) is usually linked to days of the week, the sample showing an implicational hierarchy: (21) days of week > times of day This hierarchy can be nicely connected to the role of institutional administration and the language of commerce: Between the two categories, it is days of the week which tend to be the property of the public domain, governing the schedule of activities of individuals in relation to institutions, more so than times of the day, which have a greater autonomous role within the private domain. Once again we find that languages that do not have a strong tradition of relying on external languages for affairs of the public domain – Yiddish, Macedonian Turkish, Khuzistani Arabic – show no borrowings in these fields. Manange once again occupies a similar position, showing here too resistance toward borrowing and behaving much like a language with a tradition of literacy and institutional discourse. For adjective comparison, our sample clearly confirms the hierarchy postulated by Elšík and Matras (2006) for Romani dialects (omitting the least borrowable value “positive”, which is the default form of the adjective and is not usually accompanied by any overt derivational marker): (22) superlative > comparative Formal means of constructing both the superlative and the comparative are borrowed in Domari, Otomi, and Vietnamese, as well as Western Neo-Aramaic. Languages in which borrowing is limited to the superlative include Indonesian, with a Javanese particle paling, Rumungro, which borrows the Hungarian superlative prefix leg, and Kildin Saami, which has the Russian the superlative marker same. Yiddish, also confirming the hierarchy, shows an interesting case of hierarchical distribution of matter- and pattern-replication: While the comparative shows pattern-replication – greyser fun mir ‘bigger from me’ – the superlative shows matter borrowing, replicating the Russian marker same: same groys ‘biggest’.
60
Yaron Matras
8. Constituent order and syntax Contact-induced change in word order is generally not common in our sample. Evidence for change in basic word order is found in Rapanui, with a tendency toward change in Quichua and Otomi and sometimes in Khuzistani Arabic (here we are dealing mainly with relaxation of discourse constraints). It is plausible that in their earlier history Hup, Rumungro, and Domari showed different word-order patterns, though concrete evidence is lacking. The most common change in word order appears to affect possessive constructions. Examples are Domari, Macedonian Turkish, Rapanui, Quichua, Rumungro, and Likpe (see above, under Nominal structures). This is understandable, given the fact that a change in the position of possessor and possessed does not affect the position of the verb and so it leaves the organization of the predication intact. The position of adjectives is affected by contact in Domari, sometimes in Quichua, and in urban Manange. Relative clauses change their position relative to the head in Macedonian Turkish and in Nahuatl, as well as, arguably, in Domari, if we compare the language with attested Indo-Aryan languages. The position of the copula appears more vulnerable to change than the position of lexical verbs, as seen in Macedonian Turkish, Rumungro (tendency only), and North-eastern Neo-Aramaic. From this, we might propose the following tentative hierarchy, based partly on frequency and partly on pure prediction, for the likelihood of word order to be affected by contact-related change: (23) nominal constituents (possessor, adjective) > copula predications > verbal predications Note that the hierarchy is sensitive to the presence of a lexical verb as initiator of the predication – a factor which impedes borrowing. Thus the most borrowable are structures that do not involve a full predication, or at least not a verbal one; these are followed by non-lexical predications, while predications that contain full lexical verbs appear last. In the area of clause structure, one of the frequent changes observed is the emergence of copula clauses: Otomi borrows the Spanish copula ta, Hup uses a possible Tukano loan as a copula, and Indonesian creates copulas on a Sanskrit model. This makes sense, if one considers that non-universality of (present-tense) copula predications, and the fact that a clash of systems, and so pressure toward convergence, is more likely to occur here than in other clause types.
The borrowability of structural categories
61
As far as syntax-relevant grammatical vocabulary is concerned, negation particles are borrowed in Domari, Quichua, Guaraní, Biak, Hup. Connectors and conjunctions were dealt with above (under Other Parts of Speech), and it was seen that most languages in the sample show some kind of contact influence on clause combining strategies. This is true especially of coordination, where almost all languages are affected, some by mere borrowing of connectors, others by changing connectivity strategy, as in the case of Rapanui (from serialization to connectors), or Macedonian Turkish (from converbs to connectors introducing finite clauses). A new type of adverbial subordination is attested only in Macedonian Turkish, while relative clauses are re-structured in Macedonian Turkish, Domari, Yiddish (to some extent), and Tasawaq. Like connectors, relative particles appear to be borrowing-prone, and we find loans in Nahuatl, Rapanui, Domari, Otomi, Guaraní, Indonesian, and Kildin Saami. On the whole, then, the sample languages do not offer an extreme wealth of data on contact-induced change in clause or sentence structure. In particular, colonial languages and languages of administration do not seem to have the effect on syntactic structures as they may have on other domains of structure, in particular grammatical vocabulary. Those cases where we do find farreaching changes in syntactic typology tend to be languages in a prolonged situation of stable multilingualism, as in the case of Macedonian Turkish or Domari, confirming Thomason and Kaufman’s (1988) prediction on the link between prolonged and intense cultural contact, and significant typological disruption.
9. Lexicon All languages borrow lexical elements. Based on authors’ responses to the questionnaire, we can make a statement about the likelihood of a certain word-class element to be affected by contact. This hierarchy is not implicational, as there is no evidence to suggest that borrowing in a lower-ranking category necessarily entails borrowing in higher-ranking categories. Rather, it is based on frequency; (24) shows how frequently selected word classes occur among the list of word classes affected by contact in the sample: (24) nouns, conjunctions > verbs > discourse markers > adjectives > interjections > adverbs > other particles, adpositions > numerals > pronouns > derivational affixes > inflectional affixes
62
Yaron Matras
In some salient features, this hierarchy resembles other hierarchies proposed in the literature.14 Thus, nouns appear at the top of the list, unbound grammatical vocabulary is rather high on the list, and bound morphology is low, with derivational morphology outranking inflectional morphology. Note however some differences to the hierarchy proposed by Muysken (1981) and others: Conjunctions and discourse markers occupy a high position, and outrank some of the lexical categories. Numerals outrank pronouns and derivational morphology in this particular sample, much of it composed of languages in contact with a colonial language. No differentiation is made between coordinating and subordinating conjunctions in our discussion, but it is clear that the position even of subordinating conjunctions is far higher on the hierarchy that that assigned to it in Muysken’s data. We must, however, treat the meaningfulness of such a full hierarchy with caution. We have seen evidence that the presence of numerals on the list can be biased by the type of contact situations that are selected, and the presence or absence of a tradition of literacy in the recipient languages. The presence of pronouns on the list is largely a product of the function of terms of address and their degree of lexicalization, or alternatively of the clash of systems that distinguish exclusive/inclusive reference, and those that don’t. These and more are coincidental circumstances that influence the contact situation around particular categories, and which will promote or demote certain categories on the hierarchy, depending on the degree of presence of languages in the sample that answer to certain sociolinguistic and structural criteria. Thus, the more reliable hierarchies are those that provide a picture of the susceptibility of category values to borrowing, while the comparison among categories is not entirely free of arbitrary factors.
10. Concluding remarks Noteworthy is the extent of borrowing across the different languages of the sample. If we take as an indicator 36 prominent categories representing various aspects of structure – from phonology, through to morphology, unbound grammatical vocabulary, lexicon, and syntax15 – and assign scores to languages based on the number of categories that show some kind of contact influence, then the scores range from 31 (Rumungro, Guaraní), to just 6 (Yaqui) and 7 (Biak) (see Table 1). Table 1 shows the overall “borrowing score” for each language. It also shows the borrowing score among 11 categories representing Other Parts of
The borrowability of structural categories
63
Speech (OPS),16 an indicator of mostly unbound grammatical lexemes, and the proportion of OPS among the categories affected by contact. The majority of languages range within a 0.05 distance from the proportional 0.305 share of OPS among the category total, indicating that there is, on the whole, a rather predictable share of OPS among borrowed categories. Note however that some languages show a disproportionately high level of borrowing of OPS: Yaqui, Biak, Likpe, Mosetén, Jaminjung score between 0.40.5, meaning that OPS account for up to 50 percent of borrowed categories. Note that all these languages tend to have overall borrowing scores of between 615, that is, on the lower side of the borrowing range. This might be interpreted as indirect Table 1. So-called “borrowing scores”, and the proportion of Other Parts of Speech among categories affected by contact. Language
Total score
OPS score
OPS/Total ratio
Sampled Yaqui Biak Manange K’abeena Likpe Mosetén Mac. Turkish Rapanui Khuz. Arabic Kildin Saami Jaminjung Vietnamese Nahuatl Tasawaq Purepecha Western Neo-Aramaic Hup Kurmanji Kurdish Yiddish Domari Otomi Quichua Indonesian Guaraní Rumungro
36 6 7 10 10 10 11 12 13 13 15 15 17 18 18 19 20 21 21 22 24 25 26 26 31 31
11 3 3 3 3 4 5 3 3 4 5 7 6 5 7 7 6 5 6 7 9 9 8 9 10 10
0.305 0.500 0.428 0.333 0.333 0.400 0.455 0.250 0.230 0.307 0.333 0.467 0.352 0.278 0.389 0.368 0.300 0.238 0.286 0.318 0.375 0.360 0.307 0.346 0.323 0.323
64
Yaron Matras
evidence that borrowing begins with OPS, before continuing to other categories. The lowest OPS scores, for Rapanui, Hup, and Macedonian Turkish (0.230.25), represent languages that undergo structural changes as a result of contact but in which borrowed OPS are under-represented. These figures generally confirm the predictions and observations that unbound grammatical morphemes are high on the borrowability scale compared to other categories; though they also allow us to conclude that contact influence is rarely limited to them, and that it is not impossible for a language to display even a certain amount of resistance toward borrowing of OPS. It is also interesting to note that a number of categories occupy an entirely peripheral position in the borrowing behaviour of languages in this sample. They include bound case markers, bound tense markers, bound person markers as well as in most cases unbound person markers (deixis and anaphora; exceptions being reciprocal and reflexive pronouns, and “lexicalized” pronouns as in Indonesian), demonstratives and expressions of place deixis. In the following discussion it will hopefully become clear that the absence of borrowing in these domains is not taken to mean that constraints exclude them from being borrowed. Rather, our focus is on those categories that do show a more salient and frequent tendency to be affected by borrowing, and our agenda is to explain why speakers are motivated to borrow forms and structures in those categories. The absence of borrowing among other categories may be left to be interpreted as just that: the absence of any particular motivation to converge the two systems around these particular categories. Typological features are neither excluded nor even rarely affected by borrowing in the sample. A number of languages undergo considerable typological convergence: Macedonian Turkish, Domari, North-eastern NeoAramaic, and Rumungro. Ongoing shifts in morphological typology can be detected in a number of other languages, too. Although statistically, unbound grammatical morphemes are more likely to be borrowed than typological features, there is no direct interdependency between any specific value or category that falls within these respective groups of structures. Implicational hierarchies of the kind postulated above only apply among the values of the same category. But since the likelihood of borrowing is different for different categories, there may be a quasi-implicational relationships across categories in different structural domains. Thus, since connectors are frequently borrowed, but re-structuring within the TMA domain is rare, we might expect a language that shows contact-induced re-structuring in the domain of TMA to show borrowed connectors as well. Such an expectation is based on the higher borrowing frequency of connectors.
The borrowability of structural categories
65
Despite the lack of any direct functional link between the borrowing of connectors and the restructuring of TMA categories (or any other example of contact influence), the challenging question remains, why certain categories are more susceptible to change in situations of language contact, than others. It is here that borrowing hierarchies, especially the implicational relations among paradigm values of the same structural category, can shed some light. The fact that borrowing within such categories often follows a non-arbitrary, predictable course, suggests that semantic-pragmatic features that distinguish among category values participate in motivating borrowing. The relations among borrowed values can thus help us illuminate the motivation behind borrowing, and so help us make sense of the different degrees of susceptibility of categories to the borrowing process. Let us, for this purpose, review the hierarchies. A first set of hierarchies might be grouped together based on a general notion of frequency, referential meaning, and usage context of the borrowed structural material. This group is rather diverse. The more frequent adoption of new consonants over new vowels (1) is conditioned by the mere diversity of consonants and the fact that they tend to outnumber vowels in each of the sample languages. The greater likelihood that phonemes be adopted as part of loanwords than as independent phonological features (3), is similarly a practical issue relating to the need to accommodate loanwords. Borrowing as a utilitarian enrichment of means of expression also belongs here. The greater likelihood of borrowing of nouns over other parts of speech (as expressed in 24) is a product of the likelihood of nouns to express new concepts and to name objects and institutions (see already Weinreich 1953: 37). The high borrowability of lower ordinals (12) is connected to the fact that they mark exclusivity by assigning lower figures a special lexical item rather than a grammatical derivational procedure, with borrowing supplementing the procedure of lexical creativity. Certain usage contexts may favour borrowing, if there is a close association with the contact language in certain domains. Thus, borrowed numerals are more likely to be used in formal than informal contexts (9), higher numerals as well as mathematical “zero” are more likely candidates for borrowing, being reserved primarily to more formal-institutional contexts (10), and names of days of the week are more likely to be borrowed than times of the day (21). An additional theme, which groups together another bundle of hierarchies, may be defined as accessibility, cognitive complexity, and expectedness. Low accessibility and/or high complexity correlate with the borrowing susceptibility of peripheral as opposed to core local relations (4), of higher
66
Yaron Matras
numerals (10) (reinforcing the usage-based motivation cited above), of independent (factual) embedded events over dependent (non-factual) ones (15), and of linked independent events (purpose clauses, causal clauses) over linked dependent ones (adverbial subordinations) (14). Low expectedness can be brought in connection with some of those, and in particular with the contact-susceptible properties of contrast (13), concessive subordination (14), phasal change (16), restrictive focus (17), and superlative (22). In all of these cases, the speaker’s assertive authority is potentially reduced as a result of the speaker venturing into propositional domains involving a degree of uncertainty or unexpectedness. Not unrelated are the properties around which external circumstances reduce the speaker’s confidence and control even more overtly. These include the high borrowing susceptibility of conditionality over other subordinations (14), of participant-external modality over participant-internal modals (5), of modality itself over aspect and tense, as well as of aspect/aktionsart (the internal structure of the event, independent of the speaker’s perspective) over tense, and of the future over other tenses (6), as well as of indefinites (which rely on a presuppositional domain) over deictics (which rely on the speaker’s own orientation perspective)17 (20). Finally, we find a set of hierarchies that operate at the level of the interaction, where those structures are more borrowing-prone that are more tightly connected to the emotive level of the discourse or speech act rather than to the content level of the word or phrase. Such is the case with prosodic features over segmental phonological features (2), discourse markers over other particles (18), connectors over other parts of speech (as in 24), causal argumentation over other forms of subordination (14), and even the positive answer particle over the general (i.e. also content-bound) negative particle (19). What do these three themes – accessibility/expectedness, external dependency, and interaction-level operations – have in common, and why are they especially susceptible to contact-related change? In order to answer this question, we must return to our hypothesis about what constitutes “borrowing” in the first place (cf. Section 1). Borrowing, we had said, is a strategic compromise which bilinguals adopted in conversation and which has become socially acceptable. Social acceptability is a pre-condition for change, since language is the collective, socio-cultural product and asset of a community. But there is no reason to assume that social attitudes should in any way prejudice contrastive connectors over additive connectors, or temporal indefinites over time deixis. The reason for the hierarchical arrangement of
The borrowability of structural categories
67
categories in respect of their borrowing susceptibility has to do with the first part of our definition of borrowing, namely the part that describes borrowing as a strategic compromise adopted in conversation. It is here that speakers are naturally inclined to prioritize when handling the control mechanism that selects certain (“language-particular”) structures in certain sets of interactions. Maintaining the demarcation boundary between repertoire components (or “languages”) is a burden on the mental processing of language in conversation, and yet it is a social requirement. Compromise is sought when the tension assumes its most extreme forms: when the burden of controlling the language selection mechanism coincides with other sources of tension in the interaction itself. Such tension emerges when the speaker’s assertive authority is at stake and a special effort is needed in order to win over the hearer’s confidence: When expressing unexpected chains of arguments, when contradicting or challenging presuppositions, when assuming responsibility for propositional content that lies beyond the domain of secure knowledge, or when directly intervening with hearer-sided processing by monitoring and directing turns and speech acts (e.g. through prosody or discourse markers). Since the conversational tension around such processing tasks cannot itself be reduced, bilingual speakers’ only alternative is to eliminate the need to distinguish between sub-components of their linguistic repertoire – or “languages” – and to unify the structures that trigger the appropriate processing operations. The result is a fusion of the two systems of structures around the relevant functions (see Matras 1998). The trigger for borrowing around these kinds of structures – those that cluster around the themes defined above as “accessibility”, “expectedness”, “interaction-level”, and so on – is thus neither social acceptability, nor prestige, nor gaps in the recipient language. Rather, it is the need to reduce the cognitive load when handling a complex linguistic repertoire. Social acceptability is merely an accompanying condition for spontaneous innovations to become anchored in the long-term speech behaviour of the community. In this respect, the susceptibility of a great number of grammatical categories to borrowing is pre-determined by their language processing function, and therefore universal. One of the most striking findings of the present investigation is the fact that so many hierarchies that were identified for the cross-linguistic sample presented in this volume, were a perfect or near-perfect match to those identified by Matras (2002) and by Elšík and Matras (2006) for the sample of Romani dialects in contact with a variety of different languages, and, to the extent that material for comparison was available,
68
Yaron Matras
also with hierarchies proposed for other samples. This clearly supports their universal predictive power. Moreover, the fact that a sample consisting of multiple recipient languages shows virtually the same results as a sample with a controlled recipient language (Romani) suggests that the structure of the recipient language plays only a secondary, perhaps even just a peripheral role in determining preferences of borrowing. The primary role is played by the functionality of the categories and the extent of bilingual pressure, i.e. the extent to which bilinguals need to make frequent decisions on language choice. Note that this sharpens the focus of what Thomason and Kaufman (1988) had referred to somewhat more bluntly as the intensity of cultural contact, helping us to move toward a more specific characterization of relevant patterns of communicative interaction. To be sure, the structures of the languages involved, especially the recipient language, may play a certain role in the borrowing process. But this role must be seen primarily as an imposition of constraints on what is essentially a universal process, motivated by cognitive features of language processing. Such constraints might include the presence of a competing structure on one side of the paradigm (as with the Jaminjung contrastive marker); or the availability of literacy as a factor strengthening the coherence of the recipient system and thus reinforcing demarcation boundaries and helping to resist borrowing; or indeed the presence of social attitudes that block language mixing. On the other hand, the fact that languages like Malay employ a plethora of lexical means, and not just deictic and anaphoric expressions, to refer to participants creates a motivation for renewal of this inventory of expressions and so also for borrowing in the domain of (socalled) personal pronouns, which is not typically found in languages that rely on participant deixis and anaphora. The principal conclusion that must be drawn from the above observations is that different borrowing motivations apply to different functional categories. With some, the motivation is lexical enrichment. With others, it is the fusion of elements of formal discourse with the language that dominates formal discourse, while in a series of categories the motivation is a reduction in the tension surrounding certain language processing tasks. Though neither gaps nor social prestige are primary motivators for borrowing, both are indirectly involved, as the process of “borrowing” can be defined as a license to speakers to dismantle the mental demarcation boundaries that separate their individual “languages” and, around a particular selection of categories, to make full use of their entire repertoire of linguistic structures and forms irrespective of the setting of the communicative interaction.
The borrowability of structural categories
69
Acknowledgement This chapter was written during a research visit to the Research Centre for Linguistic Typology at La Trobe University, Melbourne, made possible through a Distinguished Fellowship at the Institute for Advanced Study, La Trobe University, and an International Linkage Fellowship of the Australian Research Council.
Notes 1. I use the term “language” in quotes in this context since it is not obvious that multilingual speakers process language in the form of separate systems; it is safer to assume that multilingual speakers have an overall repertoire of linguistic forms, to which constraints are attached concerning the situations and conversational constellations in which those forms may be used, for various purposes. The notion of a separation of “linguistic systems” on the part of the language user is therefore somewhat of an abstraction. 2. We assume that borrowing always begins with at least some degree of bilingualism, however rudimentary, or at least with an exposure to settings of communicative interaction that require the selection of a separate inventory of forms and structures. Once a certain behaviour pattern is adopted by those speakers who interact in a variety of settings – and so have access to two (or more) “languages” – new forms and structures may diffuse into the speech patterns of monolinguals as well, or may survive the historical decline of widespread bilingualism. Such latter process may strengthen our ability to identify borrowings, but it is not a pre-requisite for borrowing. 3. On the problem of establishing “borrowability” on the basis of corpus frequency, see already Weinreich’s (1953: 3536) critical remarks. 4. Since the occurrence in a corpus of a low-ranking category presupposes that of the higher-ranking category, occurrences of the higher-ranking category will always outnumber those of the lower one. 5. Throughout I use the notation “greater than” (>) to denote the value that is more likely to be affected by contact induced change (in a frequency-based hierarchy), and which in an implicational hierarchy constitutes a pre-requisite for the borrowing of any item specified to the right of it and marked “lesser than”. 6. The one exception being Kriol, which has a phonological system that is very similar to Jaminjung. 7. In both Romani and Domari, genetically related material (deriving from IndoIranian postposed adverbial specifiers) undergoes a similar development toward agglutinative case markers (cf. Matras 2002). 8. The model is in fact areal, and is also shared by Persian and Western Armenian, and to some extent by Levantine Arabic as well.
70
Yaron Matras
9. Rapanui uses Spanish tiene que which seems to express both necessity and obligation; Imbabura Quichua has pudi- “can”, which could well cover both ability and possibility. 10. We also have no evidence to uphold the (frequency) hierarchy proposed by Wichmann and Wohlgemuth (forthc.) (note that prominence of strategies is arranged from left to right): light verbs < indirect insertions < direct insertion < paradigm transfer. But we have no grounds on which to challenge this hierarchy, either. 11. See Bakker (1997), however, on constraints that prevent the isolation of the Algonkian verb to a bare stem, in the context of Cree/French contact (albeit in connection with the formation of the mixed language Michif, not with borrowing in the conventional sense). 12. Like all Romani dialects, Rumungro too uses Greek derivation markers for form ordinals from cardinals, but the hierarchy applies to the borrowing of ordinal word forms. 13. Wilco van den Heuvel, p.c. 14. Compare with integrated hierarchy presented by Muysken (1981), repeated by Winford (2003): 51: nouns > adjectives > verbs > prepositions > coordinating conjunctions > quantifiers > determiners > free pronouns > clitic pronouns > subordinating conjunctions 15. The full list is: Consonants, vowels, morphological typology, alignment type, local relations, classifiers/gender, possession, plurality, definiteness, diminution/ augmentation, nominalization, case marking, tense categories, tense marking, aspect categories, aspect marking, aktionsart categories, aktionsart marking, mood categories, modal verbs, voice and valency, numerals, personal pronouns, demonstratives, indefinites, interrogatives, connectors, subordinating conjunctions, phasal adverbs, focus particles, discourse markers, time deixis, adjective comparison, constituent order, syntax, basic cultural vocabulary. 16. These are: Numerals, personal pronouns, demonstratives, indefinites, interrogatives, connectors, subordinating conjunctions, phasal adverbs, focus particles, discourse markers, time deixis, adjective comparison. 17. More precisely, indefinites can be said to engage the hearer more actively in supplementing an imaginary knowledge domain in which the missing context can be situated: consider an indefinite expression such as ‘anywhere’, where it is up to the hearer to construct an image of possible locations that satisfy vague contextual criteria. With deixis, on the other hand, the speaker is confident that speaker and hearer share a very particular perspective. Thus, ‘here’ leaves no room for ambiguity, or for hearer-sided creativity.
The borrowability of structural categories
71
References Aikhenvald, Alexandra Y. and R. M. W. Dixon (eds.) 2001 Areal Diffusion and Genetic Inheritance. Oxford: Oxford University Press. 2006 Grammars in Contact. A Cross-linguistic Typology. Oxford: Oxford University Press. Aikhenvald, Alexandra Y. 2002 Language Contact in Amazonia. Oxford: Oxford University Press. 2006 Grammars in contact: A cross-linguistic perspective. In: Alexandra Y. Aikhenvald and R. M. W. Dixon (eds.), Grammars in Contact. A Cross-Linguistic Typology, 166. Oxford: Oxford University Press. Bakker, Peter 1997 A Language of Our Own: The Genesis of Michif, the Mixed Cree– French Language of the Canadian Métis. Oxford: Oxford Univer sity Press. Campbell, Lyle 1993 On proposed universals of grammatical borrowing. In: Henk Aertsen and Robert Jeffers (eds.), Historical Linguistics 1989: Papers from the 9th International Conference on Historical Linguistics, 91109. Amsterdam: John Benjamins. Dixon, R. M. W. 1995 Complement clauses and complementation strategies. In: F. R. Palmer (ed.), Grammar and Meaning: Essays in Honour of Sir John Lyons, 175220. Cambridge: Cambridge University Press. 2006 Complement clause types and complementation strategies in typological perspective. In: R. M. W. Dixon and Alexandra Aikhenvald (eds.) Complementation. A Cross-Linguistic Typology, 147. Oxford: Oxford University Press. Elšík, Viktor, and Yaron Matras 2006 Markedness and Language Change: The Romani Sample. Berlin: Mouton de Gruyter. Field, Fredric 2002 Linguistic Borrowing in Bilingual Contexts. Amsterdam: Benjamins. Givón, T. 1990 Syntax. A Functional-Typological Introduction. Amsterdam: Benjamins. Haugen, Einar 1950 The analysis of linguistic borrowing. In Language 26 (2): 210231. Heath, Jeffrey 1984 Language contact and language change. In Annual Review of Anthropology 13: 367384.
72
Yaron Matras
Matras, Yaron 1998 Utterance modifiers and universals of grammatical borrowing. Linguistics 36 (2): 281331. 2002 Romani: A Linguistic Introduction. Cambridge: Cambridge University Press. 2004 Layers of convergent syntax in Macedonian Turkish. Mediterranean Language Review 15: 6386. Matras, Yaron and Jeanette Sakel 2007 Investigating the mechanisms of pattern-replication in language convergence. Studies in Language 31 (4): 829865. Moravcsik, Edith 1975 Verb borrowing. Wiener Linguistische Gazette 8: 330. 1978 Language contact. In: Joseph H. Greenberg, Charles A. Ferguson and Edith A. Moravscik (eds.), Universals of Human Language, Vol. 1: 93122. Stanford: Stanford University Press. Muysken, Pieter 1981 Halfway between Quechua and Spanish: The case for relexification. In: Arnold Highfield and Albert Valdman (eds.), Historicity and Variation in Creole Studies, 5278. Ann Arbor: Karoma. Ross, Malcolm 2001 Contact-induced change in Oceanic languages in north-west Melanesia. In: Alexandra Aikhenvald and R. M. W. Dixon (eds.), Areal Diffusion and Genetic Inheritance: Problems in Comparative Linguistics, 134166. Oxford: Oxford University Press Stolz, Thomas 1996 Grammatical Hispanisms in Amerindian and Austronesian languages. The other kind of Transpacific isoglosses. Amerindia 21: 137160. 2007 Allora. On the recurrence of function-word borrowing in contact situations with Italian as donor language. In: Jochen Rehbein, Christiane Hohenstein and Luaks Pietsch (eds.), Connectivity in grammar and discourse, 7599. Amsterdam: Benjamins. Stolz, Christel, and Thomas Stolz 1996 Funktionswortentlehnung in Mesoamerika, Spanisch-Amerindischer Sprachkontakt. In Sprachtypologie und Universalienforschung 49 (1): 86123. 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): 177. Thomason, Sarah G. 2001 Language Contact: An Introduction. Edinburgh: Edinburgh University Press.
The borrowability of structural categories
73
Thomason, Sarah G., and Terrence Kaufman 1988 Language Contact, Creolization and Genetic Linguistics. Berkeley: University of California Press van der Auwera, Johan 1998 Phasal adverbials in the languages of Europe. In: Johan van der Auwera with Dónall P. Ó Baoill (ed.), Adverbial Constructions in the Languages of Europe, 25145. Berlin: Mouton de Gruyter. van Hout, Roeland, and Pieter Muysken 1994 Modelling lexical borrowability. Language Variation and Change 6: 3962. Weinreich, Uriel 1953 [1968] Languages in contact. The Hague: Mouton. Wichmann, Søren, and Jan Wohlgemuth Forthc. Loan verbs in a typological perspective. In: Thomas Stolz, Dik Bakker and Rosa Salas Palomo (eds.), Aspects of Language Contact. Berlin/ New York: Mouton de Gruyter. Winford, Donald 2003 An Introduction to Contact Linguistics. Oxford: Blackwell.
Grammatical borrowing in Tasawaq Maarten Kossmann
1. Background1 Tasawaq (tásàwàq) is the main language of the date palm oasis of In-Gall (íngàl), about 100 km west of Agadez in the desert of Niger (Western Africa). It is sometimes stated that Tasawaq is also spoken in Tegidda-n-Tesemt, an important salt extraction site in the region. This is true, but only in the sense that salt exploitation is seasonal labor (Bernus and Bernus 1972: 23, 30), and that in the rest of the year the great majority of salt miners stay in their homes in In-Gall. The number of speakers is unknown, but probably lies between 2,000 and 10,000. Since 1991, Tasawaq has been recognized as an official language of Niger (Sidibé 2002: 186). This status has no practical consequences. Tasawaq is a Northern Songhay language. Songhay is a close-knit language group, which is commonly regarded to be part of the Nilo-Saharan language phylum (e.g. Bender 1997). Other affiliations have been proposed, and it constitutes one of the better candidates in Africa for an isolated family. All Northern Songhay languages have been heavily affected by language contact. In the case of Tasawaq, the main language of influence is southern Tuareg (also called Tamajeq), the language of the main nomadic group in the desert around In-Gall. Tuareg is a Berber language belonging to the Afroasiatic language phylum. Languages of minor influence on Tasawaq are Arabic (Semitic, Afroasiatic), the language of religion and in earlier times also of long-distance trade, and Hausa (Chadic, Afroasiatic), the lingua franca in this part of Niger, and the language of the only urban center in the region, Agadez. Fulfulde, spoken by another nomadic group in the desert around In-Gall, does not seem to have had any influence on Tasawaq, neither grammatically, nor lexically. According to speakers of the language most Tasawaq speakers also speak Tuareg and Hausa (Sidibé 2002: 194). On the other hand, non-native speakers of Tasawaq seem to be extremely rare. Tuareg has influenced Tasawaq, both at the lexical and the grammatical level. Lexical influence includes the introduction of many items which are considered to be ‘basic’ according to most researchers who have an opinion on this, e.g. body-part terms such as ‘finger’, ‘heart’, ‘tongue’, ‘knee’ and ‘tooth’.
76
Maarten Kossmann
Northern Songhay languages in general, and Tasawaq in particular, have been analyzed as “mixed” languages, with Songhay and Tuareg components (Nicolaï 1990, Wolff and Alidou 2001). This is based on a number of arguments. In the first place, the high degree of lexical influence, in the second place the existence of separate systems for elements of different origin in some sub-systems of the morphology of Tasawaq, and finally a number of important structural features which would have a Tuareg background, e.g. the genitive adposition Hn` and SVO word order. Neither of these features are decisive for ranking the language under the “mixed” languages. For example, Northern Berber languages such as Riffian Berber – which nobody would call “mixed” in the sense of Michif or Ma’á – have similar percentages of loanwords in basic lexicon and have separate morphological sub-systems on the basis of etymological origin. The remaining part of the argument, the structural features, are not decisive, as this type of structural borrowing is attested elsewhere in non-mixed languages, and also because the adduced evidence for borrowing is not always very strong. Therefore, I will consider Tasawaq a Songhay language which was strongly influenced by Tuareg, rather than a mixture of the two languages for which no basic language can be identified.
2. Phonology The most conspicuous phonological influence of Tuareg on Tasawaq is the introduction of pharyngealized consonants, both in Tuareg and in Songhay lexemes. In Proto-Songhay pharyngealization was certainly not present. In Songhay lexemes, pharyngealized consonants mainly occur in the vicinity of a and o; in lexemes borrowed from Tuareg, there are no restrictions on their occurrence. It should be noted, however, that according to Robert Nicolaï (p.c.), who worked with a number of different informants, in most Tasawaq idiolects (among others the one described by Nicolaï 1979), consonantal pharyngealization is absent. Further influence of Tuareg and other languages mainly concerns the distribution of certain consonants. Thus, in the Songhay part of the lexicon q and γ are in near complementary distribution, q being found in word-initial position before o and in word final position after a and o, while γ appears elsewhere. Due to the introduction of especially Arabic lexemes, in which both sounds are phonemic, this distribution has become blurred. Another conspicuous effect of language contact is the change in relative frequency of certain types of consonant cluster. In the Songhay part, most consonant clus-
Tasawaq
77
ters start with a nasal or r. Other clusters are possible, but rare. The influx of foreign lexemes from languages which are less restrictive in their clustering procedures has raised the relative frequency of the other consonant clusters considerably. There exist interesting similarities between Tasawaq and Hausa phonology. In the first place, they share a phonological rule by which the short mid vowels e and o are lowered to a in all contexts except before pause (cf. Newman 2000: 399). The underlying vowel quality reappears in contexts where the vowel in question is lengthened, e.g. Tasawaq: (1)
γáy dàr ‘I stretched out’ γáy dààr-á ‘I stretched it out’ γáy dáb ‘I closed’ γáy dééb-à ‘I closed it’ γáy dàs ‘I touched’ γáy dòòs-á ‘I touched it’
In the second place, Hausa and Tasawaq have similar tonal features. Both languages have three tones, High, Low and Falling. Different from other Songhay languages such as Zarma, but similar to Hausa, Rising tone does not exist in Tasawaq. Moreover, both languages have a strong dislike for all-Low contours in polysyllabic words. In Hausa, such contours are rare and mainly occur in loanwords (Newman 2000: 606). In Tasawaq, underlying all-Low contours (which are frequent) are automatically changed to sequences with an initial Falling tone in polysyllabic words. Because of phonotactic restrictions on the occurrence of Falling tones, it is simplified to a High tone in certain syllable types. The basic all-Low contour reappears in bound syntactic contexts, e.g. in nouns when they are followed by a numeral: (2)
bângù ‘well’ bàngù hínká ‘two wells’
It is difficult to decide whether these similarities between Hausa and Tasawaq phonology are due to Hausa influence on the language, or whether they constitute independent developments.
3. Noun morphology In its noun morphology, Tasawaq makes a strict distinction between native elements and foreign elements. This is most conspicuous in plural formation. Native Songhay words have no lexical plurals; plurality is obligatorily
78
Maarten Kossmann
marked by means of an element H-yo, which occurs in noun-phrase final position.2 This is illustrated in the following examples: (3)
a. dábdè ‘piece of clothing clothing’ (underlying form: dàbdè) b. dàbdá-yo ‘clothes’ clothes-pl c. dàbdè sídày-yo ‘red clothes’ clothes red-pl
With borrowed words, on the other hand, plurality is marked on the noun rather than on the noun phrase. These plurals cannot immediately be followed by the suffix H-yo. This suffix reappears when the noun is not the last constituent of the noun phrase, e.g. (4)
a. tákàrdè ‘(sheet of) paper’ (< Tuareg) paper b. sìkárdààwà n ‘sheets of paper’ (< Tuareg) papers c. sìkárdààwà n sídày-yo ‘red sheets (of paper)’ papers red-pl
The plural formations found in borrowed elements reflect Tuareg morphology. They can be of two types. In the first type, there are both changes in the initial syllable of the noun, and changes elsewhere – either vowel changes in the stem or suffixes. This is illustrated by the following examples: (5)
singular plural singular plural àsáágù ìsúúgà < Tuareg əsegu isuga ‘comb’ táágày sígàyà n < Tuareg tagăyt šigăyyen ‘palm frond’
The second type is characterized by the suffix -(t)à n. As in Tuareg, this type is found with a small set of Tuareg nouns, but especially constitutes the canonical way of integrating loanwords from other languages than Tuareg, e.g. (6)
singular àládày àlkìtáb
plural àládààyà n àlkìtáábà n
< Hausa àládèè ‘pig’ < Arabic al-kitaab ‘book’
Tasawaq
79
The etymological split between a class of nouns of Songhay origin, which exhibits no lexical number marking, and a class of borrowed nouns with singular–plural distinctions is relatively strong. Only eleven Songhay nouns were found, which receive plurals according to the Tuareg fashion of treating loanwords (i.e. by suffixing -(t)à n), e.g. (7)
singular plural gwánsì gwánsìtàn < Songhay ‘snake’
A certain number of borrowings from Arabic and Hausa are treated as if they were Songhay nouns, e.g. (8) singular plural àssàbí àssàbí-yo < Arabic ṣabiiy ‘child’ bàṛqòòní bàṛqòòní-yo < Hausa bàrkòònóó ‘pepper’ Another interesting influence from Tuareg is found in the marking of natural gender. In the Songhay part of the language, the only non-phrasal way of expressing natural gender is by means of suppletive stems. This is only found in a few cases; many items which would be gendered in other languages are left unmarked for gemder in Tasawaq. Examples: (9) àlzírày ízè báàbà áàrù báynà
‘male or female in-law’ ‘son, daughter’ ‘father’ náànà ‘mother’ ‘man’ wây ‘woman’ ‘male slave’ ṭààmú ‘female slave’
< Songhay < Songhay < Songhay < Songhay < Songhay
Tuareg has a derivative gender system, in which masculine denotes male animates and larger (in relation to the corresponding feminine form) inanimates, while feminine denotes female animates and smaller inanimates. In Tasawaq, this system is reflected in a consistent manner in borrowings denoting animates. The size difference with inanimate concrete nouns, on the other hand, is only reflected in a few lexicalized pairs, and cannot be considered a feature of Tasawaq grammar. The natural gender difference is illustrated by the following pairs: (10) àbóóbàz àgéélìm ááràb
‘male cousin’ tàbóóbàz ‘female cousin’ < Tuareg ‘male orphan’ tàgéélìm ‘female orphan’ < Tuareg ‘Arab man’ tááràb ‘Arab woman’ < Tuareg
80
Maarten Kossmann
4. Adjectives Songhay languages have a special class of words, which are only used as nominal modifiers. I will refer to them as “adjectives”. Many of them are derived from verbs by special morphological devices. Tuareg, on the other hand, has no class of adjectives. The type of modification carried out by adjectives in Songhay, is provided by relative clauses in Tuareg. In subject relatives, Tuareg uses a special form of the verb, the so-called participle. With stative verbs, which are very common in Tuareg adjective-like relative clauses, the participle (m.sg.) is marked by a suffix -ăn. Tasawaq has retained the Songhay system of modification with adjectives, most of which are morphologically derived from verbs. There are several regular formations of adjectives. Adjectives based on monosyllabic verbs (always with the shape CVC) are formed by lengthening the (underlying) vowel of the verb and adding a suffix -o. The lexical tone is replaced by a L-H tone pattern, e.g. (11) nàq fár qwáṣ zìr
‘to press’ ‘to open’ ‘to cut’ ‘to wipe’
nààqó fèèró qòòsó zììró
‘pressed’ ‘opened’ ‘cut (adj.)’ ‘wiped’
A similar morphological device is found with disyllabic verbs ending in a vowel. In these verbs, the final vowel is substituted by -o, and the lexical vowel pattern is replaced by a L-H contour, e.g. (12) síírí ‘to be crooked’ sììró ‘crooked’ fúmbú ‘to stink’ fùmbó ‘stinking’ qwàrnó ‘warm’ qwárnò ‘to be warm’ These two related morphological devices reflect Songhay patterns. All verbs to which they apply have a Songhay origin. The other verb types have a different morphology, which consists of the suffixing of -à n, under some circumstances accompanied by vowel lengthening in the preceding syllable. The lexical tone pattern of the verb is retained. Verbs of this class include both original Songhay verbs and verbs borrowed from Tuareg, e.g.
Tasawaq
(13) bààráy kàkáy fáṛàṭ yízmàm gílìllìt fùsús
‘to change’ ‘to build’ ‘to sweep’ ‘to squeeze’ ‘to be round’ ‘be light’
bààráyà n kàkááyà n fáṛàṭà n yízmàmà n gìlíllìtà n fùsúúsà n
‘changed’ ‘built’ ‘swept’ ‘squeezed’ ‘round’ ‘light’
81
< Songhay < Songhay < Tuareg < Tuareg < Tuareg < Tuareg
The suffix -à n reflects the masculine singular suffix of the Tuareg ‘participle’ (subject relative verb form). It should be noted that Songhay verbs which have the required structure for this type of adjective formation are quite rare, while on the other hand all Tuareg loan-verbs belong to this class. Thus, although the distribution of adjectival formations is ruled by structure and not by etymology, there is a clear etymological scission between the class of -o final adjectives (all of which have a Songhay background) and the -à n final adjectives, most of which are based on Tuareg verbs.
5. Verbs The verb in Tasawaq is entirely Songhay in its structure. Different from Tuareg, mood, aspect and negation are not marked in the verb stem. As in other Songhay languages, these categories are expressed by porte-manteau morphemes, which immediately precede the verb. The positive Perfective is unmarked. (14) γá b-sì ṭàkááfùṛ. 1sg imperfective-speak French ‘I speak French.’ Unlike other Northern Songhay languages, the subject is not obligatorily expressed by a pronoun when a lexical subject is present. The Tasawaq construction probably reflects proto-Songhay at this point, e.g. (15) sáy í-n nààná-yo sìní … well 3pl-of mother-pl say ‘And their mothers said …’ Phrases with a lexical subject followed by a subject pronoun are quite frequent, e.g.
82
Maarten Kossmann
(16) àžéémùr à ṇáṣ calf 3sg be.fat ‘The calf is fat.’ This may represent influence from Tuareg, which has obligatory subject inflection, or Hausa, where the subject is obligatorily expressed in a portemanteau morpheme combining pronominal and aspectual information. Tuareg verbs are borrowed according to two strategies. In about half of the borrowed verbs, a form without any reflex of Tuareg person–number affixes is used. This class includes all verbs with more than two syllables as well as a number of disyllabic verbs, e.g. (17) gílìllìt ‘to be round’ < Tuareg gələllət ‘be round’ (Short Imperfective)3 In the other half, the Tuareg 3sg:m prefix y(ə)- is taken over as y(i)- preceding the verb form. This is only found in disyllabic verbs. In Tasawaq the initial y does not refer to person, but is part of the verb stem, e.g. (18) γáy yízmàm < Tuareg y-əẓmăm 1sg press 3sg:m-press:perfective ‘I pressed.’ ‘He pressed.’ In Tuareg, aspect is marked by different vowel patterns in the verb stem. This provides us with the opportunity of deciding which Tuareg aspectual form was used as the basis of the Tasawaq verb. With verbs without the y- prefix, this turns out to be a difficult question, as all kinds of vowel patterns are found and no specific form can be discerned as the basis of borrowing. With y- initial verbs, on the other hand, one finds an interesting distribution (see Kossmann fc.). Tasawaq y- initial verbs which refer to actions reflect the (Positive) Perfective aspectual forms of Tuareg. They all share similar vowel patterns (mainly i ~ a, reflecting the Tuareg Perfective scheme ə ~ ă) and a High–Low tone pattern (reflecting Tuareg penultimate stress), e.g. (19) yígmàm ‘to chew tobacco’ < Tuareg ‘y-əgmăm 3sg:m Perfective yílmàq ‘to swim’ < Tuareg ‘y-əlmăγ 3sg:m Perfective yínḍàb ‘to shoot’ < Tuareg ‘y-ənḍăb 3sg:m Perfective Tasawaq stative y-initial verbs reflect another Tuareg aspectual form, the Resultative. This is shown by the vowel pattern (mainly i ~ a, reflecting the Tua-
Tasawaq
83
reg Resultative scheme ə ~ a), and by the Low–High tone pattern, reflecting Tuareg final stress, e.g. (20) yìgdá ‘to be right’ < Tuareg yəg’da 3sg:m Resultative yìggád ‘to be shy’ < Tuareg yəg’gad 3sg:m Resultative yìláz ‘to be ugly’ < Tuareg yə’laz 3sg:m Resultative One remarks that in Tuareg every verb may appear in any of the aspectual stems. In Tasawaq, this aspectual difference has been lexicalized as a difference in tone class. Both Tuareg and Songhay use affixation for verb derivation. In Tasawaq only one Songhay-based verb derivation exists, the causative suffix -n`dá, for example, (21) káání ‘to sleep’ káán-ìndá ‘to put to sleep’ (< Songhay) In the course of borrowing, many originally derived Tuareg verbs have been borrowed into Tasawaq. In a few cases both a derived verb and an underived verb have been borrowed. This is relatively rare, and far from systematic. Tasawaq does not have the intricate interaction between underived Songhay-based verbs and derived Tuareg-based verbs as found in some other Northern Songhay languages, such as Tadaksahak (Christiansen and Christiansen 2002).
6. Verbal nouns The morphology of Tasawaq verbal noun formation is split according to the etymological origin of the verb. Most Songhay verbs derive their verbal noun either by zero derivation (in one specific class tonal change), or by the suffixation of a suffix -yo (tone uncertain), e.g. (22) verb ḅáq ḍà n bá n
verbal noun ḅáq ḍâ n bá n-yo
origin ‘to break / the breaking’ Songhay ‘to sing / song’ Songhay ‘to finish / end’ Songhay
Tuareg verbs are taken over together with the corresponding Tuareg verbal noun. The great deal of variation and irregularity in Tuareg verbal noun formation is reflected in Tasawaq, e.g.
84
Maarten Kossmann
(23) verb fáṛàṭ kírìnzìt yìlkám zìrgín
verbal noun àfáṛàṭ àkírìnzì àlákàm tàzárgàn
‘to sweep/sweeping’ ‘to claw/clawing’ ‘to follow/following’ ‘to be dirty/dirtyness’
origin Tuareg Tuareg Tuareg Tuareg
7. Other word classes The only obvious influence of Tuareg on the pronominal system of Tasawaq is the introduction of the reflexive element ímà n ‘self’ (< Tuareg iṃan ‘self, spirit’, constructed with a genitival phrase (‘his self’), e.g. (24) γá b-ṣíṛìnkìṭ γá nn ímà n-sí. 1sg imperfective-comb 1sg of self-to ‘I am combing myself.’ The corresponding constructions in Tuareg and in Songhay are similar, both using a genitival phrase. Songhay languages normally have a noun meaning ‘head’ where Tasawaq uses ímà n. Numerals 4 to 10, as well as the decades, are borrowed from Arabic; the numerals 100 and 1000 are borrowings from Tuareg. There is an interesting syntactic difference between the cardinal numerals from 1 to 19 and the cardinal numerals from 20 onward. In the first group, the numeral follows the head noun, thereby reflecting Songhay syntactic patterns, e.g. (25) bàngù hínká bàngù sábàγà
‘two wells’ ‘seven wells’
In the second group, the numeral precedes the head noun and is linked to it by means of a particle Hn`, e.g. (26) γàssìrín in ṭáṛṛày ‘twenty roads’ xàmsín ìn bàngù ‘fifty wells’ téémàdá n bàngù ‘a hundred wells’ This closely reflects the Tuareg pattern, in which the genitive preposition n occurs between the higher numeral and the noun.
Tasawaq
85
Different from Tuareg practice, in both constructions the quantified noun has the singular form (thus ṭáṛṛày ‘road’ rather than ṭáṛṛààyà n ‘roads’). For most other quantifiers no certain loan origin could be discerned, the main exception being àlkúl ‘every’, which was borrowed from Arabic. Simple postpositions all derive from Songhay. A possible exception is the genitival adposition Hn, which in Tasawaq relates a possessor (in first position) to a possessed (in last position), e.g. (27) hááwí n gí cow of grease ‘grease of a cow’ Other Songhay languages mostly use Possessor–Possessed constructions without a linking adposition. As Tuareg has a genitival preposition n, it is generally assumed that this was borrowed into Tasawaq (as well as into the other Northern Songhay languages). In the opinion of the present writer, this is far from evident. In the first place, the Tuareg construction has the inverse structure, i.e. Possessed–Possessor. In the second place, the tone pattern of the Tasawaq adposition remains unexplained if one considers it a borrowing. I consider a Songhay origin (maybe somehow related to the old Songhay genitival pronoun wánè ‘that of’) a serious alternative. Many spatial relationships are expressed by means of complexes containing a preposition and a nominal element. This nominal element is often a borrowing from Tuareg, e.g. (28) kùsú nn ámmàs < Tuareg aṃṃas ‘the inside’ pot of inside ‘inside the pot’ A number of coordinating and subordinating particles have been borrowed. These include most coordinators: kó ‘or’ (< Hausa), mé ‘or’ (< Tuareg), àmmá ‘but’ (< Hausa < Arabic); the preposition n`dá ‘and, with’ (mainly used in NP coordination), however, has a Songhay background. Subordinating particles less commonly have a foreign origin; I have noted wàlá ‘even if’ (< Tuareg < Arabic) and compound tún gá ‘because’ (from Hausa tún ‘because’ and Songhay gá ‘on’).
86
Maarten Kossmann
8. Syntax All Northern Songhay languages have relatively strict S–Aux–V–O word order, while most Songhay languages have an alternation of S–Aux–O–V word order with less frequent S–Aux–V–O word order, mainly found with low-transitivity two-place verbs (Heath 1999: 161). Because of this difference in word order it has sometimes been claimed that Northern Songhay changed its word order under the influence of Tuareg. This claim is difficult to substantiate. In the first place, Tuareg is a VSO language, so the influence would have to be indirect. In the second place, S–Aux–V–O is also attested in Songhay outside the Northern group, in Timbuktu Songhay (Koyra Chiini; Heath 1999b). One part of syntax where Tuareg influence is conspicuous is the formation of relative clauses. In Songhay languages outside Northern Songhay, relative clauses are marked by means of a relative particle kaŋ. This marker is quite similar to a relative pronoun: in adpositional relatives the postposition follows kaŋ, and in most Songhay languages there is no (other) pronominal reference to the head noun in the relative clause. In Tuareg, a different system is found. The relative clause is marked by the use of certain verbal forms (the ‘participle’ with subject relatives) and certain syntactic features (esp. clitic fronting). There is no pronominal reference to the head of the relative clause, except with subject relatives: the ‘participle’ is marked for number and gender. There exists a three-way opposition between relative clauses neutral to definiteness, overtly definite relative clauses, and overtly indefinite relative clauses (Galand 1974). The neutral relative clauses have no overt linking to the head noun, while the definite and indefinite relatives use different pronominal markers as a linking device. In Tasawaq there exist two relative constructions, a construction without relative linker, and a construction which uses a pronominal element à-γó, lit. ‘this one’. In the construction without linker, the head noun may bear the normal deictic clitic -γo, ‘this’ when it is definite; when the head noun is indefinite, no elements come in between the noun and the relative clause. The head of an à-γó relativization is always indefinite. The following examples of subject relatives illustrate this: (29) ààrù-γó [gáw àssáγàl né bí] M à sí. man-proximal [work work here yesterday] PN 3sg be ‘The man [that worked here yesterday] is M.’ (definite head; no relative linker)
Tasawaq
87
(30) γáy gùn(á) bàrá-fó [b-gáw àssáγàl]. 1sg see man-one [imperfective-work work] ‘I saw a man [who was working].’ (indefinite head; no relative linker] (31) áfàẓò hún-kàt [àγó fìrízì]. reed grow-ventive [relative green] ‘Green reed grew.’ (indefinite head; relative linker] The constructions and differentiations are quite similar to Tuareg patterns. Tuareg patterns are also reflected in the way Tasawaq treats extraction from a postpositional phrase (for Tuareg cf. Heath 2005: 633ff.). Both Tuareg and Tasawaq in such cases put the adposition (without proniminal reference) on the left edge of the relative clause. Depending on the type of relative clause, this adposition follows a relative particle or stands alone, for example, (32) γá b-ní n àššááhì [kúná súúkàr à ssí]. 1sg imperfective-drink tea [on sugar 3sg be.not] ‘I drink tea without sugar.’ (Lit. I drink tea [on which there is no sugar]) (33) γáy gùn(á) tùgúzì-fóó [àγó gá àssàbí bárà]. 1sg see tree-one [relative on child be] ‘I saw a tree [on which there was a child].’ 9. Conclusion Tasawaq is strongly influenced by Tuareg and, to a lesser extent, Arabic and Hausa, both in its matter and in its patterns. In lexical borrowing, one finds an interesting feature, which consists of the introduction of foreign morphology through borrowing. Thus in nominal plural formation and in verbal noun formation, elements of different etymological origin have completely different morphologies, reflecting their original morphology. Something similar has probably been the background of the distribution of two adjective formations, one using Songhay morphology, the other using borrowed Tuareg morphology. Pattern borrowing is also found a lot. As the reconstruction of Songhay syntax is a difficult matter, it is often difficult to decide whether certain patterns derive from Songhay (with changes due to internal factors), or whether they have been taken over from Tuareg.
88
Maarten Kossmann
Notes 1. All Tasawaq data in this chapter were collected by the author during fieldwork in Niger in Fall 2003. The fieldwork was carried out with one single informant, Mrs. Ibrahim, born Nana Mariama Aweïssou, a school teacher in her twenties, originary from In-Gall, now resident in Agadez. I wish to thank her for her time and patience. Mrs. Ibrahim has fluent command of Tasawaq, Hausa and French, but does not speak Tuareg. Her idiolect is unusual, it seems, in a number of respects, esp. the presence of phrayngealized consonants, which most speakers of Tasawaq seem to lack (Robert Nicolaï, p.c.). The chapter was written in the framework of the NWO (Netherlands Organization for Scientific Research) research project “Tuareg and the Central Sahelian Languages. A History of Language Contact”. Other sources on Tasawaq are Alidou (1988), Nicolaï (1979, 197984, 1980), Wolff and Alidou (2001). I thank Robert Nicolaï, who gave me the opportunity to listen to his Tasawaq recordings, which helped me a lot in clarifying a number of questions. Transcription follows general practice in the field of African linguistics, with the exception of the following. Nasalization of a vowel is indicated by a superscript n following the vowel. Nasalization is in most cases an allophone of the nasal n, but is unsystematic in word-final position. Pharyngealization is indicated by a subscript dot. Superscript H indicates a floating High tone. In Tuareg transcriptions ă indicates a short low central vowel. Tone is consistently marked; absence of tone marking reflects the author’s uncertainty about the transcription. 2. The tone of the final syllable is difficult to hear, and has been left unmarked in the examples. The application of the preceding floating High tone, on the other hand, is well established. 3. For ease of reference, I adopt the terminology in Heath (2005) for the different aspectual stems. Other researchers on Tuareg use different terms.
References Alidou, Ousseïna 1988 Tasawaq d’In-Gall. Esquisse linguistique d’une langue dite “mixte”. Mémoire d’Études et de Recherches sous la direction de Prof. Dr. Ekkehard Wolff, Université de Niamey. Bender, M. Lionel 1997 The Nilo-Saharan Languages. A Comparative Essay. Munich: Lincom. Bernus, Edmond, and Suzanne Bernus 1972 Du sel et des dattes. Introduction à l’étude de la communauté d’In Gall et de Tegidda-n-tesemt. Études nigériennes 31. Niamey: Centre Nigérien de Recherches en Sciences Humaines.
Tasawaq
89
Christiansen, Niels and Regula Christiansen 2002 Some verb morphology features of Tadaksahak. SIL Electronic Working Papers. (http://www.sil.org/silewp/abstract. asp?ref=2002005). Galand, Lionel 1974 Défini, indéfini, non-défini: Les supports de détermination en touareg. Bulletin de la Société de Linguistique de Paris 69 (1): 205224. Heath, Jeffrey 1999a A Grammar of Koyraboro (Koroboro) Senni. Cologne: Rüdiger Köppe. 1999b A Grammar of Koyra Chiini. Berlin/New York: Mouton de Gruyter. 2005 Tamasheq (Tuareg of Mali). Berlin/New York: Mouton de Gruyter. Kossmann, Maarten Forthc. The borrowing of aspect as lexical tone: y-initial Tuareg verbs in Tasawaq (Northern Songhay). To appear in Studies in African Linguistics. Newman, Paul 2000 The Hausa Language: An Encyclopedic Reference Grammar. New Haven/London: Yale University Press. Nicolaï, Robert 1979 Le songhay septentrional (études phonématiques). Bulletin de l’IFAN, 41, série B; part 1: 304370; 539567; 829866. 197984 Sur la phonologie des langues “mixtes” du songhay septentrional. Comptes rendus du GLECS, 2428: 395412. 1980 Le songhay septentrional (études prosodiques). In: Itinérances … en pays peul et ailleurs. Mélanges réunis à la mémoire de Pierre François Lacroix, I, 261289. Paris: Société des Africanistes. 1990 Parentés linguistiques (à propos du songhay). Paris: Éditions du CNRS. Prasse, Karl-G., Ghoubeïd Alojaly, and Ghabdouane Mohamed 2003 Dictionnaire touareg-français (Niger). Copenhagen: Museum Tusculanum Press. Sidibé, Alimata 2002 Analyse critique de quelques opinions sur l’idiome des isawaghan: Le tasawaq. Mu ƙara sani. Revue de l’Institut de Recherches en Sciences Humaines (Université Abdou Moumouni, Niamey) 10/12: 185197. Wolff, H. Ekkehard, and Ousseïna Alidou 2001 On the non-linear ancestry of Tasawaq (Niger). Or: how “mixed” can a language be? In: Derek Nurse (ed.), Historical Language Contact in Africa, 523574. (Special volume of Sprache und Geschichte in Afrika 16/17).
Grammatical borrowing in K’abeena Joachim Crass
1. Background K’abeena is a Highand-East Cushitic (HEC) language spoken by some 35,000 people living in and around the town of Wolkite. This small town is situated some 160 km south west of Addis Ababa, the capital of Ethiopia, at the northwestern edge of the Gurage region. The Gurage region, a settling of people speaking South-Ethiosemitic languages, is surrounded by areas that are inhabited by people speaking Cushitic languages. These Cushitic languages are – apart from K’abeena – Alaaba, Hadiyya, Libido, and Oromo. The closest related languages to K’abeena are Alaaba, Kambaata and T’imbaaro. These four languages form one subgroup of HEC. The other four subgroups are (1) Burji, (2) Gedeo, (3) Sidaama, and (4) Hadiyya and Libido. K’abeena is a member of the Ethiopian linguistic area (ELA). It is in contact mainly with the South-Ethiosemitic languages Amharic, Chaha, Ezha, Muher, Wolane and with the Lowland East Cushitic language Oromo. Oromo is spoken to the north, Chaha to the south, Ezha to the southeast and Muher and Wolane to the east of the K’abeena speaking area. However, this is only an approximation. In fact, the main town Wolkite and the surrounding rural area are multi-ethnic, i.e. no clear boundaries can be drawn between settling areas of various groups. Almost all K’abeena speak Amharic as a second language, which is the lingua franca of Ethiopian towns (Meyer and Richter 2003) and of the Gurage region. The other contact languages are known by individuals who live or grew up in the respective contact area. Many K’abeena speak three or four languages. The South-Ethiosemitic languages belong to different subgroups of SouthWest-Semitic. Amharic belongs to one branch of Transversal South Ethiopic, Wolane to the second branch of Transversal South Ethiopic, called East Gurage. Chaha, Ezha and Muher belong to different groups of Outer South Ethiopic (Faber 1997: 6). HEC and the Gurage languages form a sub-area of the ELA, which we name HEC-Gurage sub-area. The existence of this sub-area is proposed by Zaborski (1991), who calls it Gurage-Sidamo sub-area. Unfortunately Zaborski does not discuss its features (for recent discussion cf. Bisang 2006, Crass and Bisang 2004).
92
Joachim Crass
The K’abeena reached their actual settling area in the second half of the nineteenth century, after splitting of from the Alaaba (Braukämper 1980). Culturally the K’abeena do not differ considerably from the people speaking Ethiosemitic languages. The staple food is Ensete edulis, known as “false banana”. Whereas the K’abeena are exclusively Muslims, the Ethiosemitic peoples are mainly but not exclusively either Christians or Muslims. This chapter deals with features found in K’abeena and other languages of the area. However, in most of the cases it is difficult to define the source of these features. Several features represent grammaticalization processes or pattern-replication. However, this does not mean, that these features should be excluded as contact-induced features. Especially in the case of rare or unattested grammaticalizations, contact-induced change is one possible way of explaining the similarities (cf. Bisang 1996, Heine 1994, Heine and Kuteva 2003). Furthermore it is not possible to decide whether these features belong to the ELA or to the HEC-Gurage sub-area of the ELA. The other languages dealt with in this chapter are Amharic, Libido, Oromo, Muher, Wolane, Gumär (Outer South Ethiopic) and Zay (East Gurage). Gumär and Zay are not direct contact languages of K’abeena but they are members of the HEC-Gurage subarea. The examples to illustrate the features are all taken from K’abeena. The data of all languages is an outcome of research conducted by the author of this chapter and by Ronny Meyer.1 The research was initiated by Crass in undertaking a comparison between K’abeena and Amharic. It later was extended to other languages of the HEC-Gurage sub-area. Crass provided data on Libido, Meyer provided data on Gumär, Muher, Wolane, Zay, and Oromo. The findings of this research will be published in Crass and Meyer (2007).
2. Phonology Within the sound system system of K’abeena /p/ is most probably introduced due to contact with Ethiosemitic languages via loanwords, to which this marginal morpheme seems to be restricted. The same is true for [ä] which I do not consider to be a phoneme in K’abeena; rather, it is a phonetic variant of /a/. The distribution is not easy to formulate (for a discussion cf. Crass 2005: 25f.). A phonological feature of the ELA is the presence of ejectives (for extensive discussion cf. Crass 2002). K’abeena possesses four ejectives, namely the ejective plosives /p’/, /t’/ and /k’/ and the ejective affricate /c’/. For Proto-HEC only /k’/ is reconstructed by Hudson (1989: 11). On the basis of this finding, one can suppose that ejectives were introduced into the consonant inventory
K’abeena
93
in course of time, possibly due to contact. However, there is no evidence that ejectives were introduced in K’abeena in recent time. If Hudson’s reconstruction holds true, the introduction of ejectives must me very old. Palatalization of alveolar consonants to post alveolar or palatal consonants as a morphophonological process is another areal feature, which might have spread due to contact. In K’abeena the following palatalizations occur: alveolar plosive /t/ to post-alveolar affricate /c/, alveolar plosive /d/ to post-alveolar affricate /j/, alveolar ejective plosive /t’/ to post-alveolar ejective affricate /c’/, alveolar fricative /s/ to post-alveolar fricative /sh/, alveolar fricative /z/ to postalveolar fricative /zh/, and the alveolar nasal /n/ to the palatal nasal /ñ/. In Amharic, in addition, the alveolar lateral /l/ is palatalized to the approximant /y/.
3. Nominal structures The ablative case marker, which itself is not considered to be a result of language contact, is used with verbs to form ‘since’-temporal and real conditional clauses in all languages except Oromo (for the grammaticalization of an ablative case marker to a ‘since’-temporal clause marker, see Haspelmath 1997: 66ff., Heine and Kuteva 2002: 35). In Muher and Gumär the use of the ablative case marker is restricted to ‘since’-temporal clauses. In K’abeena the ablative case marker is the suffix -VVcci. Example (1) includes the function as ablative case marker and as marker for ‘since’-temporal clauses, example (2) the function as real conditional clause marker. (1)
jarman-íicci ’ameeccoomm-íicci kabare. Germany-abl come.prv.1s-abl today.gen ’agana saaminta ’ikko. month.acc week.acc be.prv.3s.m ‘It is five weeks ago since I came from Germany.’
(2)
c’aata k’ama’yoomm-íicci ’óssuti ’affaa-’e-ba. khat.acc chew.prv.1s-abl sleep.nom hold.ipv.3s.f-1s-neg ‘If I chew khat I cannot sleep.’
4. Verbal structures The past tense marker is used in K’abeena and in all other investigated languages of the area to mark the apodosis of irreal or counterfactual conditional
94
Joachim Crass
clauses. This is remarkable because a past marker mostly occur in the protasis (Fleischman 1989: f.) In K’abeena the past tense marker is the suffix -kk’i. Example (3) shows the function as past tense marker, example (4) and (5) the function as a marker of the apodosis. (3)
’anni bokkóoni ’agáre-’e yiye-he father.gen house.loc wait.imp.s-1s say.cnv.1s-2s ’oro’yoommi-kk’i-ba-indo? go.prv.1s-pst-neg-q ‘Didn’t I leave by saying to you: “Wait in the house of my father!”?’
(4)
beréta t’eenoo ’ubbo-ba-’ikkáani t’aafaa yesterday rain.nom fall.prv.3s.m-neg-cnd Tef.acc ’udunnáammi-kk’i. thresh.ipv.1s-pst ‘If it had not rained yesterday we would have threshed Tef.’
When no adverb indicates tense, a past or a non-past interpretation is possible. (5)
t’eenoo ’ubbo-ba-ikkáani t’uma-ha-kk’i. rain.nom fall.prv.3s.m-neg-cnd good-cop-pst ‘It would have been good if it did not rain.’ ‘It would be good if it did not rain.’
The prospective aspect (cf. Comrie 1976: 64f.) is marked in K’abeena by a verbal noun in the dative (cf. example (6)) and in Ethiosemitic languages by a verbal noun with possessive suffixes. In both cases the verbal noun is followed by a copula. In Ethiosemitic languages two morpheme orders occur. Either the copula precedes the possessive suffix (e.g. Wolane) or it follows the possessive suffix (e.g. Muher). (6)
’áni timhirtíta shuuliiháa-ti. 1s.nom study.acc finish.vn.dat-cop ‘I am about to finish my studies.’
The prospective aspect is distinguished formally by the intentional. The latter is expressed with a subordinate clause followed by a copula. In K’abeena, the
K’abeena
95
predicate of the subordinate clause is a converb expressing purpose (example (7)). In Ethiosemitic languages the non-independent imperfective is marked either with a purpose or a locative marker. (7)
hokkoppaati ’intotáa-ti. afternoon.snack.acc eat.cnv.purp.1p-cop ‘We intend to eat our afternoon snack.’
A formal distinction between prospective aspect and intentional is found in K’abeena, Amharic, Gumär, Muher, Wolane, and Zay but not in Libido and Oromo. In the latter two languages the prospective aspect and the intentional are not differentiated morphosyntactically. Libido uses a converb plus a copula for both categories, i.e. the same construction used in K’abeena only for the prospective aspect. In Oromo, however, a construction consisting of a verbal noun in the dative or a verbal noun marked with a possessive morpheme followed by a copula is used. The experiential perfect is expressed by a construction including the respective verb ‘know’ in the main clause and its complement expressed by a converb. The latter expresses the event experienced by the subject. (8)
’ameerikáani ’orootéeni kasseenta-’i? ’ee, ameerikáani America.loc go.cnv.2p know.prv.2p-q yes America.loc ’oróo’ni kansóommi. go.cnv.1p know.prv.1p ‘Have you ever been in America?’ ‘Yes, we have [the experience to have] been in America.’
All languages except Oromo lack a verb ‘have’. To express possession the respective verb ‘to exist, to be present’ is used. The possessor is marked with the dative case, the possessum with the nominative. (9)
kii bíkku c’úulu yóo-’e. 2s.gen size child.nom exist.3-1s.obj ‘I have a child of the same age/size as yours.’
This construction is used in all languages except Libido to express obligation, too. In this case the subject is a verbal noun. Libido cannot make use of this construction, because it does not have object agreement markers.
96
Joachim Crass
(10) ’oró’u yóo-’e. go.vn.nom exist.3-1s.obj ‘I have to go.’ Converbs occur in many languages of the ELA. Typical is that converbs usually cross-reference the subject. According to Tosco, the presence of the converb in most Ethiopic Semitic languages is probably the result of old Cushitic influence … Within Cushitic, the converb is typically found in the languages of the highlands … It is likewise found in the Omotic languages of the highlands, … and can therefore be considered a genuine areal feature. (Tosco 2000: 345)
The number of converbs differs considerably in the languages. Whereas in Ethiosemitic languages generally one or two converbs occur, the number of converbs in HEC languages is much higher. In K’abeena two mainly sequential converbs, one progressive converb, one negative converb, one affirmative and one negative purpose converb occur. Furthermore, K’abeena marks objects on the verb according to the primary/secondary object pattern, i.e. in the case of ditransitive verbs the recipient is marked on the verb, not the theme. This pattern is attested also for Ethiosemitic languages. (Ethio-)Semitic verbs are integrated into K’abeena on the basis of the 3s.m of the perfective, which is the citation form of verbs in Semitic languages. This is remarkable because the 3s.m is the citation form only according to scientific tradition, not in the language use. Normally, the citation form is the verbal noun. However, the 3s.m seems to have a similar function in the language use. The final vowel in (Ethio-)Semitic verbs, which marks the 3s.m, is replaced by the suffix -u which marks verbal nouns, the citation form of verbs in K’abeena. Example: From the Amharic verb kättäbä ’he vaccinated’ the final vowel -ä is replaced by the suffix -u. Therefore, the citation form in K’abeena is kattabu ‘vaccinate’. Other examples are K’AB ’anabbabu ‘read’ from AMH anäbbäbä, K’AB tamaaru ‘lern’ from AMH tämarä, K’AB t’aafu ‘write’ from AMH ts’afä or t’afä. When the second radical of a three-consonantal verb is not geminated in K’abeena, the verb is definitely not borrowed from Amharic, because in this language the second consonant is always geminated in the perfective. The source can only be one of the Gurage languages. However, in these languages, the situation is not homogeneous. Consonant
K’abeena
97
gemination vs. non-gemination is a feature which yields different verb classes. Furthermore, in some cases, languages can be excluded as source, because sound shift took place. An example is the K’abeena verb faradu ‘judge’. In Dobbi, Muher, Mäsqan and Soddo (like in Amharic) the verb is färrädä. In Chaha the cognate verb is fänädä, in Ezha fännädä and in Endegegn, Silt’e, Wolane and Zay färädä (Leslau 1979: 241). Since Wolane is the only contact language of K’abeena, the verb most probably is borrowed from Wolane. The source of verbs of the religious domain is often unclear. They may be borrowed either directly from Arabic or via Amharic or another contact language. An example is saggadu ‘pray’, borrowed from Amharic säggädä or another language with gemination of this verb in the perfective. Here Arabic can be excluded as direct contact language because the verb is realized as sadžada. i.e. without gemination of the second consonant, which in addition is a palatal affricate and not a velar plosive. In other cases the source is not clear. The K’abeena verbs katabu ‘write’ and k’ara’u ‘read’ might be borrowed directly from the Arabic verb kataba und qara’a. The two verbs katabu and k’ara’u are used mainly in religious contexts or by religious people, the verbs t’aafu ‘write’ and ’anabbabu ‘read’, borrowed from Amharic, in nonreligious contexts.
5. Other parts of speech Complementizers are grammaticalized out of a similative marker, which is not considered to be the result of language contact. Primarily they mark complement clauses (example (11)) but they may be used to mark affirmative and negative purpose clauses as well (example (12) and (13)). This feature is found in K’abeena, Libido, Oromo, Amharic, Gumär, Muher, Wolane, and Zay. However, in all languages it is not the main type to mark purpose clauses. (11) moggó-gga híilu ríccu yoo-ba. theft-simil bad thing.nom exist.3-neg ‘There is nothing being as bad as theft.’ (12) shukúru ga’áta ’ameetánu-gga dagáammi. Shukur.nom tomorrow come.ipv.3s.m-simil know.ipv.1s ‘I know that Shukur will come tomorrow.’
98
Joachim Crass
(13) ’áti k’ama’áanti-gga k’aawwáanka-si buuru 2s.nom drink.ipv.2s-simil coffee.loc-3s.m butter.acc wartó-si. put.prv.3s.f-3s.m ‘She put butter into the coffee in order that you may drink it.’ According to Heine and Kuteva (2002: 91) the “directionality proposed here [i.e. the grammaticalization of a complementizer to a marker of purpose clauses, J.C.] has not yet been established beyond reasonable doubt. More data to substantiate this hypothesis are required”. The fact that the grammaticalization of a complementizer to a marker of purpose clauses occurs in the languages of the HEC-Gurage sub-area shoes that this grammaticalization is more frequent than has been considered to be. However, the overall rarity makes it reasonable to consider the occurrence in this area to be due to language contact. The (Ethio-) Semitic noun wäk’t/wak’t ‘time’ is used in K’abeena as the head of a relative clause to form a temporal clause. (14) kesa da’iyoommi wak’ti tassh yiyo-’e. 2s.acc meet.prv.1s.rel time.acc ideo say.prv.3s.m-1s.obj ‘I was happy when I met you.’ The Ethio-Semitic noun məknəyat ‘reason’ is used as the head of a relative clause to form a causal clause. (15) ’ámru dunkiyo-’i mikiññaati daggoom-ba. Amru.nom be.late.prv.3s.m-rel reason.acc know.prv.1s-neg ‘I don’t know why Amru was late.’ The (Ethio-)Semitic noun säbäb/sabab ‘reason’ is used as the head of a relative clause to form a causal clause. The head noun of the relative clause is the subject of a complement clause. (16) c’úulu-’i harafáta ’ameetu’náani child.nom-1s.pss eid.al.adha come.neg.cnv.3s.m fakk’oo sabábati ma ’ikkó-gga stay.away.prv.3s.m.rel reason.nom what.acc be.prv.3s.f-cmpl zaaññiyóommi. not.know.prv.1s
K’abeena
99
‘I don’t know what was the reason why my child did not come to the Eid al-Adha.’ In one case the Amharic denominal derivational suffix -äñña producing adjectives (and nomina agentis) is borrowed in K’abeena. Suffixed to the K’abeena verb ’it-u ‘eat’ it yields ’itañña ‘glutton, stodger’. K’abeena and most of the other languages extensively use ideophones. These ideophones are verbalized by the respective verb ‘say’. Partly, the inventory of ideophones is identical in form and meaning in K’abeena and its contact languages. In the following, I list a few ideophones of K’abeena and Amharic, including the respective citation form of the verb ‘say’ which is in K’abeena the verbal noun yu and in Amharic – according to scientific tradition, not in the language use – the 3s.m of the perfective alä. K’AB ’illimm yu, AMH əlləmm alä ‘disappear suddenly, vanish, disappear from sight’; K’AB bogg yu, AMH bogg alä ‘flare, blaze, appear suddenly (light)’; K’AB sillimm yu, AMH səlləmm alä ‘fall into a swoon, be in a coma’, K’AB will yu, AMH wəll alä ‘desire or crave something (food, drink, smoking), have a momentary strong desire for something’ (the English translations are from Kane 1990).
6. Constituent order SOV word order is an areal feature of the ELA. Most scholars consider Ethiosemitic languages to have shifted their word order because of intensive contact with Cushitic, especially central Cushitic languages.
7. Syntax Another feature found widespread in the languages of the area is the fact that copulas differ in main and subordinate clauses. In K’abeena two types of main clause copulas occur (cf. Crass 2003). One type is gender inflected, namely -ha for masculine and -ta for feminine. The other type of copula, namely -ti, is uninflected. In subordinate clauses (relative clauses, complement clauses, adverbial clauses) the fully inflected verb ’ihu ‘be’ occurs. (17) ’ísu rosisaanco-ha. 3s.m teacher.acc-cop.m ‘He is a teacher.’
100
Joachim Crass
(18) gu’mára halaale ’ikko-’i riccu hasaawwiyye! always true be.prv.3s.m-rel thing.acc talk.imp.p ‘Tell always the truth!’ [Tell always thing, which are true!]
8. Lexicon The ELA is characterized by several interesting types of lexicalizations. Hayward (1991) distinguishes three categories of lexicalizations, which he exemplifies with data on Amharic, Oromo and Gamo, an Omotic language cluster spoken in South West Ethiopia. According to Hayward (1991: 140) these lexicalizations reinforce “the very real cultural unity of Ethiopia” (cf. Hayward 2000). The three categories are (1) single-sense lexicalizations, (2) lexicalizations with two or more distinct senses and (3) lexicalizations involving similar derivations. The first category comprises “single-sense lexicalizations of typically indigenous concepts”, the second category lexicalizations “showing inter-linguistic matching across the three languages” and the third category lexicalizations with a “similar (parallel) ‘derivational pathway’.” To the first category belong mainly nouns such as lexical items for seasons of the year, categories of terrain, categories of dung/excrement, super-categories for birds, types of borrowing and skin colour classification of people of the region. Furthermore, this category includes the suppletive imperative of the verb ‘to come’ (Ferguson’s feature G17) and particles with the meaning ‘Take this!’ which have no obvious etymological relationship to a verb. The second category, lexicalizations with two or more distinct senses, is predominantly comprised of verbs and some nouns. Examples: the respective verbs have the basic meaning ‘hold, catch’ and the secondary meaning ‘start, begin’. The respective verbs with the basic meaning ‘play’ have the secondary meaning ‘chat’. The third category includes verbal derivations, compound verbs, i.e. ideophones verbalized by the respective verb ‘say’, possessive constructions including two NPs, and ideomatic expressions. Examples for verbal derivations are the causative of the respective verb ‘want’ having the meaning ‘need’, the causative of the respective verb ‘enter’ having the meaning ‘marry’ and the causative of the respective verb ‘pass the night’ having the meaning ‘administer’. Compound verbs are ‘become silent’, ‘hurry up’ and ‘jump up suddenly’. Possessive constructions including two NPs have a word by word meaning and a metaphorical meaning. Examples are ‘son of man/people’ having the meaning ‘mankind, human being’ and ‘land of man/people’ with the meaning ‘foreign country’. Idiomatic expressions are ‘regain/recover control,
K’abeena
101
take courage’ being composed of the noun ‘heart’ and the verb ‘return (intr.)’, and ‘catch cold’, of which the noun ‘cold’ is the subject and the experiencer the object of the verb ‘catch’. K’abeena shares at least several of these lexicalizations. I deal only with some lexicalizations of Hayward’s second and third category, which I consider especially interesting. The verb ’afu has the basic meaning ‘hold, catch’ and the secondary meaning ‘begin start’, the verb ’alapp’u ‘play’ the secondary meaning ‘chat’. Hayward’s examples of verbal derivations as part of the third category are attested also in K’abeena. The causative of verb hasu ‘want’ is hasisu ‘need, be necessary’, the causative of the verb ’a’yu ‘enter’ is ’a’isu ‘marry’ and the causative of the verb garu ‘pass the night’ is gasshu ‘administer’. Most of the borrowed nouns are expressions of cultural goods of different kinds (food, cloths, topics related to Islam) and some abstract nouns. Examples for abstract nouns are K’AB suusi from AMH sus ‘mania, passion, rage’ (dt. Sucht) K’AB ’umuri from ARAB cum(u)r ‘age’, K’AB keerti / hayraati from ARAB khayr (engl. ‘good, benefit’), K’AB haali from ARAB ḥaal ‘situation’ (Crass 2005: 5260). Examples for nouns of mainly typical muslim cultural heritage: K’AB maskiida ‘mosque’ from AMH mäsgid, K’AB sheet’aani ‘devil’ from AMH säyt’an or ARAB shayt’an, K’AB t’uuri ‘punishment by God which befalls a wrongdoer’ from AMH t’ur, K’AB kitaaba ‘book’ from ARAB kitaab. Examples for vegetation are K’AB ababa ‘flower’ from AMH abäba, K’AB baarzaafi ‘eucalyptus tree’ from AMH bahər zaf (Remark: AMH bahr zaf literally means ‘tree of the sea/ocean’. The lexeme bahər in compounds often designates foreign origin of an item (Kane 1990: 855). An example for geography is K’AB alamíta ‘world’ from AMH aläm. Examples for tools are K’AB akaafa ‘shovel’ from AMH akafa, K’AB billaawa ‘(kind of) knife’ from AMH billawa, K’AB faasa ‘axe’ from AMH fas. An example for minerals is K’AB work’a ‘gold’ from AMH wärk’. Examples for body parts (and related abstract nouns) are K’AB angooli ‘brain’ from AMH ang(w)äli [ango:li], K’AB k’albi ‘heart, mind, intelligence, reason’ from AMH k’älb, K’AB nafséeta/nabséeta/nafsíta/nabsíta (-éeta and -íta are different flexion classes of the noun) ‘soul, life’ from AMH näfs, K’AB sabri ‘patience’ from ARAB sabr. An example for cultural goods of different kinds is K’AB birati ‘metal’ from AMH bərät. Furthermore, there are contact phenomena in areal cultural vocabulary, especially in relation with the Ensete plant (cf. Crass and Meyer 2005). Other word classes are interjections, e.g. K’AB ciff, AMH cəff, an interjection to chase away cats and fillers, e.g. K’AB bal, AMH bäl ‘well’. In the case
102
Joachim Crass
of interjections to call or chase away animals a remarkable correspondence is attested in many languages of the area. An interesting case is the K’abeena abstract noun ma-ricc-oomáta ‘essence, nature’. It has the following structure: The question word ma ‘what’ is followed by the noun ricc-u ‘thing’ yielding “what-thing”. To this the abstract noun suffix -oomáta is added. In Amharic, the structure of the respective abstract noun mənənnät is similar. Here the question word mən ‘what’ is derived by a suffix for abstract nouns, namely -(ə)nnät. Greetings are expressed by identical patterns, namely by questions. In order to greet somebody in the morning, one says in K’abeena garee gal-ti ‘Did you pass the night well?’, i.e. garee ‘well’ precedes the verb gal-u ‘pass the night’, inflected for the second-person singular. The equivalent expression in Amharic is dähna addär-k, i.e. dähna ‘well’ precedes addär-ä ‘he passed the night’, in this example inflected for the second-person singular masculine. In the evening one asks in K’abeena garee hos-si ‘Did you pass the day well?’. The equivalent in Amharic is dähna wal-k.
9. Conclusion K’abeena shares a lot of phonological, grammatical and lexical features with other languages spoken in the highlands of Ethiopia. This fact allows it to refer to this area as the ELA. The core area comprises the languages of the highlands of Ethiopia. The more distant a given language is situated from this core area the fewer features it has. However, the areal status of individual features is not accepted generally (for the discussion cf. Crass 2002, Crass and Bisang 2004, Tosco 2000, Zaborski 1991). Furthermore, the existence of the ELA is denied by Tosco (2000). However, since only a relatively small number of languages are described adequately, these findings must be considered preliminary.
Abbreviations AMH acc abl ARAB cmpl
amharic accusative ablative Arabic complementizer
cnd cnv cop ELA f
conditional converb copula Ethiopian linguistic area feminine
K’abeena gen HEC ideo imp ipv K’AB loc m neg nom obj p
genitive Highland East Cushitic ideophone imperative imperfective K’abeena locative masculine negative nominative object agreement plural
prv pss pst purp q rel s simil vn 1 2 3
103
perfective possessive past purpose question relative clause singular similative verbal noun first person second person third person
Note 1. This research was undertaken in the scope of the Collaborative Research Center 295 Cultural and linguistic contacts: Processes of change in North Eastern Africa and West Asia (Sonderforschungsbereich 295 Kulturelle und sprachliche Kontakte: Prozesse des Wandels in historischen Spannungsfeldern Nordostafrikas/Westasiens).
References Bisang, Walter 1996 Areal typology and grammaticalization: Processes of grammaticalization based on nouns and verbs in east and mainland south east Asian languages. Studies in Language 20 (3): 519597. 2006 Linguistic areas, language contact and typology: Some implications from the case of Ethiopia as a linguistic area. In: Yaron Matras, April McMahon and Nigel Vincent (eds.), Linguistic areas: Convergence in historical and typological convergence, 7598. Houndmills: Palgrave Macmillan. Braukämper, Ulrich 1980 Geschichte der Hadiya Süd-Äthiopiens (Studien zur Kulturkunde 50). Wiesbaden: Steiner. Crass, Joachim 2002 Ejectives and pharyngeal fricatives: Two features of the Ethiopian language area. In: Baye Yimam, Richard Pankhurst, David Chapple, Yonas Admasu, Alula Pankhurst, and Birhanu Teferra (eds.), Ethiopian studies at the end of the second millennium. Proceedings of the 14th
104
Joachim Crass
International Conference of Ethiopian Studies, 611 Nov. 2000, 1679– 1691. Addis Ababa. Addis Ababa: Institute of Ethiopian Studies. 2003 The copulas of K’abeena: Form, function, and origin. Afrika und Übersee 86: 2342. 2005 Das K’abeena. Deskriptive Grammatik einer hochlandostkuschitischen Sprache (Cushitic Language Studies 23). Köln: Köppe. Crass, Joachim, and Walter Bisang 2004 Einige Bemerkungen zum äthiopischen Sprachbund und ihre Relevanz für die Areallinguistik. In: Walter Bisang, Thomas Bierschenk, Detlev Kreikenbom, and Ursula Verhoeven (eds.), Kultur, Sprache, Kontakt, 169198. Würzburg: Ergon. Crass, Joachim, and Ronny Meyer 2005 Die Komplexität kultureller und sprachlicher Kontakte am Beispiel der Nomenklatur zur Ensete-Pflanze. In: Walter Bisang, Detlev Kreikenbom, and Ursula Verhoeven (eds.) Prozesse des Wandels in historischen Spannungsfeldern Nordostafrikas/Westasiens. Akten zum 2. Symposium des SFB 295, 15.10.–17.10.2001, 411427. Würzburg: Ergon. 2007 Ethiopia. In: Bernd Heine and Derek Nurse (eds.), A Linguistic Geography of Africa. Cambridge: Cambridge University Press. 228249. Faber, Alice 1997 The genetic subgrouping of the Semitic languages. In: Robert Hetzron (ed.), The Semitic Languages. 315. London, New York: Routledge. Fleischman, Suzanne 1989 Temporal distance: A basic linguistic metaphor. Studies in Language 13: 150. Haspelmath, Martin 1997 From Space to Time: Temporal Adverbials in the World’s Languages. München, Newcastle: Lincom. Hayward, Richard 1991 À propos patterns of lexicalization in the Ethiopian language area. In: D. Mendel and U. Claudi (eds.), Ägypten im afro-asiatischen Kontext. Aufsätze zur Archäologie, Geschichte und Sprache eines unbegrenzten Raumes. Gedenkschrift Peter Behrens, 139156. (Afrikanistische Arbeitspapiere. Sondernummer 1991). Köln: Institut für Afrikanistik. 2000 Is there a metric for convergence? In: Colin Renfrew, April McMahon, and Larry Trask (eds.), Time Depth in Historical Linguistics Vol.1, 621640. Cambridge: McDonald Institute for Archeological Research. Heine, Bernd 1994 Areal influence on grammaticalization. In: Martin Pütz (ed.), Language contact and language conflict, 5568. Amsterdam/Philadelphia: John Benjamins.
K’abeena
105
Heine, Bernd, and Tania Kuteva 2002 World Lexicon of Grammaticalization. Cambridge: Cambridge University Press. 2003 On contact-induced grammaticalization. Studies in Language 27 (3): 529572. Kane, Thomas Leiper 1990 Amharic-English dictionary. Wiesbaden: Harrassowitz. Leslau, Wolf 1979 Etymological Dictionary of Gurage (Ethiopic). Volume III. Etymological Section. Wiesbaden: Harrassowitz. Meyer, Ronny, and Renate Richter 2003 Language Use in Ethiopia from a Network Perspective (Research in African Studies 7). Frankfurt: Lang. Tosco, Mauro 2000 Is there an “Ethiopian language area”? Anthropological Linguistics 42 (3): 329365. Zaborski, Andrzej 1991 Ethiopian language subareas. In: Stanislaw Pilaszewicz and Eugeniusz Rzewuski (eds.), Unwritten Testimonies of the African Past. Proceedings of the International Symposium Held in Ojrzanów N. Warsaw on 78 November 1989, 123134 (Orientalia Varsoviensia 2). Warsaw: Wydawnictwa Universytetu Warszawskiego.
Grammatical borrowing in Likpe (Sɛkpɛlé) Felix K. Ameka
1. Background Sɛkpɛlé is the auto-denomination of the language spoken in the area known as Likpe which is to the east and north-east of Hohoe (the district capital and an Ewe (Gbe) speaking town) as far as the Togo border in the northern part of the Volta Region of Ghana (see Map 1). There are approximately 23,000 residents in the area who speak the language (District Assembly Representative 1998 figures). A small percentage, living in the two modern migrant villages, speaks the language as a second language. If we add the other native speakers in the diaspora, there may well be more than 30,000 speakers of the language today. Sɛkpɛlé is one of the fourteen languages most recently characterized as Ghana–Togo–Mountain (GTM) languages (Ring 1995) that were first recognized as a group and referred to as Togorestsprachen by Struck (1912) and, in English, as “Togo Remnant languages”,1 for instance by Westermann and Bryan (1952: 96). Their genetic classification and cultural history have remained controversial (see e.g. Nugent 1997, 2005). The distinctive typological features of these languages that separate them from the surrounding languages such as Ewe and Akan include (i) the active noun class system similar to that of the Bantu languages which they have inherited, undoubtedly, from Proto-Niger-Congo, (ii) head marking at the clause level through subject cross-referencing on the verb, and (iii) their highly agglutinative nature especially their derivational verb morphology. Some scholars wonder whether the group of languages is not a socio-cultural or a typological grouping masquerading as a genetic unit (Blench 2001, Maddieson 1998). Nevertheless, they are classified as belonging to Kwa (Niger-Congo). It is difficult, however, to establish the Ghana-Togo-Mountain (GTM) languages as a group in relation to Kwa. Heine (1968) subclassified the GTM languages into Ka-Togo and Na-Togo subgroups. The current view is that the two subgroups branch out individually from Proto-Kwa as in Figure 1 (Williamson and Blench 2000). Sɛkpɛlé is a Na-Togo language and has two major dialect divisions, namely, Sɛkpɛlé and Sekwa.2 Table 1 shows the dialects and the villages where they are spoken.
108
Felix K. Ameka Ega Avikam Alladian Ajukru Abidji Abbey Attié Ebrie
Potou
Mbatto Krobu Potou-Tano
Abure, Eotilé
West Tano
Akan Nzema-Ahanta
Tano Central Tano
Bia
Anyi, Baule, Anufɔ Efutu-Awutu
Guan
South
Larteh-Cherepong-Anum Northern Guang Ga
Proto-Kwa
Dangme Lelemi-Lefana Akpafu-Lolobi Lipke, Santrokofi Na-Togo
Logba Basila, Adele Avatime Nyangbo-Tafi
Ka-Togo
Kposo, Ahlo, Bowiri Kebu, Animere Ewe
Gbe
Gen, Aja Fon-Phla-Phera
Figure 1. Proto-Kwa (from Williamson and Blench 2000: 29)
A large part of the Likpe-speaking population is multilingual in Sɛkpɛlé, Ewe – the dominant lingua franca – and Akan, not to mention (Ghanaian) English (cf. Ring 1981). Ewe, the dominant lingua franca, co-exists with Likpe in all spheres of life in the home, the community, in the church, in the market,
Likpe (Sɛkpɛlé)
109
Table 1. Sɛkpɛlé dialects and their geographical distribution Language
Sɛkpɛlé
Dialects
Sekwá
Sub-dialects Villages
Sɛkpɛlé L2 communities Situnkpa
Bakwa Todome Nkwanta
Alavanyo Wudome
Semate
Avedzime Mate Agbozume Abrani Koforidua
Sela Bala Kukurantumi
© Likpe Local Committee
Map 1. Likpe Traditional area
110
Felix K. Ameka
at school and other public ceremonies such as funerals and marriages. Akan is less prominent, although it is the lingua franca in a neighboring community. Likpe was first written in about 1933 using the Ewe orthography (see Dogli 1933). The next time there was writing was with the new wave of literacy and language development under the auspices of the Ghana Institute of Linguistics, Literacy and Bible Translation (GILLBT) in 1987. Today there are pamphlets of stories, proverbs and literacy materials as well as a pamphlet containing a translation of the letters of Paul (New Testament). There are also cassettes containing some drama and text in Likpe. Likpe is used partially as medium of instruction and is taught outside the regular curriculum in some villages after school hours to both Primary and Junior Secondary School pupils. The focus of the present chapter is on the grammatical changes in Sɛkpɛlé due to the influence of Ewe, the dominant lingua franca, and areal convergence. That is, the impact of the neighbouring languages on Likpe grammar. Likpe seems to have innovated constructions such as the present progressive and an Undergoer Voice construction due to Ewe influence. At the same time there are several patterns that are found in Likpe due to pressures of areal adaptation, for instance, postpositions and arguably, serial verb constructions have emerged in the language through this mechanism. There are also grammatical items such as intensifiers and particles and interjections which are also shared among the neighbouring languages, but Likpe has borrowed some connectives and a complementizer directly from Ewe (see also Ameka 2007).
2. Typology Likpe is an agglutinative language with some head marking at the clause level and dependent marking in the NP (except for qualifiers), properties presumably inherited from Proto Niger-Congo. Features of nouns, including class and number, are marked by prefixes on nouns. Ewe, on the other hand, is predominantly isolating with agglutinative features. Plural marking in Ewe is by an enclitic which attaches to the element in the NP that occurs before the Intensifier. Likpe marks plural number for a small set of nouns by a suffix. The use of such a structure in Sɛkpɛlé is due to its contact with Ewe (see Section 3). The functional load of the morphological process of Reduplication seems to have increased in Likpe due to its contact with Ewe. Some adjectival modifiers in Sɛkpɛlé are formed by verbal reduplication similar to the pattern found in Ewe. For instance,
Likpe (Sɛkpɛlé)
(1)
111
a. Reduplication of a property verb to form a qualifier ná ‘become.black, dirty’ ná-ná ‘dirty’ bù ‘become.wet, rotten’ bùbù ‘wet, rotten’ b. Reduplication of action verbs to form result-state qualifiers là ‘cut’ là-là ‘torn’ f ‘to arrive newly’ f-f ‘new’
Gerund formation of verb and complement NP structures involves the reduplication of the verb with the NP preposed to it. Such a process is available in Ewe and could have influenced the process. In fact it appears that the gerund involving reduplication is in competition with the ‘older’ form of gerund formation involving the use of the class marker for deverbal nominals. Compare both types of gerund formation shown in (2): (2)
a. bi-sí-tk-tk Compare Ewe te-ʃa-ʃa cm-yam-red-be.on yam-red-plant ‘yam planting’ ‘yam planting’ b. bi-sí bu-tkə cm-yam cm-be.on ‘yam planting’
The gerunds formed by reduplicating the verb part seem to be entering the language through translation, especially of texts mediated through Ewe.
3. Nominal structures Likpe noun stems tend to participate in sg/pl class pairing system for number marking. Some kin terms do not have plural counterpart classes. Kin terms belonging to ego’s parents’ generation and above and proper names are suffixed with the form -m ‘pl’ to signal their plurality. This form is heterosemically related to the third-person plural pronoun. (3)
a. Kofi kú Áma-mə´ ə-diə. name com name-pl scr-quarrel ‘Kofi and Ama and co quarrelled.’ b. ambe ‘mother’ ambe-mə´ ‘mother-pl’ éwú ‘grandmother’ éwú-mə´ ‘grandmother-pl’
112
Felix K. Ameka
This Likpe structure is a replication of the Ewe pattern where plural number is marked by a clitic =wó ‘pl’, which is attached to the last element in the NP before the intensifier, and which is also in a heterosemic relation to the ‘3pl’ pronominal form wó. The use of the Ewe form as an associative plural, that is, N=wó means ‘N and co’, e.g. Áma=wó [Ama=pl] ‘Ama and co’, could have served as the model for the copy.
4. Verbal structure Likpe, like its closest genetic relatives, marks tense, aspect and mood categories by prefixes on the verb (including for example past progressive). However, it has developed a present progressive periphrastic construction similar to the one found in Ewe which has the form: Subject -l ‘hold’ (NP) Gerund. (4a) below is a clause in the aorist with no segmental marker for the category, while (4b) is an instantiation of the progressive construction in relation to the state of affairs represented also in (4a) (see Ameka 2002). (4)
a. o-té ka-m. 3sg-sell cm-rice ‘She sold rice.’ b. ɔ-l ka-m bo-té. 3sg-hold cm-rice cm-sell ‘She is selling rice.’
Compare the Ewe equivalent of (4b) in (4c) in terms of the structure and order of the elements. The phonetic similarity between the operator verb in the two languages could have facilitated the adaptation of the structure into Likpe. (4)
c. Ewe é-le mlu dzráx-ḿ. 3sg-be.at:pres rice sell-prog ‘She is selling rice.’
An operator for the expression of necessity has also been adopted into the language. This may be due to areal rather than Ewe specific influence. Moreover, the construction is one of the structures in which Likpe uses a complementizer that is borrowed from Ewe, namely b ‘quot’, as illustrated in (5).
Likpe (Sɛkpɛlé)
(5)
113
é-hiɛ˜´ bə´ u-tsyi wə ú-su u-bíkə. impers-need quot 3sg-carry 3sg 3sg-go 3sg-bury ‘It was necessary that he (Skunk) should take her (his mother) to go and bury.’
There is no special marking of verbs that are borrowed. They are only phonologically adapted; for example, the verb form that is used in example (5) above has the form hia ‘need’ in both Ewe and Akan, but is adapted in Sɛkpɛlé as hiɛ ‘need’. Likpe uses several patterns for marking voice/valency that conform to areal patterns: the reflexive is formed by the use of a pronoun plus a grammaticalized form of the word for ‘body’, i.e. əsúə (similar to the structure one gets in Akan, but not in Ewe). The use of 3pl to express general subjects and impersonal passive meanings is also an areal pattern. There is an Undergoer Voice construction expressed periphrastically where the Undergoer-like argument is linked to the subject position of the clause and of an operator verb nɔ ‘hear’.3 The operator verb takes a nominalized verb complement with the Actor-like argument, if expressed, functioning as its first object (and like the Goal argument in a double object construction). The semantics of the construction belongs to the semantics space of the so-called ‘potential passives’. There is a similar periphrastic construction in Ewe but its operator verb is a modal verb nyá, grammaticalized from the verb ‘know’. The Actor-like argument in the main event, if expressed, is marked as a dative (experiential) object in the Ewe construction, as illustrated in (6). (6)
Ewe nynu-a nyá kp-ná (ná-m). woman-def mod see-hab dat-1sg Lit: ‘The woman is see-able (to me).’ i.e. ‘The woman is beautiful (to me).’
(7)
Likpe u-sió -m á-nɔ (mɛ) bó-be. cm-woman agr-det scr-hear 1sg cm-look Lit: ‘The woman hears (me) looking.’ i.e. ‘The woman is beautiful (to me).’
114
Felix K. Ameka
(8)
Likpe n-t á-nɔ bú-nə í-tə´ be-tsyúé. cm-alcohol scr-hear cm-drink 3sg:impers-give cm-some Lit: ‘Alcohol hears drinking give some.’ i.e. ‘Alcohol-drinking is enjoyable to some (people)’, ‘Some people like drinking alcohol.’
One argument in support of the Ewe construction potentially influencing the Likpe construction comes from the periphrastic nature of the construction in Likpe. One would have expected an affixal marking of such a meaning on the verb in Sɛkpɛlé. Secondly, if the Actor is expressed it takes the form of a Goal argument similar to the dative marking in Ewe. Thirdly, there is a variation on the expression of the Actor-like argument in the Sɛkpɛlé construction which replicates the Ewe pattern, as illustrated in (8) (see Ameka 2005a for further details on the construction in both Ewe and Likpe).
5. Other parts of speech Likpe has borrowed a few grammatical items from Ewe and also makes use of several forms that are found throughout the Lower Volta Basin area, i.e., the area in which Kwa languages are spoken. It has borrowed the contrast connector gaké ‘but’ from Ewe and adapted it as appropriate as kaké ‘but’ in the Sɛkpɛlé dialect since unlike Sekwa, there are no [−anterior] [+voice] consonants in the Sɛkpɛlé dialect. Languages in the area tend to have two or three disjunction markers, one of which tends to be used in interrogative contexts. Likpe has two disjunction markers: nyé ‘this or that, it does not matter which’ and lee ‘this or that, I don’t know which’. It appears the latter form is influenced by one of the disjunction markers in Ewe, the form lóó ‘this or that, I don’t know which’. Because of their ignorative feature they tend to be used in interrogative contexts and can link phrases or clauses. A hint that this may be the case is that Tuwuli, another GTM language, has a cognate disjunctive marker nye ‘or’ used in general contexts while in interrogative contexts the other form mbɔe is used (Harley 2005). Likpe seems thus to have adapted the Ewe form for use in interrogative contexts. Likpe has also borrowed a complementizer b ‘comp, quot’ from Ewe which it uses in addition to its own complementizer ŋkə ‘quot’. Sometimes the Ewe and the Likpe complementizers are doubled. Compare the use of
Likpe (Sɛkpɛlé)
115
both forms in similar contexts in (9) and (10) both taken from a Likpe settlement history narrative. (9) sé ɔfu kɔdzó -m le-te ŋkə məə-tsyá when name name agr-det dep-know quot 3pl-too a-slé eto be-tídi bé-ni ko ŋkə oo, atúu … cm-church poss cmpl-person 3pl-cop int quot interj welcome ‘When Ofu Kwadzo got to know that they too were church people, he said oo welcome, (he and them will work together).’ (10) nyã bəə-b bə-tu m nyã bə´ə oo ka-sɔ kpé. and 3pl-come 3pl-meet 3pl and quot interj cm-land be.in ‘And they came to meet them and they said “oh there is land”.’ While the relativizer itself is not borrowed, Likpe seems to be developing the use of the definiteness determiner as an optional relative-clause-final marker. This is an areal pattern. The interesting thing is that in the other languages like Ewe, Akan, Ga etc. the same form that is used in this relative clause context is used to mark all other background information constituents such as left-dislocated topic NPs and preposed adverbial clauses such as conditionals and temporals. In Likpe, as we shall see below, there are different forms for marking these, one of which is a borrowed form from Ewe. (11) kɔsídá kó l-yɛ (mə´) week agr dep-pass det ‘the week which passed, the past week’ In addition, two adverbial connectors are borrowed from Ewe: álé bé ‘so that’ (example 12) and tógb bé ‘although’. (12) álé-bé ŋko ni kasé min-yi ba-kpɛlé eto akokosa n. thus-quot this cop how 1sg-know cmpl-Likpe poss history emph ‘So this is how I know the history of Likpe (to be).’ Various intensifying words (focus particles) found in Likpe are forms that have diffused through the West African littoral area. However, the Likpe forms resemble more closely the forms used in Ewe than the others. Thus words such as ko ‘only, just’, boŋ ‘rather’, tsyá ‘also’ are shared with Ewe. Similarly, interjections such as ah㈠‘now I know’ and fillers like oo, an
116
Felix K. Ameka
utterance-initial vocative particle, o …, answer particles like ee ‘yes’, o ‘no’; agreement-signalling particles like yoo ‘OK’ and the palatal click with nasal release, both as an agreement marker and signalling continuation in the sense of ‘I understand’, are all shared by various languages in the area. Additionally utterance final particles for expressing attitudinal meanings; e.g. ló ‘I advise you’, are widespread in the area. Some of these forms give evidence of convergence among the languages in the area and their occurrence in Likpe can be attributed to areal adaptation. A form ma- used for ‘privative’ derivations in Ewe has occurred in some derived words in Likpe (see example 13). But it is unclear whether it is copied from Ewe or it is retained from some ancestor language since other GTM languages, e.g. Tuwuli (Harley 2005) seem to have similar forms. (But of course they could also have taken it from Ewe.) (13) a. Likpe lə-fə n bə-n-sí ko kasɔ´-ma-nɔ-ma-nɔ cm-time agr 3pl:hab-lig-sit int under-priv-hear-priv-hear ə-b-lu-f m bə-tsyú l ntí. scr-vent-leave-dir 3pl cmpl-neighbour loc midst ‘The time they stayed there then misunderstanding emerged among their neighbours.’ b. Ewe nú-gɔme-ma-se-ma-se thing-under-priv-hear-priv-hear ‘misunderstanding’ Likpe has also adopted the background information marking particle lá ‘tp’ from Ewe (although it is an item that is consciously recognized and edited out of texts). (14) kasé mi-nɔ nyã ní b bó ba-kpɛlé lá how 1sg-hear 3sg cop quot 1pl cmpl-Likpe tp bo kə-síə-kɔ fefe ka búu-siə ní atébubu. 1pl cm-sit-place last agr 1pl:past-sit cop name ‘How I heard it is that we the Likpe people, our last place of settlement where we stayed was Atebubu.’ The background information particle in Ewe is in a heterosemic relation to the definiteness marker and is used at the end of left dislocated NPs, connectors as well as preposed adverbial phrases and clauses and relative clauses. In
Likpe (Sɛkpɛlé)
117
Likpe, the Ewe form la´ occurs in all these contexts. The Likpe indigenous form of marking background information in these contexts is to lengthen the phrase-final vowel.
7. Syntax One of the features of Likpe grammar that could have emerged due to areal pressure from the surrounding languages like Ewe and Akan is verb serialization in a single clause. Dimmendaal (2001: 386) claims that the spread of serial verb constructions (SVC) to the GTM languages could account for the reduction in verb derivational morphology that is in progress in these languages (cf. Hyman 2004). Perhaps an indication that this may be so is that the features of SVCs in Sɛkpɛlé share some features with Akan SVCs and other features with Ewe SVCs, and there are other features that are common to all three languages. For instance, the shared subject argument is expressed with each verb in the SVC in both Likpe and Akan, but expressed only once in an Ewe SVC. Negation, on the other hand, is expressed only once in an SVC in both Likpe and Ewe but recapitulated with each verb in Akan. For all the languages the verbs in an SVC should be marked for aspect and modality values that are semantically compatible (see Ameka 2005b). Likpe uses locative verbs in the expression of predicative possession. The verbs are kpé ‘be.in’, t ‘be.at’ and tk ‘be.on’. These are used to express both general location and ‘have’ possession (see Ameka 2007b for a discussion of their use and semantics). With each of these verbs there are two constituent orders for the possessive use: one in which the Possessor is linked to the Subject position and the possessed phrase linked to the object function. The second order involving Figure–Ground reversal has the possessed linked to the subject position and the possessor to the post-verbal object position. The two orders are illustrated for the verb kpé ‘be.in’ below. (15) a. Possessor–verb–possessed o-kpé a-fokpá 3sg-be.in cmpl-footwear ‘He has shoes.’ b. Possessed–verb–possessor a-fokpá kpé wə cmpl-footwear be.in 3sg ‘He has shoes.’
118
Felix K. Ameka
It is possible that the second order in (15b) is due to Ewe influence. Ewe uses a general locative verb le ‘be.at:pres’ in a periphrastic construction to express predicative possession: ‘have’. The structure has the form possessed–verb– possessor NP + possessive postposition si ‘hand’. This Ewe structure is the only order possible for expressing predicative possession in Ewe using the locative schema.
8. Lexicon A number of constructions that one finds in Likpe grammar are based on semantic formulas that are available in other languages in the area. For instance, a verb–noun collocation which literally translates as ‘see/look way/road’ is used to express the idea of ‘hope’ as illustrated for Likpe in (16). (16) ó-be ku-sú ŋkəə mba uuka-wuuns-ko é-bu-b 3sg-look cm-way quot those 3sg:phab-help-assoc 3pl:fut-come ba-wuuns-ko w u-bik wo ambé -m. 3pl-help-assoc 3sg 3sg-bury 3sg mother agr-det ‘He hoped that those who he used to help would come to help him bury his mother.’ Similarly various verb–verb collocations in serial verb constructions form semantic formulas for the expression of particular meanings in the serializing languages. One such formula is for the expression of the sense of ‘believe’. It is composed of two verbs the first of which is invariably a ‘receive/get’ verb and the second an ingestion or imbibing verb such as ‘eat’ or ‘hear’. Likpe makes use of the ‘receive hear’ pattern which is the same pattern that we find in Ewe. The Likpe form is exemplified in (17). (17) n-fo n-nɔ míə yɔɔ-lkɛ. 1sg-get 1sg-hear 1sg:quot 3sg:fut-be.good ‘I believe it will be good.’ A formula for expressing the notion of ‘begin’ relates to making contact with the bottom of the situation that is began. Such expressions occur in several languages in the area.
Likpe (Sɛkpɛlé)
119
9. Conclusion Various changes have taken place and continue to take place in Likpe grammar due to its contact with Ewe on the one hand, and due to pressures of areal adaptation, on the other. The present progressive construction, plural marking on kinship terms, Undergoer Voice construction, and constituent order for predicative possession are constructional patterns that have been directly replicated from Ewe. Patterns involving serial verb constructions and postpositions, intensifiers and various semantic formulas occur in Likpe due to their presence in the convergence area. Furthermore, Likpe has borrowed lexical as well as grammatical items from Ewe replacing indigenous terms in some cases. It is significant that the grammatical items that have been borrowed have discourse structuring or organizational functions such as a contrast marking conjunction, a disjunction marker, and reason expressing connectives. Nouns and verbs are also borrowed. Borrowed verbs do not receive any special treatment. Nouns borrowed into Likpe are integrated on the basis of form and meaning into the noun class system. Thus a noun like agbeli ‘cassava’ which in Ewe is made up of a prefix a- and the stem and which is singular or collective is borrowed into Likpe and analysed as a plural noun to fit the a- plural class and a singular form is formed as le-gbeli ‘one tuber of cassava’. It is in this domain that the borrowing of nouns has an effect on the grammar of Sɛkpɛlé.
Abbreviations agr assoc cm com comp cop dat dem det dep dir emph
agreement marker associative verb extension noun class marker comitative complementizer copula dative (preposition) demonstrative determiner pragmatically dependent subject cross-reference directional emphatic particle
fut impers int interj lig loc neg past phab pl poss pot pres
Future impersonal intensifier interjection ligature locative negative marker past tense past habitual plural possessive marker potential present
120
Felix K. Ameka
priv prog q red
privative progressive propositional question reduplicative
scr subject cross-reference marker sg singular tp background topic marker
Notes 1. They have also been referred to as “Central Togo” (Dakubu and Ford 1988). 2. The findings reported here are based on my own field investigations that I have been conducting in various Likpe communities intermittently over the past decade or so. I am very grateful to my consultants, especially E. K. Okyerefo, Justina Owusu, Cephas Somevi, Comfort Atsyor and the late A. K. Avadu, for patiently teachinjg me their language. 3. I use the terms Actor and Undergoer following their usage in Role and Reference Grammar (RRG) see Van Valin and La Polla (1997: 141147). To characterize the objects in a double object construction, I use the terms “Theme” and “Goal”. The latter is a cover role for the object with the semantic relation of recipient, beneficiary, maleficiary etc. or the dative more generally.
References Ameka, Felix K. 2002 The progressive aspect in Likpe: Implications for aspect and word order in Kwa. In: Felix K. Ameka and E. Kweku Osam (eds.), New directions in Ghanaian Linguistics, 85111. Accra: Black Mask 2005a “The woman is seeable” and “The woman perceives seeing”: Undergoer voice constructions in Ewe and Likpe. In: M. E. Kropp Dakubu and E. Kweku Osam (eds.), Studies in the Languages of the Volta Basin, Volume 3, 4362. Legon: Department of Linguistics, University of Ghana. 2005b Multiverb constructions on the West African littoral: Microvariation and areal typology. In: M. Vulchanova and T. A. Åfarli (eds.), Grammar and Beyond: Essays in Honour of Lars Hellan, 1542. Oslo: Novus Press. 2007a Grammars in contact in the Volta Basin (West Africa): On contact induced grammatical change in Likpe. In: A. Y. Aikhenvald and R. M. W. Dixon (eds) Grammars in Contact: A Cross-Linguistic Typology, 114– 142. Oxford: Oxford University Press.
Likpe (Sɛkpɛlé) 2007b
121
The coding of topological relations in verbs: The case of Likpe (Sɛkpɛlé). In: Felix K. Ameka and Stephen C. Levinson (eds.), The Typology and Semantics of Locative Predication: Posturals, Positionals and Other Beasts, 10651103 (Linguistics 45 (5/6), special issue). Blench, Roger M. 2001 Comparative Central Togo: What have we learnt since Heine? Paper presented at the 32nd Annual Conference on African Linguistics: Berkeley, 2325 March 2001. Dakubu, M. E. Kropp, and Kevin C. Ford 1988 The Central-Togo languages. In M. E. Kropp Dakubu (ed.), The languages of Ghana, 119154. London: Kegan Paul. Dimmendaal, Gerrit J. 2001 Areal diffusion and genetic inheritance: An African perspective. In: A. Y. Aikhenvald and R. M. W. Dixon (eds.), Areal diffusion and genetic inheritance, 358392. Oxford: Oxford University Press. Dogli, A. (Rev.) 1933 Likpe Catechism No 1 and 2. Keta: Catholic Printing Office Heine, Berndt 1968 Die Verbreitung und Gliederung der Togorestsprachen. Berlin: Dietrich Reimer. Harley, Matthew 2005 A descriptive grammar of Tuwuli: A Kwa language of Ghana. Ph.D. thesis, SOAS. Hyman, Larry 2004 How to become a Kwa verb. Journal of West African Languages 30 (2): 6988. Maddieson, Ian 1998 Collapsong vowel harmony and doubly articulated fricatives: Two myths about the phonology of Avatime. In: Ian Maddieson and Thomas Hinnebusch (eds.), Language History and Linguistic Description in Africa, 155166. Trenton: Africa World Press. Nugent, Paul 1997 Myths of Origin and Origins of Myth: Politics and the Uses of History in Ghana’s Volta Region. Berlin: Das Arabisch Buch. 2005 A regional melting pot: The Ewe and their neighbours in the GhanaTogo borderlands. In: Benjamin Lawrence (ed.), The Ewe of Togo and Benin, 2945. Accra: Woeli. Ring, Andrew J. 1981 Ewe as a Second Language: A Sociolinguistic Survey of Ghana’s Central Volta Region. Legon: Institute of African Studies. 1995 Lɛlɛmi tone. Papers from GILLBT’s seminar week 30 January–3 February 1995, Tamale, 1995: 1626.Tamale: GILLBT Press.
122
Felix K. Ameka
Struck, R. 1912
Einige Sudan-Wortstämme. Zeitschrift für Kolonialsprachen 2: 233– 253, 309323. Van Valin, Robert D. Jr, and Randy J. La Polla 1997 Syntax: Structure, Meaning and Function. Cambridge: Cambridge University Press Westermann, Dietrich, and Margaret A. Bryan 1952 Languages of West Africa. Handbook of African Languages, Volume 2. London: Oxford University Press, for the International African Institute. Williamson, Kay, and Roger Blench 2000 Niger-Congo. In: Bernd Heine and Derek Nurse (eds.), African Languages: An Introduction, 1142. Cambridge: Cambridge University Press.
Grammatical borrowing in Katanga Swahili Vincent A. de Rooij
1. Background1 Katanga Swahili, also known as Shaba Swahili, is a contact variety of Swahili spoken in the urban centers of Southern Katanga, dr Congo. Katanga Swahili resulted out of contacts between speakers of closely related Bantu languages.2 Despite the strong structural similarities of these languages, the phonology, tma system, and morpho-syntax of Katanga Swahili have been restructured considerably. Elsewhere (de Rooij 1997), I argued that, apart from adult second language learning, nativization, i.e. first language acquisition by locally born children, may have played a role in the genesis and development of Katanga Swahili. I also argued that selective simplification (Kapanga 1993) cannot have been the sole restructuring process since even features which are characteristic of both the ‘lexifier’ language, the East African Coast variety of Swahili, and the genetically related ad/substrate languages, such as agglutinative verbal morphology and the noun-class agreement system, have been lost to varying degrees in Katanga Swahili. Katanga Swahili is widely used as a first language in Southern Katanga and is spoken by an estimated number of at least 2 million people.3 In the cities of the Southern Katanga Copperbelt, multi-lingualism is widespread. Apart from Katanga Swahili, many speak French, Congo’s official language, and some ‘ethnic’ language (Kabamba 1979). Katanga Swahili is used in all informal settings: it is used in the domestic sphere but also in informal public settings (public transport, markets, shops). Informal notes and letters are often written in Swahili but since Katanga Swahili has no standard orthography, writing is often done in idiosyncratic ways (cf. Blommaert 1999, 2004; Fabian 1990). Books and newspapers in Swahili are widely available but the variety used in these publications is very similar to the Standard Swahili of Tanzania and is almost like a foreign language to speakers of Katanga Swahili. According to Fabian (1986) Katanga Swahili was well established as a language distinct from the East African variety of Swahili by 1940. Katanga Swahili in its present form, however, cannot be taken as representing the language as spoken around 1940, although there are probably no dramatic
124
Vincent A. de Rooij
differences between the two stages. Between 1940 and the present day, the urban centers of Southern Katanga and Elisabethville (the colonial name of present-day Lubumbashi) in particular, have absorbed large numbers of migrants, mainly from the neighbouring Kasai provinces (Fetter 1976: 173– 176). The massive influx of Luba-Kasai, Songye, Kanyok and Kete speakers from Kasai, all of whom have/had to learn Swahili, does/did of course have an impact on the language. It should also be noted that the majority of these migrants’ children acquire/d Katanga Swahili as their first language instead of their parents’ native language. Therefore, Katanga Swahili as it is spoken today is to be seen as the outcome of processes of second language learning and nativization that have been going on since the 1940s.
2. Phonology Katanga Swahili has a symmetrical five-vowel system (see Table 1), also found in ad/substrate languages. Just like East Coast Swahili, it does not have phonemic vowel lengthening nor does it have grammatical, apart from one exception, or lexical tone which are all prominent in the ad/substrate languages. Phonetic values of /e/ and /o/ range, depending on the environments they occur in, from [e] to [ɛ] and from [o] to [ɔ ] respectively. East Coast Swahili dental and velar fricatives do not occur in Katanga Swahili. In East Coast Swahili, they are found exclusively in Arabic borrowings. Many of these words denote Islamic concepts and therefore play no role in Katanga where Islamic influence is virtually non-existent. In the few generally used words of Arabic origin, dental fricatives /ð/ and /θ/ become /z/ and /s/ respectively. The velar fricative /γ/ occurs in Katanga Swahili as /k/. ECS glottal fricative has also been lost. Several other ECS phonemes have not been fully retained in Katanga Swahili either, most probably due to ad/sub-
Table 1. Katanga Swahili vowels
close close-mid open
front
back
i
u e
o a
Katanga Swahili
125
Table 2. Katanga Swahili consonants
plosives nasals fricatives affricates trills tap lateral approximant glides prenasalized consonants
labiobilabial dental
postalveolar alveolar palatal
pb m
t
n s
f
velar
glottal
k tʃ
l w mp mb
j mf mv
nt nd ns nz
nʃ ndʒ
nk ng
Note: indicates ECS phoneme with weak phonemic status in Katanga Swahili
strate influence from Bemba and Luba-Kasai, where they do not function as phonemes (cf. Kashoki 1968 and Burssens 1939). In Luba-Kasai and Bemba we do not find: voiced palatal affricate /ʤ/, voiced velar plosive /g/, glottal fricative /h/, and alveolar vibrant /r/. In Katanga Swahili, these are commonly replaced by /j/, /k/, ø, and /l/ respectively. Furthermore, in Bemba we do not find: voiced alveolar plosive /d/, voiced labio-dental /v/ and alveolar fricative /z/. These are often replaced by /l/, /f/, and /s/ respectively. /s/, /z/, /t/ are often palatalized, especially by speakers with a Luba-Kasai background, to [ʃ], [ʒ], and [c~tʃ] respectively when followed by /i/. Furthermore, the ECS preverbal tense affix -li- is normally pronounced as [ri] or[ɾi]. Katanga Swahili has a range of prenasalized consonants that are also found in substrate languages (cf. Bostoen 1997: 91). The status of some of these prenasalized consonants as phonemes remains to be settled (de Rooij 1997: 335). It seems clear that contact with ad/substrate languages has resulted in a phonological system that has drifted away from the East Coast Swahili system and has become more similar to the systems of ad/substrate languages. Phonemes that do not occur in ad/substrate languages have been lost, partly or completely. The clearest example of this process is /h/ which is almost categorically left unarticulated. Other phonemes have blended into one, where the phoneme that is absent in one or more ad/substrate languages has been lost or weakened. An example is the blending of East Coast Swahili velar plosives /g/ and /k/ into /k/.
126
Vincent A. de Rooij
3. Nominal structures Nominal structures in Katanga Swahili are more analytic than those in ECS and ad/substrate languages. Katanga Swahili has retained the typically Bantu nounclass agreement system, but has done so in a reduced and simplified form. Table 3 lists the noun class prefixes of ECS and Katanga Swahili. Noun classes 1 through 10 are arranged pairwise where the even numbered class prefix denotes plurals and the odd numbered class prefix singulars (e.g. class 7 ki-tabu ‘book’ versus class 8 bi-tabu ‘books’). The number of noun class prefixes in Katanga Swahili has increased in comparison to ECS: although it has lost one (Cl. 10 collapses with Cl.6) it has added three (Cl. 11, 12, 13) which have been borrowed from Luba-Kasaï and Bemba. The differences in morphophonemic shapes of noun class prefixes 1, 2, 3, 5, 8, 11, and 14 can also be attributed to ad/substrate influence.
Table 3. Noun class prefixes in Katanga Swahili, ECS, Luba-Kasaï and Bemba (some variants of prefixes not shown for reasons of clarity) Noun class
ECS
Katanga Swahili
Luba-Kasaï
Bemba
1 2 3 4 5 6 7 8 9 10 10 → 6 11 12 13 14 15 16 17 18
mwammi-
mubamumiji-/ri-/∅makibiN-/∅-
mubamumidimatbiNN-
mubamumilimatfiniNniN-
lukatubukuɸakumu-
lukatubukupakumu-
makivi-
u-
ukupakumu-
malukatubukupakumu-
Katanga Swahili
127
It should be noted that infinitives are morphologically marked as nouns by noun class 15 prefix ku-. The locative classes 16, 17, and 18 have a different status than they have in ECS. In Katanga Swahili, pa-, ku-, and mu-, occur as pre-prefixes and may function as prepositions, as shown in (13) where mu is followed by an NP consisting of a demonstrative (ile) and a plural noun (mashiku ‘days’). (1)
(…) mais u-na-kufwa mu ile ma-shiku. but you-tma-die loc dem 6-day ‘(…) but you will die during that period.’ (Félicien/VDK1: 8/38)
In ECS, on the other hand, locative phrases are formed by suffixing a general locative affix -ni to a noun, while pa-, ku-, and mu- can only be affixed to noun modifiers. The use of locative prefixes as pre-prefixes in Katanga Swahili must be attributed to ad/substrate influence, since it occurs in all Central Bantu languages (Grégoire 1975: 17). The semantics of Katanga Swahili pa-, ku-, and mu- is the same as in the ad/substrate languages and ECS where, roughly speaking, pa- expresses a general locative meaning (at), ku- expresses direction (toward), and mu- expresses being inside of (in). Noun–Adjective agreement has been simplified radically: most adjectives have only one generalized form that is used with nouns from different classes. Agreement is marked most strongly in classes 1,2,7,8,12,13,14, especially in the plural classes 2 and 8 among these, but ultimately depends on the strength of the generalized form. This phenomenon does not occur in neighboring languages and can, therefore, not be the result of borrowing. Reduction and simplification is also found in subject and object agreement on the verb. In Katanga Swahili object concord markers co-indexing non-human objects are very seldomly used, except for classes 7/8. The generalized, but not categorical (Bostoen 1997: 106), use of i-, as a subject marker in classes 3 through 10 is striking. This restructuring cannot be explained by invoking ad/substrate influence, because the ad/substrate languages make use of the same agreement system as ECS, where markers have roughly the same morphophonemic shape as the prefixes of the nouns they refer to. The use of marker i- seems to correlate strongly with the feature [−human]: it does not occur with nouns belonging to classes 1/2, denoting human beings while its use is favored in all other classes except classes 7/8 and 11 through 14. Noun class prefixes 11 through 14 stand apart from the others in that they are used productively to derive nouns with very specific meanings (e.g. diminutives).
128
Vincent A. de Rooij
4. Verbal structures Katanga Swahili has the preverbal tma affixes that are typical of Bantu languages. According to Schicho (1988,1990) the ECS preverbal tense affixes that have survived in Katanga Swahili have lost much of their meaning as realizations of tense and may in many cases be regarded as a kind of dummyelements that have to be realized for morpho-syntactic reasons. The most frequently used preverbal tense affixes in Katanga Swahili are: -na- marking present tense, -li- or -ri- marking past tense, and -ta- marking future tense. According to Schicho (1988: 568), in a narrative sequence time reference needs to be marked only once by a tense affix on a verb, by sentenceinitial adverbs, or may even be left unexpressed if time reference can easily be inferred from contextual information. Schicho claims that the following distinctions provide the basis for what he calls the Aspect-dominated tma system of Katanga Swahili: 1. [+anterior] (including [perfect/resultative]) 2. [−anterior, −posterior] 3. [posterior/irrealis] 4. [progressive] (including [habitual], [intensive], [durative], [iterative]) tma is often expressed by auxiliary verbs. [+anterior] with perfective/resultative aspect can be expressed by using -toka (mu-) ‘leave, quit (loc)’, and -isha ‘finish’, also shortened to -sha as shown in (2). -isha is also used in this way in ECS but -toka (mu) is not. (2)
mi-na-isha ku-pakala vernis. I-tma-finish inf-apply varnish ‘I have already varnished it.’ (Schicho 1988: 569)
[posterior/irrealis] can be expressed by -tafuta ‘look for’ and by -enda ‘go’ as in (3). (3)
(h)a-ba-ta-enda ku-ra nani, ku-ra nkuku. neg-they-tma-go inf-eat filler inf-eat chicken ‘they will not eat ehm, eat chicken.’ (Michel/M1: 5/09)
Katanga Swahili
129
[progressive] with habitual, intensive, durative, or iterative aspects can be expressed by the verb -anza ‘start, begin’ as in (4) or by the copulative element -ko- used as a preverbal affix as shown in (5). (4)
a-na-anza ku-fan(ya) ma-bêtise. he-tma-start inf-do 6-stupidity ‘He started (and went on) doing foolish things.’ (Dédé m./M1: 7/08)
(5)
disons be-ko-na< be-ko-na-soigner eh? let’s say they-cop-tma< they-cop-tma-care for tag ‘Let’s say they provide medical care, don’t they?’ (Félicien/VDK1: 7/17)
In ECS -isha is used to mark perfective/resultative aspect but the use of -anza ‘start’ to mark habitual/durative aspect and -enda ‘go’ or -tafuta ‘look for’ to express inchoative aspect can be connected to semantically similar verbs in Luba-Kasai (Bostoen 1997: 121). In ECS and the ad/substrate languages, negation is expressed by a verbal prefix. In Katanga Swahili this strategy is still used with one important difference: in most cases, a sentence-final second negation element is added (cf. Schicho 1992: 84). In (6) we find (h)apana, originally a negative locative copula in ECS, but nowadays used almost exclusively as a negative answer to a yes/no question meaning. (6)
mais zamani (h)a-ba-ku-anz-ak-e ku-rima vile but long ago neg-they-tma-start-int-fin inf-cultivate thus (h)apana. neg ‘But long ago they didn’t cultivate (the fields) like that.’ (Papa Tshibangu/VT3: 1/04)
Periphrastic constructions, however, do also occur making use of Swahili verbs with an inherently negative meaning as in (7) and (8). This use of negative verbs does not occur in ECS but has been attested in the following ad/ substrate languages: Bemba, Luba-Katanga, Luba-Kasai, Lala, and Lamba (Kamba Muzenga 1981: 56). (7)
kama u-na-kosa kw-enda ku-bar hii. if you-tma-refuse inf-go loc-bar dem ‘If you don’t go to that bar.’ (Schicho 1990: 478)
130
Vincent A. de Rooij
(8) a-na-acha ku-tumika. he-tma-discontinue inf-work ‘He doesn’t work any more/longer.’ (Schicho 1990: 478) Another loss of verbal morphology in favor of periphrastic constructions is evident in relativization. In ECS, two relativization strategies occur: a relativizer may be infixed or suffixed on the verbal complex, or a relative pronoun with a relative concord suffixed to it may be used as in (9). This construction alternates with the one in (10) where the relative concord is infixed in the verbal complex: (9) mtu amba-ye a-li-mw-ona simba a-na-ogopa sana. man pron-rel he-past-it-see lion he-prs-fear much ‘The man who saw the lion is very frightened.’ (Vitale 1981: 90) (10) mtu a-li-ye-mw-ona simba a-na-ogopa sana. man he-past-rel-it-see lion he-prs-fear much ‘The man who saw the lion is very frightened.’ (Vitale 1981: 90) In Katanga Swahili we find a demonstrative, agreeing with the noun it modifies, which functions as a relativizer. It may be placed before or after the noun. Examples are given in (11)–(12). (11) u-ko d’accord na (h)ii mambo e-ko-na-sema? you-cop in.agreement with dem things she-cop-tma-say? ‘Do you agree with the things she’s saying?’ (Dédé m./M1: 2/10) (12) (h)aba ba-filles tu-ko-na-ona ba-na-zunguluka (h)umu. dem 2-girls we-cop-tma-see they-tma-walk around here ‘these girls we see walking around here.’ (Fidélie/M1: 4/10) The development of relativizers from deictic elements is a wide-spread phenomenon in contact-induced languages (Romaine 1988: 250). In the case of Katanga Swahili, this common trend has undoubtedly been reinforced by ad/ substrate pressure from Bemba (de Rooij 1997: 330).
5. Other parts of speech Numerals in dates and years are mostly in French. Discourse markers, including items that are traditionally classified as conjunctions, constitute another
Katanga Swahili
131
group of items that occur almost exclusively in French (De Rooij 2000). Frequently occurring are bon as a marker of topic development and transition, non as a quotation marker, mais as a marker of contrast, puisque and parce que as markers of causal relation and alors, donc and et puis as markers of conclusion and succession. Examples of donc, et puis, are given in (13); mais occurs here as an element initiating a switch to French. In (14), mais is used repeatedly in a monolingual Swahili fragment. (13) njo eh Mungu, ni Mungu w-a richesse, top eh God cop God 1-conn wealth ‘God, is a god of riches, (0.5) (0.5) ni Mungu w-a or, ni Mungu cop God 1-conn gold cop God is a God of gold, is a God w-a argent. (0.5) ↓donc (h)ii richesse 1-conn silver (0.5) so dem wealth of silver. (0.5) therefore, all this riches yote (h)ii i-na-tu-appartenir shi all dem it-tma-us-belong to we belongs to us, we who are ba-toto yake. (1.0) et puis, Mungu 2-children poss.3sg (1.0) furthermore God His children. (1.0) furthermore, God a-shi-na Mungu w-a bu-chafu mais he-neg-cop God 1-conn filth but is not a God of squalor mais (but) c’est un Dieu de la propreté. it’s det God of det cleanliness it’s a God of cleanliness.’ (Fidélie/DM13) (De Rooij 2000: 461 [glosses and translation slightly adapted, VDR])
132
Vincent A. de Rooij
(14) njo ku-sema: lu-fu ya bamalaika: c’est que top inf-say 11-death conn 2-angel it.means.that ‘It means that it was the death of angels: it means that mu-ntu: a-ju-e ma-neno: mais a-na-kufa tuu 1-man he-know-neg 6-word but he-tma-die just/merely a human being who does not understand what is happening but still bule. mm? mu-ntu yee hapana ku-jua ma-neno: for.nothing tag 1-man he neg inf-know 6-word he dies a senseless death. see? a man who does not understand what mais yee a-na-kufa: bule. but he he-tma-die for.nothing is happening but he dies a senseless death.’ (Fabian 2000: #5) [translation slightly adapted, VDR]) Question words, demonstratives, connectors, and free pronouns, all remain Swahili. In De Rooij (2000) it is suggested that quasi wholesale borrowing of French markers has pragmatic and syntactic explanations. The pragmatic explanation holds that French markers stand out better in Swahili environments and, hence, are more salient and effective as discourse structuring devices. Using Fench markers in Swahili is, furthermore, not a problem from a syntactic perspective since markers are not or hardly integrated in the morphosyntactic frame of sentences.
6. Conclusion It is clear that features of several Bantu languages have been borrowed into Katanga Swahili: auxiliary verbs to express aspect, locative noun prefixes (that are similar to prepositions), negative verbs to express negation, and an analytic relativization strategy. It is not always possible to identify one language as the source language because of the strong structural similarities between potential source languages. It is clear, however, that those elements that have been borrowed into the language can all be described as analytic. The change from an agglutinative morphosyntax to a more analytic structure is all the more striking since all of the languages involved are related Bantu languages, all of them agglutinative in structure. This particular change may
Katanga Swahili
133
be explained as a result of strategies promoting semantic transparency in first language acquisition and second language learning under less than ideal learning conditions. The other major instance of borrowing, the use of French discourse markers, fits a general pattern in language contact, well-documented in many studies (see e.g. the contributions to Maschler, ed. 2000).
Abbreviations and transcription conventions (0.5) ↓ 2,6-[noun] cop dem fin inf int
pause (in seconds) sharp low-falling pitch contour class 2 or 6 noun prefix copula demonstrative final vowel of verb form infinitive intensive aspect
loc neg past prs rel tag tma top
locative negation past tense present tense relativizer tag tense/mood/aspect topicalizer
Notes 1. This chapter is based in part on de Rooij 1995, 1996, 1997. I am grateful to those who commented on these earlier publications and to the editors of this volume for their comments. All usual disclaimers apply. Unless indicated otherwise, all examples in this chapter were selected from the author’s fieldwork data. Fieldwork was carried out in Lubumbashi, dr Congo, in 1991 (June–October) and 1992 (June–December) with grants from the Institute for Functional Research into Language and Language Use (IFOTT) and the Netherlands Foundation for the Advancement of Tropical Research (WOTRO). The financial support of both institutions is hereby gratefully acknowledged. 2. In 1971, Katanga was renamed into Shaba as part of Mobutu’s policy of ‘zaïrianization’. After Mobutu’s removal from power in 1997, this renaming was undone and the province was given back its former name, Katanga. 3. Due to the fact that nearly all research on Swahili as spoken in Katanga has been carried out in Lubumbashi, the capital of Katanga, some authors refer to the language as ‘Lubumbashi Swahili’ (see e.g. Gysels 1992, Polomé 1968, Schicho 1982). How and to what degree Swahili as spoken on the Copperbelt resembles varieties of Swahili spoken elsewhere in Katanga remains an open question.
134
Vincent A. de Rooij
References Blommaert, Jan 1999 Reconstructing the sociolinguistic image of Africa: Grassroots writing in Shaba (Congo). Text 19: 175200. 2004 Writing as a problem: African grassroots writing, economies of literacy, and globalization. Language in Society 33 (5): 643671. Bostoen, Koen c. 1997 Het Shaba-Swahili: Geschiedenis en bronnen [Shaba Swahili: History and sources]. MA Thesis, Ghent University. Burssens, Amaat 1939 Tonologische Schets van het Tshiluba (Kasayi, Belgisch Kongo). Antwerpen: De Sikkel. Fabian, Johannes 1986 Language and Colonial Power: The Appropriation of Swahili in the Former Belgian Congo, 18801938. Cambridge: Cambridge University Press. 1990 History from Below: The ‘Vocabulaire of Elisabethville’ by André Yav (Texts, translations, and interpretive essay). Amsterdam: John Benjamins. 2000 The history of Zaire as told and painted by Tshibumba Kanda Matulu in conversation with Johannes Fabian, Third Session, Part 2. Archives of Popular Swahili 2 (7). http://www2.fmg.uva.nl/lpca/aps/tshibumba3b.html (3 August 2007). Fetter, Bruce 1976 The Creation of Elisabethville 19101940. Stanford: Hoover Institution Press. Grégoire, Claire 1975 Les locatifs en bantu. (Annales du Musée de l’Afrique Centrale, Série in-8º, Sciences Humaines, no 83.) Tervuren: Musée de l’Afrique Centrale. Gysels, Marjolein 1992 French in urban Lubumbashi Swahili: Codeswitching, borrowing, or both? Journal of Multilingual and Multicultural Development 13 (1/2): 4155. Kabamba, Mbikay 1979 Stratigraphie des languaes et communications à Lubumbashi. Problèmes sociaux zaïrois 124125: 4774. Kamba Muzenga 1981 Les formes verbales négatives dans les langues bantoues. (Annales du Musée de l’Afrique Centrale, Sciences Humaines 106.) Tervuren: Musée de l’Afrique Centrale.
Katanga Swahili
135
Kapanga, André Mwamba 1993 Shaba Swahili and the processes of linguistic contact. In: Francis Byrne and John Holm (eds.), Atlantic Meets Pacific: A Global View of Pidginization and Creolization, 441458. Amsterdam: John Benjamins. Kashoki, Mubanga E. 1968 A Phonemic Analysis of Bemba: A Presentation of Bemba Syllable Structure, Phonemic Contrasts and Their Distribution. (Zambian Papers 3.) Manchester: Manchester University Press. Maschler, Yael (ed.) 2000 Discourse Markers in Bilingual Conversation. Special issue of International Journal of Bilingualism 4 (4). Möhlig, Wilhelm J. G. 1981 Die Bantusprachen im engeren Sinn. In Bernd Heine, Thilo C. Schadeberg, und Ekkehard Wolff (eds.), Die Sprachen Afrikas, 77116. Hamburg: Helmut Buske Verlag. Polomé, Edgar C. 1968 Lubumbashi Swahili. Journal of African Languages 7 (1): 1425. Romaine, Suzanne 1988 Pidgin and Creole Languages. London: Longman. De Rooij, Vincent A. 1995 Shaba Swahili. In: Jacques Arends, Pieter Muysken, and Norval Smith (eds.), Pidgins and Creoles: An Introduction, 179190. Amsterdam/ Philadelphia: John Benjamins. 1996 Cohesion through Contrast: Discourse Structure in Shaba Swahili/ French Conversations. Amsterdam: Ifott. 1997 Shaba Swahili: Partial creolization due to second language learning and substrate pressure. In: Arthur K. Spears and Donald Winford (eds.), The Structure and Status of Pidgins and Creoles, 309339. Amsterdam/Philadelphia: John Benjamins. 2000 French discourse markers in Shaba Swahili conversations. International Journal of Bilingualism 4 (4): 447467. Schicho, Walter 1982 Syntax des Swahili von Lubumbashi. Vienna: Afro-Pub. 1988 Tense vs. aspect in Sango and Swahili of Lubumbashi. In: Mohammad Ali Jazayery and Werner Winter (eds.), Languages and Cultures: Studies in Honor of Edgar C. Polomé, 565579. Berlin: Mouton de Gruyter. 1990 AUX, Creole und Swahili von Lubumbashi. Zeitschrift für Phonetik, Sprachwissenschaft und Kommunikationsforschung 43 (4): 476483. 1992 Non-acceptance and negation in the Swahili of Lubumbashi. African Languages and Cultures 5 (1): 7589. Vitale, Anthony J. 1981 Swahili Syntax. Dordrecht: Foris.
Grammatical borrowing in Khuzistani Arabic Yaron Matras and Maryam Shabibi
1. Background The Khuzistani dialect of Arabic (henceforth Kh. Arabic) is spoken natively by over 3 million people, who constitute roughly 7 percent of the population of the province of Khuzistan in south-western Iran. The dialect is the easternmost representative of the continuum of Mesopotamian dialects of Arabic, which cover the river lands of southern Iraq in the west (Ingham 1982: 14). Arab settlement in the area is believed to go back to the beginning of the Christian era. In the centuries following the advent of Islam, the Arabic language enjoyed the status of the literary language of religion, scholarship, and administration, as well as being the primary language of everyday communication in the province. This changed with the coming to power of Reza Shah Pahlavi in 1926 and the introduction of an intensive campaign favouring Persian as the only official state language. The policy included the settlement of a Persian-speaking population in the province. Unlike other dialects of Arabic, Kh. Arabic has not attracted much attention within the linguistic community. Ingham (1997) devotes a chapter to the dialect, focusing, however, on an introductory discussion of phonology and vocabulary only; and Shabibi (2006) provides an overview of the structures of the dialect along with an analysis of contact-induced developments in morphosyntax. Persian is now the only language of education, local media and newspapers, administration, and most urban commerce in the province of Khuzistan. Arabic is the language of the family and Arabic-speaking neighbourhoods, though even as an informal language it is now in decline, and Persian is the preferred language of the younger generation born since the 1970s. All educated adult speakers of Arabic are bilingual, and Arabic monolingualism is limited to the uneducated older generation, and to the older generation in rural communities. Arabic literacy is limited by and large to reading the Qur’an, and to a very basic level of instruction in Modern Standard Arabic, though even most educated Arabs have no active command of Modern Standard Arabic. There is, however, considerable exposure to Arabic-language satellite media, and so to the broadcast (oral) version of Modern Standard Arabic. In some
138
Yaron Matras and Maryam Shabibi
cases, Khuzistani Arabic speakers are able to read modern Arabic by drawing on their exposure to these media, combined with their basic familiarity with the Arabic script and with Classical Arabic (Qur’an).
2. Phonology The only apparent phonological contact phenomenon in Kh. Arabic is the interchange of /ɣ/ and /q/, in words such as /ɣarīb, qarīb/ ‘close’ (cf. Modern Standard Arabic /qarīb/ ‘close’, /ɣarīb/ ‘strange’). This matches the realization in Persian of etymological /q/ as /ɣ/. Phonemes that are otherwise absent from the Arabic system, most notably /p/, are retained in Persian loanwords: panjara ‘window’,
3. Morphological typology A major change under Persian influence is the levelling of the status of attributes. In Arabic, adjectival attributes follow the head noun, and agree with the head noun in gender, number, as well as in definiteness: (1)
Standard Arabic (and other dialects) a. walad kabīr boy big.m ‘a big boy’ b. l-walad l-kabīr def-boy def-big.m ‘the big boy’
Nominal attributes, by contrast, are conjoined by means of the attributive Iḍāfa-construction, whereby only the dependent (genitive) noun is overtly marked for definiteness: (2)
Standard Arabic (and other dialects) walad l-mudīr boy def-director ‘the director’s son’
Khuzistani Arabic
139
In Persian, both types of attributes are treated in the same way: The attribute (whether adjectival or nominal) follows the head, and an attributive particle (the Ezāfe marker) mediates between the two: (3)
Persian a. pesar-e bozorg boy-ez big ‘the big boy’ b. pesar-e modīr boy-ez director ‘the director’s son’ c. xūne-ye sefīd house-ez white ‘the white house’ d. moʔallem-e madrese teacher-ez school ‘the school teacher’
The pattern in Kh. Arabic matches the Persian arrangement (note that, as in other dialects of Arabic, the definite article l- assimilates to dental consonants, resulting in gemination of that consonant): (4)
Khuzistani Arabic a. walad č-čibīr boy def-big.m ‘the big boy’ b. walad l-modīr boy def-director ‘the director’s son’ c. bīәt l-abyaḍ house def-white ‘the white house’ d. moallәm-at l-madrәsa teacher-f.cons def-school ‘the school teacher’
Note that in the adjectival attributive construction in (4a and 4c) overt definiteness agreement between noun and adjective is lacking, just like in the
140
Yaron Matras and Maryam Shabibi
genitive attribute construction in (4b and 4d). Based on the Persian model, Kh. Arabic has reanalysed the definite article in such constructions as a marker of attribution, which matches the Persian (definite) Ezāfe marker -(y)e. Its distribution now resembles that of the Persian Ezāfe attributive marker: It appears, like Persian -(y)e, between the two constituents of the attribution, and it is used to link both adjectival, and nominal attributes. Further evidence that the functions of the Persian construction are mapped onto Arabic structures is provided by the position of the feminine Construct State or Iḍāfa-marker -at, seen in (4d) in a position that is not untypical of Arabic as a whole. In Arabic, the Construct State marker (still recognisable in the vernaculars only in the feminine singular) is reserved for nominal attribution, as in (4d). But in Kh. Arabic we find it in adjectival attributive constructions as well, as in (5a–b); it even attaches directly to adjectives, as in aly-at ‘high.f’ in (5b): (5)
Khuzistani Arabic a. jazīr-at l-xaḍra island-f.cons def-green ‘the green island’ b. ṭōf-at aly-at l-bīәt wall-f.cons high-f.cons def-house ‘the high wall of the house’
This matches again the distribution of the Persian Ezāfe marker -(y)e (6): (6)
Persian a. jazīre-ye sabz island-ez green ‘the green island’ b. dīvār-e boland-e xūne wall-ez tall-ez house ‘the high wall of the house’
Note that in the ‘mixed’ type, in (5b), involving both an adjectival-attribute (‘high wall’) and a genitive attribute (‘wall of the house’), the first (adjectival) attribution relies exclusively on the Construct State marker, while the second (nominal) relies on the combination of the Construct State marker with the following definite article. In fact, Kh. Arabic allows for variation in such cases, and the Construct State marker may be accompanied by a definite art-
Khuzistani Arabic
141
icle in both positions. Consider example (7), where the nouns are masculine, and there is no option of using an overt Construct State marker: (7)
a. Khuzistani Arabic walad č-čibīr l-modīr boy def-big def-director ‘the director’s big/eldest son’ b. Standard Arabic (and other dialects) walad l-mudīr l-kabīr boy def-director def-big ‘the director’s big/eldest son’ c. Persian pesar-e bozorg-e modīr boy-ez big-ez director ‘the director’s big/eldest son’
The crucial aspect of the Kh. Arabic construction is (1) to have a marker of attribution mediating between the head and its attribute, (2) to place the attribute in a position immediately following its head, and (3) to avoid any overt marking of definiteness in the adjectival attribution. In all this, Kh. Arabic copies precisely the Persian attributive construction. Contrasting with Persian, it retains a distinct marking of attribution with feminine singulars, but allows this marking to assimilate into the generic function of the attributive marker. The outcome of the process is (1) the loss of the distinction between nominal and adjectival attribution, (2) the loss of overt marking of definiteness in attributive constructions, (3) a change in the word order in complex (‘mixed’) attributive constructions (as in 5b and 7a), and, finally, (4) gender variation in the marking of the attributive construction, with optional use of the definite article to accompany the Construct State in feminine singulars in complex attributions.
4. Nominal structures The most notable contact-induced change in Kh. Arabic nominal structures is the status of the Iḍāfa-construction alluded to above. The replication of a construction type that is similar to the Persian Ezāfe leads, as discussed above, to the abandonment of definiteness agreement. The decline of overt definiteness marking can also be observed in other constructions in the language,
142
Yaron Matras and Maryam Shabibi
notably in the absence of a definite article with definite head nouns of relative clauses. This too follows a Persian model (where definiteness generally remains unmarked): (8)
Khuzistani Arabic mara lli šift-ū-ha xābar-at. woman rel saw-2pl.m-3sg.f called-3sg.f ‘The woman that you saw called.’
5. Verbal structures In the derivation of verbs, the tendency to paraphrase inchoative and causative verbs drawing on an analytic construction rather than on derivational morphology, although found in other dialects of Arabic, appears to be reinforced by Persian. Thus we find: (9)
Khuzistani Arabic š-šijra z-zɣīr-a šwayye šwayy tṣīr čibīr-a. def-tree def-small-f little little become.3sg.f big-f ‘The small tree gradually grows.’
Loan-verbs appear to be limited to the replication of Persian compound verbs consisting of a nominal stem (masdar) and a verbalizing element or ‘light verb’ (Persian kardan ‘to do’ or šodan ‘to become’). The nominal stem, often itself an Arabic loan into Persian, is replicated directly in Kh. Arabic, while as corresponding native light verb ṣaww- ‘to do’ is employed for Persian kardan, and ṣār- ‘to become’ for Persian šodan, thus: Persian taaɣībeš kard ‘he followed him’ (follow-3sg did.3sg) is rendered ṣawwā-h taɣīb (did.3sg-3sg follow). An additional change to the verb system, brought about through Persian influence, concerns the tense system. Persian has both a simple past tense, which is expressed by the person-inflected past stem of the verb, and a composite past tense, which consists of a past participle and an auxiliary. The auxiliary, based on the existential verb, may inflect for person as well as tense; the present-tense auxiliary is used to form the perfect, the past-tense auxiliary forms the pluperfect. Arabic, by contrast, has only one, simple past tense, though combinations of the past-tense existential verb with the lexical verb (usually in the imperfect or present-future) are also possible, usually expressing habitual aspect of conditional mood. Kh. Arabic copies the Per-
Khuzistani Arabic
143
sian composite past tense, drawing on inherited resources. The only available participle form in Arabic is the present participle, which inflects for gender and number (but not for person), and it is this form that serves as the basis for composite past tense in Kh. Arabic. Since the Arabic existential verb does not have a present-tense form, the only available auxiliary is a past-tense auxiliary; the construction thus matches the Persian pluperfect: (10) a. Khuzistani Arabic mәn rәħ-әt. lә-l-bīet, huwwa mā-rāyәħ čān when went-1sg to-def-home he neg-going.sg.m was.3sg.m ‘When I went home he had not gone away.’ b. Persian vaɣti raft-am xūne, ūn na-rafte būd. when went-1sg home he neg-gone was.3sg.m ‘When I went home he had not gone away.’ (11) a. Khuzistani Arabic mәn gabul šāyfat-ha čәnәt. from past seeing.sg.f-3sg.f was.1sg ‘I had seen her before.’ b. Persian az ɣabl ūn-o dīde būd-am. from past 3sg-acc seen was-1sg ‘I had seen her before.’
6. Other parts of speech A series of Persian discourse markers, fillers, tags, and focus particles are used in Kh. Arabic. Most of these elements are well integrated into Kh. Arabic and are not perceived by speakers as foreign. The category that is most obviously influenced by Persian is that of discourse markers with a primarily interaction-qualifying rather than syntactic-semantic function: xō/xōb/xōš ‘well’, xōlāse ‘in sum’, albate ‘of course’, hič ‘at all, altogether’, ham ‘indeed, well’: (12) xōb w-hāy sabab ham l-laði gabal čān … dm and-this reason dm rel once was ‘Well, and the reason that indeed once existed for this …’
144
Yaron Matras and Maryam Shabibi
(13) xolāse hīč mā-rəħ-na madrasa. dm dm neg-went-1pl school ‘After all, we didn’t go to school at all.’ (14) albate čān-an ham b-ðīč z-zamān banāt č-čān-an dm were-3pl.f dm in-that def-time girls rel-were-3pl.f yarħ-an. go-3pl.f ‘Of course there were indeed girls at the time who used to go [to school].’ These are accompanied by Persian-derived focus particles: ham ‘too’ and ham … ham ‘both … and’. (15) ðīč ənd-ha θnīən frūx ana ham ənd-ī θnīən. that.f poss-f two children I too poss-2sg two ‘She has two children, and I have two children, too.’ (16) ham ana w ham alī rəħ-na l-əl-pārk. both I and also Ali went-1pl to-def-park ‘Both Ali and I went to the park.’ (17) umm-ī ham ɣəsl-at lə-mmāīn ham naḍḍəf-at mother-1sg both washed-3sg.f def-dishes and cleaned-3sg.f l-bīət. def-house ‘My mother [both] washed the dishes and cleaned the house.’ Optional, occurring in variation alongside various Arabic-derived counterparts such as ħatta ‘even’ or lākin ‘but’, is the contrastive correlative balke ‘but [… also]’: (18) huwwa mū bass bāhūš balke šujjā ham. he neg only clever but brave too ‘He is not only clever but also brave.’ Further Persian borrowings that are generalized in Kh. Arabic are the concessive subordinating conjunctions agarče and bā īnke, both 'although/ even though’, and the factual complementizer ke ‘that’:
Khuzistani Arabic
145
(19) huwwa rāħ lwaħda l-əl-pārk agarče umm-a he went.3sg.m alone to-def-park although mother-3sg.m gall-at l-a lā-yrūħ. said-3sg.f to-3sg.m neg-go.3sg.m ‘He went to the park alone, even though his mother told him not to go.’ (20) rayyāl-na bə-l-yōm xəṭab, bā īnke θaləθtaš sana man-1pl in-def-day proposed.3sg.m although thirteen year umr-ī sawwūm rāhnamāī ubū-y qəbal b-ī age-1sg third secondary school, father-1sg accepted for-3sg.m ‘When my husband proposed, although I was [just] thirteen years old, third year of secondary school, my father agreed.’ (21) tədr-īn ke rayl-əč ala kəl-šī čaððab. know-2sg.f comp husband-2sg.f on everything lied.3sg.m ‘You know that your husband lied about everything.’ The latter, the Persian complementizer and relativizer ke, does not appear in non-factual (subjunctive) complements, where instead we find the Arabic (historical) relativizer l-laði or illi, which also continues to cover the function of a relativizer. Nonetheless, occasionally Persian ke is also found in the position of the relativizer: (22) əbən uxū ɣāzī ke huwwa w mart-a hnā … son brother Ghazi rel he and wife-3sg.m here ‘Ghazi’s nephew, who is here with his wife …’ From this we might assume a gradual process of convergence in steps, as follows: in stage 1, the Persian model of having an identical marker for complement clauses and relative clauses (ke) is copied into Kh. Arabic, with the effect of generalizing the relativizer l-laði/illi (at the expense of the historical Arabic complementizer ’inn-) to cover the function of complementizer. The result is a convergence of patterns among the two languages. In stage 2, the actual Persian marker ke is adopted into Kh. Arabic in factual complement clauses, as seen in (21). The result is a split within Kh. Arabic between factual and non-factual complements, whereas the same marker is used in both languages to introduce factual complements. Finally, in stage 3, the beginnings of which are attested in the contemporary language, Persian ke infiltrates Kh. Arabic relative clauses as well, as seen in (22).
146
Yaron Matras and Maryam Shabibi
7. Constituent order One change in constituent order has already been mentioned above, in Section 2: it concerns the shift in ‘complex’ attributive constructions, away from the Arabic norm, which allows an adjectival modifier to be separated from its head (by a nominal modifier of the complex noun phrase), toward the Persian-type constituent order, whereby each attribute must immediately follow its head. We repeat example (7) here: (7)
a. Khuzistani Arabic walad č-čibīr l-modīr boy def-big def-director ‘the director’s big/eldest son.’ b. Standard Arabic (and other dialects) walad l-mudīr l-kabīr boy def-director def-big ‘the director’s big/eldest son’ (also: ‘the big director’s son’) c. Persian: pesar-e bozorg-e modīr boy-ez big-ez director ‘the director’s big/eldest son’
A further issue related to the order of constituents in Kh. Arabic concerns the position of the copula-auxiliary /čān/, which, in the composite past tense (pluperfect), follows the lexical verb: mā-rāyәħ čān ‘he had not gone away’ (Persian: na-rafte būd) (see examples 10 and 11). Noteworthy is also the flexible position of the causal conjunction čīe ‘because’. Like its Persian counterpart čon, it can also occupy the final position in the adverbial clause expressing cause: (23) a. Khuzistani Arabic līeš mā-reħ-tī l-әl-madrasa? why neg-went-2sg.f to-def-school čān edd-ī xuṭṭār čīe. was.3sg with-1sg guests because ‘Why didn’t you go to school?’ ‘Because I had guests.’
Khuzistani Arabic
147
b. Persian čerā be madrese na-raft-ī? why to school neg-went-2sg mehmūn dāšt-am čon. guest had-1sg because ‘Why didn’t you go to school?’ ‘Because I had guests.’ Finally, we must consider what appears to be the beginning of a shift in word order, extending the contexts in which Object–Verb order is favoured to comply more frequently with the Persian type. Object–Verb order in Arabic is generally highly marked and is employed as a means to topicalize the direct object. Kh. Arabic makes use of such strategies, which include – unlike Persian, where OV prevails – the pronominal resumption of the object in a position following the lexical verb. Nevertheless, such constructions in Kh. Arabic do not necessarily express the topicalization of the object: (24) lə-bnayya d-dār naḍḍəf-at-ha. def-little.girl def-room cleaned-3sg.f-3sg.f ‘The little girl cleaned [it] the room.’ (25) haðan xālāt-ī līsāns-hən kaẓẓ-ann-a. these aunts-1sg degree-3pl.f gained-3pl.f-3sg.m ‘My aunts received [it] their degree.’
8. Lexicon The presence of numerous Persian lexical borrowings is a distinguishing feature of Kh. Arabic, setting it apart from other neighboring dialects of Arabic. Nevertheless, there is considerable sociolinguistic stratification in the use of Persian vocabulary among different groups of speakers (cf. Shabibi 1998). As the principal language of the public sphere, Persian supplies numerous lexical items in the domains of trade, institutions, tools, and other aspects of public and technical life (e.g. xarīd-o-furūš ‘trade’, pīč guštī ‘screwdriver’, lebās šū’ī ‘washing machine’, etc.). In everyday vocabulary, Persian idioms are commonly calqued in Kh. Arabic, facilitated by the fact that those idioms themselves are often based on Arabic loan vocabulary in Persian, and so even more easily replicable in Kh. Arabic: Consider Kh. Arabic wāyәd mamnūn, lit. ‘very grate-
148
Yaron Matras and Maryam Shabibi
ful’, in the sense of ‘thank you very much’, based on Persian xeyli mamnūn, or Kh. Arabic yarreti zaħma, lit. ‘you have taken trouble [on my behalf]’, also an expression of gratitude, from Persian zahmat kešīdī. Here, the fact that the languages already share a large part of their vocabulary (as a result of earlier, historical influence of Arabic on Persian), makes replication of lexical Matter redundant, and promotes in turn replication of idiomatic Patterns surrounding a pivotal word in the idiom that is already shared by both languages.
9. Conclusion Matter replication of Persian material is found in Kh. Arabic primarily in the domain of lexical vocabulary, and in part in grammatical vocabulary, covering discourse markers that operate strictly on the interaction level (i.e. not conjunctions), focus particles, a correlative particle, a complementizer and relative particle, and concessive subordinating conjunctions. Pattern replication is most notable in the emerging change of constraints on word order (extension of marked word-order patterns), the favouring of analytic constructions and emergence of a new analytic past tense (pluperfect), and the reduction of overt marking of definiteness. Perhaps the most remarkable contact-induced change, one which strongly affects the typology of attribution in the language, is the identification of Kh. Arabic grammatical morphemes in attributive constructions – the Construct State marker (visible in the feminine singular only) and the definite article that appears between head and attribute – with the Persian attributive particle, and the consequent merger of two historically distinct attributive constructions – adjectival and nominal – into a single type, replicating the state of affairs in Persian.
Abbreviations acc comp cons cop def dm ez f
accusative complementizer construct state marker copula definite article discourse marker Persian Ezāfe attributive marker feminine
indef m neg past pl poss rel
indefinite article masculine negation simple past (perfective) plural possessive expression relative particle
sg
singular
Khuzistani Arabic
149
References Ingham, Bruce 1982 North East Arabian Dialects. London: Kegan Paul International. 1997 Arabian Diversions: Studies on the Dialects of Arabia. Reading: Ithaca Press. Shabibi, Maryam 1998 Variation in the use of Persian loan words among Iranian Arabic speakers. Unpublished Msc. Dissertation. University of Edinburgh. 2006 Contact-induced grammatical changes in Khuzistani Arabic. Ph.D. thesis. University of Manchester.
Grammatical borrowing in Domari Yaron Matras
1. Background Domari (also Domi, sometimes also Qurbati) is the Indo-Aryan language spoken by a population of commercial nomads in the Middle East. The language retains some archaic Indo-Aryan features, such as the Middle Indo-Aryan present-tense conjugation, but also shows some radical re-structuring in the past-tense conjugation and in syntactic typology. In this respect it resembles Romani, also an Indo-Aryan diaspora language of originally peripatetic groups. The self-appellations rom and dom are also related, both deriving from Indic ḍom. However, some isoglosses separating the two language appear to be rather ancient, and it is highly unlikely that they both split from the same ancestor language after leaving India. The present chapter deals with the only dialect that has been extensively documented – that of the Palestinian Doms of Jerusalem (see Matras 1999; Macalister 1914). The language is confined strictly to oral use within the family and, to a limited extent, with members of other Dom communities. The precise history and date of arrival of the group in the region remain unknown. The language shows a layer of Kurdish influence, and the community has a sense of affinity to another, Arabic-speaking population of commercial nomads who are referred to as “Kurds”. Arabic has been the principal contact language for many centuries. The Jerusalem community began shifting to Arabic in the 1960s, and individuals who were raised since this period are largely monolingual in Arabic, with only passive exposure to Domari. It is estimated that out of a total number of between 6001000 community members, only around 10 percent speak Domari fluently; the language is thus endangered or even moribund. Whatever information is available suggests that this is also the situation at least in urban Dom communities elsewhere in the Middle East.
2. Phonology The Domari sound system strongly resembles that of its contact language, Palestinian Arabic, though it is not always obvious that this is due to borrowing
152
Yaron Matras
or convergence. All consonants with the exception of /p/, /v/ (interchangeable with /w/), /tš/ and /g/ are shared with Arabic. Uvular /q/ appears also in the pre-Arabic component and may be the outcome of Iranian (Kurdish) influence (e.g. qišṭoṭa, alongside kišṭoṭa ‘small’). The glottal stop /ʔ/ is distinctive only in the postposed negation marker -éʔ: (1)
n-mang-am-éʔ neg-want-1sg-neg ‘I don’t want.’
The pharyngeals [ħ] and [ʕ] appear to be restricted to Arabic-derived lexical loans, but pharyngealization of dentals /t, d, s, z/ is transferred to the preArabic component as well: [wɑ:ṭ] ‘stone’, [ḍɑnḍ] ‘tooth’. Consonant gemination also appears independently of the Arabic component: [tilla] ‘big’. Under Arabic influence, the affricates /tš, dž/ are being reduced to sibilants /š, ž/. In the case of the voiced affricate, a similar change has fairly recently taken place in the Arabic dialect as well, and some variation is still observable. All vowel sounds with the exception of [ɔ] and [ʌ], both rather infrequent, are shared with Arabic. As in Arabic, there is variation in the realization of the short vowel phonemes /i/ [i, , ɪ] /a/ [a, æ, ɑ] and /u/ [u, ʉ, ʊ], and of long /ā/ [a: æ:, ɑ:], with back vowels preferred in the vicinity of uvular and pharyngealized consonants, as in [ṭɑ:ṭ] ‘Arab’, but [tat] ‘heat’. As in Arabic, the vowels /u/ and /i/ are often interchangeable: džuwir/džiwir ‘woman’ (also a feature of Kurdish). Prothetic and epenthetic vocalization around initial consonant clusters can also be regarded as a general regional phenomenon. Prosody and intonation are largely shared with Arabic.
3. Morphological typology As a New Indo-Aryan language, Domari will have undergone a re-structuring of its past tense formation leading presumably at some stage to the emergence of split morphological ergativity. Evidence to this effect is the use of the originally oblique form of the pronoun for ama ‘I’, and the construction of the past-tense conjugation, based on the attachment of what once were oblique personal clitics (kard-o-m ‘I did’ < *karda-o-me ’done-by-me’) (cf. Matras 2002: 145151). The language is, however, no longer ergative, though this might be attributed to the general drift away from ergativity in northwestern Indo-Aryan frontier languages as well as in Iranian, and there is no con-
Domari
153
crete evidence linking it with the influence of Arabic, which is not ergative, either.
4. Nominal structures The overall similarity in word-order rules results in a similar positioning of nominal objects in the sentence. The most extensive Arabic influence on nominal structures is in the domain of local relations, more specifically the almost wholesale borrowing of Arabic prepositions. We can assume that Domari lacked prepositions altogether before contact. Local relations expressions that are not borrowed are expressed either by case suffixes, or by genitivepossessive location expressions which consist of a location adverb inflected for (oblique) possession and the Locative case, preceding a head in the Ablative. There is only a very small set of such inherited location adverbs, all expressing strict spatial relations: mandža ‘in’, bara ‘out’, paš ‘behind’, agir ‘in front’, atun ‘above’, axār ‘below’, and čanč- ‘next to’: (2)
čanč-is-ma kury-a-kī next.to-3sg.obl-loc house-obl.f-abl ‘next to the house’ (lit. ‘in its-side from-the-house’).
Arabic-derived prepositions may be integrated into this format, as long as the meaning is stative: (3)
žamb-is-ma lāč-a-ki next.to-3sg.obl-loc girl-obl.f-abl ‘next to the girl’ (< Arabic žamb ‘next to’).
Temporal and more specified spatial relations are generally expressed through Arabic prepositions, the nouns appearing in the Ablative case (serving as a general prepositional case): (4)
baʕd wars-ak-ki after year-indef-abl ‘after a year’ (< Arabic bad ‘after’)
The Arabic prepositions ma ‘with’, min ‘from’, la ‘to’, fī ‘in’and ind ‘at’ compete with the synthetic cases Associative, Ablative, Dative, and Locative
154
Yaron Matras
respectively. Occasional doubling of case may be observed, e.g. Arabic fī ‘in’ + Locative case -ma (see example 6), though in general these prepositions too trigger the default Ablative/Prepositional case on the noun (example 5): (5)
maʕ bɔy-im-ki with father-1sg.obl-abl ‘with my father’ (< Arabic ma ‘with’)
(6)
fī šare-ma in street-loc ‘on the street’ (
Non-Arabic expressions are preferred only with pronominal clitics: (7)
a. ab-us-ke for-3sg-ben ‘for him’ b. minšān zirt-an-ki for child-obl.pl-abl ‘for the children’ (< Arabic minšān ‘for’)
(8)
a. nkī-man at-1pl ‘by us/at ours’ b. ʕind wud-as-ki at old-obl.m-abl ‘at the old man’s’ (< Arabic ind ‘at’) Domari has two formats for the possessive construction:
(9)
a. bɔy-im kuri father-1sg.obl house b. kury-os bɔy-im-ki house-3sg.nom father-1sg.obl-abl ‘my father’s house’
The second, (9b), in the order head–determiner, is by far the more widespread, and contrasts with the normal Indo-Aryan (including Romani) determiner– head construction. It matches however the Iranian type (cf. Kurdish mal-a
Domari
155
bav-ê min ‘house-det father-det me’) as well as the Arabic type bēt abū-y ‘house father-1sg’, and is matched even more closely by the (less frequent) Arabic construction bēt-ō la-ʔabū-y ‘house-3sg to-father-1sg’. As in Arabic, citation forms of many inalienable nouns must include possessive marking, thus bɔy-om ‘my father’ for ‘father’, cf. Arabic abū-y. Arabic plural suffixes are often retained with Arabic nouns, and usually doubled by a Domari plural ending: Arabic mislim-īn ‘Muslim-s’, Domari mislim-īn-e. Finally, the overuse of the Domari demonstratives aha/ihi/ehe with nouns, and a slight erosion of their deictic focusing quality, as in ‘and this man went into this house, to fetch this jar of water’, resembles the tendency in Arabic discourse toward generalization of the reduced demonstrative hā- (< hāda/ hādi/hadōl), which tends to accompany the Arabic definite article in similar contexts (Domari has no definite article).
5. Verbal structures Domari differs from Arabic in its structure of past tenses, but the two languages share a distinction between the present indicative and subjunctive (Domari lah-ami ‘I see’, lah-am ‘that I see’; Arabic b-ā-šūf ‘I see’, ā-šūf ‘that I see’), as well as the absence of an explicitly structured future tense. For aspectual distinctions, Domari relies directly on Arabic-derived auxiliaries expressing habitual, inceptive, and iterative aspect, which retain their Arabic tense and person inflection. Unlike in Arabic, however, these auxiliaries are followed by an indicative, not subjunctive, form of the main lexical verb: (10) kunt aw-ami. was.1sg come-1sg ‘I used to come.’ (11) ṣārat mangišk-ari. began.3sg.f beg-3sg ‘She began to beg.’ (12) baqēt kamk-ami. continued.1sg work-1sg ‘I continued to work.’
156
Yaron Matras
The conditional is formed with the Arabic auxiliary kān with the Domari anterior past: (13) a. Domari law ēr-om xužoti kān laher-d-om-s-a. if come.past-1sg yesterday cond see-past-1sg-3sg-ant b. Arabic law až-īt mbāreħ kān šuft-ō. if came-1sg yesterday cond saw.1sg-3sg ‘If I had come yesterday I would have seen him.’ Modal expressions, with the exception of sak- ‘to be able to’, are all borrowed from Arabic. The modal expression for obligation and necessity lāzim, and that for possibility, mumkin, are impersonal. The expression for desire/ intention, bidd-, retains its Arabic person inflection, and is followed by the Domari subjunctive. Arabic loan-verbs are integrated into Domari by incorporating the isolated stem of the subjunctive form – e.g. -štrī- from a-štrī ‘that I buy’, -īš- from a-īš ‘that I live’ – into the inflected loan-verb integration markers -k- from -kar- ‘to do’, for transitives, and -h(r)- ‘to (have) become’, for intransitives: štrī-k-ami ‘I buy’, īš-hr-omi ‘I live’. Domari and Arabic both lack a verb ‘to have’. Although contact influence will not have been the source of the absence of ‘to have’, the specific Domari possessive expression wāšī-m ‘with-me, at-mine’ for ‘I have’ (rather than a construction of the type ‘to-me there-is’, as in other Indo-Aryan languages) does resemble Arabic ind-ī ‘at-mine’. There are also similarities in the organization of existential predications. While Arabic has nominal predications in the present tense and lacks a present-tense copula, Domari, in contrast with its overall SVO structure, retains an enclitic copula, which, in the past tense, is modified by the (non-clitic, non-final) Arabic-derived copula kān: (14) a. Domari ama mišta-hr-omi. I ill-cop-1sg b. Arabic ʔana ayyān. I ill ‘I am ill’
Domari
157
(15) a. Domari ama kunt mišta-hr-om-a. I was.1sg ill-cop-1sg-ant b. Arabic ʔana kunt ayyān. I was.1sg ill ‘I was ill.’ Like Arabic (but also Kurdish), Domari lacks infinitives in modal constructions, and employs a present-subjunctive form of the embedded verb instead: (16) bidd-ī dža-m kury-a-ta want-1sg go-1sg.subj house-obl-dat ‘I want to go home’
6. Other parts of speech Domari shows a massive amount of Arabic loans in its grammatical vocabulary. All numerals above ‘5’, with the exception of ‘10’ and ‘100’, are derived from Arabic (e.g. sitte zirt-ēni ‘six children-pred’), and some speakers also use Arabic numerals for ‘3, 4, 5’. The quantifiers akam ‘few’ and kull ‘all’ (alongside Kurdish-derived giš) are from Arabic, as are most indefinite pronouns, with the exception of ek-ak lit. ‘one’ for ‘somebody, anybody’. Most Arabic indefinite pronouns show only a shallow level of grammaticalization, and are similar in structure to the nouns from which they derive (e.g. ħāža ‘thing, anything’, wāħad ‘one, anyone’, maħall ‘place, somewhere’). They are integrated into Domari by adding the Domari indefinite article -ak: maħal-ak ‘somewhere, anywhere, nowhere’, ħāža-k ‘something, anything, nothing’. Other indefinites, such as dāʔiman ‘always’, are borrowed directly. Expressions for the days of the week, dates, and usually also seasons are Arabic-derived, with times of the day showing a mixture of Turkish (ṣabaħtan ‘morning’), Arabic (zuhur ‘midday’), and Indic (arāti ‘night’). Domari borrows the Arabic interrogatives ʔayy ‘which?’, qaddēš ‘how much?’, and waqtēš ‘when?’ (though not from Jerusalem Arabic; they appear to have been adopted prior to settlement in the city, from Beduin or rural dialects). Further pronouns borrowed from Arabic include the reciprocal baḍ (laherde baḍ ‘they saw one another’), and the third-person resumptive pronoun iyyā-, which retains Arabic-derived gender and number inflection
158
Yaron Matras
(mana illi torim iyyā-h ‘the bread that you gave me [it]’; see below, on relative clauses). All connectors (coordinating and subordinating conjunctions), discourse markers, interjections, particles, fillers, and tags in Domari are Arabic, and assume the same position in the utterance as they do in Arabic. They include ū ‘and’, fa ‘and so’, bass/lākin ‘but’, wila ‘or’, wala ‘nor’, iza / law ‘if’, lamma ‘when’, qabel mā ‘before’, bad mā ‘after’, badēn ‘and then’, yanī ‘that is’, the phasal adverbs (e.g. lissa ‘still, yet, no longer’) and focus particles (e.g. bass ‘only’, kamān ‘too’), the particles ʔa ‘yes’ and la ‘no’, and more. The factual complementizer inn- carries Arabic inflection and, like in Arabic, may either be impersonal, with a default third person masculine singular marker, or agree with the subject of the complement clause: (17) ama sin-d-om inn-o/inn-ak atu īšhr-ori hinēn. I hear-past-1sg comp-3sg.m/comp-2sg you live.2sg here ‘I heard that you live here.’ While place deictics remain Indic, many temporal deictic expressions are borrowed from Arabic, including hallaʔ ‘now’ and badēn ‘later’. In the adjective, comparative and superlative forms are borrowed wholesale from Arabic, including their lexical forms, rendering all adjectives in the language (except those whose positive forms are also Arabic-derived) suppletive: (18) a. tilla–akbar–akbar wāhed ‘big bigger biggest’ b. qišṭoṭa–azɣar–azɣar wāhed ‘small smaller smallest’ This can be explained by the motivation to borrow a comparative/superlative procedure, but the inability to segment the Arabic comparative/superlative into analysable morphemes (cf. Arabic kbīr–akbar ‘big–bigger’).
7. Constituent order Constituent-order rules in Domari are on the whole fully compatible with Arabic. Both languages have flexible word order, with a tendency toward SV(O) in isolated, categorical sentences, and with other patterns as options.
Domari
159
Since most Indo-Aryan and Iranian languages are SOV, this can be considered a clear case of convergence with Arabic. The only exception appears to be in present-tense copula clauses, discussed above (examples 1415). The more frequent Domari possessive construction shows the order possessed– possessor noun (example 9b), matching that of Arabic. The original position of the Domari adjective was in front of the noun (see Macalister 1914). Adjective-noun constructions are still encountered (20a), but they are greatly outnumbered in all contexts by an alternative construction in which the noun is followed by the adjective, to which a non-verbal predication marker is attached (19b): (19) a. ēr-ī qišṭoṭ-ī šōnī. came-3sg.f little-f girl ‘A little girl came.’ b. ēr-ī šōnī qišṭoṭ-ik. came-3sg.f girl little-pred.f ‘A little girl came [A girl came, being little].’ The order of (19b) – originally a marked construction – matches the noun– adjective order found in Arabic. The position of local relation expressions is mixed. While adverbial constructions based largely on inherited material may have the modifier in a position following the noun (example 21), Arabic-derived prepositions occupy the same pre-nominal position as in Arabic (21): (20) kury-is-ma bara house-3sg.obl-loc outside ‘outside the house’ (21) min kury-a-ki from house-obl-abl ‘from the house’
8. Syntax In both languages, non-verbal predications allow subject and predicate to appear in adjacent positions, which might be interpreted as ongoing convergence on the part of Domari with the structure of Arabic nominal clauses:
160
Yaron Matras
(22) a. Domari ama mišta-hr-omi. I ill-cop-1sg b. Arabic ʔana ayyān. I ill ‘I am ill’ (23) a. Domari wuda bizzot-ēk. old.m poor-pred.m b. Arabic l-xityār miskīn the-old.m poor.m ‘The old man is poor.’ Another sign of convergence in non-verbal predications is the fact that, while Domari retains Indo-Aryan negation particles (na/n) elsewhere, in non-verbal predications is adopts Arabic mišš: (24) pandžī mišš bizzot-ēk. 3sg neg poor-pred.m ‘He is not poor.’ The Arabic verbal negator mā is used for the negation of modal and auxiliary verbs borrowed from Arabic, as well as in the vicinity of Arabic-derived prepositions and particles: (25) warik-ar-a mlāy-ēk minšan mā džan-ad-is. wear-3sg-ant veil-pred.m so.that neg know-3pl.subj-3sg ‘She used to wear a veil so that one would not recognise her.’ Clause combining rules in the two languages are by and large identical, Domari drawing entirely on the pool of Arabic connectors and subordinating conjunctions: (26) qabel mā dža-m xałłaṣk-ed-om kam-as. before comp go-1sg.subj finish-past-1sg work-obl ‘Before I left I finished my work.’
Domari
161
(27) iza wars-ari, n-aw-am-eʔ. if rain-3sg neg-come-1sg-neg ‘If it rains, I shall not come.’ (28) na kil-d-om bara liʔann-hā wars-ari. neg go.out-past-1sg out because-3sg.f rain-3sg ‘I did not go out because it is raining.’ (29) ū daʔiman/ yaʕnī/ kunt ama kury-a-m-ēk wala. and always that.is was.1sg I house-obl-loc-pred.f and.not kil-šami wala aw-ami. exit-1sg and.not come-1sg ‘And I was always/ I mean/ at home, not going out nor coming.’ (30) kānū lamma qayišk-ad-a kunt wēšt-am-a wāšī-san. was.3pl when cook-3pl-ant was.1.sg sit-1sg-ant with-3pl ‘When they were cooking I used to sit with them.’ As in Arabic, there is no infinitive in Domari, and the verbs of modal clauses normally appear in the subjunctive (see example 16). Finally, relative clauses in Domari take over the Arabic structure, including both the uninflected relativizer illi, and the presence of an Arabic-derived resumptive pronoun with third person head nouns, which agrees with the head noun in gender and number, retaining Arabic inflection: (31) mana illi to-r-im iyyā-h bread rel gave-2sg-1sg res-3sg ‘the bread you gave me’ (32) ple illi to-r-im iyyā-hum money rel gave-2sg-1sg res-3pl ‘the money(pl) you gave me’
9. Lexicon All semantic domains of the Domari lexicon adopt Arabic loans. The only limitations on lexical borrowing are in the domain of grammatical vocabulary, more specifically referential and deictic pronouns, and place deictics,
162
Yaron Matras
both of which categories seem uninfluenced by borrowings. Alongside Arabic loans, Domari has also retained some Kurdish lexicon (zara ‘boy’, Kurdish zaro), as well as Turkish lexical loans (qapi ‘door’, Turkish kapı).
10. Conclusion It would be useful at this stage to remind the reader that under “borrowings” and “loans” we understand those (Arabic-derived) forms in Domari, for which the language has no inherited alternative; they are thus distinguished from ad hoc switches or mixing patterns. This said, the extent of Arabic borrowing into Domari can be described as nothing less but massive. It is indeed easier to point out those domains in grammar in which borrowing is not found; even for those, exceptions or some hedging of another kind can usually be found: There is no borrowing of case inflection (but Arabic has no synthetic case markers), of synthetic tense marking (though Arabic-derived modality and aspect auxiliaries retain their Arabic tense inflection), of person marking on verbs, prepositions, or nouns (though Arabic-derived modality and aspect auxiliaries retain their Arabic person inflection, and Arabic person agreement markers may appear with Arabic-derived complementizers such as inn- ‘that’ or liʔann- ‘because’, as well as on the resumptive pronoun iyyā-). There is also no borrowing of definite articles (which exist in Arabic but not in Domari, but may occasionally accompany Arabic nouns in Domari discourse), of personal pronouns, or of demonstratives. These domains thus appear as “resistant to borrowing” – at least in the history of Domari so far; but given the extent of grammatical borrowing in the language, we may have a tentative indication of those domains of grammar which the forces of contact-induced change may find more difficult to infiltrate. Already the presence of (at least some) Arabic-derived items in the Domari set of lower numerals, in verbal negation, and in existential constructions, put Domari on the extreme side of the continuum for grammatical borrowing. A remarkable feature of Domari–Arabic contact is the reliance on the borrowing of actual linguistic matter, or MAT-borrowing. While in some domains this is the obvious choice, it is not at all self-evident that Domari should use Arabic-derived prepositions, inflected aspectual auxiliaries, or even subordinating conjunctions. The absence of language-internal grammaticalization processes to replicate the Arabic model (pattern replication or PAT-borrowing) in these domains indicates considerable flexibility within the speech community; it appears to allow itself to shift and re-define the
Domari
163
demarcation boundaries between the two separate sets of forms, rules and constructions – the “internal code” Domari, and the “external code” Arabic – which together constitute the speakers’ linguistic repertoire, and to maintain a boundary that is almost symbolic, drawing only on a limited amount of everyday vocabulary items, deictic and anaphoric reference tools, and the structuring of tense and of person agreement, as the almost exclusive components of the linguistic instrument used to flag and negotiate in-group identity. The other linguistic-mental processing operations, most notably those associated with discourse and utterance organization and clause combining, rely entirely on Arabic structures; for these operations, the two codes are inseparable, having undergone “fusion” (cf. Matras 1998). In this respect, the absence of PAT-borrowing in a series of grammatical domains might be interpreted as a kind of “weak resistance” against the collapse of cross-linguistic demarcation boundaries, or perhaps as “full acceptance” of fusion. One outstanding domain that relies on pattern replication is the formation of non-verbal predications. The presence of nominal sentences in Arabic, but not in Domari, is a major typological difference between the languages. Here, Domari accommodates by replicating at least one principal feature of the Arabic nominal sentence, namely the placement of Subject and Predicate in adjacent positions, not separated by a verb. The verbal element in Domari then follows the predicate; somewhat ironically, this is also the only construction type in which Domari resists full accommodation to Arabic word-order rules, maintaining a verbal copula in enclitic position. As discussed above, in the past tense this difference too is minimized, once again by resorting to MAT-borrowing of Arabic copula forms.
Abbreviations abl ant ben comp cond cop dat det f indef loc
ablative anterior benefactive complementizer conditional copula dative (possessive) determiner feminine indefinite article locative
m neg nom obl past pl pred res sg subj
masculine negation nominative oblique simple past (perfective) plural (non-verbal) predication marker resumptive pronoun singular subjunctive
164
Yaron Matras
References Macalister, R. A. S. 1914 The Language of the Nawar of Zutt, the Nomad Smiths of Palestine. (Gypsy Lore Society Monographs 3). London: Edinburgh University Press. Matras, Yaron 1998 Utterance modifiers and universals of grammatical borrowing. Linguistics 36: 281331. 1999 The state of present-day Domari in Jerusalem. Mediterranean Language Review 11: 158. 2002 Romani: A Linguistic Introduction. Cambridge: University Press.
Grammatical borrowing in Kurdish (Northern Group) Geoffrey Haig
1. Background Kurdish is the cover term for a bundle of closely related west Iranian languages, spoken across a large area of the Middle East centering at the intersection of the Turkish, Iranian and Iraqi national borders. The number of speakers is variously estimated at between 20 and 40 million. Traditionally, three major dialect clusters are identified: The Northern Group, often referred to as Kurmanji (also spelt Kurmanjî, Kurmanci, Kurmancî); the Central Group, often referred to as Soranî; and the Southern Group. In terms of numbers of speakers, the Northern Group is the largest, encompassing all the Kurds of Turkey1 and Syria, plus the northernmost Kurds of Iraq (Zakho, Dohuk), Kurds of west Iran around Lake Urmia, plus outliers in Azerbaijan, Armenia, and Georgia. The Central Group includes most of the Kurds of Iraq around the cities of Suleimania, Kirkûk, and Erbil, plus speakers in Iran around the cities of Sanandaj and Kermanshah. While the distinction between Northern and Central Group Kurdish is not controversial, the exact demarcation of the Southern Group remains hotly disputed, but I will not enter these issues here. This chapter is concerned solely with the Northern Group of Kurdish. Speakers of the Northern Group have maintained long-standing relations with speakers of many languages. Alongside the national languages such as Arabic, Armenian, Azerbaijani, Georgian, Persian, Turkish and Russian, there has been contact with numerous minority languages, for example varieties of Eastern Neoaramaic, some indigenous languages of the Caucasus, Turkoman, varieties of Romani (see for example the Chapter on Domari), to name but a few. Obviously it is not possible to cover the full range of contact situations and outcomes in the space of this chapter. Instead I will be focussing on the Kurdish of Central Anatolia, and restricting the analysis to the impact of the (now) major contact language, Turkish. The areas considered are Muş, Erzurum and Tunceli, where contact with Turkish has traditionally been fairly strong, and where the number of other languages involved is somewhat less than in many parts of the the Kurdish speech zone. Of the
166
Geoffrey Haig
three different local varieties considered, the Erzurum and Muş ones appear close enough to be identified by their respective speakers as “my dialect”, but the Tunceli variety shows some distinct features which are, to my knowledge, not found elsewhere. In the interests of brevity, I will refer to these varieties collectively as Central Anatolian Kurdish (CAK), although the term is not an established one in Kurdology. In assessing the impact of Turkish, I have taken the Kurdish of Zakho and Dohuk (North Iraq), often referred to as Bahdinî (in assorted spellings), as a benchmark for a less Turkish-influenced brand of Kurdish, against which Kurdish influence on Central Anatolian Kurdish can be gauged. In addition to my own data, I draw on the results of other publications on Kurdish–Turkish language contact (Dorleijn 1996, Bulut 2002, 2005, 2006, Matras 2002, Haig 2001, 2006, forthc.). Two final caveats need to be mentioned before we proceed to the data. First, Armenian must have been a considerable influence on CAK up to the beginning of the twentieth century, but it is unfortunately not possible to address the issue of Armenian influence here. Second, comparisons with Turkish are generally drawn on the basis of colloquial standard Turkish. But in fact, local varieties of Turkish from the area differ in many respects from the standard. However, these dialects represent “a rather neglected spot on the Turkological map.” (Bulut 2002: 51), so one often has little choice but to fall back on the convenient fiction of the colloquial standard. Nevertheless, for the local variety of Turkish from Erzurum at least, a reasonably reliable source is available, Gemalmaz (1995), which I refer to at some points. The sociolinguistics of Kurdish in Turkey is extremely complex, variegated, and poorly described. Prior to the founding of the Turkish Republic in 1923, relations between the two speech communities were not marked by any great prestige assymetry. In fact, in the partly autonomous regions of Anatolia, Kurdish enjoyed considerable prestige as the language of many powerful landowners and religious leaders, and was learned as a second language and used as a lingua franca by speakers of many other speech communities. However, as a result of the nationalist currents accompanying the founding of the Turkish Republic, the status of Kurdish deteriorated rapidly, and the language has been officially non-existent for much of the Republic’s history (see Haig 2003, 2004). The advent of compulsory schooling, military service, and the intrusion of mass-media to the most isolated parts of Kurdistan have led to large-scale language shift, and a drastic reduction in the number of children acquiring Kurdish fully as an L1. The recent changes in the wake of EU-fueled reforms, while of considerable symbolic importance, have done little to reverse these trends. Unfortunately, there is as yet no serious empir-
Kurdish (Northern Group)
167
ical research on these ongoing developments so I am obliged to draw on the personal observations of speakers I have worked with over the past years in assessing the situation. The speakers who provided the narratives from which most of this data has been taken are all Kurdish native speakers, between 50 and 75 years old (two males, two females). All except one (a woman in her seventies, from Erzurum) are also fluent speakers of Turkish.
2. Phonology The segmental phonology of Turkish consists to a large extent of crosslinguistically unmarked elements, most of which are present in the phoneme inventories of the neighbouring languages anyway. It is thus difficult to pinpoint phonological influence of Turkish on Kurdish. The best candidate for contact influence in the vowels is the partially systematic use of fronted rounded vowels /y/ and /ø/ in some varieties of Kurdish (see Bulut 2005: 225226 for further discussion). There is little evidence for the transfer of Turkish vowel harmony into Kurdish. Syllable structure in Kurmanji is somewhat less constrained than in Turkish, but vowel epenthesis (rather than consonant deletion) is the usual strategy for breaking up consonant clusters, as it is in Turkish. As in Turkish, hiatus is generally avoided. Kurdish phonology does show features that bespeak of contact influence, but not very obviously from Turkish. For example, Kurdish in Central Anatolia exhibits a three-way distinction among the stops – between voiced, voiceless aspirated, and voiceless non-aspirated – giving rise to a Caucasus-style three-way stop distinction. There is some disagreement whether the relevant phonetic parameter is ejective vs. non-ejective (according to some Soviet authors) or voice–onset time (MacKenzie 1961; Kahn 1976), or both (Jastrow 1977). There is also disagreement on its origin; I consider contact influence from Armenian the most likely source. A second notable characteristic of Central Anatolian Kurdish, at least of Muş and Erzurum, is the widespread presence of pharyngealized segments, not only as expected in borrowings from Arabic (e.g. dæʢwat ‘wedding celebration’), but also in words that historically do not have pharyngeal segments, for example mʢær ‘snake’, tʢæl ‘bitter’, mʢæhinə ‘mare’. In a sense, the pharyngeals are extraneous to the basic phonology: they are restricted to individual lexical items, they play no part in morphology, their functional load is very limited, and there is considerable cross-speaker and cross-dialect variability in the extent of their presence.
168
Geoffrey Haig
3. Morphological typology Central Anatolian Kurdish is both prefixing and suffixing, with a small measure of Ablaut, restricted to the past stem formation of a closed set of irregular verbs. Turkish is exclusively suffixing, with practically no stem-vowel alternations. Morphological alignment in Kurdish is split ergative (ergative with past tenses of transitive verbs), thus contrasting with Turkish, which is accusative throughout. There is evidence that ergativity in Kurdish is disappearing, but this need not necessarily be linked to contact influence (as suggested by Dorleijn 1996), as similar tendencies can be observed throughout Iranian (Haig 2007). A likely instance of contact influence is the complete loss of clitic pronouns in the Northern Group. Such mobile pronominal clitics are a highly salient feature of most West Iranian languages, including Persian. They can be traced back to the clitic genitive/dative pronouns of Old Iranian, and are thus a deeply rooted genetic trait of Iranian. Crucially, such clitics are mobile (special clitics in the sense of Zwicky 1977). For example, in Central Kurdish, personal pronouns in oblique functions frequently cliticize and move to other constituents: (1)
nân=mân lagal bi-xô! food=1pl.clitic with imp-eat:pres:2s ‘Eat a meal with us!’
The first-person plural clitic pronoun mân is syntactically a complement of the preposition lagal, but it is phonologically hosted by the noun nân. As mentioned, mobile pronominal clitics are one of the most salient features of West Iranian. But they are completely lacking in the Northern Group of Kurdish, as they are in Zazaki, likewise spoken in Central Anatolia. The presence or absence of such mobile pronominal clitics appears to follow an areal distribution, having crossed, for example, genetic boundaries in the case of varieties of Turkic spoken in Iraq under heavy contact pressure from Iranian and Semitic languages with pronominal clitics (Bulut 2005: 227228). It seems that there is an areal isogloss in the Iranian languages between languages with mobile pronominal clitics, and those that lack them, with non-clitic languages concentrated in the west (Northern Group Kurdish and Zazaki) and north (languages of the Caspian, Don Stilo, p.c.). It is not unreasonable to link this to Turkic influence; the Turkish of Turkey and Azerbaijan has lacked
Kurdish (Northern Group)
169
such mobile pronominal clitics throughout its attested history, cf. the Anatolian Kurdish and Turkish versions of (1): (2)
bi me va nan bi-xwe! circ 1pl:obl circ food imp-eat:pres:2s
(3)
bizim-le yemek ye! 1pl:gen-com food eat:imp.2s
Otherwise, contact influence on the morphology is difficult to ascertain, and there is no attested borrowing of actual bound morphology (but see the conditional clitic below), unless attached to a Turkish base.
4. Nominal structures The structure of the noun phrase in Kurdish differs from that of Turkish in most respects. Turkish is consistently head-final, and marks case relations through a single phrase-final affix or postposition. Nouns lack gender, and there are no agreement phenomena between elements of the noun phrase. Kurdish on the other hand is largely head-initial, while dependents to the head are linked to it via the Ezāfe-particle, an unstressed vowel, sensitive to the gender and number of the head. There is a binary case distinction between the unmarked Direct and the marked Oblique case in the Northern Group of Kurdish. The Oblique has different realizations, depending on the gender of the noun, the presence or absence of determiners, and is subject to some lexical exceptions (e.g. Ablaut in a small set of masculine nouns). In Central Anatolian Kurdish, these rules appear to have become simplified, so that in most environments the Oblique is expressed through a single suffix [-i:],2 and Ablaut for case marking is completely absent. The Oblique case marker thus ends up superficially resembling the Turkish Accusative suffix, particularly that of the local Turkish dialects, where the Accusative suffix tends towards a unified [-i] realization rather than the standard Turkish realization which varies according to the laws of vowel harmony. Other case relations (Instrumental, Benefactive, Comitative, local relations etc.) are expressed through adpositions. Old Iranian was predominantly prepositional (this is certainly true for Old Persian, pace Harris and Campbell
170
Geoffrey Haig
1995: 140), and the Southern and Central Groups of Kurdish are also clearly overwhelmingly prepositional. However, there are rather interesting signs of a contact-induced shift in the patterning of adpositions. In addition to the basic inventory of prepositions, the Northern Group also has a number of circumpositions, for example: (4)
Benefactive / Indirect Object: ji te ra ‘for you’ Locative: di dinyayê da ‘in the world’
While circumpositions are common to many West Iranian languages, in some varieties of Central Anatolian Kurdish, the prepositional part of the these circumpositions is omitted, leaving the sole marker of case a postposed phrasal clitic, or postposition:3 (5)
Benefactive / Indirect Object: te ra ‘for you’ Locative: dinyayê da ‘in the world’
The Tunceli “postposition” -ra is the functional equivalent of the Turkish dative suffix, while the Tunceli locative -da covers most of the ground that the Turkish locative -da/-de does.4 Note that the expression of locative is thus actually phonetically almost identical in the two languages, though this is a coincidental outcome of the process outlined above, rather than actual borrowing of matter. Thus Central Anatolian Kurdish has actually acquired two postpositions via restructuring of its indigenous circumpositions. Given the lack of such forms in other varieties of the Northern Group, there seems little doubt that we are dealing with Turkish influence. Central Anatolian Kurdish has also actually borrowed two Turkish postpositions, göre and ait, but it does not use them as postpositions: Tu. postposition X-dat göre X-dat ait
Ku. preposition or noun (li) gora X aitê X
‘according to X’ ‘belonging to X’
A dubious candidate for matter borrowing of a postposition is Turkish sonra, an adverb meaning ‘after, later’, but also used as a postposition. In Central Anatolian Kurdish a word şûnda is used in an almost identical manner, for example as a postposition, where the NP it governs requires the Kurdish preposition ji ‘from’:
Kurdish (Northern Group)
171
(6) ji wê roj-ê şûnda from that:obl day-obl after ‘from that day on’, ‘after that day’ This calques the Turkish expression (same meaning): (7) o gün-den sonra that day-abl after However, it is by no means certain that şûnda really is borrowed from Turkish (claimed for example by Chyet 2003); the phonological differences are difficult to account for, and some authors consider it to be related to the native noun şûn ‘place’. This may therefore be a case of a conspiracy of phonological similarity and functional parallels leading to identity of usage across the two languages, rather than straightforward borrowing. Possessive constructions in Central Anatolian Kurdish are Ezāfe constructions, with the possessor following the possessed, as in: (8) heval-ek-î min friend-indef-iz 1s:obl ‘a friend of mine’ This contrasts with the head-final construction in Turkish (same meaning): (9) ben-im bir arkadaş-ım 1s-gen a friend-poss1s No variety of the Northern Group known to me shows clear evidence for contact-induced change in the possessive NP. For predicative expressions of possession, however, there are notable similarities. Both languages lack a lexical ‘have’ verb. For predicative expressions of possession, a possessive construction such as those just shown is combined with an existential predicate: (10) heval-ek-î min he-ye. friend-indef-iz 1s:obl existent-cop(3s) (11) ben-im bir arkadaş-ım var. 1s-gen a friend-poss1s existent-cop(3s) ‘I have a friend.’
172
Geoffrey Haig
While this could be mere chance similarity, it is notable that the Zakho and Dohuk dialects of the Northern Group, with much less exposure to Turkish influence, regularly use a different possessive construction, using a fronted oblique Possessor (own fieldwork, see also MacKenzie (1962: 320) for further examples): (12) te kalem he-ye? 2s:obl pen existent-cop(3s) ‘Do you have a pen?’ This type of construction, historically certainly older than (10), is now entirely lacking in Central Anatolian Kurdish, and was never present in Turkish.
4.1. Comparative forms of adjectives Historically Kurdish has an adjectival suffix for the comparative degree, as in dirêj-tir ‘long-er’ and mezin-tir/meztir ‘larg-er’, etc. The Central Group also has a suffix for the superlative degree, but this is absent in Anatolian Kurdish. Instead, Superlative is expressed periphrastically, for example through a partitive construction (of-them the larger) etc. In Turkish, comparative is expressed through the particle daha, and superlative is expressed through the particle en, both combining with the basic form of the adjective: (13) uzun, daha uzun, en uzun
‘long, longer, longest’, etc.
In Central Anatolian Kurdish, the Turkish comparative and the superlative particles are frequently borrowed, and in my data, the native comparative suffix can then be omitted (though according to Dorleijn (1996) and Bulut, (In print), there is variation in this regard): (14) daha mezin, en mezin, etc.
‘larger, the largest’
In both languages, the standard of comparison is expressed through an ablative marker (preposition ji in Kurdish, case marker in Turkish). However, the comparative suffix is usually retained when the standard of comparison is explicitly mentioned, see Dorleijn (1996: 53). Thus at least for the generation of speakers I have gathered data from, the original comparative suffix is still available as an option, certainly for adjectives of high frequency.
Kurdish (Northern Group)
173
5. Verbal structures On the face of it, the verb system of Kurdish and that of Turkish are very different. Turkish verbs display a rich range of suffixes and enclitics distinguishing voice, tense, evidentiality, negation, interrogative, mood, and aspect, while person marking constitutes the final layer of suffixation. Kurdish has a system of two stems, a present and a past stem, which combine with prefixes and suffixes distinguishing two moods, and with the past stem, an aspect distinction. Future tense in Kurdish is expressed through a clausal clitic plus a subjunctive form of the verb. Passive and causative are expressed via auxiliary verbs only. But the most striking distinction between the two verb systems is that while Turkish possesses a rich inventory of productive non-finite forms (infinitives, participles, converbs, verbal nouns), Kurdish has practically no non-finite verb forms. The sole exceptions are the so-called infinitive (actually just a conventionalized citation form, with low frequency and productivity in syntax), and a secondary participle (see below). There are nevertheless traces of contact influence from Turkish on the Central Anatolian verb system. The first is the use of a tense form based on a participle in [i:] plus the clitic copula person endings, to express evidentiality. There is some variation in the extent that this is systematically applied, but certainly for the Tunceli dialect which I investigated, forms such as hat-i-ye ‘come-part-cop(3s)’ are clearly intended to express an evidential, an unwitnessed past, corresponding to the Turkish form gel-miş ‘s/he came (but I did not see it)’. Thus what is borrowed is a semantic category, expressed using the forms felt by speakers to come closest to the relevant expression units in the donor language. A straightforward case of matter borrowing in the grammar of verbs is the Turkish clause-final clitic conditional marker with the forms =se/=sa (only the former form is borrowed), used to mark the protasis of a conditional clause. An example from Tunceli showing the use of this form is the following (see Bulut 2006 for further examples and discussion): (15) eer bapir-ê min ew-na ne-girt-ine cem if grandfather-iz:m 1s:obl dem-pl neg-take:pst-pl to xa=sa ew-na di-mir-in. refl=cond dem-pl prog-die:pst-pl ‘If my grandfather had not taken them in they would have died.’
174
Geoffrey Haig
5.1. Borrowing of verbs Kurdish shares with most Iranian languages a lack of productive morphological derivation of new verbs. Thus there is no morphology that would be functionally comparable to, say, English -ize or -ify (as in priorit-ize and beaut-ify etc.), so other means are necessary to create new verbs.5 In Kurdish, as in most Iranian and Indo-Aryan language, new verbs are created using the light verb strategy: a non-verbal element (noun, adjective, adverb, or even a phrase of some sort) is combined with the light verbs kirin ‘do’, or bûn ‘be, become’ (a couple of other lexical verbs also operate as light verbs, but are ignored here). The resulting expressions are tightly bound, compound-like units that behave in some, though not all respects, like a simple verb form (see Haig 2002 for details). This strategy is ideally suited for incorporating borrowed items, be they nouns, verbs or anything else in the donor language. For example, Arabic verbal nouns are incorporated into Kurdish in this fashion (e.g. fehm kirin ‘understanding do’=‘understand’, from an Arabic masdar, or qayîl bûn ‘saying be’=‘to be in agreement, consent’, from an Arabic participle). A striking regularity in Kurdish is that many of the items borrowed into such light verb constructions are Turkish -mIş verb forms. In Turkish, the -mIş suffix forms both participles with perfective meaning, and finite verb forms with evidential meaning. In the latter case, the -mIş suffix is followed by person agreement. It seems likely that the participles are the point of departure for the borrowed forms, which never have any additional morphology. The borrowing of Turkic -mIş verb forms into Iranian languages is a wellestablished phenomenon, found also for example in Tadjik (through contact with Uzbek). In Kurdish, such forms are found throughout Anatolia, and are very common in the texts collected by Le Coq and Lerch from the nineteenth century – see Bulut (2006) for extensive discussion. However, they are considerably less widespread in the Kurdish of Zakho and Dohuk, corresponding to the lesser degree of Turkish influence in these areas. In Turkish itself the vowel in the -mIş suffix has four possible realizations, according to the rules of vowel harmony. In Kurdish, however, it is generally always realized with the same short vowel, somewhere between [i] and [ö] (simply transcribed with /i/ in the examples). The suffix has not to my knowledge been extended to non-Turkish stems. Examples of borrowed verbs attested in my Central Anatolian Kurdish texts are the following (see Bulut (2006) and Dorleijn (1996: 4952) for further discussion and examples):
Kurdish (Northern Group)
175
(16) annamiş kirin ‘understand’; düşünmiş kirin ‘think’; nişanlanmiş bûn ‘get engaged’; başlamiş kirin ‘begin’; sevinmiş bûn ‘be happy’; tanışmiş bûn ‘know, get to know’; sokmiş bûn ‘be inserted’; dayanmiş bûn ‘endure’; dinlemiş kirin ‘listen’ It needs to be stressed that the -mIş verb forms used in Kurdish do not have the same semantics as the Turkish verb forms on which they are apparently based: they have lost their perfective participial sense, and certainly do not have any sense of evidentiality. In effect, they are a tense-neutral kind of action nominal. While some are simply centuries-old, established borrowings, acquired as part of the Kurdish lexicon during L1 acquisition, it appears that bilingual speakers command a rule which permits them to access the Turkish repertoire for additional verbs, yet to deploy them in Kurdish with the appropriate change in meaning. In addition to the well-known borrowing of -mIş verb forms, Central Anatolian Kurdish has also borrowed Turkish bare verb stems. The ones attested in my data are: (17) karış ‘mixing up’ (used as noun); say kirin ‘count’; şaş kirin ‘make an error’, ‘be confused’;6 inan kirin ‘believe’; bekle kirin ‘wait’; kapat kirin ‘close’ Another area of matter borrowing in the verbal system are impersonal expressions for obligation and necessity. For example gerek (in Kurdish reduced to gere) and lazim ‘necessary’, the latter originally from Arabic, are commonly borrowed. In standard Turkish at least, such expressions are combined with nominalized clauses (lit. my-going is-necessary). Kurdish of course lacks such nominalizations (compare the lack of non-finite verb forms discussed above), so the syntax of such impersonal particles looks different: they are put clauseinitially and followed by a finite clause in the subjunctive (lit. is-necessary I go). Interestingly, this syntactic pattern is in fact a feature shared throughout the area, and is also found in local varieties of Turkish (Bulut 2005: 229).
6. Other parts of speech As discussed in the preceding sections, words of all major classes are borrowed (nouns, verbs, adjectives), numerals, adverbs, connectors, discourse
176
Geoffrey Haig
particles. No borrowing is attested of personal pronouns, of demonstratives, of interrogatives, or of negation particles. Numerals up to 20 are generally Kurdish, but there is great variation among different speakers, and even with individual speakers, the choice of language for numerals will vary according to speech situation and context. Dates will very often be given in Turkish. Pronouns (personal, possessive and indefinite) and demonstratives are generally Kurdish throughout, though some are shared anyway (e.g. her ‘each, every’ in both languages). Turkish uses a subordinator ki, originally of Iranian origin, while Kurdish uses an etymologically related subordinator ke, ku, and sometimes ki (though in a somewhat different set of contexts to the Turkish one), so that in practice it may be difficult to distinguish borrowing from native use. Postpositions were briefly discussed above; Turkish has no prepositions which could be borrowed, but at least two Turkish relational nouns, orta ‘middle’ and yan ‘side’, are borrowed and in some dialects of Kurdish used as prepositions meaning ‘between’ and ‘beside’ respectively. A further source of borrowed elements concerns the vocabulary of time (much of it originally of Arabic provenience): Turkish zaman, vakit ‘time’, ara ‘interval, space’ (Tu. o arada ‘at that period’, calqued in Central Anatolian Kurdish as wê arê da), sonra (as an adverb) ‘after, later’. The largest group of borrowings concerns discourse markers and conjunctions of various types. The following Turkish words (some of Arabic origin) are used in spoken Anatolian Kurdish to varying degrees (given in standard Turkish orthography): (18) ama ‘but’, fakat ‘but (more stongly contrastive), artık ‘still, yet, already’, yani ‘I mean, you know’, çünkü ‘because’, demek ‘that means’ (Kurdified to demêga), sırf ‘only’, yalnız ‘just, only’ (usually with metathesis of the liquid and nasal) However, Kurdish retains a distinct set of such words which most speakers are able to use: lê ‘but’, êdî (same meaning as Tu. artık), bes ‘but, well, enough’, loma ‘because’ (borrowed from Arabic). They are therefore not replaced wholesale by Turkish elements, but Turkish elements are used alongside them, with the frequency and distribution of the Turkish elements varying heavily from speaker to speaker, and speech situation to speech situation. The heaviest concentration I have seen is in the Tunceli data, where the speaker’s narratives are regularly interspersed with Turkish discourse markers, adverbs, and discourse-regulating phrases such as:
Kurdish (Northern Group)
177
(19) ondan sonra ‘after that’; ister istemez ‘whether one likes it or not’; en sonunda ‘finally’; tekrar ‘again’; yoksa ‘or else’; işte (general filler) Connectors and subordinating conjunctions are only very sparingly used in Turkish, and many of those that are used are borrowings from other languages. In Central Anatolian Kurdish, virturally no connectors are used, simple juxtaposition being the favoured means of clause combining – see Section 8 for further discussion.
7. Constituent order Both Turkish and Kurdish share a (largely) verb-final constituent order in the clause, but elsewhere, constituent order diverges. For the NP, see Section 4, in subordination, it is generally the main clause that precedes the subordinate clause in Kurdish, while the opposite is found in (Standard) Turkish. However, in the spoken vernacular both languages prefer asyndetic sequences, so the differences are less noticeable. Generally, there is little violation of the Kurdish constituent-order rules, except when entire Turkish phrases are inserted, which can mostly be interpreted as code-switching. However, the Tunceli dialect shows a striking change of word order in a minor construction, which appears to be clearly influenced by the corresponding Turkish construction. As it is one of the few fairly clear instances of contact-induced change in constituent order in Kurdish, it is worth looking at a little closer. In the interest of clarity, I will restrict myself to the past tenses, where the changes can be observed most clearly. In Turkish, there is no lexical copula verb for static senses of ‘be’ in main clauses. Rather, we find a clitic tense marker, in the next example realized as =ti, to which a person agreement suffix is added (here zero for the third person singular): (20) Murat mühendis=ti. Murat engineer=pst(3s) ‘Murat was (an) engineer.’ For processual senses (‘become’), however, an inflected form of the lexical verb olmak is required: (21) Murat mühendis ol-du. Murat engineer become-pst(3s) ‘Murat became an engineer.’
178
Geoffrey Haig
In the Northern Group of Kurdish, on the other hand, the same lexical verb (bûn) is used in both static and processual senses. The semantic difference is normally indicated through a change of word order. With the static sense, the verb follows the copula complement: (22) Murat mezin bû. Murat big be:pst(3s) ‘Murat was big/tall.’ In the processual sense, the verb precedes the copula complement. Compare (23) with (22): (23) Murat bû mezin Murat be:pst(3s) big ‘Murat became big/grew up.’ Thus most varieties of the Northern Group mark the semantic distinction between static and processual senses of the copula through a change in word order, whereas Turkish marks the distinction through the opposition full verb vs. clitic. In the Tunceli dialect of Central Anatolian Kurdish, however, we find the semantic opposition marked in the same manner as in Turkish. With static senses of bûn, the initial [b-] of the verb is elided and the verb cliticizes to the copula complement: (24) Gund-ê ma pir rind=û. Village-of us very fine=be:pst(3s) ‘Our village was very beautiful.’ With processual senses of bûn, however, the full form of the verb with initial [b-] is required, but unlike in other varieties of the Northern Group, it follows the copula complement: (25) kili darmadaxin bû ew gund-da. everything haywire be:pst(3s) that village-loc ‘Everything went haywire, in that village.’ Note that two changes must have happened to bring Tunceli into line with Turkish in this respect: a fairly natural phonological change (lenition of [b] to [w]/zero), together with cliticization, which affected the sentence final,
Kurdish (Northern Group)
179
static sense of bû. The originally non-sentence-final, processual bû on the other hand,was not affected by the phonological change, but it did undergo a word-order shift, from after the copula complement to before it.
8. Clause combining and complex clauses Standard Turkish has non-finite verb forms for clause combining (converbs), for complementization (participles, infinitives), and for relative clause formation (participles). Kurdish, on the other hand, has practically no non-finite verb forms, so the languages differ radically as far as the available inventory of forms is concerned. Nevertheless, in actual usage the differences are to some extent levelled out. As Bulut (2002, 2005) points out, Turkish dialects from east Anatolian make more sparing use of converbs than the standard vernacular (see below on coordination).7 Investigation of the spoken Turkish provided in Gemalmaz (1995) show that the kind of nominalizations used in the standard (written) language for complementation, and for relative clauses are extremely scarce, but are nevertheless possible. Nevertheless, the texts of Gemalmaz (1995) show frequent use of non-finite adverbial clauses consisting of a verbal noun plus a semantically bleached noun or adverb, formally corresponding to standard Turkish git-tiğ-im zaman go-ptcpl-poss1s time ‘when I go/went’, or git-tik-ten sonra go-ptcpl-abl after ‘after (someone) went’ (see also Menz 2002). There is nothing corresponding to this type of adverbial clause in Kurdish; rather, a temporal adverb (e.g. gava ‘the time (that)’=‘when’) occurs clause-initially. However, in the domain of clause coordination, Turkish and Kurdish show one striking similarity: coordination of sequential clauses is achieved in both languages through simple unmarked juxtaposition of finite clauses. Both local varieties of Turkish, and the Kurdish of the region, scarcely use an and-type conjunction for clause combining.8 Other varieties of both languages do have such a form (standard Turkish ve, Kurdish û). Thus in the most basic type of loose clause combining, there is a striking unity in the neighbouring languages, but it does not extend greatly into the expressions of more syntacticized inter-clausal relations (adverbial, subordination, relative clauses etc.).
180
Geoffrey Haig
9. Conclusion Despite centuries of coexistence of Turkish and Kurdish speakers in Anatolia, the core grammars of Central Anatolian Kurdish and Turkish have remained quite distinct: constituent order in the NP, inflectional morphology, gender system, alignment in past tenses, means of subordination. As a rough indication of Turkish influence, it is instructive to compare the grammar of CAK with the Bahdinî dialect of Kurdish spoken in North Iraq. The major grammatical differences are the following (some were discussed in the chapter, others were not): a. Loss of non-canonical subjects with predicates of possession, desire, necessity (cf. example (12) above); b. loss of gender distinctions in some environments (not discussed above); c. emergence of a unified plural marker (not discussed above); d. increase in the number and frequency of circumpositions and postpositions; e. increase in text frequency of the use of pre-verbal adpositional arguments (not discussed above). The changes generally involve a loss of constructional variants, or changes in the frequency of constructional variants, rather than the introduction of completely new structures, either through matter or pattern borrowing. Taken individually, they are examples of commonplace language changes that readily occur in the absence of contact pressure. Yet it is a simple fact that all the changes noted result in a grammar that is significantly closer to that of Turkish. It did not have to be that way, and it certainly seems more than coincidence. What we have then is the cumulative effect of small changes, each of which serve to push the entire grammar a little further in a certain direction. This type of gradual, cumulative change may be typical for the type of long-standing coexistence on more or less equal footing that characterized Turkish–Kurdish language contacts up the beginning of the twentieth century. For example, code-switching and early Kurdish–Turkish bilingualism may well have been quite unusual among the rural population in the preRepublican era (indeed, one of my speakers was still a Kurdish monolingual), so we might expect the contact outcome to be quite different to that found in, for example, very small and threatened minority languages surrounded by a dominant language, where bilingualism has been the norm for an extended period. The Tunceli data nevertheless show the results of very intensive bor-
Kurdish (Northern Group)
181
rowings, possibly reflecting the region’s relative proximity to the Turkishspeaking west, but this needs closer monitoring. While the present chapter has been based on data from speakers all over 50, it seems likely that the speech of the younger generation (under 30), most of whom have attended Turkish-speaking schools and have had early exposure to Turkish mass media, will differ significantly in the type and extent of borrowing. However, the current situation is characterized not merely by an increase in Turkish influence, but also by parallel developments which render the analysis considerably more complicated: the emergence of Kurdish mass-media, and the resurgence of Kurdish nationalism which has led to the politicization of language issues, and a backlash of purism against Turkish influence. These factors serve to make the evaluation of Turkish influence on the speech of younger Kurdish speakers a very complex topic for future research.
Notes 1. Note that the Northern Group of Kurdish does not encompass the closely-related Iranian language Zaza(ki), spoken in Central Anatolia and referred to by some authors as a variety of Kurdish. 2. There is considerable cross-speaker variation here. Some speakers still distinguish two forms of the Oblique suffix, phonologically [e:] and [i:], depending on the gender of the noun. 3. The same phenomenon is found in varieties of Kurdish further North in Azerbaijan and the Caucasus generally. 4. The postpositional particles vary as to whether or not they trigger the Obliqe case of their complement NP, but the details are too complex to discuss here – see Bulut (2006: 101104) for discussion. 5. Kurdish must at some stage have had productive denominal verbalizing morphology, as there are verbs in the language obviously derived from borrowed nouns (e.g. zewicîn ‘to marry’ from Arab. zQwdZ), but these processes are now defunct. 6. Widespread in all varieties of the Northern Group. 7. I am very grateful to Christiane Bulut for a number of important observations on the Turkish dialects of Eastern Anatolia (p.c.). 8. In Kurdish, an enclitic form of û / u is sometimes used in combining NPs: dê=u bav ‘mother and father’, and it may occasionally cliticize to a verb form (use of clitic ‘and’ is widespread in colloquial Persian, see Stilo 2004). But not all cases of a postverbal clitic û have an immediately following second clause, so it is difficult to ascertain the function of the clitic in these cases.
182
Geoffrey Haig
References Bulut, Christiane 2002 Evliya Çelebi as a linguist and dialectologist: seventeenth century East Anatolian and Azeri Turkic dialects. In: N. Tezcan and K. Atlansoy (eds.), Evliya Çelebi ve Seyahatname, 4963. Gazimağusa, North Cyprus: Doğu Akdeniz Üniversitesi. 2005 Zum Kopierverhalten türkischer Übergangsdialekte. In: Walter Bisang, T. Bierschenk, D. Kreikenborn, and U. Verhoeven (eds.), Prozesse des Wandels in historischen Spannungsfeldern Nordostafrika/Westasien, 221233. Würzburg: Ergon. 2006 Turkish elements in spoken Kurmanji. In: Hendrik Boeschoten and Lars Johanson (eds.), Turkic Languages in Contact. Proceedings of the Wassenaar Meeting, Feb. 1996, 95121. Wiesbaden: Harrassowitz. Chyet, Michael 2003 Kurdish–English Dictionary. New Haven: Yale University Press. Dorleijn, Margreet 1996 The Decay of Ergativity in Kurmanci. Language Internal Or Contact Induced? Tilburg: Tilburg University Press. Gemalmaz, Efrasiyap 1995 Erzurum ili ağızları, II. Cilt: Inceleme, Metinler ve Dizinler. Ankara: Türk Dil Kurumu. [The dialects of the Erzurum region, Volume 2: Analysis, Texts, Glossaries] Haig, Geoffrey 2001 Linguistic diffusion in modern East Anatolia: From top to bottom. In: Alexandra Aikhenvald and R. M. W. Dixon (eds.), Areal Diffusion and Genetic Inheritance: Problems in Comparative Linguistics, 195224. Oxford: Oxford University Press. 2002 Complex predicates in Kurdish: Argument sharing, incorporation, or what? Sprachtypologie und Universalienforschung/Language Typology and Universals 55 (1): 2548. 2003 Sprachenvielfalt und Sprachenpolitik am Rande Europas: die Minderheitensprachen der Türkei. In: Dieter Metzing (ed.), Sprachen in Europa. Sprachpolitik, Sprachkontakt, Sprachkultur, Sprachentwicklung, Sprachtypologie, 167186. Bielefeld: Aisthesis. 2004 The invisibilisation of Kurdish: The other side of language planning in Turkey. In: Stefan Conermann and Geoffrey Haig (eds.), Die Kurden: Studien zu ihrer Sprache, Kultur und Geschichte, 121150. Hamburg: EB-Verlag. 2006 Turkish influence on Kurmanjî: Evidence from the Tunceli dialect. In: Lars Johanson and Christiane Bulut (eds.), Turkic-Iranian Contact Areas. Historical and Linguistics Aspects, 279295. Wiesbaden: Harrassowitz.
Kurdish (Northern Group) 2007
183
Alignment shift in Iranian languages. A Construction Grammar Approach. Berlin: Mouton. Forthc. Spoken Kurdish from the Muş and Erzurum regions. Wiesbaden: Reichert. Harris, Alice, and Lyle Campbell 1995 Historical Syntax in Cross-Linguistic Perspective. Cambridge: Cambridge University Press. Jastrow, Otto 1977 Zur Phonologie des Kurdischen in der Türkei. Studien zur Indologie und Iranistik 3: 84106. Kahn, Margaret 1976 Borrowing and Regional Variation in a Phonological Description of Kurdish. Ann Arbor, Michigan: Phonetics Laboratory of the University of Michigan. MacKenzie, David 1961 Kurdish dialect studies, Volume I. London: Oxford University Press. 1962 Kurdish dialect studies, Volume II. London: Oxford University Press. Matras, Yaron 1998 Utterance modifiers and universals of grammatical borrowing. Linguistics 36 (2): 281331. 2002 Kurmanjî complementation: Semantic-typological aspects in an areal perspective. Sprachtypologie und Universalienforschung/Language Typology and Universals 55 (1): 4963. Menz, Astrid 2002 The dialects of Erzurum. Some remarks on adverbial clauses. Turkic languages 6: 199214. Stilo, Don 2004 Coordination in three Western Iranian languages. Vafsi, Persian and Gilaki. In: Martin Haspelmath (ed.), Coordinating Constructions, 270330. Amsterdam: Benjamins. Zwicky, Arnold 1977 On Clitics. Bloomington: Indiana University Linguistics Club.
Arabic grammatical borrowing in Western Neo-Aramaic Werner Arnold
1. Background The modern Aramaic dialects are the remnants of a wide variety of old and middle Aramaic dialects that dominated the Middle East in antiquity. The western variety of Aramaic survived only in three villages in the Qalamūn mountains in Syria, namely Maʿlūla, Baxʿa and Jubbʿadīn. Although the three villages lie close to each other, there are remarkable differences in the language, so that one can speak of three different dialects. In the phonology the dialect of Baxʿa is more archaic than the two other dialects. Jubbʿadīn is the most progressive dialect. On the other hand the dialects of Jubbʿadīn and Maʿlūla are in morphology and vocabulary more archaic than that of Baxʿa, where gender distinction is lost in the plural of verbs, adjectives and pronouns. The population of Maʿlūla is Christian with a small Muslim minority. Baxʿa and Jubbʿadīn are purely Muslim villages. There are no significant differences in the dialect between Muslims and Christians in Maʿlūla. Today the language, known as Western Neo-Aramaic (or Maʿlūla Armaic), is spoken by a maximum of 10,000 people. Among all Neo-Aramaic languages Western Neo-Aramaic (WNA) is the only language with a growing number of speakers. The language is a vernacular, not written, and only spoken in everyday life, within the village and the families. The introduction of a writing system (with Hebrew characters!) by a teacher of English in Maʿlūla is still in the beginning and not accepted by everbody in the village. The language of instruction and religious worship is Arabic. Therefore all inhabitants of the three villages speak Arabic as a second mother tongue. While in Baxʿa und Jubbʿadīn an Arabic dialect is spoken which is very similar to the dialect of the neighbouring villages, the people of Maʿlūla adopted the city dialect of Damascus not later than in the nineteenth century. The sound changes which are connected with the adoption of the new dialect did not effect the Aramaic language of the village, including the vocabulary which was borrowed from Arabic until that time. The fact that nearly all Arabic loans in Maʿlūla originate from the period before the change from the rural dialect to the city dialect of Damascus shows that the contact between the Aramaeans and the Arabs
186
Werner Arnold
was intimate during the centuries and that the required Arabic vocabulary was incorporated into the Aramaic dialect to such a high degree, that after the change to the dialect of Damascus, only a limited number of words were borrowed. One can find words mainly from the standard language and only sporadically words originating in the dialect of Damascus in this last period of loans. Arabic has been the only contact language for more than a thousand years, therefore NWA contains only a few Turkish and Greek loans. The fact that Aramaic and Arabic have a long common history and that they are very closely related languages that widely correspond in phonology and morphology, facilitates the mutual adoption and assimilation of loans, but makes the identification of borrowings from one to the other very difficult. The present work is based predominantly on my own fieldwork in the three villages that has been carried out in the years 19851987. More results of that research are published in Arnold (1990, 2000, 2002).
2. Phonology 2.1. Consonants The most noticeable sound shifts which occurred in WNA concern the socalled Begadkephat consonants. In an earlier stage of the Aramaic language the phonemes /b/ [b], /g/ [g], /d/ [d], /k/ [k], /p/ [p] und /t/ [t] had the spirant allophones /v/ [v], /ġ/ [ɣ], /¦/ [ð], /x / [x], /f/ [f] and /§/ [θ] which appear basically after vowels. In WNA the acoustic difference between spirant and plosive pronunciation is preserved. However, it is fixed for each single word and within each root so that the former allophones have become phonemes. The spirants are preserved with the exception of the voiced labiodental fricative [v] which shifted to the voiced bilabial plosive [b] most likely under the influence of Arabic which does not know a phoneme /v/. The old voiced plosives were devoiced and the old voiceless plosives [k] and [t] were palatalized. The same sound change that occurred within Aramaic can also be observed in the majority of the Arabic loans: (1)
*d [d] → t [t] ḳerta < qird ‘monkey’ *t [t] → č [ʧ] čamam < tamām ‘completely’ *b [b] → p [p] xappōza < xabbāz ‘baker’
Western Neo-Aramaic
187
The phonemes /¦̣/ [ð~] and /ž/ [ʒ] (Baxʿa /ǧ/ [ʤ]) are restricted to Arabic loans: (2)
_̣arfa < _̣arf žayša, Baxʿa ǧayša < ǧayš
‘skin bag for butter’ ‘army’
We might posit a pronunciation [g] for the Arabic phoneme /ǧ/ [ʤ] for the earliest time of language contact between Aramaic and Arabic in Syria; in any case, this Arabic /ǧ/ was treated like Aramaic /g/ and shifted after consonants to the voiceless consonant [k] and in word initial position and after vowels to the spirant ġ [ɣ]: (3)
initial position: ġmōʿča < gamāʿa ‘crowd’ after vowels: farraġ < farrag ‘he looked’ after consonants: mawkʿa < mawgaʿ ‘pain’
In all later borrowings Arabic /ǧ/ appears in all positions as ž [ʒ] (in Baxʿa ǧ [ʤ]). The consonants [d], [g] and the glottal stop ʾ [ʔ] in word-central position are attested only in very few loans of Arabic and European origin which are not fully assimilated to the Aramaic system of phonology. The assimilation of [n] to the following consonant is a very old phenomenon of the Aramaic language. Younger speakers try to avoid the assimilation of [n] under the influence of Arabic, where [n] is normally not assimilated to the following consonant, if the consonant [n] can be reconstructed from other derivations of the same root, as in the following example: (4)
nōḥeč yiḥḥuč → yinḥuč
‘he comes out’ ‘he should come out’
2.2. Vowels The vowel system of WNA is more complicated than the Arabic system. The Arabic short vowels [i] and [u] appear as in WNA in loans of Arabic origin in stressed syllables as [e] and [o]. The Arabic long vowel ā [a:] appears in Western Neo-Aramaic in general as ō [o:] and is shortened to [a] in unstressed syllables:
188
Werner Arnold
(5)
Arabic siʿr > burǧ > qāḍi >
Aramaic séʿra ‘price’ (but pl siʿrō´ ) bórža ‘tower’ (but pl buržō´ ) ´ ḳō_̣ya ‘judge’ (but pl ḳa_̣yō´ )
In some loans the Arabic imāla (i-umlaut) ā [a:] → ē [e:] is attested (wētya < wādi ‘valley’) as in many Arabic dialects in the area (Arnold and Behnstedt 1993: 98105).
2.3. Stress and syllable structure Aramaic words stress only the final syllable or the penultima. In analogy to the Arabic geminated stems a vowel a is inserted after the geminated radical in Baxʿa and Jubbʿadīn, so that the stress is now on the antepenultima: (6)
Baxʿa and Jubbʿadīn záppani Maʿlūla zappni
‘I sold’
3. Nominal structures (the integration of loans) Masculine singular nouns of Arabic origin receive the ending -a, feminine singular nouns the ending -a or -ča. Masculine plural nouns have the ending -ō(ya), the feminine plural-ending is -(y)ōa. After numerals, an enumeration plural is formed: (7)
Arabic qism
Aramaic (Maʿlūla) ḳesma ‘part’ ḳismō ‘parts’ iᵊr qism ‘two parts’ samaka samᵊka ‘fish’ (sg) samkōa ‘fish’ (pl) arč samkan ‘two fish’
Some Arabic loans receive the original Aramaic diminutive endings -ōna (m) and -(a)nīa (f), which express diminution no longer:
Western Neo-Aramaic
(8) Arabic muʿallim muʿallima
189
Aramaic mʿallmōna ‘teacher’ (m) mʿallmanīa ‘teacher’ (f)
As Arabic and Aramaic are two closely related Semitic languages, they share most of their nominal patterns. Only a few forms are borrowed from Arabic as miCCaC/muCCaC and maCCūC: (9) Aramaic < Arabic mufčḥa < miftāḥ ‘key’ makčūba < maktūb ‘letter’ Gender distinction is lost in Baxʿa and among some speakers in Maʿlūla and Jubbʿadīn too, in all plural forms of the adjectives and of the independent or suffixed pronouns. This can be explained only by the influence of the neighbouring Arabic villages, which do not know gender distinction in these forms at all.1
4. Verbal structures The two old Semitic tenses perfect and imperfect are preserved2 probably under the influence of Arabic (Correll 1978: 153). They are used to express preterite tense and subjunctive exactly as in the Arabic dialects of Syria. Different from Arabic are the two new tenses, the present tense and the perfect that have developed from the old participles. Future tense or optative is expressed in Maʿlūla and Baxʿa with the auxiliary verb batt- (< Arabic badd-)3 with Aramaic person inflection and following subjunctive just as in Arabic: (10) Aramaic: batt-e yīxul Arabic: badd-o yākol will/want-3sg eat-3sg.subj ‘He will/wants to eat.’ To express a progressive action the Arabic prefix ʿam- (< ʿammāl) with present tense is used:
190
Werner Arnold
(11) Aramaic: ʿamšō mōya Arabic: ʿambišrab mayy ‘He is drinking water.’ Verbs of Arabic origin are treated like Aramaic verbs. They are supplied with the same inflectional affixes as Aramaic and carry the same object suffixes. In fact only the radicals of the Arabic root are borrowed and treated as an Aramaic root. Under the influence of the Arabic dialects in the surrounding villages Baxʿa has lost gender distinction in all plural forms of the verb. The masculine form was generally adopted. The following paradigm gives the preterite tense of the basic stem of the verb i_̣ḥek ‘to laugh’: (12) Coll. Arabic _̣aḥak _̣aḥak-at _̣aḥak-u _̣aḥak-t _̣aḥak-ti _̣aḥak-tum _̣aḥak-tin _̣aḥak-t _̣aḥak-na
Aramaic (Maʿlūla) i_̣ḥek _̣iḥk-a i_̣ḥek _̣iḥk-ič _̣iḥk-iš _̣iḥk-ičxun _̣iḥk-ičxen _̣iḥk-i _̣iḥk-innaḥ
‘he laughed’ ‘she laughed’ ‘they laughed’ ‘you (m sg) laughed’ ‘you (f sg) laughed’ ‘you (m pl) laughed’ ‘you (f pl) laughed’ ‘I laughed ‘we laughed
The Arabic verbal roots are integrated into the WNA system of verbal stems exactly like Aramaic roots within the traditional Aramaic system of verbal stems. This is the case with the following Arabic stems: (13) I stem II stem IV stem V stem
Arabic loan iḍḥeḳ ‘he laughed’ ḥammel ‘he loaded’ aġreḳ ‘he fell asleep’ čḥammal ‘he endured’
Aramaic word išmeʿ ‘he heard’ baššel ‘he cooked’ arkeš ‘he woke up’ čzappan ‘he was sold’
All Arabic stems can be incorporated into the Aramaic stem system. Arabic stems, which do not correspond to an Aramaic stem are converted in the following way:
Western Neo-Aramaic
(14) III stem VI stem VII stem VIII stem X stem
Arabic šāraṭ tarāfaq infaǧar iftaham istaqbal
WNA > šōreṭ > črōfeḳ > inᵊfžar > if ᵊčham > sčaḳbel
191
‘to bet’ ‘to accompany’ ‘to explode’ ‘to understand’ ‘to accept’
The Arabic VII. and VIII. stems can be formed also from Aramaic roots to express the passive to the Aramaic I. stem and have replaced with some few exceptions the old Aramaic passive stem E§pʿel: (15) Aramaic fḥ
I stem VII stem Aramaic nġb I stem VIII stem
ifaḥ inᵊfaḥ inġab inᵊčġab
‘to open’ ‘to be opened’ ‘to steal’ ‘to be stolen’
On the other hand the Aramaic passive stem ččaCCaC (< ettaCCaC) can also be formed from Arabic roots: (16) Aramaic wrx awrex ččawrax
Arabic wqf awḳef ‘to prolong’, ‘to erect’ ččawḳaf ‘to be prolonged’, ’to be erected’
Many borrowings of the Arabic I stem appear in Western Neo-Aramaic in the IV stem. The reason for this change is unclear, but it should be mentioned that the IV stem in Western Neo-Aramaic is very productive whereas in the Arabic dialects of Syria the IV stem has disappeared nearly completely: (17) Arabic I stem bada > Aramaic IV stem abᵊt
to begin
5. Other parts of speech All ordinal numerals are borrowed from Arabic and have with the exception of awwal ‘first’ the Arabic imāla ā > ē (ēni ‘second’, ēle ‘third’, rēbeʿ ‘fourth’, etc.). Western Neo-Aramaic borrowed from Arabic the reflexive ḥōl- (< ḥāl) which is used beside Aramaic nefš-. Furthermore the Arabic reciprocal baʿ_̣-
192
Werner Arnold
is borrowed from Arabic. All of them are used with the Aramaic pronominal suffixes. The dativus ethicus is attested in Western Neo-Aramaic and in the Arabic dialects of the area and may be a result of language contact: (18) Aramaic _mex-le šaʿa
Arabic nām-lo sāʿa ‘He slept himself for an hour.’
The Arabic coordinating conjuctions fa ‘and so’, walla ‘or’ and lakin (also lakan, lakinni, lakōn; in Jubbʿadīn lačin, ličin, līčin) or bass for ‘but’ are borrowed from Arabic. The subordinating conjunction _ōb ‘if’ is a calque of Arabic inkān in which in is translated by Aramaic _ and kān by the Aramaic equivalent ōb (Spitaler 1938: 117). The Arabic word is also used in the form nkōn. Other borrowings are iza and law ‘if’, innu (also inne and inni) ‘that’ and lamma ‘when’ (beside Aramaic mi_), illa ‘except’ and others. For the times of the day only _̣ahwa (< ḍaḥwa) ‘late morning’ and ʿaṣᵊr ‘afternoon’ are of Arabic origin. The days of the week are preserved in Maʿlūla and Jubbʿadīn, but they are replaced by the Arabic names in Baxʿa. Adjectives with the Arabic prefix m-, the infix -č- (< Arabic -t-) and the suffix -ōnay (< -āni) expressing affiliation (amrikōnay ‘American’) were incorporated into the Aramaic vocabulary and among young speakers have sometimes replaced the inherited Aramaic form: (19) Aramaic adjective: išmeʿ ‘audible’ Young speakers: mašmuʿ (Arabic masmūʿ) Old Aramaic has no pattern to express elative terms so that WNA was forced to realize comparative and superlative forms by adoption of the Arabic morphological pattern ʾaCCaC: (20) Arabic ṣaġīr aṣġar minn-o aṣġar wāḥid
WNA izʿur ‘small’ azʿar menn-e ‘smaller than he’ azʿar aḥḥa_ ‘the smallest’
6. Syntax The relative particle ti (also či, Baxʿa ći, all < *dī) is Aramaic but the way of subordinating relative clauses is fully compatible with Arabic. In syndetic
Western Neo-Aramaic
193
relative clauses the antecedent of the relative clause is determined and followed by the relative particle while asyndetic relative clauses have no relative particle and the antecedent is indetermined. In old Aramaic such asyndetic relative clauses are unattested (Correll 1978: 117) and must be considered as borrowed from Arabic. (21) a. Asyndetic wō rōʿya ʿamraʿēl ʿizzōye. ep-pret shepherd-indet herd-3sg.pres.pm goat-pl-sf.3sg.m ‘There was a shepherd, (who was) herding his goats.’ b. Syndetic hanna ġamla ti ṭʿille. dp.3sg.m camel rp carry-perf.3sg.m-sf.3sg.m ‘This camel, which has carried him.’ The coordination of circumstantial clauses (Arabic ḥāl) by means of the conjunction w- (and) is not attested in old Aramaic but occurs frequently in WNA and in the modern Arabic dialects of the area (Grotzfeld 1965: 101). Correll (1978: 147) argues that this construction is borrowed from the Arabic dialects as no other Aramaic dialect has similar circumstantial clauses. (22) w hū ʿammallex 4 willa ōle sōblᵊ cp ip-3sg.m walk-3sg.m.pres-pm lo! come-3sg.m-pret mayor-cs blōta. village ‘While he was walking, lo! – the mayor of the village came.’
7. Conclusion Aramaic and Arabic are two closely related Semitic languages with a long common history. The fact that morphemes in the Semitic languages normally consist of three radicals facilitates the mutual adoption and assimilation of loans, but makes the identification of borrowings from one to the other very difficult. The Arabic dialects spoken in the surroundings of the WNA speech island themselves are very much influenced by Aramaic and sometimes preserve Aramaic words which no longer occur in WNA. For many centuries this type of aramaicized Arabic was the only contact language of WNA, and Correll (1978) believes that the conservatism of Arabic is the reason for the
194
Werner Arnold
archaic structure of WNA in comparison to the eastern Aramaic dialects, where under the influence of Turkish and Iranian languages the old verbal system collapsed. WNA has enriched its vocabulary with thousands of words from Arabic and has incorporated the Arabic system of verbal stems, but the integration of Arabic borrowings was performed by a total adaptation to the Aramaic phonological and morphological system. In general, WNA has preserved its linguistic heritage or has developed independent from Arabic.
Abbreviations c cp cs det dp ep f indet ip m
consonant (radical) coordination particle construct state determined demonstrative pronoun existential particle feminine indetermined independent personal pronoun masculine
perf pl pm pres pret rp sf sg subj WNA
perfect tense plural progressive modifier present tense preterite relative pronoun suffix singular subjunctive Western Neo-Aramaic
Notes 1. In the inflexion of the verb, gender distinction in plural forms is also lost. 2. This is not the case with eastern Aramaic dialects. 3. In Jubbʿadīn the auxiliary verb bēl- of unclear origin is used.
References Arnold, Werner 1990 Das Neuwestaramäische. V. Grammatik (Semitica Viva 4/V), Wiesbaden. 2000 The Arabic dialects in the Turkish province of Hatay and the Aramaic dialects in the Syrian mountains of Qalamūn: Two minority languages compared. In: Jonathan Owens (ed.), Arabic as a Minority Language, 347370. Berlin, New York. 2002 Zur Geschichte der arabischen Lehnwörter im Neuwestaramäischen. In: Norbert Nebes (ed.), Neue Beiträge zur Semitistik. Erstes Arbeitstreffen der Arbeitsgemeinschaft Semitistik in der Deutschen. Morgen-
Western Neo-Aramaic
195
ländischen Gesellschaft vom 11. bis 13. September 2000 an der Friedrich-Schiller-Universität Jena. Ed. by (Jenaer Beiträge zum Vorderen Orient 5), Wiesbaden, 511. Arnold, Werner, and Peter Behnstedt 1993 Arabisch-Aramäische Sprachbeziehungen im Qalamūn (Syrien). Eine dialektgeographische Untersuchung mit einer wirtschafts- und sozialgeographischen Einführung von Anton Escher (Semitica Viva 8), Wiesbaden. Correll, Christoph 1978 Untersuchungen zur Syntax der neuwestaramäischen Dialekte des Libanon (Maʿlūla, Baḫʿa, Ǧubbʿadīn); mit besonderer Berücksichtigung der Auswirkungen arabischen Adstrateinfllusses; nebst zwei Anhängen zum neuaramäischen Dialekt von Ǧubbʿadīn (= Abhandlungen für die Kunde des Morgenlandes XLIV, 4), Wiesbaden. Grotzfeld, Heinz 1965 Syrisch-arabische Grammatik (Dialekt von Damaskus) (Porta linguarum orientalium, Neue Serie VIII). Wiesbaden. Spitaler, Anton 1938 Grammatik des neuaramäischen Dialekts von Maʿlūla (Antilibanon). (Abhandlungen für die Kunde des Morgenlandes XV, 4). Leipzig.
Grammatical borrowing in North-eastern Neo-Aramaic Geoffrey Khan
1. Background Aramaic, which was one of the major Semitic languages in the pre-Islamic Middle East, still survives today in various vernacular dialects. These NeoAramaic dialects can be divided into four main subfamilies, which include (1) the Western group spoken in Maʿlūla and various other villages in the region of Damascus, (2) the Ṭuroyo group, spoken in Ṭūr ʿAbdīn in southeastern Turkey and in the village of Mlaḥsō in southern Turkey, (3) Mandaic, spoken in the city of Ahwāz, Iran, and the surrounding region, and (4) the north-eastern group. North-Eastern Neo-Aramaic (NENA)1 contains a far greater diversity of dialects than any of the other groups. These were spoken across a wide area encompassing northern Iraq, north-western Iran, south-eastern Turkey, Armenia and Georgia. A large proportion of the speakers of these dialects, however, have been supplanted from their original places of residence due to political events during the twentieth century and now live in a diaspora of émigré communities in various parts of the world. On account of this, many of the dialects are now facing extinction. The NENA group includes dialects spoken by Jews and Christians. The Jewish dialects in all cases differ from the Christian dialects, even where the Jews and Christians lived in the same town or region. There are considerable differences, for example, between the Jewish dialect and Christian dialect in the towns of Urmi, Salamas, Sanandaj and Sulemaniyya, in which the two communities lived side by side. In other geographical areas, such as Zakho and the surrounding region, the differences between the dialects of the two communities are of a lesser degree.2 These dialectal cleavages between confessional communities appears to have evolved not only through social divisions but also through different migration histories. Where Jewish and Christian communities existed side by side in towns such as those mentioned above, in some cases it can be established that the settlement of the Jews in the town was earlier than that of the Christians, who were more recent immigrants from the villages in the surrounding countryside.
198
Geoffrey Khan
The NENA dialects exhibit extensive grammatical borrowing from the non-Semitic languages with which they have been in contact for many centuries. The main source of borrowing is Kurdish, an Iranian language that was spoken in numerous dialects across the NENA region. Some features have been borrowed also from Arabic and Turkic languages. Grammatical borrowing is very diverse across the NENA group and much of it still remains unstudied. For this reason I shall here take as a case study one particular dialect of the group that exhibits widespread borrowing and has been fully described, viz. the Jewish dialect of Sulemaniyya (North West Iraq).3 There was a large Aramaic speaking Jewish community in Sulemaniyya since the foundation of the town in 1784 by Ibrāhīm Pāshā Bābān. The majority of the Jews came from the village of Qaradax, situated twenty miles to the south. Between 1950 and 1952 the entire community, which consisted of approximately 500 households, migrated to Israel. Today the dialect is still spoken by only a few elderly immigrants in Israel and is likely to be totally extinct within the next few years. The Muslims of Sulemaniyya are almost entirely Kurdish speaking, with a small minority of inhabitants whose first language is Turkmen. The Jews of the town spoke the local Kurdish dialect in order to communicate with the Muslim inhabitants. The source of the grammatical borrowing in the Aramaic dialect can in some cases be identified as the Sulemaniyya Kurdish dialect. In other cases, however, it appears to have arisen by contact with other Kurdish dialects or other Iranian languages in more remote regions. Sometimes the source is Arabic or a Turkic language. Some of these features may have been transferred to the Sulemaniyya Jewish Aramaic dialect indirectly through other Jewish Aramaic dialects. A few elements have been borrowed also from Hebrew, the traditional language of Jewish education in the community.
2. Phonology A number of phonological changes in the Aramaic dialect appear to be contact-induced. In the Aramaic dialect, for example, the original interdental s and _ shift to the lateral l:4 (1)
*bea > bela ‘house’ *ʾida > ʾila ‘hand’
North-eastern Neo-Aramaic
199
This shift is found in all Jewish NENA dialects that were spoken in communities east of the Zab river (the so-called ‘trans-Zab group’).5 The shift of * to the lateral /l/ seems to have been preceded historically by a shift to the voiced stop /d/. This is shown by the fact that in some dialects in which the shift to /l/ has taken place the reflex of * has remained as /d/ in some words. For instance, in the Jewish dialects of Urmi, Ruwanduz and Rustaqa the root ‘to come’ is ʾdy (< *ʾy), e.g. ʾidyele ‘he came’. The process resulting in the lateral /l/ described above, therefore, can be regarded as resulting from the weakening of the articulation of /d/. Interdentals are absent in Sulemaniyya Kurdish and post-vocalic /d/ is to an interdental approximant /đ/ (MacKenzie 1961: 3, 8) but there is no shift of dentals to the lateral l. Such a shift, however, can be identified in the Mukri dialect of Kurdish spoken in north-western Iran (Kapeliuk 1997). The phoneme /w/ often has a labio-dental realization [v] in the Aramaic dialect but the corresponding phoneme in the local Kurdish is always realized as a bilabial. The labio-dental realization is a development found in Farsi and also in the Aramaic dialects of western Iran. In the Aramaic dialect an /l/ in the environment of pharyngalized consonants sometimes shifts to /r/: (2)
parṭixwa < palṭixwa ‘We used to go out.’
This has been found only in the speech of women informants. A parallel phonological shift is attested in the Kurdish speech of some women from Sulemaniyya and regularly in the Kurdish dialects spoken in Arbel, Koy Sanjaq and Ruwanduz (MacKenzie 1961: 4, 28). After a vowel, /d/ and /z/ may freely alternate with each another. This process appears to be restricted to specific lexical items and is best treated as an alternation of phonemes rather than as allophonic alternation: (3)
ʾidyo ~ ʾizyo ‘today’ qadome ~ qazome ‘tomorrow’ guda ~ guza ‘wall’
In the speech of the closely related Aramaic dialect of the Jews of Ḥalabja the three-way alternation d ~ _ ~ z is sometimes heard, with the intermediate interdental:
200
Geoffrey Khan
(4)
xădir ~ xă_ir ~ xăzir ‘he becomes’ guda ~ gu_a ~ guza ‘wall’
The weakening of the stop /d/ to an interdental /¦/ or sibilant /z/ may have been stimulated by contact with the Kurdish dialects of the region, in which the articulation of /d/ is weakened to an interdental approximant in postvocalic position. The shift of etymological /z/ to /d/ would, therefore, have to be regarded as a back-formation. It is also relevant to note that the alternation d ~ _ ~ z is found in the Jewish Persian dialects of western Iran.6 The stress patterns of the Aramaic dialect are similar to those that are found in the Sulemaniyya Kurdish dialect. In both dialects stress is generally placed at the end of a word. Exceptions to this tend to be found in the same categories of words: (5)
Vocative Aramaic táta ‘father!’ Kurdish mā´mwastā ‘teacher!’ Past verbs Aramaic híyen ‘They came.’ Kurdish hátin ‘They came.’ Stress groups combining two or more words Aramaic tré-yome ‘two days’ Kurdish dé-rož ‘two days’
3. Nominal structures The Jewish Aramaic dialect disinguishes two genders (masculine and feminine). This contrasts with the Kurdish dialect of Sulemaniyya, which makes no morphological distinction in gender. Under the influence of the Kurdish dialect, however, the original gender distinction in the third-person singular pronoun of the Aramaic dialect has been lost: (6)
Aramaic Kurdish ʾo aw ‘he/she/it’
The demonstrative adjectives in Aramaic have also lost the distinction between singular and plural in imitation of Kurdish:
North-eastern Neo-Aramaic
(7)
201
Aramaic Kurdish ʾay am ‘this/these’ ʾo aw ‘that/those’
Many nouns have been borrowed by the Aramaic dialect from Kurdish. Since the Kurdish dialect makes no gender distinctions, they have been assigned a gender in the Aramaic dialect. The grammatical gender of all loanwords that refer to human beings corresponds to the sex of the referent. The majority of loanwords that refer to inanimate objects or small animals are construed as feminine in gender, e.g. šāx (f.) ‘mountain’, qali ‘carpet’ (f.), jiji (f.) ‘hedgehog’ The same applies to loans from Arabic which have entered the dialect through Kurdish, e.g. kĭteb (f.) ‘book’, majlis (f.) ‘meeting’. There is a sizeable residue of inanimate loans that are construed as masculine in gender. The gender assignment of these appears to have a semantic basis, in that most of the nouns in question either denote (i) a long, thin entity, e.g. qamīš (m.) ‘cane’, top (m.) ‘gun’ or (ii) a collective or non-solid entity, .e.g. xaḷūz (m.) ‘coal’, čay (m.) ‘tea’. Many borrowed nouns are adapted to Aramaic morphology by adding an Aramaic nominal ending. In most cases the Aramaic masculine ending -a is added, irrespective of the gender assignment: (8)
boqa (f.) ‘frog’ < Kurdish boq lăša (m.) ‘body’ < Kurdish laš
In a few cases the Aramaic feminine ending -ta is added to Kurdish loans that have been assigned feminine gender: (9)
masita (f.) ‘fish’ < Kurd. masi dargušta (f.) < Kurd. darguš
The Aramaic dialect uses a definite article suffix -ăke that has been borrowed from Kurdish. In the local Sulemaniyya Kurdish dialect the form of the article is -aka. The form -ăke is likely to have its origin in a form of the particle with an oblique case marker -aka-y, which is found in Kurdish dialects lying to the North and North East of Sulemaniyya (MacKenzie 1961: 5758). Furthermore, the morphological behaviour of the particle in the Aramaic dialect is different from that of the particle in Kurdish. In the Aramaic dialect it cannot take any further suffixes. In Kurdish, by contrast, it may take plural (-ān) and pronominal suffixes:
202
Geoffrey Khan
(10) Aramaic barux-ăke barux-awal-ăke barux-an
Kurdish dost-aka ‘the friend’ dost-akān ‘the friends’ dost-aka-mān ‘our friend’
Possessive constructions in which a head noun is qualified by a determiner noun are normally formed by juxtaposing the two. The Aramaic genitive particle d-, which regularly occurs in such constructions in some NENA dialects, is rarely used. The construction without the d- corresponds to a functionally equivalent construction in Sulemaniyya Kurdish whereby a head noun is linked to a following determiner noun by means of a compound vowel -a (MacKenzie 1961: 64): (11) Aramaic Kurdish brona mălik kur¯-a pāšā
‘the son of the king’
The final -a of the Aramaic noun is the general nominal ending, but in this environment it is identified with the Kurdish compound vowel -a. The Iranian Ezāfe particle is occasionally used in genitive constructions when the head noun is a loanword: (12) maktab-i hulaye ‘school of the Jews’
4. Verbal structures In both the Aramaic and Kurdish dialects the copula verb ‘to be’ is expressed by an enclitic on the predicate that is inflected for person and number like a verb. Although there are signs of the emergence of such a enclitic in earlier Eastern Aramaic, its full development and acquisition of verbal inflection are apparently due to the influence of Kurdish: (13) Aramaic Kurdish ʾo jwan-ya aw jwān-a ‘She is beautiful’ ʾat jwan-yat to jwān-ī ‘You are beautiful’ The simple past tense of the Aramaic dialect has acquired an ergative type of inflection by contact with Kurdish. The intransitive verbs have active subject inflection, which resembles the inflection of present tense verbs, whereas
North-eastern Neo-Aramaic
203
transitive verbs have a passive type of inflection, with the verb agreeing with the patient of the action and the agent expressed by an oblique agentive pronominal affix that does not serve as the grammatical subject of the verb. The oblique pronominal affix is identical to the pronominal object affix of the present form of the verb. These structures correspond to what is found in the Sulemaniyya Kurdish dialect: (14) Second person Aramaic Simple past intransitive mīl-et ‘You (ms.) died’ mīl-at ‘You (fs.) died’ mīl-etun ‘You (pl.) died’ Simple past transitive qṭil-lox ‘You (ms.) killed him’ qṭil-lax ‘You (fs.) killed him’ qṭil-laxun ‘You (pl.) killed him’ Kurdish Simple past intransitive mird-ī ‘You (s.) died’ mird-in ‘You (pl.) died’ Simple past transitive kušt-it ‘You (s.) killed him’ kušt-tān ‘You (pl.) killed him’
Present mel-et ‘You (ms.) die’ mel-at ‘You (fs.) die’ mel-etun ‘You (pl.) die’ Present + pronominal object qăṭil-lox ‘He kills you (ms.)’ qăṭil-lax ‘He kills you (fs.)’ qăṭil-laxun ‘He kills you (pl.)’ Present a-mir-ī ‘You (s.) die’ a-mir-in ‘You (pl.) die’ Present + pronominal object a-t-kužē ‘He kills you (s.)’ a-tān-kužē ‘He kills you (pl.)’
In addition to this simple past tense, consisting of a past stem and inflectional endings, the Aramaic dialect also has a compound perfect consisting of a passive participle and an enclitic form of the verb ‘to be’. The same two types of past conjugation are found in the Kurdish dialect: (15)
Aramaic Simple past: mīl ‘He died’ Compound past: mila-y ‘He has died’
Kurdish mird ‘He died’ mirduw-a ‘He has died’
Although there are clear contact-induced resemblances between the Aramaic and Kurdish dialects, a closer look reveals several differences. In the Jewish
204
Geoffrey Khan
Aramaic dialect, for example, the agentive pronominal suffix on the simple past transitive verb is bonded to the verb. In Kurdish, on the other hand, this agentive suffix is moveable and is generally attached to a word that occurs earlier in the clause. The transitive predication ‘the man killed the dog’, for example, is expressed in the two dialects as follows: (16) Aramaic gor-ăke kalb-ăke qṭil-le. man-the dog-the killed-past:3ms-3ms:obl. Kurdish pyāw-aka sag-aka-y kušt. man-the dog-the-3ms:obl killed-past:3ms The compound past tense in the Kurdish dialect has an ergative form of inflection whereas the corresponding conjugation in the Jewish Aramaic dialect has an active type of inflection in both the intransitive and transitive verbs: (17) Aramaic gor-ăke mila-y. man-the died-cop:3ms ‘The man has died.’ gor-ăke kalb-ăke qiṭl-u-y. man:the dogs.the killed-3pl:obl-cop:3ms ‘The man has killed the dogs’ Kurdish pyāw-aka mirduw-a. man-the died-cop:3ms ‘The man has died.’ pyāw-aka sag-akān-ī kuštuw-in. man:the dogs.the-3ms:obl. killed-cop:3pl ‘The man has killed the dogs.’ It is worth noting that in a few Jewish NENA dialects in neighbouring western Iran, the compound past verbal form is ergative in inflection. This is the case, for example, in the Jewish dialects of Sanandaj and Kerend, e.g.
North-eastern Neo-Aramaic
205
(18) Jewish Sanandaj gor-ăke kalb-ăke qiṭl-en. man:the dogs.the killed-cop:3pl ‘The man has killed the dogs.’ The Kurdish dialect of Sulemaniyya does not, in fact, have a pure ergative system regarding case-marking and verb agreement in either of the past tenses, since it is lacking an ergative case marker on the agent noun. The agent is not marked with an oblique case but is extraposed and resumed by an oblique pronominal element. This feature is shared by the Jewish Aramaic dialect of Sulemaniyya, but the Aramaic dialect has moved even further away from the ergative system than the Kurdish one. In the Aramaic dialect not only is the compound past conjugation inflected actively but even the simple past conjugation is beginning to be treated as an active form. One reflection of this is that the patient of the action is sometimes given oblique inflection. This is regularly the case, for example, when the object is a first- or second-person pronoun: (19) qṭil-le ʾillox. killed-3ms:obl 2ms:obl. ‘He killed you.’ One linguistic innovation that is found in the verbal system of the Jewish Aramaic dialect is the expression of the present progressive by an infinitive combined with a enclitic form of the verb ‘to be’: (20) šaqola-yet. taking-cop:2ms ‘You are taking.’ This appears to have developed from a construction in which the infinitive is combined with the locative preposition b-, which is still preserved in some NENA dialects, e.g. Christian Urmi: (21) bi-šqala-vit. in-taking-cop:2ms ‘You are taking.’
206
Geoffrey Khan
No direct parallels to this infinitive construction are found in Sulemaniyya Kurdish, or indeed in other Kurdish dialects but there are close parallels to the Aramaic construction in languages spoken further north, such as Turkish, Eastern Armenian (Chyet 1995: 246) and some Iranian dialects belonging to the Tati group spoken in north west Iran (e.g. Chali, cf. Yar-Shater 1969: 225), e.g. Turkish almak-ta-sın ‘You are taking’ (taking-loc-cop:2S.); Chali xordan-u-ind ‘They are eating’ (eating- loc-cop.3pl).
5. Other parts of speech A large number of other grammatical words have been borrowed from Kurdish. These include a variety of particles and adverbials, including subordinating clausal conjunctions such as ʾagar ‘if’, taku ‘in order that’, nakun ‘lest’ and the relative particle ga, connectives such as ʾinja ‘then’, ham ‘also’, yan ‘or’, čunga ‘because’, modal particles such as the volitive particles ba and mar expressing deontic modality in verbs, the phasal particle heštan ‘still’, ‘yet’, the comparative particle biš ‘more’, the negative modifier of nouns hič ‘no’, and a variety of adverbials, e.g. čannakaw ‘suddenly’, dubara ‘again’, har ‘always’. The Kurdish post-verbal particle awa (after a vowel wa) is widely used in the Aramaic dialect after Aramaic verbs. It often expresses the sense of ‘returning back’, ‘restoring’, ‘repetition’ or ‘completion’: (22) hiye-wa. come:past:3ms-particle ‘He came back.’ (23) qadome der-awa. tomorrow return:fut:3ms-particle ‘He will return tomorrow.’ The Kurdish inclusive particle -iš has been borrowed. In the Aramaic dialect, however, the particle is not as integrated into the morphology as it is in Kurdish. This is reflected by the fact that it is always an external affix in word-final position in the Aramaic dialect, whereas in Kurdish it precedes a pronominal suffix (MacKenzie 1961: 128): (24) Aramaic nošew-iš
Kurdish xō-š-i ‘also himself’
North-eastern Neo-Aramaic
207
Some particles in the Aramaic dialect are borrowed from Kurdish but do not correspond exactly to what is found in Sulemaniyya Kurdish. On some occasions we seem to be dealing with doublets. This applies, for example, to the particles gal ‘with’ and laga ‘at the home of’ in the Aramaic dialect, both of which appear to be related to the particle lagaḷ in the Sulemaniyya Kurdish dialect. A particle of Aramaic original may imitate the use of an Aramaic particle. This applies, for example, to the Aramaic particle k- which is attached to present base verbs in the indicative mood. In the present state of the dialect the particle has, in fact, been lost by phonetic attrition, but is still preserved in some verbs. The etymology of the particle is clearly Aramaic (cf. Babylonian Talmudic Aramaic qā- < qāʾem ‘rising up’), but its use is likely to be an imitation of the use preverbal present indicative particles in Kurdish, in the case of Sulemaniyya Kurdish this is a-.
6. Syntax The basic word order of the Jewish Aramaic dialect is SOV when the verbal arguments are free-standing nominals, which corresponds to that of the Kurdish dialect (see examples 16 and 17). Various borrowed subordinating particles are used to introduce subordinate clauses in the Aramaic dialect. The Kurdish subordinating particle ka-, which is used in the Kurdish dialect of Sulemaniyya and the surrounding region (MacKenzie 1961: 131), has been borrowed by the Aramaic dialect (pronounced either ka- or ga-) to introduce various types of subordinate clause. Attributive relative clauses are often introduced by this Kurdish particle. The native Aramaic relative particle d- has been completely lost in relative clauses after head nouns. As in Kurdish, the relative clause follows its head: (25) ʾo-baxta ka-xăzitta ga-doka šwawt-i-ya. that-woman who-see:2ms in-there neighbour-1s:obl-cop:3fs ‘The woman whom you see there is my neighbour.’ The Kurdish particle ka- is sometimes used in the Aramaic dialect as a subordinating temporal conjunction with the sense of ‘when’: (26) ga-wīš-wa, ʾanten-wa-le. when-dried:3ms-pastp took:3pl:imp-pastp-3ms:obl ‘When it had dried, they took it.’ (R:40)
208
Geoffrey Khan
The particle ka- is also used as a complementizer to introduce factive complement clauses, mainly after the verb ‘to know’: (27) kăyen-wa ga-ʾo brata ʾil-d-o-brona know:imp:3pl-pastp comp-that(dem) girl objm-that(dem)-boy gba. loves:imp:3fs ‘They would know that the girl loves the boy.’ The protasis of a conditional constructions in Aramaic is generally introduced by the Kurdish particle ʾagar ‘if’: (28) ʾagar la-gbitta,ʾana kun-naw ta-naš xet. if neg-love:imp:3ms:subj-3fs obl to-person other ‘If you do not love her, I shall give her to some other person.’ (R:165) Temporal ‘when’ clauses are frequently introduced by the Kurdish nominal waxta ‘time’: (29) waxta ʾana menna, ʾay tre-brone ruwwe, băxew la-koli time I die-1ms dem two-sons big care neg-do:fut:3pl l-yal-ax. objm-children-2fs.obl ‘When I die, these two older boys will not look after your children.’ (R:17) The Arabic particle ḥatta, or its variant form hatta with a laryngeal, is used as a clausal conjunction, generally with the sense of ‘until’ and introduces an event that marks the endpoint of the activity or situation denoted by the main clause: (30) rĭqa-le bari ba-ṣiwa ḥatta ran:past-3ms.obl after-1s.obl with-stick until di-le-lli. hit:past-3ms.obl-1s.obl ‘He ran after me with a stick until he beat me.’ Syntactically subordinate clauses that have a concessive sense are generally introduced by the Hebrew particle afillu:
North-eastern Neo-Aramaic
209
(31) ʾafillu ʾo barux-i-ye, ʾana yaridew although he friend-1s.obl-cop.3ms. I help-3ms.obl la-kunna. neg-do:fut.1ms ‘Although he is my friend, I shall not help him.’
7. Lexicon There is a high degree of lexical borrowing in the Aramaic, especially in the category of nouns. On the basis of a sample corpus of material, the broad assessment of the proportion of loanwords in the lexicon of the various categories of words is as follows: nouns, 67%, adjectives, 48%, particles, 53%, verbs, 15%. The vast majority come from Kurdish. A few originate in other languages, such as Arabic and Turkish, though most of these are likely to have been transmitted through Kurdish. The Arabic loans sometimes exhibit features that are distinctive of Arabic lexemes in Kurdish, such as the pronunciation of the tā’ marbūṭa as -at (e.g. ḥukmat ‘government’, sa‘at ‘hour’) and the use of broken plurals with a singular sense (e.g. tujār ‘merchant’). The Aramaic dialect exhibits also many calques of Kurdish expressions. Verbs constitute the most resistant category to loaning, as is commonly the case in language contact situations. The greater facility with which nouns are loaned is clearly demonstrated by the so-called ‘phrasal verbs’, which consist of an inflected verbal form combined with a noun or particle as its complement. These are nearly all based on a Kurdish model. In most cases, however, only the noun or particle complement is a direct loan of a Kurdish word. The verb is an Aramaic calque of the Kurdish: (32) Aramaic rek lpl swal ʾwl sayr ʾwl gargurke ʾwl xafad ʾxl gorane ʾmr wa dmy
Kurdish rek kawtin swal kirdin sayr kirdin gargurke kirdin xafat xuwardin gorani wittin wā zānīn
‘to agree’ ‘to beg’ ‘to look at’ ‘to crawl’ ‘to be distressed’ ‘to sing’ ‘to think’
Borrowed verbs, by contrast, are inflected fully with the Aramaic verbal inflection. The existence of a rich inflectional morphology in verbs is no doubt
210
Geoffrey Khan
one reason why the verbal section of the lexicon has been more resistant than nouns to borrowing. One of the few verbs in the Aramaic dialect that have been loaned from Kurdish is dyy (< Kurd. dān). This is used in various phrasal verbs that are based on Kurdish models, e.g. (33) Aramaic bāz dyy čirike dyy čapḷe dyy
Kurdish bāz dān ‘to jump’ čirike dān ‘to shout’ čapḷe dān ‘to clap’, ‘applaud’
The Kurdish verb dān has a wide range of meanings, including ‘to give’, ‘to hit’, ‘to put’. The corresponding Aramaic form, dyy, regularly occurs in the Aramaic phrasal verbs that have Kurdish models with dān. The distribution of dyy in the Aramaic dialect outside of this context, however, is not as wide as that of Kurdish dān. When used independently of phrasal expressions, the verb dyy in Aramaic most commonly has the sense of ‘to hit’. It is not used in the meaning of ‘to give’, which is one of the basic senses of the Kurdish source word dān. The Aramaic dialect retains the native verb to express this meaning (present indicative: kul, past: hiwle) and so has resisted a complete lexical transfer. The verb borrowed from Kurdish in some cases has taken on the full basic meaning of the Aramaic equivalent but nevertheless the dialect retains the Aramaic verb and uses it in a slightly different meaning. This applies, for example, to the fate of the Aramaic verb prx. In earlier Aramaic *prḥ had the sense of ‘to fly’ and, indeed, this sense is still retained by the verb in some NENA dialects. In the Jewish dialect of Sulemaniyya, however, the meaning ‘to fly’ is expressed by the Kurdish loanword fry (< Kurd. firīn). The native verb prx, nevertheless, is still retained in the sense of ‘to jump over’, which is connected conceptually with the notion of ‘flying’. On some occasions the contact with Kurdish has not brought about a direct borrowing of lexical items but rather as given rise to the development of a phonetic resemblance between morphological forms in the two languages. In the Aramaic dialect, for example, the verb ‘to come’ (< *ʾy) undergoes irregular phonetic contraction and loses its original middle radical * completely. As a result, the base of the present conjugation of the verb resembles phonetically the corresponding Kurdish verb: (34) Aramaic k-e (3ms indicative) k-en (3pl indicative)
he (3ms subjunctive) hen (3pl subjunctive)
North-eastern Neo-Aramaic
Kurdish e (3s indicative) en (3pl indicative)
211
b-e (3s subjunctive) b-en (3pl subjunctive)
One may perhaps include here the irregular loss of the final /m/ in the word ʾidyo ‘today’ (< *ʾid-yom) under the influence of the vocalic pattern of the Kurdish equivalent imro. Another case of phonetic convergence is the Aramaic form gbe, which functions as an invariable verb form expressing necessity. The form is derived from the Aramaic verb ‘to want’. Its function is identical to that of the Kurdish invariable verb abe (MacKenzie 1961: 106), which it resembles phonetically, e.g. (35) Aramaic gbe hezex
Kurdish abe biçīn
‘We must go’
Some lexical items in the Aramaic dialect are non-Semitic loanwords yet do not correspond to what is found in the local Kurdish dialect of Sulemaniyya. This is the case, for example, with the two basic kinship terms in the Jewish Aramaic dialect tata ‘father’ and lala ‘maternal uncle’. They are not found in the local Kurdish dialect, but parallels can be found in several other languages of the region, the nearest being Hawrami, spoken in the Hawraman mountains. Similarly a loanword in the Aramaic dialect may be identified in the Kurdish dialect of Sulemaniyya but with a different meaning, whereas it is used with the same meaning as is found in the Aramaic dialect in a more remote linguistic source. This applies to the word baba, which, in the Aramaic dialect of Sulemaniyya, usually means ‘grandfather’ rather than ‘father’. In the Kurdish dialect of Sulemaniyya the cognate word has the sense of ‘father’ (bāb, bāw), but in Hawrami it is used with the sense of ‘grandfather’ as in the Aramaic dialect: (36) Jewish Sul. Ar. tata lala baba
Sulemaniyya Kurdish bāw, bāwk xal ba-pīr
Hawrami tata ‘father’ lalo ‘maternal uncle’ baba ‘grandfather’
On some occasions a Kurdish loanword in the Aramaic dialect has a different range of meaning from what it has in the source language. The word čolăka, for example, is used in the Aramaic dialect with the meaning of ‘bird’. In
212
Geoffrey Khan
Kurdish it has the specific meaning of ‘sparrow’, the general term for bird’ being mal.7 The loanword mayīn is used in the Aramaic dialect as a general word for ‘horse’, whereas in the Kurdish source language it refers specifically to a ‘mare’. Most surviving speakers of the Aramaic dialect who have been resident in Israel since the 1950s use Hebrew words in their Aramaic speech. A large proportion of these words are taken from Modern Hebrew. Particularly common are the connective particles ʾaz ‘then’ and ʾaval ‘but’. As is the case with lexical transfer from Kurdish, Hebrew verbs are not so freely borrowed. Speakers prefer to form verbal phrases containing a Hebrew nominal element and an Aramaic verb: (37) mazmīn koli-wa-le. invite-heb.participle do-imp.3pl-pastp.3ms.obl ‘They would invite him.’ Certain Hebrew words that occur in the informants’ speech, however, existed in the Aramaic dialect before the immigration of the Jewish community to Israel. These can usually be distinguished by phonetic features that are not characteristic of Modern Hebrew. In such words, for example, vocalic šewa is pronounced /ă/, as in băraxa ‘blessing’, nădaba ‘charity’, băli ‘without’. Consonantal gemination is pronounced, as in ʾafillu ‘even if’, keʾillu ‘as if’, sukka ‘booth’. Beth is pronounced as a stop where it is a fricative in Modern Hebrew, as in tob ‘good’, kabod ‘respect’. In some cases a Hebrew word has undergone a phonetic process under the influence of Kurdish, which demonstrates that it is a heritage from the Aramaic dialect as it was spoken in Kurdistan, e.g. mira < mila ‘circumcision’.
Abbreviations comp cop dem fut imp loc neg
complementizer copula demonstrative future imperfective locative negator
obj objm obl pastp pres subj
object object marker oblique past particle present subject
North-eastern Neo-Aramaic
213
Notes 1. The term was coined by Hoberman (1988: 557) to replace ‘Eastern Neo-Aramaic of earlier classifications (cf. Socin 1882: v; Duval 1896: 125; Tsereteli 1977, 1978). This was necessary in order to distinguish the north-eastern dialects from modern Mandaic, which is as distant typologically from them as the western Neo-Aramaic dialects. 2. Cf. Hopkins (1993: 65). 3. The dialect is described in Khan (2004). All the data in this chapter are taken from this description, which is based on extensive fieldwork undertaken in Israel with Jewish immigrants from Sulemaniyya. 4. The transcription follows that adopted in Khan (2004). Vowel length is largely predictable from the syllable structure. As a general rule, in open syllables a vowel is long and in closed syllables it is short. Vowels following this principle are not marked by diacritics. The breve and macron signs are used only when the vowel is short in an open syllable or long in a closed syllable respectively. 5. For the main characteristics of the trans-Zab group, see Mutzafi (2004: 910). 6. I am grateful to Don Stilo (personal communication) for drawing my attention to these phenomena in the Kurdish and Jewish Iranian dialects. 7. This is similar to the semantic relationship between Arabic ʿuṣfūr ‘sparrow’ and Hebrew ṣippor ‘bird’, which are cognates.
References Chyet, M. L. 1995 Neo-Aramaic and Kurdish: An interdisciplinary consideration of their influence on each other. Israel Oriental Studies 15: 219249. Duval, R. 1896 Notice sur les dialectes néo-araméens. Mémoires de la société de linguistique de Paris 9: 125135. Hobermann, R. D. 1988 The history of the Modern Aramaic pronouns and pronominal suffixes. Journal of the American Oriental Society 104: 221231. Hopkins, S. 1993 The Jews of Kurdistan in Eretz Israel and their language. Peʿamim. Studies in Oriental Jewry 56: 5074. Khan, G. 2004 The Jewish Neo-Aramaic Dialect of Sulemaniyya and Halabja. Leiden: Brill. MacKenzie, D. N. 1961 Kurdish Dialect Studies, Volume 1. London: Oxford University Press.
214
Geoffrey Khan
Mutzafi, H. 2004 Socin, A. 1882
The Jewish Neo-Aramaic Dialect of Koy Sanjaq. Wiesbaden: Harrassowitz.
Die neu-aramäischen Dialekte von Urmia bis Mosul. Texte aund Übersetzungen. Tübingen: Laupp. Tsereteli, K. G. 1977 Zur Frage der Klassifikation der neuaramäischen Dialekte. Zeitschrift der Deutschen Morgenländischen Gesellschaft 127: 244253. 1978 The Modern Assyrian Language. Moscow: Nauka. Yar-shater, E. 1969 A Grammar of Southern Tati dialects. The Hague: Mouton.
Grammatical borrowing in Macedonian Turkish Yaron Matras and Şirin Tufan
1. Background The variety described here is representative of the Turkish dialects spoken in the Republic of Macedonia, especially those in the west of the country, and to a considerable extent also of Rumelian or Balkan Turkish as a whole (cf. e.g. Matras 1998, 2004; Friedman 2003). The Balkan or Rumelian dialects of Turkish descend directly from Ottoman Turkish and are generally considered mutually comprehensible with Standard Turkish (henceforth ‘Tk.’); there are even direct historical links with Anatolian Turkish (cf. Caferoğlu 1964). We draw here primarily on data from the dialect of Gostivar, a city in the western part of the Republic of Macedonia – henceforth GT for ‘Gostivar Turkish’ (for a comprehensive description see Tufan 2007).1 Turkish is the native language of the Turkish ethnic minority in the various Balkan countries. It is the first language of many Muslim Romani communities, and it is also spoken by some Albanians, Macedonians, and other ethnicities as a second or third language.2 As the official language of the Ottoman Empire, Turkish was a lingua franca and the language of administration and trade in the Balkans for more than half a millennium (between the fourteenth and early twentieth century). With the final collapse of the Ottoman Empire (1912), Turkish became a minority language. In Macedonia, it was not until the 1950s that its status became regulated and Turkish-language education, cultural institutions, and media received state backing. The form of Turkish taught at school was Standard Turkish, while the vernacular continued to be used in the private domain. Turkish speakers in the region are generally biand often trilingual, speaking, in western Macedonia, alongside the state language, also Albanian. Over the past century, and especially since the 1950s, the importance of the state language and its relevance to career progression, education, and mobility has grown immensely, and this is reflected in the amount and the nature of Macedonian lexicon that has found its way into the local varieties of Turkish. In today’s Republic of Macedonia, Turkish speakers have direct contact with Standard Turkish not only through schooling, but also through satellite television and the internet, which are present in almost every Turkish household. Successive waves of emigration to Turkey in recent
216
Yaron Matras and Şirin Tufan
decades have further fortified personal ties with Turkey, and visits to Turkey are frequent, resulting in even greater exposure to Standard and Anatolian Turkish.
2. Phonology Among the consonants, we find the dental-alveolar affricate /ts/, which has its source in Macedonian and Albanian. It is found not only in loanwords (Albanian-derived tsapo ‘goat’, Macedonian-derived tsevka ‘pipe’) and in borrowed affixes (Macedonian feminine-agentive -itsa), but it is also transferred occasionally into native Turkic words: tsıs ‘shut up’ (cf. Tk. sus). Initial consonant clusters are permitted in GT which do not appear in Tk.: GT (also Macedonian and Albanian) Stambol ‘Istanbul’, Tk. İstanbul. There are, on the other hand, also cases of simplification. The surrounding non-Turkic languages simplify Turkish geminates in Turkish borrowings (cf. Friedman 2003: 58), and this trend is also found in GT: /dükan/ ‘shop’, Tk. /dükkan/ ‘shop’; /akıli/ 'clever', Tk. /akıllı/. As in the neighbouring languages, there is a weakening of /h/, though the origins of this development in Western Rumelian Turkish are thought to be in the features carried by immigrants from northeast Anatolia (Németh 1956: 21): GT /ayvan/ ‘animal’, Tk. /hayvan/; GT /paali/ ‘expensive’, Tk. /pahalı/; GT /saba/ ‘morning’, Tk. /sabah/. Recent contact with Standard Turkish appears to have triggered the re-introduction of /h/, and variation is commonly found, especially in grammatical function words such as /em, hem/ ‘and’, /er, her/ ‘every’, or /ep, hep/ ‘all’. In line with the absence of vowel-length distinctions in both Macedonian and Albanian, there is a tendency in GT to shorten ‘double’ or ‘lengthened’ vowels, which appear in Turkish in loans of Persian and Arabic origin: thus /galiba/ ‘probably’ (Tk. /ga»liba/), /hala/ ‘yet’ (Tk. /ha»la/). The loss of /ö/ – which does not exist in the contact languages – may also be a contact-induced phenomenon. In GT historical /ö/ is usually realized as /ü/ or as /o/: GT ürenci ‘student’, Tk. öğrenci; GT dort 'four', Tk. dört.
3. Nominal structures The feminine derivational markers -ka and -(i)tsa are borrowed from Macedonian, and are productive with Turkish word stems: arkadaş ‘friend’ (gender-neutral, and by default masculine), arkadaş-ka ‘female friend’; koyşi
Macedonian Turkish
217
‘neighbour’, koyşi-ka ‘female neighbour’; yalanci ‘liar’, yalanci-tsa ‘female liar’. The suffix -(i)tsa is further extended to denote a female affiliated with an identified male, thus: dayo ‘maternal uncle’, day-tsa ‘maternal uncle’s wife’; Muzafer-itsa ‘Muzaffer’s wife’. The extended distribution of the inherited diminutive suffix -çe appears to be influenced by the presence of a similar form in the neighbouring languages: kış-çe ‘little girl’, Macedonian devoj-če. The case of the dependent in possessive constructions is also affected by contact. The possessor often appears in the ablative case, still accompanied, as in Tk., by possessive inflection on the object of possession (the head), while in Tk. the possessor appears in the genitive: (1)
a. Gostivar Turkish kıskardeş-i güvegi-den sister-3sg.poss groom-abl ‘the groom’s sister’ b. Standard Turkish damad-ın kız kardeş-i groom-gen sister-3sg.poss ‘the groom’s sister’
The construction seems to copy the propositional marking of the possessor in Macedonian, which appears either in the ablative or dative: (2)
Macedonian a. sestra-ta na zet-ot sister-def to groom-def b. od zet-ot sestra-ta from sister-def groom-def ‘the groom’s sister’
4. Verbal structures The copula in GT appears, like in Macedonian, as an independent verb, and not, as in Tk., in an enclitic form. Since this concerns issues of constituent order, the position of the copula will be discussed further in Section 6. A characteristic feature of the verb in Rumelian Turkish is the loss of the modal infinitive, and the reduction of converbal forms in general. As a strategy of clause linkage, this issue is discussed in Section 7 on ‘Syntax’.
218
Yaron Matras and Şirin Tufan
Gostivar Turkish continues the general Turkish pattern of forming new verbs by incorporating lexical nouns from the contact language, and integrating them with a light verb which differentiates valency. Both et- ‘do’ and yap- ‘make’ are employed with transitives, and ol- ‘become’ with intransitives: yaparsın komparatsiya 'you compare', privatizir oldi ‘it was privatized’. Idiomatic structures are often copied as loan-blends, involving Matter replication of a Macedonian noun, accompanied by a translation of the Macedonian verb: rutina alayim ‘I shall get into the habit’, lit. ‘I shall take a routine’, Macedonian da zemam rutina.
5. Other parts of speech A number of conjunctions and particles are borrowed from Macedonian and Albanian. Matras (2004) notes for the Turkish dialects of eastern Macedonia that the Macedonian additive conjunction i is regularly used when conjoining phrases, while Turkish ve is limited to conjoining constituents (as in example 3). Note that the adversative conjunction ama is identical in Turkish and Macedonian, Macedonian having borrowed it from Turkish. The Slavic contrastive-addition marker a indicates opposition between two phrases: (3)
İlk-okul-i ve orta-okul-i bitır-dı-m Türkçe first-school-acc and middle-school-acc finish-past-1sg Turkish dil-ın-de, a fakulted-i bitır-dı-m language-poss-loc and/however university-acc finish-past-1sg Makedonce dil-ın-de. Macedonian language-poss-loc ‘I finished primary and secondary school in Turkish, but university in Macedonian.’
Another use of a is for disjunction: (4)
Amerika a Alman yatırım-i dır. America or German investment-poss is ‘It is an American or German investment.’
It is possible that this function results from a blend between the Macedonian contrastive-additive a, and the Albanian-derived question particle a, which is also borrowed into GT:
Macedonian Turkish
(5)
219
A git-tı-n Stambol-a? q go-past-2sg Istanbul-dat ‘Have you been/ did you go to Istanbul?’
The Albanian requestive particle lu(te)m is also borrowed: (6)
Gel benım-le lum. come me-inst req ‘Please come with me.’
Subordinating conjunctions are mainly grammaticalized interrogatives and thus of Turkish origin, but the presence of kose ‘as if’ seems to indicate a contamination of Macedonian kako ‘as if’ and Albanian kinse ‘as if’, possibly reinforced by the similarity to the Turkish conditional verbal augment -se. Possibly, an Albanian model sepse ‘because’ is also behind the use of se as a subordinator of cause (‘because’).
6. Constituent order Although on the whole still an SOV language, flexibility of word order in Turkish is exploited in GT to extend pragmatically restricted variants to wider contexts, thereby increasing harmony between GT and its contact languages in the organization of utterance structures. Word order shift has acquired different degrees of stability with different constructions. In the possessive construction, the order head–modifier has become the preferred order in GT, mirroring the order in the Macedonian and Albanian constructions: (7)
a. Gostivar Turkish ruba-lar-i damad-ın clothes-pl-3sg.poss groom-gen b. Macedonian ališta-ta na zet-ot clothes-def to groom-def c. Albanian teshat e dhandrit clothes att groom
220
Yaron Matras and Şirin Tufan
d. Standard Turkish damad-ın eşya-lar-ı groom-gen clothes-pl-3sg.poss ‘the groom’s clothes’ The object of comparison is expressed in GT with the help of a preposition neka ‘like’, grammaticalized from the interrogative ne kadar ‘how much’, copying the Macedonian preposition kolku ‘as much’. It is positioned, as in Macedonian, between the attribute and the object of comparison: (8)
a. Gostivar Turkish güzel neka Meryem beautiful like Meryem b. Macedonian ubava kolku Merjem beautiful like Meryem c. Standard Turkish Merye kadar güzel Meryem as.much beautiful ‘as beautiful as Meryem’
This is the only obvious indication of a shift, in any construction, from the postpositional structure of Turkish, to prepositions. In verb phrases, the most stable case of word-order convergence with the neighbouring languages concerns the position of the copula. Whereas the Turkish copula is enclitic, GT tends to preserve a more conservative independent copula stem in i-, which, however, occupies the position between the subject and the predicate noun, as in the contact languages: (9)
a. Gostivar Turkish Sen (i)-sın küçük bir kış-çe. you cop-2sg small indef girl-dim b. Macedonian Ti si edno malo devoj-če. you cop.2sg indef small girl-dim c. Albanian Ti je nji vajz ë vogël. you cop.2sg indef small att girl
Macedonian Turkish
221
d. Standard Turkish Sen küçük bir kız-sın. you small indef girl2sg ‘You are a small girl.’ This is the general rule in the copula construction, irrespective of the word class or case of the predicate (e.g. adjective, locative noun, etc.): (10) Siz i-dı-nız ev-de. you cop-past-2pl house-loc ‘You were at home.’ In other constructions, deviation from verb-final order is much less pragmatically marked, and much more frequent, than in colloquial Tk., indicating a drive toward harmonization of the utterance planning procedures with those of the contact languages. Consider the following sentences, in which direct and indirect objects follow the verb without any inference of de-focusing or de-topicalization (which would be the reading accompanying such constructions in Tk.): (11) Ben gür-dü-m korkuli rüya. I see-past-1sg scary dream ‘I saw a scary dream.’ (12) Ben ver-dı-m bikate ekmek sizın dort tene I give-past-1sg little bread 2pl.poss four item beygir-ınız-e. horse-2pl.poss-dat ‘I gave your four horses some bread.’ (13) Onlar gid-ecek-ler dügün-e benım-le. they go-fut-3.pl wedding-dat me-inst ‘They will go with me to the wedding.’ The default position for objects that constitute new topical information in lexical predications remains, however, the pre-verbal position:
222
Yaron Matras and Şirin Tufan
(14) Parlament-ın-de var-dır iki dil. parliament-3sg-loc exist-cop.3sg two language Arnaut-lar Arnautçe konuş-ur. Albanian-pl Albanian speak-aor Makedon-lar Makedonce konuş-ur. Macedonian-pl Macedonian speak-aor Azınlık-lar Makedonce konuş-ur. minority-pl Macedonian speak-aor Bir tek Arnaut-lar Arnautçe konuş-ur. one only Albanian-pl Albanian speak-aor ‘In Parliament, there are two languages. The Albanians speak Albanian. The Macedonians speak Macedonian. The minorities speak Macedonian. Only the Albanians speak Albanian.’
7. Syntax Some of the most remarkable changes that have affected Rumelian Turkish – a characteristic feature of this group of Turkish dialects – is the adoption of clause combining strategies that are similar to those employed in the surrounding Indo-European languages. Essentially, these are based on the juxtaposition of finite clauses, linked through independent semantic markers that introduce the subordinate clause (subordinating conjunctions). This system replaces almost entirely the Turkic system of converbs and nominal embedding. Modal complements are not introduced by a conjunction, but make use of the historical optative, which, now expressing dependency on the main verb, serves as a subjunctive, with the complement clause generally following the main clause (see also Matras 1998, 2004): (15) a. Gostivar Turkish Yarın ist-er-ım oyna-(ya)-im dügün-de. tomorrow want-aor-1sg play-subj.1sg wedding-loc b. Macedonian Utre saka-m da igra-m na svadba-ta. tomorrow want-1sg comp play-1sg at wedding-def
Macedonian Turkish
223
c. Albanian Nesër dua të luj në darsëm. tomorrow want.1sg comp play.1sg in wedding d. Standard Turkish Yarın düğün-de oyna-mak isti-yor-um. tomorrow wedding-loc play-inf want-prog-1sg ‘I want to dance at the wedding tomorrow.’ The finite embedded predicate in the subjunctive replaces the historical Turkish infinitive. The same type of construction is used in manipulation clauses (modal complements with different subjects): (16) Daa çok sev-er-ım anlat-ır-sın kimse. more much like-aor-1.sg tell-aor-subj.3sg somebody ‘I prefer somebody to narrate it [to me].’ Factual or epistemic complements, which in Tk. may be expressed through either finite clauses, or nominalizations, always appear as postposed finite clauses, introduced by the subordinator ki, which is also common in Tk.: (17) Hised-ıl-mes ki vardır sonbaar. feel-pass-neg.aor comp exist.cop.3sg autumn ‘It does not feel like autum.’ In this manner, GT aligns itself with the other Balkan languages also in respect of the distinction between factual and non-factual complements. While the other languages have complements that specialize for factual/epistemic and non-factual/subjunctive (e.g. Macedonian deka vs. da, Greek oti vs. na, Bulgarian če vs. da, Romani kaj vs. te, and so on), in GT the opposition is expressed by using the inflected subjunctive on the verb in modal complements, and the ki complementizer (and indicative mood) in epistemic complements. Relative clauses also undergo re-structuring in Rumelian Turkish. Like the other Rumelian Turkish dialects, GT shows a relativizer ne, derived from the interrogative ‘what’, which mediates between the head noun and the finite, postposed relative clause (see Matras 1998, 2004). This replaces both the Turkish gerundial relative clause, and its finite counterpart in ki. The formation once again matches that of the principal contact language Macedonian, where the relativizer is equally derived from the interrogative ‘what’:
224
Yaron Matras and Şirin Tufan
(18) a. Gostivar Turkish O kış-çe ne gel-di biz-de şimdi yaşa-r that girl-dim rel come-past 1pl-loc now live-aor.3sg Stambol-da. Istanbul-loc b. Macedonian Devoj-če-to što dojde kaj nas sega živee vo İstanbul. girl-dim-def rel came at us now live.3sg in Istanbul c. Standard Turkish Biz-e gel-en kız şimdi İstanbul-da yaşı-yor. 1pl.dat come-ger girl now Istanbul-loc live-prog.3sg ‘The girl that came to (visit) us now lives in Istanbul.’ Like relative clauses, embedded clauses in GT are finite, usually postposed to the main clause, and introduced by an interrogative, functioning as a conjunction; Turkish-type nominalizations of embedded propositions are not found. Adverbial clauses show a mixed pattern in relation to convergence tendencies. One type of adverbial clause shows an overwhelming tendency to copy the Indo-European subordination type: postposed finite subordinate clauses introduced by a conjunctions. To this end, a series of grammaticalization processes take place giving rise to new subordinating conjunctions. The semantic relations involved in clause combinations of this type are those of time (introduced by açin ‘when’ in GT, or by ne zaman ‘when’ in other dialects of Macedonian Turkish), location (introduced by nerde ne ‘where’ < lit. ‘where what’, cf. Macedonian kade što lit. ‘where what’), reason (introduced by niçin ‘because’ < ‘what-for’, cf. Macedonian zošto lit. ‘for-what’), manner (introduced by kose ‘as if’, possibly a contamination of Macdeonian kako ‘how’, Albanian kinse ‘as if’, and Turkish -se ‘if’; see above), and comparison (introduced by neka ne ‘as much as’ < ne kadar ne ‘how much what’, cf. Macedonian kolku što). Purpose clauses and final clauses are equally finite, and show the verb of the subordinated clauses in the subjunctive. They are introduced respectively by the complementizer ki, directly reinforcing the subjunctive (cf. Macedonian prepositional reinforcer za da), and the conjunction çaki ‘until’. A second type of clause linkage remains largely unaffected by contactinduced restructuring. This involves conditional clauses (‘If I pass my exam my dad will buy me a bicycle’), and concessive clauses (‘Although I want to go to Antalya, I won’t be able to go’). Both are marked by the conditional marker -se (on its own for conditional clauses, with addition of de or hem ‘too’ for concessive clauses), which is added to a finite subordinated clause.
Macedonian Turkish
225
Thus, where Turkish already operates with finite subordinations, there appears to be no motivation to re-organize the structure of clause linking.
8. Lexicon Despite multilingualism in the region where GT is spoken, lexical borrowing is predominantly from Macedonian, reflecting the growing importance of the state language in the past two to three generations especially. Lexical borrowing from Macedonian naturally affects in the first instance semantic areas belonging to the public domain, such as names for institutions (teatar ‘theatre’, fakultet ‘faculty’, univerzitet ‘university’, vodovod ‘water board’, kanalizatsiya ‘infrastructure’, militsiya ‘police’, armiya ‘army’, ordinatsiya ‘dental surgery’, klinika ‘clinic’, autobuska stanitsa ‘bus station’), terms for practitioners and professionals (elektriçar ‘electrician’, stomatolog ‘dentist’, sestra ‘nurse’, apotekarka ‘pharmacist’, direktor ‘director’, student ‘student’, privatnik ‘having business in the private sector’), academic subjects and professions (meditsina ‘medicine’, farmatsiya ‘pharmacy’, stomatologiya ‘dentistry’, ispit ‘exam’, praktiçno ‘practical exam’, poen ‘mark’, matura ‘graduation’), construction and technology (kuyna ‘kitchen’, patos ‘flooring’, şifunyer ‘cabinet’, parno ‘central heating’, radiator ‘radiator’, garaja ‘garage’, satelitska ‘satellite’, kaseta ‘cassette’, elektrika ‘electricity’), as well as miscellaneous domains (maçka ‘cat’, şatka ‘duck’, sok ‘fruit juice’, zvuçnost ‘sound’, spetsiyalizatsiya ‘specialization’, tragediya ‘tragedy’, etc.).
9. Conclusion It is interesting to note once again that Turkish has only been a minority language in Macedonia for some three to four generations now. The fact that Matter borrowing is limited to a rather small number of discourse particles and conjunctions, may be a reflection of this recent retreat of Turkish from public life, and its replacement, to a considerable degree, by Macedonian. The lexicon, of course, reflects the recent dominance of Macedonian-speaking society in the public domain, employment, technology, and so on. Nevertheless, the restructuring of clause combining strategies based largely on a Macedonian model constitutes a radical departure from the Turkic syntactic type, and it is most certainly much older than the retreat of Turkish as the language of the public domain. Rather, the changes in this domain reflect
226
Yaron Matras and Şirin Tufan
century-old multilingualism. It appears that in daily communication, speakers were under pressure to organize complex utterances in a compatible way across the various languages that constituted their linguistic repertoire. What is essentially an economy-driven motivation – reducing multiple pattern types across the linguistic repertoire to just one – might be understood as a harmonization of utterance-organization strategies (Matras 2004). The two areas that are most obviously affected are clause combining strategies, and to a somewhat lesser extent, word order. With the former, it is the packaging of supplementary information through finite subordinations that prevails, and to this end a series of grammaticalization processes are triggered, exploiting elements of the inherited lexicon, often following the Macedonian model (‘replica grammaticalization’ in the terms of Heine and Kuteva 2005). The latter, word order, involves harmonization of strategies of mapping information status at the level of the linear organization of the utterance. Here, some constructions, such as possessive noun phrases and existential (copula) predications, appear more vulnerable to the pressure toward harmonization than others. Nevertheless, even word order in the basic verb phrase shows a partial relaxation of the pragmatic constraints on the appearance of postverbal objects. This in turn provides an extended scope to employ such constructions, which resemble the word order rules of the contact language. We may speculate that it was possible for pattern-replication of this type to emerge in the vernacular language long before Turkish retreated as the official language of the public sphere: it exploited constructions that pre-existed, to some extent at least, in colloquial usage, such as semi-embedded finite optative constructions (see discussion in Matras 1998 and 2004), or finite subordinations introduced with ki, or pragmatically-marked constructions involving de-topicalization of direct and indirect objects (in post-verbal position). Pattern replication was thus a kind of compromise, allowing speakers to maintain language loyalty while assisting the levelling of certain language processing strategies within the multilingual repertoire. We suggest that this latter aspect is a crucial component of the history of linguistic areas, for which the Balkans have long served as a prototype example.
Abbreviations abl acc aor
ablative accusative aorist
att cop dat
attributive marker copula dative
Macedonian Turkish def dim fut gen indef inf inst loc neg
definite article diminutive future genitive indefinite marker infinitive instrumental locative negation
past pl poss prog q rel req sg subj
227
past tense marker plural possessive progressive interrogative particle relative particle requestive singular subjunctive
Notes 1. Examples are taken from Tufan’s fieldwork in Gostivar; observations are based partly on fieldwork data collected by Matras among speakers from Stip. 2. Figures or even estimates of numbers of speakers in the entire region are difficult to obtain. Ethnic Turks in the Republic of Macedonia itself number around 70,000.
References Caferoğlu, Ahmet 1964 Anadolu ve Rumeli ağızları ünlü değişmeleri. TDAYB. 133. Friedman, Victor A. 2003 Turkish in Macedonia and Beyond. Wiesbaden: Harrassowitz Heine, Berndt, and Tania Kuteva 2005. Language Contact and Grammatical Change. Cambridge: Cambridge University Press. Matras, Yaron 1998 Convergent development, grammaticalization, and the problem of ‘mutual isomorphism’. In: Winfried Boeder, Christoph Schroeder, and Karl-Heinz Wagner (eds.) Sprache in Raum und Zeit, 89103. Tübingen: Narr. 2004 Layers of convergent syntax in Macedonian Turkish. Mediterranean Language Review 15: 6386. Németh, Gyula 1956. Zur Einteilung der Turkischen Mundarten Bulgariens. Sofia: Bulgarische Akademie der Wissenschaften. Tufan, Şirin 2007 Language convergence in Gostivar Turkish (Macedonia). PhD thesis, University of Manchester.
Grammatical borrowing in Kildin Saami Michael Rießler
1. General The present chapter deals with borrowings in Kildin-Saami.1 The principal contact language for Kildin is Russian, and has been so at least since the end of the Middle Ages. There are no grammatical borrowings detected from any languages other than Russian. However, many of the contact phenomena dealt with may also be traced in the other Kola Saami languages Akkala, Skolt, and Ter, which are facing the same contact-linguistic environment or have even been exposed to stronger assimilation pressure. But since Kildin is the best documented and the most accessible of the Kola Saami languages this investigation will be restricted to observations on this language. Most of the data used for this investigation have been taken from existing descriptions of Kildin. Examples given without reference come from my own field notes and have been cross-checked with native speakers. The orthographic representation of examples follows the standards set in the dictionary written by Kuruč, Afanasjeva, and Mečkina (1985).
1.1. Linguistic background Saami is a branch of the Uralic language family. All Saami languages are fairly similar in grammatical structure and lexicon. They form a dialect chain stretching from central and northern Scandinavia to the eastern tip of the Kola Peninsula. Kildin belongs to the group of East Saami languages. The other subgroups of Saami are Central Saami (spoken in the northern parts of Finland, Sweden and Norway) and South Saami (spoken in Central Scandinavia), each of which includes several languages. One characteristic of the phonology of most Saami languages is the occurrence of preaspirated voiceless stops and affricates [p, t, k, ¿, ʧ]. Negation in Saami is expressed by means of an inflected negation auxiliary followed by the non-finite main verb in a special connegative form. Phrase structure in Saami is for the most part head-final, including the predominant occurrence of postpositions instead of prepositions and strict head-finality in
230
Michael Rießler
noun phrases with noun, adjective, and pronoun modifiers. Relative clauses, however follow the noun they modify. In the verb phrase a shift from SOV to SVO word order seems to be taking place. The change in the order of verb and direct object as well as the introduction of prepositions and relative clause constructions are probably contact-induced. However, these changes go back to Common Saami tendencies and are subsequently not dealt with in the present investigation. One Kola Saami characteristic – as compared to the western Saami languages – is the relatively large consonant inventory, which is mostly due to the fact that almost all consonants have a phonologically distinct palatalized counterpart. As for the nasal and lateral dentals /n/ and /l/ there is an opposition not only to the respective palatalized phonemes, but also to the nasal and lateral palatals /ɲ/ and /ʎ/; consider the minimal triples mānn / maÃnÃ/ ‘moon; month’ – mannҍ /manÃ/ ‘egg’ – mann' /maɲÃ/ ‘daughter in law’ and pāll /paÃlÃ/ ‘ball’ – māll' /maÃlÃ/ ‘juice’ – māll'j /maÃʎÃ/ ‘rust’. The existence of a phonological opposition between palatal and palatalized consonants seems to be uncommon cross-linguistically (cf. Stadnik 2002: 31, elsewhere).
1.2. Sociolinguistics and geography Kildin is currently spoken on the Kola Peninsula in the northwestern-most part of the Russian Federation by no more than 700 people. The language is endangered due to language shift to Russian and is hardly ever heard in public life nowadays. Only elder Saami use their mother tongue in conversation with family members, relatives or friends. Among the younger generation, there is a strong decline in active language competence due to the lack of a vibrant speech community and the lack of any social motivation for learning and using Saami (an overview on the current socio-political situation of the Saami in Russia is given in Scheller 2006). The integration of the Kola Saami into the Russian Empire, their adaptation to Russian culture, and their conversion to Orthodox Christianity began as early as the fifteenth century. Nevertheless, despite the longstanding assimilation pressure, the territorial communities of the Saami were able to preserve their social, economic and cultural identity – at least in the central and northern areas – until the end of the nineteenth century. By the first half of the twentieth century, the Saami culture was on the verge of destruction. The tapping of mineral resources and the military armament of the region
Kildin Saami
231
were connected with an immense influx of manpower from Russia and other republics of the Soviet Union. The dissolution of the traditional Saami communities – as the result of forced integration of Saami reindeer herders into large new agricultural co-operatives – and the resettlement of the Saami for socio-political, economic and military reasons led to a dispersion of the original speech communities. The former compact Saami settlements and coherent local speech communities were replaced by mixed communities of Saami speaking different local varieties, together with non-Saami (above all Komi and Russians). As a result, within a few decades, the indigenous Saami people became a tiny, scattered minority without any great influence on political decisions (for an overview on Kola Saami history and further references see Kulonen, et al. 2005: 261265).
1.3. Contact induced change in Saami Along with the changes in Kola Saami culture and society, we witness a number of contact-induced linguistic changes. These changes – almost exclusively of Russian origin – concern nearly all domains of the grammar but are especially strong in the lexicon. Still, contact-induced change in Kola Saami has not yet been the subject of a systematic investigation. Certain contactinduced features of Russian origin are mentioned in works on Kola Saami, for example in Kert's (1971) grammar of Kildin. Russian influence on Kola Saami is also the subject of a published conference abstract by Klaus (1977) dealing with borrowed adverbs and the use of Russian numerals in Saami speech. The most considerable listing of contact-induced features in Kildin is found in a paper by Kert (1994). But even here it is mostly lexical borrowings which are dealt with. Only a few grammatical features are listed briefly by the author, such as borrowed phonemes in loanwords, the borrowed superlative particle, and borrowed function words (Kert 1994: 112). All these features are dealt with in more detail in the respective sections below. Another phonological feature which Kert believes to be borrowed from Russian is palatalization (Kert 1994: 111). This idea is shared by Stadnik (2002: 34, 165; without reference to Kert). Neither of the two authors, however, offer any explanations for the contact-linguistic mechanisms behind the proposed development. Instead of being caused by borrowing I find it much more reasonable to assume that Kildin Saami palatalization is triggered by another, probably language-internal development, namely the apocope of the reduced stem-final vowels.
232
Michael Rießler
2. Phonology Kildin has borrowed some phonemes along with Russian loanwords. In most cases, however, the phonological distinctiveness of these phonemes is weak since they only occur in loanwords and there are almost no real minimal pairs available. Consider, for example, the innovative voicing opposition in wordinitial plosives, sibilants, and the labio-dental fricative, i.e. (original) /p, t, k, ʃ, s, v/ – (innovative) /b, d, g, ʒ, z, f/, cf. purrk /purÃk/ ‘lower part of the reindeer's antlers’ – būrka /buÃrka/
3. Nominal structures 3.1. New domains of illative case In Kildin the illative case has expanded its domain to impersonal constructions such as in example (1). This use of illative is clearly influenced by the equivalent use of a dative construction in Russian (Szabó 1984: 3637). (1)
a. Russian Tebe ne nado znat′. 2sg.dat neg is_nessessary.3sg to_know b. Kildin Saami Tonnҍe e= be tīdtҍe. 2sg:ill.sg neg is_nessessary.3sg to_know ‘You don‘t need to know.’
The equivalent construction in other Saami languages would normally be one with a main verb and the subject in nominative case (e.g. NSa dárbbahit ‘to need, to have use of’). However, this use of illative in Kildin is probably not a very recent innovation since similar examples are found in all Kola Saami
Kildin Saami
233
languages (according to Itkonen's (1958) dialect dictionary of the Kola Saami languages). Another, though archaic, impersonal verb in Kildin, given by Itkonen (1958: 33) as glk (~gɛlk, galk) 3sg.prs ‘need, shall’ (cf. NSa galgat inf ‘to shall, to have to’) requires either illative or genitive. Since similar impersonal constructions – with either genitive or partitive – occur in Finnish it cannot be ruled out that their use in the Kola Saami languages was originally calqued, based on the Finnish model. Russian influence could than have caused the replacement of an original genitive or partitive (depending on the verb) in these impersonal constructions. Note also that the impersonal verbs bedt and ebe (e=be ← ejj bedt [neg:3sg.prs is_nessessary]) are borrowed from Finnish (or Karelian) pitää ‘to need’ which is used in a construction with a partitive subject. In a construction expressing the age of someone or something in years, illative case is obligatory as well (2a). This expression is also clearly modeled on the analogous usage of dative in Russian (2b) and is not equivalent to the construction used in other Saami languages, cf. the North Saami example (2c). (2)
a. Kildin Saami mɨnnҍe l'ēv kōll′m gke. 1sg:ill be.3pl three year-part b. Russian mne tri god-a. 1sg:dat three year(m)-gen.m.sg lit. ‘to me are three years.’ c. North Saami mun lea-n golbma jagi. 1sg be-1sg:prs three year\gen.sg ‘I am 3 years old.’
3.2. Diminutives and augmentatives A case of pattern-borrowing in the nominal morphology of Kildin can be found in the adoption of a second diminutive form. The two diminutives of Kildin are formed by means of the suffixes -a and -enč. Whereas the first diminutive suffix (3b) simply marks diminution (without expressing the speaker's attitude to the noun in question), the second diminutive (3c) has rather expressive semantics and could be characterized as a complimentary form.
234
Michael Rießler
(3)
Diminutive and complimentary in Kildin a. koabp' ‘ditch’ b. koab'-a ‘small ditch’ ditch-dim1 c. koab'-enč ‘dimple’ ditch-dim2
The model of the differentiated diminutive seems to have been borrowed from Russian, where graded diminutive forms are quite frequent. The system of graded diminutive (and augmentative) forms with single and combined suffixes is on the whole very typical of Russian as well as of other Slavic languages. In (3b) the feminine diminutive suffix -ka is attached to the noun jáma ‘ditch’ resulting in a form with the “simple” diminutive reading ‘small ditch’. (3c) constitutes a diminutive form as well, although with the rather expressive meaning ‘nice small ditch, dimple’. (4)
Diminutive and complimentary in Russian a. jáma ‘ditch’ ditch(f) b. jám-ka ‘small ditch’ ditch-dim1:f c. jám-oč-ka ‘nice small ditch; dimple’ (complimentary) ditch-dim2-dim1:f
The existence of a second diminutive (with complimentary semantics) in Kildin is clearly innovative since the other Saami languages only have the one simple diminutive form. The complementary (=dim2) suffix -(e)nč of Kildin goes back to the Proto-Saami diminutive suffix *[-ńʒ´e] as does the diminutive suffix -š / -ž(a)- of Central and South Saami (Korhonen 1981: 320), cf. NSa mánná ‘child’, mánáš ‘little child’, mánážat ‘little children’. The Kildin Saami innovative diminutive suffix -a, following the weak consonant stem, is identical to the weak flexional stem of the original diminutive suffix -enč, occurring, e.g., in the genitive or locative of the diminutive noun (koab'-enč ‘dimple’ – koab'-a [ditch-compl\gen:sg] ‘of the dimple’ – koab'a-s't [ditch-compl-loc:sg] ‘in the dimple’). The diminutive koab'a dim ‘small ditch’ is thus homophone with the genitive of the second diminutive koab'a compl\gen:sg ‘of the dimple’. Obviously speakers of Kildin generalized the genitive (or weak stem) form of the original diminutive noun as the unmarked diminutive form at some point. At the same time the meaning of the original
Kildin Saami
235
diminutive form of -enč has been shifted to that of a complementary form. Russian influence did not necessarily trigger the first step in the development of the new diminutive form. On the contrary, giving up the original diminutive suffix -enč in favor of -a might just be the result of the overall tendency towards phonological reduction and loss of suffix codas in Kildin. Historically, -a does not belong to the suffix but is a reflex of the original second syllable vowel, cf. NSa niibi ‘knife’ – niipa-š dim and KSa njjp ‘knife’ – njp-a dim – njp-enč compl. The archaic diminutive suffix -enč and the innovative suffix -a had probably been in free variation for a while (and might still be in the speech of some Kildin speakers) before the first one acquired a secondary meaning. But in any case, the reinterpretation of the original diminutive as complementary must be the result of Russian influence. In Russian (and other Slavic languages), it is not only diminutives which have graded semantics, but augmentatives do as well. (5)
Gradated augmentatives in Russian a. dóm ‘house’ house b. dóm-išče ‘large house’ house-aug c. dóm-iško ‘worthless/bad (large) house’ (pejorative) house-aug.dim
Kildin performs analogously here as well. Kildin has an augmentative suffix, although I find that derived augmentative nouns are used less frequently than diminutives in Kildin. The most common augmentative suffix in Kildin is -p'ihk /-pikÃ/ [-pik]. A second augmentative suffix with a more expressive, pejorative meaning is -p'agka /-pagÃa/. End'ukovskij (1937: 139) also gives pejorative forms on -p'ikenč ?/-pigenʧ/, which is obviously a combination of an augmentative and a diminutive suffix. My language consultants do not use this form. However, I find it tempting to believe that -p'agka is a secondary development out of an original -pikenč and I suggest the following development of the suffix (exemplified by the noun pēr̥̥r t ‘house’): (6)
a. aug pēr̥t-p'ihk [house-aug] b. pejor1 pēr̥t-p'igk-enč [house-aug-dim1] c. pejor2 pēr̥t-p'agk-a [house-aug-dim2]
236
Michael Rießler
The weak consonant stem in (5a) pēr̥t- (nom.sg pēr̥̥rt) is triggered by the suffixation of aug. Regular stem alternation rules would also account for the suffixal consonant gradation in (5b) -p'igk (← -p'ihk) as well as for the suffixal umlaut caused by -a dim2 in (5c) -p'agk (← -p'igk). The etymology of the augmentative suffix -p'ihk, however, remains obscure. There is no augmentative in western Saami languages or even in Fennic, and the Slavic augmentative suffixes have different shapes and thus cannot be the source either. Saami speakers have suggested to me that the second pejorative suffix may have been borrowed from the Russian word b'aka, which means ‘excrement’ (‘doo-doo’) in child language. I find this rather unlikely. But even if b'aka has been borrowed from Russian, this would not explain the origin of -p'ihk. In Komi there is a noun pik meaning ‘trouble, misfortune’ which can also be used as an interjection with the meaning ‘Tough luck! Too bad!’ (Lytkin 1961: 537538), cf. Ru bedá with the same meaning). Another possible source (according to Jurij Kusmenko p.c.) could be Scandinavian bäck ‘tar’. Even though the lexical source of the suffix remains unknown, the existence of an augmentative in Kildin and its derivation into a pejorative by means of extending it with a diminutive suffix is obviously due to Russian influence.
4. Verbal structures: analytic future tense Kildin seems to be grammaticalizing an analytical future tense form with the auxiliary all'ke ‘to start, to become’. Future tense is not known to occur in other Saami languages. (7)
Analytic future tense in Kildin (Kuruč 1985: 558) mjj all'k'-ep lōgk-e. 1pl fut-1pl read-inf ‘We will read.’
The construction in Kildin is obviously modeled on the Russian future tense with the future auxiliary budet'. The latter construction is clearly grammaticalized since the verb budet' has no lexical meaning. (8)
Analytic future tense in Russian my bud-em čita-t'. 1pl fut-1pl read-inf ‘We will read.’
Kildin Saami
237
According to Kuruč (1985: 558) all'ke is used as future auxiliary without any lexical meaning. This is true at least for the speech of younger Kildin Saami, who according to my own observations tend to use all'ke as future marker similar to the Russian construction with budet'. However, the lexical use of all'ke is also attested, as in the following example taken from а fairy tale recorded in 1975 (Sammallahti 1998: 148).3 (9) Sjj ell'k-enҍ kānnҍc kānҍc-es' =že šoabš-e. 3pl start-3pl.prt friend friend\gen-poss:3sg top love-inf ‘They fell in love (with each other).’
5. Other parts of speech 5.1. Pronouns: Negative indefinites In the Saami languages negative indefinites are usually derived from interrogatives by means of a suffix (or particle), cf. NSa guhte-ge [who-neg] ‘nobody’, gosa-ge [where-neg] ‘nowhere’, etc.4 Kildin exhibits a different construction with a negation prefix, borrowed from the Russian negative particle ni (~ né); consider the examples in (10). (10) Negative indefinites with ni in Russian and Kildin Ru
KSa
gloss
translation
a. niktó ni-k’ē neg-who? ‘nobody’ b. nigdé ni-kas’t neg-where? ‘nowhere’ c. ničtó ni-mī neg-what? ‘nothing’ The marker ni- is productively used with all interrogatives in Kildin, which is obviously the result of calquing the Russian model. Note that the Kildin interrogative pronouns are all inherited from at least Proto-Saami. Exactly as in the Russian source model, ni- attaches productively to interrogatives not only in nominative but in other cases as well, for example ni-mēnn [neg-what. acc:sg] ‘nothing(acc)’, ni-k'ējn [neg-who.com:sg] ‘with nobody’, ni-k'ēnn [neg-who.gen:sg] ‘nobody's’, etc. Finally, the double-negated construction in Kildin appears to be calqued from Russian as well.
238
Michael Rießler
(11) Kildin Saami ni-mēnn munn emm ujn neg-what.acc:sg 1sg neg:1sg.prs see.conn:prs Russian ničego ja ne vižu neg.what.acc:sg 1sg neg see.1sg:prs ‘I don't see anything.’ Note that ni in Russian is not a prefix since it can be separated from the interrogative when the latter is used with prepositional interrogatives, as in Ru ni s kém [neg with instr] ‘with nobody, with no one’. In Kildin, however nothing can be inserted between ni- and the interrogative. The obligatory boundedness of ni- to its host and its productive use (even though restricted to the closed class of interrogatives) suggests that ni- is a prefix in Kildin. This marker thus constitutes the only prefix in this otherwise exclusively suffixing language.
5.2. Particles and discourse markers Lexical borrowing may go hand in hand with the introduction of grammatical structures. This is obviously the case with the discourse-pragmatic structuring of conversation and narration in Kildin which looks quite similar to that of Russian due to the overall borrowing of Russian discourse markers like vot ‘well’, nu vot ‘now then’, tak ‘so’, ved' ‘you know’, etc. Another example of borrowing can be seen in the Russian topic marker =že (~ =še) ‘but; you know, you see; also’. (12) Ovvan=že ‘Ivan, you know’ Ovvan top The use of =že in Kola Saami, however is clearly not a very recent innovation since it was already mentioned in the Kola Saami grammar of Halász (1883: 40). Analogue to Russian, the enclitic =že in Kildin also occurs as part of the (lexicalized) adverb ndtše ‘also’, cf. Ru tože (← tot že ‘the same’). The first component of the adverb in Kildin, however, was inherited at the latest from Proto-Saami (cf. NSa na ‘so, well’ and ná ‘so, this way’). All discourse markers mentioned above have functions similar to those in the source language. However, these have replaced original Saami markers
Kildin Saami
239
rather than introducing a new structure of text ordering. A case of introducing a new discourse-pragmatic model along with the borrowed marker is found in the discursive use of the conjunction a ‘and, but’
5.3. Connectors Several Russian borrowings have replaced Saami connectors (which in their turn are often old borrowings from Baltofinnic into Common-Saami or ProtoSaami), cf. jesli (~ jesle) ‘if’
5.4. Adjectives: Analytic superlative Comparison in Kildin is usually expressed by means of the suffixes -a comp and -mus sup; consider the adjective nūrr ‘young’ – nūr-a comp – nūr-a-mus comp-sup.5 But the superlative of adjectives can also be formed analytically with the borrowed particle same
240
Michael Rießler
gender nor do attributive adjectives agree in case, the superlative marker occurs as an uninflected particle. (15) a. Synthetic superlative in Kildin kugk-a-mus čuekas long-comp-sup road ‘the longest road’ b. Analytic superlative in Kildin same kugk'-es' čuekas sup long-attr road ‘the longest road’ c. Double marked superlative in Kildin same kugk-a-mus čuekas sup long-comp-sup road ‘the longest road’ Whereas the innovative analytic superlative in (15b) seems to be used without a difference in meaning as compared to the original synthetic superlative in (15a), the double superlative construction in (15c) is used to further emphasize the superlative notion ‘longest’.
6. Lexical borrowings Kildin Saami has borrowed a large amount of Russian vocabulary. Nouns, verbs, and adjectives have all been borrowed. By far the largest part of loanwords from Russian belong to modern world concepts; however, Russian loanwords are found in all other semantic domains as well. There are also a fair number of function words among the recent lexical borrowings of Russian origin. Examples for borrowed function words include the adverb toal'ke /tɒlke/ [tɒlkə] (~ [tɔkə])
Kildin Saami
241
adopted without any phonological adoptation. The word for ‘car’, mašína (
7. Conclusions Contact with Russian has resulted in a few grammatical changes in Kildin Saami. Changes concerning the borrowing of actual linguistic matter (or MAT-borrowing) are mostly found at the level of discourse-pragmatic text structuring (discourse markers, coordinators, subordinators). Interestingly, almost all of the examples of borrowed function words are also found in other Uralic contact languages of Russian (cf. Majtinskaja 19781979). Apart from that, structural changes due to MAT-borrowing can found in the replacement of an original synthetic construction by a new analytic construction in adjectival morphology (the superlative particle) and, for nominal morphology, in the replacement of an original negative suffix by a borrowed negative prefix on interrogatives. Other changes in verbal and noun morphology, such as the grammaticalization of an analytic future tense and the introduction of secondary diminutive and augmentative forms are rather due to pattern replication (or PATborrowing), i.e. due to the adaptation of similar models from Russian. The core areas of Kildin morphology and phonology seem to be relatively resistant to MAT-borrowings: there are no borrowings of synthetic case, tense, and person markers, or of demonstratives or other pronouns. Furthermore, the distribution of new phonemes is clearly restricted to the loanwords with which they where borrowed.
242
Michael Rießler
Abbreviations acc attr aug com comp compl conn dat dim f fut gen ill inf instr KSa loc
accusative attributive augmentative comitative comparative complimentary connegative dative diminutive feminine future genitive illative infinitive instrumental Kildin Saami locative
m n neg nom NSa part pejor pl poss prs prt PSa Ru sg sup top
masculine neuter negative nominative North Saami partitive pejorative plural possessive present preterite Proto-Saami Russian singular superlative topic
Notes 1. I would like to thank my colleagues K. Hildebrandt, K. Kotcheva, J. Kusmenko, J. Wilbur as well as the editors of this volume for their helpful comments. The usual disclaimers apply. My fieldwork on Kola Saami was supported by the Gesellschaft für bedrohte Sprachen e.V. and the VolkswagenStiftung. 2. Note, however, how the preaspirated articulation of the voiceless stop in floht [flotÃ] is a clear example of the phonological extension adaptation of the Russian loanword according to the Saami phonological rules. 3. The stem alternation (e ← a) in the example (9) is the result of regular Umlaut. 4. The gloss neg might, however be questionable since the morpheme can indeed be added to other semantic word classes in North Saami and expresses some kind of emphasis rather than negation in most constructions where it occurs. 5. Note that the iconic comparative marking with sup attaching to comp is the result of the relatively recent apocope of the original comparative suffix -mp together with the shift of the original word stem boundary. The comparative suffix -a is part of the stem historically, cf. NSa nuorra ‘young’ – nuora-t comp – nuora-mus sup. The comparative nūr-amp ‘younger’ can still be found in older text collections and descriptions of Kola Saami languages but is now regarded as archaic by speakers of Kildin.
Kildin Saami
243
References End'ukovskij, A. G. 1937 Saamskij (loparskij) jazyk [The Saami (Lappish) language]. (Jazyki i pis'mennosti narodov Severa). Moscow-Leningrad. Halász, Ignácz 1883 Orosz-lapp nyelvtani vázlat [An outline of Russian–Lappish grammar]. Nyelvtudományi közlemények 17: 145. Itkonen, Toivo Immanuel 1958 Koltan- ja kuolanlapin sanakirja = Wörterbuch des Kolta- und Kolalappischen. (Lexica Societatis Fenno-Ugricae 15.) Helsinki: Suomalais-ugrilainen seura. Kert, Georgij Martynovič 1971 Saamskij jazyk (kil'dinskij dialekt): Fonetika, morfologija, sintaksis [The Saami language (Kildin dialect): Phonetics, Morphology, Syntax]. Leningrad: Nauka. 1994 Saamsko-russkie jazykovye kontakty [Saami–Russian linguistic contacts]. In: Pjotr Mefodievič Zajkov (ed.), Pribaltijsko-finskoe jazykoznanie, 99116. Petrozavodsk: Rossijskaja Akademija Nauk. Klaus, Väino 1977 O nekotoryx javlenijax vlijanija russkogo jazyka na saamskij jazyk [On some effects of Russian influence in Saami]. In: Issledovanie finno-ugorskix jazykov i literatur v ix vzaimosv'azax s jazykami i literaturami SSSR. Tezisy dokladov Vsesojuznogo Naučnogo Soveščanija Finnougrovedov, 27–30 okt. 1977 g., 33. Užgorod. Korhonen, Mikko 1981 Johdatus lapin kielen historiaan [Introduction to the history of the Lappish language]. (Suomalaisen Kirjallisuuden seuran toimituksia 370.) Helsinki: Suomalaisen Kirjallisuuden seuran. Kulonen, Ulla-Maija, Irja Seurujärvi-Kari, Risto Pulkkinen, and Johanna Roto 2005 The Saami: A Cultural Encyclopaedia. (Suomalaisen Kirjallisuuden seuran toimituksia 925.) Helsinki: Suomalaisen Kirjallisuuden seura. Kuruč, Rimma Dmitrijevna 1985 Kratkij grammatičeskij očerk saamskogo jazyka [A short grammatical sketch of Saami]. In: Rimma Dmitrijevna Kuruč, Nina Jelisejevna Afanasjeva, and Jekatarina Ivanovna Mečkina (eds.), Saamsko-russkij slovar' = Saam'-rūšš soagknehk', 529567. Moskva: Russkij Jazyk. Kuruč, Rimma Dmitrijevna, Nina Jelisejevna Afanasjeva, and Jekatarina Ivanovna Mečkina 1985 Saamsko-russkij slovar' = Sа¯аm'-ruušš soagknehk' [Saami-Russian dictionary]. Moskva: Russkij Jazyk.
244
Michael Rießler
Lytkin, V. I. 1961 Komia-roča slovar’ = Komi-russkij slovar’ [Komi-Russian dictionary]. Moscow: Gosudarstvennoje izdatel’stvo inostrannyx i nacional’nyx slovarej. Majtinskaja, Klara J. 19789 Zaimstvovannye elementy, ispol'zuemye v finno-ugorskix jazykax pri obrazovanii form naklonenij [Borrowed elements, used in inflectional forms in Finno-Ugric languages]. Etudes finno-ougriennes 15: 227– 231. Sammallahti, Pekka 1998 The Saami languages:. An introduction. Kárášjohka: Davvi Girji. Scheller, Elisabeth 2006 Die Sprachsituation der Saami in Russland. In: Antje Hornscheidt, Kristina Kotcheva, Tomas Milosch, and Michael Rießler (eds.), Grenzgänger: Festschrift zum 65. Geburtstag von Jurij Kusmenko, 280290. (Berliner Beiträge zur Skandinavistik 9.) Berlin: Nordeuropa-Institut der Humboldt-Universität. Stadnik, Elena 2002 Die Palatalisierung in den Sprachen Europas und Asiens. Eine arealtypologische Untersuchung. (Tübinger Beiträge zur Linguistik 461.) Tübingen: Narr. Szabó, László 1984 The Function of the Inessive–Elative and the Dativ–Illative in Kola Lappisch. Nordlyd. Tromsø University Working Papers on Language and Linguistics 8: 452.
Grammatical borrowing in Yiddish Gertrud Reershemius
1. Background The history and development of Yiddish, a west-Germanic language, is intertwined with language contact from its very beginning. Three layers of contact, historical, recent and current, can be distinguished.
1.1. Historical contact The Yiddish language originates from medieval times and developed through contact: Jewish speakers of Old High German and later Middle High German varieties enriched these vernaculars with a component, mainly in the lexicon, from Hebrew and Aramaic. These languages of the scriptures and religious practice served as written and high varieties in a situation of internal Diglossia. The spoken language was written mainly for informal purposes, such as private letters, memoirs, notes and entertaining or devotional literature, addressed to those who were unable to read and write in Hebrew, e.g. women or those who could not afford an extended education. It is not known whether in early stages the vernacular was identical with the varieties spoken by Non-Jews: a majority of scholars in the field of historical Yiddish linguistics assume that Yiddish developed on the substrate of an older spoken Jewish language, either Aramaic or a Romance-based variety (Jacobs 2005: 956).1 In their view the language of Ashkenazic Jews has always been distinct from the varieties spoken around it, and it clearly is the case that the Germanic component in Yiddish developed into a form distinct from German varieties (cf. Timm 2005 for the lexicon). Written sources of the language exist from the late thirteenth century. These sources are easily recognizable as Jewish since they were written in Hebrew letters. From the beginning of the sixteenth-century sources prove Yiddish definitely to be distinct from German varieties, in lexicon and phonology (Timm 1987).
246
Gertrud Reershemius
1.2. Recent contact Yiddish in the West (Western Yiddish) remained in contact with spoken German varieties and the slowly evolving German standard language. However, migration of Jews to Slavic-speaking areas in Eastern Europe due to persecution in the West ever since the First Crusade formed Eastern Yiddish through language contact with Slavic languages and the addition of a Slavic component, especially to the lexicon, phonology and lexical derivation. Eastern Yiddish thrived from the eighteenth century onwards, and became a fully-fledged language able to cover all oral and written linguistic domains of modern life and to contribute to world-class literature. Western Yiddish, however, was abandoned by its speakers during the nineteenth century in most parts of the Western Yiddish language area apart from some remote provinces in the Southwest and in the Northwest of the German- and Dutch-speaking countries, where remnants of a Yiddish vernacular could be found well into the twentieth century in small rural Jewish communities (e.g. Lowenstein 1969). Currently, the term ‘Yiddish’ refers to Eastern Yiddish, a language spoken by more than 10 million speakers until the Second World War.2 Recent language contact refers to the formative contact with Slavic and, to a certain extent, with Baltic languages, as well as to contact with both Standard German and spoken German varieties which took place over the centuries and which influenced both languages. Yiddish–German contact is an extremely complex issue due to the languages’ close structural relatedness and the various possible modes of contact. Therefore, an analysis cannot be attempted in the framework of this chapter.
1.3. Current contact Current contact involves Yiddish and a number of languages such as English, Israeli Hebrew, Spanish, and Flemish, since the beginning of the twentieth century. After the Holocaust – approximately 5 million of its victims were speakers of Yiddish – and anti-semitic persecution in Eastern Europe during the 1950s and 1960s, a large majority of surviving Eastern European Jews emigrated to the United States, Israel, South America or Western Europe. Most emigrants did not retain their language, although in some places Yiddish-speaking enclaves developed, consisting of strictly orthodox Jewish communities, which preserved the diglossic set-up of traditional Jewish society to a certain extent. Currently, the number of Yiddish-speakers seems to
Yiddish
247
be growing again and is estimated to be around a million worldwide (Jacobs 2005: 3). More precise figures are not available which is mainly due to the orthodox communities’ hesitance to take part in surveys. The present chapter must restrict itself to recent contact due to the fact that sociolinguistic data and analyses of the highly diverse framework of current contact is not yet at a level to draw conclusions (cf. Isaacs 1999a, 1999b). The chapter therefore focuses on contact with Slavic languages, which led to the acquisition of the Slavic component in Yiddish.3
2. Phonology The phoneme inventory of Yiddish has been influenced by recent language contact, although apart from a general tendency for palatal dental plosives to become fricatives (/c/ + /dz/ and /č/ + /dž/) the changes are usually restricted to certain geographic parts of the historic Yiddish language area: a. In North-Eastern Yiddish, new phonemes evolved: /t’/, /d’/, /s’/, /n’/, /l’/. b. In the east of Ukraine /h/ was replaced by /ɦ/, whereas it was dropped in the west. In some areas of Ukraine /h/ and /ɦ/ form oppositions. c. /l/ changed generally to /ł/, in eastern areas to /w/. Voicing might have undergone contact-induced change, as Weinreich describes: the transformation of the consonantal tense/lax opposition into a voiceless/ voiced one. … In Yiddish, the Slavic environment has had a still further impact: it was probably western Ukrainian … and Southern Belorussian … that served as a model for the distinguishing of voice even at the end of words. … final distinctive voice in Yiddish is found further north and east than in the coterritorial Slavic and Baltic languages, and the interaction of internal Yiddish structural causes with Slavic influences must be considered. (Weinreich 1958: 374)
As for vowels, in Belorussia /o/ changed to /a/ in pretone position, as for example in xalile–xolile ‘God forbid’ (Weinreich 1958: 372). Although the distinction between long and short vowels has been lost in most dialects of Yiddish, it is not considered to be due to contact with Slavic Languages (Weinreich 1958: 374). Two further areas of possible influence by recent language contact might be stress and consonantal clusters. A number of non-phonemic secondary stresses have been eliminated, and initial consonantal clustering, as for ex-
248
Gertrud Reershemius
ample in Dlile ‘Delilah’ or pgam ‘defect’, has been extended under the influence of Slavic languages (Weinreich 1958: 377).
3. Nominal structures The following nominal structures may have evolved or been influenced due to contact: the possessive, word order of the predicate, gender, diminutives and derivational suffixes.
3.1. Possessive The genitive case has disappeared in Yiddish. Instead, a periphrastic construction with the preposition fun ‘of’ is used: dos tixl fun der mame ‘mother’s headscarf’. At the same time, a possessive form was retained for persons, e.g. majn bobes bobe ‘my grandmother’s grandmother’, in which the Germanic neuter and masculine genitive suffix -s was also applied to feminine nouns. In some cases the definite article is added in the dative: dem suns froj ‘the son’s wife’.4 Lötzsch (1974: 451452) claims that the development of the Yiddish possessive could have been influenced by the flexibility of word order in the Slavic languages, particularly Russian. He admits, however, a number of differences between the Yiddish and the Russian possessive constructions.
3.2. Word order of the predicate An inflected adjective in the predicate conveys a distinct semantic meaning: ix bin krank ‘I am sick (now)’ but ix bin a kranker ‘I am a sick man’ (Weinreich 1958: 383). Weinreich claims that this distinction follows the example of long and short forms of adjectives in Russian. Another feature concerns the positioning of the adjectives after the noun they determine, following the example of Slavic languages: “In Yiddish adjectives may be placed after the noun they determine, usually with an affective connotation: dos land dos farbotene ‘the forbidden country’, more solemn than dos farbotene land; di toxter majne ‘my daughter’, with a touch of disparagement absent in majn toxter.” (Weinreich 1958: 382).
Yiddish
249
3.3. Gender Northeastern Yiddish lost the neuter gender, as did two of its contact languages, Lithuanian and Latvian (see Eggers 1998: 346347). In the case of internationalisms, borrowings from more than one language usually in the semantic areas of academic, scientific or technological discourse, gender is applied in accordance with Slavic languages, e.g. univerzitet (m) ‘university’, komitet (m) ‘committee’, zignal (m)‘signal’.
3.4. Diminutives Diminutive suffixes -inke, -enyu, -ke, as for example in muminke ‘dear/little aunt’, tatenyu ‘dear/little father’ or Avromke ‘dear/little Abraham’ (here convergence with a Germanic-component suffix -ke) are borrowed from Slavic languages. (see Weinreich 1980: 531) Yiddish has two degrees of diminutives in most nouns (-l and -ele, cf. hun ‘hen’ – hindl ‘hen / little hen’ – hindele ‘little hen’), which follows the example of Slavic languages, e.g. Polish. Although the suffixes used are Germanic, the German language(s) do not have two degrees of diminutives. Furthermore, the NortheasternYiddish dialect, which does not show the neuter gender any more, has “adopted the un-Germanic but Slavic rule that diminutives have the same gender as the base form.” (Weinreich 1958: 380)
3.5. Derivational suffixes A number of affixes were borrowed from Slavic languages:5 -še and -ke to create feminine nouns, e.g. gubernatorše ‘the governor’s wife’ or lererke ‘teacher’ -ák, -l’ák, -ńák, -áč to create pejorative forms of certain nouns, e.g. cvujak ‘hypocrite’; paskudniak ‘vicious person’ or jungač ‘thug’ -ák to describe somebody’s origin, e.g. litvak ‘someone from Lithuania’ -éc to indicate strength, e.g. boxeréc ‘strong bloke’ -čik, -eši to create endearment forms, e.g. Avromčik ‘dear Abraham’; mameši ‘dear mother’ -(e)ńu, -inke to create endearment forms in the vocative (see under 3.4)
250
Gertrud Reershemius
-ixe to create nouns for female animals; e.g. lejbixe ‘lioness’ -arńe to name localities, e.g. xasid-arńe ‘gathering place of khassidim’ -ńik (m) and -ńice (f) as personifiers, e.g. tšajnik ‘teapot’ or ejšesišnice ‘adulteress’ -úk to create pejorative expressions for professions, e.g. šusteruk ‘shoemaker’(pej.)
4. Verbal structures Yiddish makes wide use of its set of Germanic prefixes and applies them, following the example of Slavic verbs as loanblends – e.g. farčepen ‘to provoke’, where the stem is from Slavic origin and the prefix Germanic – or loan translations – e.g. opbrengn ‘to bring back’ which is modelled on Polish odniešč – or “… in verbs where the Yiddish prefix is applied productively without reference to a Slavic model.” (Weinreich 1958: 381). Loan-verbs are integrated directly into the morphological marking of Yiddish verbs. The verb is usually borrowed in the unmarked inflected form. The following structures have been influenced by contact with Slavic languages: Aspect/Aktionsart, modal particles and verbal derivational patterns.
4.1. Aspect/Aktionsart The significant development of lexical (prefixation) as well as grammatical aspect in Yiddish (semelfactive, e.g. a kuk ton – ‘to have a quick look’, and iterative, e.g. ix flejk gejn – ‘I (always) go’) is very likely to have been triggered by contact with Slavic languages and their complex inventory of grammatical aspect. It does not copy, however, the Slavic structures but “represents a further development of a Germanic system” (see for a detailed discussion Aronson 1985: 185; Albert and Meijering 2001).
4.2. Modal particles The following adverbs with mainly modal functions were borrowed from Slavic languages: “e.g. až to emphasize great quantity, het to emphasize distance, na to accompany giving, ot to accompany pointing, take ‘indeed’, xoč(be) at least (cf. Polish choćby), jakoš ‘somehow’, jakbe ‘as if’ (cf. jakby).
Yiddish
251
Note also the peculiar adverbs male (+interrogative) ‘no matter (what, who, etc.)’ and same (+superlative) ‘very’ (e.g. der same grester ‘the very biggest’).” (Weinreich 1958: 390). As for verbal derivational patterns, the Slavic affix -ke can function as a verbalizer of interjections, e.g. bom-ke-n ‘to say “bom”’ (Weinreich 1958: 378).
5. Other parts of speech In the areas of pronouns, particles and discourse markers, and adjectives and adverbs, Yiddish shows signs of contact with Slavic languages.
5.1. Pronouns The deictic particle ot is borrowed from Slavic and serves as a demonstrative intensifier in Yiddish in combination with the der, di, dos-paradigm: e.g. ot der jingl ‘this very boy’. The indefinite pronoun abi is borrowed from Slavic: (1)
Zej hobn do tsu ton nit mit abi-vemen. you have here to do.inf not with pron-who.dat ‘You are not dealing with anybody here’. The indefinite pronoun ljade is borrowed from Slavic:
(2)
Di eltere bojes zajnen ojx grejt a ljade tog tsu antlojfn. the older boys are also ready a pron day to run off.inf ‘The older boys are ready to run off any day.’
The construction -(s’)nit -iz follows a similar construction in Slavic languages: (3)
Zi hot gor kejn kojex nit fartraxtn zix she has intensif no strength not think.inf refl iber velxe-nit-iz injonim.6 about which-not-be.3sg issues. ‘She hasn’t got the strength to think about any issues.’
252
Gertrud Reershemius
Interrogative pronouns: the construction ver ... ver is formed according to the example of Russian: (4)
Guthartsike balebostes hobn zi ba zix gehaltn, kindhearted housewifes have her with refl keepparticiple ver a xojdeš, ver lenger.7 pron a month, pron longer. ‘Kindhearted housewifes kept her with them, some for a month, some for longer.’
The intensifying enclitic -že ‘then’ after the interrogative pronoun vos, e.g. vos-že, is borrowed from Slavic: (5)
Vos-že est du? what-intensif eat:2sg you? ‘What is it you are eating?’
An increased use of reflexivity can be observed in Yiddish for two reasons: (a) a verb which etymologically was not reflexive, acquires reflexivity because the Slavic equivalent is reflexive, e.g. endikn zix ‘to come to an end’ from pol. kończyć się (example from Eggers 1998: 312). (b) As in Polish, the reflexive pronoun zix can be used as a “solitude marker”, e.g. ix gej zix – ‘I am walking along by myself’ (Katz 1987: 125).
5.2. Particles and discourse markers A number of connectors, expressing addition and contrast, have been borrowed from Slavic: “Yiddish has … the conjunctions to ‘in that case’, tsi ‘whether’ (Belor., dial. Pol. cy), (ńi)xaj ‘let…’, xoč ‘although’, (a)xibe ‘unless’, i…i ‘both… and’ (dial. even i ‘and’, in enumerations), (bodáj…) abí‘(let…) so long as’; …jak…abí- ‘be… as it may, but’-, dial. pokevanen ‘until’.” (Weinreich 1958: 390). Among these connectors, tsi ‘whether’ functions as a subordinating conjunction, xoč ‘although’ as a concessive clause marker. The multifunctional focus particle ot is borrowed from Slavic languages in the meaning of ‘just’, e.g. ot derzelbiger ‘just the one’. It is also used as a place deictic, ‘here’ ot (in pointing) and ‘there’ ot (in pointing). In the field of interjections, nu ‘well?’, ‘come on!’ is of Slavic origin. Regionally, further discourse markers borrowed from various Eastern Euro-
Yiddish
253
pean contact languages may have been in use, e.g. in code-switching or semihabitual code-switching.
5.3. Adjectives and adverbs Affixes -ske, -(ev)ate are borrowed from Slavic languages as adjectivizers. The infixes -ičk-, -ink-, -en-, originating from Slavic, do not change the particular part of speech, but add a certain semantic feature and can be used to modify adjectives. Comparative forms in Yiddish show similarities to Polish and/or Russian constructions: Yidd.: greser fun mir; Pol.: większy ode mnie – ‘taller than me’.Yidd.: er iz mer enlex geven afn general vi af a jidn; Russ.: on byl bolee poxož na – ‘he was more like a General than a Jew’ (Eggers 1998: 317). One of three ways to form the superlative is an analytical construction with the borrowed particle same + positive: majn same belibter frajnt ‘my most beloved friend’ (Lötzsch 1974: 455).
6. Constituent order It is unclear whether constituent order in Yiddish has been influenced by recent contact with Slavic or Baltic languages. The features which might have developed due to contact are the following:
6.1. Possessor–possessed order Yiddish allows for the positioning of the possessive pronoun after the noun as an emphatic way of addressing a person, e.g. bruder majner! – ‘my brother!’, which is also possible in Slavic languages and is more common than in Germanic, where it appears very rarely and only in highly stylized forms, e.g. German: Vater unser! ‘Our Father!’
6.2. Adjective–noun order The adjective can follow the head noun (marked form), e.g. dos land dos farbotene – ‘the forbidden land’ (Eggers 1998: 313). Some scholars, for
254
Gertrud Reershemius
example Eggers (1998), claim that this construction is copying the generally freer word order of Slavic languages. It needs to be taken into account, however, that spoken German varieties (e.g. Bavarian) know this construction as well, e.g. in order to emphasize as in der Bub der narrische – ‘the foolish boy’. The construction could therefore equally well be a development within the Germanic component. Additionally, Yiddish does not share the German “finite-verb last” rule for subordinated clauses. Some scholars claim that this is due to contact with Slavic languages (e.g. Weinreich 1958: 383 or Eggers 1998: 313, without going into detailed discussion of this point). Ebert (1998) shows that constituent order in German subordinated clauses is a fairly recent phenomenon, since it only became the norm by the end of the sixteenth century. He comes to the conclusion that it did not originate in spoken language but in the written varieties used by Chanceries. Yiddish word order may be reflecting structures common in older, spoken varieties of German (Reershemius 2005). Yiddish word order in subordinated clauses could have developed from within the Germanic component, but independently from an emerging German standard language. Current developments in spoken German underline this point, since subordinate clauses introduced by weil – ‘because’ and obwohl – ‘although’ tend not to follow the “finite-verb-last” rule of Standard German any more.
7. Syntax In the fields of negation, coordination, and embedding, recent language contact needs to be considered as a formative factor.
7.1. Negation Double negation (or poly-negation) in Yiddish is often mentioned as a phenomenon which developed under the influence of Slavic languages. Double negation, although not allowed in Standard German, is however a common structure in older variants of German, as well as in many modern spoken German varieties. Therefore, double negation in Yiddish can be considered as a structure developed from within the Germanic component (see e.g. Eggers 1998: 315316). The fact that Slavic contact languages contain the struc-
Yiddish
255
ture as well might have supported it being established in Yiddish. Weinreich (1958: 383) draws attention to another construction related to negation which can be traced back to the influence of Slavic languages: after verbs expressing fear, the subordinated clause is negated, e.g. ix hob mojre er zol nit kumen – ‘I am afraid he might come’, which is clearly uncommon in Standard German as well as in spoken and (known) older varieties of German.
7.2. Coordination and adverbial clauses A number of coordinating conjunctions are borrowed from Slavic languages (see Section 5). The concessive subordinator xoč ‘although’ is borrowed from Slavic.
7.3. Relative clauses In Yiddish, interrogative pronouns function as relative pronouns, as in Slavic languages. It is unclear, though, whether Yiddish follows the example of Slavic here, since Standard German and spoken German varieties also use interrogatives as relative pronouns in addition to the deictic paradigm der, die, das. Word order in relative clauses follows Slavic examples: (6)
A dire fun fir tsimern, ejnem fun velxe er hot ajngeštimt a flat of four rooms, one of rel pron he has agreeparticiple tsu fardingn mir.8 to rent out me. ‘A flat of four rooms, one of which he agreed to rent out to me.’
As in some spoken Slavic varieties, Yiddish adds an anaphoric pronoun to the relative pronoun vos in order to supply information about gender, number and case, e.g. ... verter, vos er alajn hot šojn in zej ništ geglejbt ... – ‘words in which he himself did not believe’ (example from Lötzsch 1974: 458). In generalizing relative clauses, Yiddish also follows structures from Slavic languages, especially Russian, e.g. ... un vi kurts di švajgendike rege zol noxdem nit zajn ... ‘however short the period of silence afterwards might be’ (example from Lötzsch 1974: 458).
256
Gertrud Reershemius
8. Lexicon Lexical borrowing is probably the area where Yiddish has been influenced most through recent language contact with Slavic, and to a certain extent Baltic languages. Bin-Nun (1973) estimates that the Yiddish lexicon contains approximately 10 to 15 percent of words borrowed from Slavic languages. Nouns, verbs, adjectives, adverbs, pronouns, adpositions, conjunctions, discourse markers, interjections, particles and derivational affixes have been borrowed into the semantic and communicative domains of religion, fauna and flora, trade, geography, kinship terms, tools and body parts. Weinreich (1958: 386387) also lists parts of house, household items, family, clothes and food. The deictic particle ot can be used as a spatial expression for ‘here’ (in pointing) and ‘there’ (in pointing).
9. Conclusion Recent language contact has led to a high degree of lexical borrowing in Yiddish from the Slavic languages. Yiddish phonology was clearly influenced by Slavic, although some features remained regional. In the area of morphosyntax, many structures may or may not have evolved through contact. Closer analysis shows that they could just as well be the result of internal developments in the Germanic component, which might, however, have been triggered by contact. The discussion of these features underlines the fact that it is insufficient to compare Yiddish only with Standard German. German spoken varieties, which existed and exist to a certain extent independently from the German Standard language, as well as older varieties, need to be taken into account because they sometimes provide alternative explanations to structures in Yiddish which have been put down to contact. Contact clearly is not the only reason for the distinct development of the Yiddish language. Grammaticalization processes within the Germanic component, which could develop fairly unrestrictedly because of the absence of a standardized written language for centuries, also need to be taken into account (cf. Reershemius 1997). By far the most productive area of contact with Slavic languages has been word derivation, which has led to a unique blend of Slavic and Germanic elements in Yiddish.
Yiddish
257
Abbreviations Belor. dat dial. f inf intensif LCAAJ m
Belorussian dative dialectal feminine infinitive intensifier Language and Culture Atlas of Ashkenazic Jewry masculine
NE Yiddish nom Pej. Pol. pron ref rel pron sg Yidd.
Northeastern Yiddish nominative pejorative Polish pronoun reflexive relative pronoun singular Yiddish
Notes 1. Paul Wexler’s (1991) hypothesis that Yiddish is a relexified variant of Sorbian has been rejected by most scholars in the field, see e.g. Eggers (1998) and the responses to Wexler’s article in International Journal of the Sociology of Language 91 (1991). 2. According to Bin-Nun (1973: 8590) Yiddish was used by eight million speakers worldwide in the 1930s, of which 6 million were based in Eastern Europe. The LCAAJ, vol.1 (1992: 10) estimates that shortly before the Second World War Yiddish had 10.5 million speakers with seven million living in Eastern Europe. 3. The chapter is based on a corpus of spoken Yiddish in Israel. The data was collected in 1988/89 and will be accessible in digitized form through the Phonogramme Archive (Austrian Academy of Arts and Sciences) in Vienna. 4. Examples taken from Weissberg (1988: 127). 5. Examples from Weinreich (1958) and Eggers (1998). 6. Examples from Lötzsch (1974: 456457). 7. Example from Lötzsch (1974: 456). 8. Example from Lötzsch (1974: 458).
References Albert, Ruth, and Henk D. Meijering 2001 Hat das Jiddische ein Aspektsystem? In: Sprache und Text in Theorie und Empirie. Beiträge zur germanistischen Sprachwissenschaft. Festschrift für Wolfgang Brandt, 2840. Stuttgart: Steiner. Aronson, Howard I. 1985 On Aspect in Yiddish. General Linguistics 25: 171188.
258
Gertrud Reershemius
Bin-Nun, Jechiel 1973 Jiddisch und die deutschen Mundarten. Unter besonderer Berücksichtigung des ostgalizischen Jiddisch. Tübingen: Moor. Ebert, Rolf 1998 Verbstellungswandel bei Jugendlichen, Frauen und Männern im 16. Jahrhundert. Tübingen: Niemeyer. Eggers, Eckhard 1998 Sprachwandel und Sprachmischung im Jiddischen. Frankfurt a. Main: Lang. Isaacs, Miriam 1999a Haredi, haymish and frim: Yiddish vitality and language choice in a transnational multilingual community. Internationa Journal of the Sociology of Language 138: 930. 1999b Contentious partners: Yiddish and Hebrew in Haredi Israel. International Journal of the Sociology of Language 138: 101121. Jacobs, Neil 2005 Yiddish. A Linguistic Introduction. Cambridge: Cambridge University Press. Katz, Dovid 1987 Grammar of the Yiddish Language. London: Duckworth. Herzog, Marvin, Vera Baviskar, Ulrike Kiefer, Robert Neumann, Wolfgang Putschke, Andrew Sunshine, and Uriel Weinreich (eds.) 19922000. The Language and Culture Atlas of Ashkenazic Jewry (LCAAJ). Vol. 1 Historical and Theoretical Foundations, vol. 2 Research Tools, vol. 3 The Eastern Yiddish – Western Yiddish Continuum. Tübingen: Niemeyer. Lötzsch, Ronald 1974 Slawische Elemente in der grammatischen Struktur des Jiddischen. Zeitschrift für Slawistik XIV: 446459. Lowenstein, Steven 1969 Results of Atlas Investigations among Jews in Germany. In: Marvin I. Herzog, Wita Ravid and Uriel Weinreich (eds.), The Field of Yiddish. Studies in Language, Folklore and Literature (3rd edn.), 1635. The Hague: Mouton. Reershemius, Gertrud 1997 Biographisches Erzählen auf Jiddisch. Tübingen: Niemeyer. 2005 Einige Bemerkungen zur Bewahrung von Merkmalen des älteren Deutsch im Jiddischen. In: Holger Briel and Carol Fehringer (eds.), Field Studies: German Language, Media and Culture. Frankfurt/Main et al: Lang, 1127. Timm, Erika 1987 Graphische und phonische Strukturen des Westjiddischen. Tübingen: Niemeyer.
Yiddish 2005
259
Historische jiddische Semantik. Die Bibelübersetzungssprache als Faktor der Auseinanderentwicklung des jiddischen und des deutschen Wortschatzes. Tübingen: Niemeyer. Weinreich, Uriel 1958 Yiddish and Colonial German in Eastern Europe. In: American Contributions to the Fourth International Congress of Slavicists, Moscow, September 1958, 369419. The Hague: Mouton. Weissberg, Joseph 1988. Jiddisch. Eine Einführung. Frankfurt/Main: Lang. Wexler, Paul 1991 Yiddish: The fifteenth Slavic Language. A study of partial language shift from Judeo-Sorbian to German. International Journal of the Sociology of Language 91: 9150.
Grammatical borrowing in Hungarian Rumungro Viktor Elšík
1. Background1 The language under description is a variety of Romani (Indo-Aryan) spoken by long-settled Roms (Gypsies) of southern Slovakia and northern Hungary, which is classified as the Northern (non-Vendic) subgroup of the South Central group of Romani dialects (cf. Boretzky 1999; Elšík, Hübschmannová, and Šebková 1999) and usually refered to as “Rumungro” in Romani linguistics. The variety I chose to describe is one of the few Rumungro varieties whose speakers are Hungarian bilinguals.2 It is the language of some 1,350 Rom inhabitants of the Hungarian village of Selice (Hungarian Sók, Romani Šóka) in southwestern Slovakia. In addition, there are about 150 Roms in the village who speak a different (a North Vlax) dialect of Romani. The former Roms are referred to as Rumungri (originally ‘Gypsy-Hungarians’) by the latter group, who are called Pojáki (originally ‘Poles’) by the Rumungri. Both groups use the ethnonym Rom for their own group and both are called cigányok ‘Gypsies’ by Hungarians, although the Hungarian villagers clearly differentiate between magyar cigányok ‘Hungarian Gypsies’ (i.e. the Rumungri) and oláh cigányok ‘Romanian Gypsies’ (i.e. the Pojáki). At present, both Rom groups taken together slightly outnumber the Hungarian population of the village. Until recently, however, the Hungarians were in a demographic majority and they remain the socially, economically, and politically dominant group in the village. Rumungro is prevalently an oral language; some Rumungri are able to write letters or text messages in Rumungro but the language is not used for regular written communication. Nor is it used in massmedia or in formal education. Although Romani in general is an officially recognized language in Slovakia, there is no recognition of the Rumungro dialect specifically and, so far, there have been no attempts at its standardization. The Rumungro of Selice is the language of family and in-group communication among the local Rumungri and the language of inter-group communication between the Rumungri and the local Pojáki. While the latter learn Rumungro as their second dialect of Romani (and speak a distinct ethnolect of it), the Rumungri usually do not learn the dialect of the Pojáki. Many Hungarian villagers understand
262
Viktor Elšík
Rumungro well, although only a few have some active competence in it and they are rarely fluent speakers. While all Selice Rumungri born before 1975 or so are native speakers of Rumungro, in some families children are presently spoken to only in Hungarian or Slovak, and left to acquire some competence in Rumungro in adolescent and adult peer groups, if at all. Thus, Rumungro of Selice is not a safe language, though it is not seriously endangered yet. All school-age or older L1 speakers of Selice Rumungro are multilingual. First of all, they are fluent and highly competent in Hungarian, which they use especially in their everyday communication with the Hungarian villagers. Some young children may be monolingual in Rumungro, although early acquisition of Hungarian appears to be the prevailing pattern nowadays. In addition, most Rumungri are fluent in Slovak, the official and dominant language of Slovakia, which they use outside of the village. Also, most have acquired at least passive competence in Czech through their exposure to Czech massmedia and employment-related stays in the Czech part of the former Czechoslovakia (in the 1960s–1980s almost all families of the Selice Rumungro community spent ten to thirty years there). Though both Hungarian and Slovak (and to some extent Czech as well) may be classified as current L2s of Selice Rumungro, it is clear that Hungarian enjoys a special sociolinguistic status: inter alia it is the language of the secondary ethnic identity of the Selice Rumungri, who frequently refer to themselves as “Hungarian” Roms, accepting the attribute ascribed to them by Hungarians. As evidenced by lexical borrowings, Rumungro shares with other Romani dialects previous contact with West Iranian (Persian and/or Kurdish), Ossetic, Armenian, and especially Greek; the latter language also had an enormous impact on Romani grammar. On the other hand, most South Slavic loanwords in Rumungro are dialect-specific within Romani. Some of them can be identified as Serbian–Croatian or even Ikavian Serbian– Croatian (Elšík, Hübschmannová, and Šebková 1999). Linguistic contact of Rumungro with Hungarian is likely to have lasted for at least two centuries. Widespread multilingualism of the Selice Rumungri in Slovak and Czech did not develop before the 1920s and 1950s, respectively. While these secondary current L2s have contributed only a few marginal loanwords, Hungarian has exerted, and continues to exert, a strong lexical and grammatical influence on Rumungro. The present chapter will focus on Rumungro borrowings from Hungarian, although borrowings from other contact languages, both pre-Hungarian and “post”-Hungarian (i.e. Slovak and Czech), will also be discussed.
Hungarian Rumungro
263
2. Phonology The inventory of Rumungro phonemes is identical to that of Hungarian, with two exceptions. First, Rumungro retains distinctive aspiration in voiceless stops and affricates, e.g. čór- [ʧo:r] ‘steal’ vs čhor- [ʧhor] ‘pour’, which is absent from Hungarian. Second, Hungarian rounded front vowels are usually replaced with their unrounded counterparts in loanwords, e.g. csütörtökön [ʧytørtøkøn] > čiterteken [ʧiterteken] ‘on Thursday’, although some speakers now tend to retain them in certain loanwords. Both vowel and consonant inventories of Romani have been enlarged due to contact with Hungarian. Instances of contact-induced phoneme loss are rare: they include the merger of the voiceless uvular fricative [χ] with the glottal fricative /h/ [h] and the merger of the palatal lateral [lj] with the palatal approximant /j/ [j], e.g. *[χaljam] >hájam [hɒ:jam] ‘we ate’. On the other hand, contact with Hungarian has given rise to several phonemic distinctions and numerous new phonemes in Rumungro. A major contact-induced change has been the development of distinctive phonological quantity: vowel length, e.g. phirav- [phirav] ‘wear’ vs phírav[phi:rav] ‘make [so.] walk’, and consonant gemination, e.g. čuča [ʧuʧa] ‘empty’ (an inflectional form) vs čučča [ʧuʧ:a] ‘breasts’. Both types of quantity have spread to the pre-Hungarian lexical component, although some individual geminates remain restricted to the Hungarian component. The inventory of vocalic qualities, too, has been enlarged due to contact. Although the open-mid front vowels – the short /ë/ [æ] and the long /e˝/ [æ:] – are mostly restricted to Hungarian loanwords, they are phonologically distinct from their closed-mid counterparts, e.g. dë [dæ] ‘but’ vs de [de] ‘give!’. In addition to the phonological quantity difference, the long /á/ [ɒ:] is distinguished through phonetic rounding from the short /a/ [a], as it is in the local Hungarian dialect. Contact with Hungarian has also triggered the development of a series of palatal consonants from palatalized dentals or palatalized velars, e.g. *[tatjar] > taťar- [tacar] ‘make warm’, *[kjhil] > ťhil [chil] ‘butter’. The Hungarian-origin phonemes play an important role in morpho-phonological alternations. In addition, several morpho-phonological rules are borrowed. For example, a morpheme-initial palatal approximant triggers gemination and a shift to a palatal of a preceding morpheme-final dental stop, as it does in Hungarian, e.g. kafid-i [kafidi] ‘table’ → {kafid-ja} kafiďď-a [kafiÔ:a] ‘tables’. Rumungro also borrows vowel harmony from Hungarian, although it remains restricted to a single type of alternation that affects only a few indigenous affixes, e.g. farkašš-a [farkaʃ:a] ‘wolves’ vs ke˝mívëšš-ë [kæ:mi:væʃ:æ]
264
Viktor Elšík
‘bricklayers’, bika-ha [bikaha] ‘with a bull’ vs këčkë-hë [kæʧkæhæ] ‘with a goat’. Apart from the development of long vowels and geminate consonants, the syllable structure of the pre-Hungarian component has remained unaffected by contact with Hungarian. On the other hand, there is no adaptation of Hungarian loanwords in terms of their syllable structure. The distribution of long vowels in Rumungro suggests that they developed before the Hungarian-induced general shift of stress to word-initial position, e.g. *[barvalÕo] > *[barva:lÕo] > barválo [bÕarvɒ:lo] ‘rich’. Intonation patterns are largely identical to those of the local Hungarian dialect.
3. Typology The typological profile of Asian (Proto-)Romani was altered rather significantly already before the arrival of its speakers to Europe. Matras (2002: 196) argues that, for example, the development of interrogative-based relativizers or the reduction of non-finite constructions could have taken place in a western Asian convergence area, i.e. before the contact of Romani with Greek in Asia Minor. The latter language nevertheless remains the major source of typological innovations that are shared by Romani as a whole: the development of a proclitic definite article, the emergence of prepositions (or a significant expansion of their inventory), the shift to a basic predicate–object order, and more (cf. Matras 1994, 2002: 198199). Post-Greek L2s have had a less significant impact on major typological parameters of Rumungro. While Romani possessed a single prefix in its Greek period, matter borrowing of several pronominal prefixes from South Slavic and Hungarian, of a superlative prefix from Hungarian, and a grammaticalization of another pronominal prefix due to pattern replication from Hungarian (see Sections 4 and 6), has increased the number of prefixes in Rumungro by eight. Outstanding syntactic developments due to contact with Hungarian include the creation of a class of preverbs (see Section 5), the “re-introduction” of non-finite subordinate constructions (see Section 8), and various modifications in word-order patterns (see Section 7).
4. Nominal structures Nouns are commonly borrowed into Rumungro. Pre-Greek and some (presumably early) Greek noun loans show full morphological integration and are
Hungarian Rumungro
265
structurally indistinguishable from indigenous nouns; they have a so-called oikoclitic morphology. Some (presumably late) Greek and post-Greek noun loans, on the other hand, have so-called xenoclitic morphology (Elšík and Matras 2006: 324333), which is characterized, above all, by borrowed nominative inflections, mostly of Greek origin. These inflections were extracted from lexical loans of nominative noun forms, and extended to later loanwords as well. For example, the xenoclitic nominative singular feminine suffix -a was extracted from Greek-origin nouns, e.g. cip-a ‘skin’ < tsip-a, and extended to nouns borrowed from South Slavic, e.g. péť-a ‘oven’ < Serbian– Croatian pēć, and Hungarian, e.g. virág-a ‘flower’ < virág. Hungarian does not contribute any xenoclitic noun inflections. Like nouns, adjectives, too, are commonly borrowed. The distinction between xenoclitic and oikoclitic adjective inflection, which is reconstructable for earlier post-Greek stages of Romani, has been lost in Rumungro due to internal analogical developments (Elšík and Matras 2006: 329). Borrowed adjectives now inflect like indigenous adjectives, showing Indo-Aryan inflections, e.g. žut-o ‘yellow’ < Serbian–Croatian žut. Unlike earlier adjective loans, adjectives borrowed from Hungarian contain the overt adaptation suffixes -av- or -n- of South Slavic origin, in addition to the indigenous inflections, e.g. sirk-av-o ‘grey’ < szürke, kík-n-o ‘blue’ < kék. Although Rumungro possesses indigenous means to derive manner adverbs from adjectives, most manner adverbs corresponding to Hungarian-origin adjectives are lexical loanwords from Hungarian rather than internal derivations, e.g. okoššan ‘wisely’ < okosan (cf. okoš-n-o ‘wise’ < okos). Extraction from lexical borrowings is the source of several derivational affixes in Rumungro. There are three borrowed noun-deriving affixes that are productive, in addition to several lexically restricted ones, which will be left out of the present discussion. First, the South Slavic-origin suffix -kiň- derives nouns denoting female humans from loanwords of Hungarian human nouns, e.g. sakáč-kiň-a ‘female cook’ ← sakáč-i ‘cook’ (< szakács ‘cook’). Second, the Hungarian-origin suffix -áš- derives action nouns from internal verb derivations in -áz- (see Section 5), e.g. ďij-áz-áš-i ‘singing’ ← ďij-áz-in- ‘to sing’ (← ďíl-i ‘song’), and from a few underived indigenous verbs, e.g. muk-áš-i ‘divorce, separation’ ← muk- ‘to leave, let’. Finally, the Hungarian-origin prefix mí- derives nouns denoting artificial objects from nouns denoting natural objects of the same type, e.g. mí-dand ‘artificial tooth’ ← dand ‘tooth’. The prefix has been extracted from loanwords of Hungarian compounds consisting of the noun mű ‘creation, artificial thing etc.’ and a body-part noun, e.g. mí-këňek-a ‘artificial elbow’ < mű-könyök.
266
Viktor Elšík
Extraction of adjective- and adverb-deriving affixes is also attested. The Greek-origin suffix -(i)k- derives several semantic types of relational adjectives from nouns, e.g. dévl-ik-o ‘divine’ ← dél ‘God’, ninc-k-o ‘German adj’ ← ninc-o ‘German n’ (< Serbian–Croatian nimac), meňassoň-ik-o ‘bridal’ ← meňassoň-a ‘bride’ (< Hungarian menyasszony). Ethnic adjectives in -(i)kthen form manner adverbs by the suffix -a, which is likewise of Greek origin, e.g. ninc-k-a ‘in (a) German (way)’. The Serbian–Croatian-origin suffix -astis a fully productive means to derive attenuative adjectives, e.g. gull-ast-o ‘rather sweet’ ← gull-o ‘sweet’, míl-n-ast-o ‘rather deep’ ← míl-n-o ‘deep’ (< Hungarian mély). The Hungarian-origin suffix -óš- derives a few active adjectives from causatives of a class of indigenous verbs, e.g. dara-v-óš-n-o ‘frightening’ ← dara-v- ‘to frighten’ (← dara- ‘to fear’); the obligatory presence of the adaptation suffix -n- after -óš- shows that the latter has been extracted from adjectives borrowed from Hungarian. Finally, a complex interplay of matter borrowing from Hungarian and internal re-analysis has resulted in the development of the suffix -šon, which derives manner adverbs from a class of similative adjectives, e.g. roman-iká-šon ‘in a rather Gypsy way’ ← roman-ikán-o ‘Gypsy-like’ (← roman-o ‘Gypsy adj’). Affix extraction, however, does not appear to be responsible for the borrowing of the superlative prefix lëg- from Hungarian, which derives the superlative from the comparative, e.g. lëg-bar-eder ‘the biggest’ ← bar-eder ‘bigger’ (← bár-o ‘deep’), lëg-míl-n-eder ‘the deepest’ ← míl-n-eder ‘deeper’ (← míl-n-o ‘deep’ < Hungarian mély). The superlative prefix must have been borrowed directly, without the mediation of lexical borrowing, since Rumungro superlative forms are internal derivations rather than borrowings of Hungarian superlatives, such as leg-mély-ebb ‘the deepest’. The direct borrowing of the superlative prefix was probably facilitated by the resulting structural isomophism between the Rumungro and the Hungarian degree derivation, viz. derivation of comparatives by suffixation and of superlatives by further prefixation. Although the Rumungro comparative suffix is preHungarian (Indo-Aryan or, more likely, Iranian, cf. Matras 2002: 196), dialect comparison with other Romani dialects suggests that the retention of synthetic comparatives in Rumungro is likely to be due to contact with Hungarian. An unambiguous instance of pattern replication from Hungarian is the creation of associative plurals in Rumungro human nouns, which are distinct from their regular plurals, e.g. ke˝mívëš-ingere ‘bricklayer and his work team’3 vs ke˝mívëšš-ë ‘bricklayers’ (← ke˝mívëš-i ‘bricklayer’); the category, undocumented in other Romani dialects, replicates an identical distinction in
Hungarian Rumungro
267
Hungarian, e.g. kőműves-ék vs kőműves-ek (← kőműves). Probably due to pattern replication from the genderless Hungarian, Rumungro has lost feminine derivation with nouns denoting (higher) animals: for instance, the masculine gra ‘horse’ (originally also *‘stallion’) has lost its inherited feminine counterpart *gras-n-i ‘mare’ and is now a generic designation of the species. There are numerous instances in Rumungro of pattern replication concerning the syntax and semantics of case markers. To mention just a couple of examples: the dative case, which encodes beneficiaries and some recipients, is now also used to encode certain predicate complements (Rumungro 1a, Hungarian 1b), and the inferior spatial preposition tal ‘under’ also encodes temporal telic extent (2ab). Finally, the Hungarian model has triggered the grammaticalization of indigenous spatial adverbs into a series of separative prepositions, cf. the Rumungro preposition anglal (3a), and the Hungarian postposition elől ‘from the front of’ (3b). (1)
a. Romes-ke man hajovav. Gypsy.obl-dat 1sg.acc feel.1sg b. Cigány-nak érzem magam. Gypsy-dat feel.1sg emph.1sg ‘I feel as a Gypsy.’
(2)
a. Tal o pándž dí ári sasťíja. under def five day outward get.healthy.pfv.3sg b. Öt nap alatt gyógyult meg. five day under get.healthy.pret pfv ‘S/he recovered in five days.’
(3)
a. Naššov anglal mre jakha. get.lost.imp from.the.front.of 1sg.gen:pl eye.pl b. Tűnj a szemem elől. get.lost.subj def eye.1sg.poss from.the.front.of ‘Get out of my sight!’, lit. ‘Get lost from the front of my eye(s)!’
5. Verbal structures Verbs are commonly borrowed into Rumungro. Pre-Greek and early Greek loan-verbs show full morphological integration and are structurally indistinguishable from indigenous verbs. Post-Greek loan-verbs, on the other hand,
268
Viktor Elšík
are marked out by a specific adaptation marker, the Greek-origin suffix -in-, which is added to an inflectional stem of the source verb, e.g. vič-in- ‘to shout’ (< Serbian–Croatian vič-), dógoz-in- ‘to work’ (< Hungarian dolgoz-), and followed by regular indigenous inflections. The suffix was extracted from lexical borrowings of Greek verbs with the present stem in -in-. Though none of these have been retained in Rumungro, the suffix has been extended to those Greek loan-verbs that originally contained a different suffix, e.g. rum-in- ‘to damage, spoil’ < Greek rim-az- ‘to ravage’. Dialect comparison suggests that the suffix -in- was originally specialized for non-perfective adaptation of some transitive loan-verbs in Romani (Matras 2002: 130). In Rumungro, however, it has developed into a general, aspect- and valencyneutral, verb-adaptation marker.4 Nonce loan-verbs from Slovak (or Czech) show a distinct pattern of morphological adaptation: their infinitive stems get adapted by the Hungarian-origin adaptation suffix -ál-, in addition to the regular adaptation suffix -in-, e.g. sledov-ál-in- ‘to observe, follow’ (< Slovak sledova-).5 The adaptation suffix -in- is absent in passive participles of morphologically adapted borrowed verbs. Instead, the participles contain the Greekorigin participle suffix -ime, e.g. rum-ime ‘spoiled’, vič-ime ‘shouted’, téle dógoz-ime ‘worked away’ (lit. ‘downward worked’), sledov-ál-ime ‘observed, followed’. The suffix was extracted from Greek lexical borrowings and extended to post-Greek loan-verbs and several indigenous verb classes, e.g. d-ime ‘given’ (← d- ‘to give’; cf. Elšík and Matras 2006: 331332). Another borrowed non-finite marker is the Hungarian-origin infinitive suffix -ňi (cf. Elšík and Matras 2006: 179), which has been extracted from lexical borrowings of Hungarian infinitives and extended to a class of non-borrowed verbs, viz. those derived by the suffix -áz- (see below). Like the participle suffix -ime, the infinitive suffix -ňi is incompatible with the adaptation suffix -in-, e.g. ďij-áz-ňi ‘to sing’ (← ďij-áz-in- ‘to sing’, stem). Unlike Hungarian infinitives, the Rumungro Hungarian-origin infinitives do not allow any nominal inflection (see Section 7 for syntactic details). Extraction from Hungarian loanwords is also the source of two Rumungro verb-deriving affixes. The suffix -áz-, in conjunction with the following adaptation suffix -in-, is a productive means to derive intransitive verbs from preHungarian nouns, e.g. paramis-i ‘fairy-tale’ (< Greek) → paramis-áz-in- ‘to tell fairy-tales’. The second extracted derivational affix is the causative suffix -tat-. Rumungro allows three different structural types of causatives of Hungarian loan-verbs: lexical borrowing and adaptation of Hungarian causatives, e.g. dógoz-tat-in- ‘to make [so.] work’ (< dolgoz-tat); internal deriv-
Hungarian Rumungro
269
ation from a non-causative loan-verb by an indigenous causative marker, e.g. dógoz-in-av-; or, most commonly, a combination of both types, which results in double causative marking, e.g. dógoz-tat-in-av-. This pattern of double causative marking has also been analogically extended to some classes of inherited and internally derived verbs, e.g. ďij-áz-in- ‘to sing’ → ďij-áz-tat-in-av‘to make [so.] sing’. While morphological causatives have been inherited from Indo-Aryan, dialect comparison within Romani suggests that their retention and productivity in Rumungro is due to pattern replication from Hungarian (cf. Hübschmannová and Bubeník 1997). Morphological frequentatives, on the other hand, represent a novel category in Rumungro: the existence of suffixal frequentatives in Hungarian has triggered the development of the Romani transitive suffix -ker- ~ -ger- into a valency-neutral frequentative marker in Rumungro, e.g. ťin- ‘to buy’ → ťin-ger- ‘to buy frequently’. Pattern replication from Hungarian is also responsible for the creation of a specific class of preverbs (coverbs, verbal particles), i.e. free adverbial forms that encode spatial, aktionsart or lexical modification of the verb. Many, though not all, preverb constructions are exact translations of their Hungarian models, e.g. téle thov- < le-tesz [downward put] ‘pass (e.g. an exam)’, but Rumungro ánde sov- [inward sleep] vs Hungarian el+alsz- [away sleep] ‘fall asleep’.6 Most preverbs arose through grammaticalization of pre-Hungarian spatial adverbs and many, though not all, of these still retain their spatial functions as well (cf. also Schrammel 2005). In addition, numerous preverbs are matter borrowings from Hungarian (see Section 6).
6. Other parts of speech In addition to lexical verbs, nouns, adjectives, and manner adverbs, Rumungro has borrowed numerous function (or less lexical) words from its different L2s. The modal particle of possibility šaj ‘can’ is likely to be of West Iranian origin (Matras 2002: 196). Greek is the source of the cardinal numerals efta ‘seven’, ofto ‘eight’, ëňňa ‘nine’, and trianda ‘thirty’ and the ordinal trito ‘third’; the quantifier buka ‘a little, a piece of’; the address particle more ‘hey, man!’; the temporal deictic particle paleg ‘then, after that’ (< ‘again’); and the temporal adverb táha ‘tomorrow’. Serbian–Croatian provided the quantifiers dosta ‘enough’, sako ‘every’, and cilo ‘whole’; the distributive particle po; the optative/permissive particle nek ‘let’, which has also been grammaticalized into a subordinator (cf. 11e,f); the focus particle ni ‘not
270
Viktor Elšík
even, neither’ and the related coordinator ni – ni ‘neither – nor’; the negative pronoun ništa ‘nothing’; and the preverb préku ‘through; across, over’, which has been grammaticalized within Rumungro from a borrowed spatial adverb (cf. Section 5). Most function words have been borrowed from Hungarian, the current L2. Hungarian is the source of numerals (see below), the quantifier čepo ‘few, little; a few, a little’ (< ‘a drop of’), the degree words igën ‘very, very much’ and túl ‘too, too much’, the generic obligative particle musaj ‘one has to’, numerous preverbs (e.g. át ‘through; across, over’ or sít ‘apart’), and a few marginal postpositions (e.g. sërint ‘according to’ or fëlé ‘in the direction of’). Rumungro commonly borrows inflectional forms of Hungarian nominals, including pro-words, which function as adverbs in the recipient language, e.g. aňňira ‘to that extent’ (< sublative of annyi ‘that much’), magátú ‘by oneself’ (< ablative of the reflexive–emphatic pronoun maga), idëgembë ‘abroad’ (< inessive/illative of idegen ‘foreign country’). Especially temporal adverbs of this kind are abundant, e.g. akármikor ‘anytime whatsoever’, tavaskor ‘in the spring’ (< temporal case of akármi ‘anything whatsoever’ and tavasz ‘spring’), díbë ‘at noon’, márciušba ‘in March’ (< inessive of dél ‘noon’ and március ‘March’), serdán ‘on Wednesday’ (< superessive of szerda ‘Wednesday’). Borrowing from Hungarian is extensive in discourse-related function words, such as repetition adverbs (újbú or újra ‘again, anew’), utterance-level adverbs (talán ‘perhaps’, bistoš ‘certainly’, përsë ‘of course, sure’, bizoň ‘indeed’), phasal adverbs (még ‘still’ and má ‘already’), focus particles (iš ‘also, too’, čak ‘only’, ippën ‘just’, pont ‘exactly’, ëgís ‘entirely’), affirmative answer particles (the regular hát ‘yes’, and the contrary-to-expectation dë ‘but yes’), interjections (ëhë), fillers (hát), sequential discourse markers (no), and more. Borrowed coordinators and subordinators are common and will be discussed in Section 8. In addition to function words, Rumungro has borrowed several functionword affixes, only some of which will be discussed here. The Greek-origin suffix -t- derives regular ordinals from cardinal numerals, e.g. dúj ‘two’ → dúj-t-o ‘second’. The Hungarian-origin suffix -tú marks separative orientation in local adverbs and posterior–durative relation in temporal adverbs, e.g. ánglal ‘in/to the front’ → ánglal-tú ‘from the front’, ídž ‘yesterday’ → idž-al-tú ‘since yesterday’. The South Slavic-origin prefix ni- and the Hungarian-origin prefixes vala-, akár-, and minden- apply to interrogative pro-words, e.g. káj ‘where’ → ni-kháj ‘nowhere’ (negative), vala-káj ‘somewhere’ (specific indefinite), akár-káj ‘anywhere whatsoever’ (free-choice), and mindenkáj ‘everywhere’ (universal quantification). The Hungarian-origin prefixes
Hungarian Rumungro
271
am- and uďan- apply to deictic pro-words, e.g. asso ‘such’ → am-asso ‘such like the other’ (deictic contrast) and uďan-asso ‘just such like this/that one’ (deictic identity). All of the pronominal prefixes must have been borrowed without the mediation of lexical borrowing. There are also several instances of pattern replication from Hungarian in function words. The genderless Hungarian is the source of gender neutralization in the nominative of the Rumungro third-person singular pronoun: the original feminine form ój ‘she’ has replaced the original masculine form *óv ‘he’, assuming a gender-neutral function ‘s/he’ (cf. H ő ‘s/he’).7 On the other hand, the development of a distinction between local pro-words of stative location and direction, e.g. káj ‘where’ vs kija ‘whither’, is likely to have been modelled on an identical distinction in Hungarian. Due to a complex interplay of pattern replication and internal re-analysis, the universal-quantification prefix sa- has developed as an alternative to the borrowed universalquantification prefix minden- (see above), e.g. sa-káj ‘everywhere’. Pattern replication has also been involved in the grammaticalization of the reciprocal pronoun jékh-ávr- [one-(an)other-] ‘each other’, which is a compound of an identical structure as the Hungarian reciprocal pronoun egy-más. The expression of the phasal expression ‘no longer’ as a negation of ‘already’ is clearly modelled on Hungarian.8 In syntax, adnominal cardinal numerals (optionally in case of ‘one’) have lost case agreement with their head nouns due to Hungarian influence, e.g. dúj (*dúj-e) muršenca [two (*two-obl) man.pl.soc] ‘with two men’. A final note concerns borrowing of Hungarian numerals. Two types of loans must be distinguished: morphologically integrated loanwords, which have no inherited, pre-Hungarian alternative (the cardinals nulla ‘zero’, ëzeri ‘thousand’, and miliomo ‘million’, the ordinal e˝šéno ‘first’, and most fraction numerals), and morphologically unintegrated loanwords, which alternate with inherited numerals. The unintegrated numerals allow or require, due to Hungarian influence, the singular of some of their head nouns, viz. of some Hungarian-origin nouns denoting currency units: contrast pándžvárdeš hallér-ja ‘fifty hellers’ (indigenous numeral, plural noun) with ëtvën hallér-i ‘fifty hellers’ (Hungarian numeral, singular noun) < ötven hallér. Note that the latter construction is not necessarily a code-switch, as the singular noun is morphologically adapted in Rumungro. The alternation between inherited and borrowed expressions also concerns various de-numeral derivations and compounds, e.g. tritóneste [third.loc.sg.m] or harmadikán (< Hungarian) ‘on the third [day of a month]’, eftavardešberšiko or hëtvënívëšno (< Hungarian) ‘seventy-year-old’.
272
Viktor Elšík
7. Constituent order Linear order of the predicate, its arguments and adverbial adjuncts is flexible in Romani, being largely determined by pragmatic factors (cf. Matras 1995, 2002: 167174). While syntactic non-configurationality is also characteristic of Rumungro, numerous aspects of Rumungro clause-level order appear to have been borrowed from Hungarian, likewise a non-configurational language. A prominent example is the tendency to position focussed constituents immediately before the finite verb; this frequently results in clause-final position of the copula in non-verbal predications (4; second line). (4)
Odá hi gadžikano sokáši. that.m cop.pres.3 nonGypsy(a) habit Romano sokáši tista áver hi. Gypsy(a) habit sheer other cop.pres.3 ‘That’s a nonGypsy habit. The Gypsy habit is completely different.’
On the other hand, linear order at the noun phrase level is syntactically determined in Rumungro: all types of adjectival modifiers, including descriptive adjectives, adnominal possessors, demonstratives, and numerals, always precede their head nouns. While the modifier–noun order prevails in all Romani dialects (cf. Matras 2002: 165167), it has been fully grammaticalized in Rumungro due to contact with Hungarian. The alternative noun–modifier order is simply ungrammatical, except in cases of afterthought whereby the postposed modifier is a substantivized apposition. Rumungro exhibits an etymological split in the order of adpositions: while those borrowed from Hungarian are postposed to their object noun phrases, adpositions of preHungarian origin always remain preposed.9 An analogical split occurs with focus particles translatable as ‘also, too’: the indigenous te is preposed to the focused element, while the Hungarian-origin iš is postposed.
8. Syntax A number of clause-level syntactic features that Rumungro shares with Hungarian is due to a typological or areal similarity between the two languages, rather than due to immediate borrowing from Hungarian into Rumungro. For example, both languages have uninflected pre-verbal negators, allow prodrop, and use a copula verb in non-verbal predication (though, unlike Hun-
Hungarian Rumungro
273
garian, Rumungro does not allow copula deletion in the third-person present affirmative). Rumungro also shares with Hungarian negative agreement of the predicate with negative pro-words; this is clearly a post-Greek pattern in Rumungro, though South Slavic is a more likely source than Hungarian. The major structural domain of syntactic borrowing from Hungarian into Rumungro is clause combining and phrase combining. Rumungro borrows all of its coordinating conjunctions with the exception of conjunctive coordinators, which are pre-Hungarian: plain disjunctive vaď ‘or’, contrastive disjunctive vaď – vaď ‘either – or’, free-choice alternative ha – ha ‘whether – or’, and several connectors with adversative and contrastive functions, e.g. dë ‘but’, azomba ‘however’, mégiš ‘still, even so’, hanem ‘but rather’, and meg and pëdig ‘but, in turn’ (5). Borrowed adverbial subordinators include the causal mërt and mivël ‘since, because’ (6), and several non-simultaneous temporal subordinators: the posterior mire and miëlét ‘before’, the posterior– durative még ‘until’, and the anterior durative mióta ‘since’ (7). (5)
Dë ón na džan ánglal, hanem téle džan. conj 3pl neg go.3pl to.the.front conj downward go.3pl ‘But they are not progressing, they are rather sinking.’
(6)
Mivël čoháni ssa, conj witch cop.3sg.pret na tromalahi and-i khangéri te džan. neg dare.3sg.rem in-def.f church(f) comp go.3pl.subj=inf ‘Since she was a witch, she did not dare to go to the church.’
(7)
Mióta džukela hi amen, náne amen mačka. conj dog.pl cop.3.pres 1pl.acc cop.neg.3.pres 1pl.acc cat ‘Since we have kept dogs, we do not keep a cat.’
Clausal complements of predicates of utterance, propositional attitude, (acquisition of) knowledge, immediate perception and the like, are introduced by the Hungarian-origin general subordinator hoď (8a). Like in Hungarian, this subordinator is also employed to introduce several types of adverbial clauses (8b: reason clause) and, optionally, embedded interrogative clauses (8c) and embedded polar questions (8d). The latter are – obligatorily, unless an alternative construction is used – marked by the question enclitic -i, which is also borrowed from Hungarian. The subordinator hoď may also precede various pre-Hungarian subordinators that introduce embedded
274
Viktor Elšík
commands and other clausal complements of manipulative predicates (8e), and purpose clauses (8f). Unlike in Hungarian, however, the subordinator hoď cannot introduce such clauses by itself. (8)
a. Halíjom, hoď má n- aná le understand.pfv.1sg comp already neg bring.1sg.fut 3sg.m.acc uppe gódi. on brain ‘I understood that I will not persuade him any more.’ b. Daráhi, hoď našlíja o lóvo. fear.1sg.rem comp get.lost.pfv.3sg def.m money(m) ‘I was afraid that the money had gotten lost.’ c. Na šunde láčhe, (hoď ) ko vičinel taj so. neg hear.pfv.3pl well comp who shout.3sg and what ‘They did not hear well who was shouting and what.’ d. Na džanav, (hoď ) muká -i man tutar te neg know.1sg comp let.1sg.fut -q 1sg.acc 2sg.abl comp čumiden. kiss.3pl.subj=inf ‘I do not know whether I will let you kiss me.’ e. Phenďa mange, (hoď ) khére nek áčhovav. say.pfv.3sg 1sg.dat comp at.home opt stay.1sg.subj ‘S/he told me to stay at home.’ f. Site le papaleg uppe alakhes, (hoď) káj nek must 3sg.m.acc again upward find.2sg comp where opt džanesahi le te phenen. know.2sg.rem 3sg.m.acc comp say.3pl.subj=inf ‘You have to discover it again, in order to be able to say it.’
Due to pattern replication from West Iranian or Greek, complement clauses of modal predicates were finite in the early European stages of Romani: the subordinate verb was introduced by an indigenous non-factual complementizer and showed subject person–number agreement with the matrix verb (Matras 2002: 161). Pattern replication from Hungarian has resulted in a development of a non-finite complement form in Rumungro, through fossilization of a frequent finite form of the subordinate verb: the subordinate verb now invariably shows third plural subjunctive inflections, irrespective of the person–number of the matrix verb. This non-finite construction, which may be termed the subjunctive infinitive (or the “new” infinitive, Boretzky
Hungarian Rumungro
275
1996), encodes not only clausal complements of modal predicates (cf. 6, 8f) but also clausal complements of some manipulative verbs (cf. 8d) and tightly integrated same-subject purpose clauses. The Hungarian-origin infinitive in -ňi (see Section 5) is used in identical syntactic contexts as the subjunctive infinitive (9a–c). Like the infinitive in Hungarian, the Hungarian-origin infinitive in Rumungro does not allow any complementizer. (9)
a. Kam-áhi dógoz-ňi. want-1sg.rem work(v)-inf ‘I would like to work.’ b. Muk-j-a l-a ďij-áz-ňi leave-pfv-3sg 3sg-f.acc song-v-inf ‘S/he let her sing.’ c. Dža-s huhur-áz-ňi. go-1pl mushroom-v-inf ‘We are going to go and collect mushrooms.’
Pattern replication from Hungarian has also occurred in relative clauses. Although Selice Rumungro relativizers are formally identical to interrogatives, whereas Hungarian relativizers are not, the former partly copy the “ontological” restrictions of the latter: human head nouns usually select a person pro-word (‘who’) as a relativizer in Rumungro, while non-human head nouns mostly select a thing pro-word (‘what’).
9. Lexicon Out of a much larger inventory of early loanwords into Romani (as attested in different Romani dialects), Rumungro of Selice retains ca. 20 loanwords from Iranian languages, ca. 10 loanwords from Armenian, and over 35 loanwords from Greek. In addition, there are over 30 loanwords from South Slavic, which are mostly not shared with other dialects of Romani. Most of the pre-Hungarian loanwords are nouns, while verbs and adjectives are less numerous; only relatively few pre-Hungarian function loanwords have been retained (see Section 6). While there are a few stable noun loanwords from the secondary L2s of Selice Rumungro speakers (e.g. pepšo ‘black pepper’ from Czech), and while nonce borrowing of nouns and verbs from these languages is rather common, the by far most important current source of loanwords is Hungarian.
276
Viktor Elšík
Hungarian loanwords include basic vocabulary in domains such as body parts, bodily functions, kinship, or physical properties (e.g. ‘knee’, ‘to breathe’, ‘son-in-law’, ‘weak’). Unlike some Romani varieties that employ internal word-formation processes to create a layer of secret vocabulary in certain semantic domains (cf. Matras 2002: 223), Rumungro does not seem to avoid loanwords (such as čëndéri ‘policeman’) in these domains. Instances of pattern replication without matter borrowing in complex referring expressions are exceptional, e.g. sobota-kurko [Saturday-Sunday] ‘weekend’ calquing local Hungarian szombat-vasárnap. An overwhelming majority of Hungarian compounds are borrowed rather than translated, e.g. fog-orvoš-i ‘dentist’ < fog-orvos [tooth-doctor], though translations of lexicalized preverb–verb collocations are common (see Section 5). Some Hungarian compounds may be decomposed into adjective–noun collocations, e.g. világ-ik-o háború [world-adj-nom.sg.m war] ‘world war’ < világháború [world-war]. Phraseological idioms are commonly translated from Hungarian; for an example see (3). As several Hungarian types of greetings and similar expressions are missing in the traditional Rumungro culture, some speakers have started to fill in the “gap” by using Hungarian expressions, e.g. szia ‘hi; bye’, jó étvágyat ‘bon appetit’. Some indigenous politeness expressions are used in wider contexts due to cultural contact. For example, palikerav ‘I thank; I greet’ is not traditionally used after being served a meal or coffee at home, but some Rumungri would now use it in this context, as the local Hungarians do.
10. Conclusions The sociolinguistic situation of all Romani varieties is highly favourable to contact-induced developments, since almost all Romani speakers are bilingual in the relatively prestigious languages of the dominant “matrix” populations and since, at the same time, Romani linguistic ideologies are relatively tolerant of borrowing in most functional domains. Moreover, the long-settled Roms of the Hungarian regions of Slovakia have developed a strong orientation towards Hungarian cultural models, which facilitates Hungarian-induced linguistic changes in the few extant Hungarian Rumungro varieties, including Rumungro of Selice. This concluding section is an overview of various types of contact-induced developments that have affected the structure of this particular Romani variety.
Hungarian Rumungro
277
Both matter borrowing and pattern replication are well attested in Rumungro. Lexical matter borrowing, i.e. borrowing of syntactically free symbolic form–function units, is common in Rumungro with all grammatical classes of content words (verbs, nouns, adjectives, and manner adverbs) and with most classes of function words. Borrowed adpositions are rare, however, and there is no matter borrowing of personal pronouns or of the definite article. Also, only adverbial categories of reflexive, deictic, interrogative, and indefinite pro-words are lexically borrowed (so not, for example, adnominal or pronominal demonstratives). Lexical matter borrowing of paradigmatically related pairs (or sets) of words may result in what I have termed affix extraction, i.e. indirect or lexical affix borrowing. Note that affix extraction assumes not only an adoption of an affix within loanwords from a certain L2 and its (potential) paradigmatic identification, but also its analogical, language-internal, extension to other etymological compartments within the L1 lexicon. There are numerous instances in Rumungro of lexical matter borrowing of morphologically complex Hungarian words (e.g. derived frequentative verbs) whose affixes do not extend to non-Hungarian bases, and which are therefore not considered to be instances of affix borrowing. Categories whose affixal markers did get extracted in Rumungro are nevertheless numerous, and include derivational as well as inflectional categories. Extracted affixes of pre-Hungarian origin are: nominative noun inflections; a passive participle marker; non-inflectional loan-verb and loan-adjective adaptation markers; and markers deriving: feminine human nouns; relational and attenuative adjectives; ethnic adverbs; and ordinal numerals. Affixes extracted from Hungarian loanwords include: an infinitive inflection; a non-inflectional loan-verb adaptation marker; and markers deriving: action and artificial nouns; active de-verbal adjectives; denominal and causative verbs; similative adverbs; and several unproductive derivational markers. The patterns of analogical extension of the extracted affixes to different etymological compartments of the Rumungro lexicon are rather varied, and they are not discussed in any detail in this chapter. I should only like to point out here that several extracted affixes appear to have been “activated” to apply to loanwords from a chronologically following L2. For example, the South Slavic-origin suffixes -av-, -n-, and -kiň- apply to Hungarian loanwords (see Section 4), and the Hungarian-origin suffix -ál- applies to Slovak and Czech loanwords (see Section 5). Also left out of discussion were the details of various processes of language-internal re-analysis that are involved in extraction. For example, the Rumungro de-nominal verb-deriving suffix
278
Viktor Elšík
-áz- (see Section 5) does not correspond to any allomorph of its Hungarian source, as its extraction involved a re-analysis of its boundary, e.g. cigarett-áz-in- ‘to smoke cigarettes’ ← cigarett-a ‘cigarette’ (< cigarettá-z ← cigaretta). Although affix extraction is the source of a greater part of borrowed Rumungro affixes, there are also affixes the borrowing of which appears not to have been mediated by lexical matter borrowing. This is the case of the Hungarian-origin superlative prefix (see Section 4) and of several pronominal prefixes of Hungarian and South Slavic origin (see Section 6). These affixes must have been borrowed directly, “by themselves”, since there are no paradigmatically related pairs of lexical borrowings that could have served as a source of their extraction. The process of direct affix borrowing, whose possibility is sometimes claimed to be in need of demonstration (cf. Winford 2003: 6164), appears to be restrained by certain structural factors. Note especially that direct affix borrowing only takes place in Rumungro when the resulting morphological construction is, in effect, a “semicalque” on a semantically equivalent L2 construction. For example, the Rumungro proword vala-káj ‘somewhere’ consists of a directly borrowed indefiniteness prefix and an indigenous local interrogative base, which “calques” the local interrogative base of the Hungarian model vala-hol. Several types of selective borrowing are attested in Rumungro. First, only some inflectional forms of Hungarian nominals may be borrowed without a parallel borrowing of the base forms of these nominals, e.g. the sublative új-ra ‘again, anew’ (lit. ‘onto a/the new one’) but not the nominative *új ‘new’. Not surprisingly, the borrowed inflectional forms are those that fulfil adverbial or discourse-related functions. Second, only some allomorphs of an affix, or alternatives within an affix paradigm, may be borrowed. While structural factors such as the degree of transparency in the source language are known to play a role here (Winford 2003: 9197), sometimes functional factors are clearly involved as well. For example, Rumungro borrows the distal deictic-contrast prefix am- from Hungarian (see Section 6) without borrowing its proximal, equally transparent, counterpart em-. Finally, several function words are borrowed only in some of their source functions. Sometimes differences in the distribution of the source word appear to be responsible for selective borrowing, as in the case of the Hungarian question clitic (see Section 8). In other instances, however, selective borrowing reveals functional motivations. For example, the Hungarian coordinators meg and pëdig ‘and; but, in turn’ have only been borrowed in their adversative uses, in which meg is postpositive and pëdig prepositive (see Section 8). Their con-
Hungarian Rumungro
279
junctive uses, in which meg is prepositive and pëdig postpositive, are unattested in my Selice Rumungro text corpus. Selective borrowing confirms that borrowing is motivated by functional, as well as structural, factors. In addition to matter borrowing, Rumungro frequently replicates grammatical patterns (constructions and categories) of its L2s without necessarily borrowing the linguistic matter that encodes these patterns. As discussed in Section 3, Greek was the major structural model for Romani in this respect. There are not many constructions in present-day Rumungro whose origin is South Slavic: the negative agreement with, and the de-interrogative structure of, negative pro-words are rare examples. Numerous syntactic patterns, on the other hand, have been modelled on Hungarian, the current L2: the socalled subjunctive infinitive; the syntactic category of preverbs and many individual preverb constructions; encoding of various case relations; absence of case agreement in numeral constructions; negation of phasal adverbs; ontological restrictions on relativizers; certain pragmatic and syntactic aspects of linear constituent order; and more. Pure replication of morphological constructions is rare, being represented especially by occasional translations of Hungarian compounds, including the reciprocal pronoun. However, replication from Hungarian is responsible for the creation or elaboration of some morphological categories (associative plurals in nouns, frequentatives in verbs, and orientation in spatial adpositions and pro-words) and for the reduction of others (gender in anaphoric pronouns and feminine derivation in nouns denoting animals). Also, the retention and productivity of some inherited morphological categories (degree in adjectives and causatives in verbs) are likely to have been motivated by pattern replication from Hungarian. Matter borrowing and pattern replication frequently go hand in hand, conspiring, so to speak, to make the L1 more like the L2. To mention some examples: Hungarian-origin adpositions retain their postpositioning in Rumungro; unintegrated Hungarian numerals tend to retain their property of requiring singular head nouns; the replicated category of preverbs is enhanced by a few lexically borrowed members; direct affix borrowing results in “semicalques” of the model constructions (as discussed above); new phonemes, which are first adopted within loanwords, may be later extended to other lexical compartments, copying to some extent the phonotactics and morphophonogical rules of the model language; and so on. However, matter borrowing and pattern replication may also result in competing constructions, as in the case of the two Rumungro infinitives (see Section 8), one of which (the Hungarian-origin infinitive) does not allow a complementizer, while the other (the replicated subjunctive infinitive) requires one.
280
Viktor Elšík
Abbreviations 1 2 3 abl acc adj comp cond conj cop dat def emph f fut gen imp inf itr loan
first person second person third person ablative accusative adjective(-deriving marker) complementizer conditional conjunction copula dative definite article or conjugation reflexive–emphatic pronoun feminine future genitive imperative infinitive intransitive loan-verb adaptation marker
loc m n neg nom obl opt pfv pl poss pres pret q rem sg soc subj tr v
locative masculine noun negator or negative form nominative oblique optative–permissive particle perfective inflection or particle plural possessive present preterite question particle remote tense singular sociative subjunctive transitive verb(-deriving marker)
Notes 1. The chapter is based on my linguistic research on Hungarian Rumungro that was carried out during short but numerous fieldtrips to Selice, Slovakia, between 1997 and 2007. I wish to thank the late Milena Hübschmannová for introducing me to the Selice Rom community; Július Lakatoš and Alena Krészová for their hospitability and native speaker expertise; the Roma Culture Initiative of the Open Society Institute, Budapest, for their financial support of my Rumungro research in 20012002; and Adéla Gálová for her help with Hungarian example sentences. The descriptive sources on Hungarian that I have consulted include Abondolo (1988), Kenesei, Vago, and Fenyvesi (1998), Siptár and Törkenczy (2000), and Tompa (1968). 2. Although all Rumungro varieties have been influenced by Hungarian, most Rumungro speakers presently live in ethnically Slovak parts of Slovakia and are Slovak bilinguals; an overwhelming majority of Rumungro communities in Hungary and in the Hungarian parts of Slovakia have undergone language shift to Hungarian.
Hungarian Rumungro
281
3. The Rumungro associative plurals are similar in form to nominative plural agreement (Suffixaufnahme) forms of inflectional genitives of the respective nouns, though they differ from them in some interesting structural details (see Elšík and Matras 2006: 322323). 4. The Greek-origin suffix *-(V)s-, which appears to have been the marker of perfective adaptation of all loan-verbs and of non-perfective adaptation of intansitive loan-verbs (Matras 2002: 130), has acquired novel functions in Rumungro: it is now an integral part of the suffix -(i)sal-, which serves as a stem extension in several valency-changing or aktionsart derivations, e.g. cid-isaj-ov- ‘to stretch itr’ (anticausative) ← cid- ‘to pull’, térň-isaj-ár- ‘to make young’ (factitive) ← térn-o ‘young’, khand-isaj-ov- ‘to stink intensively’ (intensive) ← khand- ‘to stink’. 5. Although Kenesei, Vago, and Fenyvesi (1998: 357358) describe the Hungarian suffix -ál- as a de-nominal verb-deriving marker, it in fact a verb-adapting suffix, which is synchronically distinct from the de-nominal verb-deriving suffix -(V)l. 5. When preposed to the verb they modify, Hungarian preverbs are orthographic prefixes. Nevertheless, they are syntactically free elements. 6. However, oblique case forms of the pronoun have remained differentiated for gender, e.g. the accusative le ‘him’ vs la ‘her’ (cf. Hungarian őt ‘him, her’). 7. The expression of ‘not yet’ as a negation of ‘still’ is congruent with Hungarian, but is likely to be pre-Hungarian. 8. This contrasts with the contact-induced postpositioning of inherited prepositions in some Romani dialects influenced by postpositional languages such as Turkish or Finnish (cf. Matras 2002: 206).
References Abondolo, Daniel 1988 Hungarian Inflectional Morphology. Budapest: Akadémiai Kiadó. Boretzky, Norbert 1996 The “new infinitive” in Romani. Journal of the Gypsy Lore Society, Fifth Series, 6, 151. 1999 Die Gliederung der Zentralen Dialekte und die Beziehungen zwischen Südlichen Zentralen Dialekten (Romungro) und Südbalkanischen Romani-Dialekten. In: Halwachs and Menz (1999: 210276). Elšík, Viktor, Milena Hübschmannová, and Hana Šebková 1999 The Southern Central (ahi-imperfect) Romani dialects of Slovakia and northern Hungary. In: Halwachs and Menz (1999: 277390). Elšík, Viktor, and Yaron Matras 2006 Markedness and Language Change: The Romani Sample. Berlin: Mouton de Gruyter.
282
Viktor Elšík
Halwachs, Dieter W., and Florian Menz (eds.) 1999 Die Sprache der Roma. Perspektiven der Romani-Forschung in Österreich im interdisziplinären und internationalen Kontext. Klagenfurt: Drava. Hübschmannová, Milena, and Vít Bubeník 1997 Causatives in Slovak and Hungarian Romani. In: Yaron Matras, Peter Bakker, and Hristo Kyuchukov (eds.) The Typology and Dialectology of Romani, 133145. Amsterdam/Philadelphia: John Benjamins. Kenesei, István, Robert M. Vago, and Anna Fenyvesi 1998 Hungarian. London/New York: Routledge. Matras, Yaron 1994 Structural balkanisms in Romani. In: Norbert Reiter, Uwe Hinrichs, and Jiřina van Leeuwen-Turnovcová (eds.), Sprachlicher Standard und Substandard in Südosteuropa und Osteuropa. Wiesbaden: Harrassowitz. 195210. 1995 Connective (VS) word order in Romani. Sprachtypologie und Universalienforschung 48 (1): 189203. 2002 Romani: A Linguistic Introduction. Cambridge: Cambridge University Press. Schrammel, Barbara 2005 Borrowed verbal particles and prefixes: A comparative approach. In: Barbara Schrammel, Dieter W. Halwachs, and Gerd Ambrosch (eds.), General and Applied Romani Linguistics: Proceedings from the 6th International Conference on Romani Linguistics, 99113. Munich: Lincom Europa. Siptár, Péter, and Miklós Törkenczy 2000 The Phonology of Hungarian. Oxford: Oxford University Press. Tompa, József 1968 Ungarische Grammatik. The Hague: Mouton. Winford, Donald 2003 An Introduction to Contact Linguistics. Oxford: Blackwell.
Grammatical borrowing in Manange Kristine A. Hildebrandt
1. Background Manange, also known by its endonym ŋjeshaŋ, ŋjeshaŋte, or ŋjaŋmi ‘our language/our people,’ is a Bodish language of the Bodic subphylum of TibetoBurman. It is spoken in northern central Nepal, and it is grouped with other Tamangic (or ‘Gurungic’ or ‘TGTM’) languages, shown in Figure 1 (van Driem 2001; Bradley 1997; Noonan 2003).1 Manange is spoken by members of a single ethnic group of under 5,000 speakers, located in the northern Manang district. Geographically, Manang is known as the Inner Himalayan Valley, as it is surrounded to the south, east and west by the Annapurna mountain range. Manang is culturally and linguistically heterogeneous, divided into three ethnic group areas: Gyasumdo to the south, the high elevation Nar valley to the north, and the upper ŋjeshaŋ valley in the west (Snellgrove 1961). Although Manange peoples live in all portions of the Manang District, the ŋjeshaŋ valley is considered the traditional area of Manange habitation. Both Gurungs and Mananges (or Manangis, Manangpas, Manangbas and Manangbhots by Indic peoples) are the dominant ethno-linguistic groups of Manang.
Tibeto-Burman
Bodish
Tebetan Complex (incl. Sherpa)
Ghale
Figure 1. Genetic affiliation of Manange
“Tamangic” Tamang Seke Gurung Kaike Thakali Gyalsumdo Manange Chatyal Nar/Phu
284
Kristine A. Hildebrandt
In terms of endangerment status, the Manange language can currently be considered as small but relatively viable, with some prospect for endangerment (using (Kincade 1991) as a model). Although the speaker population is under 5,000, there seems to be continued transmission of Manange to younger generations (albeit bilingual), combined with some small-scale displacement via emigration of some generations of speakers from traditional Manang to urban Kathmandu. Factors contributing to an observed small-scale loss of Manange include the rise of access to formal education in Nepali, as well as the general prestige of Nepali in terms of socio-economic advancement. Factors contributing to retention of the language include positive within-ethnic group identity and various prestige factors, including the comparative wealth and social status that Mananges have accrued as entrepreneurs. The history of language contact in Nepal is complex, and the results of this long-term inter-mingling of languages have had varying consequences on typological and genetic features of different Tibeto-Burman languages located there. Noonan (2003) charts the different types of grammatical borrowing in a number of Tibeto-Burman languages from different sub-phyla. Of the three main types of contact scenarios, the oldest situation is between Himalayish languages (including Kiranti, Kham, Magar, Chepang, Newar, and others), whose speakers have been long-time residents of Nepal. A more recent type of contact is between speakers of the Tibetan-type languages of the Bodish sub-group, including Manange (i.e. within-family contact). These peoples are more recent immigrants to Nepal, having migrated within the last two millennia, and occupying territories that are in close proximity. A still more recent, and different, type of contact situation in Nepal has been between speakers of Tibeto-Burman languages and Indo-European languages like Nepali. Although Nepali was already well-established in western Nepal, there has been more recent contact of this third type in eastern and central Nepal. Now, as the influence of Nepali (and perhaps other non-TibetoBurman languages) spreads throughout Nepal, cross-family contact is as (or more) likely as within-family contact. Despite the rather (geographically) remote location of the Manang villages, there is evidence that Mananges have been in regular contact with speakers of other languages (Indic and Sinitic) for a long time. In 1956, David Snellgrove, of the School of Oriental and African Studies, undertook a six month journey through Nepal to update map information originating from the Survey of India and to study Buddhist art and scriptures. He spent some time in the Manang District, and he was initially intrigued by the lack of surprise displayed by Mananges when they first encountered him. He also noted that
Manange
285
Manange youths “spoke Nepali willingly and fluently” (1961: 205). Snellgrove also noticed silks from mainland China and Singapore adorning the walls of local gompa buildings, suggesting some trade-oriented contact with other Asian peoples. Snellgrove soon learned that Mananges had significantly more contact with the world beyond the Nepalese borders than did many other indigenous groups, holding posts in the Indian Army and having unique travel rights to Malaysia and Singapore. In recent generations, it has become commonplace for many Mananges to migrate temporarily or permanently to the Kathmandu valley, or to lower elevations within Manang during winter, to benefit from longer growing seasons. In winters, women and children especially, stay in low elevation villages where Nepali (Indo-European) is spoken, and men may travel to other regions in Nepal or to India (or beyond) for work (Rogers 2004). Although some Mananges (and other peoples) do remain in Manang year-round, this number seems to be declining as the years go by. As a result, for part of the year, many Mananges are surrounded by, and use, Nepali either in urban Kathmandu or in other lower elevation villages in Manang. Another relevant factor for Manange language contact with Nepali is education (Hildebrandt 2003, 2006). There is one school in each larger Manang village nowadays, and instruction is in Nepali. In addition, a number of adults who live in Manang (traditionally men, but increasingly women too) have had some education either in Kathmandu or abroad. These opportunities for formal education have lead to frequent and long-term contact with other languages, like Nepali, Hindi and increasingly, English. Recently, Manang has become a tourist hot-spot because the popular “Annapurna Long Circuit” bisects the district. As a result, a tourist-driven economy has emerged where wealthy Mananges build elaborate lodges to host foreign trekkers. Other related tourist-oriented businesses have grown in the area, including guided tour operations, porter services, and a solar-powered cyber-cafe. Some aspects of this new economy are grounded in Nepali language use (e.g. interaction with tour guides and porters), and so the economic benefit of speaking Nepali has grown there. Not all Mananges benefit equally from this new trekking economy. Many Mananges still live traditional, subsistence-farming lives, usually because they live in areas that are too far off the main trekking route to benefit from the tourist industry in the way that more strategically located residents can. These Mananges claim to use Nepali only sometimes, (e.g. with outside visitors) Another observation is the recent immigration of Tibetans, Lhomis and Nar-Phus to the Manang villages. They have come to Manang in search of
286
Kristine A. Hildebrandt
better economic opportunity; they rent vacated (Manange-owned) houses and farm the land in a kind of share-cropping situation. Mananges report that these new residents adopt Manange for local use, or else use Nepali with them. My own (limited) interaction with these people has been in the Manange language, and not in their traditional languages, nor in Nepali. This report focuses on the one-way effects in Manange of language contact with Nepali. Although there is preliminary evidence of contact-induced changes between Manange and other T-B languages (e.g. some lexical borrowing from unknown dialects of Tibetan), the effects of contact with Nepali are easier to pinpoint and document. Further investigations can reveal the potential effects of this contact on the Nepali spoken by different segments of the diverse Manange ethnic community.
2. Phonology The phonological structure of Manange in many ways typifies that of the Bodish languages: there is no contrastive voicing opposition for obstruents, there is an alveolar and post-alveolar opposition in consonant place of articulation, and there is a velar nasal in word-initial position (e.g. 4ŋi ‘two’; 1ŋʌ ‘1.sg’). Manange also has a voiceless plain and aspirated retroflex plosive, that while contrastive in word-initial position in basic vocabulary (e.g. 1ʈu ‘sit/stay’ vs. 4 ʈu ‘six’), is still marginal in overall lexicon frequency. There is considerable evidence that the retroflex is one of the more reliable features of South Asia as a linguistic area (Masica 2001; Noonan 2003). It has probably entered into the Bodish languages via contact with Indic languages (which in turn acquired it from Dravidian). The most interesting case of contact-induced structural change in Manange phonology is not obviously borrowing from Nepali, but rather a case of loss or simplification (likely via analogical leveling). This has been documented in Hildebrandt (Hildebrandt 2003; 2004) as a phonetic and phonological merger of the tone system. The properties of the tone system employed by more linguistically conservative speakers is as follows. All words (both native and old loans, both mono- and disyllabic) fall into one of four tones, illustrated in (1). As (1) shows, tone /1/ and /2/ words show low and high pitches, respectively. The words from the two (falling) contour tones have an additional defining property in that with tone /3/ words, if the initial consonant is an obstruent, it is unaspirated, and with tone /4/ words, the initial obstruent is aspirated. However, this distinction is not retained with sonorant-initial words, which
Manange
(1)
287
Manange tones Tone
Pitch properties
Initial onset consonant properties
Example
1 2 3 4
Low Level High Level Very High Falling Mid-High Falling
N/A N/A Unaspirated if Obstruent Aspirated if Obstruent
ʈu ‘sit/stay’ ʈu ‘thread’ ʈu ‘cereal’ ʈu ‘six’
are found in all four tones without any aspiration or voicing differences (e.g. the near-minimal set 1ŋje ‘chew’; 2ŋi ‘seven’; 3ŋje ‘milk’; 4ŋje ‘spill’). With Mananges who have grown up in a more intense environment of Nepali bilingualism (mainly those Mananges who were born/raised in Kathmandu), the structure of the tone system shows marked changes. Hildebrandt (2003) demonstrates that urban speakers show a large-scale phonetic merger of the two contour tones into a two-way high-low opposition. In addition, the conceptualization and grouping of words into different melody groups is considerably fuzzier than for rural Mananges of the same age group. It is not at all obvious that the altered tone system is simply one symptom of a larger process of language loss (i.e. shift to Nepali). Manange in the urban environments appears to be maintained in a situation of diglossia, whereby its place in Manange life is firmly rooted in domestic, private environments, while Nepali is the language of necessity in public domains. Nevertheless, this structural result (along with others described below) appears to be a consequence of such a maintenance scenario. It is also not obvious that urban Manange is borrowing anything from Nepali phonology. Nepali has no tone, and in fact has a four-way obstruent voicing distinction (voiceless plain, voiceless aspirated, voiced plain and voiced aspirated). There is no evidence that urban Mananges are incorporating obstruent onset voicing into their production of Manange. Rather, lexical frequency may play a role in determining which words evidence phonetic pitch merger and in determining the pitch properties of the emergent two-way system. There is some evidence of a possibly emergent iambic stress pattern (noncontrastive) in Manange words, perhaps via contact with Nepali. In Nepali polysyllabic words, (phonetic) stress is initial if all syllables of word are of equal syllable weight (Acharya 1991). If non-initial syllables are of certain (progressively heavier) weight, then stress falls right-ward, suggesting an iambic tendency.
288
Kristine A. Hildebrandt
In Himalayish T-B languages, stress and tone favor an overall trochaic (initial) pattern (cf. Bickel 1998; 2003). For example, in Kiranti languages, main stress is almost always initial. In Bodish languages, the tone feature of the initial element (syllable, morpheme) is carried across all other bound units. In Manange, the tone pattern retains this initial/trochaic preference, and most disyllabic words carry a main stress on the initial syllable. But there are also some disyllabic (nouns) that are clearly stressed on the final syllable, as in example (2), with pitch re-set, vowel amplitude/intensity and duration as indicators of this stress (cf. Hildebrandt 2003; 2004). (2)
Final Main Stress Form Meaning 2 nja.Õta ‘chain’ 2 to.Õsoŋ ‘now’ 3 ŋjo.Õkroŋ ‘breast’ 3 to.Õɾe ‘grave’ 4 ko.Õʈe ‘button’ 2 tʃep.Õkjel ‘vulture’ 3 ŋo.Õkɾoŋ ‘forehead’
3. Nominal structures I treat the different patterns of case-marking in Manange in this section, even though it is essentially a topic of argument structure alignment, and appears in the typology section of the database. Manange is like the other Bodish (Himalayish) and Indo-European languages of the South Asian linguistic area in showing some variation of a (split) ergative–absolutive alignment pattern (cf. Masica’s ‘ergative belt’ 2001: 250251). Split ergativity is actually reconstructed by DeLancey (1989) back to Proto T-B, so its presence in Manange is not necessarily attributed to contact with Nepali. However, the different patterns of ergative case marking in the rural and urban Manange communities is of interest. Rural Mananges show a pattern in their speech of split ergativity that aligns with modality. The A argument of a transitive verb in realis mode (i.e. perfective and perfective progressive aspects, simple present ‘tense’) hosts the ergative enclitic, while in irrealis mode (future, immediates, deontics, etc.) is absolutive (zero) marked, shown in examples (3)–(5) (examples from Hildebrandt 2004: 99100)
Manange
(3)
Realis 1 2 mriŋ=tse naka 2puŋ 2kol-tsi. woman=ERG chicken egg boil-perf ‘The woman boiled the egg.’
(4)
Irrealis (future) 1 mriŋ*=tse 2naka 2puŋ 2kol(-pʌ). woman*=ERG chicken egg boil(-nom) ‘The woman will boil the egg.’
(5)
Irrealis (immediate) 1 1 1 ŋʌ*=tse nʌkju=ri 2prim-pi lʌ-tsi 1(sg)*=ERG dog=loc hit/kick-imm do-perf ‘I prepared to/was about to hit/kick the dog.’
289
With urban Mananges, there is no such split. The A arguments of transitive verbs host the =tse enclitic, regardless of any aspect or modality distinctions. In this sense, the ergative–absolutive pattern is different both from rural Manange and from Nepali, as Nepali has a split-ergative system aligning with aspect (arguments of perfective-marked transitive verbs show -ley ergative marking while imperfective verbs show absolutive/zero marking). It appears then that there is a process of overgeneralization of ergative casemarking for urban Mananges, whereby it has become a general marker of transitivity. A second likely contact pattern with nominal structures is shown by different patterns of NP constituent ordering across the rural and urban Manange communities. This more closely resembles what Matras and Sakel term ‘pattern borrowing’, or the adoption of a strategy or structure from another language (2007). Nepali has a separate class of lexical adjectives and they are pre-nominal in order (e.g. miʈo kana ‘tasty food’). And in fact this is the general pattern with Indic languages of South Asia. In Bodish languages, the situation is slightly more complex, as both NA and AN order are attested. Nar-Phu is strongly NA in order, while Chantyal and Tamang show AN. It is generally assumed that the pre-nominal order in Bodish is the newer pattern via contact with Indic languages (Bickel 2001; Masica 2001; Noonan 2003).
290
Kristine A. Hildebrandt
Manange again shows a now-familiar split across speaker communities. The rural population shows overwhelmingly NA ordering in attributive NPs (e.g. 1nʌkju 1tjʌ-pʌ dog big-nom ‘(the) big dog’). This pattern is also supported in Hoshi with a speaker from Prakaa Manang (Hoshi 1986: 212). Urban Mananges show overwhelmingly AN ordering (e.g. 1tjʌ-pʌ 1nʌkju big-nom dog) (Hildebrandt 2004; Genetti and Hildebrandt 2004).
4. Verbal structures The changes to verbal structures in Manange again fit a pattern-borrowing type, and one of these (the dependency between negation and aspect marking) is evident mainly in the urban community of Manange speakers. In most Bodish languages the periphrastic strategy of valency increasing (causation) is the main (and usually) only available strategy (Noonan 2003), shown for Manange in example (6) (example from Hildebrandt 2004: 107). (6)
Manange Periphrastic Causation Strategy 1 1 amʌ=tse lʌ-tse 1ŋʌ=tse 1taŋ 1pja-tsi. mother=erg do-cc 1sg=erg floor clean-perf ‘My mother made me clean the floor.’
In the periphrastic strategy, the first clause contains an ergative-marked A and the verb 1lʌ ‘do’, which hosts the clause chaining suffix -tse. The matrix clause carries the aspect (perfective) suffix -tsi. Both rural and urban Mananges use this strategy, and they also employ another (less productive) causation strategy that is more morphological in structure, shown in (7) (example from Hildebrandt 2004: 106): (7)
Manange Morphological Causation Strategy 1 ŋʌ=tse 3tʃʌ 1le 1lʌ-tsi. 1.sg=erg tea warm do-perf ‘I made the tea warm/warmed the tea.’
Here, the verb 1lʌ ‘do’ is used in a compound structure and it carries the aspect affix. This structure is noticeably absent from other Bodish languages (except for Chantyal, which also shows other structural borrowing from Nepali). In Nepali causation is signaled only through morphological means (with a suffix -āu):
Manange
291
(8) Nepali Morphological Causation (Acharya 1991: 168) Subhadrā suśīla-lāī bhāta khũw-āu-chin. Subhadrā Suśīla-dat rice eat-caus-3sgpres.fem ‘Subhadra makes Susila eat rice’ The second contact-induced change is seen only with the urban Manange speakers, so they are apparently modeling their pattern of (Manange) verbal inflection based on that in Nepali. For rural Manange speakers, the morphological coding of aspect on the verb is dependent on negation: negative marked (prefixed) verbs do not show aspect marking, with the resulting difference in (9) and (10): (9) Affirmative2 1 ŋʌ=tse 1kola=ri 3ʃitaŋ 1lʌ-tsi. 1.sg=erg child=loc scold do-perf ‘I scolded the child.’ (10) Negative 1 ŋʌ=tse 1kola=ri 3ʃitaŋ 1a-lʌ. 1.sg=erg child=loc scold neg-do ‘I did not scold the child.’ Urban Mananges do not acknowledge this dependency, and both negated and non-negated verbs can host the full range of aspect morphology (e.g. 3ʃitaŋ 1 lʌ-tsi and 3ʃitaŋ 1a-lʌ-tsi).
5. Other parts of speech Numerals in Manange follow a base-ten system (2tʃu ‘ten’; 4ŋiʃu two-ten ‘twenty’; 2sumtʃu three-ten ‘thirty’; 4plitʃu four-ten ‘forty’, 4ŋʌtʃu five-ten ‘fifty’, 4ʈuktʃu six-ten ‘sixty’, etc.). Consecutive counting within the individual bases follows a pattern of addition of single units to the multiple (e.g. 1 tʃukre ten-one ‘eleven’, 1tʃuŋi ten-two ‘twelve’, 1tʃupsẽ ten-three ‘thirteen’ etc.; 4ŋiʃu 4kri two-ten-one ‘twenty one’, 4ŋiʃu 4ŋi two-ten-two ‘twenty-two’, 4 ŋiʃu 2sẽ two-ten-three ‘twenty-three’ etc.). There is evidence from other T-B languages that such a decimal system in Manange may be a recent innovation, formed under pressure from similar systems in Indic languages. Tamang (Tamangic), in comparsion, has a semi-
292
Kristine A. Hildebrandt
complete vigesimal system, and Dzongkha, a Tibetan dialect and the national language of Bhutan, has a complete vigesimal system to the fourth power of the base (Mazaudon 2003). The decimal system in Manange has probabily been in place for awhile, as the phonotactic alternations between simple and complex numerals indicate. For example, the numeral ‘three’ in its bare form is 2sẽ, and in a compounded form ‘thirty’ is realized with a word-medial bilabial nasal coda (2sumtʃu). A similar situation is found with 4ʈu ‘six’, which is realized as 4ʈuktʃu ‘sixty’ with a velar plosive in word-medial coda position. Coda consonants are rare in Manange, due to diachronic erosion of syllable-edges (this diachronic development is frequently attested in many other Tibeto-Burman languages), and these alternations suggest that the lexicalization of these numerals in such a decimal structure took place at a stage when final codas were still present. Bodish languages are unlike other T-B languages (e.g. Himalayish) in that they do not possess a numeral classifier system. Manange seems to have borrowed its single classifier -ta from Nepali. Nepali has a classifier system of two: -janā for human count nouns: -ʈa for non-human count nouns (Acharya 1991: 100). Urban Mananges (optionally) use a segmentally altered form of the non-human classifier for both human and non-human count nouns: (11) Classifier 4 1 ŋi-t ha kola two-class child ‘two children’ 4 3 ʃi-t ha pʌle one-class leg ‘one leg’ In addition to the numerals, one of the interrogative pronouns in Manange shows matter borrowing (the borrowing of form rather than strategy) from Nepali: 2puŋ 2kʌtti (egg + many). The loanword 2kʌtti (< Nepali kati ‘few/ some/little bit’) is used in other parts of speech in Manange, for example as a loan-verb 2kʌtti 1lʌ many do ‘to try’.
6. Clause combining/syntax There is a some evidence of Nepali contact phenomena in Manange clause combining strategies. In Bodish languages, one productive way of signaling
Manange
293
adverbial clause linkage is via the use of converbal constructions, whereby one of two verbs is the matrix verb, and the other (with a non-finite, converbal marker) verb conveys manner information. This is seen in elicited structures such as (12) (Hoshi 1986: 301) and in narrative structures (13) (Hildebrandt 2004; 136), where there is a dual reading of sequential actions where the second action comes about in a causal relationship with the first. The converb is marked with the clause chaining suffix -tse and the following (resultative) verb (clause ) takes finite aspect marking.3 (12) Elicitation 1 juŋ 4tsoŋ 1lʌ-tse 2kje 1kʌ-tsi stone sell do-cc profit come-perf ‘I sold stones and made a profit.’ (or, ‘Because I sold stones, I made a profit.’) (13) Narrative 1 u 3ja 2tipal=ko 2ʃʌmlepre 1jʌ-tse 1lʌ-tse dist yak some=def forget go-cc do-cc 1 kim=ko ʌle 1lʌ-tse ʌtse tẽ 1ʈu 1mi. 3.pl=def seq do-CC like.this then stay evid ‘Having forgotten (about their friends), having done this, those yaks stayed in the valley.’ Converbal structures similar to the ones above are found in abundance in other Bodish languages.4 The use of another linker 2ta 3pi-na (lit. ‘what say-adv’) in ‘because’ adverbials in Manange appears to be a structural calque of Nepali kina bhane ‘because’ (lit. ‘why say’) (Hildebrandt 2004: 110): (14) With 2ta 3pi-na 1 2 ŋʌ=tse kristin=ri taŋ 1pin-tsi 2ta 3pi-nʌ 1.sg=erg Kristine=loc gift give-perf what say-adv 4 1 nese ki 2manaŋ=ri 1jʌ-pʌ-ro. tomorrow 3.sg Manang=loc go-nom-rep ‘I gave Kristine a gift because (it is said) she will go to Manang tomorrow.’ Mananges also make use of Nepali word/phrasal and clausal coordinators ra and ani. In Manange, words and phrases can be coordinated via (Bodish) tẽ or via Nepali ra:
294
Kristine A. Hildebrandt
(15) Borrowing of tẽ (Hildebrandt 2004: 109) 1 3 pʌ=ko tẽ pje=ko. husband=def conj wife=def ‘the husband and the wife.’ (16) Borrowing of ra (adapted from Hildebrandt 2004: 78) 1 4 4 nʌkju ɾʌ ʃi polpʌ=ɾi 1mo 1mu. dog CONJ one frog=indef cop evid ‘There was a dog and a frog.’ Clause coordination can be signaled via clause chaining (-tse) or by juxtaposing the first clause, along with the final, bare verb before the second coordinated clause, as (17) shows (unpublished text data): (17) Bare verb clause coordination tjʌpʌ 2prĩ 4ŋi ra 1nẽ 1ten-tsi. yeast put two day alone put/leave-perf ‘I put the yeast in (the mash) and left it alone for two days.’ One additional strategy of clausal coordination found only in conversation data is with the Nepali clause coordinator ani. Example (18) is taken from a conversation between two people who are talking about making homebrewed beer: (18) Clause coordination with ani Grandma: 2taŋ=ko 1kʌ-tsi nʌ 1a-tʌ? smell=def come-perf or neg-become? ‘Did (an alcohol) smell come or not?’ 2 Auntie: 1a-tʌ ani pe=ko neg-become and.then beer=def 3 2 naŋ=ri tsaŋ-tsi. inside=loc fill-perf ‘No, it didn’t, and then I put the beer inside (of a pot).’ In (18) the clause coordinator follows a bare verb ‘become’ and precedes a clause with an inflected verb. Likely the verb is bare because it is negated. In non-negated clauses, ʌni can conjoin finite clauses. In such cases, this would be a case of both matter and pattern borrowing. It is matter borrowing in that the coordinating form from Nepali is used. It is also pattern borrowing
Manange
295
in that the clauses on both sides of the conjunction may be finite, independent clauses (as in Nepali), whereas with other clausal coordination strategies in Manange the verb of the first clause is always non-finite (either bare or marked with -tse converb/clause chaining morphology).
7. Lexicon In a study on loanwords in Manange, Hildebrandt (2007) has found that approximately 12 percent of the ‘lexicon’ included in this study show some evidence of being borrowed. Of these, the vast majority, approximately 93 percent, come from Nepali, with the remaining loans from English or from other Tibeto-Burman languages.5 Most Nepali loans are nouns (84%), and with a small amount of verbs, property concepts and function words rounding out the list. Of the nouns, most belong to semantic fields like ‘animals’, ‘clothing/ grooming’ and ‘the modern world’, and (19) lists some examples of these. (19) Nepali loanwords in Manange6 Manange 1 tuŋ 2ŋ 1 poro 1 gohi kutti golbera makai kʌrila 2 cak 2 ʃakre
Nepali dumsi parewa gohi lamkhuʈʈe golbheɽa makai kaĩɽo cakk shakti
Meaning ‘porcupine’ ‘pigeon/dove’ ‘crocodile’ ‘mosquito’ ‘tomato’ ‘corn’ ‘gourd’ ‘buttocks’ ‘brain/mind’
Borrowed property concepts (both true adjectives and verb-like adjectives) include those listed in (20) (20) Borrowed property concepts Manange alo suntala 2 mi 2kʌtti sitʌri sita
Nepali ‘potato + orange/fruit’ kati = emphatic sittei si:dha
Meaning ‘orange’ ‘furry’ ‘free/no charge’ ‘straight (path)’
296
Kristine A. Hildebrandt
Even though verbs constitute a smaller amount of Nepali loans, they are somewhat more interesting structurally, as there are different strategies employed to enter them into Manange inflectional morphology. Some Nepali loans take a ‘dummy’ root (2ti), which itself can host aspect and modality morphology. Examples are shown in (21), where the bold-faced element is the Nepali loan-verb root. (21) Borrowed Verbs with 2ti 7 Manange rok 2ti-pʌ tʃuk 2ti-pʌ kelai 2ti-pʌ
Nepali rok-nu tʃukyau-nu kelau-nu
Meaning ‘to forbid’ ‘to separate’ ‘to sift/clean’
Other loans from Nepali are the first element in a verbal compound structure where the second piece is one of a small set of (native) semantically empty verbs (e.g. 1lʌ ‘do’; 2prĩ ‘hit/put/affect’). The second verb hosts the aspect/modality morphology: (22) Borrowed Verbs with 1lʌ or 2prĩ Manange 1 hai 1lʌpʌ 3 pu 1lʌ-pʌ poke 2prim-pʌ
Nepali hai aaunu puknu ‘untie’
Meaning ‘to yawn’ ‘to whistle’ ‘to tie’
The compounding process per-se is not borrowed from Nepali; in fact, verbal compounding is quite typical of languages throughout South Asia (and beyond). One ongoing question is what the motivation is behind the different Manange verb-words. One possible answer is that the verb 1lʌ ‘do’ occurs with loan-verbs that are transitive in Manange (where the A argument is ergativemarked). The verb ‘do’ does in fact function as a periphrastic causative marker in Manange (Hildebrandt 2004: 106107). Along these lines, a number of the Nepali source verbs in this database have the diphthong portion aau in the stem. Acharya (1991: 167168) notes that the derivational morpheme -aau- in Nepali derives a base verb form into a causative or ergative form (e.g. bannu ‘to become’ > banaaunu ‘to cause someone or something to become, to make’). In the Manange loan-verb cases, however, it is not obvious that the aau marker in the Nepali source forms is functioning as a causative marker here, as these Nepali verbs have no non-causative, (i.e. non-aau) counterparts
Manange
297
(i.e. searches through Turner’s dictionary have not revealed forms like semantically related, but non-transitive patnu or kelnu). In addition, some borrowed verbs (like ‘drown’ and ‘yawn’) are not transitive in Manange, as there is no ergative-marked A argument. As such, it is currently not clear what motivates the different verb-words in these constructions. Additional work in this area of loanword integration could yield clearer patterns and functions of these verb-words.
8. Conclusion The relatively long period of (both punctuated and regular) contact between Nepali and Manange has resulted in a number of structural changes to Manange. Some of these changes can be considered pattern borrowings, whereby a strategy is modeled on Nepali (e.g. the lack of a negation-finiteness dependency on verbs). Other changes are matter borrowings, where a form from Nepali is incorporated into Manange (e.g. loanwords, phrasal and clausal conjunction, the numeral classifier). Still other changes are not clearly borrowing at all, but rather structural loss or pattern (over-) generalization in the urban community of speakers, perhaps due to infrequent and interrupted access to Manange in a scenario of asymmetrical bilingual maintenance (e.g. the tone merger, the lack of a split-ergative pattern). An obvious next step in the documentation of contact-induced change is to more systematically note the ways in which the Nepali of both urban and rural Mananges may be altered. Such cases of Tibeto-Burman substratal influence on Nepali, while not regularly recorded, have been noted previously (e.g. Genetti 1999; Bickel 2001).
Abbreviations Superscript numerals indicate the tone category membership of the adjacent word. 1 first person 2 second person 3 third person adv adverbial caus causitive cc clause chainer
class conj cop dat def erg evid fem imm
classifier conjunction copula dative definite ergative evidential feminine immediate (irrealis)
298
Kristine A. Hildebrandt
indef loc neg nom perf
indefinite locative negative nominalizer perfective
pres rep seq sg
present reported speech seqiential singular
Notes 1. I would like to thank the Manange community for its ongoing assistance with my ongoing study of their language. I wish to also thank Michael Noonan and Balthasar Bickel for feedback and advice on this account. All errors are my own. 2. This strategy was not confirmed by me as a regular pattern until a 2004 fieldtrip, and these examples are from my field notes. The ‘urban’ pattern (negated verbs inflected for aspect) is also observed by Hoshi (2006) with a speaker from Prakaa Manang. 3. I have adapted Hoshi’s transcription of segments and tones to fit with other Manange examples. 4. Yet another “Bodish” way to signal causation is via nominalization of the clause of causation (Hildebrandt 2004: 118) 1
ŋʌ=tse 4mwi 3kjʌ=ri 1pim-pʌ 3kjʌ 3kola 3kju-pʌ. 1.sg=erg money 2.sg=loc give-nom 2.sg dress buy-nom ‘Because I gave you money, you will buy a dress’ 5. This percentage is from a database list of approximately 1,100 meanings. This study stands as a contribution to the Loanword Typology Project, organized by Martin Haspelmath and Uri Tadmoor at the Max Planck Institute for Evolutionary Anthropology (http://www.eva.mpg.de/lingua/files/lwt.html). 6. Some loans fall into the existing tone system (usually either the low level /1/ or high level /2/ tone), and words without a tone numeral mean that the tone status/ features are not yet established. 7. The au portion of the Nepali verb is the causative affix. The 2ti morpheme occurs frequently (but not exclusively) with causative-marked loan-verbs from Nepali.
References Acharya, Jayaraj 1991 A Descriptive Grammar of Nepali and An Analyzed Corpus. Washington, DC: Georgetown University Press. Bickel, Balthasar 1998 Rhythm and feet in Belhare morphology. Rutgers Optimality Archive Working Paper No. 287.
Manange 2001
299
The Tibeto-Burman substratum of Nepali (Indo-Aryan). Substrate Workshop, Leipzig Germany. 2003 Prosodic tautomorphemicity in Sino-Tibetan. In: David R. Bradley, Randy LaPolla, Boyd Michailovsky and Graham Thurgood (eds.), Language Variation: Papers on Variation and Change in the Sinosphere and in the Indosphere in Honour of James A. Matisoff, 8999. Canberra: Pacific Linguistics. Bradley, David R. 1997 Tibeto-Burman languages and classification. In: David Bradley (ed.), Tibeto-Burman Languages of the Himalayas (Papers in Southeast Asian Linguistics No. 14, Pacific Linguistics Series A86), 171. Canberra: Australian National University. DeLancey, Scott 1989 Verb agreement in Proto-Tibeto-Burman. Bulletin of the School of Oriental and African Studies 51 (2): 315333. Genetti, Carol 1999 Variation in agreement in the Nepali finite verb. In: Yogendra P. Yadava and Warren W. Glover (eds.), Topics in Nepalese Linguistics, 542– 555. Kalimadi, Kathmandu: Royal Nepal Academy. Genetti, Carol, and Kristine A. Hildebrandt 2004 The two adjective classes in Manange. In: R. M. W. Dixon and Alexandra Y. Aikhenvald (eds.), Adjective Classes, 7496. Oxford: Oxford University Press. Hildebrandt, Kristine A. 2003 Manange tone: Scenarios of retention and loss in two communities. Linguistics, University of California. 2004 A grammar and glossary of the Manange language. In: Carol Genetti (ed.) Tibeto-Burman Languages of Nepal: Manange and Sherpa, 241. Canberra: Pacific Linguistics. 2006 Maintenance and merger in Manange. Manuscript. 2007 Loanwords in Manange, a Tibeto-Burman Language of Nepal. Manuscript. Hoshi, Michiyo 1986 An outline of the Prakaa grammar: A dialect of the Manang language. 187317. Tokyo: Institute for the Study of Languages and Cultures of Asia and Africa. Kincaid, M. Dale 1991 The decline of native languages in Canada. In: R. H. Robins and E. M. Uhlenbeck (eds.), Endangered Languages, 157176. Oxford and New York: Berg.
300
Kristine A. Hildebrandt
Masica, Colin P. 2001 The definition and significance of linguistic areas: Methods, pitfalls and possibilities (with special reference to the validity of South Asia as a linguistic area). In: Rajendra Singh (ed.), Yearbook of South Asian languages and Linguistics, 205267. London/New Delhi: SAGE. Matras, Yaron, and Jeanette Sakel 2007 Investigating the mechanisms of pattern replication in language convergence. Studies in Language 31 (4): 829865. Mazaudon, Martine 2003 Les principes de construction du nombre dans les langues Tibétobirmanes. In: J. François (ed.), La pluralité, 91119. Paris: Société de Linguistique de Paris. Noonan, Michael 2003 Recent language contact in the Nepal Himalaya. In: David Bradley, Randy LaPolla, Boyd Michailovsky and Graham Thurgood (eds.), Language Variation: Papers on Variation and Change in the Sinosphere and in the Indosphere in Honour of James A. Matisoff, 6588. Canberra: Pacific Linguistics. Rogers, Clint 2004 Secrets of Manang. Kantipath, Kathmandu: Mandala Publications. Snellgrove, David L. 1961 Himalayan Pilgrimage: A Study of Tibetan Religion by a Traveller Through Western Nepal. Boston: Shambhala. van Driem, George 2001 Languages of the Himalayas: An Ethnolinguistic Handbook of the Himalayan Region, Containing An Introduction to the Symbiotic Theory of Language. Leiden: Brill.
Grammatical borrowing in Indonesian Uri Tadmor
1. General1 1.1. Classification Malay-Indonesian is a member of the Malayic subgroup of Western MalayoPolynesian, a branch of the Austronesian language family. Malayic languages are spoken throughout the Malay-Indonesian archipelago, and similar forms of standard Malay-Indonesian serve as the national languages of Indonesia, Malaysia, Brunei, and Singapore. The number of speakers of Malay-Indonesian is nearly 250 million, making it by far the most widely spoken language in Southeast Asia. The name “Indonesian” (or Bahasa Indonesia, ‘the Indonesian language’) refers to standard Malay as used in Indonesia, and also to regional varieties of the language which have been developing throughout the country in recent decades.2 It would not be possible to discuss all varieties of Indonesian here, not only because of space limitations, but also because many of them are poorly documented. Therefore the discussion will focus mostly on standard Indonesian, the most widely used variety of Malay-Indonesian. When relevant to the discussion, a few other varieties will be mentioned and specifically noted as such.3
1.2. Sociolinguistic position The vast majority of Indonesians know at least some Indonesian, and most use it on a regular basis. However, the standard language is not acquired as a first language. Where Indonesian is used as a home language, it is invariably in the form of a local colloquial variety. In this sense, all speakers of standard Indonesian are at least bilingual or bidialectal.4 Children acquire the standard language early on from its use on television and in school. In recent years the Jakarta dialect has been making inroads into areas that have previously been the sole domain of standard Indonesian. It is also used in youth magazines, on the Internet, in text messaging, and recently in advertisements as well. However,
302
Uri Tadmor
for any formal purposes standard Indonesian is still the overwhelming choice. Indonesian is the sole official language of Indonesia, and is used in all government communication, both oral and written. Practically all published work (books, newspapers, magazines) is in Indonesian, as are most product markings and instructions, public signs, and even personal letters. Spoken Indonesian is widely used as a lingua franca among people who belong to different ethnolinguistic groups.
1.3. Sources of foreign influence on Malay-Indonesian The earliest foreign language known to have been in significant contact with Malay-Indonesian was Sanskrit. The oldest Malay inscriptions (7th c. AD) are intertwined with Sanskrit texts, and even the Malay sections contain many Sanskrit loanwords. However, these inscriptions are too few and too fragmentary to reach any definite conclusions about grammatical borrowing. Sanskrit continued to be used in the Malay-speaking world for centuries as a literary and liturgical language, but gradually disappeared from use after the introduction of Islam. Newer Indo-Aryan languages such as Hindi–Urdu as well as Dravidian languages such as Tamil, brought to the archipelago by merchants, missionaries, and immigrants, were also in contact with MalayIndonesian, and have left traces in the form of numerous loanwords. Chinese pilgrims and traders have been visiting Indonesia for well over a thousand years, and Chinese communities have also existed throughout the archipelago for many centuries. Various southern Chinese languages are spoken in Indonesia, and have influenced colloquial varieties of Indonesian, although in the standard language their influence has been limited and purely lexical. Traders from the Near East first arrived in Indonesia during the second half of the first millennium ad. Eventually the Arabic and Persian languages were to have a strong impact on Malay-Indonesian. However, this did not take place until several centuries later, when local inhabitants began converting to Islam. The influence of Arabic has been especially strong, in the form of a large number of loanwords. Most did not enter the language from spoken Arabic, but rather from Arabic literature (as well as from Persian and Indian literature, where Arabic loanwords abound). Because many religious and other texts were translated into Malay-Indonesian from Arabic, sometimes word for word, Arabic has also had some grammatical influence on MalayIndonesian.
Indonesian
303
The earliest Europeans with a substantial presence in Indonesia were the Portuguese, who first arrived in the first half of the sixteenth century. There are numerous Portuguese loanwords in Indonesian, most originating from creolized varieties rather than from metropolitan Portuguese. Some colloquial varieties of Malay-Indonesian underwent structural interference from these creoles, but the structure of the standard language was not affected. The next Europeans to appear on the scene were the Dutch, who sent their first expedition in 1598. They eventually came to control all of present-day Indonesia, and remained until the mid-twentieth century. The use of Dutch in Indonesia was rather limited and only a fraction of the population ever gained any fluency in the language. Nevertheless, since Dutch-speaking Indonesians formed to the influencial elite, Dutch had a strong impact on the Indonesian lexicon, and a lesser one on its grammar as well. Following independence, English quickly replaced Dutch as the most widely studied European language in Indonesia. Although English instruction in general has not been very successful, many members of the upper classes have a good command of the language, which they acquire in elite schools as well as abroad. Indeed, code-switching between English and Indonesian has become common among well educated Indonesians. English is also heard daily on television and in movie theaters, so most Indonesians have at least some exposure to it. In addition to coming in contact with foreign languages, Malay-Indonesian has been in contact with hundreds of local languages, principally via its role as a lingua franca throughout the archipelago. The most influential of these local languages overall was Javanese, which has existed in a state of quasi-symbiosis with Malay-Indonesian for over a millennium. Today, native speakers of Javanese form the largest group of speakers of Indonesian. Another language that has had some influence on Standard Indonesian is Minangkabau, a Malayic language of western Sumatra. Many Indonesian authors and educators, especially during the early formative years of modern standard Indonesian, were native speakers of Minangkabau. Sundanese, spoken in western Java, has had a strong impact on the nearby Jakarta dialect, and through it on the standard language as well. It is important to note that many borrowed features and words – especially the oldest and best-integrated ones – entered the language not via widespread bilingualism, but rather through written literary languages used by small minorities. Such changes affected the language of the elite first, before slowly spreading to the general community.
304
Uri Tadmor
2. Phonology5 The structural domain arguably most affected by contact has been the phonology. Contact-induced phonological changes were introduced principally via two means. Many words borrowed into Malay-Indonesian contained sounds and sound combinations previously unknown in the language. Initially, loanwords were assimilated to the existing phonological structure, but when borrowing from a particular language was extensive (as was the case with words from Sanskrit, Arabic, and Dutch), this eventually led to changes in phonotactics, and even to the introduction of new phonemes. In addition, the widespread use of Malay-Indonesian as a second language has played an important role in phonological interference, as speakers transferred features from the phonology of their native languages into Malay-Indonesian. This type of interference was much stronger in Indonesia than in Malaysia, since the great majority of Indonesians did not speak Malay historically. The most important language in this category is Sundanese.
2.1. Consonants The inherited phoneme inventory (that is, Indonesian phonemes which reflect Proto Malayic phonemes) includes 16 consonants: /b, d, Ô, g, p, t, c, k, m, n, ɲ, ŋ, l, r, s, h/. Indonesian orthography – which is used in this chapter for citing Indonesian words – represents these consonants with characters identical to their IPA equivalents, with three exceptions: the voiced palatal stop /Ô/ is spelled j, the palatal nasal /ɲ/ is spelled ny, and the velar nasal /ŋ/ is spelled ng. The glottal stop was not yet a phoneme in Proto Malayic, and even in modern Indonesian it is only marginally distinctive. A phonetic glottal stop may be inserted before word-initial vowels after pause. In addition, in medial position there are words in which it clearly phonemic. These are all loanwords, chiefly from Arabic. After a consonant it occurs, for example, in the word Jumat [jumʔat] ‘Friday’, in contrast with words such as rumah [rumah] ‘house’; after a vowel, in the word syair [çaʔir] ‘prayer’, in contrast with words such as kain [kain] ‘cloth’.6 In final position it occurs in a few kinship terms, e.g. (bapak [bapaʔ] ‘father’, kakak [kakaʔ] ‘elder sibling’) and exclamations (tidak [tidaʔ] ‘no!, not’,7 masa [masaʔ] ‘really?!’), which contrast with the large number of words which end in vowels. In such words the glottal stop originated in an exclamatory intonation that involved glottalization.
Indonesian
305
In the case of kinship terms, its origin would be vocative forms, which are exclamatory by definition.8 Other than in these words, final glottal stop hardly occurs in Indonesian.9 The phonemicization of the glottal stop is therefore the result of a combination of internal and external factors. Two other phones which probably phonemicized under the influence of borrowing were the glides w and y. In Proto Malayic, w and y never contrasted with u and i, respectively, and are thus best analyzed as underlying vowels which undergo gliding in certain environments. In early Sanskrit loanwords, Sanskrit v ([w])10 was represented by Malay b in syllable initial position, e.g. baca ‘read’ (< vaca ‘speaking’), and by u in other positions, e.g. suara ‘sound, voice’ (< svara ‘sound’). Sanskrit y was represented by j in initial position, e.g. jasa ‘meritorious service’ (< yaśas ‘honor’), and by i in other positions, e.g. setia ‘loyal’ (< satya ‘true, faithful’). In later Sanskrit loanwords, however, there is no longer an assimilation of semivowels, and they are left intact, even in syllable-initial position: wanita ‘woman’ (< Sanskrit vanita), bahaya ‘danger’ (< Sanskrit bhaya). This pattern is repeated in loanwords from other languages as well: the earlier the loan, the greater the chance that the semivowels would be represented by a stop. Initial glides are now found in a few native words, such as ya ‘yes’ (from ia ‘3sg’) and yang ‘relative pronoun’ (from ia '3sg' + ng 'ligature'). Malaysian also has yu ‘shark’ (from the earlier hiu, still used in Indonesian) and yuran ‘fee’ (cf. Indonesian iuran). Unassimilated semivowels also occur in a large number of later loanwords from languages other than Sanskrit, especially Arabic. In addition to phones which phonemicized under the influence of borrowing, modern Indonesian also has several consonant phonemes which were borrowed outright. All loan phonemes consist of fricatives, which is not surprising, considering that Proto Malayic was very poor in fricatives (the only fricative phonemes were *h and *s). The labiodental f is now used by most Indonesians, and certainly forms part of the phoneme inventory of standard Indonesian. It probably first entered the language via Dutch loanwords, such as famili ‘relatives’ (< familie ‘family’) and filem ‘film, movie’ (< film). This phoneme is also present in many Arabic loanwords, but initially Arabic f was represented by p as in pikir ‘think’ (< Arabic fikr ‘thinking, cognition’) and peduli ‘caring’ (< Arabic fuḍūlī ‘inquisitive, busybody’). In Dutch loanwords, too, f was initially represented by p; the words for ‘relatives’ and ‘film’ cited above were realized as [pamili] and [piləm], respectively, when first borrowed. The addition of /f/ to the phoneme inventory also added a new place of articulation. Before its incorporation, Indonesian had a series of bilabials, but no labiodentals.
306
Uri Tadmor
The loan phoneme /ç/, spelled sy, first emerged as the equivalent of Arabic ʃ in loanwords, for example in the words syair ‘poem’ (< Arabic ša‘ir) and syukur ‘thank (God)’ (< Arabic šukr). Before then, ʃ in Arabic loanwords was represented by s, as it still is in some well-established older loanwords like serikat ‘union’ (< Arabic širkat-). Indeed, the words for ‘poem’ and ‘thank (God)’ just cited are still realized sair and sukur by many speakers. Later, sy was also used to represent ʃ in English loanwords, e.g. syuting ‘shooting, filming’, syok ‘shock’. Another loan phoneme is /z/. Initially it was represented by j in loanwords from Arabic, for example jiarah ‘to visit a grave or a holy site’ (< Arabic ziyārah ‘a visit’), jaman ‘time, period’ (< Arabic zamān ‘time’). Such pronunciations persist, but many educated speakers now realize these words as ziarah and zaman, respectively. In European loanwords, z was initially represented by s as in bensin ‘gasoline’ (< Dutch benzine), even if the original orthography with z was maintained as in zébra [sebra] (< Dutch zebra). Increasingly, z is retained unassimilated in European loanwords too (now originating mostly from English). The loan phoneme /x/ has had a similar history. Initially, the voiceless velar fricative x in Arabic loanwords was represented by k, as in kabar ‘news’ (< Arabic xabar) and Kamis ‘Thursday’ (< Arabic [yaum al]-xamīs). Later, representing x with h became the norm, although the pronunciation k still persists as well. Some speakers realize Arabic x in loanwords with a spelling pronunciation – that is , as a sequence of k and h. Finally, people with an Islamic educational background often preserve the original sound [x]. Thus the word akhir ‘end’ (< Arabic āxir) is pronounced variously as [akir], [ahir], [akhir], and [axir]. Dutch also has the phoneme x, and some speakers with a Dutch educational background (or from groups strongly influenced by Dutch) retain x in Dutch loanwords in Indonesian, e.g. [spaxeti] ‘spaghetti’, [bioloxi] ‘biology’, [texəl] ‘tile’. Finally, /v/ has been claimed as a (loan) phoneme by some authorities, for example the official dictionary of Indonesian (KBBI 2002). However, orthographic v is invariably realized as f or p in Indonesian. (The phoneme /v/ does, however, occur in English loanwords in Malaysian.)
2.2. Vowels The Indonesian vowel system includes six vowels: /a, e, i, o, u, ə/. Indonesian orthography represents these vowels in a straightforward manner, with the
Indonesian
307
exception of /e/ and /ə/, which are both spelled /e/. In this chapter, the two are distinguished: /ə/ is spelled e, and /e/ is spelled é. The schwa – realized as a mid central vowel in modern Indonesian – is reconstructed in Proto Malayic only between consonants,11 in sharp contrast to all other vowels, which occur at word edges as well as before and after other vowels. This casts a strong doubt on its historic status as a phoneme. Moreover, even its original phonetic nature is not clear (in all probability it was realized as a short a). Therefore it is possible to analyze schwa in premodern Malay-Indonesian as phonetically inserted (epenthetic), later undegoing phonemicization under the influence of Dutch loanwords, in which a mid-central vowel can occur word finally, as in halte bis ([haltə bəs/bis] ‘bus stop’.12 This phonemicization has affected native vocabulary as well. Schwa now occurs at the end of the very common word ke ‘to’ (a clitic, from Proto Malayic *ka) and also before other vowels, but only at morpheme boundaries), as in keindahan ‘beauty’ (from indah ‘beautiful’ + the noun-forming circumfix ke-an). The process lying behind the emergence of the vowel phonemes é and o is not fully understood process, but it was probably influenced by language contact. These vowels did not occur (at least not as phonemes) in Proto Malayic, and almost all occurrences of é and o in inherited Indonesian morphemes reflect *i and *u, respectively. However, no satisfactory phonological explanation for this phoneme split has been put forward, and counterexamples can be found for any hypothetical conditioning environment. One possibility is that the phonemicization was caused by random over-distinction of phonetic lowering by second-language speakers of Malay, who were transferring into it the phonemic distinction of their native language.
2.3. Phonotactics The syllable structure of Indonesian has been profoundly affected by borrowing. In Proto Malayic, the syllable shape was (C)V(C). There were no true consonant clusters, i.e. sequences of consonants at syllable edges. The only permissible sequences of two consonants ran across syllable boundaries, and even then they were restricted to homorganic nasal clusters13 (for example in *bantu ‘help’, *tanda ‘sign’, *jumpa ‘meet’) or /r/ followed by another consonant (as in *terbang ‘fly’, *pergi ‘go’). Due to massive lexical borrowing, Indonesian now allows two or even three consonants in the onset and coda, as in the following examples: pré.tél ‘dismantle’ (< Javanese prétél), stro.bé.ri
308
Uri Tadmor
‘strawberry’ (< English strawberry), fals ‘(to sing) off key’ (< Dutch fals), korps ‘corps’ (< Dutch korps). Other phonotactic constraints of Indonesian have also undergone change due to contact. As just mentioned, sequences of nasal+oral consonants in Proto Malayic were homorganic. Loanwords were initially assimilated to this pattern, as in mungkin ‘maybe’ (< Arabic mumkin) and ingkar ‘repudiate’ (< Arabic inkār). However, due to borrowing, modern Indonesian allows heterorganic nasal-oral sequences in newer loanwords, as in angpao ‘gift envelope’ (< Hokkien âŋ-pao) and tanpa ‘without’ (< Javanese tanpa). Nonhomorganic nasal-oral sequences can also be found in newly coined words, as in the clipped forms amdal ‘environmental impact study’ (from analisa mengenai dampak lingkungan) and Ménhankam ‘Minister of Defense’ (from menteri pertahanan dan keamanan).
3. Typology Historically, Malay-Indonesian was an agglutinative language, a fact still reflected in the numerous affixes and clitics of the standard language. Colloquial varieties are on the whole much more isolating, and use far less productive affixation. Some do not even use the verbal active and passive prefixes meng- and di-, which are the most commonly used affixes in standard MalayIndonesian. With the decreased use of affixation and the increased use of roots as free morphemes, the distinction between different word classes has also been eroding. These processes, while most evident in colloquial varieties, have also been affecting standard varieties. Although internal factors may be partly responsible for these changes, external factors were also involved. A decrease in morphological complexity is an integral part of the process of pidginization, and indeed pidginized varieties of Malay-Indonesian exhibit a sharply reduced productive morphology. Such varieties have served as the basis for creolized varieties of Malay-Indonesian, which in turn have been influencing the standard language. The fact that Malay has served as a regional lingua franca for over a millennium is also part of the reason why its morphosyntactic structure has lost some of its complexity, since it involved widespread and often imperfect second-language acquisition. It should also be noted that the vast majority of current speakers of Malay-Indonesian have acquired it as a second language (even if eventually it may have become their dominant language). A process of language shift has been having similar affects, as increasing numbers of
Indonesian
309
speakers have been abandoning their ancestral home languages and switching to Malay-Indonesian as their home language instead. This long historical process received an impetus in the middle of the twentieth century, when Malay-Indonesian became the national language of newly independent Indonesia and Malaysia. Other changes observed in the history (and prehistory) of Malay-Indonesian, such as the increased frequency of SVO sentences, may also be related to this historical process. Contact-induced processes of simplification in various standard languages, including in standard MalayIndonesian, are discussed in detail in McWhorter 2007. In addition to the general simplifying and markedness-reducing affects associated with pidginization, use as lingua franca, widespread second language acquisition, and language shift, the typology of Indonesian has also undergone more specific changes due to contact. In an insightful yet largely overlooked paper,14 Becker and Umar (1980: 97) discussed syntactic change in Indonesian. They observed ‘a general systemic change in Indonesian which has been going on for a long time: the change from a focus system of topicalization to a subject system of topicalization.’ Several types of change are mentioned, some of which are discussed in this chapter. Becker and Umar conclude that ‘[t]he kinds of syntactic changes we are observing in Indonesian may be among the most important impositions of the colonial – and neocolonial – period [...]’ (1980: 100). This may be an overstatement, but there is little doubt that these changes they discussed have been taking place, and that language contact is at least partially responsible for them.
4. Nominal structure 4.1. Case marking and prepositions Indonesian has overt case marking only in the pronominal system, and on a very limited scale at that. Classical Malay had three enclitic pronouns which could roughly be described as having a genitive function: -ku (1sg), -mu (2), and -nya (3sg). In addition to expressing possession, they were also used with certain prepositions. These clitics were the counterparts of the free morphemes aku (1sg), engkau (2sg), kamu (2pl), and ia (3sg). The genitive use of these clitics persists in modern Indonesian: rumah-ku 'my house', untukku ‘for me’, rumah-mu 'your house', untuk-mu ‘for you’, rumah-nya 'his/ her/its house', untuk-nya ‘for him/her/it’. However, these enclitics have also developed an accusative function, which they did not have in early Classical
310
Uri Tadmor
Malay: me-lihat-ku 'to see me', me-lihat-mu 'to see you', me-lihat-nya 'to see him/her/it'. The probable source of this change is to be found in literally translated Arabic religious texts. Arabic has a complete set of pronominal suffixes, which have both accusative and genitive functions.15 These suffixes were translated into Malay using pronominal clitics, which had previously been used only with a genitive function. An interesting case of genitive marking that owes its development to language contact is the so-called possessive -nya construction. The enclitic -nya, as just seen above, is the oblique counterpart of the third-person pronoun ia/ dia. It also fills various other functions, one of which is marking the head in possessive construction. This is patterned after a similar construction in Sundanese and Javanese. For discussion and examples, see Section 4.4. There are no clear examples of prepositions that were borrowed into Malay-Indonesian as such. However, several words which were borrowed into Indonesian as content words were later grammaticalized as prepositions. The colloquial preposition sama ‘with’ is derived from Sanskrit sama ‘equal’. In early Classical Malay, sama only meant ‘same (as)’. However, some varieties of Bazaar Malay, while maintaining its original sense of ‘same’, also developed into a preposition with a basic comitative function (‘with’), replacing the Malay comitative preposition dengan, but also filling various other functions.16 The comitative function of sama was then transferred into the standard language in the form bersama (with the literary prefix ber-), meaning ‘(together) with’. The function of some inherited prepositions has been extended under the influence of similar prepositions in other languages. The benefactive preposition untuk ‘for’ is now also used in certain temporal expressions, such as untuk selama-lamanya ‘forever and ever’, untuk pertama kalinya ‘for the first time’. This is probably under the influence of the Dutch benefactive preposition voor ‘for’, which can also mark temporal clauses, such as voor eeuwig en altijd ‘for ever and ever’, voor het eerst/voor de eerste keer ‘for the first time’. Another preposition whose function has been influenced by Dutch is dari ‘from’. In modern Indonesian the original ablative function of this preposition has been extended to include a genitive function, as in rumah dari Presiden ‘the President’s house’ and daftar dari kata-kata. This extension of meaning, considered ungrammatical by purists but nevertheless widely used, is patterned after the Dutch ablative preposition van, which can also express possession, as in het huis van de President ‘the President’s house’ and een lijst van woorden ‘a list of words’.
Indonesian
311
The marking of spatial relations is remarkably free of borrowing. A notable exception is antara ‘between’, from Sanskrit antara ‘intermediate space or time’. It has been claimed that pada, a preposition with a variety of functions, is derived from Sanskrit pada ‘foot’ or the related word pāda ‘site, position’, but the evidence is not altogether convincing (Gonda 1952: 396). Finally, the Arabic vocative particle yā has been borrowed into Indonesian as ya. It was first borrowed as part of fixed expressions such as Ya Allah! ‘My God!’ (< Arabic yā aḷḷāh) and Ya Robbi! ‘Oh my Lord!’ (< Arabic yā rabbī). Later, ya came to be used with native words, as in Ya Tuhan! ‘Oh my God!’ and Ya Ampun! ‘Oh my goodness!’
4.2. Sex and gender There is no gender marking per se in Malay-Indonesian, and even natural sex distinction was not lexicalized in early Malay, except for a few pairs of basic kinship terms such as father/mother and husband/wife. However, due to borrowing, there are now quite a few words which have distinct male and female forms. Some Sanskrit loanwords in Indonesian come in pairs, with male forms ending in -a and female forms ending in -i (female): putra ‘son’ : putri ‘daughter’, siswa ‘male high school student’ : siswi ‘female high school student’, saudara ‘male sibling or relative’ : saudari ‘female sibling or relative’; dewa ‘god’ : dewi ‘goddess’. This pattern has been extended to the native word pemuda, which originally meant ‘young person’, but now means ‘young man’, because the final -a was reinterpreted as a male ending. A new word pemudi ‘young woman’ was created as its female counterpart by false analogy. Another pair of Sanskrit-derived suffixes with distinct male and female forms is -wan/-man and -wati, which respectively denote male and female habitual agents, e.g. karyawan ‘(male) worker’ : karyawati ‘(female) worker’, wartawan ‘(male) journalist : wartawati ‘(female) journalist’. The bases of these words themselves are also derived from Sanskrit, but these suffixes can be used with non-Sanskrit bases, as in ilmuwan ‘scientist’ (ilmu ‘knowledge, science’, < Classical Arabic [al]-‘ilmu ‘[the] knowledge’). Some loanwords from Arabic also distinguish between male and female forms. These consist mostly of terms related to Islam such as soleh ‘pious (Muslim man)’ : solehah ‘pious (Muslim woman)’, haji ‘a man who has made the pilgrimage to Mecca’ : hajjah ‘a woman pilgrim’. Some more examples are provided in Section 4.3.
312
Uri Tadmor
In most loanwords with distinct male and female forms, the female form is not in common use, and the male form in fact serves as the unmarked member (which can refer to females as well, especially colloquially). However, in some common kinship terms borrowed from Dutch, the male–female distinction is strictly maintained in Indonesian. Such terms include mama ‘mother’ (< Dutch mamma) and papa ‘father’ (< Dutch pappa); tante ‘aunt’ (< Dutch tante) and om/oom ‘uncle’ (< Dutch oom); oma ‘grandmother’ (< Dutch oma) and opa ‘grandfather’ (< Dutch opa).
4.3. Number and numerals In Proto Malayic, the only set of words with formal number distinction were pronouns: *aku ‘1sg’, *kami ‘1pl’, *kau ‘2sg’, *kamu ‘2pl’, *ia ‘3sg’, *sida(?) ‘3pl’.17 Other than that, Malay-Indonesian does not distinguish between singular and plural forms. Collective nouns can be formed by reduplication: anak ‘child/children’ : anak-anak ‘group of children’, rumah ‘house/ houses’ : rumah-rumah ‘group of houses’. There is a trend in academic and journalistic writing to use reduplicated forms as the equivalents of English plurals, especially in Malaysia. This reinterpretation of collectives as plurals is due to the influence of Dutch and English, both of which have morphological plural forms. Two particles used to form collectives were borrowed from Javanese. In formal Indonesian para (< Javanese para) is used to form collective nouns, as in penonton ‘viewer’ : para penonton ‘the viewers, the audience’, penumpang ‘passenger’ : para penumpang ‘the passengers (of a particular vehicle or transport service)’. In colloquial Jakarta Indonesian (and increasingly in literary Indonesian as well), another particle, pada, is used to form collective-subject verbs, as in meréka pergi ‘they went’, meréka pada pergi ‘they went (referring to a group)’. The word pada (in this sense) was borrowed from Javanese padha ‘same, equal’, from which the collective-forming function developed in Javanese before being borrowed into Indonesian. In some cases plural forms of words were borrowed but used without number distinction (that is, even when the referent is singular). Examples of such Arabic loanwords are (satu) huruf 18 ‘letter [of the alphabet]’ (<ḥurūf ‘letters’, the singular form being ḥarf), and (seorang) ulama ‘Islamic scholar’ (< ‘ulamā’ ‘scholars’, the singular form being ‘alīm [which was also borrowed without number distinction as alim]). Dutch plurals borrowed without number distinction include (satu) karcis ‘ticket’ (< kaartjes ‘tickets’, the sin-
Indonesian
313
Table 1. Arabic loanwords distinguished for number and gender in Indonesian masc.sg ‘attendee’
hadir (
fem.sg
masc.pl
fem.pl
hadirah (
hadirin (
hadirat (
Note: In general Indonesian (i.e. not used in an Islamic context), hadir is an adjective meaning ‘present’, while hadirin means ‘audience’.
gular is kaartje), and (seorang) politisi ‘politician’ (< politici ‘politicians’, the singular is politicus [which was also borrowed without number distinction as politikus]). Similar examples of English loanwords include (satu) tips ‘tip’, (seorang) fans ([fens])‘fan’. The only cases of plural forms which are actually used as such (i.e. in opposition to singular forms) are a few Arabic terms of human reference used by speakers with a strong Islamic background. Examples are provided in Table 1. The numeral system has also been impacted by borrowing. Tiga ‘three’ ultimately derives from Indo-Aryan via a Dravidian language. The Old Malay word for ‘three’ was tlu, which goes back to Proto Austronesian. The Sanskrit loanword laksa ‘ten thousand’ (< lakṣa ‘a hundred thousand’) is rarely used in modern Malay-Indonesian, but juta, also borrowed from Sanskrit (< ayuta ‘ten thousand’), is the only word for ‘million’. The Dutch loanword milyun ‘million’ (< miljoen) is now obsolete, but milyar ‘billion’ (< miljard) and trilyun ‘trillion’ (< triljoen) are commonly used with reference to monetary amounts (the intrinsic value of the Indonesian currency is extremely low). Another numeral borrowed from Dutch – or possibly English – is lusin ‘dozen’, ultimately derived from Dutch dozijn or English dozen via Chinese Bazaar Malay (where the change d > l is common).19 In Old Malay, numerals between 10 and 20 were expressed by simple juxtaposition, e.g. sa-puluh dua (lit. ‘one-ten two’) ‘twelve’. In modern Malay-Indonesian, these numerals are formed with the special element belas ‘-teen’, e.g. dua belas (lit. ‘two-teen’) ‘twelve’. This pattern was borrowed from Old Javanese, along with the element belas itself (< Old Javanese welas). A similar pattern was also borrowed from Javanese for numerals between 21 and 29, with the element likur ‘score and...’: se-likur ‘twenty-one’, dua-likur
314
Uri Tadmor
‘twenty-two’, etc., but these forms are rarely used in modern Malay-Indonesian. Finally, some Sanskrit-derived bound numerals are used in compounds, such as éka- ‘one’ (< Sanskrit eka), dwi- ‘two’ (< Sanskrit dvi), tri- ‘three’ (< Sanskrit tri), catur- ‘four’ (< Sanskrit catur), and panca- ‘five’ (< Sanskrit paṅca). They occur not only with Sanskrit-derived bases, but also with bases from other sources, as in dwifungsi ‘dual function (of the military; fungsi is from Dutch) and caturwulan ‘trimester’ (lit. ‘four months’; wulan is from Javanese).
4.4. Possession Possession can be expressed in various ways in Indonesian. In standard Indonesian (as in earlier forms of Malay), possession is indicated by simple head-initial juxtaposition, as in rumah ibu ‘mother’s house’ (rumah ‘house’, ibu ‘mother’). In colloquial Jakarta Indonesian, the common way to indicate possession is by attaching a third-person pronominal clitic to the head. This pattern is modeled after similar constructions in Javanese and Sundanese: (1)
Indonesian: rumah-nya bapak Javanese: omah-é bapak Sundanese: imah-na bapa house-3 father ‘father’s house’
This pattern has spread from colloquial to standard Indonesian, but only when the head is a verb. In such cases, the addition of -nya nominalizes the verb. (2)
pecah-nya pesawat break-3 airplane ‘the breaking up of the plane’
In some varieties of Bazaar Malay (and in speech forms that developed from it or were influenced by it), a different possessive construction is used. The order of the head and possessor is reversed, and punya ‘to have’ is inserted between them: bapak punya rumah ‘father’s house’ (this can also mean ‘father has a house’, depending on the context). In Malacca (which was the focal point for the dissemination of Bazaar Malay throughout the archipelago) this construction may have been patterned after Hokkien, as can be seen
Indonesian
315
in the following example from Baba Malay (a creole based on Bazaar Malay), adapted by Baxter (1988: 92) from Lim (1981: 4552): (3)
Baba Malay: gua punya rumah 1s + punya + house Hokkien: guà é chhǔ 1s + é + house ‘my house’
Finally, as mentioned in Section 4.1, the ablative preposition dari can also be used in certain contexts to indicate possession, under the influence of Dutch van.
5. Verbal structure Indonesian has only two affixes that can be considered inflectional, but they occur with very high frequency. These are the active prefix meng- and the passive prefix di-. Derivational affixes are much more numerous. None of the basic affixes of standard Indonesian exhibit evidence for borrowing. Categories such as tense, mood, and aspect are not grammaticized in Indonesian. Tense is optionally expressed by particles such as akan (to indicate future actions) and telah (to indicate past actions). Such particles are used with increasing frequency, even in contexts where they would not be used in earlier forms of the language, and this may be due to the influence of Western languages in which tense marking is obligatory. This trend is especially strong in standard Malaysian, where the influence can be clearly traced to English. The very common perfective particle sudah developed from a verb meaning ‘to complete, to finish’, which may have been derived from Sanskrit śuddha ‘cleansed, cleared, acquitted’. Many other particles with modal or modal-like meanings are also derived from loanwords, e.g. mesti ‘have to, must’ (< Javanese mesthi ‘inevitable’), perlu ‘need, must’ (< Classical Arabic (al-)farḍu ‘(the) duty’), bisa ‘can’ (< (dialectal) Javanese bisa ‘can’, possibly of ultimate Sanskrit origin), réla ‘willing to’ (< Arabic riḍa’ ‘agreement, consent’), mungkin ‘maybe, possible, probably’ (< Arabic mumkin ‘possible’), pasti ‘certain(ly)’ (< Javanese pesthi ‘predestined fate’); niscaya ‘certainly’ (< Sanskrit niścaya ‘certainly’), suka ‘to like’ (< Sanskrit sukha ‘pleasure’).
316
Uri Tadmor
6. Other parts of speech 6.1. Pronouns Several Indonesian pronouns are derived from loanwords, although it is not clear whether the pronominal system as a whole has been affected. The deferential first-person pronoun saya is derived from Sanskrit sahāya ‘companion’. It is used side by side with the inherited pronoun aku, now relegated to informal, intimate, or literary/poetic use. The introduction of honorific distinctions into the pronominal system may have been due to contact, although it is also possible this was due to an internal process. The originally plural secondperson pronoun kamu developed a secondary meaning as honorific secondperson pronoun (singular or plural), and this process may have preceded the grammaticalization of saya. Also, the third-person singular honorific pronoun beliau does not appear to be a loanword. So it is probable that rather than introducing a new category, the loanword saya simply filled the empty firstperson slot in an already existing paradigm of honorific pronouns. The third-person plural pronoun meréka (in earlier Malay, marika) was borrowed from Old Javanese marika, a distal demonstrative that was also used as a third-person pronoun. A third-person plural pronoun cannot be reconstructed in Proto Malayic with certainty. Some Malayic isolects in Borneo have forms going back to *sida, but this etymon has cognates in some nonMalayic languages of Borneo, and may have been borrowed from them into Bornean Malayic. If the slot was empty in prehistoric Malay, it would help explain why a loanword was used to fill it. In colloquial varieties of Malay-Indonesian, there are several other borrowed pronouns. In the Jakarta dialect, the general 1sg pronoun is gué, from Hokkien goá; 2sg is lu, from Hokkien lù. These two pronouns are well known throughout Indonesia (and in Malaysia too, where they are sometimes used in Bazaar Malay used with or by ethnic Chinese).20 The English first-person pronoun I and especially the second-person pronoun you are sometimes used by Indonesians when addressing foreigners. They are also used among westernized Indonesians. In urban peninsular Malay, both I and you are in very common use. Pronouns of Dutch origin are also used by certain groups of Indonesian speakers. These are éke ‘1sg’ (< ikke ‘1sg [child language]’), yéy ‘2sg’ (
Indonesian
317
special jargon. From there these pronouns spread to the jargon of trendy young women, known as Bahasa Gaul (‘language of socializing’). Sometimes they are also used in writing, as in (4), taken from a column in a leading Indonesian daily.22 In this excerpt, in which the writer is debating with his friends whether his writing style is critical enough, second-person reference expressions have been highlighted: (4)
“Ohh . . . jij salah. Kita tuh enggak menganggap lo begitu, Mas. Kita kan tahu dari dulu Mas memang sukanya gigit-gigitan, kan?” komentar spontan teman saya. “Maksud sampean dari dulu saya anjing?” balas saya. ‘“Oh, you’re wrong. We don’t think you’re like that. We’ve known all along that you like to bite, right?”, commented my friends spontaneously.’ “You mean I’ve always been a dog?”, I replied.’
In this short excerpt, four different expressions are used for second-person reference, all loanwords: jij from Dutch, lo from Hokkien, Mas (lit. ‘elder brother’) and sampéan from Javanese. Traditionally, pronouns have been considered to be impervious (or at least highly resistant) to borrowing. As Thomason and Everett (2001: 301) explain, because pronouns form a closed set and form a tightly structured system, linguists assumed that borrowing into the set would disrupt the system. (There may have been another, more prosaic reason for this wrong assumption: pronouns are rarely borrowed in European languages.) However, Thomason and Everett go on show (citing copious examples) that “given appropriate social circumstances, pronouns and even whole pronominal paradigms are readily borrowed” (2001: 301). Two social factors may have contributed to widespread pronoun borrowing in Malay-Indonesian (and in other Southeast Asian languages). One is the tendency to adapt to the speech of one’s interlocutor by using structural features and lexical items perceived as belonging to the interlocutor’s language, including pronouns. This may have been part of the motivation behind the initial borrowing of gua and lu from Hokkien. Another factor is lexicalized politeness; some pronouns (such as saya from Sanskrit and sampéan from Javanese) were initially borrowed as the honorific counterparts of existing pronouns, while others (such as I and you from English) can be used when the speaker wishes to avoid having to make a choice between an honorific pronoun and a derogatory one.
318
Uri Tadmor
6.2. Interrogatives Some basic interrogatives of Indonesian can be reconstructed in Proto Malayic, such as apa ‘what?’, siapa ‘who?’, mana ‘which?, where?’. However, there are no reconstructed forms for ‘when?’ and ‘how?’. The most common Indonesian word for ‘when?’, kapan, was borrowed from Javanese. The much earlier loanword bila ‘when?’ (< Sanskrit velā ‘time’) is now used only in formal or poetic contexts (although it is still in general use in standard Malaysian and various Malay dialects). In the word bagaimana ‘how?’ (lit. ‘like which’), the element bagai is borrowed from Tamil vakai ‘kind, manner, method’ (the second element, mana ‘which?, where?’, is Malay).
6.3. Particles and their functions The borrowing of function words may have repercussions on the grammar, because they may introduce new categories and distinctions into the language. While many Indonesian function words are derived from loanwords, relatively few were borrowed as such. Most were borrowed as content words, and later grammaticalized within Malay-Indonesian. Examples of both types, as well as for changes in the meanings of function words under the influence of language contact, are discussed below.
6.3.1. Conjunctions It is not possible to reconstruct coordinating conjunctions such as ‘and’, ‘or’, and ‘but’ in Proto Malayic. The conjunction dan ‘and’ is probably a contracted form of dengan ‘with’;23 it is commonly used in Classical Malay and in standard Indonesian, but rarely in colloquial varieties. Another literary form with a similar meaning is serta ‘and, also, together with’, from Sanskrit sārtha ‘company’. The word atau ‘or’ (< Sanskrit athavā ‘or, rather’) is used in both standard and colloquial varieties of Malay-Indonesian. Similarly, tapi ‘but’ (literary tetapi), from Sanskrit tathāpi ‘nonetheless’, is commonly used in formal and informal varieties of Indonesian. The use of Sanskrit loanwords to fill such basic functions is indicative of the extent of the influence of Sanskrit on early Malay.
Indonesian
319
6.3.2. Complementizer and relativizer The commonly used complementizer bahwa is a loanword (< Sanskrit bhāva ‘being, state’). It is possible that Proto Malayic did not use a complementizer to introduce indirect speech or thought. However, bhāva is not used as a complementizer in Sanskrit, so the development of a complementizer in MalayIndonesian is only indirectly due to contact. The literary relativizer nan was also borrowed, from the closely related language Minangkabau. It occurs in modern standard Indonesian, but not in early Classical Malay.
6.3.3. Adverbial clause markers Most adverbial clause markers in Indonesian are borrowed or contain borrowed elements. These include sementara ‘while (simultaneously)’ (< Sanskrit sam-anantara ‘immediately contiguous to or following’), waktu ‘when (referring to the past)’ (< Classical Arabic [al-]waqtu ‘[the] time’), walau(pun) ‘although’ (< Arabic walau ‘and [even] if’), meski(pun) ‘although’ (< Creole Portuguese maski ‘although’, < Portuguese mas que), karena ‘because’ (< Sanskrit kāraṇa ‘cause’), sebab ‘because’ (< Arabic sabab ‘reason, cause’), seperti ‘like’ (< Malay se- + Sanskrit prati ‘nominal prefix expressing likeness’), guna ‘in order to’ (< Sanskrit guṇa ‘quality, use’), supaya ‘in order to’ (< Malay se- + Sanskrit upāya ‘means’), agar ‘in order to’ (< Persian agar ‘if’), pasal ‘regarding, concerting, about’ (< Arabic faṣl ‘part’), soal ‘regarding, concerning, about’ (< Arabic su’āl ‘question’). The fact that so many adverbial clause markers in Indonesian are borrowed or based on loanwords may be an indication that this category of function words did not exist in early Malay, and that its presence in modern Indonesian is due to borrowing.
6.3.4. Focus particles Borrowed focus particles include cuma ‘only’ (< Tamil cummā ‘vaguely, gratuitously, freely’), saja ‘only, just’ (< Sanskrit sahaja ‘natural’), sama ‘same (as)’ (< Sanskrit sama ‘equal’), persis ‘precisely’ (< Dutch precies ‘precise(ly)’), pas ‘exactly’ (< Dutch pas ‘just now’), saban ‘each, every’ (< Javanese saben ‘each, every’), and colloquially even [ifən] ‘even, even though’ (< English even).
320
Uri Tadmor
6.4. Situation-bound expressions Borrowed greetings include Halo! ‘Hello!’ (< Dutch Halo!), Selamat! (‘Congratulations!’, and also part of many every day greetings like Selamat pagi! ‘Good morning!’, < Arabic salāmat-), and Asalamualaikum! ‘Peace be upon you!’ (traditional Islamic greeting, < Classical Arabic as-salāmu ‘alaikum!). The fact that Indonesian has many borrowed greetings may be because in traditional Malay society (and in southeast Asia in general) there were few specific situational greetings. Upon meeting an acquaintance, one would say something like ‘Where are you off to?, ‘ or ‘Where are you coming from?’, or ‘Have you had lunch yet?’, as indeed many Southeast Asians still do informally. The concept of specific greeting expressions for different times of day and situations appears to have been borrowed. Other borrowed situation-bound expressions and interjections include Sori! ‘Sorry!’ (< English ‘Sorry!’), Maaf! ‘Excuse me, forgive me!’ (< Hindi–Urdu māf, ultimately from Arabic mu‘āf ), andWow! ‘exclamation expressing admiration’ (< English Wow!). An expression that stands out by not being borrowed is Terima kasih! ‘Thank you!’, but the common reply Sama-sama! ‘You’re welcome!’ is based on the loanword sama ‘equal’ (< Sanskrit sama).
7. Constituent order As mentioned in Section 3, SVO sentences have become increasingly frequent in Malay-Indonesian. Contact with languages where this is the prevalent word order (basically English and Dutch) may have been involved in this process, although the evidence is not conclusive. In the Bazaar Malay of Jakarta, and consequently in some varieties of Jakarta Indonesian, demonstratives are preposed rather than postposed (as they are in standard Malay-Indonesian and in traditional Malay dialects), e.g. rumah ‘house’, ini rumah ‘house this’ (rather than rumah ini). This may reflect a Creole Portuguese substrate (cf. kaju ‘house’, iste kaju ‘this house’), reinforced by Chinese word order (cf. Hokkien chhǔ ‘house’, cí lè chhǔ ‘this house’). This analysis is supported by the fact that preposed demonstratives are typical of ethnic Chinese speakers of Jakarta Indonesian. Some idiomatic expressions of time and frequency appear to reflect a Sanskrit word order. One such expression is pertama kali ‘(for the) first time’ (lit. ‘first time’, rather than the expected kali pertama [which also occurs but is less usual]). This is probably not a coincidence, since both constituents –
Indonesian
321
pertama ‘first’ and kali ‘time’ – are of Sanskrit origin. By analogy, in other expressions containing kali ‘time(s)’, the modifier also precedes the head: kedua kali ‘second time’, lain kali ‘another time’.
8. Syntax Most matters pertaining to syntactic borrowing have already been discussed in previous sections. This section discusses two contact-related phenomena which have not been treated yet: the emergence of copulas and locative relative clauses.
8.1. Copulas Early Malay did not have copulas. A nominal subject and a nominal predicate could be simply juxtaposed, although frequently the topic marker pun marked the subject. In recent centuries, several copula-like expressions have emerged, possibly under the influence of Western languages which require a copula with nominal predicates. The most common are adalah (from ada ‘exist’ + -lah ‘comment marker’) and merupakan (originally a verb meaning ‘to take the form of’). Ialah (from ia ‘3sg’ + lah ‘comment marker’) also occurs, but is more common in Malaysian and in earlier Indonesian literature. The choice of copula depends on idiomaticity, but in (5) all three are permissible: (5)
Kemiskinan adalah/merupakan/ialah tantangan bagi Indonesia. poverty copula challenge for Indonesia ‘Poverty is a challenge for Indonesia.’
Recently there has been a trend of using copulas with adjectival predicates as well, especially in Malaysia. This is probably due to the influence of English.
8.2. Locative relative clauses Locative relative clauses are a relatively recent development in Indonesian, and are due to the influence of Dutch. Instead of the relativizer yang, such
322
Uri Tadmor
clauses use a locative interrogative (usually di mana ‘where’, but ke mana ‘whither’ and dari mana ‘whence’ also occur): (6)
rumah di mana dia tinggal house at which 3sg live ‘the house where he lives’
Purists view such constructions as ungrammatical, and advocate using the word tempat ‘place’ instead of di mana ‘where’, as in (7). However, this construction is also without precedent in early Malay, and thus also owes its emergence to foreign influence, albeit indirect. (7)
rumah tempat dia tinggal house place 3sg live ‘the house where he lives’
9. Lexicon Lexical and semantic borrowing (including calquing) are often treated together in the literature, but in fact they constitute different phenomena. Lexical borrowing consists of adopting morphs from another language. This usually also invovles adopting meanings as well, but in principle a morph can be borrowed without a meaning that is different from the source word’s. Thus, for example, Maori Wīwī ‘French’ is from French ‘oui, oui!’ ‘yes, yes!’, but there is not indication that it ever meant ‘yes’ in Maori, or that it was used as a name in French. In other words, only the morph has been borrowed, without its meaning. Semantic borrowing, on the other hand, consists of a change of meaning in morphemes which already exist in the language, and does not involve a transfer of morphs. In fact, semantic borrowing is better viewed as a type of structural borrowing (since it affects the semantic structure).
9.1. Lexical borrowing A large number of loanwords can be found in all varieties of Malay-Indonesian and in all semantic fields. Most older loanwords from Sanskrit and Arabic were borrowed directly from written sources. Newer loanwords from European, local, and other languages were usually borrowed via direct
Indonesian
323
contact with their speakers. English constitutes an intermediate case: some words were borrowed from speech, while others were borrowed from written sources. Some examples of loanwords in Indonesian from various sources are listed below. Sanskrit: suami ‘husband’, istri ‘wife’, kepala ‘head’, muka ‘face’, kunci ‘key’, gula ‘sugar’, kerja ‘work’, cuci ‘wash’, pertama ‘first’, semua ‘all’. Arabic: badan ‘body’, dunia ‘world’, nafas ‘breathe’, lahir ‘be born’, kuat ‘strong’, séhat ‘healthy’, kursi ‘chair’, waktu ‘time’, pikir ‘think’, perlu ‘need’. Chinese (Hokkien): cat ‘paint’, toko ‘store’, hoki ‘lucky’, téko ‘teapot’, mi ‘noodles’, kécap ‘soy sauce’, giwang ‘earrings’. Persian: kawin ‘marry’, domba ‘sheep’, anggur ‘grapes, wine’, pinggan ‘dish’, gandum ‘wheat’, saudagar ‘merchant’. Portuguese and Portuguese Creole: garpu ‘fork’, kéju ‘cheese’, sepatu ‘shoes’, jendéla ‘window’, méja ‘table’, roda ‘wheel’, bola ‘ball’, minggu ‘week’, dansa ‘dance’, séka ‘wipe’. Dutch: open ‘oven’, sup ‘soup’, handuk ‘towel’, kamar ‘room’, mobil ‘car’, gelas ‘glass’, duit ‘money’, koran ‘newspaper’, nécis 'neat', bor 'to drill'. English: koin ‘coin’, bolpoin ‘pen’, strés ‘stressed out’, tivi ‘television’, tikét ‘ticket’, pink ‘pink’, gaun ‘formal dress’, komputer ‘computer’, notes ‘notepad’, flu ‘flu’, stop ‘to stop’, cas ‘to charge’. Old Javanese: bapak ‘father’, ibu ‘mother’, meréka ‘they’, daging ‘meat’, rusak ‘damaged’, masuk ‘enter’, murah ‘cheap’, antar ‘bring/take’, ratu ‘queen’, pasti ‘sure’.
9.2. Semantic borrowing Some Indonesian words of Malay (i.e., inherited) origin have changed their meaning based on the meanings of similar-sounding words in Javanese and Table 2. Replaced meanings in some Indonesian words Word
Meaning in Malaysian
Meaning in Indonesian
Meaning in Javanese and Sundanese
sulit butuh gampang
secluded, secret penis illegitimate child
difficult need easy
difficult need easy
324
Uri Tadmor
Table 3. Indonesian expressions calqued from Dutch Indonesian expression
Literal meaning
Metaphorical meaning
benang mérah sayap kiri sayap kanan isapan jempol
‘a common theme’ ‘(politically) radical’ ‘(politically) conservative’ ‘a made-up story’
luar biasa
‘red thread’ ‘left wing’ ‘right wing’ ‘something sucked out of the thumb’ ‘out of usual’
Dutch expression
Literal meaning
Metaphorical meaning
rode draad linker vleugel rechter vleugel iets uit de duim zuigen
‘red thread’ ‘left wing’ ‘right wing’ ‘to suck something out of the thumb’ ‘out of usual’
‘a common theme’ ‘politically radical’ ‘politically conservative’ ‘to make up a story’
buitengewoon
‘extraordinary’
‘extraordinary’
Sundanese. Often the semantic change is subtle and thus hard to detect, but sometimes the borrowed meaning can completely replace the original one, as in the examples in Table 2. People often borrow words to represent concepts which are not yet lexicalized in their language, but sometimes a word is borrowed even if the language already has a word for that concept. In such cases the original word may become obsolete, but it can also be retained side by side with the newer loanword. Since it is not economical for a language to have two words with the same meaning, one of the words may undergo a semantic change. For example, Indonesian borrowed the Javanese words sapi ‘cattle’ and tawòn ‘bee’, but has also retained the inherited Malay words lembu ‘cattle’ and lebah ‘bee’. The inherited words then underwent semantic narrowing; lembu is now used for particular kinds of cattle, while lebah is used for the honey bee. Some Indonesian metaphorical compounds have equivalents in many languages of Southeast Asia and beyond. In such cases it is difficult to trace their origin, and there is also the possibility that they arose independently, especially if their semantics is transparent. However, many compounds that have opaque compositional semantics can be traced to Dutch. A few of the many examples of such calqued expressions are provided in Table 3.
Indonesian
325
10. Conclusion Malay-Indonesian has undergone extensive interference from other languages. This was due to long term contact with foreigners, as well as to its use as a lingua franca throughout the Malay-Indonesian archipelago. The domain most affected by borrowing has been the lexicon, while structural borrowing has been moderate. The impact of borrowing has been particularly strong on the phonology, significantly expanding the inventory of phonemes as well as radically changing the syllable structure and phonotactics. There are also some morphosyntactic contact phenomena, but on a smaller scale. In particular, the use of enclitic pronouns as direct objects and the emergence of copulas and locative relative clauses appear to due to contact. The main sources of borrowing into Malay-Indonesian were Sanskrit and Arabic (often via a third language), and more recently some local languages, Dutch, and English. Borrowing from Sanskrit and Arabic was almost exclusively through writing, while borrowing from other languages was mostly through speech. These two kinds of contact resulted in different kinds of interference phenomena. Taken together, these contact-induced changes have resulted in a modern Indonesian that is markedly different from its Malay predecessors. Notes 1. The author has taught Indonesian at several institutions, and has been working as a linguist in Indonesia since 1999. 2. In this chapter ‘Indonesian’ is used as shorthand for ‘standard Indonesian’. 3. Other than standard Indonesian, varieties of Malay-Indonesian mentioned in this chapter include Old Malay (the language of the oldest Malay inscriptions, especially those of the seventh century); Classical Malay (the language of the written literature of the 17th–19th centuries, from which modern standard Indonesian developed); Malay dialects (regional varieties spoken by ethnic Malays); Bazaar Malay (pidginized forms of Malay used for inter-ethnic communication); Urban Peninsular Malay (used among ethnic Malays in Kuala Lumpur and some other cities of West Malaysia, influenced by Bazaar Malay); and Malaysian (used here to refer to standard Malay as spoken in Malaysia). 4. In a diglossic situation where speakers use standard Indonesian in more formal situations and a colloquial variety of Indonesian as a home language, the two can be said to form the two ends of a continuum. 5. It would be unwieldy to furnish references for each of the many loanwords cited here. Suffice it to note that most of the Arabic loanwords mentioned here appear
326
6.
7. 8. 9.
10.
11.
12. 13.
14.
15. 16.
17. 18.
19. 20.
21.
Uri Tadmor in Jones 1978; most European loanwords appear in Grijns et al. 1983; and most Sanskrit loanwords appear in de Casparis 1997. The glottal stop in [jumʔat] ‘Friday’ and [çaʔir] ‘prayer’ may be deleted in rapid speech, but crucially a glottal stop can never be inserted in words such as rumah ‘house’ and kain ‘cloth’. The sense of ‘not’ developed from the original sense ‘no!’. Similar phenomena can be found in other Austronesian languages; see Blust 1979. In many Malay dialects as well as in standard Malaysian a final k is realized as a glottal stop in inherited vocabulary. However, this is not the case with standard Indonesian, nor with the most important colloquial variety, Jakarta Indonesian. In both, final /k/ is realized [kÝ]. The convention of using v in the transliteration, which I follow here, is based on the current pronunciation of Sanskrit. Historically, this was a bilabial semivowel much like English w, and this was probably the sound that was borrowed into Malay. Otherwise, it would be difficult to explain why v should be represented by u in non-initial position (see below). The initial schwa reconstructed in words such as *əmpat ‘four’ and *ənam ‘six’ only occurs before sonorants, and is therefore better analyzed as representing syllabicity: mÞpat, nÞam. Indeed, this is how these words are still realized in most Malay-Indonesian variants, including the standard ones. Similarly, schwa phonemicized in Malaysian under the influence of loanwords, but the process there was more complex. The only apparent exception is -ngs-. This can be explained by the phonetic realization of s in Malay and by Malay phonotactics, but a detailed explanation would be beyond the scope of this chapter. Embarrassingly the present author also overlooked this paper, only becoming aware of it just before submitting the final version of this chapter for publication. The sole exception is 1sg, which has different accusative and genitive forms. The semantic expansion from ‘same’ to include ‘with’ is fairly transparent: two persons doing the same action together can be viewed as doing it with each other. It is not clear whether Proto Malayic actually had a 3pl form (see Section 6.1). In this and following examples, satu ‘one’ or seorang ‘one+numeral classifier for humans’ are added, to demonstrate the fact that they can be used with a singular meaning. According to Scott Paauw (p.c.), a similar form occurs in Chinese Pidgin English. An example for the popularity of these pronouns in Indonesian comes from the title of the television series Gue sihir lu! (‘I put a spell on you!’), shown on the SCTV channel at the time of writing. Sometimes the original Dutch orthography is retained, as in (4).
Indonesian
327
22. From the colunm Kurang Tajam (‘not sharp enough’) by Samuel Mulia, Kompas online edition, 10 December 2006, accessed at http://www.kompas.co.id/ver1/ Kesehatan/0612/10/120132.htm. The original spelling has been retained. 23. This etymology was suggested to me by David Gil.
References Baxter, Alan N. 1988 A Grammar of Kristang (Malacca Creole Portuguese). (Pacific Linguistics B95.) Canberra: The Australian National University. Becker, Alton L., and Umar Wirasno 1980 On the nature of syntactic change in Bahasa Indonesia. In: Paz Buenaventura Naylor (ed.), Austronesian Studies: Papers from the Second Eastern Conference on Austronesian Linguistics. (Michigan Papers on South and Southeast Asia Number 15). Ann Arbor: The University of Michigan. Blust, Robert 1979 Proto-Western Malayo-Polynesian Vocatives. Bijdragen tot de taal-, land- en volkenkunde 135: 205251. de Casparis, J. G. 1997 Sanskrit Loan-Words in Indonesian. Published for the Indonesian Etymological Project as NUSA 41. Jakarta: Universitas Katolik Indonesia Atma Jaya. Gonda, J., 1952 Sanskrit in Indonesia. The Hague: Oriental Bookshop. Grijns, C. D., J. W. de Vries, and L. Santa Maria 1983 European Loan-Words in Indonesian. Published for the Indonesian Etymological Project by the Koninklijk Instituut voor Taal-, Land- en Volkenkunde, Leiden. Jones, Russell 1978 Arabic Loan-Words in Indonesian. Published simultaneously by the Indonesian Etymological Project as Cahier d’Archipel 2, SECMI, Paris. Produced at the School of Oriental and African Studies, University of London. KBBI 2002 Kamus Besar Bahasa Indonesia [The Great Dictionary of the Indonesian Language]. Jakarta: Balai Pustaka. Lim, Sonny 1981 Baba Malay: The language of the ‘Straits-Born’ Chinese. MA thesis, Monash University, Australia.
328
Uri Tadmor
McWhorter, John H. 2007 Language Interrupted: Signs of Non-Native Acquisition in Standard Language Grammars. Oxford: Oxford University Press. Thomason, Sarah G., and Daniel L. Everett 2001 Pronoun borrowing. Berkeley Linguistics Society: Proceedings of the Annual Meeting 27: 301315.
Grammatical borrowing in Biak Wilco van den Heuvel
1. Background Biak (in older sources also Numfoors(ch), Mafoors(ch) or Myfoors(ch)) is an Austronesian language of Papua, the easternmost province of Indonesia. The language has around 70,000 speakers, most of whom live on the islands Biak, Supiori, Numfor and the smaller islands around. In addition, the language is spoken in a number of settlements and on several islands along the North coast of the Bird’s Head peninsula, reflecting a long-lasting prominent position of Biak people in trade throughout the area. The data presented in this chapter are based on different periods of fieldwork in the village Wardo, West-Biak. The long history of contact with both Austronesian and non-Austronesian languages has led to a number of similarities between Biak and other languages in the same linguistic area. A brief discussion of Biak as part of the linguistic area of Eastern Indonesia can be found in Van den Heuvel (2006: 1112), who also discusses the relation between Biak and the languages of the Bird’s Head. The present chapter, however, focuses specifically on the unidirectional influence of local Malay/Indonesian on Biak. Although Malay has served as a sort of ‘supra-regional’ language throughout Indonesia from before the arrival of the Portuguese (Adelaar and Prentice [1996]), the influence of Malay on Biak used to be small. From at least the beginning of the eighteenth century, Biak people played an important role in trade, functioning as intermediaries between the West (especially Tidore) and the people from the Bird’s Head Peninsula (Van den Heuvel 2006: 23, and the references cited there). Although Malay was used in contacts with Tidore, Biak people appear to have used their own language in contacts with the local population, which can be deduced from the relatively high number of borrowed words in the local languages. After the end of the nineteenth century, however, with the intensification of Dutch colonial rule over the region and the conversion of the Biak people to Christianity, the use of Malay gained ground. Malay was used as the administrative language during Dutch colonial rule, in education, and by preachers from outside Biak. Since the 1960s, Indonesian, which is a standardized form of Malay, has been the of-
330
Wilco van den Heuvel
ficial language. This has lead to a gradually increasing use of Indonesian (and local Malay) to the expense of Biak. Nowadays, in the main town of the island (Kota Biak), Indonesian and local Malay (hereafter LM) are the dominant language both inside and outside the family. In Kota Biak, the use of Biak is restricted to communication between older Biak speakers (older than 55 years of age). In the villages that are more remote from town, Biak is used by younger generations also. In the village Wardo, where most fieldwork was conducted, the Biak language is used actively by people of approximately 30 years of age and older. For all generations in the village, code-switching between Biak and Indonesian/LM is common. The corpus is restricted to speakers that are still (relatively) fluent in the language. Although code-switching between Biak and Indonesian/LM is very common throughout the island, the present chapter focuses on one-to-one borrowing from Indonesian/LM into Biak, specifically on those cases that can be qualified as grammatical borrowing. For each of the presented phenomena, however, it will be argued briefly whether the phenomenon in question is more like borrowing or more like code-switching, based on the assumption that the demarcation between code-switching and borrowing is gradual rather than absolute (cf. Salmons 1990: 466, Gardner-Chloros 1995). Both for lexical borrowing and for grammatical borrowing, it is often not possible to distinguish between borrowing from Indonesian and borrowing from LM, as the two languages are similar in many respects. In this chapter, therefore, I will not attempt to distinguish between the two types of borrowing. As I am not aware of any influence of Indonesian/LM on Biak typology or Biak nominal structures (except from lexical borrowing), the discussion below is restricted to its influence in the realm of phonology, verbal structures and other parts of speech. The chapter focuses on the borrowing of modal auxiliaries, of the conjunction kalau ‘if’, and on the borrowing of the negative adverb bukan ‘not’.
2. Phonology Compared to Indonesian/LM, Biak lacks /l/, /h/, /t/ and /ŋ/, but has an additional voiced labial fricative /β/. In addition, unlike Indonesian/LM, Biak has a distinction between short and long vowels (the latter indicated by a diacritic sign on top of the vowel). In spite of these differences, adaptation of Indonesian/LM loans to Biak phonology is very rare. During my fieldwork, the only examples of adaptation were found with some very old people, like kapal
Biak
331
‘ship’ being realized as [kapar] and tahun ‘year’ being realized as [saun]. The non-adaptation of words from Indonesian/LM can be accounted for in several ways. On the one hand, it can be seen as an argument for their status as (insertional) code switches rather than loans. It is also possible, however, to view this phenomenon as proof of a certain convergence of Biak phonology and Indonesian/LM phonology. In that view, speakers of Biak have one shared phoneme inventory, integrating the Indonesian/LM and the Biak set of phonemes.
3. Verbal structures Whereas both Biak and Indonesian/LM have SVO order in the verbal clause, they differ in the expression of the subject. While in Indonesian the subject is indicated by either a free pronoun or a noun phrase, Biak expresses the subject by a prefix or infix on the verb, which is optionally preceded by a coreferential appositional noun phrase. This is illustrated in (1), where the noun phrase rusa nanine ‘this deer’ is optional, and in an appositional relation to the subject prefix d-. (1)
(Rusai nan-i-ne) di-ores. (deer giv-3sg.spc-this 3sg-stand ‘(This deer) stood still.’
In fact, the position preceding the verb is a sentential topic position. Whereas this position is often taken by noun phrases that are coreferent with the subject, it can also be occupied by other nominal arguments, as is the case in (2) below. Here the topic position is occupied by a preposed object, while the canonical object position is taken by a resumptive pronoun. (2)
Insape, ai-knam an-i-ne nko-kar i. then wood-tree giv-3sg.spc-this 1pl.ex-fell 3sg ‘Then, this tree we cut it.’
Considering contact phenomena, there is a clear difference between the borrowing of main verbs on the one hand, and the borrowing of auxiliary verbs on the other hand. Main verbs are integrated into Biak by prefixation with a verbalizing prefix ve-. The combination of prefix and verb, then, behaves as a normal verb, in the sense that the newly formed verb has to be combined
332
Wilco van den Heuvel
with a subject prefix. An example of the use of the verbalizer is given with ko-ve-putar ‘we turn’ in (3), where the newly formed verb ve-putar ‘vblzturn’ combines with the subject prefix ko ‘1pl.inc’. (3)
Ko-ro Yafdas ma, ko-ve-putar ve Ridge. 1pl.inc-loc Yafdas to.here 1pl.inc-vblz-turn to Ridge ‘We come from Yafdas to here, (and) we turn to Ridge.’
The prefix ve- ‘vblz-’ is used not only for the integration of loan-verbs, but also for the verbalization of both indigenous and exogenous stems belonging to other parts of speech, as illustrated in the following examples. In (4), the verbalizer is used for the verbalization of an indigenous numeral, while it is used in (5) for the verbalization of a borrowed Indonesian noun: (4)
Sko-ve-kyor. 3PC-vblz-three ‘They are three (persons).’
(5)
Ve-guru. <3sg>vblz-teacher ‘He is a teacher.’
Whereas borrowed main verbs clearly function as verbs in the recipient language, this is much less the case for the borrowed auxiliaries bisa ‘can’ and harus ‘must’. It should be noted that Biak has only one native auxiliary verb, the verb ve ‘want’, while it lacks native equivalents for the borrowed verbs bisa ‘can’ and harus ‘must’.1 The auxiliary verb behaves like a normal verb in that it combines with a prefix expressing person, number and gender of the subject. This is illustrated in the following example, where the subject-prefix ya- ‘1sg’ on the auxiliary verb is coreferential with the subject-prefix ya-‘1sg’ on the main verb: (6)
Ya-ve ya-ra ve pasar. 1sg-want 1sg-go to market ‘I am about to/want to go to the market.’
More commonly, the auxiliary combines with an “impersonal” 3sg subject marker, as in the following example, where the main verb ya-ra ‘1sg-go’ is preceded by i-ve ‘3sg-want’ rather than ya-ve ‘1sg-want’.
Biak
333
(7) Rov i-ne i-ve ya-ra ve amber night 3sg.spc-this 3sg-want 1sg-go to foreigner an-i-ra-wa. giv-3sg.spc-sea-over.there ‘This night I want to go to the foreigner seawards over there.’ [MIax] When the verb combines with a preverbal topic-NP, the auxiliary can either precede this NP or intervene between the NP and the verb.2 The auxiliary shares these structural properties with the historically related modal–aspectual adverb imbe ‘want’, whose function cannot be distinguished from that of the auxiliary. Consider (8) and (9) below, where the inflected auxiliary i-ve can be replaced by the modal adverb imbe ‘want’ without any observable change in meaning. While the auxiliary/adverb precedes the preverbal NP in (8), it follows in (9) (8) Imbe/I-ve [snon-nánki an-i-ne]NP i-bur. want/3sg-want [male-sky giv-3sg.spc-this 3sg-leave ‘This man from heaven wanted to leave.’ (9) [Snon-nánki an-i-ne]NP imbe/i-ve i-bur. [male-sky giv-3sg.spc-this want/3sg-want 3sg-leave ‘This man from heaven wanted to leave.’ We now return to the borrowings bisa ‘can’ and harus ‘must’, which are auxiliary verbs in the donor language. With respect to their position in the clause, these formatives are similar both to the auxiliary verb and to the modal adverbs. Given, however, that the borrowed formatives cannot be inflected for number, person and gender of the subject, they should be analyzed as modal adverbs rather than as auxiliary verbs. Consider (10) and (11) below. The position of bisa ‘can’ in (10) is parallel to that of imbe/ive ‘want’ in (8), while the position of bisa ‘can’ in (11) is parallel to imbe/ive ‘want’ in (9). Both in (10) and (11), the formative bisa ‘can’ is used in its bare form, not preceded by a verbalizer or a subject prefix, which makes it similar to the use of the modal adverb imbe ‘want’ in (8) and (9). (10) Vape bisa [romá-mkun an-i-ne]NP mám bavír i. but can [child-little giv-3sg.spc-this <3sg>see know 3sg ‘(All frogs looked alike,) but the child could recognize it (his own frog).’
334
Wilco van den Heuvel
(11) Inkukro sinan mko-i-ne mko-fár fa because parent 1pl-spc-this 1pl-tell cons [roma-babo ko-i-ne]NP bisa k-ák-swar epéne. [child-young 1pl.inc-spc-this can 1pl.inc-also-remember push.tight ‘Because you parents you tell so that we younger children can also remember (the stories).’ I analyze the phenomenon described here as closer to borrowing than to codeswitching. Although people are aware of the Indonesian/LM origin of these modals, they are the only available options for the concepts expressed by them, and are used rather regularly, without observable speaker variation. There are no indications that the use of modals is triggered by neighboring elements from L2. Finally, harus ‘must’ is not adapted to Biak phonology, as it is realized as [harus] despite of the absence of the phoneme /h/ in the indigenous Biak phoneme inventory. This, however, need not be seen as an argument in favor of code-switching, as non-adaptation is the rule rather than exception. As suggested above, Biak speakers employ an integrated Indonesian/LM phoneme system, so that there is no need to adapt loan phonemes to the indigenous Biak phoneme-inventory.
4. Other parts of speech Within the domain of “other parts of speech”, the discussion focuses on the use of a borrowed conjunction kalau ‘if’ and the use of a borrowed negative adverb bukan ‘not’. While the discussion on bukan ‘not’ will be postponed until Section 5, this section discusses the use of kalau ‘if’, along with a number of other, minor, contact phenomena: complementizers, numerals and reference to concepts like days of the week. The section opens, however, with a brief discussion on the conjunctions atau ‘or’ and dan ‘and’. Both the conjunction atau ‘or’ and the conjunction dan ‘and’ take the same structural position in the clause as their Biak counterparts (orovaido ‘or’ and ma ‘and’ respectively) and cannot be combined with them. While the use of atau is relatively frequent (46 instances of atau ‘or’ vs. 107 instances of orovaido ‘or’), dan ‘and’ is used only occasionally (10 instances of dan ‘and’ vs. hundreds of examples of ma ‘and’), in contexts containing a lot of code switches. This shows that atau ‘or’ is on its way to becoming an established loan, while dan ‘and’ should rather be considered a code-switch. For both conjunctions, there is no observable functional difference between the
Biak
335
Indonesian/LM word and their Biak counterpart. Both the Indonesian/LM and the Biak conjunctions are used for the connection of two phrases or two clauses. Whereas the function of dan ‘and’ is quite straightforward, it should be noted that atau ‘or’ (as well as its Biak counterpart orovaido) is typically used to introduce an alternative term for the (referent of) the constituent directly preceding the conjunction. Examples of this use are given in (12) and (13). In (12) we find the Biak word ruk ‘monkey’ given as an alternative for the Indonesian kera ‘monkey’. (12) Ras oser kera atau ruk i-warpu wáwe su-yaw day one monkey or monkey 3sg-with turtle 3du-pursue fa su-ker imbyef. purp 3du-plant banana ‘One day, a monkey or ruk with a turtle ran after each other to (go and) plant a banana tree.’ In (13) the speaker describes a scene that he is watching, and gives an alternative term for the referent of ai-mun ‘wood-piece’. (13) D-ors pan-kar ai-mun=i, 3sg-stand <3sg>touch-break wood-piece=3sg.spc atau ai-snáw=i. or wood-branch=3sg.spc ‘He is standing and breaking a piece of wood, or a branch.’ Turning to the borrowed conjunction kalau ‘if’, it should be noted that this conjunction differs from the other two conjunctions both in its functional properties and in the fact that its structural properties differ from those of its Biak counterpart. Structurally, the borrowed conjunction kalau ‘if’ differs from that of indigenous ido ‘theme’ in that kalau precedes the constituent that it has scope over whereas ido follows it. The two conjunctions can be used either on their own, or in combination.3 Functionally, the two conjunctions kalau ‘if’ and ido ‘theme’ have a similar pragmatic function, but slightly different semantics. As for their pragmatic function, both kalau and ido can be characterized as conjunctions that mark the constituent that they have scope over as “setting the scene” for the clause to come. This function is illustrated in (14), where the two conjunctions are used in combination. Note that the borrowed conjunction kalau precedes the phrase ránsyo ‘sweet potato’, which then is followed by the indigenous conjunction ido:
336
Wilco van den Heuvel
(14) Kalau ránsyo ido, bisa k-án vepék i. if sweet.potato theme can 1pl.inc-eat raw 3sg ‘As for sweet potatoes, we can eat them raw.’ The conjunctions kalau and ido may have scope either over a (series of) phrase(s), as in (14) above, or over a (series of) clause(s). In the latter case, the indigenous conjunction ido allows for both a temporal and a conditional interpretation, while the use of kalau is restricted to conditional contexts. Two examples of the use of kalau introducing clauses are given in (15) and (16). Sentence (15) is the introduction of a narrative sketching the hypothetical situation of the addressee accompanying the speaker at a journey across the island Biak. (15) Kalau Wilco wa-so vo kuy-ék ro bis, if Wilco 2sg-accompany sim 2du-go.up loc bus ku-sasyar ro terminal kota (...) 1du.inc-go.out loc terminal town ‘If Wilco you followed and the two of us got on the bus, the two of us would go out of the bus terminal (…)’ While (15) above is an example of the use of kalau on its own, the following sentence is another example of the use of kalau and ido in combination. (16) Kalau nko-na rovean=no va ido, if 1pl.ex-have food=nonSP.pl not theme na nk-án vo (...) 3pl.inan 1pl.ex-eat sim ‘If we have no food, that (sago) we eat and (use it to live).’ Examples of the use of bare ido are given in (17) and (18). Whereas (17) allows for the addition of kalau, this is not possible in (18), as the sentence cannot have a conditional interpretation. (17) Na ya-ra vo ya-yar ido, w-óre. then 1sg-go sim 1sg-late theme 2sg-call ‘When I then go and am late, you call.’ (18) I-fukn kapai nan-ya ido d-óve: (...) 3sg-ask mouse giv-3sg.spc theme 3sg-say ‘When he asked the mouse, it said: (...)’
Biak
337
As for the classification of kalau as either a code-switch or a loan, the following should be noted. First, whereas the use of kalau is optional, it does not seem to be motivated by “a conscious choice for stylistic effects”. Rather, the choice for kalau seems to be motivated by (subconscious) structural pressure from the donor language to express the function of “setting the scene” by means of a sentence-initial conjunction. Once the speaker has come to the end of the constituent setting the scene, there is structural pressure from the indigenous language to mark this setting of the scene again, by using the indigenous constituent-final ido. The fact that the use of kalau does not seem to be triggered by the presence of neighboring L2 elements points towards borrowing. The use of kalau is not very frequent, however, given that it is used in only 20 cases out of an estimated 300 cases in which it could be used in addition to ido.4 Although this low frequency does not exclude an analysis of kalau as a loan, a higher frequency would be a more convincing indication of its status as a loan. Finally, following the earlier discussion on shared phoneme inventories, the absence of phonological integration – kalau being realized as [kalau] whereas Biak has no phoneme /l/ – does not say much either in favor of or against borrowing. Turning to another part of speech, the corpus contains a number of examples of the use of the Indonesian complementizer bahwa, an example of which is given in (19). (19) Karena rasul Yohanes i-fawi bahwa jemaat because apostle Yohanes 3sg-know compl congregation Efesus se-terima wós Mansern Ephesus 3pl.an-receive word Lord ve=d-ya. <3sg>pos=3sg-3sg.spc ‘For the apostle John he knew that the congregation of Ephesus had received the Word of the Lord.’ Unlike Indonesian/LM, Biak has no complementizers, but links clauses either asyndetically, as in (20), or by the use of a conjunction, as illustrated by the conjunction voi ‘but’ in (21). (20)
Mám randip=s-ya s-an rovean <3sg>see pig=3pl.an-spc 3pl.an-eat food ve=na. <3sg>pos=3pl.inan ‘He saw that the pigs had eaten his food.’
338
Wilco van den Heuvel
(21) Si-fawi va voi, mankroder an-ya sáe 3pl.an-know not but frog giv-3sg.spc <3sg>go.out bur fyom an-i. from vase giv-3sg.spc ‘They did not know, but the frog had gone out of the bottle.’ It should be noted that the use of bahwa ‘compl’ is clearly restricted to texts that contain quite a number of code switches. This suggests that the use of bahwa ‘compl’ should be analyzed as an instance of code-switching due to language loss, rather than as incorporation of a new structure into the Biak language. Considering numerals, both Biak and Indonesian/LM (like many other Austronesian languages) have a decimal system, and it is very unlikely that one of the two languages has borrowed the system from the other. Borrowing of individual numerals, however, is very common, especially for numbers higher than ten. Finally, the language uses Indonesian/LM words for reference to the days of the week, as well as for the concept ‘year’. Although some older people know of other names for days of the week introduced by missionaries, I have never come across any of these names being used spontaneously.
5. Syntax Apart from the placement of kalau ‘if’ described above, another evident contact phenomenon in syntax is the placement of the borrowed negator bukan ‘not’ in sentence-initial position, whereas the indigenous Biak negator va is used sentence-finally. The use of the Biak negator va ‘not’ is illustrated in (22): (22) Mansren Yesus i-pok fa ve-farander ko va. Lord Jesus 3sg-can cons <3sg>vblz-forget 1pl.inc not ‘The Lord Jesus cannot forget us.’ All of the sentences in the corpus that are introduced with borrowed bukan ‘not’ are at the same time closed off with the Biak negator va. An example of the use of bukan ‘not’ in combination with va ‘not’ is given with the following sentence:
Biak
339
(23) Bukan ko-fafyár biasa va. not 1pl.inc-tell usual not ‘It is not that we just sit normally and tell stories.’ The use of Indonesian/LM bukan ‘not’ and Biak va ‘not’ thus leads to double marking of negation, which however is interpreted as single negation. Both in Indonesian/LM and in Biak, the use of bukan ‘not’ serves to contradict a presumed belief, indicating that the circumstances referred to are not true. More than the bare use of indigenous va ‘not’, the additional use of borrowed bukan ‘not’ stresses the counter-expectational or “contrastive” nature of the negation (cf. Van Minde 1997: 278 on bukang, Macdonald and Darjowidjojo 1967: 160). It is not surprising, then, that in quite a number of instances the negated expression is linked to a following clause by the indigenous counterexpectational or contrastive conjunction voi ‘but’. An example of the combined use of bukan ‘not’, va ‘not’ and the contrastive conjunction voi ‘but’ is given with (24). The sentence is part of an exposition about a boy seeking for a ran-away frog. (24) Indya bukan mankroder an-ya is-ya va voi So not frog giv-3sg.spc 3sg.pred-that not but kapai nan-i-ne d-éke. mouse giv-3sg.spc-this 3sg-go.up ‘So not the frog was there, but the mouse came up.’ More than in the case of kalau ‘if’, the use of bukan ‘not’ seems to be restricted to those speakers that tend to code-switch between Biak and Indonesian. The reason for the relative frequent borrowing of the negator compared to other adverbs is very much comparable to the reasons described for kalau ‘if’ above. Speakers that are used to speak Indonesian/LM in daily life are guided by a (subconscious) structural pressure to express contrastive negation by the use of a sentence-initial adverb. As the Biak negative adverb cannot be used in this position, speakers make use of an Indonesian loan. As soon as the speaker has reached the end of the constituent that is negated, structural pressure from the indigenous language makes him/her use the Biak negator va ‘not’ in addition.
340
Wilco van den Heuvel
6. Conclusion In addition to the lexical borrowing of verbs, this chapter has described several contact-related phenomena that can be qualified as instances of grammatical borrowing. An overview of borrowed formatives is given in Table 1, which also compares their functional and structural properties with those of their Biak counterparts. Comparing the different cases of grammatical borrowing, the following can be observed. The borrowing of modals differs from the other cases in that the modals lack indigenous counterparts fulfilling the same function. The modals are similar to the conjunction atau ‘or’ in that both occupy the same syntactic position as indigenous members of the same lexical category. The borrowing of kalau ‘if’ and bukan ‘not’ (and bahwa ‘compl’, which however is better considered a code-switch), on the other hand, can be said to bring along not only a phonological form (MAT), but also a pattern (PAT) that differs from their Biak counterparts. Comparison of the last two columns Table 1. Borrowed formatives; function and structural position compared to indigenous counterparts Formative
Function
Function compared to counterpart
Stuctural position compared to its counterpart
Used in combination with or instead of counterpart
bisa ‘can’ + harus ‘must’
modal adverb
[no counterpart with same function]
same as modal adverb imbe ‘want’
[no counterpart with same function]
atau ‘or’
disjunctive conjunction
same
same
instead
kalau ‘if’
setting the scene; conditional
loan is semantically more restricted
different: initial vs. final
combination + instead
bukan ‘not’
negative adverb, implying contrast
loan is more specific (implies contrast)
different: intial vs. final
combination
Biak
341
shows that the combination of a borrowed and an indigenous formative is only acceptable (or even required) when the two occupy different structural positions in the clause.
Abbreviations an compl cons du ex giv nonSP inan inc loc
animate complementizer consecutive dual exclusive marker of givenness non-specific inanimate inclusive locative
pc pl pos pred sg sim spc u vblz
paucal plural marker of possession predicative singular simultaneous specific filler verbalizer
Notes 1. To be more precise, the language has an auxiliary verb pok ‘can’, which however is used in negative contexts only, as in ya-pok ya-rir aw va ‘1sg-can 1sglet.go 2sg not’ → ‘I cannot let you go’. For the expression of “knowing” or “being able”, the language makes use the verb fawi, as in the following i-fawi farfyáre ‘3sg-know <3sg>tell’ → ‘he can tell stories’. While the latter sentence was approved in eliciation, the non-elicited examples in the corpus are restricted to negative sentences, like sifawi siwasya va ‘3pl.an-know 3pl.anread not’ →’they cannot read’. There seems to be no alternative for the borrowed Indonesian harus ‘must’. 2. In case the auxiliary precedes the NP, it can only combine with an impersonal 3sg-subject marker, and not with a marker that reflects the person-number and gender properties of the subject of the main verb. 3. The corpus contains more than 600 sentences containing the conjunction ido. In 20 of these sentences, the indigenous conjunction ido is paired with the borrowed conjunction kalau. In addition, the corpus contains 10 sentences where kalau is used on its own, not accompanied by ido. 4. As stated in n. 3, the corpus contains at least 600 sentences containing the conjunction ido. Only in half of these sentences, it would be possible to pair ido with kalau, given the fact that the use of kalau is more restricted than that of ido.
342
Wilco van den Heuvel
References Adelaar, K. Alexander, and D. J. Prentice 1996 Malay: Its history, role and spread. In: Stephen A. Wurm, Peter Mühlhäusler, and Darrell T. Tryon (eds.), Atlas of Intercultural Communication in the Pacific, Asia, and the Americas, 673693. Berlin/New York: Mouton de Gruyter. Gardner-Chloros, Penelope 1995 Code-switching in community, regional and national repertoires. In: Lesley Milroy and Pieter Muysken (eds.), One Speaker Two Languages: Cross-Disciplinary Perspectives on Code-switching, 6890. Cambridge: Cambridge University Press. Macdonald, R. Ross, and Soenjono Darjowidjojo 1967 Indonesian Reference Grammar. Washington (DC): Georgetown University Press. Salmons, Joe 1990 Bilingual discourse marking: Code switching, borrowing, and convergence in some German-American dialects. Linguistics 28: 453480. Van den Heuvel, Wilco 2006 Biak: Description of an Austronesian Language of Papua. Utrecht: LOT. Van Minde, Don 1997 Malayu Ambong: Phonology, Morphology, Syntax. Leiden: Research School CNWS, School of Asian, African and Amerindian Studies.
Sino-Vietnamese grammatical borrowing: An overview Mark J. Alves
1. Overview1 The purpose of this chapter is to describe the influence of the Chinese language(s)2 on Vietnamese with a focus on grammatical aspects. Language contact between Vietnamese and Chinese has led not only to the borrowing of many thousands of Chinese words and influence on the phonological system of Vietnamese but also to some changes in Vietnamese grammar. This influence and borrowing was not, however, sufficient in quantity to move Vietnamese away from its Southeast Asian typological linguistic template, and many of the grammatical characteristics typical of varieties of Chinese are not part of Vietnamese grammar, thereby indicating substantial contact but not one that suggests overwhelming language contact. Following Thomason and Kaufman’s (1988: 50) typology of language contact, we can consider Sino-Vietnamese language contact a case of medium to strong cultural pressure with heavy lexical borrowing and moderate structural influence. This chapter first looks briefly at the history of language contact that Vietnamese has had with several languages. It then covers specific aspects of phonological, morphological, lexical, and grammatical borrowing, with an emphasis on Chinese linguistic influence. Overall, the grammatical influence of Chinese is primarily lexical rather than structural, and many of the grammatical words of Chinese origin were not originally grammatical in Chinese, highlighting internal innovation as a source of some of the change rather than direct influence from Chinese. The rest of this introduction describes the historical and sociolinguistic setting. Vietnamese is the national language of Vietnam, a country with over 80 million people. 90 percent of the population is ethnic Kinh, the ethnic autonym for the Vietnamese.3 As the language of the majority, Vietnamese is, and has been for the last millennium from the end of Chinese rule over Vietnam in the middle 900s CE, the language of administration, religion, education, and most aspects of daily life, although Chinese literature has nevertheless remained a constant influence for two thousand years.4 The other 10 percent
344
Mark J. Alves
of the population in Vietnam speak over 50 different languages belonging to five different language families: Austroasiatic, Austronesian, Hmong-Mien, Sino-Tibetan, and Tai-Kadai. In this multi-lingual situation, Vietnamese has also served as a lingua franca and has spread linguistic elements into languages of some ethnic minority groups.5 The orthography used in Vietnam has moved through three stages over the past two millennia. In the first stage, Chinese was the sole writing script from about 100 bce to sometime after the tenth century ce.6 In the second stage, for several centuries after the end of Chinese rule in Vietnam, the development of Chữ Nôm, a writing system based on Chinese writing techniques7 but representing colloquial Vietnamese vocabulary, had matured in a full literary tradition by the early nineteenth century. Finally, in the third stage, with the arrival of European missionaries in the seventeenth century came the creation of a romanized orthography to represent the pronunciation of Vietnamese, the seminal publication being the 1651 ‘Dictionarium Annnamiticum [sic] Lusitanum, et Latinum’ (a Vietnamese–Portuguese–Latin dictionary) by the Portuguese missionary Alexandre de Rhodes (reprinted in 1991). However, it was not until the beginning of the twentieth century that what came to be called the Quốc Ngữ alphabet completely replaced the use of the Chinese and Nôm scripts. While this was also the French-colonial era, when the French administration was encouraging the use of a Western alphabet in Vietnam, the Vietnamese themselves ultimately adopted this writing system at least in part under a Nationalist movement (DeFrancis 1977). Overall, for the past two thousand years, the primary influence in Vietnamese culture and language has come from Chinese. By the time the Chinese arrived, those living in the Red River Delta region, the presumed ancestors of the Vietnamese, had a developing civilization with possible influence from Tai-Kadai groups, though there is little evidence to portray this community vividly. From the first century bce on, successive waves of Chinese entered Vietnam, settling in Vietnam and marrying into the Vietnamese population, thereby contributing to both Vietnamese culture and the Vietnamese gene pool. The early, formative Sino-Vietnamese era during the Han dynasty set the stage for continued Chinese cultural influence which remains to the present day, primarily in the area of the lexicon through the coining of modern terminology, called ‘Sino-neologisms’ in this chapter. In particular, the written language of Chinese has brought during the three stages described above substantial additions to the Vietnamese lexicon in a wide variety of semantic categories, along with the imported culturally specific ideas and conceptual systems.
Sino-Vietnamese
345
2. Phonology Ultimately, no absolute method exists to determine precisely how Chinese has influenced Vietnamese phonology and to separate borrowing from natural internal changes. Nevertheless, considering the quantity of vocabulary borrowed from Chinese into Vietnamese and various similarities in their phonological systems, such a possibility cannot be casually dismissed. At the same time, it is best to consider also possible language-internal innovations before assuming that similar phonological aspects are the direct result of borrowing from Chinese. The aspects of Vietnamese phonology that most likely show the influence of contact with Chinese are tones and certain classes of consonants. The system of tones in Vietnamese, with a single register height split, fits the Chinese model better than that of Tai-Kadai languages, with a three-way distinction (Haudricourt 1972), supporting the position that Chinese, as opposed to Tai, could have influenced the development of the Vietnamese tone system. However, the precise progression of the development of Vietnamese tones has yet to be agreed completely upon by researchers. Haudricourt’s (1954) original ground-breaking hypothesis of tonogenesis in Vietnamese has been questioned and modified (Gage 1985, Diffloth 1990). Some have suggested that Vietnamese tones are an internal development (Alves 2001). Thurgood (2002) proposes a model for Vietnamese tonogenesis in which it is the result of interacting laryngeal features, a model which does not require language contact to account for it. Based on the archaic four-tone Vietic languages, such as Ruc and Arem (Nguyễn V. L. 1988), which have much less lexical influence from Chinese and which have preserved final /-h/ which corresponds to the hỏi and ngã tones of Vietnamese, it is possible to hypothesize that the borrowing by Vietnamese of numerous Chinese words having the shǎng shēng 上聲 tone category may indeed have influenced the development of the hỏi/ngã tone category in Vietnamese and varieties of the closely related Mường languages. The strong version of this hypothesis – that Chinese borrowing directly led to the creation of the hỏi/ngã tone category – is likely to be proven wrong, while the weak version – that Chinese influenced this tone category to a degree through extensive lexical borrowing of words with that tone category – may turn out to be more feasible. In the area of segmental phonology, it is most likely the case that the retroflex sounds /ʂ/, /ɽ/, and /ʈ/, the labial sounds /f/ and /v/, and the dental sound /z/ are, at the very least, partially the result of the massive borrowing of Chinese loanwords and the phonological changes in Chinese during some
346
Mark J. Alves
periods (in particular, the period during which palatalization was spreading throughout Chinese in the Middle Chinese era) when Chinese vocabulary was being incorporated into the Vietnamese spoken lexicon. Retroflex sounds are typologically marked among the nearby major language groups Tai and Mon-Khmer in Southeast Asia, thus supporting the notions that these sounds were not borrowed from the neighbors of the Vietnamese and that they are less likely random changes. Numerous instances of non-Chinese vocabulary which today have initial retroflex sounds are readily found in the seventeenth-century dictionary by de Rhodes as initial clusters of *tl.8 In some instances, such sounds may have come from collapsed pre-syllables, as suggested by the presumed original Mon-Khmer phonological word structure of earlier stages of Vietnamese, which finally became single retroflex consonants. This apparent monosyllabification of Vietnamese and reduction of clusters into single consonants was most likely due to a combination of natural linguistic tendencies toward the unmarked as well as contact with and massive lexical borrowing from Chinese. Finally, it is perhaps worth mentioning that Vietnamese, like varieties of Chinese, has a maximum syllable CVC structure, with the exception of the glide /w/ in CGVC syllables, also a characteristic of varieties of Chinese. More recent adaptation of Western words, mainly French and English, have brought certain non-native phonemes, such as initial unaspirated /p-/ from French (e.g. pin ‘battery’).9 Such single instances of borrowed sounds are limited to foreign loans and have not altered the Vietnamese phonological system.
3. Nouns, measure words, and nominal structures In the realm of Vietnamese syntactic structure, the only aspect which Chinese most likely influenced is noun phrase structure, specifically regarding the position of measure words before their semantically selected nouns. This order is in contrast with the post-nominal position expected for an otherwise modifier-final language, as well as in contrast with the typology of numerous languages in Southeast Asia (e.g. various Tai languages and Mon-Khmer languages native to neighboring regions in Southeast Asia west of Vietnam) which do exhibit such a sequence. Indeed, it is primarily the Mon-Khmer languages within Vietnamese borders that generally follow the measure–noun order (Jones 1969), suggesting that this is a contact effect in this region from
Sino-Vietnamese
347
China southward, whereas the regions west of Vietnam exhibit a different typological pattern. Overall, this suggests (but certainly does not prove at this point) two distinct regions of long-term language contact, with a chain of influence from Chinese to Vietnamese and then to various ethnic minorities in Vietnam. Support for this position of Chinese influence lies in two key points. First of all, Vietnamese has borrowed at least a dozen classifiers from Chinese and dozens of other general measure words (Nguyễn 1957 and Alves 2001). The borrowing of such vocabulary most likely came through both literary and spoken contact, and the number may have lent to the borrowing of the position of the elements as well. Many of these loanwords have kept their original semantic properties and take the same kinds of nouns as they do in Chinese, while others have developed new semantic constraints on co-occurring nouns.10 Table 1 contains a list of Vietnamese classifiers of Chinese origin. These are more semantically bleached and hence more grammaticalized lexical items than general measure words (e.g. bình 瓶 (píng) ‘bottle of’, bao 包 (bāo) ‘bag of’, and the like, as well as units of measurement, such as length and weight). A handful of the forms in the table are phonologically nativized
Table 1. Borrowed Chinese classifiers Vietnamese
Chinese
Category
bàn bản căn (SV gian) chiếc (SV chích)
壁 (bì) 本 (běn) 間 (jiān) 隻 (zhī)
cuốn (SV quyển) đạo đỉnh đoạn đôi môn phát tòa (SV tọa) vị viên viên
卷 (juǎn) 道 (dào) 頂 (dǐng) 段 (duàn) 對 (duì) 門 (mén) 發 (fā) 座 (zuò) 位 (wèi) 員 (yuán) 丸 (wán)
a unit for flat surfaces (table, hand, foot) a unit for scripts, reports, compositions a unit for houses (1) a unit for vehicles cars, boats, planes (2) a pair of chopsticks unit for books unit for laws, orders, decrees unit for mountains unit for sections, paragraphs, passages couple of shoes, chopsticks, husband/wife unit for a subject/field of study unit for a shot of a firearm, an injection unit for buildings unit for people of high status unit for officials unit for small, round things (pills, tablets, bullets, etc.)
348
Mark J. Alves
words, next to which are listed the Sino-Vietnamese (SV) standard readings of the same etymon. A particularly significant borrowing is the Sino-Vietnamese generic classifier cái (most likely the archaic Chinese form 丐 (gài)),11 which has a virtual article-like function in Vietnamese that, when used alone without quantification, can indicate definiteness and resembles the function of the semantically equivalent lexical item in Cantonese, the default classifier 個 (Cantonese goh, Mandarin gè). With very little doubt, this is a Chinese loanword, though the timing and means of transmission (i.e., spoken or literary sources, or a combination) is not clear. The second issue, in contrast with the numerous borrowed measure words, is that of the post-nominal modifying elements in Vietnamese, including stative verbs, relative clauses, possessive elements (e.g. the native possessive marker của), and demonstratives (e.g. the native demonstrative đó ‘that’), none show the borrowing of grammatical lexical elements. Consider the following sample noun phrase, which shows the various post-nominal modifiers in Vietnamese, all of which precede nouns in varieties of Chinese. (1)
một cái bàn mới đó của tôi one unit table new that of I ‘that new table of mine’
It is also important to recognize that Chinese has had very little influence on the numeral system of Vietnamese, in contrast with the substantial influence of the numeral systems of Tai, and through Tai, Mon-Khmer languages (i.e. the borrowing of the prime decimals, starting at 30). All basic numbers in Vietnamese, which are of Mon-Khmer origin, have consistently maintained their place in both spoken, colloquial language as well as written and/ or formal spoken language. The single lexical item expressing ‘ten thousand’ was borrowed twice from Chinese. The older Han dynasty borrowing muôn is the more colloquial form, while the same word borrowed again during the Tang dynasty, vạn 萬 (wàn), is the literary form, though in fact, neither is commonly used in Vietnamese today. Instead, the native vocabulary items ngình ‘thousand’ and triệu ‘million’ are used. Sino-Vietnamese numerals are generally restricted to specific semantic functions (e.g. ordinals for grades in school). The Sino-Vietnamese oridinal ‘fourth’ -tư (SV tứ) 四 (sì) in particular has a more a general function as the fourth day of the week and fourth month of the year. Overall, Sino-Vietnamese numerals have a peripheral rather than primary role in the Vietnamese numeric system.
Sino-Vietnamese
349
While the numeral system in Vietnamese has been unaffected by Chinese, the system of quantification has been more notably influenced. Vietnamese has accepted some Chinese lexical items of general quantification, including các 各 (gè) ‘various’ and mỗi 每 (měi) ‘each’. The status of these words as Chinese is solid as they are listed in Sino-Vietnamese dictionaries and have very similar semantico-syntactic properties. The term mọi ‘every’ is also likely a phonologically nativized earlier borrowing form of mỗi.12 The Vietnamese pronoun system has also received some influence from Chinese, but not in ways typically seen among varieties of Chinese, indicating both borrowing and innovation in Vietnamese. The three examples include the bound prenominal chúng 眾 (zhòng) (third-person plural marking before pronouns, e.g., chúng nó (plural3rd singular) ‘they’), the unbound ta (SV tha) 他 (tā) (third-person plural marking after pronouns, e.g., cô ta (miss-plural) ‘those ladies’), and the unbound y 伊 (yī) (third-person singular). Only the latter of the three is parallel with original Chinese usage,13 but the first two are bound rather than free forms. Indeed, it must be acknowledged that these lack complete certainty in their origins and can only be considered possible Chinese loans based on their phonetic and semantic similarity. The process of grammaticalization, if it turns out to be so, would need to be identified before a higher degree of certainty could be reached. Another important grammatical aspect that ultimately comes from Chinese is the use of terms of address, though their pronominal function in Vietnamese is an extreme semantico-syntactic extension of their original usage in Chinese, one which parallels that of other Southeast Asian languages. Vietnamese pronouns have for the most part been replaced by a system of familial-based terms of address – many but not all of which come from Chinese – that indicate age, gender, and degree of formality and politeness, including both standard Sino-Vietnamese (e.g. cô 姑 (gū) ‘miss (more formal)’) and some nativized Sino-Vietnamese forms (e.g. chị (SV tỉ) 姐 (jiě) ‘miss (less formal)’) (Benedict 1947). Such terms are highly grammaticalized; they have complete pronominal referential functions and in fact have ‘floating’ reference, being able to refer variously to first, second, or third person in context. While such terms do function as terms of address in Chinese, they have not gone this far in the direction of complete pronominal usage as in Vietnamese. As in the English translation of sentence (2), in which ‘Sir’ is an optional addition, in modern Chinese, such terms generally appear outside the sentence core, whereas in Vietnamese, the term of address is the subject inside the sentence matrix, not an external, optional element.
350
Mark J. Alves
(2)
anh đi đâu vậy? you Sir go where thus ‘Where are you going, Sir?’
4. Verbs One area of significant Sinitic influence is in the Vietnamese system of preverbs, especially in the systems of negation (see Table 3) and voice. One post-verbal complement has also been borrowed. A number of negation words in Vietnamese appear to be grammaticalized words of Chinese origin. First, it has been posited (Nguyễn P. P. 1996) that the negation word không is a grammaticalized form of the homophonous Sino-Vietnamese không 空 (kōng) meaning ‘void’. Usage of this form has become the dominant form over other native Vietnamese negation words, such as chẳng. Next, two lexical items, chớ (SV trừ) 除 (chú, meaning ‘exclude’ in Chinese) and đừng (SV tình) 停 (tíng, meaning ‘stop’ in Chinese), both meaning ‘don’t’ in Vietnamese, are probably of Chinese origin. Neither form had these semanticosyntactic properties when borrowed. Note that chớ is also a connector word, as discussed in Section 5. The passive-like elements borrowed from Chinese have been discussed (Matisoff 1991, Alves 2001). The Sino-Vietnamese form được (SV đắc) 得 (dé) shares some of the various functions as seen in Chinese, including passive-like, abilitative, and resultative functions (see Matisoff 1991). In addition, the two forms do ‘by’ and bị ‘negatively affected by’ show similarly passive-like functions. While do and bị are unarguably standard Sino-Vietnamese borrowings of relatively recent borrowing (perhaps early twentieth century), được is more difficult to account for in terms of its unexpected phonetic realization (i.e., both tone height and vowel quality). However, considering its significant phonetic and semantico-syntactic overlap, including shared patterns of grammaticalization in varieties of Chinese and Vietnamese, it should be included until such time as a stronger alternative explanation can be provided. The pre-verb đang is likely from Chinese 當 (dāng), meaning in Chinese ‘at the time of’. In Vietnamese, it is used to indicate an action in progress in Vietnamese. If it is Chinese in origin, it has developed a different syntactic distribution from that of Chinese (i.e., it appears between a subject and verb in Vietnamese main clauses but before a subject in dependent clauses in Chinese).
Sino-Vietnamese
351
The Sino-Vietnamese word qua (SV quá) 過 (guò) ‘to cross’ has a postverbal directional function parallel to that in Chinese, a preposition or adverblike function expressing the meaning ‘across’. The etymon is also seen as an intensifier in Vietnamese, as discussed in Section 5. However, despite the fact that Vietnamese has a rich variety of post-verbal directional elements, besides this single term, no other directional terms have been borrowed from Chinese. (3)
đi qua đường. go cross street ‘Go across the street.’
5. Adverbs, conjunctions, locative terms and others For the most part, the grammatical adverbs and conjunctions that have been borrowed from Chinese, as listed in Table 2, were borrowed grammaticalized and have largely kept their original senses with little or no modification. Nevertheless, it is worth noting that several of the items in Table 2 are phonologically nativized loans with standard Sino-Vietnamese counterparts with-
Table 2. Borrowed Chinese connective words Vietnamese
Chinese
Gloss
bèn (SV tiện) chỉ chính chứ (SV trừ) cùng (SV cộng) giá (SV giả) hiện tại hoặc nhưng sở dĩ tại thậm chí tuy nhiên và (SV hòa) vì (SV vị)
便 (biàn) 只 (zhǐ) 正 (zhèng) 除 (chú) 共 (gong) 假 (jiǎ) 現在 (xiàn zài) 或 (huò) 仍 (réng) 所以 (suǒ yǐ) 在 (zài) 甚至 (shén zhì) 雖然 (suī rán) 和 (hé) 為 (wèi)
and then only just/precisely but not (contrastive) together if/supposing currently or but the reason why… because even however and because
352
Mark J. Alves
out those developed usages. It must be admitted that more data (e.g. written records and both syntactic and phonological comparative research of Vietnamese and Chinese) will have to be provided to verify these phonological variants as genuine Chinese loanwords.14 A handful of locative terms have also been borrowed. These include (a) tại 在 (zài) ‘to be at’ (with a more formal, literary flavor), which is the original meaning in Chinese (and which has developed the additional meaning ‘because’ in Vietnamese), (b) ngoài (SV ngoại) 外 (wài) ‘outside (of)’, (c) gần (SV cận) 近 (jìn) ‘near (to)’, and (d) bên (SV biên) 遍 (biān) ‘side’.15 Notably, a number of comparative/intensifying words are likely Chinese loanword candidates that have grammaticalized since entering Vietnamese. These include (a) bằng (SV bình) 平 (píng) ‘equal to’ (originally ‘level/balanced’), (b) nhất 一 (yī) ‘most’ (originally ‘one’), (c) giống (SV chủng with the basic meaning ‘type/kind’) 種 (zhǒng) ‘resemble/similar to’, (d) thật (SV thiệt) 實 (shí) ‘truly’, and (e) quá 過 (guò) ‘extremely’ (with the basic meaning of ‘to pass over’) (Alves 2005). More data, such as examples of these in ancient Nôm writings, would be required to label these with more certainty. One item which has maintained its original semantico-syntactic properties as in Chinese is the form như 如 (rú) ‘as/like’. The Sino-Vietnamese form tự 自 (zì) means ‘to do by oneself’, though it has a much more restricted usage than in Chinese, where it has a full reflexive function.
6. Syntax Vietnamese is a topic–comment language with right-branching syntactic structures, like many other languages in Southeast Asia. Grammatical borrowing from Chinese has been primarily lexical with relatively little evidence of structural borrowing, the one exception being the position of measure words in quantified noun phrases, as discussed in Section 3.
7. Lexicon and timing of borrowing While Vietnamese has definitely borrowed a modest amount of vocabulary from French and quite probably from Tai groups before contact with the Chinese,16 neither loanword source has contributed in any way to Vietnamese grammar or syntax.17 Loanwords from each of these sources number at
Sino-Vietnamese
353
most several dozen from French (many which have fallen out of use in the post-colonial era) and a few dozen from Tai. These loanwords belong to rather restricted ranges of semantic domains. Words of a Tai origin suggest contact at a time when agricultural techniques of the Tai peoples were passed on from the ancestors of the Tai to the ancestors of the Vietnamese. Loanwords in this category include domesticated livestock such as vịt (Thai pèt) ‘duck’ and đực (Thai t ɬ`k) ‘young male animal’ and terms related to farming such as rẫy (Thai rây) ‘dry field’, đồng ruộng (Thai thûŋ) ‘field’, and mương (Thai mɬaŋ) ‘ditch’ (Nguyễn 1995: 322).18 French loanwords that remain in Vietnamese today, as opposed to vocabulary that fell out of use in Vietnamese after the French left Vietnam, are primarily introduced western cultural terms and modern (at the time of borrowing) accoutrements, such as áo sơ mi (Fr. chemise) ‘shirt’, xà phòng (Fr. savon) ‘soap’, xe ô tô (Fr. automobile) ‘automobile’, bơ (Fr. beurre) ‘butter’, pin (Fr. pile) ‘battery’ and pa tê (Fr. pâté) ‘pate’,19 among others. The European presence did lead to the spread of an alphabetic writing system in Vietnam, though lexical influence from French was nevertheless superficial, resulting in the permanent borrowing of some dozens of words (Barker 1969) and some dozens more which have fallen out of usage. More recently, some English loans have entered Vietnamese, but not in large numbers and without certainty of permanent or even long-term presence in the Vietnamese lexicon, though Vietnamese dictionaries appear to list increasing numbers of them. Some examples include mít tinh ‘meeting’, vi rút ‘virus (either computer- or health-related)’, and in tơ nét ‘internet’. Such words tend to be related to, but are not restricted to, technology. In contrast with the loanwords from other sources which were borrowed in relatively small numbers within relatively short periods of time, Chinese vocabulary in Vietnamese consists of several thousand words borrowed over a period of two thousand years.20 As discussed in Section 1, the starting point of Sinitic borrowings began possibly as early as 100 bce during the end of the Western Han dynasty. Words that were borrowed from Chinese in this era are often considered by the Vietnamese to be Nôm vocabulary, meaning essentially native Vietnamese vocabulary. This perception is due to both the substantial phonological and in some cases semantic changes over such an expanse of time and the lack of written records to verify their borrowing. The second stage of borrowing happened as a result of the spread of the Chinese rhyming dictionaries throughout East Asia during the Tang dynasty (618907). These books brought with them the entirety of literary Chinese, though this vocabulary was, of course, brought into use in spoken Vietnam-
354
Mark J. Alves
ese over time. This borrowing of Chinese through written language continued throughout the second millennium ce, though spoken language contact was likely also a factor as trade and cultural intake continued. The third stage of borrowing was in the beginning of the twentieth century, when the Japanese had been translating Western concepts by utilizing classical Chinese lexical material. This system of linguistic adaptation spread into Chinese, Korean, and Vietnamese (Sinh 1993). Over a period of several decades, thousands of new Sino-neologisms entered Vietnamese, though it was ultimately a mixture of borrowings from both Japan and China. One of the more difficult problems is, when dealing with language contact over such a lengthy period of time, determining the timing of borrowings. In some cases, such as the influence of Chinese on the Vietnamese use of the passive voice, modern written records show that this corresponds with the timing of large quantities of translations of Western writings into Chinese. In other cases, phonological evidence is the source of the identification of timing. For example, Wang Li (1948) posited that giống ‘type’ was a word borrowed most likely during the Han dynasty, and the same word was again borrowed as the Sino-Vietnamese (SV) literary reading chủng 種 (zhǒng). Mei (1970) identified this as part of a pattern of both the palatalization of initial consonants and the correspondence between the Old Sino-Vietnamese sắc tone and the Middle Chinese shǎng shēng 上聲 tone category (whereas Sino-Vietnamese proper borrowed in the Middle Chinese era was the hỏi tone, as in the previous example). Another systematic phonological correspondence is between literary Sino-Vietnamese readings having the nặng tone and the nativized cognates with the huyền tone, such as the standard, literary readings tự自 (zì) ‘from’, dụng 用 (yòng) ‘use’, and loại 類 (lèi) ‘type’ versus their nativized, spoken readings, từ, dùng, and loài respectively. Considering their otherwise similar phonetic shape indicates that they are more likely relatively recent borrowings, perhaps within the past several centuries rather than in the Han-dynasty/ Pre-Tang era. In many other instances, no patterns or written records exist to assist in verifying approximate dates of the borrowings, which weakens somewhat the argument for these words as Chinese in origin. Regardless, with the weight of the lexical evidence and numerous patterns of phonological correspondences, such words with variant pronunciations must at least be considered likely loanwords. As a result of the similarity in general of word formation patterns throughout China and Southeast Asia, bisyllabic compounds consisting of two syllables each with distinct meanings cannot be said to be the direct influence of
Sino-Vietnamese
355
Table 3. Borrowed Chinese derivational word forms Vietnamese
Chinese
Gloss
bất phản phi vô hoá học
不 (bù) 反 (fǎn) 非 (fēi) 無 (wú) 化 (huà) 學 (xué)
‘in/un-’ ‘anti-‘ ‘non-’ ‘non-’ ‘-ize’ ‘-ology’
Chinese despite the borrowing of significant quantities of Chinese bisyllabic compounds, such as the Sino-neologisms in the beginning of the twentieth century. There are not too many instances of morphological borrowing (e.g. prefixes or suffixes) since many Chinese compounds have been borrowed unanalyzed. There are, however, a number of derivational morphemes that have been borrowed from Chinese, as in Table 3, that have some limited amount of productivity in Vietnamese and can be used to create new terms in Vietnamese. The examples here are generally in line with Vietnamese typology (e.g. negation indicated before the negated aspect, as is the case in Vietnamese syntax). In contrast, when a typological pattern in Chinese lexical material does not match that of Vietnamese, expected modifications are seen. In some cases, the Vietnamese head–modifier order has been applied to Chinese loanwords, such as the order of the names of continents. The Chinese term for ‘Europe’ 歐洲 (ōu zhōu) (Europe-continent), Sino-Vietnamese Âu Châu, was used in its original Chinese order earlier in the twentieth century, but later, the native-like order Châu Âu (continent-Europe) became dominant, as was the case with expressions for other continents. In addition, there are some calques such as xe lửa ‘train’ (vehicle-fire) and máy bay ‘airplane’ (machinefly) which also follow the expected Vietnamese order. These two words have largely replaced the earlier Sino-Vietnamese forms hoả xe 火車 (huǒ chē, (fire-vehicle)) ‘train’ and phi cơ 飛機 (fēi jī, (fly-machine)) ‘airplane’. While the terms with Chinese order are not completely out of usage, the forms that follow the native Vietnamese patterns are dominant. Overall, lexical borrowing has not influenced the word-formation patterns seen in Vietnamese, which has retained its typological Southeast Asian pattern of head-modifier. Also, as mentioned below in Section 8, the reduplicating patterns common to varieties of Chinese are entirely absent in Vietnamese.
356
Mark J. Alves
8. Summary While Chinese has been the primary and most influential donor of vocabulary to the Vietnamese lexicon, this borrowing has not resulted in relexification or major restructuring of Vietnamese syntax. There does appear to have been some influence in Vietnamese phonology and some influence in the position of measure words in quantified noun phrases, but otherwise, grammatical borrowing has been largely in the area of grammatical vocabulary. The amount of grammatical vocabulary is substantial, including dozens of measure words, two dozen or so words with connective functions, a good handful of grammatical adverbs and preverbs, some locational terms, and a few other numerals and quantity expressions. To further understand the Sino-Vietnamese borrowing situation, it is also important to consider in a borrowing situation what was not borrowed. Norman’s (1988: 13) list of Sino-Tibetan comparisons in a comparative wordlist of six Chinese and non-Chinese Sino-Tibetan languages shows two dozen basic vocabulary items, none of which have replaced native Vietnamese words (with the single exception of lạnh (SV lãnh) 冷 (lěng) ‘cold’). Clearly, however much lexical influence Vietnamese has received, it has not been relexified and has retained a significant amount of vocabulary of Mon-Khmer stock. Vietnamese does not even share characteristic vocabulary of the nearby Chinese Yue languages, such as Cantonese.21 Beyond lexical borrowing, there is the minimal amount of syntactic borrowing. Were the Sino-Vietnamese contact heavy enough, we might expect to see other kinds of grammatical, structural aspects of Chinese in Vietnamese, but most of the grammatical typological characteristics common to varieties of Chinese are in fact not seen in Vietnamese. Notable characteristics of varieties of Chinese include verbal compounds of ability (verb–得 (dé)–resultative) and direction (notably, the verb plus two-syllable directional terms); the use of reduplication (A-not-A) in questions; the Chinese style of reduplication of the second syllable in two-syllable terms (ABB); the use of post-positional locative nouns; and the position of modifiers, possessives and demonstratives before head nouns. While Vietnamese does make use of a substantial number of sentence final particles with modal properties to express politeness, assertion, commands, and others, there is no strong evidence suggesting that they have come from Chinese.22 As can be seen, Sino-Vietnamese borrowing did not lead to the borrowing of characteristic aspects of most varieties of Chinese. Moreover, internal innovation, rather than borrowing in a direct sense, has led to the develop-
Sino-Vietnamese
357
ment of grammatical functions of Sino-Vietnamese vocabulary that originally lacked those specific semantico-syntactic characteristics. Questions remain about the nature of the contact through which borrowing occurred. Was this borrowing primarily through spoken contact or from texts that the Vietnamese elite spread into spoken Vietnamese? In fact, it appears that the borrowing has only been partially through spoken contact. It seems that, though bilingualism in Chinese and Vietnamese has clearly played a part in this contact situation, a significant amount of Chinese was likely transmitted into Vietnamese from Chinese writings by Vietnamese leaders, scholars, and other socially influential figures, a situation which best accounts for the reason why colloquial grammatical elements common to varieties of Chinese did not enter Vietnamese. Notes 1. Sources of data regarding borrowing into Vietnamese are mentioned when relevant. Otherwise, the data comes from the author’s research, and any discrepancies are the author’s responsibility. 2. The simplified term ‘Chinese’, as used in this chapter, can refer variously to the entire group of Sinitic languages, various subgroups within the Sinitic branch of Sino-Tibetan, or the written language used by all of those groups. Specific usages of ‘Chinese’ for the arguments in the chapter are noted as needed. 3. The term Kinh is itself Chinese in origin, jīng 京. 4. See Taylor (1983) for a more detailed discussion of the first thousand years of Sino-Vietnamese contact. 5. The issue of borrowing of Vietnamese elements has not been explored extensively in linguistic literature. 6. No records exist to demonstrate whether or not the Vietnamese had developed an indigenous writing system before the arrival of the Chinese, as some have suggested. 7. Most often, as in Chinese, Chữ Nôm characters use a combination of Chinese radicals, one with a phonetic element and one with a semantic element. See Nguyễn D. H. (1990) for more discussion. 8. See Maspero (1912), Ferlus (1981, 1992), and Nguyễn (1995) for more discussion on the historical reduction of initial clusters and development of initials in general in Vietnamese. 9. Final /-p/ is a native sound in Vietnamese. Initial /p-/ was lost most likely due to changes in Chinese from *p to *f and perhaps partly due to merging with earlier Vietnamese *β. 10. This is not unlike the variation in semantic properties of the same classifier in different varieties of Chinese.
358
Mark J. Alves
11. In Alves (2005), this was linked the generic Chinese classifier個 (gè), which is less likely the source of Vietnamese cái. Nevertheless, the phonetic and semantic similarities and the dominant use of in varieties of Chinese 個 (gè) do put into question whether gài 丐 or gè 個 is the source. 12. Another possibility is that it is a retention of the original Mon-Khmer word meaning ‘one’, which could have grammaticalized, as the cognate in the MonKhmer language, Pacoh (Alves 2005: 76). If this hypothesis is valid, this could account for the fact that Vietnamese một ‘one’ has an added /t/; it is an instance of phonological distinction due to semantico-syntactic differences. 13. The Chinese form yī 伊, while not part of all spoken varities of modern Chinese, is the standard third-person pronoun in Taiwanese. 14. Not included in the list is the quintessential Vietnamese topic–comment linking thì, which may have developed from the homophonous Chinese form 時 (shí) meaning ‘time’. The origins of this form will remain speculative until more explicit data become available. 15. While it is tempting to include Vietnamese trong ‘inside’ as a nativized form of Sino-Vietnamese trung 中 ‘inside’, there is a competing Mon-Khmer etymon (cf. Pacoh kallúng and Khmer knoŋ, with presyllabic telescoping to retroflex /ʈ/ in Vietnamese, as discussed in Section 2). 16. The French borrowings are certain since they are recent, phonologically close to their source words, and clearly connected with modern cultural innovations. While Tai-Vietnamese contact appears certain, Tai borrowings can only be determined by a systematic comparison of reconstructions going back two millennia or earlier, making such forms likely but far from certain candidates. Also, the possibility that Tai borrowings happened after Vietnamese contact with the Chinese cannot be excluded but is difficult to verify. 17. The general typological similarities between Vietnamese and other Southeast Asian languages cannot be readily attributed to any other language and can only be considered an areal affect with no clear, single direction of influence or borrowing. 18. The modern Tai forms are given for convenience. These forms have been checked for Proto-Tai forms using the ‘Proto-Tai’o’matic’ lexical database, which contains a compilation of several Proto-Tai reconstructions, at http://crcl.th.net/. 19. See last paragraph in Section 2 on loanword phonology of Western words. 20. Studies on Sino-Vietnamese vocabulary and these layers of vocabulary include the works of Wang (1948), Ðào (1979), Tryon (1979), and Pulleyblank (1981). 21. The exception is the verb “to see” thấy, most likely related to Chinese 睇 (tì), Cantonese tái. 22. Where are the Vietnamese sentence particles à, which expresses surprise, and ạ, which expresses politeness. Chinese as well as other languages in the region have sentence particles with similar unmarked phonological material, essentially eliminating the ability to determine whether borrowing has occurred. One form of note is the Vietnamese sentence-final particle mà, which suggests
Sino-Vietnamese
359
that what is said is something previously asserted and should be known by the speaker. This is not unlike Mandarin Chinese ma 嘛. Again, however, the phonological material is unmarked and harder to confirm as borrowed material.
References Alves, Mark J. 2001 What’s so Chinese about Vietnamese? Papers from the Ninth Annual Meeting of the Southeast Asian Linguistics Society, 221242. Ed. Graham W. Thurgood. 2005 Sino-Vietnamese grammatical vocabulary and triggers for grammaticalization. The 6th Pan-Asiatic International Symposium on Linguistics. 315332. Hà Nội: Nhà Xuất Bản Khoa Học Xã Hội (Social Sciences Publishing House). Barker, Milton E. 1969 The phonological adaptation of French loanwords in Vietnamese. MonKhmer Studies 3: 138147. Benedict, Paul K. 1947 An analysis of Annamese kinship terms. Southwestern Journal of Anthropology 3: 371390. Đào, Duy Anh 1979 Chữ Nôm: nguồn gốc, cấu tạo, diễn biến (Chu Nom: Origins, formation, and transformations). Hà Nội: Nhà Xuất Bản Khoa Học Xã Hội. DeFrancis, John 1977 Colonialism and language policy in Viet Nam. New York: Mouton. De Rhodes, Alexandre 1991 Từ Ðiển Annam-Lusitan-Latinh (Thường Gọi là Từ Ðiển Việt-Bồ-La). Ho Chi Minh City: Nhà Xuất Bản Khoa Học Xã Hội. (First publ. 1651). Ferlus, Michel 1981 Sự biến hóa của các âm tắc giữa (obstruentes mediales) trong tiếng Việt (Changes of medial obstruents in Vietnamese). Ngôn Ngữ Học 1981/2: 121. 1992 Histoire abrégrée de l’évolution des consonnes initials du Vietnamien et du Sino-Vietnamien. Mon-Khmer Studies 20: 11125. Gage, William W. 1985 Vietnamese in Mon-Khmer perspective. In: S. Ratankul, D. Thomas, and S. Premisirat (eds.), Southeast Asian Linguistics Presented to Andre-G. Haudricourt, 493524. Bankok: Mahidol University. Diffloth, Gérard 1989 Proto-Austroasiatic Creaky Voice. Mon-Khmer Studies 15: 139154.
360
Mark J. Alves
Diffloth, Gérard 1990 Vietnamese as a Mon-Khmer language. Papers from the First Annual Meeting of the Southeast Asian Linguistics Society, ed. by Martha Ratliff and Eric Schiller, 125139. Haudricourt, André G. 1954 Sur l’origine de la ton de Vietnamien. Journal Asiatique 242: 6982. 1972 Two-way and three-way splitting of tonal systems in some far-eastern languages. In: J. G. Harris and R. B. Noss (eds.), Tai Phonetics and Phonology, 5886. Bangkok: Central Institute of English Language. Jones, Robert B. 1969 Classifier constructions in Southeast Asia. Journal of the American Oriental Society 90 (1): 112. Maspero, Henri 1912 Études sur la phonétique historique de la langue Annamite: Les initiales. Bulletin de l’École Françoise d’Extrême-Orient 12: 1127. Matisoff, James A. 1991 Areal and universal dimensions of grammaticalization in Lahu. In: E. Traugott and B. Heine (eds.), Approaches to Grammaticalization, Volume 1, 383453. Amsterdam/Philadelphia: John Benjamins. Mei, Tsu-Lin 1970 Tones and prosody in Middle Chinese and the origin of the rising tone. Harvard Journal of Asiatic Studies 30: 86110. Nguyễn, Đình Hoà 1957 Classifiers in Vietnamese. Word 13 (1): 124152. 1966 Vietnamese–English Dictionary. Rutland, Vermont: Charles E. Tuttle Co. 1990 Graphemic borrowing from Chinese: The case of Chữ Nôm, Vietnam’s demotic script. Bulletin of the Institute of History and Philology, Academia Sinica 61: 383432. Nguyễn, Phú Phong 1996 Negation in Vietnamese and in some of the Viet-Muong Languages. Pan-Asiatic Linguistics: Proceedings of the Fourth International Symposium on Languages and Linguistics, 810 Jan. 1996, Vol. II: 563– 568. Thailand: Mahidol University at Salaya, Institute of Language and Culture for Rural Development. Nguyễn, Tài Cẩn 1995 Gíao trình lịch sử ngữ âm tiếng Việt [Textbook of Vietnamese historical phonology]. Hà Nội: Nhà Xuất Bản Gíao Dục. Nguyễn, Tài Cẩn, and Hoàng Dũng 1994 Về các từ gốc Hán cổ tiếng Việt xử lý bằng thuỷ âm tắc bên (lateral stops) [On the Vietnamese words of Old Chinese origins with lateral stops]. Ngôn Ngữ 1994 (2): 17.
Sino-Vietnamese
361
Nguyễn, Văn Lợi 1988 Sự hình thành đối lập đường nét thanh điệu bằng/không bằng trong các ngôn ngữ Việt-Mường (Trên tư liệu tiếng Arem và Rục) (The features of the comparative formation of the level/non-level tones in some Viet-Muong languages (in materials of the Arem and Ruc languages)). Ngôn Ngữ 2: 39. Norman, Jerry 1988 Chinese. Cambridge University Press. Pulleyblank, Edwin G. 1981 Some notes on Chinese historical phonology. Bulletin de l’école Françoise d’Extreme-Orient 277288. Sinh, Vinh 1993 Chinese characters as the medium for transmitting the vocabulary of modernization from Japan to Vietnam in Early twentieth century. Asian Pacific Quarterly 25 (1): 116. Taylor, Keith W. 1983 The Birth of Vietnam. Berkeley: University of California Press. Thomason, Sarah, and Terrence Kaufman 1988 Language Contact: Creolization and Genetic Linguistics. Berkeley and Los Angeles: University of California Press. Thurgood, Graham 2002 Vietnamese and tonogenesis: Revising the model and the analysis. Diachronica 19 (2): 333363. Tryon, Ray 1979 Sources of middle Chinese phonology: A prolegomenon to the study of Vietnamized Chinese. MA thesis, Southern Illinois University. Wang, Li 1948 Hanyueyu yanjiu (Research on Sino-Vietnamese). Lingnan Xuebao. 9.1.196. (Repr. 1958 in the Hanyushi Lunwenji. Beijing Kexue Chubanshe, 290406.)
Recent grammatical borrowing into an Australian Aboriginal language: The case of Jaminjung and Kriol Eva Schultze-Berndt
1. Introduction This chapter deals with the nature and extent of grammatical borrowing into an Aboriginal language of northern Australia from Kriol (also called Roper River Kriol, ROP). This is an English-lexified Creole that has arisen relatively recently out of a colonial contact situation and now functions as a lingua franca between indigenous people throughout a large area of northern-central Australia. The recipient language is a dialect cluster comprising the two closely related and mutually intelligible varieties Jaminjung and Ngaliwurru (henceforth Jaminjung for short; DJD). They are the only remaining members of the Jaminjungan or Yirram subgroup of the Mirndi family, one of the NonPama-Nyungan language families. All observations are based on my own (ongoing) fieldwork on the language since 1993. Although Jaminjung has to be considered a severely endangered language (there are possibly only around 50, mostly elderly, fluent speakers today), it will be argued that, for the remaining fully competent speakers, the extent of grammatical borrowing from the dominant language Kriol is limited to function words, especially in the domain of connectors and discourse-structuring devices, and thus fits in with the predictions made in the literature for the accessibility of grammatical morphemes to borrowing. Lexical borrowing, unsurprisingly, also occurs; the integration of verbal loans is discussed here in some detail because it exposes an interesting feature of the recipient language. In contrast, borrowing of phonological features and of structure (“pattern”) does not occur. The issue is made more complicated, however, by the many phonological and grammatical similarities between Kriol and Jaminjung which are arguably not the result of borrowing or “adoption” (in the sense of Johanson 2002) but rather of substrate influence (“imposition”) on Kriol from languages similar to Jaminjung, if not Jaminjung itself. Although the degree of substrate influence in Creole languages continues to be a debated topic, I will assume
364
Eva Schultze-Berndt
substrate influence here if the phenomenon in question is not attested in the lexifier language, English, but is widespread (in terms of “pattern”, not “matter”) within the linguistic area of which Jaminjung is a part. Such features include, for example, the phoneme inventory (Section 4), the absence of a copula, the existence of general subordinate clauses (“adjoined relative clauses”) in the functions of both nominal modification and adverbial, the pronominal system, the semantics of case categories (see Section 5 for an example), and the semantics of TAM categories (see Section 6 for an example). A second issue that arises more generally in a study of grammatical borrowing is the question of how to distinguish borrowing from code-switching. Code-switching to Kriol is common in the everyday language use of the remaining Jaminjung speakers, since the former can be considered as the dominant language even in in-group interaction in terms of its frequency of use (see also Section 2). While I do not claim to have solved this problem, and will raise it again in the concluding section, I attempt here to at least make explicit the criteria according to which certain items have been considered borrowings rather than switches. One often-cited criterion, that borrowings are used even by monolingual speakers of the recipient language, is obviously not applicable, since there are no monolingual speakers. On the other hand, the criterion that only those items should be considered borrowings that have no equivalent in the receiving language seems too strict to account for the actual patterns observed. The criterion of recognisability also seems somewhat too strict, since Jaminjung speakers usually recognise a word as “English”, and point this out in situations which are likely to trigger purist sentiments, such as dictionary production. The main criteria employed here to identify borrowings as opposed to switches are therefore the following: (a) The item is used frequently and across different speakers. (b) The item regularly occurs in utterances with Jaminjung as a matrix language; the criterion for matrix language status, in turn, is the use of a Jaminjung inflecting verb (see also Section 6.2 on the word class status of verbs in Jaminjung). Kriol auxiliaries are never found in co-occurrence with an inflecting verb in Jaminjung. (c) The item also occurs as a single-word item in utterances that otherwise show no influence from or switches into Kriol, i.e. the item does not just occur as part of whole phrases or clauses in Kriol. (Alternatively, these instances could of course be regarded as insertional code-switching in the sense of Muysken 2000).
Jaminjung and Kriol
365
(d) The item, when it occurs, is completely prosodically integrated (not associated with hesitation phenomena). Sections 3 to 8 of this chapter provide a discussion of grammatical borrowings from Kriol into Jaminjung, contrasted with similarities due to substrate influences from northern Australian languages in Kriol. They are preceded by an account of the sociolinguistic situation that brought the two languages into contact (Section 2), and followed by a brief section on lexical borrowings (Section 9) and concluding remarks (Section 10).
2. Sociolinguistic background The traditional country of the Jaminjung and Ngaliwurru people is situated roughly between the present-day settlements of Victoria River Crossing and Timber Creek in the Northern Territory. The Jaminjung were hunter-gatherers with a nomadic life-style; they engaged in trade and maintained ceremonial and inter-marital relationships with members of neighbouring language groups. The resulting multilingualism has led to linguistic convergence across different language families similar to that described by Heath (1978) for East Arnhem Land. Structural convergence manifests itself in the areas of phonology and prosody, marking of grammatical relations, tense–aspect–mood categories, complex predicate formation, complex sentence formation, the use of discourse particles, and pragmatic conditioning of word order. A discussion of these earlier layers of borrowings is however outside the scope of the present chapter. The first people of European descent came to the area in 1834, and the establishment of cattle stations began in the 1880s. There can be no doubt that the early contact history throughout northern Australia was in many cases violent, including massacres on the part of the colonizers (Rose 1991), as well as being accompanied by the spread of previously unknown diseases. Sooner or later most of the indigeneous inhabitants would be forced – by actual coercion or by the desire for safety – to join the work force of the cattle stations, as essentially unpaid labour. Almost all older Jaminjung and Ngaliwurru speakers worked on cattle stations earlier in their lives, as stockmen, cooks, builders, domestic workers, or police trackers. After the introduction of equal wages, forced labour gave way to unemployment, and land and power still remain largely in the hand of non-indigenous people.
366
Eva Schultze-Berndt
The genesis and spread of Kriol and the ongoing process of language shift from the dozens of Aboriginal languages of northern Australia to this new language can be regarded as a consequence of the disruptions described above. Kriol originates in the English-based Pidgin used between the first colonizers and the indigenous inhabitants of the Sydney area (Troy 1993; Tryon and Charpentier 2004) which subsequently spread inland and north, and then further to the Pacific Islands. The records of the Pidgin used in the Northern Territory in the late 19th and early twentieth century cited by Harris (1986) bear a close resemblance to Kriol as it is spoken today, which in turn has many similarities with Pacific Pidgin and Creole languages such as Tok Pisin and Bislama. A stabilization and standardization of the Pidgin in northern Australia was brought about by the necessity of communication between the increasing numbers of Aboriginal people working on the stations, the (primarily) English-speaking pastoralists, and the non-English-speaking colonists. According to some authors (Munro 2005: Ch. 2), Kriol emerged, with concomitant substrate influences, mainly as the result of this stabilization. Others, in particular Harris (1986), assume that creolization occurred abruptly early in the twentieth century at an Anglican mission at Roper River (close to the present-day Ngukurr), which provided refuge to the survivors of several language groups. Harris (1986: 306312) argues that a peer group of children, who lived in dormitories and were thus separated from adults for large parts of the day, needed a common language, and adopted and creolized the existing Pidgin. World War II and later the collapse of the pastoral industry led to increasing mobility which favoured the adoption of Kriol as the lingua franca it is today (Munro 2000). If Harris’ account of creolization at Roper River is correct but if one allows, as Munro (2005) does, for substrate influence, eight languages from four different Non-Paman-Nyungan language families (Marra, Alawa, Warndarrang, Ngandi, Ngalakgan, Nunggubuyu, and Mangarrayi) are the plausible substrate languages originally spoken at Roper River Mission (Harris 1986: 230233). All of these are unrelated (or at least not demonstrably related) to the Mirndi family which includes Jaminjung. One therefore has to distinguish substrate influence during the alleged process of creolization, and substrate influence in other regions of northern Australia to which the new language subsequently spread. Strictly speaking, it is only in the latter sense that Jaminjung can be counted as one of the substrate languages. However, due to the areal convergence already mentioned, Jaminjung shares many of the features of the above languages. Of the features associated with substrate influence
Jaminjung and Kriol
367
on Kriol by Munro (2005: Ch. 5), these are e.g. a minimal/augmented pronominal system, a distinction between punctual–perfective and continuous– imperfective in the past tense, a modal character of the “future” marker and a system of locative, possessive and instrumental cases. While Kriol as spoken in Roper River is the best-described variety, the existence of regional varieties is described by authors such as Hudson (1983), Sandefur and Harris (1986) and Rhydwen (1993, 1996). Munro (2000) arrives at the conclusion that the differences between them are slight and mostly limited to small differences in the phonemic inventory, in phonetics and prosody, and in the lexicon. This is confirmed by my own comparison of published descriptions and texts of Roper River Kriol (Sandefur 1979, 1991; Sandefur and Sandefur 1981, Munro 2000, 2005) with the Kriol spoken by Jaminjung speakers. However, a more detailed study of the use of grammatical constructions and a comparison with the traditional languages spoken in the region might well reveal more differences between the varieties than the fairly superficial comparisons undertaken so far. As already mentioned, Kriol is now the dominant language for most Jaminjung speakers, as it is used not only as a lingua franca between members of different language groups, but also as the in-group language of cross-generation communication. Moreover, a growing number of first-language speakers are monolingual, or bilingual in Kriol and English. All of the remaining Jaminjung speakers are also bilingual in Kriol, but also usually in at least one other indigenous language of the region. Even when the traditional languages are spoken among members of the older generations, code-switching with Kriol is very common, and has been for some time (cf. McConvell 1988). None of the older speakers is literate in either Jaminjung or Kriol, since both have only ever been used as an oral medium of communication within the family and the larger community, and neither is used in the media or in education, at least in the area under consideration. In education and written communication (the role of which in daily life is, however, limited), English is used exclusively. Acrolectal Kriol or Aboriginal English are used by many people, especially younger people, to speak to outsiders, and English is generally understood, but in the area under consideration, most Kriol speakers’ active command of standard English is limited. Consequently, grammatical and lexical influence in Jaminjung is quite clearly only from Kriol, not from English. In concluding this section, it may be worth pointing out that Kriol is not a language of prestige and power, in multiple respects. While non-indigenous, English-speaking people still tend to consider Kriol a degenerate form of
368
Eva Schultze-Berndt
English, indigenous people, especially the older ones, tend to equate Kriol with English and to perceive the use of the language as a threat not only to the traditional languages, but also to their identity (see e.g. Schmidt 1990: 113). This has not stopped the considerable spread of Kriol, which now has several thousands of speakers, as it does, paradoxically, fulfil a function as a symbol of indigineous identity in contrast to the dominant language, English, especially among younger people (Schmidt 1990: 111115, Munro 2000). In the remaining sections of this chapter, I will discuss the extent of Kriol borrowings into Jaminjung in the domains of phonology (Section 3), typological features (Section 4), nominal structures (Section 5), verbal structures (Section 6), other parts of speech (in particular function words, Section 7), constituent structure and other syntactic patterns (Section 8) and the lexicon (Section 9). The findings are summarized in Section (10).
3. Phonology While the phoneme inventories and phonotactic constraints of Kriol and Jaminjung are very similar, this is quite certainly not due to borrowing but to substrate influence. For example, like most Aboriginal languages of the area, the basilectal variety of Kriol lacks fricatives. Thus, lexemes derived from English lexemes with the phoneme /s/ in Kriol have an alveo-palatal stop (e.g. jup < Engl. soup), in turn non-existent in English. (Speakers of Kriol may well adopt an acrolectal register, including a phonological system which is more similar to English, when interacting with English speakers). Similar observations can be made for prosody, although there has not been much research on the prosody of either Kriol or the traditional Victoria River languages.
4. Typological features Jaminjung and Kriol have quite distinct typological characteristics, and again, no contact influence from Kriol to Jaminjung seems to have taken place as far as these are concerned (but see below on ergativity). As a Creole, Kriol is mostly isolating, with little derivational morphology and, arguably, no inflectional morphology. TAM categories are marked periphrastically, by auxiliaries such as the past perfective marker bin. The basic constituent order is SVO (see also Section 8), and core arguments in transitive clauses are distinguished by word order only, following a nominative–accusative system. Subjects are
Jaminjung and Kriol
369
always overtly expressed, at least by a pronoun. Oblique arguments and adjuncts are marked by prepositions (e.g. the comitative preposition gotim and the locative preposition la). Example (1), which was offered by a speaker as a translation of (2) below, illustrates these basic characteristics. (1)
thei bin oldei faind-im, thet.. goana, (...) 3pl aux.pst habit find-tr dem goanna ‘They used to find the goanna (the dogs did).’
In contrast, as illustrated in (2), the basic morphological type of Jaminjung is agglutinating to fusional, and the language has a rich system of verbal inflections. Bound pronominals and mood markers are prefixed to the verb stem, whereas tense–aspect marking is achieved by suffixation or stem suppletion. The constituent order, on the sentence level at least, is determined by pragmatic considerations only; compare the two lines of (2). Arguments and adjuncts take case markers which probably have to be analysed as enclitics rather than suffixes. The alignment type is ergative; the absolutive case is unmarked (and is left unglossed in the examples). (2)
burra-ngayi-rna yirrag wirib-di jarl, malajagu (...) 3pl:3sg-see-impf 1pl.excl.obl dog-erg track.down goanna jarl burra-ngayi-na yirrag, malajagu. track.down 3pl:3sg-see-impf 1pl.excl.obl goanna ‘They used to track them down for us, the dogs, the goannas (...) they tracked the goannas down for us.’
Like in a number of other Australian languages, ergative marking is “optional” in that around 30 percent of transitive agents are unmarked. The possibility that “optionality” of ergative marking is a result of contact, as suggested for other contact situations e.g. by Schmidt (1985), Bavin and Shopen (1985) and Meakins and O’Shannessy (to appear), cannot be excluded with certainty, for lack of historical data. On the other hand absence of ergative marking has a semantic and pragmatic effect similar to that described by McGregor (1992, 1998) for Warrwa and Gooniyandi, here considered as an internal feature of the languages. A further notable feature of Jaminjung and surrounding languages, which distinguishes them from Kriol, is a part of speech system where inflecting verbs such as -ngayi ‘see’ in (2) form a closed class, but can be combined with members of an open class of “uninflecting verbs”, of which jarl in (2) is
370
Eva Schultze-Berndt
an example. This phenomenon will be discussed in more detail in the context of integration of loan-verbs in Section 6.2.
5. Nominal structures No influence from Kriol has been discerned in the word order within Jaminjung noun phrases, which is relatively free. Nominal inflectional morphology is absent from Kriol anyway, and derivational morphology such as the nominalizing suffix -bala, has not been borrowed into Jaminjung. Very rarely, Kriol prepositions appear to be “doubling” a Jaminjung case marker; this is a phenomenon in need of further investigation. With the exception of ergativity (see Section 4), the semantics of Kriol cases (expressed by prepositions) and Jaminjung cases (expressed by suffixes/enclitics) is very similar, but this is quite clearly a substrate phenomenon and not due to borrowing. One example is the use of a general locative case in Jaminjung, shown in (3), which translates as the general locative preposition la in Kriol, illustrated in (4). Unlike English spatial prepositions, both Jaminjung -gi and Kriol la are unspecific about the relationship, e.g. containment or attachment, between located object and ground object. (3)
durlurl-ma ga-yu, jalbud-gi. push-cont 3sg-be.prs house-loc ‘He/she is making a noise in the house.’
(4)
(...) wan thei meik nois la haus rel 3pl make noise loc house ‘when they make a noise in the house’
6. Verbal structures and the integration of verbal loans 6.1. Verbal structures What was said in Section 5 for nominal structures also holds for verbal structures. No grammatical morphemes associated with the verb in Kriol are borrowed into Jaminjung. Structurally, too, the complex predicate system of Jaminjung is clearly different from the auxiliary-verb system of Kriol. Semantically, there are many similarities between the categories expressed by
Jaminjung and Kriol
371
function words in Kriol and by affixes or stem suppletion in Jaminjung, but this can be put down to substrate influence, not borrowing. An example is the expression of habituality by an imperfective suffix in Jaminjung in (1) and by the function word oldei in Kriol in (2). The examples in (5) and (6) illustrate the past potential category, formed by a combination of the past imperfective (here expressed by stem modification) and potential prefix in Jaminjung, and by a combination of past auxiliary bin and potential auxiliary wanna (< Engl. want to) in the Kriol translation. (5)
jurruny-ni yirr burru-bila. hand-erg/instr pull 3pl:3sg-pot:get/handle.impf ‘They were going to pull it with their hands (a cart).’
(6)
wal thei bin wanna pul-im intit. well they aux.pst pot pull-tr tag ‘Well, they were going to pull it, weren’t they?
6.2. The integration of verbal loans Jaminjung speakers not only borrow Kriol nominals, but also verbs quite extensively (see Section 9). Verb integration exhibits some interesting features due to the nature of the verbal system of Jaminjung, which (like that of surrounding languages, see e.g. Dixon 2001, McGregor 2002, Schultze-Berndt 2003) relies on the combination of two distinct parts of speech. Inflecting or generic verbs inflect for person and tense–aspect–mood (see also Section 4), but form a closed class of around 35 members. Most concepts for actions and states are expressed by members of a distinct part of speech (here called “uninflecting verb”, but also termed “preverb” or “coverb” in the literature), which do not inflect and form an open class. Lexical items which function as verbs in Kriol are integrated into Jaminjung as uninflecting verbs, and, just like their native equivalents, form complex verbs in combination with a semantically appropriate inflecting verb. This therefore functions as a verbal classifier (see Schultze-Berndt 2000 and McGregor 2002 for detailed arguments) at the same time as functioning as integrator, and as an indicator of verbness itself. For example, concepts of physical or psychological manipulation involve the transitive verb -angu, glossed as ‘get/handle’ in (7) (compare (13) and (23)), and motion concepts involve one of the inflecting verbs of motion, the most common one being -ijga ‘go’ in (8) (compare (24)).
372
Eva Schultze-Berndt
(7)
jalig-di.. lukabta-im bun-ngangu ngidbud-gi \ child-erg look.after-tr 3pl:1sg-get/handle.pst night-loc ‘The children looked after me at night.’
(8)
janyungbari-bina yagbali-bina, shift-im yirr-ijga-ny \ another-all place-all shift-tr 1pl.excl-go-pst ‘To another place, we moved over.’
The inflecting verb also expresses valency (since the paradigm of pronominal affixes distinguishes transitive from intransitive verbs); for example, the uninflecting verb shiftim shown in (8) can also combine with the transitive verb -arra ‘put, cause to be in a location’, to express the meaning of ‘shift/move something over (tr)’. Formally, this system can be described as periphrastic marking by a combination of a loan-verb stem with an additional marker. The form that is borrowed is an uninflected stem (but could hardly be anything else since there are no verbal inflections in Kriol). A small number of derivational suffixes (the transitivity marker -im and the progressive/continuous suffix -(a)bat) are borrowed with the stem but are not productive in the recipient language. The strategy of integrating Kriol verbs described here is extremely frequent and productive.
7. Other parts of speech The borrowing of “matter” from Kriol into Jaminjung affects lexical items as well as some function words; bound grammatical morphemes are not borrowed, and neither are pronouns, adpositions and auxiliaries. (It should be pointed out that the borrowing of auxiliaries has been ruled out by definition here, since utterances containing a Kriol auxiliary in combination with Jaminjung lexical items were regarded as essentially Kriol with switching to Jaminjung, according to the criteria set out in Section 1.) For occurrences of Kriol pronouns and adpositions, there are examples where the boundary between borrowing and code-switching is perhaps less clear – for example, I have found a few examples of isolated Kriol adpositions in a sentence that was otherwise Jaminjung – but these are very infrequent in the data. The remainder of this section is devoted to a discussion of borrowed grammatical morphemes, including particles functioning as connectors (7.1.), subordinating conjunctions (7.2.), focus particles (7.3.), negators (7.4.), and discourse-structuring devices (7.5.), discussed in turn below. All lexical bor-
Jaminjung and Kriol
373
rowings are discussed in Section 9, including the equivalents of English adverbs and numerals, since both in Kriol and in Jaminjung these belong to the nominal category and not to any minor part of speech.
7.1. Connectors The borrowed additive particle en (< Engl. and) and the disjunctive particle o (< Engl. or) may be used both as clause and NP connectors, illustrated only for en in (9) and (10). Since Jaminjung does not have additive or disjunctive connectors, these borrowings appear to fill a structural gap. The traditional strategy of juxtaposition continues to be used frequently though, both in Jaminjung and in Kriol. (9) yawayi, buji jalig birang thanyung warrng, yes cond child behind another walk en ngiyi jungulug jwinging ga-yu, yuno, hai. and here one swinging 3sg-be.prs you.know high ‘Yes, if another child is walking behind, and here one is swinging, you know, high, (then the one might get hit by the swing).’ (10) yurrg-bayan burra-rra-nyi yirrag warlarladbari-ni, talk-cont 3pl:3sg-put-impf 1pl.excl.obl rdp:old.man-erg en mululurru-ni \ and rdp:old.woman-erg ‘They were talking to us, the old men and old women.’ The Kriol borrowing ani (< Engl. only) is frequently used as a contrastive connector, i.e. as a translation equivalent of English but, as in (11). In this case, however, the borrowing does not fill a structural gap, but replaces a corresponding Jaminjung contrastive particle yiga, illustrated in (12). In some instances, the particle bugu ‘just, only’ is also used in a contrastive function, as in (13), in addition to its more common restrictive function. The semantic extension from a restrictive particle ‘only’ to a contrastive connector is thus likely to be due to substrate influence. (11) jarlwab nga-mama-na ani jarlig guyawud burr-agba:, ngilijja:, save 1sg:3sg-have-impf only child hungry 3pl-be.pst crying ‘I was going to save it (the food) but the children were hungry, (and) crying.’
374
Eva Schultze-Berndt
(12) majani yawurr-irdbaj! yiga langa-marnany jubard \ maybe irr:3pl-fall but ear-priv shut ‘(We tell them) “(…) you all might fall!” – but they don’t listen, (as if) deaf.’ (13) yathang gujang ngarrgina Nangari, right mother 1sg:poss subsection bugu=biya bulgarding, Jimij-nyunga just=now father subsection-source ‘All right, my mother is a Nangari, only/but the father is (the child of someone) from the Jimij subsection.’
7.2. Subordinating conjunctions No subordinating conjunctions exist in Jaminjung for finite clauses except for the general subordinator =ma which is a second position clitic, and the postverbal conditional clitic =wunthu ‘if’. The latter is sometimes, but rarely, replaced with the Kriol equivalent buji (< English suppose) in clause-initial position, as shown in (9) above. Other types of subordination, e.g. causal and purposive clauses, involve case marking on uninflecting verbs in a non-finite clause. The only borrowed conjunction with a high frequency is the Kriol causal conjunction dumaji (< Engl. ‘too much’), illustrated in (14). This is again a case of a loan apparently filling a structural gap. (14) mali rolimap yurra-ngu:, damarlung, cloth roll:tr:up 1pl.incl:3sg-get/handle.pst nothing gumard marring dumaji \ road bad because ‘We rolled up our blankets (in order to go on a camping trip), (but) nothing (= we didn’t go after all), since the road was bad.’
7.3. Focus particles and phasal particles The term “focus particle” is used here in the sense of e.g. König (1991) to refer to particles with an additive, restrictive etc. function which are associated with a focal constituent. An additive function (‘also, too’) is fulfilled by
Jaminjung and Kriol
375
the two Kriol borrowings in Jaminjung, tu, used much like its English source too, and igen (< Engl. again), illustrated in (15). These also have a Jaminjung equivalent, the clitic =gayi ‘also, too’. (15) janyungbari ga-yu=ngardi \ janyungbari igen girrgirrlang \ another 3sg-be.prs=sfoc another too galah ga-ruma-ny, Lijuna-ngunyi igen \ 3sg-come-pst place.name-abl too ‘There is another one, another (mythological) galah as well, he came, (him) too from Legune.’ (from a myth) Kriol ani (<Engl. only) is used as a restrictive particle alongside the Jaminjung equivalents bugu ‘just, only’ and =biji ‘only’, but, as shown in Section 7.1, also as a contrastive connector. (16) ngayin-marnany, ani darrmad meat-priv only freshwater.crocodile (Context: “bring us kangaroo meat!”) – “(There is) no (proper) meat, only freshwater crocodile (meat)!” A borrowed particle/clitic that comes closest to a pure focus marker is the sequential discourse marker =na (< Engl. now). Its two, related, functions are to mark a new (hence focal) event in a sequence, as in (17), and a shift in topic (i.e. a topic sequence), as in (18) (see also Graber 1987). This does not fill a structural gap, but rather appears in the same positions and with the same functions as the Jaminjung clitic =biya(ng) (compare its use in (26), (29), and (30)), which points again to substrate influence. (17) burru-yu=na\ gurrany burr-angga bunug-bunug=na \ 3pl-be.prs=now neg 3pl-go.prs rdp-steal=now julany-nyunga, smoke-source (children, after being punished for stealing by “smoking” them) ‘They stay (in one place) then, they don’t go stealing then, because of that smoke.’ (18) nami=na, guny-b-ijga \ Bulla-bina 2sg=now 2du-pot-go place.name-all ‘You, you two will go, to Bulla.’
376
Eva Schultze-Berndt
As a phasal particle, Kriol yet (< Engl. yet) is frequently used as a negative polarity item, as in English (19), but also in the meaning of ‘still’ (20). For both functions, there does not seem to be a Jaminjung equivalent except for the much more strongly grammaticalized restrictive clitic =(w)ung (SchultzeBerndt 2002), also shown in (20). (19) gurrany yirrgbi-nyunga yet. neg talking-source yet ‘(We have) not talked yet.’ (i.e. the speaker is aware that the linguist was hoping to do some recording) (20) mani ga-yu=wung yet. money 3sg-be.prs=restr yet ‘Money is still there.’
7.4. Negation The Kriol particles indicating a negative answer, na:(wu) or nomo ‘no’ are used frequently in Jaminjung utterances. It is unclear whether the ‘yes’ particle, yawayi, common to several languages of the area, arose from contact influence. (21) na: jurlag ga-gba=ni the mugmug walthub jarriny-gi, no bird 3sg-be.prs=sfoc there owl inside hole-loc (Context: The boy thought “maybe the frog is here in the hole.”) ‘No, a bird was there, an owl inside the hole.’ (from a Frog Story) In contrast, the Kriol negation particles neba/neva (< Engl. never), employed for sentence negation mainly in the past indicative, and nomo (< Engl. no more), employed for sentence negation mainly in the nonpast and in nonindicative moods, and for constituent negation, are never used in Jaminjung utterances in my data, with one crucial exception: there are a number of examples, including (22), where the particle nomo is used for metalinguistic negation instead of the Jaminjung general negative particle gurrany. (22) wirr gani-ma-m jarriny-ngunyi, nomo dibard \ move.out 3sg:3sg-hit-prs hole-abl neg jump ‘It comes out of the hole, not “it jumps”.’ (about a frog, correcting ESB’s suggestion of dibard ganiyu ‘it jumped out’)
Jaminjung and Kriol
377
Jaminjung and Ngaliwurru also have the special independent negative forms damarlung (J) and gara (Ng). They are used on their own as a negative answer to a request, but may also follow negated statements for emphasis, or follow positive sentences to indicate that the action described did not lead to the desired result. The latter function is illustrated in (14) for damarlung, and in (23) for its Kriol equivalent, najing/nathing (< Engl. nothing). Despite the existence of the Jaminjung forms, najing is employed so frequently that it can justifiably be regarded as a borrowing. (23) wardany-ni=ma skrejimbat yirra-ngu, hand-erg=subord scratch:tr:cont 1pl.excl:3sg-get/handle.pst nathing \ nothing ‘We were scratching (i.e. digging) with our hands, (but) to no avail.’ (looking for yam) The frequency of use of najing is in striking contrast to the absence of the Kriol sentence negator in Jaminjung clauses, mentioned above. The answer may be that najing, in addition to expressing the concept of a negative or insuccessful action, also functions as a discourse structuring device, since like its Jaminjung equivalents it usually occurs in an intonation unit of its own. Further borrowings of this nature are discussed in the next subsection.
7.5. Discourse-structuring devices One of the most striking uses of Kriol borrowings is as discourse-structuring and paragraph-marking devices – even though in most cases, Jaminjung equivalents exist. Subsumed under this heading are tags, fillers, and paragraph-boundary markers. The Kriol tag (y)intit (< Engl. isn’t it) is ubiquitous in everyday speech although it also has equivalents in Jaminjung (ngi’) and Ngaliwurru (ngali). (Y)intit is used irrespectively of the polarity or the verb of the utterance, or the person and number of the subject; an example is (6) in Section 6.1. A further tag yuno (< Engl. you know ), illustrated in (9) occurs much less frequently. The marking of an end of a paragraph or a transition frequently involves a switch to Kriol. For example, the Kriol particles binij (<Engl. finish), which marks the end and culmination of an episode, and thetsol (< Engl. that’s all),
378
Eva Schultze-Berndt
which has a more metalinguistic function in indicating a switch in overall topic, are used very frequently, even though an equivalent (yathang ‘finished, enough, all right’) does exist in Jaminjung. The use of binij is illustrated in (24). (24) aligeita=biya mit-im=ung ga-ruma-ny=nu, alligator=now meet-tr=restr 3sg-come-pst=3sg.obl jum gan-uga \ binij \ grab 3sg:3sg-take.pst finish ‘the alligator came to meet it (a kangaroo), and grabbed it, finished’ Another particle, olrait (< Engl. all right), marks a transition between two clauses which have a stronger thematic coherence, e.g. where the continuation is a natural consequence of the beginning, as in (25). (25) bard-bard nganth-arra-m larriny-ni, rdp-cover 2sg:3sg-put-prs paperbark-erg/inst olrait, mirrbba ga-ngga gunjarlg-di. all.right buried 3sg-go.prs ground-erg/inst ‘You cover it up with paperbark, allright (i.e. when that’s done), it is covered with earth.’ (meat to be roasted) Another Kriol particle used to establish cohesion is wal, used in a similar manner to its source in the lexifier language, well, but without clear equivalent in Jaminjung. The following example is from a myth about Emu and Brolga, where Emu incites Brolga to kill all her children but two, falsely pretending that she herself only has two children. The particle wel here indicates a shift in local topic from Emu to Brolga. (26) ngayug=guji jirrama ngawuny-nganja, jalig, jirrama, 1sg=first two 3sg:3du-take.prs child two gani-yu=nu, gumurrinji-ni \ 3sg: 3sg-say/do.pst=3sg.obl emu-erg wal gudarrg-di=biya jalig burrb ganurru-mangu niwina \ well brolga-erg=now child finish 3sg:3pl-hit.pst 3sg.poss ‘I already have (lit. ‘take’) (only) two children, two, she said to her, the emu did. Well, the brolga then killed all her children (except for two).’
Jaminjung and Kriol
379
To sum up this section: except in a few cases, borrowed function words were shown not to fill structural gaps, but to correspond fairly closely to existing equivalents in Jaminjung, much more so in any case than to the source item in the lexifier, English. This points to a scenario of substrate influence in terms of “pattern” on the Kriol forms, and subsequent borrowing of “matter” to Jaminjung, partly replacing the equivalents. The question about the motivation for such borrowings will be adressed in the concluding section (Section 10).
8. Constituent order and other syntactic patterns Information structure and its influence on constituent order has scarcely been studied for either Kriol or Jaminjung and other traditional languages of the area. On the surface, Jaminjung has pragmatically conditioned word order, while Kriol has the fixed basic constituent order SVO. However Kriol allows for certain alternative word orders which appear to show substrate influence. To give just one example, both Jaminjung and Kriol allow for discontinuous or “split” noun phrases. One of the contexts in which these occur – in both languages – is relatively easily identified. This is in an annuntiative thetic sentence (Sasse 1987, 2006), i.e. an utterance announcing the presence or appearance of some entity or situation “out of the blue”. Discontinuous noun phrases – with one component appearing before and the other after the verb – are used to emphasize a quality of the entity whose existence or appearance is announced. The Jaminjung and Kriol examples in (27) and (28) are translation equivalents, produced on the same occasion by the same speaker. (27) warrgad ga-ram=yirrag mayi. long 3sg-come.prs=1pl.excl.obl man ‘A tall man is coming for us.’ (28) longbala kaming thet men theya. long/tall coming that man there ‘A tall man is coming there.’ No influence of Kriol constituent order on Jaminjung could be discerned in terms of consituent order, or for that matter on any other syntactic patterns except where a pattern is created by the borrowing of one of the function words
380
Eva Schultze-Berndt
for which no Jaminjung equivalent exists (Section 7). Another potential area of influence, mentioned in Section 4, is optional ergativity. The overall impression is that similarities of pattern between the two languages are due to substrate influence, not borrowing.
9. Lexical borrowings As already pointed out, and illustrated in many of the examples, Kriol lexical items are frequently borrowed into Jaminjung, although the boundary between borrowing and single-item or insertional code-switching is difficult to draw. I am not aware of calques in Jaminjung on the basis of Kriol expressions. Borrowed individual lexemes include verbs as well as nominals; the latter category also includes items that are numerals, adjectives, or spatial and temporal adverbs in English.
9.1. Nominals It comes as no surprise in a situation of relatively recent contact between two cultures with a very different material basis that nominals borrowed from Kriol into Jaminjung include many names for introduced animals, plants, tools and other artefacts, such as buliki ‘cow, cattle’, barrigi (< English paddock) ‘fence’, kroba ‘crowbar’, jangayi ‘shanghai; sling shot’, teip ‘tape recorder’, and eroplein ‘aeroplane’. On the other hand, many Jaminjung coinages or semantic extensions of existing words are in use for introduced items alongside with Kriol words – e.g. diwu-ngarna ‘plane’, lit. ‘fly-assoc’ = ‘thing for flying’, jalwany-ngarna ‘tape recorder’, lit. ‘talk-assoc’ = ‘thing for talking’, wagurra ‘rock, money’ (Kriol mani), bagarli ‘paperbark; paper, paper money’ (Kriol peipa) or jingil ‘plant juice, meat juice, soup; petrol/ diesel’ (Kriol petrol). There are also some clear instances of borrowing of temporal nominals (including those originating in English adverbs). Days of the week are generally borrowed (including expressions such as penjin dei ‘the day of the pension payments’, and peidei ‘pay day’). Kriol expressions are used for times of the day but alongside Jaminjung equivalents. The most frequent time-of-day expression from Kriol is alibala ‘early, in the morning, the next day’ (from English early + fellow).
Jaminjung and Kriol
381
(29) alibala=biyang gud buny-agba, morning=now rise 3du-be.pst ‘In the morning then the two got up.’ Spatial nominals including deictics, on the other hand, are used but mostly in those places identified as switches rather than borrowings, with a few possible exceptions such as (21) and (30). (30) bily=biya ga-gba=nu tharrei=guji \ burst=now 3sg-be.pst=3sg.obl there=first ‘(The ulcer) burst on him only there (in hospital).’ Since Jaminjung only has numerals corresponding to ‘one’, ‘two’ and ‘three’, it comes as no surprise that numerals are borrowed from Kriol. Interestingly, however, ordinal numerals, which are completely absent from Jaminjung, do not seem to be borrowed, in other words, ordinal numerals are not used at all. Other Kriol quantifiers are used occasionally, despite the existence of Jaminjung equivalents, but their low frequency points to code-switching rather than borrowing. One example is (31). (31) janju binka-ni wuju.. plenti nga-manggu gagawuli, that river-loc small plenty 1sg:3sg-hit.pst long.yam ‘At that small creek I got lots of long yam.’
9.2. Verbs The borrowing of Kriol verbs into Jaminjung as uninflecting verbs was already discussed in Section 6.2 with respect to their integration. Borrowed items include not only verbs for trade and money exchange such as bayim ‘buy:tr’, for activities related to Western technology such as kikap ‘kick start (of motor)’ or to handling of cattle such as masterim ‘muster:tr’, but (as shown e.g. in (7), (8), (23) and (24)) also verbs denoting traditional everyday activities for which Jaminjung equivalents exist and continue to be used.
10. Summary and discussion This chapter has provided an overview of lexical and in particular grammatical borrowings from Kriol, an English-lexified Creole spoken in northern
382
Eva Schultze-Berndt
Australia, to Jaminjung, a Non-Pama-Nyungan language of northern Australia. As demonstrated in Section 2, this situation is one of intense contact and strong functional pressure from the dominant language, Kriol although this is not a language associated with power and prestige. The contact situation with Kriol is special in another way, as language shift to Kriol set in not long after the Creole itself came into being. It was argued here that Kriol shows many influences in “pattern” – themselves areal features – from the substrate languages, e.g. in its phonology, the categories expressed by grammatical markers, and even constituent order; these were therefore not considered as borrowing. The functional pressure from Kriol has led to a fairly high frequency of borrowings of “matter”, both of lexical and of grammatical morphemes. As far as the lexical level is concerned, it was shown that both verbs and nominals (including adjectival and adverbial nominals) are borrowed. Borrowing is not restricted to terms for introduced items, but on the other hand core vocabulary such as kinship terms are not borrowed. Verb integration relies on a pre-existing system of classificatory inflecting verbs which are semantically generic and form a closed class; Kriol verbs are integrated as uninflected verbs. On the grammatical level, borrowing of “matter” is restricted to free or clitic forms, including connectors, subordinating particles, particles associated with focus, and discourse-structuring particles. A sentence negation marker from Kriol was shown to be restricted to metalinguistic negation. Other free forms, notably auxiliaries and adpositions, are not borrowed, at least according to the criteria set out in Section 1. Bound grammatical forms are not borrowed at all, not surprisingly in part since Kriol does not have inflectional affixes. However, it does have derivational affixes which are not borrowed, i.e. never occur in Jaminjung utterances outside Kriol lexical forms. The findings thus confirm hierarchies of borrowability as previously suggested in the literature (e.g. Weinreich 1968, Heath 1978, Thomason and Kaufman 1988). More specifically, they confirm the high borrowability of grammatical morphemes with discourse-regulating functions (Stolz and Stolz 1996, Matras 1998). This is all the more striking as only few cases of these borrowings – some sentence connectors and the phase particle yet – fill true “structural gaps” in the recipient language. In many other cases, Kriol forms, whose functions are very different from their English lexical sources and thus likely to be modelled on the indigenous substrate languages, take the place of existing Jaminjung equivalents. If the model proposed by Matras (1998) is valid – he argues that the use of discourse-regulating markers from the dom-
Jaminjung and Kriol
383
inant language results from “fusion” of the respective grammatical systems in this domain in the bilingual due to cognitive pressure – one may well question whether it makes much sense to draw a strict boundary between codeswitching and borrowing in a situation like the one described here where all Jaminjung speakers are bilingual in the dominant language, Kriol.
Abbreviations Apart from the abbreviations used in interlinear glosses listed below, the symbol \ is used to indicate a final (falling) intonation contour, and a comma to indicate a nonfinal intonation contour, at a prosodic break. 1, 2, 3 1st, 2nd, 3rd person abl ablative case (starting point of motion) all allative case assoc associative derivational suffix (“associated with X”) aux auxiliary cond conditional marker cont continuous activity; derivational marker on uninflecting verbs contr contrastive focus marker dir directional suffix on locative nominals (outward direction) du dual erg/inst ergative/instrumental case excl exclusive (pronominal category) impf imperfective aspect incl inclusive (pronominal category) irr irrealis mood loc locative case neg negative/negation obl oblique pronominal pl plural poss possessor/possessive pot potential mood priv privative marker, “without” prs present tense pst past perfective tense/aspect rdp reduplication restr restrictive marker sfoc sentence focus marker sg singular (pronominal category) source source, origin (case)
384
Eva Schultze-Berndt
subord tag tr
subordinator (non-specific) tag question transitive marker (in Kriol)
References Bavin, Edith, and Timothy Shopen 1985 Warlpiri and English: Language in contact. In: Michael G. Clyne (ed.), Australia, Meeting Place of Languages, 8194. Canberra: Pacific Linguistics. Dixon, R. M. W. 2001 The Australian linguistic area. In: Alexandra Y. Aikhenvald and R. M. W. Dixon (eds.), Areal Diffusion and Genetic Inheritance: Problems in Comparative Linguistics, 64104. Oxford: Oxford University Press. Graber, Philip 1987 The Kriol particle ‘na’. Working Papers in Language and Linguistics 21: 121. Harris, John W. 1986 Northern Territory Pidgins and the Origin of Kriol. Canberra: Pacific Linguistics C89. Heath, Jeffrey 1978 Linguistic Diffusion in Arnhem Land. Canberra: Australian Institute of Aboriginal Studies. Hudson, Joyce 1983 Grammatical and Semantic Aspects of Fitzroy Valley Kriol. Darwin: Summer Institute of Linguistics. Johanson, Lars 2002 Structural Factors in Turkish Language Contacts. Richmond: Curzon. König, Ekkehard 1991 The Meaning of Focus Particles: A Comparative Perspective. London: Routledge. Matras, Yaron 1998 Utterance modifiers and universals of grammatical borrowing. Linguistics 36 (2): 281331. McConvell, Patrick 1988 Mix-im-up: Aboriginal codeswitching old and new. In: Monica Heller (ed.), Codeswitching: Anthropological and Sociolinguistic Perspectives, 97124. Berlin: Mouton de Gruyter.
Jaminjung and Kriol
385
McGregor, William B. 1992. The semantics of ergative marking in Gooniyandi. Linguistics 30: 275318. 1998 Optional ergative marking in Gooniyandi revisited: Implications to the theory of marking. Leuvense Bijdragen 87 (43): 491534. 2002 Verb Classification in Australian Languages. Berlin: Mouton de Gruyter. Meakins, Felicity, and Carmel O’Shannessy Forthc. Ordering arguments about: Word order and discourse motivations in the development and use of the ergative marker in two Australian mixed languages. Paper submitted. Munro, Jennifer M. 2000 Kriol on the move. A case of language spread and shift in Northern Australia. In: Jeff Siegel (ed.), Processes of Language Contact: Studies from Australia and the South Pacific, 245270. Saint-Laurent, Quebec: Collection Champs linguistiques. 2005 Substrate language influence in Kriol: The application of transfer constraints to language contact in northern Australia. Armidale: Ph.D. diss., Department of Linguistics, University of New England, Armidale. Rhydwen, Mari 1993 Kriol: The creation of a written language and a tool of colonisation. In: Michael Walsh and Colin Yallop (eds.), Language and Culture in Aboriginal Australia, 155168. Canberra: Aboriginal Stud`ies Press. 1996 Writing on the Backs of Blacks: Voice, Literacy and Community in Kriol Fieldwork. Brisbane: University of Queensland Press. Rose, Deborah 1991 Hidden Histories. Black Stories from Victoria River Downs, Humbert River, and Wave Hill stations, North Australia. Canberra: Aboriginal Studies Press. Sandefur, John 1979 An Australian Creole in the Northern Territory: A description of Ngukurr-Bamyili dialects. Work papers of SIL–AAB, Series B, Volume 3. Darwin: Summer Institute of Linguistics. 1991 A sketch of the structure of Kriol. In: Susanne Romaine (ed.), Language in Australia, 204212. Cambridge: Cambridge University Press. Sandefur, John, and John Harris 1986 Variation in Australian Kriol. In: Joshua Fishman (ed.), The Fergusonian impact, 180190. Berlin: Mouton de Gruyter. Sandefur, John, and Joy Sandefur 1981 Introduction to conversational Kriol. Work papers of SIL–AAB, Series B, Volume 5. Darwin: Summer Institute of Linguistics.
386
Eva Schultze-Berndt
Sasse, Hans-Jürgen 1987 The thetic/categorical distinction revisited. Linguistics 25: 511580. 2006 Theticity. In: Giuliano Bernini and Marcia L. Schwarz (eds.), Pragmatic Organization of Discourse in the Languages of Europe, 255– 308. Berlin: Mouton de Gruyter. Schmidt, Annette 1985 The fate of ergativity in dying Dyirbal. Language 61: 378396. 1990 The Loss of Australia’s Aboriginal Language Heritage. Canberra: Aboriginal Studies Press. Schultze-Berndt, Eva 2000 Simple and complex verbs in Jaminjung. A study of event categorisation in an Australian language. Ph.D. diss., University of Nijmegen. 2002 Grammaticalized restrictives on adverbials and secondary predicates: Evidence from Australian languages. Australian Journal of Linguistics 22 (2): 231264. 2003 Preverbs as an open word class in Northern Australian languages: Synchronic and diachronic correlates. In: G. Booij and J. van Marle (eds.), Yearbook of Morphology 2003, 145177. Dordrecht: Kluwer. Stolz, Christel, and Thomas Stolz 1996 Funktionswortentlehnung in Mesoamerika. Spanisch-amerindischer Sprachkontakt (Hispanoindiana II). Sprachtypologie und Universalienforschung 49 (1): 79123. Thomason, Sarah G., and Terrence Kaufman 1988 Language Contact, Creolization, and Genetic Linguistics. Berkeley: University of California Press. Troy, Jakelin 1993 Language contact in Early Colonial New South Wales 1788 to 1791. In: Michael Walsh and Colin Yallop (eds.), Language and Culture in Aboriginal Australia, 3350. Canberra: Aboriginal Studies Press. Tryon, Darrell T., and Jean-Michel Charpentier 2004 Pacific Pidgins and Creoles: Origins, Growth and Development. Berlin: Mouton de Gruyter. Weinreich, Uriel 1968 Languages in Contact: Findings and Problems. The Hague: Mouton.
Grammatical borrowing in Rapanui Steven Roger Fischer
1. Background Rapanui is the language spoken by about one-fourth of the descendants of the original East Polynesian settlers of Easter Island in the south-eastern Pacific. In its phonology, morphology, syntax and lexicon, Rapanui is a typical Polynesian language. That is, vowels prevail over a limited consonantal inventory; only open syllables are allowed; particles and lexemes constitute the morphological system; verbal and nominal frames shape all syntactic units; and nearly the entire lexicon shares cognates within related lexica of other East Polynesian islands. Rapanui is the sole indigenous language of Easter Island; there are no dialects. (The island has only the single community of Hanga Roa.) Rapanui was also the island’s only language until foreign settlement in 1864/1866. Until the 1930s, when Chilean Spanish began to intrude more decisively, nearly every indigenous Easter Islander spoke only Rapanui. (Easter Island was annexed by Chile in 1888.) As of 1966, with the granting of Chilean civil rights to the Rapanui people of Easter Island, Spanish became the preferred language on the island, whereupon Rapanui approached extinction. Within the past ten years in particular, however, Rapanui has made a significant comeback, although bilingual and syncretic styles of discourse Rapanui have yielded a “Spanish Rapanui” as well as a “Rapanui Spanish”, with many mixed forms. Easter Island children remain mostly monolingual Spanish speakers, although many possess passive knowledge of some Rapanui. Of the island’s resident population of c. 4,000 in 2006 – 1,800 indigenous Rapanui, 2,200 mainland Chileans and others; their command of Rapanui unknown, a further 2,200 indigenous Rapanui live abroad, mostly in Chile – perhaps around 800 (or 1,000 if one includes expats) claim Rapanui competence; only about 500 (650) of these would be fluent in Rapanui. Recent Rapanui educational and cultural programmes targeting children allow the hope that the number of Rapanui speakers might increase, if only marginally, in future.
388
Steven Roger Fischer
2. Phonology The Rapanui sound system closely resembles its main contact language, Chilean Spanish (“castellano” on the island). Phonological adaptation often occurs, however, when alien Spanish forms are transferred into Rapanui, whereupon the Rapanui speaker chooses a relevant approximation of place and manner of articulation. For example, Spanish /g/ and /x/ will usually be interpreted as Rapanui /k/; Spanish /d/ as Rapanui /r/; Spanish /s/ and /č/ as Rapanui /t/. Spanish consonant clusters will either be simplified (/s/ will regularly be omitted) or expanded through vowel insertion to create a Polynesian open syllabic structure (also often with omission and/or re-articulation): Spanish canasto ‘basket’ is Rapanui kanato, and Spanish pobre ‘poor (one)’ is Rapanui poere. There are no rigid rules in this process. Makihara (2001a: 195) points out, for example, that Spanish olvida ‘(he/she) forgets’ can be Rapanui /orvida/, /orvira/, /orovida/ or /orovira/. In Rapanui it does appear that Spanish /l/ and /d/ are replaced with /r/ more often than Spanish /g/ and /x/ are replaced with /k/. Older speakers, who learnt Spanish only imperfectly, tend to be those who replace Spanish with Rapanui consonants more regularly and frequently than younger speakers, who are more fluent in Chilean Spanish. However, in certain formal contexts even fluent Spanish speakers will pronounce borrowed Spanish elements using Rapanui phonology, this in order to effect a “nativization” for social or psychological reasons. Nearly all such alterations tend to be conscious and strategic; there has been no communitywide systemization of Spanish phonological alteration. Although all vowel sounds are shared with Spanish, Spanish /e/ is regularly pronounced as /ɛ/, Spanish /o/ as /ɔ/ in Rapanui. Most Rapanui speakers, even elderly ones, now appear to share Spanish prosody and intonation, with rare exceptions.
3. Morphological typology Ergative–absolutive distinctions characterized the early stage of Proto-Polynesian (c. 1000 bc). But already by the much later Proto-East Polynesian stage (c. ad 500) such functions had been replaced by ever more important nominative–accusative distinctions. This eventually rendered Rapanui, once distilled from South-east Polynesian, very clearly a nominative-accusative language (though it is possible that vestigial ergative elements remain in Modern Rapanui). East-Polynesian -Cia (that is, consonant + /ia/) passive desinence was lost entirely in Rapanui, but for a small number of lexical-
Rapanui
389
ized passives. Reduplication, as in ve’ave’a ‘very hot’ and ’iti’iti ‘very little, small’, follows standard inherited East Polynesian practice, whereby beyond their main function as intensifiers such reduplications can also convey, in verbs, both plural agency and repetitive action.
4. Nominal structures As with all Polynesian languages, Rapanui marks possession using alienable–inalienable distinctions in personal pronouns, demonstratives and certain prepositions that alternate between the vowels /a/ and /o/, according to formal and intuited rules of usage (Fischer 2000). With Spanish contact, however, alienable/inalienable possessives /a/ and /o/ are generalizing to /o/. For example, (1)
ta’a pōki 2sg.poss child ‘your child’
will now sometimes be heard as to’o pōki, tu’u pōki or even tū pōki (corresponding to Spanish tu ‘your’), seemingly resulting from a perceived notion that additional marking of possessiveness is no longer necessary in Rapanui, as Spanish does not have this alienable–inalienable distinction. Rapanui /o/ already commands a greater domain in possessive marking than does /a/; thus, /o/ has been chosen as the “default marker” in this weakening of the possessive marking system as a result of Spanish contact. Postpositive positioning of qualifiers (nominal and adjectival) is demonstrating a shift toward prepositive positioning in Rapanui, and this not only in direct borrowings or calques. For example, Rapanui motore vaka reproduces English ‘motorboat’; Rapanui “should” have here vaka motore as a mixed calque. (A particularly rampant intruder due to massive tourism of late, English is being widely spoken on the island now, too.) Also, Rapanui kē vece imitates Spanish algunas veces ‘sometimes’. The Rapanui demonstrative phrase te me’e nei ‘this thing’ is now more often nei me’e after Spanish esta cosa. In similar fashion, although ara nei ‘this road’ (lit. ‘road this’) can still be heard, nā hare ‘that house’ after Spanish esa casa is now preferred. In Rapanui, locative predicates now follow more closely the Spanish model, in existentials and in adjectival predicates, as in the following three examples (2: locative predicate; 3: existential; 4: adjectival predicate):
390
Steven Roger Fischer
(2)
ki rote vai kava rlt inside/art water bitter ‘to the sea’
(3)
Ai te maika ’i nei. ext art banana rlt dem ‘There are bananas here.’
(4)
Nā hare, hare ’iti’iti. dem house house small ‘That house is small.’
As with all Polynesian languages, Rapanui has a phrase-structure grammar, with no case marking. It also has no gender marking, nor has it systematically borrowed gender marking from Spanish. (Isolated items are beginning to intrude, however, such as gender-marked diminutive suffixes.) Rapanui marks the possessed within the noun phrase, but, as mentioned above, is currently reducing the /a/ and /o/ alienable–inalienable possessive distinctions to “default” /o/. This process might stem from those indigenous Easter Islanders, all of whom speak Spanish, who intuit here a Spanish de which, for all such cases of possession (except possessive pronouns), would employ a preposition instead to mark this function. That is, a native Rapanui speaker would now “feel” that there is no longer a need to make an alienable–inalienable distinction, as no such need engages the identical Spanish statement. It is for this reason, then, that /o/ is now commonly fulfilling this function as the “default” translation of Spanish de in analogous contexts. After the Spanish model, not only has Rapanui shifted the syntax of the attributive demonstrative to prepositive position (see the example nei me’e above), it now frequently uses definite/indefinite articles like Spanish as well: Rapanui te for Spanish el/la/lo and Rapanui ’etahi ‘one’ for Spanish un/una/ uno. Traditionally following the head, the Rapanui agent will now, again following the Spanish model, often precede the head. For example, an original Rapanui construction (5)
He ’aroha, he tatangi ararua. ml suffer ml weep both ‘Both suffer and weep.’
would now, in imitation of Spanish, more likely be phrased
Rapanui
(6)
391
Ararua he ’aroha, he tatangi ’ā. both ml suffer ml weep prog ‘Both suffer and weep.’
Rapanui te tangata ‘mankind (in general)’ is now being understood, after Spanish el hombre, to mean ‘the man’, with specificity and definiteness. It appears that original Rapanui te ‘the’ was, as in all Polynesian languages, more generic marker than definite article. However, under Spanish influence this is being reanalysed. The entire Rapanui system of definiteness/indefiniteness has recently been reinterpreted in deference to the Spanish perception, a fascinating, perhaps even alarming, psycholinguistic phenomenon. In the case of diminutives, Spanish -tita and -tito (and a few others, observing, interestingly enough, a gender distinction otherwise not found in Rapanui) might rarely be suffixed to some Rapanui words and names. However, this is code-switching, not borrowing, and it remains on Easter Island very infrequent and individual.
5. Verbal structures Like all Polynesian languages, but quite unlike Spanish, Rapanui has no modal verbs. “Modality” is usually achieved through simple periphrasis. For example: (7)
Ko te rivariva he oho koe atu. foc art good ml go 2sg dir ‘You should/must go.’ (lit. ‘It is good you go away.’)
As of quite recently, frequent use is being made of Spanish tiene que ‘he/she/ it has to’ in order to show obligation in Rapanui; tiene que is simply used here as a lexicalized borrowing, without relevant inflection. And puē from Spanish puede ‘he/she/it can’ is often used to express ability and possibility as well; again as a lexicalized borrowing, without relevant inflection, it is fast becoming “native” Rapanui. It is possible that Spanish ir ‘to go’ has prompted the increased use of Rapanui oho ‘go’ to indicate future, rather than oho’s most common function to indicate simple motion; but this new use of oho would certainly be individual and infrequent, too. For example, one can now hear:
392
Steven Roger Fischer
(8) He oho au, he ha’uru. ml go 1sg ml sleep ‘I’ll go to bed.’ (lit. ‘I go, sleep.’) When Spanish verbs are used in Rapanui, it usually entails direct integration without extra marking (and nearly always with some phonetic borrowing to reproduce the Spanish verb perfectly, if possible), verbness meaning and/ or infinitive–nominal distinction. (Inflection is generally ignored by Rapanui speakers, also when speaking Spanish.) When Easter Islanders use Spanish verbs while speaking Rapanui, it almost always involves code-switching of some kind, not borrowing. Because of massive Spanish contact in particularly the past 40 years, there has perhaps been an increased use of the periphrastic passive voice by Rapanui speakers. Traditionally, Rapanui dropped its inherited Polynesian -Cia passive desinence, though the passive is very common in other Polynesian languages (particularly Māori); only isolated lexical vestiges of a marked passive now remain in Rapanui. It is doubtful whether Rapanui possessed such a robust periphrastic passive voice before 1966, the year Easter Islanders achieved Chilean citizenship and full civil rights. The periphrastic passive one now hears on the island seems to be directly patterned after Spanish. For example: (9) Ku tike’a ’ā ia ’i te ūka. t/a see res 3sg rlt art girl ‘He/She/It was seen by the girl.’ (10) He kī mai ’i te tangata. ml say dir rlt art man ‘It is being said by the man.’ Since Rapanui has no verb meaning ‘to have’, Spanish tengo is often used in discourse for first-person singular, as tengo au ‘I have’. Uninflected Spanish tiene ‘he/she/it has’ is then used for most other declensions, irrespective of person or number. It is of value to note that, for this, Old Rapanui would have used such an inherited construction as: (11) He hare ’o’oku. ml house 1sg.poss ‘I have a house.’ (lit. ‘It is (There is) house of mine.’)
Rapanui
393
Old Rapanui had no copula. With increased intercourse with Tahiti at the end of the nineteenth century, Tahitian ’ē ‘and’ was borrowed. Tahitian ’ē is used to connect the subject and predicate; hence, it is a connector which is functioning as a copula: (12) ’Ē he tu’u mai ki nei he adapta au ’ē he puē mo coorc ml arrive dir rlt here ml adapt 1sg coorc ml mod ben adapta. adapt ‘And when I arrive here I adapted and could adapt’ (Makihara 2001a: 212) Although Tahitian ’ē is still favoured, which is now regarded to be “native” Rapanui, today one can often hear in Modern Rapanui discourse Spanish y as copula, which appears to be slowly replacing Tahitian ’ē: (13) He tike’a au i ’ā Juan y Leo y José y ku riri ml see 1sg acc prs Juan coorc Leo coorc José coorc t/a anger ’ā. res ‘I see Juan and Leo and José who are angry.’ In the wake of recent massive Spanish contact, the verbal semantics of fluent Rapanui speakers has sometimes been re-evaluated. For example, Spanish recibe ‘he/she/it receives’ is now Rapanui recibe (for all declensions), which describes only the physical act of receiving – as through the Chilean post. This borrowed verb does not embrace, however, the intricate Polynesian social obligations attending the communal act of receiving, which process would still demand the use of the Rapanui verb rava’a.
6. Other parts of speech In formal Rapanui speech, there are hardly any Spanish borrowings; this allows us to construe that, in Rapanui, the use of Spanish entails almost exclusively code-switching, not borrowing. With other parts of speech, numerals are almost entirely Tahitian, having replaced most of the Old Rapanui numerals already at the end of the nineteenth century; Tahitian’s decimal system was very close to Old Rapanui’s decimal system, needing only min-
394
Steven Roger Fischer
imal replacements and minor phonological alterations. In normal Rapanui discourse (not formal speech), there is frequent use of Spanish expletives or such discourse fillers as Bueno!, but this, too, is code-switching and bilingualism, not a “contact phenomenon” within Rapanui as such. In addition, Spanish no is used often as a discourse marker, with an emotive emphasis – No! – that sets it apart from standard Rapanui ’ina ‘no’, which is far milder in tone. Relative to connectors, see example (13) above illustrating Rapanui’s use of Spanish y ‘and’. Furthermore, Spanish pero ‘but’ is now used in Rapanui, too, as originally Rapanui had no ‘but’. For example: (14) Ko au ’i nei, pero he oho. foc 1sg loc here but ml go ‘Here I am, but I’m going.’ Spanish o is now standard Rapanui ō for ‘or’, too: (15) Hoki ku hanga ’ā mo te maika ō te ’uhi ō ’ina. int t/a desire res ben art banana or art yam or not ‘Do you want the banana or the yam or not?’ Spanish porque ‘because’ is also frequently used in Rapanui, replacing original ’o te aha (lit. ‘of the what’). As with the Bueno! example above, there is frequent bilingual code-switching in Rapanui, as evidenced by the common use of Spanish ya ‘already’ and entonces ‘therefore’, for similar reasons. Spanish siempre ‘always’, nunca ‘never’, jamás ‘ever, never’ are routinely heard as well in Rapanui, which has no specific words for these temporal concepts, simply nō ‘only’ and ’ina ‘not’ (the latter borrowed early from Tahitian); in this particular case, the borrowing of these Spanish temporal items is semantic supplementation. Sometimes one can also hear Spanish/ English anti-, but this is infrequent and is often linked to topical, foreign (i.e., non-Rapanui) themes: anti-nuclear, for example. Such Spanish words and phrases as además ‘furthermore’, en corto tiempo ‘shortly’, sino que ‘but’, ante ‘before’, juto (from Spanish justo, with /x/ phonetic intrusion) ‘just’ and a few others are now common in Rapanui discourse, too. A very recent introduction is the calque ō ’ina ‘or not’, patterned after Spanish o no. Much of the above-mentioned clearly involves code-switching. However, both puē (from Spanish puede) ‘can’ and tiene que ‘has to’ have, within the past decade, virtually become grammaticalized in the Rapanui language.
Rapanui
395
7. Constituent order Spanish influence seems to be causing Rapanui’s original VSO order to become more SVO. This applies also to VS becoming SV; that is, transitivity appears to be irelevant in the process of “Rapanui hispanicization”. However, there is great individual variation. In Old Rapanui and early Modern Rapanui, any fronted topic required introduction by the Rapanui possessive particle ’a. Today, this fronting function of ’a is now largely dispensed with, primarily because Spanish demands no such particle in this position. In addition, postpositive possessors are now almost exclusively prepositive, again probably in consequence of Spanish influence. Three generations ago, frequent use of postpositive possessors were made, although prepositive possessors prevailed. For example, archaic te hare ’o’oku ‘my house’ (lit. ‘the house of mine’) would now only be Modern Rapanui to’oku hare, which agrees with Spanish mi casa. Spanish ante ‘before’ is used in Rapanui, suffixed with progressive particles, as ante + ’ā or as ante + ana to form a noun phrase than means ‘before’, ‘previously’, or ‘in the old days’. For example: (16) Ante ’ā, tangata ta’e rahi te ha’aura’a. before prog man neg many art reason ‘It’s because there weren’t many people before’ (Makihara 2001a: 205) (17) Ante ana ho’i he kompania Williamson. before prog emph ml company Williamson ‘In the old days, there was the Williamson Company’ (Makihara 2001a: 205) Ante can also function, without progressives, before perfective particles and prepositions: (18) Ante ’i oho mai nei. before perf go dir here ‘Before coming here …’ (Makihara 2001a: 206) (19) Ante ki te tu’u mai o te presidente. before dat art arrive poss dir art president ‘Before the arrival of the president …’ (Makihara 2001a: 206)
396
Steven Roger Fischer
8. Syntax Other features evidence contact phenomena in Rapanui syntax. Presuming causative to produce a complex clause, the Rapanui causative prefix haka(< Proto-East Polynesian *faka-) now often prefixes Spanish borrowings and code-switchings. For example: (20) Muy peligroso mo hakafunciona. very dangerous ben caus.function ‘It is very dangerous to make (it) work’ (Makihara 2001a: 215, n. 31/g) In fluent discursive Rapanui, it is also becoming more frequent to initiate a statement with a Spanish adverb or adverbial phrase, as is also habitual in Spanish discursive speech (this is not a characteristic of any Polynesian language, including Rapanui). Note, for example: (21) Entonces ka orvire [olvida] tātou i rā parte. therefore imp forget 1pl.incl acc dem part ‘Therefore, let us [incl] forget that part …’ (Makihara 2001a: 212) (22) Y además to’oku mana’u en corto tiempo he rē tātou. and in addition 1sg.poss opinion in short time ml win 1pl.incl ‘And, in addition, in my opinion we’ll shortly all win’ (Makihara 2001a: 212) Question formation in Rapanui produces insightful features as well, whereby Spanish intrudes on several levels. First, Spanish porque ‘why?’ appears to be on its way to replacing Rapanui ’o te aha in every context: (23) Porque he oho atu koe? why ml go dir 2sg ‘Why are you going away?’ Spanish will even redundantly supplement the native Rapanui interrogative: (24) Porque he aha? why ±spec what ‘Why?’ (lit. ‘Why why?’)
Rapanui
397
Yet Spanish interrogatives almost entirely occur in Rapanui only during codeswitching: an Easter Islander speaking a “purist” Rapanui will not insert a single Spanish interrogative. Clearly, this is not a case of borrowing. Tahitian might now be considered “native” on the island, but Spanish manifestly is not. Spanish remains intrusive, still being consciously manipulated for effect. Spanish no will be used as expletive or response in Rapanui; in all other grammatical negation, Rapanui ’ina, kai, ’ina kai, ta’e and the suffix -kore (lit. ‘-less’) are used. Old Rapanu used simple serialization, without connectors; now, Modern Rapanui co-ordinates multiple statements using such Spanish words as y, bueno, puē (= Spanish pues ‘then’), entonces, ni … ni, o … o. Some Rapanui speakers attempt to “sophisticate” adverbial clauses using Spanish inclusions, but, again, this is code-switching, not borrowing, as these inclusions are perceived always as intrusions. For example: (25) ’Ina au kai o’o nunca ki tū hare. neg 1sg neg enter never dir dem house ‘I have never entered that house.’ (26) Ka kai koe porque repuē tiene que oho. imp eat 2sg conj después mod.has.to go ‘Eat! Because afterwards you’ve got to go.’ There is regular inclusion of Spanish que for the subordinating conjunction ‘that’, a category Rapanui does not possess. (Sometimes this is written in Rapanui – which language is seldom written – as ke.) Rapanui relative-head constituent order VS is then consistently reversed to SV with the inclusion of Spanish que (here graphic ke): (27) Hoki ’ite e koe ke ia ’i tu’u mai ai? int know ag 2sg that 3sg perf arrive dir pho ‘Do you know that he’s arrived?’ However, standard Modern Rapanui, without intrusive Spanish que, would normally express this as: (28) Hoki ’ite e koe ’i tu’u mai ai e ia? int know ag 2sg perf arrive dir pho ag 3sg ‘Do you know that he’s arrived?’
398
Steven Roger Fischer
Again, nearly all Spanish inclusions in Rapanui involve code-switching and, in the case of relative pronouns, such inclusions can be frequent; however, the same user can also speak Rapanui without any such Spanish inclusions. Spanish adverbs, as discourse fillers, would hardly appear in “formal” Rapanui speech.
9. Lexicon Lexical intermixing can occur in nearly all but the most formal registers, to such an extreme degree that Rapanui can actually appear at times to be more correctly called “Spanish Rapanui” (Makihara 1999, 2001a). All Rapanui speakers are also fluent in Spanish, and these speakers then freely intermingle both languages, to varying degrees, throughout the day, depending on to whom they are talking and the circumstances. Nonetheless, there is also Rapanui speech devoid entirely of Spanish. One cannot generalize, therefore, the process of Spanish “borrowing” on Easter Island. Lexical supplementation occurs in Rapanui more frequently than lexical replacement. Again, this involves primarily code-switching. That is, the Rapanui speaker might wish, in any given circumstance, to “impress” her or his listener(s) by using the Spanish rather than the Rapanui word(s). Spanish expressions are commonly used interchangeably with Rapanui expressions. However, seldom are set formulae split into both languages at once, unless perhaps to inject humour. True lexical copying on Easter Island nearly always involves previously unknown introductions to Rapanui culture. In this sense, a “hispanicization” of the Rapanui lexicon has occurred, albeit in the form of lexical supplementation. However, whenever a “purist’s Rapanui” is being spoken very little of the indigenous lexicon will include even Spanish replacements, much less supplements. In purely Rapanui contexts – that is, in those situations not involving foreign objects, introductions or situations – not one word of Spanish need be spoken at all. Educated Rapanui people dedicated to reviving their indigenous language are currently seeking such traditional contexts in order to promote just such a “purist’s Rapanui”. And there is very much an intuited sense of what this “purist’s Rapanui” should comprise, although this latter language, for over a century no longer Old Rapanui, is a model of Polynesian language intertwining as it is in fact a late nineteenth-century Rapanui– Tahitian hybrid.
Rapanui
399
10. Conclusion At this point in time, no fewer than five resident languages engage Polynesian-oriented Islanders on Easter Island: Old Rapanui, which occurs today only in public performance (this language is largely unintelligible); Modern Rapanui, in both public and private domains; Spanish Rapanui, the local syncretism which uses much code-switching and also relies on a greater number of Spanish borrowings than does Modern Rapanui; Rapanui Spanish, the local syncretic compromise which is also used as an ethnic emblem; and Chilean Spanish, the language of the still dominant power and now prevalent resident population (2,200 to 1,800). As a result, the active language continuum on Easter Island in 2006 would be: Modern Rapanui > Spanish Rapanui > Rapanui Spanish > Spanish. The first and second of these languages, Modern Rapanui and Spanish Rapanui, are not distinctly delineated languages per se, but robust speech varieties that interact with one another daily. Modern Rapanui is thus more of a blanket designation for the local Rapanui–Tahitian intertwining that is continuously changing. It is, at the very least, not moribund but apparently enjoying an evident, though still small and tentative, renaissance. Spanish, when used in Modern Rapanui, is wielded as a tool: normally, but certainly not exclusively, for code-switching. Of course, the Modern Rapanui lexicon has borrowed a considerable amount of Spanish (Fischer 2001); but such borrowings are limited mainly to introduced technology as well as to hitherto unknown concepts and perceptions deriving from non-Polynesian spheres of activity, with some exceptions, as illustrated above. Some grammatical borrowing from Spanish figures in Rapanui as well. Owing to the paucity of Rapanui speakers (c.8001,000), there has been no attempt as yet to translate Spanish borrowings systematically and/or formally into some kind of standardized, “received” Rapanui. This might conceivably be achieved through expanding the semantic domain of the Old Rapanui lexicon. However, under present circumstances there would be little advantage in doing this. Any reversal of the current dynamic of the hispanicization process will simply be the demonstration of fewer Spanish intrusions and the gradual return to a Polynesian exclusivity. Fluent speakers of Modern Rapanui are, on the whole, conscious of a diminishing of code-switching particularly in more formal – that is, more “Rapanui” or locally ethnic – contexts. More children are now being formally integrated into indigenous language use. The hitherto private domain
400
Steven Roger Fischer
of Rapanui use is becoming once again increasingly public. New language projects are raising local awareness of the island’s original language, and promoting more active use in the larger community in general. And the island’s imminent political autonomy will certainly open a new domain of Rapanui use, which must increase not only its status, but also its application as both a spoken and written language.
Abbreviations acc aff ag art asp ben caus conj coorc dat dem dir emph ext foc imp incl int
accusative affirmative particle agentive article aspect marker benefactive causative conjunction coordinating conjunction dative demonstrative particle directional emphatic existential focus imperative inclusive interrogative
loc ml mod neg perf pho poss prog prs res rlt t/a 1sg 2sg 3sg 1pl ±spec
locative preposition mainline aspect particle modal negative perfective aspect phoric possessive progressive aspect person/place marker resultative relational particle tense-aspect particle first-person singular second-person singular third-person singular first-person plural ± specific
References Du Feu, Veronica 1996 Rapanui. (Routledge Descriptive Grammars.) London and New York: Routledge. Du Feu, Veronica, and Steven Roger Fischer 1993 The Rapanui language. In: Steven Roger Fischer (ed.), Easter Island Studies: Contributions to the History of Rapanui in Memory of William T. Mulloy, 165168. (Oxbow Monograph 32.) Oxford: Oxbow Books.
Rapanui
401
Fischer, Steven Roger 1992 Homogeneity in Old Rapanui. Oceanic Linguistics 31: 181190. 1997 The Rapanui language. In Rongorongo: The Easter Island Script. History, Traditions, Texts, 358361. (Oxford Studies in Anthropological Linguistics 14.) Oxford: Clarendon Press. 2000 Possessive markers in Rapanui. In: Steven Roger Fischer (ed.), Possessive Markers in Central Pacific Languages, 333344. (Special dual edition of Sprachtypologie und Universalienforschung 53.) Berlin: Akademie Verlag. 2001 Hispanisation in the Rapanui language of Easter Island. In: Klaus Zimmermann and Thomas Stolz (eds.), Lo propio y lo ajeno en las lenguas austronésicas y amerindias, 313332. (Lengua y Sociedad en el Mundo Hispánico 8.) Frankfurt: Vervuert, Madrid: Iberoamericana. Forthc. Reversing hispanicization on Rapa Nui (Easter Island). In: Thomas Stolz, Dik Bakker and Rosa Salas Palomo (eds.), Hispanisation. Berlin/New York: Mouton de Gruyter. Makihara, Miki 1999 Bilingualism, social change, and the politics of ethnicity on Rapanui (Easter Island), Chile. Ph.D. dissertation, Department of Linguistics, Yale University. 2001a Modern Rapanui adaptation of Spanish elements. Oceanic Linguistics 40: 191223. 2001b Rapanui–Spanish bilingualism. Rongorongo Studies 11: 2542. 2004 Linguistic syncretism and language ideologies: Transforming sociolinguistic hierarchy on Rapa Nui (Easter Island). American Anthropologist 106: 529540. Forthc. Rapa Nui ways of speaking Spanish. Language in Society 34.
Grammatical borrowing in Nahuatl Una Canger and Anne Jensen
1. Background Nahuatl is a Uto-Aztecan language spoken in Central Mexico by around 1 million people. The early missonaries invented an orthography for the language, thanks to which we have access to sources written in Nahuatl as early as the 1540s. Hence, we are in a position to trace changes in Nahuatl during the last 500 years. Since the Spanish invasion in 1521, Nahuatl speakers have been in contact with Spanish culture including the language. Spanish is the official language in Mexico, and Nahuatl is not a written language anymore. The use of Nahuatl nowadays is restricted to oral communication among family members and friends and in the Nahuatl communities. Not all children acquire Nahuatl, so most speakers belong to the middle and old generations, and all Nahuatl speakers are bilingual in Nahuatl and Spanish. Apart from English, Spanish is the only language that Nahuatl speakers are in contact with at the present time. Since Nahuatl belongs to the Mesoamerican linguistic area (Campbell et al. 1986), it shares features with other Mesoamerican languages, e.g. a vigesimal number system and V-first. Evidently, these features are contact-induced changes in Nahuatl, the speakers of which are assumed to have been the last ones to settle in Central Mexico. The geographic extent of the area in which Nahuatl dialects were and are spoken is rather large, thus Nahuatl speakers have been in contact with a range of other Mesoamerican languages. However, the results of those contact situations still have to be explored. In the present chapter, we focus on the impact of Spanish on the grammar of Nahuatl.1 The predominant possible influence from Spanish appears in the domains of nominal structures, among these the expression of spatial and other relations, constituent order, and to a lesser extent syntax. Nahuatl is a polysynthetic language with agglutinative morphology, and these two features have been exposed to little or no contact-induced changes. Concerning phonology, Spanish phonemes, e.g. voiced stops, occur only in loanwords that have not been adapted to the Nahuatl system. The voicing of intervocalic stops found in some modern dialects is not necessarily a contact phenomenon since such a change may take place without contact. However, while wovel
404
Una Canger and Anne Jensen
length was the most prominent prosodic feature in sixteenth century Nahuatl, in the twentieth century Nahuatl vowel length is less prominent, and fixed word stress is now generally recognized (cf. Canger ms). The following information is based on data from two dialects, North Puebla (NP) and North Guerrero Nahuatl (NG).
2. Nominal structures The first change is in the number category. With a couple of exceptions,2 inanimate nouns in sixteenth-century Nahuatl were not inflected for plural. In modern dialects, however, most inanimate nouns are inflected for plural. In (1), the plural morpheme -tin is suffixed to the inanimate noun root kal‘house’; the quantifier miyek ‘much’ is marked for plural, too, by the morpheme -in: (1)
aʔmo miyek-in kal-tin. neg much-pl house-pl ‘There are not many houses.’ (NP Nahuatl)
This structural change is likely to be contact-induced since animate nouns in Spanish are inflected for plural. The other change concerning nominal structures is the emergence of a new part of speech, prepositions. In sixteenth-century Nahuatl, most spatial and other relations are encoded by means of a single postposition, -k(o) ‘at’, ‘on’, ‘in’, which attaches to inanimate noun roots, and by relational nouns. A relational noun was preceded by a possessor prefix referring to the head of construction. The possessor prefix third-person singular is attached to the relational noun -pan ‘on’ in (2), referring to the head λaʔtoʔka:ti:λanλi ‘rulermessenger’: (2)
inin λaʔto:l-li i:-pan ø-m-iʔtoa:-ya dem word-abs.s poss.3s-on sub.3-refl-say-impf in λaʔtoʔka:-ti:λan-λi. def ruler-messenger-abs.s ‘These words were said about the ruler-messenger.’ (CF IV, folio 203 recto)
The same applies to the possessor prefix in (3), which is attached to the relational noun -na:wak ‘near’, ‘at’:
Nahuatl
(3)
405
ti-wel-la-mati-s in i:n-na:wak sub.1pl-well-obj.indef-know-fut def poss.3pl-near to-te:kw-yo:-wa:n siwa:-pi-pil-tin. poss.1pl-lord-deri-poss.pl ‘You will be happy near our goddesses, the cihuāpipiltin.’ (CF VI, folio 143 recto)
In some modern dialects, relational nouns have lost the possessor prefix and function as simple prepositions. The preposition na ‘at’ in (4) is the reflex of -na:wak in NG Nahuatl: (4)
ma ya ø-m-namaka na Lupe Peña. imp go sub.3-refl-sell.pres at Lupe Peña ‘Off we go, it sells at Lupe Peña’s house.’ (NG Nahuatl)
In other dialects, the possessor prefix is maintained, but the whole word including the suffix functions as a preposition. The sentence in (5) includes i:pan, which also appears in (2). But in (5), it encodes the path of the movement encoded by the verb -wi:¢ ‘come’: (5)
ti-wi:¢ i:pan n kaʔkalaʔ-λe. sub.2s-come.pres to def village-abs.s ‘You come to the village.’ (NP Nahuatl)
In sixteenth-century Nahuatl the path was not encoded; instead the path was a meaning component of a movement verb. An example of that is given in (6); the path FROM is not encoded explicitly: (6)
ka o:sto:-yoʔ i:-λan ø-wi:¢. part cave-deri poss.3s-below sub.3-come.pres ‘It (the turquoise) comes from within the mines.’ (CF XI, folio 203 verso)
In Spanish, however, the path is encoded, in (7) by the preposition a ‘to’: (7)
Vienes al pueblo. come.pres.2s to.det.mask village ‘You come to the village.’
406
Una Canger and Anne Jensen
Since Spanish has prepositions, the reanalysis of Nahuatl relational nouns as prepositions may be ascribed to the influence from Spanish. The consequence of this reanalysis is far-reaching concerning the grammar of modern Nahuatl because a new part of speech, prepositions, has emerged. But this change has left the language (or its speakers) in a process of change: In NP Nahuatl some of the relational nouns still form a category of their own,3 at the same time they are used as prepositions. Moreover, some Spanish prepositions have been borrowed; see (31) in Section 6. In sixteenth-century Nahuatl the prevalent order of possessed and possessor is the one shown in (2), i.e. the possessor succeeds the possessed. This order may have facilitated the reanalysis of relational nouns as prepositions.
3. Verbal structures Apart from a few exceptions, Spanish verbal structures have not been borrowed, neither have Nahuatl verbal structures undergone changes. Spanish verbs have been borrowed from an early stage of the contact between Nahuatl speakers and Spaniards. Loan-verbs are derived from Spanish infinitives, which are treated like Nahuatl noun roots by means of the suffix -oa – as shown in (8) with the Spanish verb cantar ‘sing’: (8)
san ke:man ø-wenti-ʔ ø-kwi-kwi:ka-ʔ only when subj.3-be.drunk-subj.pl subj.3-red-sing-subj.pl ø-cantar-oa-ʔ subj.3-sing-deri-subj.pl ‘Only when they (the women) are drunk, do they sing, do they sing.’ (NP Nahuatl)
One verbal structure borrowed from Spanish forms part of a new mode category, it encodes obligation. The verb -piya ‘to have’, ‘ to guard’4 is used as an auxiliary, the obligatory subject prefix is attached to the auxiliary and to the main verb (there is no infinitive in Nahuatl), as shown in (9): (9)
ti-k-piya ti-k-či:wa-s mo-tarea. sub.2s-obj.3s-have.pres sub.2s-obj.3s-do-fut poss.2s-homework ‘You have to do your homework!’ (NP Nahuatl)
Except for the finite form of the main verb in the Nahuatl construction, it is a calque of the Spanish construction encoding obligation:
Nahuatl
407
(10) tienes que hacer tu tarea. have.pres.2s to do your homework ‘You have to do your homework!’ In sixteenth-century Nahuatl, future tense was used for encoding obligation, but the future tense is currently used for encoding another member of the mode category, potential – as it applies to Spanish. The main verb in (11) is -yes, the future form of -kaʔ ‘to be’: (11) ø-ye-s i:-aška Pedro. sub.3s-be-fut poss.3s-property Pedro ‘It is probably Pedro’s property.’ (NG Nahuatl) The emergence of potential within the mode category is a structural change in the sense of a change of the over-all pattern of the category. The periphrastic construction encoding obligation in (9), however, is not, since obligation was encoded in sixteenth-century Nahuatl. In (9) above, the morpheme -s expressing future tense appears. This morphological marking of future tense now co-exists with a periphrastic construction which also expresses future tense. The verb yaw ‘go’ is used as an auxiliary to which a subject prefix is attached, the auxiliary is succeeded by the main verb to which the obligatory subject prefix is attached. According to the transitivity of the main verb, an object prefix may also be attached as is the case in (12), in which the main verb is iʔkitia ‘weave’: (12) n-ya ni-k-iʔkitia. sub.1s-go.pres sub.1s-3s.obj-weave.pres ‘I will weave it’, ‘I am going to weave it.’ (NG Nahuatl) This is a calque of the Spanish periphrastic future shown in (13): (13) voy a tejer. go.1s.pres to weave ‘I will weave’, ‘I’m going to weave.’ However, its an open question, to which extent the tense category as such has undergone contact-induced changes.
408
Una Canger and Anne Jensen
4. Other parts of speech The vast majority of Spanish loans appear in other parts of speech. Nahuatl had – and still has now – a vigesimal number system, but except for the numerals below 10 Spanish numerals are preferred. Some Nahuatl quantifiers are still used in everyday speech, one of these is miyek ‘much’, ‘many’ (see (1) in Section 2), but in NP Nahuatl, the Spanish quantifier poco ‘(a) little’ has been substituted for the Nahuatl one. This is not the case in NG Nahuatl, where the Nahuatl quantifier te¢i ‘(a) little’, ‘a few’ is used. Apart from a few Nahuatl connectors still in usage, the connectors introducing adverbial clauses are all Spanish. However, the syntax of adverbial clauses remain Nahuatl. Moreover, in NG Nahuatl the three Spanish coordinating connectors have been borrowed, i.e. y ‘addition’ (14), o ‘disjunction’ (15), and pero ‘contrast’ (16). (14) in tepewaši ø-li:l-ti-k y ya:l ø-limpio. def tepehuaje sub.3-black-deri-partc and pron.3s sub.3-clean ‘The tepehuaje (a tree) is black, and it is clean.’ (NG Nahuatl) (15) ø-k-ilwia i:-naʔnaʔ o i:-taʔtaʔ sub.3-obj.3s-say to.pres poss.3s-mother or poss.3s-father ni-k-elewia inon ičpoka-λ. sub.1s-obj.3s-like.pres that young.girl-abs.s ‘He (the young man) says to his mother or his father, “I like that young girl.”’ (NP Nahuatl) (16) niʔʔ-k-tila:na ok-se-pa i:ka no-derecha sub.1s-obj.3s-pull.pres still-one-time with poss.1s-left pero am ø-ki:sa in nekwal-mekal. but neg sub.3-come out.pres def necualmecate ‘I pull it once more with my left (hand), but the necualmecate does not come out.’ (NG Nahuatl) In NP Nahuatl, o ‘or’ and pero ‘but’ have also been substituted for Nahuatl connectors, and in both dialects the two connectors may connect phrases, clauses and sentences. Spanish y ‘and’ is not used in NP Nahuatl. Instead the speakers use (i:)wan ‘and’. In (17) two sentences are coordinated:
Nahuatl
409
(17) ni-k-či-čipe:wa i:šwa-to:ma-λ wa:n luego este sub.1s-obj.3s-red-peal leaf-tomato-abs.s and then ah ni-k-¢in-ko-koto:na čil-le. sub.1s-obj.3s-bottom-red-cut chili-abs.s ‘I peal leaf tomatoes and then, ah, I cut off the stalk of the chili fruits.’ (NP Nahuatl) Apart from that, (i:)wan also coordinates phrases, i.e. like the second wan in (18): (18) inon costumbre ø-ki-piya n pueblo that practice sub.3-obj.3s-have.pres def village a:kin ø-mo-namik-tia who sub.3-refl-meet-caus.pres bien ø-ki-ø-λokolia-ʔ miyek cosas well sub.3-obj.3s-obj.3s-donate-subj.pl many things wan telpoka-λ ø-ki-ø-λokolia-ʔ cervezas and young.man-abs.s sub.3-obj.3s-obj.3s-donate-subj.pl beers wan guajolotes. and turkeys ‘That practice the village has: one who marries – they donate him/her many things, and young man [sic!], they donate beers and turkeys.’ (NP Nahuatl) The first wan shows the third level of coordination – it combines two succeeding, units of discourse larger than a sentence the contents of which are related to each other. In sixteenth-century Nahuatl, -wa:n occurs as a relational noun meaning ‘in the company of’,5 it was used for connecting phrases and to a certain extent clauses and sentences. Still, -wa:n does not function at discourse level for combining succeeding “chunks” – this function was carried out by aw ‘and’, which is not found in NP Nahuatl. Thus, the use of wan in NP Nahuatl resembles the function of Spanish y, and this fact may be considered a contactinduced change. In addition to the three aforementioned connectors, the Spanish adverbs luego ‘subsequently’, ‘then’ (see (17) above) and entonces ‘then’ (19) are used as discourse organizing devices in NP Nahuatl,6 whereas NG Nahuatl also maintains the Nahuatl word okino ‘then’, ‘thus’ (20):
410
Una Canger and Anne Jensen
(19) ni-k-elewia inon ičpoka-λ sub.1s-obj.3s-like that young girl-abs.s iwkon ø-k-iʔto-s n novio in that way sub.3-obj.3s-say-fut.s def fiancé entonces ø-ya-s i:-čan n novio then sub.3s-go-fut.s poss.3s-home def fiancé ‘“I like/want that young girl”, the fiancé will say it in that way, then the fiancé will go (to his) home.’ (NP Nahuatl) (20) okino in ok-se: kwaw-oʔλal ø-ki-piya then def still-one wood-stick sub.3-obj.3s-have.pres in hilo len-i:ka ni-k-wapana in hilo def warp what-with sub.1s-obj.3s-fasten def warp okino in kwaw-oʔλal in len-i:pa o then def wood-stick def what-on past ø-m-parejo ya ø-ki:sa sub.1-refl-align.pret.s now sub.3s-come out.pres ‘Then the warp has another warp rod to which I fasten the warp, then the warp rod on which I have aligned the warp, comes out.’ (NG Nahuatl) The discourse markers pues ‘then’, ‘well’, bien ‘well’ and bueno ‘well’, ‘sure’ and the hesitation marker este (see (17)) are borrowed from Spanish.
5. Constituent order The predominant constituent order of clauses in sixteenth-century Nahuatl is VS, or rather predicate–subject, since nouns and other parts of speech may function as predicates; it is still discussed whether VOS or VSO is the unmarked order. As it appears from e.g. (17) and (19) in Section 4, VO-order (and the order predicate–subject) is still used in modern Nahuatl, there are, though, occurrences with SV- and OV-orders; in (14) above the subject in tepewaši ‘tepehuaje’ (a tree) precedes the predicate li:ltik ‘black’, and in (18) the object inon costumbre ‘this practice’ precedes the verb -piya ‘has’. Still, such constituent orders are also found in sixteenth-century Nahuatl, thus nothing can be concluded with regard to Spanish influence on constituent order in main clauses. With respect to subordinate clauses, the constituent
Nahuatl
411
order in modern Nahuatl is identical to the order found in sixteenth-century Nahuatl, albeit subordinating devices have been borrowed from Spanish. Within some categories of phrases, changes have taken place. A noun phrase may precede a modifier or vice versa in sixteenth-century Nahuatl. The noun phrase in ¢a¢apa:sli we:yi lit. ‘the batten big’ from NG Nahuatl (see (24) in Section 6), and the noun phrase se: pueblo kwal¢in ‘a village nice’ from NP Nahuatl in (30) are two modern examples of the order head– modifier. However, the preferred constituent order is modifier–head. The abandonment of the head–modifier order may not be due to Spanish influence, since that constituent order is found in Spanish, too. Another category of phrases includes quantifiers. In sixteenth-century Nahuatl, the quantifier moči/noči ‘all’ occupied clause-initial position, while the quantified noun phrase occupied a position subsequent to the verbal complex. This applies to the subject in (21) which is quantified by the initial močintin ‘all’: (21) moči-ntin ø-m-a:yo:-iʔčiki:-ya-ʔ in okič-tin all-pl sub.3-refl-fontanel-scrape-impf-sub.pl def man-abs.pl i:wan in siwa:-ʔ and def woman-abs.pl ‘All the men and (all) the women shaved their heads.’ (CF X, folio 138 recto) In modern Nahuatl the noun phrase always occurs immediately after noči: (22) después este n pastel ø-mo-repartir-oa noči gente. then ah def cake subj.3-refl-distribute-deri all people ‘Then, ah, the cake it is distributed (to) all (the) people.’ (NP Nahuatl) The change from the originally discontinuous phrase shown in (21) to a continuous phrase like the one in (22) may be contact-induced, since a quantifier and a noun (phrase) always form a continuous phrase in Spanish. However, the development may have taken place independently. Phrases including other quantifiers, e.g. miyek ‘much’ (see (1) in Section 2), were continuous and the current constituent order including moči/noči may be a result of the speakers’ wish for analogue phrase structures.
412
Una Canger and Anne Jensen
6. Syntax One of the characteristic syntactic features of sixteenth-century Nahuatl is the existence of prehead relative clauses, which co-existed with post-head relative clauses. The head of the relative construction in (23) is in amiʔi:yo:¢in in amoλaʔto:l¢in ‘your honoured breath, your honoured words’, the preceding relative clause is inside the square brackets: (23) ma: ti-k-to-mak-toka-tin part sub.1pl-obj.3s-refl-give-follow-vet.pl [in nika:n ø-wa:l-ki:sa] [conn here sub.3-dir-go out in am-iʔi:yo:-¢in in amo-λaʔto:l-¢in. def poss.2pl-breath-hon def poss.2pl-word-hon ‘Let us avoid pretending that your honoured breath, your honoured words [that are coming out here] are given to us.’ (FC III, folio 36 verso) Pre-head relative clauses are absent in modern Nahuatl, and may have been abandoned due to Spanish influence; Spanish had and still has only post-head relative clauses. As it appears from (23), relative clauses in sixteenth-century Nahuatl are formed by gapping;7 the relative clause is introduced by the connector in. In NG Guerrero, that strategy is still found. The relative clause in (24) is introduced by in: (24) m-ø-piya in ¢a¢apa:s-li we:yi y ok-se: sub.1s-obj.3s-have.pres def batten-abs.s big and still-one ¢a¢apa:s-li [in ø-ki-toka]. batten-abs.s [conn subj.3-obj.3s-follow.pres ‘I have the big batten and another batten [that follows it].’ (NG Nahuatl) In some cases in is omitted, and intonation is the only indication of whether a clause should be interpreted as a relative clause or as a simple sentence of its own. Still another strategy for relative clause formation is found in NG Nahuatl. Given that NP-Rel has the syntactic function of an adverbial, a relative pronoun combined with a preposition is used. This is the case in (20) in section
Nahuatl
413
(2). The two relative clauses are introduced by len ‘what’ (an interrogative pronoun), to which the prepositions i:ka ‘with’ and pa ‘on’ are attached. (former relational nouns). Irrespective of the syntactic function of NP-rel, relative clauses are introduced by a relative pronoun in NP Nahuatl. It is also an interrogative pronoun, which shows whether the NP-head is human or not. The composite pronoun λeinon ‘what’ is used when NP-head is -human: ašan]. (25) inon noči1 [λein-on1 ni-k-mati dem all what-dem sub.1s-obj.3s-know.pres now ‘That is all [that I know now].’ (NP Nahuatl) When the NP-head is +human, the pronoun a:kin ‘who’ introduces the relative clause: (26) pero noč-ten1 [a:kin1 ø-ki-piya-ʔ servicio] but all-pl [who sub.3-obj.3.s-have-pl service o:-ø-λa-šλaw-keʔ past-sub.3-obj.indef-pay-pret.pl ‘But all [who have service = electricity] have paid.’ (NP Nahuatl) Finally, the speakers use Spanish que for introducing relative clauses: (27) pero ye kin i:-tiempo que ø-walla-s inon ampliación but already just poss.3s-time that sub.3-come-fut dem extension [que ø-teč-segurar-o-kiw]. [that sub.3-obj.1pl-guarantee/secure-deri-intro ‘But soon it was already the (its) time that that extension [that came to secure us] would come.’ (NP Nahuatl) In sixteenth-century Nahuatl headless relative clauses were introduced by the connector in followed by an interrogative pronoun. If the referent was non-human, the interrogative λein ‘what’ was used. If the referent was human, the interrogative a:kin ‘who’ was used: (28) no: ø-ki-a:na-ti:w [in λein also sub.3-obj.3s-grab-dir.imperf [conn what no-yo:l-ka:-w ni-k-nemi:-tia]i . poss.1s-live-partc-poss.s sub.1s-obj.3s-live-caus.pres ‘He will also go to grab [whatever animal I raise].’ (CF XI, folio 8 recto)
414
Una Canger and Anne Jensen
(29) inin λaʔto:l-li i:i-tečpa ø-m-iʔtoa [in a:kin dem word-abs poss.3s-at sub.3-refl-say.pres [conn who iλaʔ senkaʔ λasoʔ-λi ø-k-iʔλakoa]i. something very valuable.thing-abs.s sub.3-obj.3s-destroy.pres ‘These words are said about [whoever destroys something very valuable]’ (CF VI, folio 199 verso) Assuming that the abandoning of in as a general subordinator NP Nahuatl was the first step of a change, then there would have been no marking of a restrictive relative clause, whereas a headless relative clause would still have been introduced by one of the interrogatives a:kin and λein. The speakers may have expanded the use of the interrogatives to introduce restrictive relative clauses as well. Later they may have borrowed the Spanish subordinator que – which may be on its way to becoming the preferred item for introducing relative clauses, as it is in other Nahuatl dialects. As mentioned in Section 4, other categories of subordinate clauses have not been affected by the contact with Spanish. However, Nahuatl is a polysynthetic language, and it may be in the process of drifting towards a less polysynthetic structure due to Spanish influence. In Section 3 two new periphrastic constructions were shown: In (9) the construction used for encoding obligation – a calque of the corresponding Spanish construction; in (12) the periphrastic construction expressing future tense was shown. There are at least two further instances of periphrastic constructions that may be devices for the possible drift mentioned above. The first is shown in (30), and consists of a main clause with the verb -elewia ‘like’, ‘want’ and a complement clause introduced by the particle ma: (30) ni-k-elewia ma ø-mo-či:wa se: pueblo kwal¢in. sub.1s-obj.3s-like part sub.3-refl-make.pres one village nice ‘I like that it becomes a nice village.’ (NP Nahuatl) The other construction is formed by a main clause with the verb -neki ‘want’ and a subordinate clause in which the verb is in the future tense: (31) ya o ø-ki-nek ø-mo-mač-ti-s now past sub.3-obj.3s-want.pret sub.3s-refl-know-caus-fut.s mas entonces hasta Huauhchinango. more then to Huauhchinango ‘Now he wanted to learn more, then (off) to Huauhchinango.’ (NP Nahuatl)
Nahuatl
415
The complex construction in (31) is also found in sixteenth-century Nahuatl, co-existing with a composite verb form, in which -neki is an auxiliary, the main verb occurs in the future tense: (32) aw is ø-nel-li ø-mo-toli:nia-ʔ in and here sub.3-truth-abs.s sub.3-refl-afflict-sub.pl def ikno:-k wa:w-λi in ikno:-o:se:lo:-λ orphan-eagle-abs.s def orphan-jaguar-abs.s in ø-miki-s-neki-ʔ in conn sub.3-die-fut-want-sub.pl conn aʔ-ø-nemi-s-neki-ʔ neg-sub.3-live-fut-want-sub.pl ‘And here is (the) truth: the poor eagle (and) the poor jaguar that want to die, that do not want to live, suffer.’ (CF VI, folio 17 recto) The composite verb form in (32) is not found in the two Nahuatl dialects in question, and its abandonment in favour of the complex construction may be caused by the contact with Spanish as the constructions in (30)–(31) resemble Spanish constructions with the verb querer as main verb.
7. Lexicon Spanish loanwords are adopted in many domains of the Nahuatl lexicon including kinship terms. Presumably, the amount of loanwords correlates with a speaker’s age, in the sense that children and younger adults use more loanwords than older speakers. However, in the domain of grammatical vocabulary – demonstratives, (emphatic) pronouns and place deictics – no borrowing has taken place.
8. Conclusion Discourse organizing morphemes and expressions have been borrowed from Spanish, some prepositions have been borrowed too, just like subordinating connectives introducing adverbial clauses have been borrowed. Prepositions as a new part of speech have emerged from relational nouns, possibly due to the influence from Spanish. However, currently (some) relational nouns co-exist with prepositions. Furthermore, Spanish verbs and nouns have been
416
Una Canger and Anne Jensen
borrowed into Nahuatl, but in general they have been adapted to the Nahuatl structure of verbs and nouns, i.e., Nahuatl subject, object and possessor affixes are attached to the borrowed items. The marking of tense and direction has not been affected by Spanish. Furthermore, there are only few obvious contact-induced changes in the syntactic structure of Nahuatl, for example, relative clause formation. Changes of constituent order within some types of nominal phrases have taken place, and the changes may or may not be the result of contact with Spanish. Negation at all levels shows no influence from Spanish, the same applies to non-verbal predication. Nahuatl speakers still use nominal sentences. It seems as if the polysynthetic structure of Nahuatl is becoming less polysynthetic as some polysynthetic structures have been substituted by perifrastic constructions corresponding to Spanish ones. However, a comparison with the development of other polysynthetic languages exposed to a long contact with Spanish may reveal whether the development in Nahuatl is unique or a common feature.
Abbreviations abs caus conn def dem deri dir dire
absolutive non-possessed causative connector definite demonstrative derivational morpheme directional directional inflection with a final meaning fut future hon honorific imp imperative impf imperfective inflection, which may not have strictly imperfective meaning indef object prefix referring to no referent intro introverse verb conjugation with aspectual and final meaning irr irrealis
neg obj part past
negation object particle augment, present in the preterite and the pluperfect pl/pl plural poss possessor prefix showing the person and number of the possessor, or possessum suffix showing the number of the possessed pres present pret preterite pro emphatic pronoun partc participle red reduplication refl reflexive prefix s/s singular sub subject vet vetative
Nahuatl ¢ č λ ʔ
= = = =
417
dental affricate with dental-alveolar release dental affricate with palato-alveolar release dental affricate with lateral release glottal stop
Notes 1. Nahuatl has had an immense influence on Mexican Spanish with regard to particular domains of the lexicon, e.g. flora and fauna. Next to no research has been carried out on the Nahuatl impact on Spanish grammar. 2. tepe:λ ‘mountain’, and ciλa:lin ‘star’. 3. Some of them form part of composite nouns, as it is the case in sixteenth-century Nahuatl: ø-kaʔ λa:l-pan ø-kaʔ i:pan λa:l-li sub.3-be.pres earth-on subj.3s-be.pres on earth-abs.s ‘it is on the ground’ ‘it is on the ground’ 4. Before contact with Spanish -piya meant only ‘to guard’, but undoubtedly due to Spanish influence it acquired the additional meaning ‘to have’ 5. However, i:-wa:n may already have been reanalyzed as a connector in sixteenthcentury Nahuatl. In (21) in Section 5, the possessor prefix i:- third-person singular does not agree with the plural of the noun siwa:- ‘woman’. 6. Moreover, the Spanish adverb después ‘subsequently’, ‘then’ has been borrowed into NP Nahuatl. 7. We do not conceive of the obligatory subject and object affixes as elements expressing NP-rel.
References Campbell, Lyle, Terrence Kaufman, and Thomas C. Smith-Stark 1986 Meso-America as a linguistic area. In: Language 62 (3): 530558. Canger, Una 1980 Five Studies Inspired by Nahuatl Verbs in -oa. Copenhagen: Reitzel/ The Linguistic Circle of Copenhagen. 1988 Nahuatl dialectology: A survey and some suggestions. In: Language 54 (1): 2872. N.d. (Changing) word prosody in Nahuatl. MS. Carochi, Horacio, S. J. 1983 [1645] Arte de la lengua mexicana con la declaracion de los adverbios della. Facsimile of 1645 edition. México: Universidad Nacional Autónoma de México.
418
Una Canger and Anne Jensen
Cerón, Isaías Mendoza, and Una Canger 1993 In tequil de morrales. Copenhagen: Reitzel. Jensen, Anne Forthc. The emergence of progressive aspect in modern Nahuatl. To appear. Forthc. Syntactic changes in Nahuatl. To appear. Launey, Michel 1986 Categories et operations dans la grammaire nahuatl. Ph.D. dissertation, Université de Paris-IV. 1992 Introducción a la lengua y a la literatura náhuatl. México: Universidad Nacional Autónoma de México. Facsimile edition: CF = Codice Florentino: el manuscrito 21820 de la Collección Palatina de la Biblioteca Medicea Laurenziana (en facsimil). Florencia: Biblioteca Medicea Laurenziana 1979.
Grammatical borrowing in Yaqui Zarina Estrada Fernández and Lilián Guerrero
1. Background1 Yaqui, including Mayo, its major dialect, along with two other distinct languages of this family, Tarahumara and Guarijio, constitute the Taracahitan branch of the Uto-Aztecan linguistic family.2 Yaqui is spoken mainly in Mexico by more than 15,000 people living along the Yaqui River in the Central West part of the State of Sonora. Across the US–Mexican border, in Pascua, Arizona, just south of Tucson, there is an estimated 1,000 speakers of this language. The Yaqui speakers presently in Arizona migrated to the US at the beginning of the twentieth century. The traditional Yaqui settlements in Mexico are eight small towns: Cócorit, Bácum, Tórim, Vícam, Pótam, Ráhum, Huirivis, and Belén; the first six were founded by Spaniards beginning in 1617, although the first Spanish contact goes back to 1523, when Diego de Guzmán tried to conquer this original, indigenous nation. Since then, Yaqui has been in continuous contact with Spanish.3 Nowadays, most Yaqui people speak Spanish, but with different degrees of competence. In such a contact situation, Yaqui is the minority or vernacular language, and Spanish (or English in the US) the dominant language. The degree of bilingualism is typically asymmetrical. There are a few speakers, most of them elderly, who do not seem to understand or speak Spanish in Mexico, or English in the US, and who might be considered as monolingual in Yaqui. The Yaqui are generally known as an indigenous group that has demonstrated strength, pride, and a demanding character throughout four hundred years of Spanish occupation. It has probably waged more military revolts against the Spanish or Mexican governments than any other group, particularly from 1608 till 1929. The Yaqui are also among the few native groups that do not allow others to photograph them or record their festivities. Currently, the Yaqui language is spoken within a family context, during religious rituals and ceremonies, as well as in traditional government events. Most of the situations in which Yaqui is spoken take place among people belonging to the same ethnic group, but in other everyday activities, e.g. political, educational, or economic, the speakers make use of Spanish. From 1994
420
Zarina Estrada Fernández and Lilián Guerrero
till now they have conducted a bilingual program in order to teach Yaqui in all elementary schools in the Yaqui area. Most of the data considered for this chapter are the results of Estrada’s own field notes, while preparing a Yaqui–Spanish dictionary (Estrada et al. 2004), while preparing a language documentation archive (Estrada and Buitimea, in press),4 and also while preparing a collection of several discourse genres, now in progress.5
2. Phonology The Yaqui sound system has five vowels, fifteen consonants, two of which are glides. In comparison with other Uto-Aztecan languages, the Yaqui phonological system is quite simple and it resembles the Spanish one.6 Vowels in Yaqui are the same as in Spanish: /a/, /e/, /i/, /o/ and /u/. The complexity of the system is found in long vowels, in their syllabic weight, and in their combination with the glottal stop. In the adaptation of loanwords, vowel lengthening replaces a stressed vowel from Spanish: (1) (2) (3) (4) (5)
boosio faajam kameeyo serbeesa taasa
‘goiter’ ‘belt, girdle’ ‘camel’ ‘beer’ ‘cup’
(Spanish bocio) (Spanish faja) (Spanish camello) (Spanish cerveza) (Spanish taza)
The Yaqui consonant system differs from the one in Spanish only in a few features: lack of the labiodental fricative /f/, lack of the dental stop /d/, presence of the labiovelar fricative /bw/, presence of the glide /w/ which also replaces the Spanish velar stop /g/, presence of an aspirated laryngeal fricative /h/ written by the Yaqui from Sonora as <j>, and finally the absence of the trill /r/ and the palatalized /ñ/. The phonological impact of some of those features in the adaptation of loanwords can be illustrated as follows:7 (6) (7) (8) (9)
arorno kape kareta wolpo
‘adornment’ ‘coffee’ ‘cart’ ‘gulf’
(Spanish adorno) (Spanish café) (Spanish carreta) (Spanish golfo)
Yaqui
421
3. Nominal structures There are two phenomena related to nominal structures that arise in elements borrowed into Yaqui. Following Matras and Sakel (2007), both involve the direct borrowing of morphemes (i.e., mat), rather than remodeling of the structure (i.e., pat). The first phenomenon is not a general one, since it applies randomly to a limited number of Spanish plural nouns. As shown in the examples (10) through (13), some nouns are borrowed as plural forms marked by -s; this suffix however, has lost his functional value in Yaqui due to semantic bleaching: (10) (11) (12) (13)
aaso-s laabo-s waeba-s ankele-s
‘garlic’ ‘nail’ ‘guava’ ‘angel’
(Spanish ajo) (Spanish clavo) (Spanish guayaba) (Spanish ángel)
One possible explanation is that Yaqui adapts non-plural nouns from Spanish (e.g. aasos ‘garlic’ < Spanish ajo) as if they were plural or collective entities. Evidence for the semantic bleaching of the Spanish plural suffix is observed in example (14), where the lexical item waka-s ‘cow-pl’ is additionally pluralized by means of the Yaqui plural suffix -im: (14) waka-s-im ‘cows’ (Spanish vaca) Examples of non plural or non collective nouns borrowed from Spanish but bearing the plural suffix -im from Yaqui are illustrated below, where due to morpho-phonological factors, -im sometimes reduces to -m: (15) (16) (17) (18)
kuchi’-im kajtiila-m santo-m ornia-m
‘knife-pl’ ‘castle- pl’ ‘Saint- pl’ ‘stove-pl’
(Spanish cuchillo) (Spanish castillo) (Spanish santo) (Spanish hornilla)
The following examples show that Spanish mass nouns are also adapted into Yaqui as plural nouns: (19) (20) (21) (22)
aina-m chicharoon-im pan-im aros-im
‘flour- pl’ ‘pig skin- pl’ ‘bread- pl’ ‘rice- pl’
(Spanish harina) (Spanish chicharrón) (Spanish pan) (Spanish arroz)
422
Zarina Estrada Fernández and Lilián Guerrero
The second phenomenon concerns the borrowing of the suffix -ero which derives agentive nouns. The suffix occurs in Yaqui as either -eo, -e’o or -ero; Dedrick and Casad (1999: 77) document the suffix as -leo ~ -reo, but they do not identify the suffix as originally from Spanish. Two examples from these authors are kanoa-reo-m ‘boat-agt-pl’ (from Spanish canoa ‘boat’) and kuču-leo ‘fish-agt’ (from Yaqui kuchi ‘fish’). (23) (24) (25) (26) (27) (28)
bantear-eo apar-eo tampar-eo kapint-eo labele-eo bak-e’o
‘who carries the flag’ ‘who plays the harp’ ‘who plays the drum’ ‘carpenter’ ‘who plays violin’ ‘cowboy’
(Spanish abanderado) (Spanish arpero) (Spanish tamborilero) (Spanish carpintero) (old Spanish ravel) (Spanish vaquero)
Most of the time, the meaning of the loanword is maintained in Yaqui, although semantic extensions may be observed. For instance, the Spanish word vacas ‘cow’ in (14) is first borrowed as wakas ‘cow’, it then extends its meaning to ‘meat’; later when the agentive suffix is added, it derives the noun wakareo meaning ‘butcher, the person who sells meat’. A recent loanword from the same Spanish noun vaca is shown in (28), where the phoneme /b/ from Spanish is preserved in the derivation of the agentive noun ‘cowboy’ (or ‘horseman’). The use of this derivational suffix is very productive within the language, since it is also used to derive agentive nouns from non-Spanish nouns and verbs, as demonstrated in: (29) (30) (31) (32) (33)
kuta-reo tajkai-reo bwik-reo ji’ik-reo ji’ojte-reo
‘woodcutter’ ‘tortilla maker’ ‘singer’ ‘dressmaker’ ‘writer’
(Yaqui kuta ‘wood’) (Yaqui tajkai ‘tortilla’) (Yaqui bwike ‘sing’) (Yaqui ji’ike ‘sew’) (Yaqui ji’ojte ‘write’)
4. Verbal structures The Yaqui verbal structure is barely affected as the result of contact, except for one case of direct loan of a morpheme (mat): the verbalizer suffix -oa. This suffix, originally borrowed from Nahuatl, is used to derive finite verbs from Spanish infinitival forms. Some examples are presented below:
Yaqui
(34) (35) (36) (37) (38) (39) (40)
abogar-oa wantar-oa kombilar-oa piar-oa leiar-oa passar-oa pensar-oa
‘to advocate, argue in favor’ ‘to hold’ ‘to mix’ ‘to lend’ ‘to read’ ‘to pass’ ‘to think’
423
(Spanish abogar) (Spanish aguantar) (Spanish combinar) (Spanish fiar) (Spanish leer) (Spanish pasar) (Spanish pensar)
The adaptation of Spanish verbs via the Nahuatl suffix -oa, is relatively new in Yaqui. Karttunen and Lockhart (1976: 32) mention that the -oa strategy (along with two related forms, -uia and -ltia) was generalized in Nahuatl around 1700. This process may have been expanded into Yaqui a bit late, since the grammar of Tomas Basilio, published by Buelna (1989),8 only provides three verbal loans, none of them with the suffix -oa (and all involving the Yaqui verbalizer suffix -te): capom-te ‘to cut testicles’ < Spanish capon; manso-te ‘to tame’ < Spanish manso ‘tame’; and compes-ec-te or pes-ec-te ‘to confess’ < Spanish confesar plus -ek ‘to have’. A slightly different explanation is provided by Dedrick and Casad (1999: 143), who argue that the suffix -oa may have been derived from the Yaqui verb jooa ‘to do, to make’ (hooa in Dedrick and Casad’s orthography). Later on their grammar, and mainly based on Karttunen (1984: 4), the authors recognize that some of the verbs borrowed from Spanish “were mediated through Nahuatl” (Dedrick and Casad 1999: 325326). In addition to jooa, Estrada et al.’s Dictionary (2004) lists only four non-Spanish verbs ending in the suffix -oa, shown in (41)–(43). Rather than cases of morphological derivation through the suffix -oa, these verbs might be the case of a phonological coincidence: (41) bamsoa ‘to hang’ (42) bo’ojoa ‘to walk’ (< bo’o ‘road’ and jooa ‘to do’, ‘to make’) (43) jaboa ‘to become full’ Examples showing the full adaptation of those verbs are provided in (44). As with any other verbal stems, tense–aspect–modal operators, as well as voice morphemes are properly added to these verbal loans. Furthermore, the Spanish loanwords taking the suffix -oa have been also expanded into other Uto-Aztecan languages: e.g. Huichol, e.g. panchar-oa ‘to iron’ (Chablé p.c.).9
424
Zarina Estrada Fernández and Lilián Guerrero
(44) a. jumak jente-ta alborotar-oa-k … perhaps people-acc uphold-vblz-pfv ‘quizás (eso) alborotó a la gente …’ ‘perhaps (the fact) uphold the people ’ b. junaman bea te desarmar-oa-wa-k … there then 1pl disarmed-vblz-pass-pfv ‘allá entonces fuimos desarmados …’ ‘We were then disarmed …’
5. Other parts of speech Besides the cases commented on before, most Spanish lexical influence is found in the category of ‘other parts of speech’, especially numerals, function words, discourse markers, vocatives and idioms. Here, we find both, elements taken over directly from Spanish, i.e. mat, and elements remodeling the Yaqui structure, i.e. pat. In fact, this last category opens the discussion on when lexical and grammatical borrowing ends and code-switching begins. The numeral system in Yaqui is vigesimal (base 20), and evenly involves basic, derived and compound items. The basic numerals, 16, are senu or wepul, woi, baji, naiki, mamni, busani; nine is batani. The next three are derived: woobusani ‘seven’, wojnaiki ‘eight’, and wojmamni ‘ten’. Numbers from 10 through 20 are compound nouns: all take the stem for ten as the first part of the compound plus the adverbial locative element ama ~ aman ‘there’ followed by the cardinal number, e.g. wojmamni aman senu ‘eleven’. Above 20, the numerals take the word takaa ‘body’ (i.e. 20 fingers) as the basic unit, e.g. senutakaa ‘twenty’. The rest of the system gets more complex as in woumamni bajisi aman mamnitakaa ‘seven hundred’. For this reason, numbers above ten are often borrowed from Spanish. The most common function word borrowed from Spanish is the conjunction o ‘or’: (45) a. ume tomt-i kateme achai-m-ta-ka o dem-pl born-adjvz walk.pl-nmlz father-pl-acc-sub or mala-m-ta-ka… mother-pl-acc-sub ‘Those youngsters from today, their fathers or mothers…’ b. o ankeliito-m-ta-k-uni kaa aet ju’unea-tek… or angels-pl-acc-sub-also neg 3sg.obl know-cond ‘Or when the children didn’t understand…’
Yaqui
425
c. o jiba yo’ota bea au nattemai-ne… or always elder-acc then 3sg.dat ask-fut ‘Or they always will ask to/for an elder…’ Another common borrowed subordinator is pos ~ poj ‘well’, ‘then’, ‘so’ (< Sp. pues). Such discourse particles correspond to what Mushin (2001) considers a hesitation word. That is, pos communicates an act of hesitation, where the speaker gives place to a suspension of opinion or action, an act of doubt or vacillation when the speaker or narrator looks for a brief moment to think on what it follows or on what was just said. The clause in (46a) is a good evidence to show the adaptation of this discourse particle into the Yaqui grammatical system, since the second position enclitic subject pronoun =ne is attached at the end of poj. In (46c), pos appears in second position, immediately following (bwe)ta ‘but’. (46) a. Poj… inepo jewi im naa weye… well 1sg.nom yes loc dir go.sg.hab ‘Well…, I live here…’ b. Poj ne kaa in=ejkuela-k kaita so 1sg.nom neg 1sg.nom=school-have neg ejtudio-ta ne jippue-k… studies-acc 1sg.nom have-pfv ‘so…, I haven’t had school, no studies I had…’ c. Ta pos kaa itepo a=kulpa-k… but then neg 1pl.nom 3sg.acc=guilt-pfv ‘But we are not guilty…’ Yaqui makes use of several strategies to express conditional clauses. However, the Spanish marker si ‘if’ is commonly used in this type of constructions. As in Spanish, the particle si usually introduces the protasis clause: (47) a. si nee a=mabett-ne-’u… if 1sg.acc 3sg.acc=accept-fut-nmlz ‘If I were accepted…’ b. junama bea ne economia-ta bea ne there then 1sg.nom economics-acc then 1sg.nom nattemae-k si ama aayu-k o kaa ama aayu-k. ask-pfv if there exist-pfv or neg there exist-pfv ‘(And) there, then, I asked if there was economics or not.’
426
Zarina Estrada Fernández and Lilián Guerrero
In Spanish, the comparative conjunction komo ‘than’ (< Sp. como) has more than one function, one to express similarity, another to express comparison. Yaqui uses the suffix -su for the first function as in (48a), and the postposition benasi for the second one, as in (48b): (48) a. joan wikia-ta wi~wikosa-su-k. John string-acc rdp~belt-mod-pfv ‘John used a rope like a belt.’ b. aapo ousi tu’i-si tekipan-oa em=usi-ta 3sg.nom int good-mod work.prs-vblz 2sg.gen=son-acc benasi. comp ‘He works as well as your son.’ When komo occurs as a loanword, it resembles a comparison particle, but it also functions as an evidential marker expressing a greater degree of certainty. In the following examples komo might be translated into English as ‘more or less’, ‘as’, ‘like about’, or ‘like’: (49) a. tekipanoa-reo-tu-kan komo todo el tiempo, jewi! work-agt-cop-sub like all det time, yes ‘(It was a good) worker like all the time, yes!’ b. binwatu komo setentai ocho wasukte… time_ago like seventy eight years ‘(I was born) like about seventy eight years ago…’ c. junum bea ne komo junak tiempo-po kaa ejkuela-wa… then then 1sg.nom like moment time-loc neg school-pass ‘Then, about that period there was no school…’ d. komo iane eme inim neu aane into=t like now 2pl.nom here 1sg.dat exist.hab and=1pl.nom nau=aane. together=exist.hab ‘as now that you are here with me and we are together…’ Another borrowed particle is the Spanish temporal preposition hasta ‘up to’ as a locative, limitative particle illustrated in (50a, b), and hasta que ‘until’ as a temporal, limitative marker in (50c). Although the preposition hasta has in Mexican Spanish two possible meanings, till and when, the second one has not yet been documented in Yaqui:
Yaqui
427
(50) a. jeka-po chaasime asta junum ian “Ten Jawee-po”. wind-loc hang-go.sg.prs till there today mouth open-loc ‘It went rolling down up in the air till what is today the “open mouth” mountain.’ b. asta Merida-wi pino-m ja’awe-ka’a-po lutula. till Merida-dir pine-pl be-pfv-loc right ‘Till Merida, right there where the pines are.’ c. asta ke n-a=bwise-k. until that 1sg.nom-3sg:acc=get-pfv ‘Until the time I got it.’ Contrary to what has been documented by Lindenfeld (1971, 1982) very few cases of the Spanish subordinator que have been found in our database which is mainly based on oral discourse materials, except for one speaker, Agripina Amarillas, which makes an extensive use of this loanword: (51) Jiokot te a=pasa-roa-k ke te trajte-ta not-good 1pl.nom 3sg.acc=spend-vblz-pfv sub 1pl.nom dish-acc ama wo’ota-tua-wa-k munim jitasa joona-po jo’o-wa-me loc throw-caus-imprs-pfv beans that oven-loc make-imprs-rel ‘We spend it so bad that we were forced to throw away the beans on the oven that were prepared by somebody…’ Vocative expressions are also among those elements borrowed commonly from Spanish. Such elements occur with certain frequency in creative or poetic texts. Some of the vocatives are: no ‘no’, ‘not’ <Sp. no; aai! ‘oh!’ <Sp. ¡ay!; aa! ‘oh!’ <Sp. ¡ah!: (52) a. in(to)-te-m-emo=waa-ek? No! and-1pl.nom-lig-rflx=sister-have no ‘…and they sisters among them? No!’ b. ¡Aai! ¡Semalulukut kaa ko’okosi yaapo maisi empo hey hummingbird neg fragile look like 2sg.nom emo yoa’e! rflx tremulous.imp ‘Hey! Hummingbird you don’t look fragile when you move yourself (your wings) tremulous!’
428
Zarina Estrada Fernández and Lilián Guerrero
c. ¡Aa! Bwe’ituk te kaabe-ta-mak etejo-machi!… oh because 1pl.nom none-acc-com talk-seem ‘Oh! Because we are not (sure) to whom he seems to be talking with!’ d. Ah! Caramba, neu eela kaate ume yo-im. ah great 1sg.dat behind be.pl.prs dem.pl yori-pl ‘Oh! Great, the Mexicans are near me!’ Furthermore, in a small collection of procedural discourses, where the preparation of distinct foods is described, the particle ori appears as a hesitation word.10 Such word has been probably borrowed from the Spanish temporal conjunction ora. The alternative forms in Yaqui ori and orita seem to be motivated by their position within the sentence – first or last. Unfortunately, there is no way to demonstrate at this time that ori ~ orita are truly Spanish loans. Examples are provided in (53): (53) a. Ori, Loma Wamochil-po into ket pa~pajko-ria-wa… mmhm Loma Guamúchil-loc and also rdp~-festivity-appl-imprs ‘And, mmhm, in Loma de Guamúchil (the festivities) are also celebrated.’ b. muuni-m ota-k-a-me o kesum-k-a-me… beans-pl bone-have-a-nmlz or cheese-have-a-nmlz paapa-m wakas-ek-a-me orita…, ala! potatoes-pl meat-have-a-nmlz mmhm yes ‘Beans with bone or cheese… potatoes with meat, mmhm, yes…’ The vocative words borrowed from Spanish and illustrated within this section are quite common in Yaqui narratives, although their occurrence varies according to the speaker or the discourse type. Every individual speaker will give a distinct communicative force to the discourse according to their own personal choice or attitude, that is, according to the pragmatic context; all of the vocatives are fully adapted into the phonology of Yaqui.
6. Constituent order Yaqui’s constituent order has received, in general, little or minimal influence from Spanish. However, the conjunction komo illustrated before in (49) seems to have introduced a change in the constituent order within compara-
Yaqui
429
tive clauses involving the Yaqui postposition benasi. The example in (54) illustrates benasi before the adjective teebe ‘tall’, which is the usual position for komo in Spanish:11 (54) aapo nee benasi teebe. 3sg.nom 1sg.acc comp high ‘He is as taller as I am.’ In contrast with the position of benasi in (54), examples in (55) illustrate the regular postnominal order of benasi: (55) a. Ket-kea ili ito aet womta-la-ta benasi… also-only dim rflx 3sg.dir scare-adjvz-acc comp ‘Like if we were scared a little bit (towards) ourselves…’ b. Tua te wepul mampusiam-po benasi… true 1pl.nom one finger-pl-loc comp ‘In fact there was like about only one small amount…’ (lit. like about a finger) The order of benasi in examples in (55), which is the usual position of the conjunction como in Spanish, also demonstrates the influence of this language in introducing a change in the word order (pat) of Yaqui: (56) Benasi t=a ta’a-pea. comp 1pl.nom=3sg.acc know-des ‘Like when we want to know it (the sun).’ A comparative clause where the postposition betchi’ibo is in the wordorder position considered to be the original from Yaqui is given in (57): (57) Jume kuchum che’a su~sua-k-an nee betchi’ibo. dem.pl fish.pl more rdp~think-st-pasc 1sg.acc postp ‘The fish were smarter than me.’
7. Conclusions After four hundred years of Spanish influence on the Yaqui language and culture, the language shows few cases of grammatical borrowing. Language
430
Zarina Estrada Fernández and Lilián Guerrero
contact between Yaqui and Spanish is strong at the lexical level, but almost nil in the grammar. Yaqui has adopted a Nahuatl morphological strategy – the suffix -oa – to adapt Spanish verbs, and the Spanish derivational suffix -ero, adapted as -e’o, -eo or -ero, to derive agentive nouns. Most of the loanwords are discourse particles: conjunctions or subordinating elements. Some of them used as hesitative, hortative or vocative particles, which are typically used at the discourse level. Most of the borrowings are mat-loans. One last interesting contact phenomenon, which can be taken as a true instance of code-switching rather than of borrowing, is the usage of multimorphemic elements. That is, expressions (idioms) which include more than one word. Instances of code-switching are frequently used to give communicative force at the discourse, but they can vary according to both the topic and the speaker. Estrada, Morúa and Álvarez (2005) have illustrated some cases where Spanish expressions are used in narrative texts as a way of introducing communicative force. Good examples of code-switching are the following: (58) a. Pero no es igual, ian tajti bea ne inim weama. now dir then 1sg.nom here be-prs ‘… but it is not the same, I’m still around.’ b. Si nesio… ju’u yoi jodido! int necio det yori jodido12 ‘How stubborn… the damn white-man!’ c. soldado a huevo… soldado a huevo ‘forced to be a soldier…’ d. Bwe tua nee chingo-k ommme…! but int 1sg.acc chingar-pfv man ‘But, in fact, I fuck myself!’
Abbreviations acc adjvz agt appl caus com comp cond
accusative adjectivizer agentive applicative causative comitative comparative conditional
dat dem des det dim dir fut imp
dative demonstrative desiderative determiner diminutive directional future imperative
Yaqui imprs int loc neg nmlz nom obl pfv pl prs
impersonal intensive locative negative nominalizer nominative oblique perfective plural present
pasc postp rdp rflx rel sg st sub vblz
431
past continuous postposition reduplication reflexive relativizer singular stative subordinator verbalizer
Notes 1. We are grateful to the Max Planck Institute for Evolutionary Anthropology, in particular to Bernard Comrie, for hosting Estrada during the summers of 2006 and 2007, making it possible for us to carry this work; we also thank Fred Field for his useful comments on an earlier version of this chapter. 2. There are several articles dealing with the influence of Spanish on the structure of Yaqui: Dozier (1964), Johnson (1943), Lindenfeld (1982); two are mainly dealing with the Yaqui from Arizona (US) only Johnson’s study deals with the Yaqui spoken in Sonora (Mexico). 3. ISO 6393: yaq. 4. Crescencio Buitimea Valenzuela, who has coauthored several studies with Estrada, is a native speaker of Yaqui born in Vícam, Sonora. He is now preparing a Handbook dictionary of the Yaqui language and several books for teaching the language. 5. Estrada and Silva (2006) deal with some properties of the Pascolas’s discourse, a kind of religious discourse, which is used among the Yaqui in their festivities. The Pascola is a Yaqui man who performs the role of an anti-religious person, a clown. We want to express our gratitude to Manuel Carlos Silva Encinas for his generous guidance and for sharing with us his discourse materials in Yaqui. 6. The similarities among the phonological inventory of Yaqui and Spanish is due to a mere coincidence. See Voegelin, Voegelin and Hale (1962) as well as Whorf (1935) for studies on the reconstruction of Proto-Uto-Aztecan. Throughout this work we are using the orthography approved by the Yaqui from Sonora, Mexico. 7. Escalante (1988) provided a brief account about the adaptation of loanwords into the Yaqui from Pascua. Estrada (2005) presented a detailed analysis of the phonological adaptation of Spanish loanwords into Yaqui. 8. Tomas Basilio’s grammar, deals with Tehuelco, an extinct Cahitan dialect. 9. Chablé’s field notes are centered on the Huichol variety spoken at La Palmita, municipality of Mezquitic, Jalisco. Chablé is working on her Master thesis.
432
Zarina Estrada Fernández and Lilián Guerrero
10. A procedural discourse has been described as an activity-oriented discourse, where a series of activities, which are done by somebody, are described in a chronological order. 11. Although Lindenfeld (1971: 89) however, has considered those constructions as native from Yaqui. 12. In Spanish: ‘¡Qué necio… el yori jodido!’
References Buelna, Eustaquio 1989 [1890]. Arte de la lengua cahita por un padre de la Compañía de Jesús. México: Siglo XXI Editores. Dedrick, John M., and Eugene H. Casad 1999 Sonora Yaqui language structures. Tucson: The University of Arizona Press. Dozier, Edward P. 1964 Two examples of linguistic acculturation: The Yaqui of Sonora and Arizona and the Tewa of New Mexico. Language 32 (1): 146157. Escalante, Fernando 1988 Spanish loanwords in Yaqui. Work presented in CAIL–American Anthropology Association. Phoenix, Az. Estrada Fernández, Zarina 2005 Spanish loanwords in Yaqui. (Uto-Aztecan language from Northwest Mexico). Paper presented at the Fifth Workshop on Loanword Typology. 67 June, Leipzig, Germany. Estrada Fernández, Zarina, and Crescencio Buitimea Valenzuela Forthc. Yaqui de Sonora. Archivo de lenguas indígenas de México. México: El Colegio de México. Estrada Fernández, Zarina, Crescencio Buitimea Valenzuela, Adriana E. Gurrola Camacho, María Elena Castillo Celaya, and Anabela Carlón Flores 2004 Diccionario yaqui-español y textos: Obra de preservación lingüística. México: Plaza y Valdés. Estrada Fernández, Zarina, María del Carmen Morúa Leyva, and Albert Álvarez González 2005 Actitudes lingüísticas y contacto entre lenguas: Préstamos del español en yaqui. XIV Congreso Internacional de la Asociación de Lingüística y Filología de América Latina. 1923 Oct., Monterrey, Nuevo León. Estrada Fernández, Zarina, and Manuel Carlos Silva Encinas 2006 El discurso de los pascolas entre los yaqui de Sonora, México. Friends of Uto-Aztecan Conference (FUAC), Salt Lake City, UT. 2326 August.
Yaqui
433
Johnson, Jean Bassett 1943 A clear case of linguistic acculturation. American Anthropologist 45 (3/1): 427434. Karttunen, Frances 1984 An Analytical Dictionary of Nahuatl. Texas Linguistics Series. Austin: University of Texas Press. Karttunen, Frances, and James Lockhart 1976 Nahuatl in the Middle Years: Language Contact Phenomena in Texts of the Colonial Period. Berkeley: University of California Publications. Linguistics 85. Lindenfeld, Jacqueline 1971 Semantic categorization as a deterrent to grammatical borrowing: A Yaqui example. International Journal of American Linguistics 37 (1): 614. 1982 Langues en contact: Le yaqui face a l’espagnol. La Linguistique 18 (1): 111127. Matras, Yaron, and Jeanette Sakel 2007 Investigating the mechanisms of pattern replication in language convergence. Studies in Language 31 (4): 829865. Mushin, Ilana 2001 Evidentiality and Epistemological Estance. Narrative Retelling. Amsterdam: John Benjamins. Voegelin, Carl F., Florence M. Voegelin, and Kenneth L. Hale 1962 Typological and comparative grammar of Uto-Aztecan I (Phonology). Publications in Anthropology and Linguistics, Memoir 17 of the International Journal of American Linguistics. Indiana University. Whorf, Benjamin L. 1935 The comparative linguistics of Uto-Aztecan. American Anthropologist 37: 600608.
The case of Otomi: A contribution to grammatical borrowing in cross-linguistic perspective Ewald Hekking and Dik Bakker
1. Background In this chapter we describe the Spanish grammatical and lexical borrowings in Otomi, a language from Central Mexico. Otomi is a member of the Otomanguean family, which according to Ruhlen (1991: 37) belongs together with Uto-Aztecan and Tanoan to the Central Amerindian stock. Otomi, which constitutes together with Mazahua, Ocuilteca, Matlatzinca, Pame and Chichimeca the Otopame group within Otomanguean, is 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. We discuss two Otomi dialects, viz. the dialect of Santiago Mexquititlán in the municipality of Amealco situated in the southern part of the state of Querétaro, and the dialect of San Miguel in the municipality of Tolimán situated in the northern part of the same state. Santiago Mexquititlán is a village with a population of around 15,000 inhabitants in the mountains of Mexico’s neovolcanic axis. San Miguel is a village with a population of around 670 inhabitants in the semidesert of the Sierra Madre Oriental. In both villages the vast majority of the population are ethnic Otomis and both Otomi dialects belong to north-western Otomi, one of the larger variants of Otomi with around 33,000 speakers in total. The Otomi dialect spoken in Santiago Mexquititlán is rather similar to the Otomi dialect of the villages in the north of the state of Mexico. The dialect spoken in Tolimán is similar to the Otomi dialect of the Valle del Mezquital in the state of Hidalgo. Our description is mainly based upon a corpus of around 110,000 tokens, collected during fieldwork between 1993 and 2004, and the result of interviewing a total of 115 respondents. In this section we will give a short historical sketch of the Otomi language community. The following sections will discuss a number of borrowing phenomena on the respective levels of linguistic description. Like the majority of the larger language groups in Latin America, Otomi has a rather long contact history with Spanish. From around the year 1500
436
Ewald Hekking and Dik Bakker
onwards there has been contact between the two languages. As a result Otomi has undergone pervasive influence from that European language. It was the native language of the original inhabitants of the Valley of Mexico and the surrounding valleys. Throughout history its speakers had to confront Aztecs, Spaniards and Mestizos, speakers of Nahuatl and Spanish. Both languages belong to other language families, Uto-Aztecan and Indo-European respectively. Since the Otomis had to surrender to the Nahuas in the fifteenth century, there has been a very close contact between the Otomi and Nahuatl languages. During that contact the Nahuas developed a very negative image of the Otomis, which later was passed on to the colonial chroniclers, such as Sahagún (1999), whose Nahuatl speaking informants considered the Otomis “toscos e inhábiles” (coarse and unskillful). The word Otomi is probably a derivation of the Nahuatl word tOtomitl ‘birdhunter’ (Jiménez Moreno 1939). The Otomis themselves prefer to call their language Hñäñho, Hñöñhö or Hñähñu, and themselves Ñäñho, Ñöñhö or Ñähñu. In these names, the morpheme ñä and its variant ñö mean ‘speak’. The morpheme -ñho is probably a derivation of the adverb hño ‘well’. The morpheme h- marks the impersonal or passive voice (Hekking 1995). The Otomis have also been in contact with the Mazahuas, with whom they had a relation of equality, and the Chichimecs, in comparison with whom the Otomis probably felt superior. In this connection it is interesting to mention that the Otomis from Tolimán claim that their forbears originally spoke Chichimeca. This could mean that in the Otomi spoken in that community Chichimeca substrate might be found. As the Otomis were the second most numerous group after the Nahuas on the Mexican highlands, the Spaniards were very much interested in their conversion. Although their language was considered to be very difficult because of the fact that Otomi has considerably more vowels and consonants than Spanish, a spelling system for Otomi was developed, as well as vocabularies and grammars. Catechisms and legal documents were written in Otomi at some scale. Especially missionary friars of the Franciscan order have studied the Otomi language, such as Fray Alonso Urbano ([1605] 1990). The colonial documents written in Otomi are not easily accessible, since the authors do not always distinguish the Otomi phonemes that have no corresponding element in Spanish, especially among the vowels. After the independence of Mexico in 1813 the indigenous groups were officially no longer recognized as such and lost most of the status implied by that. As a result, many Otomis could no longer afford their education and Otomi stopped being used by the civil authorities. Only a handful of scholars continued learning and studying the language. It was in the nineteenth cen-
Otomi
437
tury that a process of language shift started. The Mexican Revolution (1911– 1917) did not lead to social change for the Otomi population, nor did it foster recognition for their language, and stop language shift. On the contrary, after a long history in which the Otomis had been degraded socially, they now belong to the lowest social levels of the Mexican society. They live in the most remote and less fertile places on the highlands from an agriculture of subsistence, reason for many of them to emigrate to the big cities, such as Mexico City, Guadalajara and Monterrey. In the twentieth century several attempts have been made to integrate the indigenous peoples in the national community by means of the officially called Educación Bilingüe. This is taken care of by indigenous teachers with a very negative attitude towards their own roots and a complete lack of knowledge about bilingual education. As a consequence, most Otomis are illiterate in their first language and very often have insufficient command of the standard variety of Mexican Spanish. Otomi 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 Otomis and the Spanish speaking world of the Mestizos has increased considerably. As a result, a rapid increase in contact phenomena from Spanish may be observed (Hekking 1995, 2001, 2002; Hekking and Bakker 1998a, 1998b, 2005; Hekking and Muysken 1995). Because of the fact that Otomi is a stigmatized language, only spoken by poor and traditional people, many Otomis do not want to convey the indigenous language to their children any longer. So far the historical background of the Otomi language community. The rest of this contribution is organized as follows. In Section 2 we describe how parts of the Otomi phonological system have changed as a result of contact with Spanish. In Section 3 we will see how Otomi has changed on a number of typological parameters. In Sections 4, 5 and 6 we have a closer look at the borrowing in nominal and verbal structures and with respect to grammatical elements. In Sections 7, 8 and 9 we will discuss how Otomi constituent order, syntax and lexicon have been affected by language contact. Finally, in Section 10 we will draw some conclusions.
2. Phonology When Spanish loanwords are inserted in an Otomi utterance, they may adjust in different degrees to the phonological patterns of the target language. The
438
Ewald Hekking and Dik Bakker
degree of phonological integration of the Spanish borrowings depends on the age of the loanword and on the degree of bilingualism of the speaker. The following adaptations of Spanish borrowings to the phonological system of Otomi were observed. a. The Spanish open central vowel a (ä in the spelling system of Hekking 2002) tends to nasalize in syllables that start with the nasal consonants m, n or ñ: animä (< anima); apenä (< apenas); ngañä (< engaña)1 b. Sibilants and voiced and unvoiced plosives at the beginning of a syllable tend to nasalize: mbakuna (
Otomi
439
l. axo (< ajo), paxa (< paja), Xuwa (< Juan) are examples of old loanwords, the palatals of which are pronounced as fricative palatals in modern Spanish Apart from these adaptations, new sounds are introduced into Otomi as well. As a result of the incorporation of unassimilated Spanish loanwords, several new segments have been added to the Otomi repertoire. Examples are the trilled alveolar vibrant rr in loanwords such as burru (< burro) and surru (< zorro), as well as the lateral l in loanwords like ladriyo (< ladrillo); lado; biskleta (< bicicleta); bwelo (< abuelo). The lateral now also appears in native items, such as lele ‘baby’. Through the adoption of Spanish and to a much lesser extent also Nahuatl loanwords the affricated alveopalatal may have been introduced in Otomi, as in chachalaka (< chachalaca (= type of bird)); chikiuite (< chiquiuitl (= basket)); chaketa (< chaqueta); chofe (< chófer) (cf. Hernández Cruz, Torquemada, and Sinclair Crawford 2004). As a possible result affricates are also palatalized in native forms, e.g. in Santiago Mexquititlán we found tx’aki instead of ts’aki (Hekking and Andrés de Jesús forthc). Furthermore we find consonant clusters which are typical for Spanish but unknown in classical Otomi, as in ektarya and septyembre. Other patterns unknown in the phonological system of Otomi are found in loanwords as well, such as syllable final consonants, and unstressed first syllables, as in prisidente and kampesinu. Sofar we have no evidence that Otomi vowel and consonant harmony, tone and intonation have been affected in any distinctive way by the contact with Spanish.
3. Typology In this section we will discuss how Otomi is changing on three typological parameters: constituent order, morphological structure and its parts-ofspeech system. As in the other Otomanguean languages, VOS and VS are the basic constituent orders at the clause level in classical Otomi (Suárez 1983; Yasugi 1995). In today’s language SVO order is frequently used (Lastra de Suárez 1994; Hekking 1995), although VS order is still very common, as is shown in example (1) (from Salinas Pedraza 1983: 105 (521)).
440
Ewald Hekking and Dik Bakker
(1)
Nubye dä meengä rä ngu nu’ä rä ’ñeei. now fut.3 go.back pos.3 house that def.sg witch ‘Now the witch will go home again.’
From a morphological point of view, classical Otomi has a rather complicated synthetic structure on the lower syntactic levels, more specifically in the noun phrase (NP) and the verb phrase (VP). At the sentence level the structure is more analytical, and it is not uncommon to find asyndetic compounding and the bare juxtaposition of constituents, with very few explicit markers of the semantic or syntactic relations, such as adpositions, coordinators and subordinators between the constituents. As a result the meaning at the clause level must often be deduced from the meaning of the main verb and the context. Also possession is marked via possessive pronouns rather than via an adposition, as shown in example (2) (from Hekking and Andrés de Jesús, forthc.). (2)
Di ’weti ár zexjo ar Xuwa. pres.1 sew pos.3 pants def.sg Juan ‘I’m sewing Juan’s pants.’
Classical Otomi uses verbal suffixes to express additional participants in the action expressed by the main verb, such as the suffix -wi for company and -bi for beneficiaries. For other functions the language has a small amount of particles at its disposal: mainly dige, ir nge, nguu and ja. Historically, these had mainly verbal functions, but nowadays they resemble Spanish prepositions in the sense that they mark objects of comparison, instrument, cause, manner and spatial orientation. This is exemplified in (3) (from Hekking and Andrés de Jesús fc). (3)
Poni ndunthe ar ngunt’uzaa come out much def.sg dust-wood ho thets’i ya zaa ir nge ar thegi. where cut def.pl wood instr def.sg saw ‘You will find much sawdust where they cut wood with a saw.’
In order to combine two or more clauses into a compound sentence, classical Otomi typically uses juxtaposition, not only for simple conjunction but also in the case of final clauses and the representation of (indirect) discourse. For relative clauses, which are always postnominal in Otomi, the gapping
Otomi
441
strategy is used, without any form of connection via a relative pronoun or a subordinator.2 All these functional gaps may have motivated Otomi speakers to adopt many Spanish prepositions, subordinators, coordinators and relative pronouns. Possibly as a side effect of this, the verbal suffixes which mark a relation between the predicate and certain adjuncts in classical Otomi are disappearing from the language (Hekking 1995: 155161). It may therefore well be the case that, as a result of intensive contact with Spanish and the ensuing high degree of bilingualism, Otomi is becoming a language with less asyndetic connecting at the clausal level and more analytical structuring on the lower syntactic levels. Let us now have a closer look at the parts-of-speech system of Otomi. Otomi could be characterized as a rigid language in the sense that the vast majority of the major lexical items fall in one of two clearly separate classes, i.e. nouns and verbs. Lexical elements of these two classes can be easily distinguished by their specific positions in the syntax and by the different types of proclitics and affixes that accompany them and are never found on the other class. Among linguists with expertise in Otomi there is no general agreement as to whether Otomi has a third and fourth class of lexemes, i.e. adjectives and adverbs. Many concepts which in Spanish and English are expressed by way of an adjective are expressed by way of a noun or a verb in Otomi. E.g. ra’yo ‘new’, dätä ‘big’, nduxte ‘naugthy’, ‘wild’, gunt’ei ‘jealous’ and ngoñä ‘bald’ are nouns, which are preceded by an article, as is shown in example (4). (4)
a. D-ar nduxte. pres.1-def.sg naughty ‘I am naughty.’ b. G-ar nduxte. pres.2-def.sg naugty ‘You are naughty.’
Other lexemes, such as dathi ‘ill’, txutx’ulo ‘small’, and johya ‘content, happy’ are verbs in Otomi, and get the usual verbal proclitics, as is shown in example (5). (5)
a. Di=johya. pres.1=be.happy ‘I am happy’
442
Ewald Hekking and Dik Bakker
b. Di=johya-he. pres.1=be.happy-pl.excl ‘We are happy’ A third group of lexemes, which are translated into English as adjectives, such as ‘thin’, ‘fat’, ‘bitter’, ‘sweet’, ‘cold’, ‘warm’, ‘yellow’, ‘red’, ‘beautiful’, ‘ugly’, ‘good’, ‘high’, and ‘low’ have the verbal suffixes -gi, -’i and -Ø to mark first-, second- and third-person (in)direct object and are preceded by the proclitic xi= that marks the third-person perfect in predicative function. As is shown in example (6), these constructions could be analyzed as impersonal constructions, i.e. ‘it is thin to me’. (6)
a. Xi=nts’ut’i-gi. prf.3=thin-1.obj ‘I am thin.’ b. Xi=nts’ut’i-’i. prf.3=thin-2.obj ‘You are thin.’ c. Xi=nts’ut’i-Ø. prf.3=thin-3.obj ‘He is thin.’
The suffixes used by this third group suggest that these lexemes are a kind of (stative) verbs. However, many linguists, notably Ecker; Hess; Voigtlander and Echegoyen; Hekking and Andrés de Jesús; Lastra de Suárez; Andrews and Bartholomew, although acknowledging that Otomi has less adjectives than for example Spanish and English, treat this group as adjectives, since they may be used adnominally. This is shown in example (7). (7)
a. ar hets’i ’ñoho def.sg tall man ‘the tall man’ b. ar nzatho ’behñä def.sg pretty woman ‘the pretty woman’ c. ar ts’ut’i nxutsi def.sg slim girl ‘the slim girl’
Otomi
443
On the other hand Palancar (forthc.) claims on the basis of his data from the Otomi spoken in San Ildefonso, in the state of Querétaro, that all lexemes which denote property concepts in Otomi are encoded as verbs and nouns and not as adjectives. He argues that examples like (7) should be analyzed morphologically, as nominal compounds, rather than syntactically, as nominal modification. As we will show in Section 4, in the corpora we have collected in Santiago Mexquititlán and Tolimán very few borrowed Spanish adjectives were attested indeed and the majority of the Spanish adjectives we did find were used predicatively, functioning either as a verb or as a noun. This borrowing behaviour of Otomi with respect to Spanish gives structural support to the view that adjectives are not a regular part of speech of this language. However, we suspect that Otomi via its contact with Spanish, a language with a multitude of adjectives, might in fact be in the process of developing adjectives as a new category. This would then be on the basis of an existing verbal subclass, i.e. transitive stative verbs. It would eventually lead to a typological shift with respect to the parts-of-speech system. Concepts which in Spanish and English are typically expressed by way of adverbs, in Otomi are also expressed by stative verbs, just like some adjective-like elements. On the basis of the derivational processes involved in this, two groups may be distinguished, exemplified by tihi ‘to run’ → ’nihi ‘quickly’, and by xi hño ‘is good’ → xi hño ‘well’, respectively.
4. Nominal structures Of the 100,541 tokens in our corpus 15,571 (14.1%) are Spanish borrowings. Just over half of these belong to a lexical category (noun, verb, adjective or adverb).3 These percentages are more or less similar for both dialects we studied. Table 1 gives the percentages were found for the four parts of speech. Since such figures may not say much on themselves, we have added the figures based on comparable data for borrowings from Spanish in two other Amerindian languages, Quechua and Guaraní. What strikes most is that Otomi seems to borrow much less lexical material, and then mainly nouns, and hardly any adjectives. Spanish nouns are almost always borrowed in their singular form, and may be accompanied by any Otomi nominal morphology, both proclitics and suffixes. This is illustrated in example (8).
444
Ewald Hekking and Dik Bakker
Table 1. Lexical parts of speech borrowed Percentages borrowed Part of Speech
Otomi
Quechua
Guaraní
Noun Verb Adverb Adjective Total
40.7 4.8 4.5 1.9 51.9
54.0 17.7 3.4 8.5 83.6
37.2 18.3 2.3 7.4 65.2
(8)
Nu’bya ya nguu ya losa-’bya, ya teja, today def.pl house def.pl concrete-act def.pl tile ya laminä yá njo’mi ya nguu. def.pl metalic.plate pos.pl roof def.pl house ‘Nowadays the houses are made of flagstone, and the roofs of the houses are made of tiles and metallic plates’
As we will see in Section 6, almost half of the borrowed grammatical material consists of Spanish prepositions. Regularly, in those contexts, Otomi particles are suppressed which were traditionally used to express the corresponding relation now taken over by the borrowed preposition. In that sense we could say that the little case marking that was present in the language is being replaced by the more analytical adpositional strategy on the basis of Spanish loans, notably to mark instrument, cause, manner and spatial orientation. Apart from this huge amount of contact-induced changes in relation marking we have also found a few examples of the introduction of gender marking. Otomi does not mark gender, but in some native words it does as a result of contact with Spanish. This is shown in example (9). The Otomi form is ’beto ‘grandchild’, which is unmarked for gender. In (2), the ‘o’ in the stem seems to be confused with the Spanish suffix that marks masculine (-o) and following the Spanish rule the Spanish feminine suffix (-a) is substituted for the -o in Otomi when the speaker refers to a granddaughter. (9)
’beto ’beta ‘grandson’ ‘granddaughter’
Over half of the borrowed Spanish adjectives (55%) we find in their canonical function of modifier of the head noun in a noun phrase. In that function they
Otomi
445
are accompanied by the usual Otomi nominal morphology. Occasionally, we find them in the position of the head noun (4%). In the rest of the cases, they occupy the position of the main verb of the sentence.
5. Verbal structures Borrowed Spanish verbs are typically treated as verbs in Otomi. The usual form is the infinitive without the final -r of Spanish: engaña (< engañar); eskohe (< escoger); eskribi (< escribir). Just like Otomi verbs, Spanish stems may be accompanied by the usual Otomi verbal proclitics and case marking suffixes. An example is given in (10). (10) a. Ntonse nu-r txuku tobe. then dem-def.sg dog still b. Mi=molesta-tho nuya kolmenä. imprf.3=bother-lim dem.3pl beehive ‘Then the dog still bothered the bees.’ Spanish verbs with vowel change and diphthongization generally keep their diphthongs ‘ie’ and ‘ue’ (cf. kwenta in 11a), although not always (konta in 11b). (11) a. Nu-ge bwelo-ga mi=kwenta. dem-dem grandfather-emph.1 past.3=tell ke hä mi=ting-ar oro ’ne-r plata. that yes imprf.3=find-def.sg gold also-def.sg silver ‘My grandfather used to tell that he used to find gold and also silver.’ b. Bi=ma bi=konta-wi yá kompañera past.3=go past.3=tell-incl.du pos.3 friend nu-’u ot’=ar wela. dem-3pl do=def.sg grandmother ‘She visited her girl-friends to tell them about the things done by her grandmother.’ Arguably as a result of language contact, Otomi is losing some of its morphological distinctions on its verbs. A striking one is the loss of the inclusive/exclusive distinction in first person dual and plural agreement marking,
446
Ewald Hekking and Dik Bakker
as in (12a). It seems to be replaced by the Spanish preposition ko (< con) ‘with’ (12b). (12) a. Xta=e-’be ma mpädi. prf.1=come-du.excl pos.1 friend b. Xta=ehe ko ma mpädi. prf.1=come with pos.1 friend ‘I have come with my friend.’ The same seems to be happening to several other verbal markers of classical Otomi, such as the suffixes -wi and -hu, which mark the company in the dual and the plural. They are also giving way to the borrowed preposition ko, as shown in (13) and (14) below. (13) a. Ar Mändo mi=ñä-wi ár nänä. def.sg Armando past.3=speak-com.du pos.3 mother b. Ar Mändo mi=ñä ko ár nänä. def.sg Armando past.3=speak with pos.3 mother ‘Armando spoke with his mother.’ (14) a. Ya nxutsi ’ñeñ-hu yá ida. def.pl girl play-com.pl pos.3 sister b. Ya nxutsi ’ñeni ko yá ida. def.pl girl play with pos.3 sister ‘The girls play with their sisters.’ This phenomenon has not only been found in Querétaro, but also in other dialects (cf. Lastra de Suárez 1992; 1997). We find it remarkable that relatively few verbs are borrowed. As was shown in Table 1, their figure is significantly lower than for Quechua and Guaraní, and can not, therefore, be a pure reflection of the frequencies of the different parts of speech in spoken Spanish. We have no ready made explanation for this other than assuming that Otomi is a nominally oriented language as opposed to especially Guaraní, which seems to be much more verbally oriented. Among the adverbs being borrowed we find different types: manner adverbs, locative and temporal ones, etcetera. They are typically functioning at the level of their use in Spanish. Thus, borrowed sentential and verb phrase modifiers are found to operate at that level in Otomi as well. Spanish adverbs
Otomi
447
which are derived productively from adjectives with the suffix -mente are usually borrowed in that form, not in the plain adjectival form.
6. Other parts of speech As has already been indicated above, we found a remarkable number of nonlexical loanwords from Spanish in Otomi. Of the 15,571 Spanish tokens in our corpus, 48.1 percent are function words in Spanish. The figures for the most frequently observed ones are found in Table 2. For contrast, we add again the figures we found for Quechua and Guaraní. Where borrowed lexical elements may be seen as fillers of semantic gaps in the lexicon which do not directly affect the structure of the grammar of the target language, the borrowing of grammatical elements may, or must have implications for its morphosyntax. We will have a closer look at some of the more striking cases.
6.1. Prepositions What strikes most is the considerable amount of prepositions from Spanish we found in our corpus, especially when we compare this with the very few we found in Quechua and Guaraní. After the noun this is the category most often borrowed. Among the 2,075 tokens there were 54 different types, a considerable part of the complete inventory of Spanish. In classical Otomi Table 2. Grammatical parts of speech borrowed Percentages borrowed Part of Speech
Otomi
Quechua
Guaraní
Preposition Coordinator Discourse marker Subordinator Pronoun Other Total
21.2 7.5 6.5 6.1 0.6 6.2 48.1
0.5 6.9 0.6 0.8 0.1 7.5 16.4
0.5 4.4 0.8 4.6 0.2 24.3 34.8
448
Ewald Hekking and Dik Bakker
there exist a small number of verbal affixes with preposition-like functions. However, in general semantic functions are not often marked explicitly and have to be deduced from the context. Borrowed Spanish prepositions make a number of functions transparent. The following cases were observed most frequently in our corpus. a. The Spanish preposition pa (< para) is often used together with or instead of the verbal suffixes -pi or -wi to mark the beneficiary. (15) Nä-r hyokunguu bi=hyok-wi dem-def.sg architect past.3=build-ben ’nar nguu pa-r ts’ut’ubi. indef.sg house for-def.sg governor ‘The architect built a house for the governor.’ Note that the definite article cliticizes to the Spanish preposition. b. The Spanish preposition ko (< con) is frequently found marking the company instead of the equivalent verbal suffix. Also in this case, double marking is common. (16) Mnde ngi=’ño-hu ko hñu ya nxutsi. yesterday past.2=walk-incl with three def.pl girl ‘Yesterday you walked with three girls.’ The same Spanish preposition is also used to mark the instrumental. Classical Otomi would employ the particle ir nge, which would take the same syntactic position as the preposition. It could be said that a grammaticalization process is under way which might eventually transform this particle, and several others in functional prepositions. (17) Ar jä’i bi=dak=ar k’eñä kon minge=r. def.sg man past.3=attack=def.sg snake with pickaxe=def.sg ‘The man attacked the snake with the pickaxe.’ c. In classical Otomi the particle ngu is used to mark the object of equation. Nowadays the Spanish preposition komo (< como) is often used instead. Very common is the use of the fused double marking komo-ngu, as in example (18).
Otomi
449
(18) Xi mi=txinga mi=mpefi much past3=work.o.s.to.death past3=work komongu ’nar meti. like indef.sg animal ‘They worked themselves to death working like an animal.’ d. In order to mark purpose, typically unmarked in Otomi, the Spanish preposition pa (< para) is often used, as shown in example (19). (19) Thoku ’nar pont’i zaa pa da t’exu ja-r ’met’e. make indef.sg cross wood for fut.3 put.on loc-def.sg roof ‘They make a wooden cross to put it on the roof.’ e. In order to mark the privative, the Spanish complex sinke (i.e. preposition sin plus the subordinator que) is frequently used, as in example (20). The standard way in Otomi employs the verb otho, which means ‘there is not’. (20) Ya nxutsi xi=mboni sinke ar nänä. def.pl girl prf=leave without def.sg mother ‘The girls have left without their mother.’ Also the Spanish prepositional construction embesde (< en vez de) is often used in such contexts. (21) Embesde-r k’ani nu’bya tam-’bya t’afi, instead.of-def.sg vegetable now buy-act sweet ya gayeta ’neh=ya refresko. def.sg biscuit also=def.sg soft drink ‘Instead of vegetables they now buy sweets, biscuits and also soft drinks.’ f. All kinds of locative relations, which are typically unmarked in classical Otomi, get Spanish prepositions such as entre ‘between’; pa (< para) ‘in the direction of’; desde ‘from’; asta (< hasta) ‘until’; a ‘to’; and de ‘from’. (22) Ya dá=pengi de Jalpa. already past.1=come.back from Jalpan ‘I have already come back from Jalpan.’
450
Ewald Hekking and Dik Bakker
(23) Bi=dexu asta mñä dige ja-r zaa. past.3=climb till on.the.top ref loc-def.sg tree ‘He climbed till the top of the tree.’ g. In classical Otomi possession is expressed by simple juxtaposition of possessed and possessor. As a result of contact with Spanish, this relationship is now often made explicit via the preposition de ‘of’. (24) Nixi Independensya nixi Reforma nixi Rebolusyon bi=nkambyo Neither Independence nor Reforma nor Revolution past.3=change yá kostumbre de ya ñäñho. pos.3 habit of def.pl Otomi ‘Neither the Independence, nor the Reform, nor the Revolution have changed the habits of the Otomis.’ h. The Spanish preposition de is also used to mark the partitive, and all kinds of relationships which may be subsumed under ‘Reference’, above all in the Tolimán dialect. In classical Otomi there are no markers for the relations concerned. In (25)–(27) we give three examples. (25) ’Na de ge’u i=ndude kaha. one of dem.3 pres.3=carry box ‘One of them carries the box.’ (26) Di=ñä-wi de byaje pa Maxei. pres.1=speak-incl.du ref trip to Queretaro ‘We talk about the trip to Queretaro.’ (27) Hoku ’nar krusi de ’nar xithe. Make indef.sg cross ref indef.sg wood ‘He makes a cross of wood.’ As shown in example (28) below, the Spanish preposition ko (< con) ‘with’ is also used to mark the ‘made of’ function. (28) Ya tsita Nt’okwä xi=thoki ko=r yeso. indef.sg saint San.Ildefonso prf.3=made.of with=indef.sg gesso ‘The saints from San Ildefonso are made of gesso.’
Otomi
451
Table 3. Spanish coordinators in the corpus Coordinator
Total tokens borrowed (speakers)
Coordination: i Adversative: pero Disjunction: o Additional negative: ni Contrastive: sino
200 (49) 188 (55) 188 (45) 89 (55) 30 (19)
6.2. Coordinators The second most frequently borrowed grammatical category is Spanish coordinators. In our two dialects we have found the types set out in Table 3. We give the number of tokens in the corpus. The figures are roughly equivalent for the two dialects. In brackets we give the number of speakers on the total of 115 that used the respective forms at least once. An example with the adversative in given in (29). (29) Hin=di pä-ka ko=r hñäñho, neg=pres.1 know-emph.1 with=def.sg Otomi pero ko=r hñämfo hä di=pä-ka. but with-def.sg Spanish yes pres.1=know-emph.1 ‘I don’t know in Otomi, but in Spanish I do know it.’
6.3. Discourse markers The third most important category of Spanish borrowings are discourse markers. We found the following Spanish elements quite frequently in paragraph initial position: pos / pwes (< pues) ‘well’ (372 tokens; 75 speakers); and este (204 tokens; 46 speakers). The latter is the masculine singular of the Spanish proximate demonstrative, which is clearly used by Otomi speakers as turn holder and hesitation marker. It is very characteristic for Mexican Spanish in the same function. Hekking and Bakker (1998) suggest that these elements give a Spanish flavour to Otomi utterances, and possibly also more status to the speaker. Following Matras (1998) we might assume that, in fact, the discourse structure of Otomi is converging towards that of Spanish discourse in situations where Spanish is supposed to be the language with the higher status. This is shown in examples (30) and (31).
452
Ewald Hekking and Dik Bakker
(30) Temu gi=mä-nge? What pres.2=say-emph.2 Pwes nuga di=mä-nga gatho ar za. well I pres.1=say-emph.1 everything def.sg good ‘“What do you think?” “Well, I think everything is okay.”’ (31) Ar Xuwa bi … este … bí=hñuxu ’nar he’mi def.sg Juan past.3 well past.3=write indef letter pa bi=mända-wi ár mpädi Enrike. for past.3=send-incl.du pos.3 friend Enrique ‘Juan … well … wrote a letter in order to send it to his friend Henrique.’
6.4. Subordinators The next class of grammatical borrowings we discuss are subordinators. Just like nominal relations, in classical Otomi subordination is often unmarked. We found a host of Spanish subordinators in the initial position of subordinate clauses. Many Spanish subordinators consist of a preposition plus subordinator que ‘that’. Examples are por que ‘because’; para que ‘so that’; ya que ‘since’; and hasta que ‘until’. Interestingly, some of these are borrowed with the subordinator (porke; yake), while in other cases only the prepositional part is borrowed (pa; hasta). The most frequently appearing borrowing in this function is pa (697 tokens used by no less than 104 speakers). Pa (< para) is also the most frequently appearing Spanish element in its prepositional function. It is not clear to us whether pa in its subordinating function should be seen as a short form of para que or whether we are dealing with the same element pa in two different syntactic functions. Other subordinators which appear with some regularity are komo (< como) ‘how’, kwando (< cuando) ‘when’, and anke (< aunque) ‘although’. We found quite a few others be it with low frequencies and for a limited number of speakers. Below we give some examples of the use of Spanish subordinators in Otomi utterances. (32) Necesita da nuya jä’i da hñunta need fut.3 dem.pl person fut.3 get.together pa da hoku ’nar mehe. for fut.3 build indef.sg well ‘These people must get together in order to build a well.’
Otomi
453
(33) No=r bätsi bi=nzoni dem=def child past.3=cry porke bi=n-tsät’i na nts’edi-tho. because past.3=refl-burn very strong-lim ‘The child cried, because it burned itself very severely.’
6.5. Pronouns There are relatively many Spanish pronouns in our data, especially when compared to the other two languages, which hardly borrow them at all. About half of the ones we found are relative pronouns. The classical Otomi way of relativizing is the gap strategy, i.e. apart from agreement markers there is no form which represents the antecedent in the relative clause, as in (34a) below. However, it is not uncommon in our corpus to find the Spanish subordinator que ‘that’ in the first position of a relative clause. This is shown in example (34b). Possibly as a result of loanshift, we may now also find relative clauses which start with Otomi demonstratives, as in (34c), or with the Otomi interrogative pronoun, as in (34d).4 (34) a. Nä-r jä’i [xi=xi-ku-ga-nu], dem-def.sg person [prf.3=say-obj.1-emph.1-emph.3 m=tiyo-ga. pos.1-uncle-1.emph b. Nä-r jä’i [ke xi=xi-ki], dem-def.sg person [that prf.3=say-obj.1 ge m-tiyo-ga-nu. dem pos.1-uncle-emph.1-emph.3 c. Ar jä’i [nä’ä bi=xi-ku-ga nuna], def.sg person [this past.3=say-obj.1-emph.1 emph.1 ge m-tiyo-ga. dem pos.1-uncle-emph.1 d. Nä’ä-r jä’i [to bi=xi-ku-ga], dem-def.sg person [who pst.3=say-obj.1-emph.1 nä’ä ma tiyo-ga-’ä. dem pos.1 uncle-emph.1-emph.3 ‘The person who has told me that, is my uncle.’
454
Ewald Hekking and Dik Bakker
We find a Spanish construction also in headless relatives, as in (35) below. In this case it is “doubled” by the newly developed Otomi demonstrative strategy. (35) Gi=tsi lo ke nä’ä gi ne. pres.2=drink sg.3.ntr what dem.sg pres.2 want ‘Drink what you want.’ Apart from the relative pronouns, there were quite a lot of Spanish indefinite pronouns in our data. We found frequent occurrences of ni’na (< ni una) ‘none’; kadu ’na (< cada una) ‘each’; kada kyen (< cada quien) ‘everyone’; kwalkyera (< cualquiera) ‘whomever’ and näda (< nada) ‘nothing’. Also interrogative pronouns and adverbials are borrowed: ke (< que) ‘what’; por ke (< por qué) ‘why’; pa ke (< para qué) ‘what for’; and komo (< cómo) ‘how’. Even the emphasizing pronoun mismo ‘self’ is borrowed, alone as in (36a), and in a double marked construction as in (36b). (36) a. mismo Kwä self God ‘God himself’ b. Enä mismo-se. say self-self ‘He said himself.’
7. Constituent order Almost certainly, the constituent order of Otomi has been influenced by Spanish. As already mentioned earlier, there is a tendency in Otomi to replace the classical VOS and VS main clause orders by the Spanish orders SVO and SV. In nominal predications, classical Otomi has the same order as Spanish, i.e. the subject precedes the predicate, as in (37). However, in the case of property assigning predicates, which have proclitic xi= and optionally also the suffixes -gi and -’i, the subject follows the predicate in the classical language. But today also in these contexts subjects may be found in pre-predicational position, just like in Spanish. This is shown in (38a/b).
Otomi
455
(37) Ar komisaryado ge def.sg political.commissioner dem ar heku-hai ja-r ehido. def.sg divide-land loc-def.sg communal-fields ‘The political commissioner is the distributor of the lands in the communal fields.’ (38) a. Xi=nkuhi ar ngo. prf.3=delicious def.sg meat ‘The meat is delicious.’ b. Ar ngo xi=nkuhi. def.sg meat prf.3=delicious ‘The meat is delicious.’ These changes in constituent order may well be in line with the restructuring of the Otomi discourse in the direction of Spanish that we suggested above with respect to the introduction of Spanish discourse markers. Both in Otomi and Spanish the order of a possessive construction is possessed–possessor, so no changes may be expected here. The same applies to the relative clause order: both languages are N–Rel.
8. Syntax We see a clear tendency in Otomi to import some highly frequent Spanish verbs and auxiliaries which mark aspect and modality. But also other changes are taking place in the area of Otomi syntax. In this section, we discuss a few examples of the most frequently attested phenomena. A recently developed periphrastic construction marks progressive aspect. It is almost certainly contact-induced, since Otomi uses the Spanish verb sige ‘continue’. This is shown in (39). (39) Him-bi=patu nu-r hñäñho, neg-past.3=change dem-def.sg Otomi siempre sige ñätho. always continue speak.Otomi ‘Otomi hasn’t changed, they always keep speaking it.’
456
Ewald Hekking and Dik Bakker
The expression of modality is affected by contact-induced constructions as well. Otomi frequently employs Spanish modal auxiliaries of necessity and possibility, such as tyene ke, (< tiene que) ‘have to’ in example (40); debe (< debe) ‘must’; pwede (< puede) ‘can’ in example (41); and nesesita (< necesita) ‘need’ in example (32) above. These are then followed by the future marking auxiliary and the main verb, just like the native alternatives mahyoni ‘is necessary’ and ar tsa ‘is possible’. (40) Pero tyene ke n-da mpefi. but have.to past-fut.3 work ‘But they had to work’ (41) Ya mi=pwede n-da mats’i ja-r ’batha. already past.3=can past-fut.3 help loc-def.sg field ‘They could already help in the field’ Another periphrastic contact-induced construction of a modal nature is the use of the calque pets’i (~tiene ‘have’) as in (42a) instead of the original form mahyoni ‘is necessary’ in (42b). (42) a. Di pets’i ga mpe-ka pres.1 have fut.1 work-emph.1 ga ’yo-ga pa nu’i gi ñuni. fut.1 walk-emph.1 for you fut.2 eat b. Mahyoni ga mpe-ka is.necessary fut.1 work-emph.1 ga ’yo-ga pa nu’i gi ñuni. fut.1 walk-emph.1 for you fut.2 eat ‘I have to work and to walk, in order that you eat.’ A new periphrastic construction developed to mark repetition. Next to the classical construction ma ’nagi ‘again’ + main verb, the construction pengi ’return’ + main verb is very common in modern Otomi. Arguably, this is a calque of the Spanish construction volver a + main verb. Double marking is common, as shown in (43). (43) Dá=pengi dá=uni ma ’nagi. past.1=return past.1=give again ‘I gave it to them again.’
Otomi
457
Yet another example of potential calquing we find in expressions of possession. There is a clear increase of the use of the verbs ’ñehe ‘to keep’, ‘to breed’ for animate and pets’i ‘to have’ for inanimate possession. The original ways to express possession in Otomi employs the verb ’bui ‘exist’ or the locative auxiliary ja ‘there is’ in combination with the possessive proclitic and the possessed noun. Both new uses are probably calques of the Spanish tener ‘to have’. (44) a. Ma=t’ixu ya ’bui yoho yá=bätsi. pos.1=daughter already exist two pos.3pl=child b. Ma=t’ixu ya ’ñehe yoho yá=bätsi. pos.1=daughter already keep two pos.3pl=child ‘My daughter has already two children.’ (45) ’Bui ’ra ya zaa pets’i yá=’ba xi=nt’axi. be indef.pl tree possess pos.3pl=sap past.3=white ‘There are some trees that have a white sap.’ In predicative constructions, which in classical Otomi would be without a copula, we sometimes find the Spanish copula ta (< está). It is used with both nominal predicates and stative verbs. Examples are the predicatively used noun u ‘salt’ in (46) and the stative verb txutx’ulo ‘to be small’ in (47). (46) Ar ’yot’u-ngo ta ar u. def.sg dry-meat cop def.sg salt ‘The dry meat is salty.’ (47) Ar tsut’äxi ar dätä ne yá=t’olo t’äxi ta txutx’ulo. def.sg she.goat def.sg big and pos.3pl=child goat cop small. ‘The she-goat is big and her kids are small.’ Among the more striking contact phenomena in the syntax of Otomi are the new strategies for the coordination and subordination of clauses. As already discussed in Section 6 Otomi borrows a huge amount of Spanish coordinators and subordinators. It uses many Spanish connectors of addition, contrast and disjunction, many Spanish adverbial clause markers of time, place, manner, purpose, cause, condition and concession, many Spanish subordinators in complement clauses and Spanish relativizers in relative clauses. Because of the introduction of these grammatical elements Otomi is now less asyndetic
458
Ewald Hekking and Dik Bakker
in clausal relations. We gave several examples of this phenomenon above. Yet another recent syntactic change that we observed is the disappearance of the question particle ha that marks the beginning of a yes/no question. It is found in example (48). (48) Yogo hin-gi=pede ir=hñäki ko ya jä’i? why neg-pres.2=tell pos.2=problem with def.pl person (Ha) g-ar gone? inter pres.2-def.sg dumb ‘“Why don’t you tell your problem to those gentlemen?” “Are you dumb?”’ We have not enough evidence that the Spanish interrogative intonation is adopted in these cases, but we have the strong suspicion that the marker is disappearing under influence of Spanish, which does not have such a marker, and uses intonation to contrast such questions with their declarative counterparts.
9. Lexicon We have seen above that only half of the borrowed elements belong to one of the four major lexical classes – noun, verb, adjective, adverb – and that the other half are borrowed from the grammatical part of the Spanish lexicon. Prominently among the latter are Spanish prepositions: they make up 44 percent of the grammatical elements borrowed, and 21 percent of all borrowed elements. Otomi does not have prepositions of itself, though it has deverbal elements with adverbial function which have a syntactic slot just before noun phrases. This is also where Spanish prepositions end up in Otomi utterances. It may be argued that from the perspective of Otomi grammar, these borrowed prepositions should be analyzed as adverbs on functional grounds, and should therefore be added to the lexical part of the borrowed inventory rather than to the grammatical part. This would bring the total contribution of lexical borrowings closer to the percentage we found for Guaraní and Quechua. When we restrict ourselves to the 1,413 meaning concepts defined by the Loan Word Typology project (LWT; Haspelmath and Tadmor forthc.) relevant for Otomi, we find the following distribution of Spanish borrowings. Table 4 gives a breakdown for the semantic fields defined by the LWT project, and only for those that have either more than 20 percent borrowings or less
Otomi
459
than 10 percent. The second column gives the total number of concepts defined for the corresponding field; the third column gives the percentage for which there is (also) a Spanish loanword; and the fourth column gives the percentage of concepts for which there is only a loanword. Not surprisingly, the category Modern World is the field with most borrowings: for over half the entries a Spanish form is used. However, it is striking that for most of these there is also an Otomi word. This is different for category 2, Animals, where loanwords are typically unique and have no corresponding word in Otomi. On the other side of the scale, we find that for Body Parts, Physical World and notions for Sense Perception, there is hardly any borrowing at all. We see that the overall percentage of borrowings is around 16 percent. The percentage of borrowed types we found in our corpus is somewhat higher, at 22.2 percent. This is to be expected, since the list of concepts of the LWT project might be seen as defining the core vocabulary of a language. Among the data that we collected in Santiago Mexquititlán en Tolimán we have found many cases of code-switching and code mixing. On the other hand, we have also found a huge amount of Spanish phrases that are not to be considered as code mixings, but composite or frozen borrowings, since they
Table 4. Spanish loans per semantic field Semantic field Modern world Animals Agriculture/vegetation Warfare and hunting Dwelling/house/furniture Clothing/personal adornment Religion and beliefs Spatial relations Moral/aesthetic Law Body parts Physical world Sense perception Total
Total number of concepts 57 105 66 40 45 57 24 74 48 26 156 68 48 1413
Percentage borrowed
Only Spanish word
53% 35% 29% 22% 22% 21% 21%
18% 29% 17% 5% 4% 11% 8%
9% 8% 8% 6% 3% 2% 16.3%
4% 6% 4% 2% 0% 0% 6.9%
460
Ewald Hekking and Dik Bakker
appear regularly and for different speakers. Most of them are noun phrases, such as agua de mais ‘maize water’; barryo kinto ‘neighbourhood number five’; el beintisinko de julyo ‘the twenty-fifth of July’; and la mera berdad ‘the very truth’. But we have also found prepositional phrases such as a los kinse diya ‘in two weeks time’; kon el tyempo ‘in due time’; and por mi parte ‘as far as I am concerned’. We came across several complex subordinators such as asi es de ke ‘so it happens that’; mbes de ke ’instead of’; and verbal phrases such as kreo ke ’I think that’ and pares ke ‘it seems that’. One could argue that these complex constructions are interpreted as unanalyzable entities by speakers of Otomi. Some of such complex borrowings are used as discourse markers, such as a ber; a lo mejor; ni modo; no mäs; and algo asi.
10. Conclusions We may conclude that Otomi has been influenced by Spanish at many linguistic levels. In the area of phonology, as a result of the adoption of unassimilated Spanish loanwords, new sounds have been introduced, such as the trilled alveolar vibrant rr, the lateral l and the affricated alveopalatal ch. Tone does not seem to have been affected by contact. As far as the lexicon is concerned, not only are many Spanish content words introduced, both nouns, verbs, adverbs and some adjectives, but also a high amount of function words, such as prepositions, coordinators, subordinators, relative pronouns and discourse markers. We have also found a large number of complete Spanish phrases. The systematic adoption of function words makes the Otomi grammar less asyndetic in its clause combining strategies. The loss of certain verbal affixes which mark inclusivity and the company, and their replacement by Spanish prepositions, make the language less synthetic in its morphological characterization. In relation to constituent order we have observed that the basic Otomi orders V–O–S, V–S and Adjectival predicate-Subject are frequently replaced by S–V–O, S–V and Subject–Adjectival predicate, the basic orders of Spanish. Otomi syntax is undergoing restructuring as a result of the introduction of several auxiliaries and a copula. Several other striking changes were observed in this area. As a result of all these changes, we estimate that Otomi would be situated somewhere between points 2 and 3 on the borrowing scale of Thomason (2001).
Otomi
461
Abbreviations 1 first person 2 second person 3 third person act actuality ben benefactive cop copula def definite dem demonstrative du dual emph emphatic excl exclusive fut future imprf imperfective incl inclusive indef indefinite
instr instrumental inter interrogative lim limitative loc locative neg negative ntr neuter obj (in)direct object pl plural pos possessive prf perfect pres present past past ref reference sg singular
Notes 1. (< engaña) should be read as ‘based on the Spanish word engaña’. Note that the Spanish forms are given in the standard spelling, not in a phonological form. 2. See Comrie (1989: 147153) for a typology of relativization. 3. We distinguish between borrowings on the one hand and code-switching on the other hand. Stretches of text which we consider to be code switches were not included in these figures. We consider Spanish prepositions as grammatical rather than lexical elements. They will be discussed in Section 6. 4. These examples stem from a questionnaire with Spanish sentences which were translated into Otomi by a number of native speakers from the two communities.
References Andrews, Henrietta 1993 The function of verb prefixes in Southwestern Otomi. The Summer Institute of Linguistics and the University of Texas at Arlington Publications in Linguistics 115. Dallas: Summer Institute of Linguistics and the University of Texas at Arlington.
462
Ewald Hekking and Dik Bakker
Bakker, Dik, and Ewald Hekking 1999 A functional approach to linguistic change through language contact: The case of Spanish and Otomi. Working Papers in Functional Grammar 71: 132. Amsterdam. Bakker, Dik, Jorge Gómez-Rendón, and Ewald Hekking Forthc. Spanish meets Guaraní, Otomi and Quichua: A multilingual confrontation. In: Thomas Stolz, Dik Bakker and Rosa Salas Palomo (eds.), Aspects of Language Contact. Berlin/New York: Mouton de Gruyter. Bartholomew, Doris 2004 Notas sobre la gramática. In: Luis Hernández Cruz, Moisés Victoria Torquemada, and Donaldo Sinclair Crawford (eds.), Diccionario del Hñähñu (Otomi) del Valle del Mezquital (Hidalgo). (Vocabularios Indígenas 45.) México, D.F.: Instituto Lingüístico de Verano. Comrie, Bernard 1989 Language Universals and Linguistic Typology. Oxford: Blackwell. Ecker, Lawrence 1952 Compendio de gramática otomí (introducción a un diccionario otomíespañol), Anales del Instituto Nacional de Antropología e Historia, tomo 4, no. 32. México: Instituto Nacional de Antropología e Historia. Guerrero, Alonso 2002 El Códice Martín del Toro. De la oralidad y la escritura, una perspectiva Otomi. Siglos XV–XVII. Tesis de Licenciatura en Etnohistoria. Escuela Nacional de Antropología e Historia. Haspelmath, Martin, and Uri Tadmor. Loanword Typology project: A collaborative project toward the comparative study of lexical borrowability in the world’s languages. Coordinated by the Max Planck Institute for Evolutionary Anthropology, Department of Linguistics. Hekking, Ewald 1995 El Otomi de Santiago M: Desplazamiento Linguïstico, Préstamos y Cambios Grammaticales. Amsterdam: Institute for Funtional Research into Language and Language Use. 2001 Cambios gramaticales por el contacto entre el Otomi y el español. In: Klaus Zimmermann, and Thomas Stolz (eds.), 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, 127151. Frankfurt am Main/Madrid: Vervuert/ Iberoamericana. 2002 Desplazamiento, pérdida y perspectivas para la revitalización del hñäñho. In: Yolanda Lastra de Suárez and Noemí Quezada (eds.), Estudios de Cultura Otopame 3, 221248. México, D.F.: Universidad Nacional Autónoma de México.
Otomi
463
Hekking, Ewald, and Dik Bakker 1998 Language shift and Spanish content and function words in Otomi. In: Bernard Caron (ed.), Actes du 16e Congres International des Linguistes. Oxford: Elsevier Sciences. 1998 El Otomi y el español de Santiago Mexquititlán: Dos lenguas en contacto. Foro Hispánico 13, Sociolingüística: Lenguas en Contacto, 4574. Amsterdam: Rodopi. 2005 Problemas en la adquisición de una segunda lengua: El Otomi frente al Español. In: Claudine Chamoreau, and Yolanda Lastra de Suárez (eds.), Dinámica Lingüística de las Lenguas en Contacto. Hermosillo: Universidad de Sonora. Hekking, Ewald, and Pieter Muysken 1995 Otomi y Quechua: una comparación de los elementos prestados del español. In: Klaus Zimmermann (ed.), Lenguas en Contacto en Hispanoamérica: Nuevos Enfoques. Frankfurt am Main: Vervuert. Hekking, Ewald, and Severiano Andrés de Jesús 1984 Gramática Otomi. Querétaro (México): Universidad Autónoma de Querétaro. Forthc. He’mi Mpomuhñä ar Hñäñho ko ya Njat’i/Diccionario Explicativo Ilustrado del Otomi del Estado de Querétaro. Hekking, Ewald, and Severiano Andrés de Jesús (eds.) 2002 Ya ’bede ar hñäñho Nsantumuriya/Cuentos en el Otomi de Amealco. Querétaro (México): Universidad Autónoma de Querétaro. Hengeveld, Kees, Jan Rijkhoff, and Anna Siewierska 2004 Parts-of-speech systems and word order. Journal of Linguistics 40 (3): 527570. Hernández Cruz, Luis, Moisés Victoria Torquemada, and Donaldo Sinclair Crawford 2004 Diccionario del Hñähñu (Otomi). México: Instituto Lingüístico de Verano, A.C. Hess, H. Harwood 1968 The Syntactic Structure of Mezquital Otomi. (Janua Linguarum. Series Practica 43.) The Hague: Mouton. Jiménez Moreno, Wigberto 1939 Origen y significación del nombre Otomi. Revista Mexicana de Estudios Antropológicos III: 6268. Lastra de Suárez, Yolanda 1989 Otomi de San Andrés Cuexcontitlán, estado de México. Archivo de las lenguas indígenas de México. México: El Colegio de México. 1992 El Otomi de Toluca. México, D.F.: Instituto de Investigaciones Antropológicas, UNAM. 1994 Préstamos y alternancias de código en Otomi y en español. In: Carolyn Mackay and Veronica Vazquez (eds.), Investigaciones Lingüísticas en Mesoamérica. México: Universidad Autónoma de México.
464
Ewald Hekking and Dik Bakker
Lastra de Suárez, Yolanda 1997 El Otomi de Ixtenco. México, D.F.: Instituto de Investigaciones Antropológicas, UNAM. Matras, Yaron 1998 Utterance modifiers and universals of grammatical borrowing. Linguistics 36 (2): 281331. Palancar, Enrique Forthc. Property concepts in Otomi: A language with no adjectives. International Journal of American Linguistics. Ruhlen, Merritt 1991 A Guide to the World’s Languages. London: Edward Arnold. Sahagún, Bernardino de 1999 Historia general de las cosas de Nueva España. México: Porrúa. Suárez, Jorge A. 1983 The Mesoamerican Indian Languages. Cambridge: Cambridge University Press Thomason, Sarah G. 2001 Language Contact. Edinburgh: Edinburgh University Press. Salinas Pedraza, J. 1983 Etnografía del Otomi. México: Instituto Nacional Indigenista. Urbano, A. [1605] 1990. Arte breve de la lengua Otomi y vocabulario trilingüe. R. Acuña (ed.). México D.F.: Universidad Nacional Autónoma de México. Voigtlander, Katherine, and Artemisa Echegoyen 1985 Luces contemporáneas del Otomi: Gramática del Otomi 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.
Grammatical borrowing in Purepecha Claudine Chamoreau
1. Background Purepecha (P’orhépecha or Tarascan) is classified as an isolated language, spoken by around 110,000 people (10 percent of them monolingual), in the state of Michoacan, in the west of Mexico. Spanish was introduced in the sixteenth century, and became the official language of Mexico, where more than about a hundred languages are still spoken. It gained more importance with the linguistic policies of the Mexican Independence and Revolution, in the nineteenth and in the twentieth centuries respectively. Spanish functions as a prestigious language, and is connected to education, a better standard of living, oral and written media, religion, administration, commerce, and employment. Nevertheless, Purepecha, in 2003, acquired (like the other indigenous languages spoken in Mexico) the status of official language. In general, Purepecha is used only orally, having been established as a written language very recently, and only used in that mode by a few individuals (specifically, the intellectual-speakers or the teachers). The language is spoken by 28 percent of the Purepecha children aged between 5 and 14, this data indicating that Purepecha is not generally transmitted to the younger generation, who prefer to learn and use Spanish. Moreover, the situation is not homogeneous in all the communities. In some villages, the language functions for communication among all family members and friends (salutations and discussions at home, in the streets, in the shops or markets, and in children’s games). In other communities, only the middle-aged and older people speak Purepecha. Spanish has been the principal contact language for many centuries, however, before the Conquest, there were speakers of other languages in the area – mostly Nahuatl (Uto-Aztecan family), and Otomi (Otopamean family). The influences of these languages in Purepecha has not been studied in detail, but one hypothesis has been offered regarding constituent order. Purepecha exhibits all the traits of an SOV language: (a) tense, aspect and modal markers follow the verb, (b) postpositions, (c) suffixes almost exclusively, (d) case markers, (e) main verbs precede inflected auxiliaries, (f) genitives can precede the head noun, and (g) relative clause can precede the head noun. Nevertheless,
466
Claudine Chamoreau
in the Lake Patzcuaro area, Purepecha has become SVO (Capistran 2002). This order has been attested since the sixteenth century, and has become progressively more widespread since that time (Villavicencio 2006). Considering that Nahuatl – and Otomi – present a verb-initial structure, this change probably has its roots in areal contact prior to the sixteenth century, with the subsequent influence of Spanish, an SVO language, continuing the process. In the present chapter, I will concentrate on the influence of Spanish contact on Purepecha, specifically, on the grammatical structure. related to this contact are found in the areas of phonology, morphological typology, nominal and verbal structures, other parts of speech, constituent order, and syntax. This chapter deals with the dialect of Jaracuaro (denoted Jr), a peninsula in Lake Patzcuaro, however, when necessary, I use data from other varieties. Purepecha varieties are more or less mutually intelligible, nevertheless, great sociolinguistic differences exist between them (Chamoreau 2005). Most of the data considered for this chapter are the result of my own field research projects carried out over a period of fifteen years.
2. Phonology In the phonological system of Purepecha, two phonemes – that are not shared with Spanish – have been influenced by Spanish: the retroflex consonant /ɽ/ and the high central vowel //. In some varieties (for example, Cuanajo), Purepecha has a phonological opposition between the retroflex /ɽ/ and the flap /r/ (e.g. jurani ‘to make somebody cough’/ juɽani ‘to come’), however, in certain varieties, this opposition no longer exists; the retroflex becomes either a flap, losing the retroflex/flap opposition, or a lateral, a phoneme probably borrowed from Spanish. Purepecha conserves an opposition, but shows a new lateral/ flap feature. In general, the lateral only appears in Spanish loanwords such as azuli ‘blue’ (from Spanish ‘azul’), or limoni ‘lemon’ (from Spanish ‘limon’). However, in some varieties (for example, Comachuen, Arantepacua), young and middle-aged speakers use the lateral (jolempiri ‘teacher’), while the older generation uses the flap (jorempiri), or the retroflex (joɽempiri). The use of the lateral in Purepecha words reveals the replacement of the Purepecha phoneme by the Spanish phoneme. Currently, Purepecha is acquiring a new phonological opposition (Chamoreau 2002a). The high central vowel // is used after /ts/, /tsh/ and /ʃ/, and a phonological opposition appears between // and the high front vowel /i/ (e.g. tsiriri ‘rib’/ tsriri ‘paste’; kheʃi ‘shoulder’/ khaʃ ‘shape’). Nevertheless, particularly in
Purepecha
467
the varieties which have lost the retroflex /ɽ/, and have transferred the lateral /l/, and in other varieties, in the case of the young and middle speakers, the high central vowel // is no longer used, and /i/ replaces // (tsiriri ‘paste’, khaʃi ‘shape’). The phonological system of these varieties has lost a vowel, and, accordingly, presents the same vowel system as Spanish (/i/, /u/, /e/, /o/, /a/).
3. Morphological typology Purepecha has not undergone an important re-structuring of its typological profile. It is an agglutinative and synthetic language that comprises a very elaborated derivational verbal system. For example, for the passive, the verb presents a derivational verbal suffix na (1), and in order to express the equational constructions (2) the e suffix is used. In certain varieties, the suffix is i: (1)
Tʃkurhi kuɽi-ra-na-ʃ-ti xwata-ɽu. firewood burn-caus-passiv-aor-ass3 hill-loc ‘The firewood was burned on the hill.’ (Jr)
(2)
Xwánu xoɽempiri-i-ʃ-ti. John teacher-pred-aor-ass3 ‘John is a teacher.’ (Jr)
Nevertheless, Purepecha exhibits new tendencies, in which analytic-periphrastic constructions appear. Passive (3) and equational (5) periphrastic structures adapt Purepechan morphemes to Spanish patterns without the transfer of linguistic material, which suggests that contact-induced grammaticalization processes have taken place. The passive periphrastic construction emerges from a patient-oriented resultative participle plus xa ‘be there’; a verb which became an auxiliary. Evidence supporting the consideration of this construction as a remodeling of the structure (PAT) includes: (a) Passive constructions involving passive participles appear in Indo-European languages, and are very rare in the Americas (Haspelmath 1994); (b) the agent is introduced as an oblique complement by using the postposition ximpo (3) in the same way as the Spanish passive construction with ser, whereas this is generally impossible in the Purepecha passive derivational construction (1); (c) The subject is always the patient, as in Spanish, whereas, in the derivational passive structure, the subject is the
468
Claudine Chamoreau
divalent-patient or the trivalent-recipient (Chamoreau 2007); (d) the younger generation use a passive periphrastic construction with the xinte ‘be’ copular verb (4), treated as the Spanish ser auxiliary, calquing the Spanish Aux.-Part. Order, whereas, in the passive periphrastic construction with xa (3), the Purepecha Part.-Aux. order is preserved: (3)
Tʃkurhi kuɽi-kata xa-ɽa-ʃ-ti xutʃari tata ximpo. firewood burn-partpp be there-ft-aor-ass3 pos1pl father inst ‘The firewood was burned by my father.’ (Jr)
(4)
Enka no u-a-ka xuramukwa-nkuni xinte-a-ti sub neg do-fut-subj law-com be-fut-ass3 ʃuka-kata. dispute-partpp ‘If he does not respect the law, he will be punished.’ (Jr)
The analytic equational construction with xinte ‘be’ is an internally motivated reanalysis, from a demonstrative to a ‘be’ verb used as a presentative (Chamoreau 2006). As a result of the influence of Spanish, many young people prefer to use the verb xinte (5) rather than use the derivational construction (2). Xinte appears essentially with nouns and pronouns. Many young people use xinte solely with adjectives (6) indicating a quality which refers to identity, and which is independent of the situation, as in ‘ser’ in Spanish. In (5) and (6), the Spanish SVAdj. order is a calque. This is a construction that is in opposition to the construction with xa (7) which expresses a relative quality dependent on the situation, as in estar in Spanish. With xa, the order is generally the Purepecha SAdj.V order, although it is possible to find the Spanish SVAdj. Order. (5)
Xwánu xinte-ʃ-ti xoɽempiri. John be-aor-ass3 teacher ‘John is a teacher.’ (Jr)
(6)
Iʃu pakanta, kwhiripu miri-kwali-s-p-ti ya ima-ni here Pacanda people forgot-mid-aor-pas-ass3 already dem-obj atʃati-ni ka myá-ntha-ʃa-p-ti eski ima xinte-p-ka riko. man-obj and think-it-prog-pas-ass3 sub dem be-aor/pas-subj rich ‘Here in Pacanda, people had forgotten this man. They used to think that he was rich.’ (Pc)
Purepecha
(7)
469
Naranʃa téri xa-ɽa-ʃ-ti. orange sweet be there-ft-aor-ass3 ‘The orange is sweet.’ (Jr)
4. Nominal structures Changes that could have arisen through the influence of Spanish include the tendency to use the plural marker itʃa, and the object case marker ni, with inanimate entities (8). Traditionally, the plural and object case markers are obligatory only for animate and definite entities (Villavicencio 2006). (8)
Ints-ku-ʃn-ti=kʃ tsɨtsɨki-itʃa-ni. give-3app-prog-ass3=3pl flower-pl-obj ‘They used to give her flowers.’ (Jr)
The Spanish diminutive suffix has become productive in Purepecha, however, only the masculine ito is used, pronounced ito or itu. Gender does not exist in Purepecha. This suffix is used with nouns, adjectives (9a), and classifiers (9b). This latter exhibits an adaptation to the Purepecha nominal phrase. (9)
a. Witsintikwa phakha-ra-ʃ-ka=ni xantiakhu-itu. yesterday stay-mid-aor-ass1/2=1 alone-dim ‘I stayed alone yesterday.’ (Jr) b. Ma itʃakwa-itu wítʃu-ni ká-ʃn-ka=ni. one long-dim dog-obj own-hab-ass1/2=1 ‘I have one little dog.’ (Jr)
5. Verbal structures There are only a few contact phenomena in the verbal structures. The most relevant of these is the transfer of the ser/estar semantic opposition (PATinfluence), being adapted as a xinte/xa dichotomy (Section 3). As a result of the influence of Spanish, the constructions with the verb xinte ‘be’ gain a greater semblance to the Spanish construction with the verb ‘ser’: passive constructions (4), equational constructions (5), and attributive constructions (6). Many young speakers integrate an idiomatic expression, dejar de ser, that they have calqued from the Spanish (10):
470
Claudine Chamoreau
(10) Xorentpherakwa xurakhu-sn-ti xinte-ni ís. education let-hab-ass3 be-inf thus ‘The education ceased to be like that.’ (Pc) In the same way, the constructions with xa ‘be there’ have adopted the values of the Spanish ‘estar’ structures: passive constructions (3), and attributive ones (7).
6. Other parts of speech Purepecha shows a significant number of Spanish loans in the category ‘other parts of speech’. Most loans are of the MAT type, but some cases of PATinfluence have been established within the numeral system and discourse markers. The numeral system in Purepecha is vigesimal, and the remodeling to a decimal system is due to Spanish influence. The numbers from 1 to 6, and 10 and 20 are generally known and used, but younger speakers prefer to use Spanish numbers except for numbers below 5. Counting and adding are generally performed using Spanish numbers. There are no contact phenomena in quantifiers. The indefinite pronoun siempre ‘always’ is highly integrated, whereas other indefinite pronouns appear only occasionally. Siempre is used, with the vowel adaptation siémpri, by all generations (11), and has gained ground relative to the Purepecha indefinite pronouns mameni and menkhu, which also express time (12) (11) Ima siémpri mí-ti-ʃn-an-ti. dem always know-face-hab-pas-ass3 ‘He had always known it.’ (Jr) (12) Xutʃari tati ménkhu xutʃi-o xa-ɽa-ʃn-ti. pos1pl father always pos1-res be there-ft-hab-ass3 ‘Our father is always at home.’ (Jr) Purepecha has borrowed two Spanish coordinating connectors: o ‘disjunction’ (13) and pero ‘contrast’ (14). The Spanish connector y ‘addition’ has not been borrowed. In this case, the Purepecha connector ka ‘and’ is used (15).
Purepecha
471
The connectors o and peru are concepts formerly unmarked in Purepecha. The native Purepecha marker ka ‘addition’ is used to combine clauses. Pero can additionally express a change of topic (16). On the phrasal level, o can be used to combine phrases (15). (13) Tʃi tʃentʃeki urapi-ʃ-ki o tʃi kawayu tuɽipi-ni. pos2 donkey be white-aor-int or pos2 horse be black-inf ‘Is your donkey white, or is your horse black?’ (Jr) (14) Xi wé-ka-ʃa-p-ka ni-ra-ni péro no 1 want-ft-prog-pas-ass1/2 go-ft-inf but neg ú-ʃ-ka ʃaná-ra-ni. be able-aor-ass1/2 walk-mid-inf ‘I wanted to go, but I was not able to walk.’ (Jr) (15) Ínts-a-sn-ti=ks ima-nki aɽi-s-ka ya tétʃakwa give-3plobj-hab-ass3=3pl dem-sub tell-aor-subj already wine ka wiraterakwa ka sáni tsíri o sáni ʃapumata. and alcohol and few corn or few roasted corn ‘They used to give them what they asked for: wine and alcohol, and a little corn, or a little roasted corn.’ (Tr) (16) Atʃati pyá-ʃ-ti yámintu ampe nénki wé-ta-ɽi-a-ka man buy-aor-ass3 all what sub want-caus-body-fut-subj ima péru ni-ntha-ʃ-ti=kʃ ya anima-itʃa. dem but go-centrif-aor-ass3=3pl already soul-pl ‘The man has bought all he will need, but (on the other hand) the souls have left.’ (Jr) Many subordinating conjunctions are borrowed from Spanish. The Spanish complementizer que, pronounced ke or ki, is never used alone. One hypothesis suggests that ke was borrowed from Spanish. Another possibility is that Purepecha also had a subordinating conjunction with the form ki, attested in the sixteenth century. A convergence between the two elements has been favoured because they present the same form. This topic had not been studied yet. The subordinating conjunction ke is employed with other borrowed markers, functioning as complex conjunctions: porki in a causal clause (17), para ke in a purpose clause (18a), sikiera ke, which is a synthetic form of the Spanish ‘si quiera’ in a hypothetical clause (19a), and sino ke in a contrast
472
Claudine Chamoreau
clause (20). These elements are analyzed as subordinating conjunctions, because they respect the Purepecha constructions; the verb of the subordinating clause is marked by subjunctive mood. Purepecha has various subordinating conjunctions, enki, eʃka and eʃki, which seem to be being progressively replaced in these constructions by ke. Nevertheless, many speakers, especially those of the middle-aged and older generations, prefer to use the Purepecha ximpoki ‘because’ instead of porki, and to use the borrowed markers para (18b) and sikiera (19b), in combination with the Purepecha conjunctions eʃka or eʃki. (17) Ima xu-ɽa-ʃn-ti porki thu yoɽi-ʃ-ka. dem come-ft-hab-ass3 because 2 call-aor-subj ‘He used to come because you called him.’ (Jr) (18) a. Xwanu xu-ɽa-ʃ-ti para ke iʃe-ka=ri. John come-ft-aor-ass3 for sub see-subj=2 ‘John came for you to see him.’ (Jr) b. Aɽi-ʃn-ti=kini para iʃki mi-ti-a-ka. tell-hab-ass3=2obj for sub know-face-fut-subj ‘He tells you that, so that you know it.’ (Jr) (19) a. Sikiera ke pirí-a-ka. provided sub sing-fut-subj ‘Let’s hope he will sing!’ (Jr) b. Sikiera iʃki tsma xu-nkwa-ka. provided sub dempl come-centrip-subj ‘Let’s hope they come back.’ (Jr) (20) No-teru=kʃ anatapu-etʃa-ni iʃe-a-ʃ-ti sino neg-more=3pl tree-pl-obj see-3plobj-aor-ass3 but ke lwegu=kʃ tʃapa-ta-a-ka. sub then=3pl cut-caus-3objpl-subj ‘They do not see any trees anymore, because somebody cut them.’ (Jr) The situation of variation between the use of ke or iʃka/iʃki in subordinating constructions is also attested with the comparative structure (see §8). Generally, para is introduced in a purpose clause with para ke or para iʃki (18a, 18b), however, it can also appear in a non-finite purpose construction (21).
Purepecha
473
(21) Thu no xatsi-ʃ-ka para xaka-khu-ni. 2 neg have-aor-ass1/2 for believe-ft-inf ‘You don’t have to believe him.’ (Jr) Spanish temporal adverbializers that have been borrowed include: hasta ‘until’ (22), desde ‘from’ (23), apenas ‘as soon as’, pronounced apenaʃ (24), luego ‘then’ (20), and entonces ‘then’, generally pronounced tonses (25). The adverbializer hasta additionally has a spatial deictic use (26). There is a native suffix of localization ɽu (26), which has an extended function (fixed and removed localization, ablative, translative, etc.). The use of hasta allows the specification of the type of localization. (22) Xima khama-ʃ-ti ya ásta wéxuɽini. then finish-aor-ass3 already until year ‘He had finished it by the new year.’ (Jr) (23) Ántʃi-kuɽi-ʃa-ka=ni désde witsintikwa. work-mid-prog-ass1/2=1 for/since yesterday ‘I have been working since yesterday.’ (Jr) (24) Petu kwhi-a-ti apenaxï thu nya-ra-ka. Peter sleep-fut-ass3 as soon as 2 arrive-ft-subj ‘Peter fell asleep as soon as you arrived.’ (Jr) (25) Tonses no ampakiti=thu=tʃka ya ni-ntha-ʃ-ti ya. then neg good=too=so already go-centrif-aor-ass3 already ‘Then the devil has left too.’ (Jr) (26) Ni-a-ka=kʃ ásta xini yoɽekwa-ɽu. go-fut-ass1/2=1pl until there river-loc ‘We will go up to the river.’ (Jr) Purepecha did not have prepositions before contact; we can assume that it was a language with only postpositions, and some suffixed case markers. So the prepositions para and por are borrowed in combination with their phrase-combining construction, i.e., they appear before the phrase or the morpheme (Chamoreau 2002b). The preposition para functions in a recipient clause (27a), and por expresses agentive (28a), causal (27a), and instrumental clauses (28a). The Purepecha postposition ximpo ‘instrumental’ can be used
474
Claudine Chamoreau
in functions similar to para (27b), or to por (28b, 29b, 30b). In all these contexts, the Purepecha marker may appear in a double construction (27c, 28c, 29c, 30c). (27) a. Ima kuɽa-tʃi-ʃ-ti=rini itʃuskuta para ama-mpa. dem ask-1/2app-aor-ass3=1obj tortilla for mother-posp3 ‘He has asked me for tortillas for his mother.’ (Jr) b. Ima kuɽa-tʃi-ʃ-ti=rini itʃuskuta ama-mpa ximpo. (Jr) c. Ima kuɽa-tʃi-ʃ-ti=rini itʃuskuta para ama-mpa ximpo. (Jr) (28) a. Mí-ti-ʃ-ti por ima. know-face-aor-ass3 by dem ‘He knows it through him.’ (Jr) b. Mí-ti-ʃ-ti ima ximpo. (Jr) c. Mí-ti-ʃ-ti por ima ximpo. (Jr) (29) a. Tstski urapiti kunti-kuɽi-ʃa-ti por kwetsapikwa. flower white bend-ref-prog-ass3 under/cause weight ‘The white flower is bending under the weight.’ (Jr) b. Tstski urapiti kunti-kuɽi-ʃa-ti kwetsapikwa ximpo. (Jr) c. Tstski urapiti kunti-kuɽi-ʃa-ti por kwetsapikwa ximpo. (Jr) (30) a. Xu-ɽa-ʃ-ka-ni por kamioni. come-ft-aor-ass1/21 by bus ‘I went by bus.’ (Jr) b. Xu-ɽa-ʃ-ka-ni kamioni ximpo. (Jr) c. Xu-ɽa-ʃ-ka-n por kamioni ximpo. (Jr) The Spanish marker komo is used in Purepecha to introduce a manner clause. (31) Pos=s komo ma atʃa=s xa-ɽa-ʃ-ti. thus=foc like a man=foc be there-ft-aor-ass3 ‘Thus, he was there like a man.’ (Jr) The Spanish phrasal adverb ya is used to mark temporal values with two different nuances: it can introduce a completive value, generally employed with the aspect aorist or the past aorist (32), or it can express a present value (33). The story in (32) is about a vulture that has transformed itself into a woman,
Purepecha
475
and the woman into a vulture. The example expresses that the vulture turned into a woman; that it was no longer an animal. In the same narrative, in (33), there is a contrast between the first verb, in the past tense, which indicates the state of the woman before, and the second verb, in the interrogative clause, which indicates a question about the present state, which is the state of the vulture. (32) Ka ma waɽiti-i-ʃ-ti ya. and a woman-pred-aor-ass3 already ‘And it is already a woman.’ (Jr) (33) Thu no xama-ʃ-p-ka lísto antiʃ=ri xa-ɽa-ʃ-ki 2 neg walk-aor-pas-ass1/2 lively why=2 be there-ft-aor-int ya. already ‘You did not used to be lively, why are you now?’ (Jr) Apart from marking temporal relations, this element ya functions like a discourse marker, with the addition connector ka ‘and’. The latter begins a clause, while the former ends one (32). Finally, many discourse markers are borrowed by Purepecha from the Spanish. The most frequent ones are the fillers: pues ‘thus, then, well’, pronounced pwes or pos (31), and bueno ‘well, sure’, pronounced wenu. It is also possible that as a result of the influence of Spanish, the use of the demonstrative inte ‘this’ is used as a filler like este in that language. This element appears in the same conditions as does inte in Spanish: it expresses a hesitation, a pause, etc. (34). It is a PAT-influence that is not connected to any direct MAT-borrowing. (34) Ximpoka=ni inte pats-nts-ka=na. because=1 em be fade-head-subj=evid ‘Because, em, I am bald, they said.’ (Jr)
7. Constituent order Constituent order seems to be influenced by areal contact prior to the sixteenth century (see Section 1), with Spanish continuing the process, and
476
Claudine Chamoreau
increasing it via the introduction of prepositions (while Purepecha had traditionally used postpositions).
8. Syntax The organization of passive and equational constructions has been influenced by Spanish (see Section 3). Many subordinating conjunctions and adverbial markers are borrowed from Spanish, in combination with the grammatical constructions they appear in, in that language. A syntactic domain which has undergone an important reorganization due to Spanish contact is the comparison of inequality. Purepecha had traditionally utilized a comparative construction of superiority of two types (Chamoreau 1995): the exceed verb of action (35), and the combination of the exceed verb of action and a coordinated polarity construction (36): (35) pedro hatztamahati juanoni ambaqueni Pedro xats-ta-ma-xa-ti xwano-ni ampake-ni. Peter surpass-caus-transf-pres-ass3 John-obj be good-inf ‘Peter is better than John.’(Peter surpasses John in being good). (Gilberti [1558] 1987: 109) (36) pedro hatztamahati ambaqueni, ca noys Pedro xats-ta-ma-xa-ti ampake-ni ka no=ʃ Peter surpass-caus-transf-pres-ass3 be good-inf and neg=foc juan xwano. John ‘Peter is better than John.’ (Peter surpasses with goodness and John does not). (Gilberti 1987: 109) Nowadays, there has been a reorganization as a result of a chain reaction triggered by Spanish interference. A cross-dialectal perspective shows that contact between the indigenous and Spanish constructions (the structure más…que) has brought about the emergence of nine constructions which may be classified in four different types (one type, the applicative construction, is not treated here, because it is marginal and corresponds to the derivative morphological characteristics of the language). I present here eight constructions organized in three types:
Purepecha
477
Type 1 is a borrowing or a PAT-influence of the Spanish comparative construction with borrowed degree and relator (37), with borrowed relator and the degree calque (38) or their Purepecha calques (39). (37) Enrike mas ʃepe-s-ti ke Pedru. Henry more be lazy-aor-ass3 sub Peter ‘Henry is lazier than Peter.’ (Cn) (38) I kamisa sáni=teru xuka-para-s-ti ke iʃu dem shirt few=more put-shoulder-aor-ass3 sub here anapu-e-s-ti. orig-pred-aor-ass3 ‘This shirt is more expensive than those from here.’ (Ih) (39) Thu sáni=teru wiria-ʃ-ka eʃki xi. 2 few=more run-aor-ass1/2 sub 1 ‘You have run more than I have.’ (Cc) Type 2 is a mixed type, employing the Purepecha polarity construction plus the Spanish comparative degree particle mas (40) or its Purepecha calque sáni=teru (41) (40) Xi xatsi-s-ka=ni mas itʃuskuta ka no thu. 1 have-aor-ass1/2=1 more tortilla and neg 2 ‘I have more tortilla than you have.’ (I have more tortilla and you have not) (Jn) (41) Iʃu sáni=teru yó-tha-ɽa-ʃn-ti ka no xini. here few=more long-leg-ft-hab-ass3 and neg there ‘Here is higher than there.’ (Here is higher and there is not) (Jr) Type 3 is a hybrid type, employing the Spanish degree mas plus the relator ke, and a locative construction with the Spanish preposition de (42), which represents an instance of code-mixing, because it only appears in a few expressions, and never alone (the Spanish preposition de ‘of’ appears in this context of comparative constructions and in some expressions, for example de veras ‘sure’). This new hybrid locative construction does not occur either in Spanish or in traditional Purepecha. This construction can occur with the Purepecha degree calque sáni=teru (43):
478
Claudine Chamoreau
(42) Inte atʃa mas khéri-e-s-ti ke de ʃo anapu yamintu. dem man more old-pred-aor-ass3 sub of here orig all ‘This man is older than (of) all the others from here.’ (Tr) (43) I tata sáni=teru=ʃ tʃhana-ʃn-ti ke de wapha-mpa dem man few=more=foc jugar-hab-ass3 sub of son-posp3 ‘This man plays more than his son.’ (Cn) This hybrid construction may also occur with the Spanish preposition entre. In (44), we can observe the presence of the borrowed marker of degree más before the quality, and the comparative relator ke, which is followed by the Spanish preposition entre. (44) Iʃu más khé-ʃ-ti ke entre xini. here more be big-aor-ass3 sub between there ‘Here is bigger than there.’ (Sat) Finally, it is unclear whether Purepecha has ever had a comparative construction of inferiority. In order to express this domain, Purepecha employs two strategies: (a) to use the borrowed Spanish comparative construction of superiority with the negation, employing either the Purepecha relator (45a) or Spanish relator (45b), and (b) to borrow the Spanish construction with the Spanish degree menos (pronounced menuʃ in certain varieties), and the Purepecha calque relator eska (46), or the borrowed degree and relator (47). (45) a. Maria sáni=taru no wiŋapi-ʃ-ti eski thu. Maria few=more neg be strong-aor-ass3 sub 2 ‘Maria is weaker (less strong) than you are.’ (Maria is not stronger than you are) (Ar) b. Maria sáni=taru no wiŋapi-ʃ-ti ke thu. (Oc) (46) Selia menos yó-tha-la-ʃn-ti eska=ni. Celia less long-leg-ft-hab-ass3 sub=1 ‘Celia is shorter (less tall) than I am.’ (Cm) (47) Xi xatsi-ʃ-ka menuʃ ke thu wé-ka-ka. 1 have-aor-ass1/2 less sub 2 want-ft-subj ‘I have less than you want.’ (Jr)
Purepecha
479
9. Conclusion The grammatical MAT-loans are numerous, and appear within their Spanish grammatical constructions. A relevant phenomenon is the typological profile of Purepecha, which shows new tendencies. Purepecha is a synthetic–agglutinative language, and, nowadays, new analytic–periphrastic constructions appear, without modifying its elaborate morphological system, but revealing a structural rapprochement to the Spanish passive and equational constructions: two distinct structures (a morphological one and a periphrastic one) may simultaneously perform the same function. Purepecha is exhibiting language-internal grammaticalization processes to replicate Spanish models. There are PAT-influences that are not connected to any direct MAT-borrowing. Other PAT-influences are the comparative constructions, linked to MATborrowing in some varieties, and showing only pattern reduplication in others.
Abbreviations aor aorist app applicative Ar Arantepacua ass assertive caus causative Cc Cocucho centrif centrifuge centrip centripetal Cm Comachuén Cn Cuanajo com comitative dem demonstrative dim diminutive evid evidential foc focus ft formative fut future hab habitual Ih Ihuatzio inf infinitive inst instrumental
int interrogative it iterative Jr Jarácuaro loc locative mid middle neg negation obj object Oc Ocumicho orig origin partpp patient-oriented participle passiv passive pas past Pc Pacanda pl plural posp kinship possessive pos possessive pred predicativizator prog progressive ref reflexive res residential Sat San Andres Tzirondaro
480 subj sub
Claudine Chamoreau subjunctive subordinating conj.
transf Tr
transference Tirindaro
References Capistran, Alejandra 2002 Variaciones de orden de constituyentes en p’orhépecha. Topicalización y focalización. In: Paulette Levy (ed.), Del Cora al Maya Yucateco: Estudios lingüísticos sobre algunas lenguas indígenas mexicanas, 349402. México: UNAM.. Chamoreau, Claudine 2002a Le système phonologique du purepecha: Une étude en synchronie dynamique. Travaux du SELF IX: 133161. 2002b Dinámica de algunos casos en purepecha. In: Zarina Estrada Fernández and Rosa Maria Ortiz Ciscomani (eds.), VI Encuentro Internacional de Lingüística en el Noreste, 271290. Hermosillo: Unison. 2005 Dialectología y dinámica: reflexión a partir del purépecha. In: Claudine Chamoreau (Coord.), Trace 47: Dinámica lingüística, 6181. México: CEMCA. 2006 En busca de un verbo “ser” en purépecha: Cadena de gramaticalización y gramaticalización en cadena. In: Rosa Maria Ortiz Ciscomani (ed.), VIII Encuentro Internacional de Lingüística en el Noreste, 6584. Hermosillo: Unison. 2007 Looking for a new participant: Passive in Purepecha. In: Zarina Estrada Fernández, Soren Wichmann, Claudine Chamoreau, and Albert Álvarez González (eds.), Studies in Voice and Transitivity. Munich: Lincom. Gilberti, Maturino 1987 Arte de la lengua de Michoacán. Morelia: Fimax. (First publ. 1558, Mexico: Juan Pablo Impresor.) Haspelmath, Martin 2004 Passive participles across languages. In: Barbara Fox and Paul Hopper (eds.), Voice: Form and Function, 151177. Amsterdam/Philadelphia: John Benjamins. Villavicencio, Frida 2006 P’orhépecha kaso sïrátahenkwa: desarrollo del sistema de casos del purépecha. México: CIESAS-COLMEX.
Grammatical borrowing in Imbabura Quichua (Ecuador) Jorge Gómez-Rendón
1. Background Imbabura Quichua (henceforth IQ) is a Quechua language spoken in the Northern Andes of Ecuador by approximately 150,000 speakers.1 The province of Imbabura ranks second among the nine Quichua-speaking provinces of the Ecuadorian Andes as for the number of speakers (Haboud, 1998: 9192). Imbabura shows also the largest number of bilingual Quichua–Spanish speakers in the country (Büttner, 1993: 48: 49). Although there are a small number of IQ monolinguals among elders, the tendency nowadays is towards increasing levels of bilingualism accompanied with the maintenance of the native language. IQ has been in contact with Spanish since the second half of the sixteenth-century in a diglossic relation. The language is vigorously spoken in most Indian settlements of the province at community and family levels. It is taught in schools as part of the Bilingual Intercultural Education Programme implemented by the Ministry of Education since 1986 with the support from international cooperation agencies and the National Indian Organization (CONAIE). In the last decades IQ has entered oral media, and regular radio broadcasting in IQ reaches all the corners of the province. The language has a unified writing system since 1980, but this is used only for textbooks of elementary education. The fact that IQ shows a strong vitality in the Ecuadorian Andes should not obscure its allochthonous origins. Quechua was brought to Ecuador by the Incas in the second half of the fifteenth century, although another Quechua variety was spoken as a lingua franca by autochthonous peoples long before. From archaeological and early historical evidence it appears that one Barbacoan language – Cara – was spoken in the present territory of Imbabura at the time of the Inca invasion. It is likely that IQ speakers were in contact with other languages of the same family – Tsafiqui and Awa Pit – through an extensive trade network at work until the second half of the seventeenth century (Caillavet 2001: 81). The present chapter focuses on a one-to-one borrowing
482
Jorge Gómez-Rendón
situation between IQ and Spanish. Contact phenomena due to substratum influence are mentioned only occasionally.
2. Phonology The phonological inventory of pre-contact IQ does not include consonants /b/, /d/, /g/, /β/ and /z/, nor medial vowels /e/ and /o/. These sounds entered the language through Spanish loanwords (e.g. kaβažu ‘horse’, didu ‘finger’, pagana ‘to pay’, poste ‘post’) and do not occur phonemically in native items. They show a high degree of integration in IQ phonology, as observed by Cole (1982: 199). A situation that has facilitated the incorporation of these sounds in the native inventory is the fact that, with the exception of /β/, they have native allophonic counterparts: thus, [b] is an allophone of /p/, [d] of /t/, and [g] of /k/, all in nasal environments; similarly [e] is an allophone of /i/ and [o] of /u/. The result is free allophonic variation in some Spanish borrowings. Typically, Spanish medial vowels are raised (/e/>/i/, /o/>/u/) or otherwise pronounced as close as possible to their Quichua equivalents. Partial assimilation is more common in words with several medial vowels (e.g. [prizidínti] ~ [prisidínte] < Sp. /presidénte/), although non-assimilated borrowings are not uncommon. Accordingly, there may be various ways to pronounce one and the same word. Different phonetic realizations depend on (i) the environment, (ii) the speaker’s level of bilingualism, and (iii) the frequency of the word.2 When ambiguities arise, these are solved by several mechanisms: e.g. misa (Sp. mesa ‘table’) and misa (Sp. misa ‘mass’) are disambiguated by the voicing of the intermediate sibilant in the second member. The assimilation of borrowings is not always rule-governed and may be idiosyncratic to a certain degree. Other contact phenomena in phonology are found at syllabic and suprasegmental levels. According to the native pattern, the main stress falls on the penultimate syllable.3 The stress pattern in borrowings depends on their degree of assimilation (e.g. kumunidá, Sp. comunidad). The retention of Spanish stress patterns may be a disambiguating strategy in some cases. The native pattern of syllable structure is CVC(V), with a limited number of consonants in coda position (/k/, /s/). Consonant clusters in onset and coda positions occur only exceptionally, often as a result of other morpho-phonemic processes. Few Spanish loanwords avoid consonants in coda position: e.g. rilújo, Sp. reloj ‘watch’. The most frequent type of clusters in Spanish loanwords
Imbabura Quichua
483
involve one of a set of plosives (/p, t, k/) plus a flap /ɾ/ like in prioste ‘sponsor of a celebration’, trabajo ‘work’, and crema ‘cream’. Loanwords with clusters in word initial position usually are not assimilated into IQ phonology – but the speaker’s level of bilingualism may be decisive. Occasionally, a vowel is inserted in between the plosive and the flap. This vowel is the same as the one following the cluster: e.g. koronika, Sp. crónica ‘story’. A final issue is the existence of certain phonetic realizations proper of IQ. These realizations make IQ different from other Ecuadorian Quechua dialects. They are claimed to come from substratum influence. In what follows I focus on the phonological survey conducted by Fauchois (1988). The main phonetic difference of Imbabura Quichua with respect to other varieties spoken in the Andean Highlands is the fricativization of plosives /p/ and /k/ in all positions except nasal (cf. supra). The resulting [f] and [j] differ in word-initial position from their aspirated counterparts [ph] and [kh] in the rest of Ecuadorian dialects, but also from their non-aspirated equivalents [p] and [k] in word- medial or word-final positions. Illustrative cases are pucuna ‘to blow’, realized as [fukuna] in Imbabura but [phukuna] in the central dialects (e.g. Tungurahua); upiana ‘to drink’, realized as [ufiana] in Imbabura but [upiana] in the Southern varieties (e.g. Saraguro, Azuay); cari ‘male’, [jari] in Imbabura, but [khari] in the central dialects (e.g. Cotopaxi); and reciprocal -naku-, realized as [-naju-] in Imbabura, but [naku] in the rest of the provinces. Voiceless [t] and voiced [d] are distinct phonemes in all Ecuadorian dialects but not in IQ, where [d] occurs mostly in allophonic variation with [t] (Fauchois 1988: 62). There exist a number of lexical localisms and toponyms with the abovementioned phonetic features: e.g. muchiju, ‘Indian hat’; Abataj, name of an Indian community. Spanish loanwords are assimilated according to the same pattern: e.g. [juiřsa], from Sp. fuerza ‘strength’, realized as [Φuiřsa] in central and Southern dialects. These facts tell us that we are before a phenomenon of substratum influence in IQ phonology. Recent research has shown that a distinct cultural and linguistic group lived in the present territory of Imbabura. There being no grammars or dictionaries available of this language, most works have focused on toponomy, anthroponomy and early Colonial documents. A short list of morphemes – both of lexical and grammatical nature – have been identified from the substratum language (Caillavet 2001: 108). Interestingly, some of them show phonetic patterns similar to those described above: [-pixal] ‘sinuosity in the landscape’; and [-tux] ‘characteristic of a burial place’. Further research is required in this field.
484
Jorge Gómez-Rendón
3. Typology Contact with Spanish has not changed the typological profile of IQ. Like other members of the Quechua family, IQ remains a typical agglutinating language. This is true not only of sociolects with minimal lexical influence from Spanish but also of relexified varieties such as Media Lengua (GómezRendón 2005). What makes IQ – and Ecuadorian dialects in general – different from other Quechua dialects is a lower degree of synthesis resulting from the loss of verb-object agreement and possessive nominal suffixes. Consider the following examples from the Peruvian varieties of San Martin and Junín in comparison with Imbabura and Ecuadorian Quechua: (1)
San Martín Quechua Ñuka-ka maka-yki. 1s-top hit-2.obj ‘I hit you.’
(2)
Junin Quechua maki-yki hand-2s. poss ‘your hand’
Imbabura Quichua Ñuka-ka kan-ta maka-ni. 1s-top 2s-acc hit-1s ‘I hit you.’ (Cole 1982: 6)
Ecuadorian Quechua kanpak maki 2s.gen hand ‘your hand’ (Cerrón-Palomino 1987: 200)
The loss of personal reference markers in IQ introduced the obligatory use of pronominal forms to mark the arguments of the predicate where other dialects use pronouns only for emphasis (cf. Section 6.2). This particular development cannot be attributed to contact with Spanish or substratum influence. The simplification of verbal morphology in Ecuadorian dialects may be interpreted as a result of koineization. Cusco Quechua was brought to present Ecuador alongside other dialects from central and Northern Peru. The presence of different dialects contributed to the emergence of a koiné (Cerrón Palomino 1987: 343), a process claimed for other peripheral areas of the Inca Empire such as Salta and Tucumán in Argentina (de Granda 2001: 207ff). No contact phenomena have been observed in the type of alignment and affixation. Nevertheless, IQ has incorporated a few Spanish morphemes, mainly through the borrowing of Spanish words with such morphemes. These include agentive -dur and diminutive -itu.4 Consider the following examples from IQ:
Imbabura Quichua
(3)
a. mam-ita mother-dim ‘dear mother’ b. huas-ita house-dim ‘little house’
(4)
a. midi-dur measure-ag ‘meter’ b. ñaupa-dur go.ahead-ag ‘representative’
485
The diminutive ending and the agentive marker occur both with Spanish lexemes (left column) and native lexemes (right column). Apart from these grammatical changes, IQ shows a large number of Spanish borrowings are assimilated into native patterns. In the classification of parts of speech elaborated by Hengeveld et al (2004), IQ is considered a language with two lexical classes, i.e. verbs and non-verbs. The class of nonverbs conflates nominal, verbal and adjectival functions. Spanish borrowings in IQ tend to match this pattern. Spanish nouns used as modifiers of noun phrases and verb phrases are not uncommon. I discussed the results of an investigation into the functional patterns of lexical borrowing elsewhere (Gómez-Rendón 2006). In spite of the considerable influence of Spanish on IQ lexicon and syntax, IQ continues to be a topic-prominent language. The drop of the topic marker -ka and its replacement with the focus marker mi is a common feature. However, this new development does not imply any loss of topic prominence (see Section 7.1).
4. Nominal structures Nominal structures influenced by Spanish have to do with case marking and NP structure. Borrowing of linguistic matter is present, though replication of patterns is the most frequent phenomena. In the following sections I discuss these structures in detail.
486
Jorge Gómez-Rendón
4.1. Case marking Contact-induced phenomena in the use and the semantics of case markers include: (i) the loss of distinction between inalienable and alienable possession; (ii) the loss of distinction between comitative ntin and instrumental wan, with the resulting conflation of both in the latter; (iii) the drop of the obligatory accusative marker on direct objects; (iv) the increasing tendency to use the plural marker on nouns after numeral modifiers; (v) the use of Spanish lexical borrowings to express local and spatial relations. The loss of distinction between alienable and inalienable possession is reflected on the gradual replacement of yuc with pac and on the alternative use of lexical strategies (5). In both cases, the use or non-use of yuc makes a difference from pre-contact IQ varieties. (5)
Ñami warmi-yuc ka-ni. already woman-poss be.1s ‘I am married already.’
(6)
Ñami kazara-shka ka-ni. already married-ptcp be-1s ‘I am married already.’
IQ has different markers for the comitative (ntin) and the instrumental (wan). Whereas ntin relates elements as if they formed one indivisible unity, wan indicates the contingent bringing together of two elements or the instrumentality of one with respect to the other. For Kaarhus (1989) the comitative– instrumental distinction entails a unique understanding of space-time relations proper of the Quichua culture. As a matter of fact, heavily Hispanicized sociolects of IQ have lost this distinction in either of two ways: (a) both case markers are used interchangeably; (b) one marker (wan) conflates both meanings. The second case, illustrated in (7) below, is far more frequent and has resulted in the reduction of the case system on the model of Spanish. (7)
warmi-wan tarpu-ngapak ri-rka-ni. woman-com sow-purp go-pst-1s ‘I went with a woman to sow.’ ‘I went with my woman to sow.’
Imbabura Quichua
487
The two meanings of (7) cannot be disambiguated without context. The use of wan and ntin resolves this ambiguity. By using ntin the speaker implies that the woman is his wife and both of them form a couple; the use of a genitive pronominal such as ñuka(pak) ‘my’ is needless in this case. Another structure influenced by Spanish is the marking of direct-object arguments. As a rule, pre-contact IQ marks direct objects with ta. On the contrary, contact varieties tend to drop this marker. Consider the following sentences lacking the accusative marker: (8) churamu-kri-n shuc ley. put-inch-3 one law ‘He/she is going to pass a law.’ (9) kankuna huasi-pi kati-nchi kay programa. 2pl.poss house-loc follow-1pl this program ‘At your home we listen to the program.’ (10) chari-nchi minimercado “Charito”. have-1pl small.market “Charito” ‘We have the small-market [called] “Charito”.’ (Fauchois 1988: 117; my glosses) In the foregoing examples the accusative marker is systematically dropped on the direct objects, which contain either Spanish borrowings (8), (9) or code switches (10). Besides, the word order is SVO and not SOV as typical of IQ. There seems to be certain connection between dropped accusative markers, deviant word orders and heavy lexical borrowing. Fauchois (1988: 117) claims that the use of Hispanicized SVO word order in IQ makes it unnecessary to mark direct objects because the element following the verb is always the object. What Fauchois fails to notice however is that post-verbal position is not assigned to objects by default and the identification of this position with objects is possible only by contrasting Spanish-like word order and IQ native word order. To this extent SVO is subsidiary to SOV and the latter remains the most frequent word order, even in contact varieties. Notice that Spanish lexical material reinforces the tendency to drop the accusative marker in SVO constructions. Other tendencies observed in contemporary IQ that may be explained in terms of contact with Spanish concern the expression of number. Plural marking in IQ is obligatory, except if numerals precede the noun heads (Cole 1982:
488
Jorge Gómez-Rendón
128). The preference in such cases is the unmarking of number. Nevertheless, marking plurality is increasingly frequent when numerals are involved, as exemplified by (11) in comparison to (12) below: (11) ñuka-ka ishkai churi-kuna chari-ni. 1s-top two son-pl have-1s ‘I have two sons.’ (12) ñuka-ka ishkai churi chari-ni. 1s-top two son have-1s ‘I have two sons.’ Double plural marking is another contact-induced development concerning number. As a matter of fact, several Spanish words have been borrowed into IQ in plural (13). In this case the Spanish plural ending -s and the Quechua plural -kuna co-occur on the same lexeme, which results in apparent double marking. However, not all cases of double marking may be interpreted in this way for several reasons. Firstly, the number of lexemes borrowed in plural is comparatively small. Secondly, some borrowings occur with or without the Spanish plural. Third, cases are found of native lexemes in which the Spanish plural ending occurs along with the native marker (14). (13) chay kosa-s-kuna-manta mana japi-ni-chu. that thing-(Sp)pl-(IQ)pl-abl neg understand-1s-neg ‘I don’t understand about those things.’ (14) riku-shka-ni kimsa alku-s-kuna Aguchu-pak patiyu-pi. see-prf-1s three dog-(Sp)pl-(IQ)pl Aguchu-poss backyard-loc ‘I saw three dogs in Aguchu’s backyard.’ It may be argued that Spanish kosas in (13) is an instance of code-switching rather than borrowing proper. This explanation however fails to explain the occurrence of the native plural. If the code-switch is in Spanish, why does IQ plural occur at all? Accepting two plural markers in one and the same noun phrase implies two competing grammars.5 On the other hand, if we consider (13) as a frozen borrowing, we have to explain the large number of borrowed lexemes with the Spanish plural, a number that goes far beyond the few examples presented in the literature (Cole 1982: 129). In all, the occurrence of double marking seems to be accountable in terms of borrowing rather than of
Imbabura Quichua
489
code-switching. The occurrence of lexical and grammatical couplets consisting of native and borrowed items with different functional distributions (cf. Brody 1987; Campbell 1993; Matras 1998) might offer an alternative explanation in which pragmatic and processing factors motivate double marking. Spanish influence on IQ case marking concerns also the expression of local relations. Two Spanish lexemes occur in local relations: ladu, ‘side’; and frinti, ‘front’. The former lexeme is especially productive in IQ. Consider the following examples: (15) kuanchi Kasko ladu kidana-ju-nchi pruyektu-wan. 1pl Casco side remain-dur-1pl project-com ‘We [the people] from Casco kept the project.’ (16) kay Imbabura ladu gente. dem Imbabura side people ‘People from this side of Imbabura.’ (17) kay Topo ladu-kuna-pak-mi siyimpre obligatorio. dem Topo side-pl-ben-emph always mandatory ‘That was always mandatory for [people] from Topo.’ (18) maijan ladu-man-shi Anglango ka-pa-rka. which side-all-dub Anglango be-hon-pst ‘On which side was Anglango?’ (19) kuanchi-ka ladu-lla kausa-shka-nchi. 1pl side-lim live-prt-1pl ‘We lived on the side.’ (20) ishkay warmi-kuna kaballu-ta chumbi-wan ladu-ladu alza-n. two woman-pl horse-acc belt-inst side.by.side lift-3pl ‘Two women, one on each side, lift the horse with the belt.’ The above examples can be classified according to the use of ladu: (a) those in which ladu modifies the head of a noun phrase, be it a pronoun (15) or a noun (16); (b) those in which ladu stands on its own, being the head of the noun phrase itself and receiving inflectional morphology (17); (c) those in which ladu is part of a postpositional phrase and accompanies a question word (18); and (d) those in which ladu modifies the main predication, either
490
Jorge Gómez-Rendón
alone (19) or in reduplication (20). In addition, ladu has an ablative meaning in (15) to (17). In these constructions the Spanish borrowing links the noun head (implicit or explicit) to another noun indicating location. Interestingly, ladu take postpositions such as manta (ablative), pi (locative), man (allative), and ta (prolative). From the difference between local relations expressed by locatives such as ladu ‘side’, in which the preceding noun does not take the possessive, and local relations expressed by native locatives such as chaupi ‘middle’, in which it does, Cole (1982: 124) concludes that expressions such as (15) to (17) are complex postpositions, one of whose components is the locative. In general, ladu may be considered a secondary locative morpheme, as it behaves exactly like other members of this class (e.g. uku ‘inside’, washa ‘behind’ or jawa ‘above’). Other non-relational uses of ladu include its use as head and modifier of noun phrases, in the latter case with the meaning of “lateral”. Finally, ladu can be adverbialized by lla or by reduplications as typical of many IQ adverbials. Though much less frequent when compared to ladu, Spanish frinti ‘front’ is used in IQ to express anterior location similarly to the native morpheme chimba ‘front’. The occurrence of frinti is rather idiolectal, however. The following example is one of the few in the corpus: (21) chay wambra-ka pungu frinti-pi shaya-ju-n. dem young-top gate front-loc stand-dur-3 ‘That youngster is standing in front of the door.’
4.2. NP structure Contact with Spanish has influenced NP Structure in the following ways: (a) the use of determiners shuk ‘one’ and kay ‘this’ to replace the native topicalizer ka, which is dropped systematically in decontextualized speech events such as radio broadcasting; (b) the occurrence of Spanish diminutive and augmentative endings in borrowed and native lexemes; (c) the borrowing of the Spanish agent nominalizer. While the first phenomenon may be classified as pattern borrowing, the last two are cases of matter borrowing. The prolific use of determiners shuk and kay at expense of topicalizer ka6 was first noticed for IQ in radio broadcasting (Fauchois 1988: 105). Interestingly, this use is found beyond the context of broadcasting. Consider the following examples:
Imbabura Quichua
491
(22) shuk gallo-mi Katacupamba-pi-ka kanta-na. one rooster-emph Katacupamba-loc-top crow-hab ‘A rooster used to crow at Katakupamba.’ (23) shuk tela tiya-n ni-k ka-ria-n. one cloth there.be-3s say-pst.hab be-dur-3s ‘They used to say that there is one cloth.’ (24) kuanchi-ka kai kosecha-pi-ka puri-rka-ria-nchi. 1pl this harvest-loc-top go-pst-dur-1pl ‘We used to go to the harvest.’ (25) primero trata-ngapa kai kabesilla-ta trata-rka-ni. first negotiate-purp this leader-acc negotiate-pst-1s ‘First, in order to negotiate, I negotiated with the leader.’ Originally, kay is a demonstrative while shuk is a numeral. Shuk (indefinite) and kay (definite) are used on the model of the Spanish contrast between indefinite un/una and definite (el/la) articles. This is only half of the explanation, though. Fauchois (1988: 106) identifies three factors leading to the overuse of kay: (1) the influence of Spanish structure on IQ whereby the speaker expresses definiteness or indefiniteness through an element (demonstrative, numeral) preceding the noun head; (2) the speaker’s difficulty to use the topicalizer ka in non-personal speech events; and (3) the need to codify additional information in the absence of extra-linguistic signs. While the first factor is clearly at work, neither the second nor the third are relevant for the examples presented here, because these were gathered in normal communication settings. Notice that the topicalizer does occur in (24). The co-occurrence of the topicalizer with the demonstrative implies that the former does not mark definiteness. Definiteness in IQ is a by-product of topic marking.7 Spanish augmentative and diminutive endings are typically used in loanwords though occur on native lexemes as well. The use of Spanish augmentative and diminutive endings has not motivated the disuse of their native counterparts (augmentatives sapa and siqui; diminutive ku). On the contrary, the Spanish ending and the Quechua affix are sometimes used contrastively in couplets. Consider the following examples, each with a lexeme of different origin.
492
Jorge Gómez-Rendón
(26) ñuka-ka shuk wawitu-lla chari-ni. 1s-top one child: dim-lim have-1s ‘I have only one little child.’ (27) Uyanza tiyimpu ñukanchi-ka papasu-wan puri-shka-nchi. Uyanzas time 1pl-top father:supl go-plus-1pl ‘In times of Uyanzas we used to go with our grandfathers.’ In (26) the Spanish diminutive ending occurs on IQ wawa ‘child’. The diminutive emphasizes how young the child is. In (27) the augmentative on Spanish papa ‘father’ does not denote any quality of the speaker’s father. Instead, it refers to the speaker’s grandfather. The low frequency of this compound indicates that grammatical borrowings are used productively even though they do not necessarily follow IQ rules. In order to form kinship terms for generations older than ego’s parents, IQ uses modifiers jatun ‘big’ or rucu ‘old’ and not augmentatives. IQ uses augmentatives on quality nouns only. These differences suggest that borrowing implies a compromise between the morphological strategies of both languages. The Spanish agentive nominalizer dur often occurs unanalyzed on borrowed lexemes such as bindidur ‘seller’, trabajadur ‘worker’ or mididur ‘meter’. In these cases it forms an indivisible unit with the root. It occurs also on native lexemes: (28) a. ñaupa-dur front-nmlz ‘spokesman’
b. michi-dur graze- nmlz ‘shepherd’
(29) a. kalpa-dur run- nmlz ‘runner’
b. yapu-dur guagra plow-nmlz cow ‘plowing ox’
While the productiveness of this Spanish nominalizer is limited in IQ, the use of compounds is attested across generations and levels of bilingualism.
5. Verbal structures The influence of Spanish on IQ verbal structures includes matter and pattern borrowing, with predominance of the former given the easy integration of
Imbabura Quichua
493
Spanish loan-verbs, modals and particles into IQ. At the same time, the occurrence of pattern borrowing in valency-changing devices and, most importantly, the gradual replacement of the nominalization strategies with finiteclause subordination are modifying the typological outline of IQ.
5.1. TMA marking An inventory of tma structures influenced by contact with Spanish includes: (i) the replacing of ngapaj with purposive chun in coreferential constructions, on the model of the Spanish subjunctive; (ii) the use of Spanish dizi- ‘say’ in reportatives and quotatives; and (iii) the use of Spanish modal verbs. In his grammar of Imbabura Quichua, Cole calls our attention to the fact that “clauses employing the verbal suffixes -ngapaj and -chun are used in roughly the same environments in which the present subjunctive is employed in Spanish” (Cole 1982, 157). The statement can be interpreted simply as a comparison to help the reader grasp the nature of these suffixes but the implications go beyond that. In order to understand this new development in IQ morphosyntax, it is necessary to remind the reader of two structural properties of IQ. All Quechua dialects use a marker of purpose in subjunctive noun clauses. Imbabura Quichua has two different markers: ngapaj, for coreferential subjects in the main clause and the subordinate clause; chun, for subjects with different referents. The following examples extracted from Cole (1982: 37f) are illustrative: (30) muna-y-man ñuka mama-ta riku-ngapaj. want-1s-cond 1s.poss mother-acc see-purp8 ‘I want that I see my mother; I want to see my mother.’ (31) muna-ni Juzi pay-paj mama-ta riku-chun. want-1s Juzi 3-poss mother-acc see-purp ‘I want José to see his mother.’ Although purposive constructions occur mainly as complements of volition verbs like muna ‘want’, the same restrictions of coreferentiality apply for other verbs. Purposive constructions in contemporary IQ do not follow this pattern. Nowadays chun tends to replace ngapak in coreferential environments. Consider the following example in which such replacement takes place in such environment:
494
Jorge Gómez-Rendón
(32) atribi-k turiro-kuna tiya-shpa-ka paykuna toria-chun. brave-nmlz bullfighter-pl be-ger-top 3pl fight-purp ‘If there were brave bullfighters to fight.’ Concurrently, the use of ngapaj becomes restricted to exclusively purposive functions. Considering the frequent use of the Spanish subjunctive and the lack of distinction between coreferential subjects in this language, it may be hypothesized that long-term contact with Spanish led to the specialization of native morphosyntactic structures. The possibility of an internal development should not be entirely discarded, but the duration and intensity of contact along with high rates of bilingualism in the speech community make a contact-induced change more likely. Of several Spanish utterance modifiers in IQ, two are used as particles with modal nuances. Consider first the following examples of Spanish tálbis ‘perhaps’: (33) chayka manchari-shpa tálbis uya-ria-nga tayta-kuna-ka. then fear-ger perhaps hear-dur-fut parent-pl-top ‘If the parents fear [the penalties], they might listen [to the teachers].’ (34) chay-pash-chari tálbis asha-gu pay-kuna-pak-pash that-adit-perhaps perhaps few-dim 3-pl-dat-adit falta-k riku-ri-n. be.missing-nmlz see-refl-3 ‘They may see that quite a few things are still missing.’ Tálbis in (33) and (34) is a phonetic assimilation of Spanish tal vez ‘perhaps’. This particle marks probability from an epistemic (33) or alethic modality (34). Notice that tálbis co-occurs with its native counterpart chari in (34). The particle may occur freely in the clause (e.g. 35). Another particle derived from Spanish is gulpi. On the one hand, the widespread use of this form across idiolects suggests it is an older borrowing. On the other hand, the difference in the meanings of gulpi in Spanish and IQ suggests a process of grammaticalization. In fact, the meaning of this particle in contemporary IQ has no semantic relation to Spanish golpe ‘blow’, even if the phonetic shape is basically the same. Consider the following examples from our corpus:
Imbabura Quichua
495
(35) na llukchi-shka-nchik tukuy-lla-tak gulpi-ta trabaja-ngapak. neg leave-prf-1pl all-lim-aff blow-advr work-purp ‘Not all of us leave at once for work.’ (36) shinallatak gulpi-lla tukuy-ta ayuda-shpa trabaja-chun however blow-lim all-acc help-ger work-purp muna-y-manta-pash. want-inf-abl-adit ‘But we help all to work just because we want it.’ In the foregoing examples gulpi is used to stress the inclusivity of the firstperson plural. This interpretation is further confirmed by the co-occurrence of gulpi with native tukuy ‘all’ in both examples. Given that IQ has lost the clusivity distinction characteristic of other Quechua languages, gulpi serves in part to fill this gap in IQ. Notice that gulpi is adverbialized in (35) and qualified for degree in (36). In the following example gulpi co-occurs with tukuy and functions as an adverbial (intensifier) without any additional morphology. (37) gulpi tukuy tandanaju-shpa llanka-na ka-n yani. blow all unite-ger work-inf be-3 think:1s ‘It is necessary that all of us work together.’ Another phenomenon with bearing on tma marking involves the borrowing of Spanish-derived verbs minishti ‘need’, kiri ‘want’ and pudi ‘be able to’ as shown in the examples below: (38) komuna-kuna-wan-ma ashtawan trabaja-na minishti-nchik. village-pl-com-aff more work-inf need-1pl ‘We need to work more with the villages.’ (39) na inkipash problema-kuna-ta tini-ngapa kiri-ni. neg whatever problem-pl-acc have-purp want-1s ‘I do not want to have any problems whatsoever.’ (40) utru iskuila-kuna-pi-pash problema-ta tiya-shka-manda other school-pl-loc-adit problem-acc to.be-prf-abl mana aprindi-i pudi-n. neg learn-inf be.able-3 ‘They cannot learn because they had problems in other schools as well.’
496
Jorge Gómez-Rendón
(41) trabaxu-pak-ka ñukanchi piko-lla-mi minishtiri-n. work-dat-top 1pl.poss pickaxe-lim-aff be.needed.3s ‘Our pickaxe is needed for work.’ Examples (38), (39) and (40) are modal verb constructions in which Spanishderived minishti ‘need’, kiri ‘want’ and pudi ‘can’ are used as auxiliaries of necessity, volition and ability, respectively. On the contrary, sentence (40) shows the same verb minishti ‘need’ used as a non-modal verb reflexivized with the suffix ri, hence its intransitive interpretation. The phonological shape and etymological origin of minishti make it a loan-verb of older import in IQ. The verb form comes from archaic Spanish menester ‘need’ as occuring in constructions like haber menester ‘to be needed’. These constructions are not used anymore in local Spanish but were used until the late eighteenth-century. Pudi and kiri are of much later import. The above examples also show that Spanish-derived verbs take IQ inflectional morphology like any native verb. A final issue to be dealt with in this section concerns evidentiality. IQ and other Ecuadorian Quechua dialects show a system of evidential values that include one type of first-hand information and three types of second-hand information including reportativity, quotativity and inference.9 Although Spanish has not influenced the structure of evidential values in IQ (but see 7.1), one case of matter borrowing is attested which consists in the replacement of the native reportative/quotative form ni ‘say’ with the Spanish verb root dizi ‘say’. This replacement is reported only for the speech of younger bilinguals. The following examples illustrate evidential and non-evidential uses of dizi: (42) Quotative evidential chayka kutichi-n “estoy buscando mi yunta de bueyes” dizin then answer-3 [I am looking for my yoken of oxen] QUOT ‘Then he/she answers, “I am looking for my yoke of oxen”.’ (43) Reportative evidential patrun da-shca rumi-ka kuri ka-shka dizin. landlord give-ptcp stone-top gold be-prf rep ‘It is said that the rock the landlord gave [to him] was gold.’ The above examples illustrate the use of Spanish-derived dizi in a variety of contexts. The semantic equivalence between the root of the loan-verb and the native root ni allows for the replacement of dizin in (42) and (43) with IQ ni without change of meaning. As typical of IQ evidentials, dizin occurs at the
Imbabura Quichua
497
end of the quote in (42) and the clause (43). Moreover, both instances of dizin carry the same tense marker of the main clause. All of these features show that ni and dizi are semantically and morpho-syntactically equivalent.
5.2. Integration of Spanish loan-verbs The integration of Spanish loan-verbs in IQ occurs through direct insertion, i.e. without extra marking. Spanish verbs are borrowed as verb roots without infinitive endings. The resulting roots are often assimilated to IQ phonology. Example (44) below illustrates this strategy. Several roots are subject to further morphophonological changes such as elision (45) or epenthesis (46) of syllables in order to conform to IQ phonotactics. Loanwords dating back to the first century of contact with Spanish are particularly interesting. On the one hand, their Spanish origin often goes unnoticed by IQ speakers due to their degree of assimilation (cf. uya below). On the other hand, old loan-verbs have fallen into disuse in local Spanish (cf. parla below). Accordingly, the loan-verbs in (44) and (45) are invariably identified as Spanish borrowings by the speakers while the loan-verbs in (46) and (47) are considered part of IQ vocabulary. (44) valora-r: value-inf > balorañuka-manta ishka-ndi cultura-ta balura-ni 1s-abl two-com culture-accvalue-1s ‘As for me, I value both cultures.’ (45) acompaña-r accompany-inf > compaña-shunchi yamta accompany-1pl.fut firewood ‘Let’s go together to collect firewood.’
compañaminga-i-ta community.work-inf-acc
(46) casa-r(se): marry-inf(+refl) > kaza+rawambra-kuna ka-shpa-ka youngster-pl be-ger-top shuk paya paya warmi-ta kazara-ri-nga one old old woman-acc marry-refl-3.fut ‘Though he is young, he will marry a very old woman.’
498
Jorge Gómez-Rendón
(47) parla-r talk-inf > parlaoi-r hear-inf > uyachay-manta parla-shka-ta uya-pa-shka-ngi-chu that-abl speak-ptcp-acc hear-hon-prf-2s-int ‘Have you heard [people] talking about that?’ The integration of loanwords involves the re-semantization of source-language meanings, because loan-verbs are not always borrowed with the same meaning they have in Spanish. This is particularly true for the early stages of contact, where bilingualism among IQ speakers was incipient. In later stages Spanish-derived verbs usually match the source-language semantics. Thus, Spanish botar ‘throw’ was interpreted in the sense of ‘to give away’. Other examples include Spanish desbaratar ‘to mess up’ used in the sense of ‘to hurt’, or tratar ‘to treat’ in the sense of ‘negotiate’. An interesting phenomenon is the verbal use of loan nouns and adjectives as verbs in IQ. This tendency is rooted in the lexical flexibility of IQ. In the following example a Spanish noun is used verbally. (48) na kai llacta shina-ka flauta-k ka-shca-n-chu nin neg this village like-top flute-hab.pst be-plus-3-neg rep medio Camuendo-ta flauta-k kashka nin. similar Camuendo-acc flute-hab.pst be.plus-3 rep ‘It is said that they did not use to play the flute as it is proper of this village [but] in the style of Camuendo village.’ (49) primero jatun flauta tiya-na cutin uchilla-gu, Castilla flauta-gu. first big flute be.inf then small-dim Castilla flute-dim ‘First it was the big flute, then the small Spanish-like flute.’ In the above example the verbal marker of habitual past is added to the Spanish lexeme flauta ‘flute’. The same lexeme is used as a noun in (49). Notice that no derivation mechanism is involved in the verbal use of flauta in (48). The trans-categorization of Spanish borrowings is not uncommon in IQ, where one often finds Spanish nouns used as adjectives and adverbs, or adjectives used as nouns and adverbs. A detailed discussion of Spanish lexical borrowings in IQ is presented elsewhere (Gómez-Rendón 2006a).
Imbabura Quichua
499
5.3. Contact-induced valency changes Certain developments in IQ verbal morphology may be attributed to contact with Spanish. These include (a) the use of reciprocal naku as a plural marker for intransitive verbs; (b) the extension of reflexive ri to cover reciprocal meanings; and (c) the use of reflexive ri on the model of Spanish impersonal se. In what follows I discuss these developments and their possible motivation by Spanish contact. As Muysken (2000b: 984) notes, reciprocal naku is used with intransitive verbs denoting actions performed together with someone else. He gives the following example: (50) puri-na[k]u10-n walk-recp-3 ‘They walk together.’ Notice that puri-n is marked for person but not for number so it may refer to singular and plural subjects alike. In (50) puri-naku-n unambiguously refers to several persons walking together. The question is whether this particular development of IQ is induced by contact with Spanish. To explain this innovative use of the reciprocal in terms of contact-induced change we need to demonstrate that (a) Spanish reciprocals can be used also as plural markers, and (b) this use serves as a model for the IQ reciprocal.11 While the reciprocal-plural relation is demonstrated by the trivial fact that every reciprocal form implies several individuals and reciprocal morphemes in fusional languages like Spanish (se, nos, os) also show number, the particular use of the reciprocal in IQ does not necessarily follow from its contact with Spanish. Other Quechua languages with a similar history of contact (e.g. Argentinean Quechua) use naku as a reciprocal only (Alderetes 2002: 5).12 Accordingly, it may be hypothesized that Spanish triggered the innovative use of the reciprocal as a verbal plural marker on the basis of the common semantics of reciprocality and plurality. From this point of view the influence of Spanish would consist in expressing both in one morpheme naku instead of two,13 i.e. a case of pattern borrowing. The taking over by naku of an additional (plural) meaning seems to have caused the reflexive ri to include reciprocity. That such extension is a gradual process is demonstrated by the fact that it is not uncommon that both markers occur in one and the same verb (51):
500
Jorge Gómez-Rendón
(51) paykuna-ka yanka-manta maka-ri-naku-nkuna. 3.pl-top firewood-abl fight-ref-rcp-pl ‘They are fighting with each other because of the firewood.’ A step towards the replacement of the reciprocal is illustrated in (52) below, where the reflexive occurs instead of the reciprocal but requires a comitative marker to signal the common action of the verb ‘coordinate’: (52) na kunbeniu tiyanchu, purki ñukanchi kurdina-ri-nchi neg agreement there.be-neg because 1pl coordinate-ref-1pl uspital-wan. coordinate-ref-1pl hospital-com ‘There is no agreement, because we cooperate with the hospital.’ Arguably Spanish served as a model on account of the existence of one paradigm of verbal morphemes expressing reflexivity and reciprocity. However, like one must be cautious in formulating a hypothesis that links this development to contact only, as there may be other, internal changes at work. Contact with Spanish is therefore one of several influencing factors and should be understood as a trigger of change. Reflexive ri shows another innovative use in IQ and other varieties of Ecuador. Notice the use of ri in the following example: (53) ñaupak Sanjuan-ka siempre-mari obligatorio ka-na front Sanjuan-top always-aff obligatory be-hab ishkai pañuelo-ta binda-ri-shka. two handkerchief-acc bandage-ref-plus ‘In former San Juan [festivals] it was obligatory to blindfold [the horse] with two handkerchiefs.’ Consider the argument structure of binda ‘bandage’ (Sp. vendar) in (53). Originally, binda is a transitive verb with agent and patient arguments. The agent and patient of binda are implicit in (53). In the context of the story it is clear that the official sponsor of the festivals used to harness a horse, and that part of the animal’s apparel consisted of two handkerchiefs. From the participants in the story, we assume that the sponsor is the agent, the horse is the patient and the handkerchiefs are the instruments. However, this distribution of arguments does not correspond to nominal and verbal morphology in (53). For
Imbabura Quichua
501
one thing, the accusative marker ta indicates that “two handkerchiefs” is the patient. This interpretation does not contradict the structure of participants in the event. Rather, the use of ri is unexpected in this context, where it cannot be interpreted as reflexive or reciprocal. What function does ri perform? In my view the Spanish impersonal se give us some clues. In this language the impersonal pronoun se is homophonous with the reflexive–reciprocal pronoun se. Furthermore, both forms are often cliticized to the verb root. In this context, the second clause in (53) seems to be a calque from the Spanish impersonal construction in (54): (54) se vendaba dos pañuelo-s. impr bandage:pst two handkerchief-pl ‘They bandage two handkerchiefs.’ If we exclude the opposite word orders (Spanish VO versus Quechua OV), both clauses show a one-to-one equivalence. From this point of view, ri is neither a reflexive nor a reciprocal but expresses an impersonal agent just like Spanish se. The replacement of otherwise different morphemes for reflexive, reciprocal and impersonal with one and the same morpheme (ri) could be explained, satsifactorily in my opinion, by contact with Spanish, in which language one and the same paradigm serves the three purposes.
5.4. Clause linking: nominalization versus subordination One of the most important contact-induced changes in contemporary IQ is the increasing replacement of embedded nominalized constructions with hierarchical, Spanish-modelled subordinated clauses. The subordination strategy in IQ makes use of Spanish subordinators including: (a) relativizer que ‘that’ after verba dicendi; (b) relative pronoun lo-que ‘that (which)’; and (c) several conjunctions like purki ‘because’ or si ‘if’. In this section I focus on the subordination of complement clauses as objects of transitive verbs and leave the discussion of the other types for Section 7. The replacement of nominalization with subordination can be understood best if we compare example (55) with the corresponding embedded construction in (56):
502
Jorge Gómez-Rendón
(55) paykuna-lla shaya-shpa paykuna apa-shka-n lo-que 3pl-lim arrive-ger 3pl take-ptcp-3pl that-which muna-shka-n. want-prf-3pl (56) paykuna-lla chaya-shpa paykuna munashka-ta apa-shka-n. 3pl-lim arrive-ger 3pl want-ptcp-3 take-prf-3 ‘Upon their arrival, they took what they wanted.’ The use of the compound pronoun lo-que ‘that which’ as a clause linker has a number of effects on the morphosyntax of the IQ clause: (a) while the clause munashkata in (56) is embedded in the main clause, munashkan in (55) is postponed to the main clause and linked to it by the pronoun; (b) the embedded construction in (56) is marked by accusative -ta and falls within the scope of apa- ‘take’; (c) whereas the verb in (55) is finite, the verb in (56) is non-finite; (d) Quechua OV word order in (56) is replaced by Spanish-like VO in (55). The foregoing use of lo-que shows that lexical borrowing of function words may have important effects on the morphosyntax of the recipient language. Other studies have shown a similar impact of Spanish conjunctions and adverbials on the matrix of Meso-American languages such as Pipil (Campbell 1987) and Nahuatl (Hill and Hill 1986). I show further effects of Spanish function words on IQ in Section 7.
6. Other parts of speech Even if loan nouns and verbs make the bulk of Spanish borrowings in IQ, the contribution of other parts of speech is not trivial and has considerable effects on the structure of the language. This section analyzes loanwords from different parts of speech and discusses the extent of their influence on other levels of IQ structure.
6.1. Numerals and quantifiers While many IQ grammars and dictionaries boast a full set of native numerals from one to hundred, their use in spontaneous everyday conversation is limited to ten in the best of cases. Above ten, speakers use Spanish numerals
Imbabura Quichua
503
even though it is not uncommon to use Spanish numerals also for five to ten. Ordinal numbers come from Spanish without exception. The following example illustrates cardinal and ordinal numbers from Spanish: (57) iskuila-manda-ka llukshi-wa-rka ña trese añu-mi school-abl-top leave-1s-pst already thirteen year-val llukshi-rka-ni, kay kuartu gradu-manda ñuka-ka leave-pst-1 dem fourth grade-abl 1s-top ‘I quit school when I was thirteen, I was there since the fourth grade.’ Apart from Spanish numerals, which are ubiquitous in any type of speech genre, IQ has borrowed several Spanish quantifiers. Unlike numerals, loan quantifiers have not replaced their native counterparts and may co-occur with them in couplets for emphasis. The most frequent Spanish-derived quantifiers are tuditu ‘all’ and alkunu ‘some’. The following examples illustrate their use: (58) alkunus chicha maltaca chaypi-mi wardaria-na. some:(Sp)pl chicha beer-top there-loc-val keep-hab.pst ‘Some people used to keep chicha beer in there.’ (59) chaimanda-mi ñuka prisidente tukushpa. therefore-val 1s president become-ger alkunas kusas-kuna-ta allichi-shka-nchi. some:(Sp)pl things-pl-acc improve-prf-1pl ‘Therefore, when I became president, we improved several things.’ (60) tuditu-mi tukuy-lla-mari tiya-naku-rka-nchi, mikuna-gu-ta all-val all-lim-aff be.sitting-rcp-pst-1pl food-dim-acc miku-shpa tiya-ura-mari eat-ger there.be-when-aff ‘At lunch time every one of us was sitting together.’ (61) ñakutin impresa tuku-rka-nchi chaymanda kumpra-rka-nchi then business begin-pst-1pl therefore buy-pst-1pl tuditu asinda all estate ‘Then we started the business and bought all [the lands of] the hacienda.’
504
Jorge Gómez-Rendón
Spanish quantifiers may be used either as noun modifiers or pronouns. Notice that alkunu in (58) is a plural masculine pronoun while alkunas is a plural feminine pronoun in (59). Both forms have been borrowed along with the Spanish markers of number and gender. One might interpret alkunas in (59) as a code switch to the extent it is accompanied by another Spanish borrowing, i.e. kusas ‘things’. Both borrowings would thus form a noun phrase inserted in the morphosyntactic frame of Spanish. But this is not the case because IQ markers occur in the same phrase. In (60) and (61) tuditu occurs without Spanish or IQ morphology. In (60) tuditu co-occurs in a couplet with IQ tukuy ‘all’, with validator mi and affirmative mari emphasizing the idea of inclusivity in the noun phrase. Whereas examples (58) to (60) follow OV word order, the second clause in (61) has VO word order. Notice that no accusative marker occurs on the direct object tuditu jazinda in (61). Arguably, the co-occurrence of lexical borrowings eventually influences IQ morphosyntax.
6.2. Pronouns In the last section I showed that Spanish-derived quantifiers are used as pronouns in IQ. In this section I show that the influence of Spanish on the pronominal paradigm of IQ goes beyond. A well-documented change involving pronouns in IQ concerns the use of native adjective kikin ‘proper’ as a polite second-person pronoun. Politeness in IQ is marked on the verb by means of the honorific affix -pa-, as shown in (62) below. Consider the following exchange extracted from an interview: (62) a. maijan iskuila-pi-tak ka-pa-rka-ngi. which school-loc-aff be-hon-pst-2s ‘Which school did you go to?’ b. ñuka-ka iskuila T.H. kay kumunidad Uksha llakta-pi-mi 1s-top school T.H. dem community Uksha village-loc-val ñuka-ka yachaju-pa-rka-ni. 1s-top study-hon-pst-1s ‘I went to school T.H. in this village of Uksha.’ Compare this strategy with the use of pronoun kikin in (63), where it cooccurs with the honorific affix. Kikin can be used in plural and receive any of a set of nominal markers. Kikin is also the base form of the possessive pronoun kikinpa.
Imbabura Quichua
505
(63) kikin-ta tapu-gri-pa-ni may-manda-ta ka-pa-ngi, 2s.hon-acc ask-inch-hon-1s where-abl-int be-hon-2s kikin-pa shuti-gu-ta-pash willa-shpa ali ka-pa-ngi-man 2s.hon-poss name-dim-acc-adit inform-ger good be-hon-2s-cond ‘I am going to ask you where you come from and be so kind to tell us your name as well.’ This development is typical of IQ and other highland varieties of Ecuadorian Quechua (CIEI 1982: 107). Arguably, kikin developed in the early stages of contact with Spanish, when social relations between Spaniards and Indians were modelled on a hierarchy of castes. Nowadays kikin is falling into disuse, being preserved only in conservative sociolects. The reason to hypothesize a contact-induced change in this case is the existence of a pronominal paradigm based on politeness distinctions in local Spanish. The intensity of contact and the higher levels of bilingualism among IQ speakers are two influencing factors. Why a similar development is not attested in other varieties of Quechua remains an open question. The presence of Spanish in the pronominal paradigm includes the subset of interrogative pronouns. Spanish-derived ura(s) ‘hour(s)’ is suffixed to the interrogative lexeme ima ‘what’ to form the loan blend imauras ‘when’, ‘at what time’. The same form is used in indirect questions, as illustrated in (64): (64) paykuna-ka yacha-n imauras-mi yaku-ka chiri chiri 3pl-top know-3 when-val water-top cold cold ka-shka-ta-pash imauras-mi yaku-ka kunuc-lla ka-n. be-prf-acc-adit when-val water-top warm-lim be-3 ‘They know when the water is very cold and when it is warm.’ The use of imauras is widespread across generations and levels of bilingualsm, which leads us to assume a comparatively earlier introduction of this form. On the contrary, the use of a pronominal duplet involving first-person pronoun ñuka and cross-reference marker wa is a late result of contact. Consider the following example: (65) wambra kashpa makanaju-ria-ni wambra-pura-kuna y chay youngster be-ger fight-dur-1s youngster-com-pl and that ñuka-ta kashtiga-wa-ria-n. 1s-acc punish-1s.obj-hab-1 ‘When I was young, I used to fight with other boys and they [his parents] punished me.’
506
Jorge Gómez-Rendón
Duplicated structures involving person reference serve contrastive purposes. However, the double marking of person in (65) suggests no emphatic or contrastive readings. We cannot ascribe this type of double marking to contact with full certainty. The loss of verb-object agreement markers in Ecuadorian Quechua was internally motivated, for which reason it is safer to view Spanish as a trigger, not as a cause.
6.3. Discourse markers and adverbial particles The number of Spanish particles and discourse markers in contemporary IQ deserves special attention. A broad inventory includes the following categories: connectors, adverbial clause markers, time deictics and discourse markers. Examples of several of these function words were presented in different parts of this chapter. The discussion in this section focuses on connectors, time deictics and discourse markers. Spanish adverbial clause markers are addressed in Section 8. The borrowing of Spanish connectors includes additive y ‘and’, contrastives o ‘or’ and dino (from Spanish de no ‘if not’) and disjunctive pero ‘but’. These connectors are used to coordinate sentences and smaller constituents. IQ has an additive of its own but lacks a function word to express contrast and disjunction. Spanish y is ubiquitous in IQ discourse while additive pash continues to be used in a variety of contexts. Unlike the Spanish conjunction, pash cannot occur in the first conjunct. The following example from Cole (1982: 79) illustrates the difference: (66) (*y) ñuka(-pash) kamlla-ta gushta-ni; and 1s(-adit) toasted-corn-acc like-1s (y) ñuka pani(-pash) kamlla-ta gushta-n; and 1s sister(-adit) toasted.corn-acc like-3 y ñuka wawki(-pash) kamlla-ta gushta-n. and 1s brother(-adit) toasted-corn-acc like-3 ‘I like toasted corn, my sister likes toasted corn, and my brother likes it too.’ Examples of contrastive o and dino are given in Cole (1982: 80). This author considers both connectors equivalent. However, the following example from our corpus shows that this is not necessarily the case and very often both occur as a single conjunctive similar in meaning to the Spanish expression o
Imbabura Quichua
507
si no ‘unless/instead’. This is illustrated in (67), where marker ka marks the topic of the previous sentence. (67) kulki-gu-ta paykuna apamu-n money-dim-acc 3.pl bring-3 or else o dino-ka kuinta-man dipusita-mu-n. or if.not-top account-dat deposit-ctrp-3 ‘They bring the money home or deposit it in the account instead.’ The Spanish disjunctive pero ‘but’ is another connector of frequent occurrence in IQ. It differs from additive and contrastive connectors in that it links sentences only: (68) ashta yarijay yarijay ka-shkan-ka nin pero payka ali much famine famine be-prf-top rep but 3s-top good jatu-k kashkanga nin. sell-nmlz be-prf-top rep ‘It is said that there was a lot of famine, but it is said that he still sold well.’ The disjunctive often co-occurs with Spanish time adverbs such as intonses ‘then’, siympre ‘always’, nunca ‘never’ or antes ‘before’. It remains to investigate whether we are dealing here with borrowings or code switches. Some Spanish connectors have been incorporated to IQ discourse without assimilation. The lack of phonetic assimilation cannot be attributed to a recent history of borrowing but to the fact that these loanwords must be perceptually salient in native discourse. Spanish time adverbs in IQ include all the days of the week. Times of the day show a mixture of native and borrowed lexicon, as shown in Table 1. Other time adverbs from Spanish are aura ‘nowadays’ (< Sp. ahora), intonses ‘then’ (< Sp. entonces), and siympre ‘always’ (< Sp. siempre). These examples show Spanish adverbs used as time deictics in IQ (see (69)–(71)). Table 1. Times of the day in Imbabura Quichua Spanish-derived
Quechua native
English
mañana tařdi nuche
--chishi tuta
morning afternoon evening/night
508
Jorge Gómez-Rendón
(69) chayka chay ladu-kuna-man-lla-mi ashtaka siympre chiri chiri thus that side-pl-all-lim-val much always cold cold ka-na-ta yacha-n. be-inf-acc know-3 ‘So it is always very cold around those places.’ (70) intonses chay tanda-kuna-ta kara-k ka-rka genti-man, then that bread-pl-acc give-hab.pst be-pst people-dat genti-man, gañan genti-man, chai jatun tanda. people-dat hacienda.worker people-dat dist big bread ‘At that time they used to give those big pieces of bread to hacienda workers.’ (71) aura-pi-mari nachu fishta-kuna-pi-ka nachu today-loc-aff neg.int festival-pl-loc-top neg.int chay kuitis-shina-lla rucu-ta ninanta reventa-chi-n. that rocket:pl-like-lim old-acc much explode-caus-3 ‘Nowadays, in the festivals, they have lots of those old fireworks.’ Aura is different from other adverbs in that it occurs with further markers including locative pi (cf. 77), topicalizer ka and affirmative mari. Notice that the original Spanish lexeme ahora means ‘now’, ‘today’ and ‘nowadays’. Only the third meaning has been preserved in IQ. The other two are covered by the native lexeme kunan. Intonses can be used also as a discourse marker in sentence boundaries. In fact, the latter use is more frequent in IQ. When used as discourse marker, intonces does not refer to a specific point in time but signals a succession of events as shown in the following example: (72) intonses chaymantaka kunan banda Santa Marianita nishkami, chay patronpa asinda korredorpi, intonses chaypi tokanajuna hashta kolonpamba, intonses chayta ña karashka jipaka, hashta hashtami bailak kana, jari, huarmi, intonses karashka jipaka amozeras mote yanushcamantaima carana. ‘And then the band ‘Santa Marianita’ stayed in the front yard of the hacienda, and then they played very hard, and then men and women danced all together, and then the servants gave toasted corn to the people.’
Imbabura Quichua
509
Table 2. Spanish discourse markers in Imbabura Quichua Spanish-derived
Meaning
osea intonses buino este diai klaro
that is then well this and then of course
Compared to other function words borrowed from Spanish, discourse markers show a modest rank in terms of frequency. Table 2 includes some Spanish discourse markers in IQ from the most frequent to the least frequent. Occasionally discourse markers are followed by the Spanish conjunctive que. Examples of this are o sea que ‘which means that’ and claro que ‘of course (that)’. Given that the foregoing markers are prolific in local Spanish discourse, their frequency in this language explains partly their frequency in IQ. Nonetheless, the primary motivation of their prolific borrowing is the pragmatic dominance of the donor language with respect to the recipient language (Matras 1998: 285). This dominance is clear for the contact situation between Spanish and Quechua in the Andes, where both languages are in a diglossic relation in respect to each other, with Spanish as the dominant language and vehicle of literacy.
6.4. Adjectives and adverbs In this section I discuss adjectival and adverbial expressions of comparison to the extent of their influence by Spanish. Other issues concerning adjectives and adverbs such as the incorporation of loan adjectives along with number and gender markers or the borrowing of manner adverbs have been addressed in previous sections. Given that the system of parts of speech in IQ makes no distinction between adjectives and adverbs (Schachter 1985: 17; GómezRendón 2006), the following discussion is valid also for adverbial comparison, even if examples are not provided for lack of space. In his description of Imbabura Quichua, Cole (1982: 93) gives the following example to show the way in which adjectival and adverbial expressions of comparison is made in IQ:
510
Jorge Gómez-Rendón
(73) Tumas-ka [Marya-ta yali-j] ali trabaja-n. Tomás-top [María-acc surpass-nmlz] good work-3 ‘Tomás works better than María.’ In (73) the standard of comparison is an embedded clause nominalized by the agentive nominalizer -j while the subject of the main clause is the compared element. The basis of comparison is expressed by the main predicate. The connection between both elements is made by the verb yali ‘surpass’. However, the comparative construction in (73) is not the only one available in IQ, where it is associated with the most conservative speakers. Other constructions attested are illustrated below: (74) chay jipa-ka Gonzalo rura-shpa saqui-na [pero más claro]cs-Sp that after-top Gonzalo make-ger leave-hab [but clearer] Galo ashtawan14 yali-shpa rura-k ka-shka ni-n Galo more surpass-ger do-nmlz be-prf rep ‘Afterwards Gonzalo stopped making [the festival] but they say that Galo used to make better [festivals than Gonzalo].’ (75) siempre runa-kuna rikunaju-shka-nchi maijan hospital-mi always Indian-pl visit-prf-1pl which hospital-val ashtawan mas maltratoka, Otavalo hospitalmi more more mistreatment-top Otavalo hospital-val ashtawan-ga yali maltrato tiya-shka. more-top surpass mistreatment there.be-prf ‘We Indian people have always visited hospitals that mistreat patients, the hospital of Otavalo mistreat patients more [than others].’ (76) Pidro uchilla ka-n, tuditu wawa-kuna-ta gana -n Pidro little be-3 all child-pl-acc win-3 ‘Pidro is smaller than all the other children.’ (lit. Pedro is small, he wins all the children) The alternative constructions of comparison differ from the traditional strategy in several ways. Compare first (74) above. This construction differs from (73) in that the standard of comparison in (74) precedes the compared element without participating in the yali predicate. In addition, yali in (74) is not nominalized but subordinated by means of the gerund marker. The comparison between both elements is made explicit by a code-switched connective
Imbabura Quichua
511
phrase pero más claro ‘but clearer’. In example (75) the standard of comparison is implicit or inferred from the preceding discourse. Although ashtawan and yali occur in the second clause, the latter appears without extra marking. Finally, (76) shows an innovative construction where the clause containing the compared element and the clause containing the standard of comparison occur one after the other without a coordinator in between. In this construction the loan-verb gana ‘win’ has replaced native yali ‘surpass’. In this gamut of alternative constructions it is possible to trace a continuum from the traditional IQ construction in (73) to the most hispanicized structure in (76). Construction (76) has been reported also for Imbabura Media Lengua (Gómez-Rendón 2001: 197) and other mixed varieties in Ecuador (Muysken 1997: 397).
7. Constituent order and syntax The affluence of Spanish loanwords in IQ goes hand in hand with less visible changes at the levels of the clause (constituent order) and the sentence (syntax). Although syntactic developments are not necessarily explained by lexical borrowing, the co-occurrence of Spanish lexical borrowing and syntactic calquing on the model of Spanish suggests a close relation between these phenomena. In syntax pattern borrowing prevails over matter borrowing, even though the former often implies the latter. Thus, for example, subordinated constructions (instead of nominalized embedded clauses) imply the borrowing of Spanish conjunctions. Several issues related to word order have been addressed in previous sections and will be not discussed here. An inventory of syntactic contact-induced changes in IQ includes: (a) Spanish SVO word order in declarative sentences and the replacement of topicalizer ka with focus particle mi; (b) Spanish SVO word order in non-verbal predicative constructions with copulas; (c) an ongoing shift from relative clause–head to head–relative clause order mediated by interrogative pronouns used as relative markers; (d) question formation on the basis of unmarked declarative sentences with Spanish-like interrogative intonation contours; and (e) the borrowing of Spanish subordinators and the replacement of nominalized clauses with adverbial subordinated clauses. Apart from these undisputed contact-induced changes, there are other minor developments in IQ not included here on account of their limited frequency.
512
Jorge Gómez-Rendón
7.1. Word order in declarative sentences and non-verbal predications In IQ the verb always occurs in sentence-final position, being immediately preceded either by the subject in intransitive constructions or the object in transitive constructions. There exists a clear tendency nowadays to Spanishlike SVO word order associated with the drop of topicalizer ka and/or the replacement thereof with the focus particle mi. Consider the following examples from Fauchois (1988: 117): (77) kallari-naku-nchik shuk mushuk semana-ta. begin-recp-1pl one new week-acc ‘We all start a new week.’ (78) churamu-kri-n shuk ley. put-inch-3 one law ‘They are going to pass a law.’ (79) kankuna huasipi kati-nchi kay programa. 3pl house-loc follow-1pl this programme ‘We watch this programme at their place.’ Example (77) shows SVO word order but marks the direct object with accusative ta. On the contrary, examples (78) and (79) not only have Spanish-like SVO word order but also lack the accusative marker on the direct object. According to Fauchois “[the new SVO word order] is almost systematic if the object is a nonce borrowing” (1988: 117; my translation). A further syntaxrelated change induced by contact is the drop of the topicalizer and the eventual replacement thereof with the focus particle mi. This change is visible in non-verbal predicative constructions involving a copula in SV word order. Consider the following example from a interview: (80) bueno ñuka shuti-mi kapan Roberto ñuka-mi ka-pa-ni well 1s.poss name-val be-hon-3 Roberto 1s-val be-hon-1s Chaupi Inti Caluquí llacta-manta. Chaupi Inti Caluquí community-abl ‘Well, my name is Roberto Tocagón and I come from Chaupi Inti Caluqui.’
Imbabura Quichua
513
In (80) the interviewee uses SV word order and drops the topicalizer ka on the subject of both sentences, instead of which he uses the focus marker (ñuka shuti-mi; ñuka-mi). The replacement of the topicalizer is common in nonverbal predicative constructions involving copulas. According to Fauchois, the drop of the topicalizer and its replacement with the focus marker is due to the fact that Quechua lacks pre-established models to present the information in long-distance communication and makes use of Spanish models (1988: 119). However, the replacement in (80) occurs in contextualized face-to-face speech. Spanish influence is obvious but the effects go beyond syntactic calquing. Therefore, this change may be associated with a new structure of the evidential system. Given that mi marks focus and first-hand information, the use of this marker as a topicalizer results in the loss of evidential marking. This explanation should be supported with additional data to be conclusive.
7.2. Head–relative clause order and relative pronouns IQ lacks relative pronouns. Relative clauses are embedded nominalized constructions preceding their heads as illustrated in the following example: (81) llaki-yuk-kuna-ta kulki-ta tapu-shpa yanapa-k runa. problem-poss-pl-acc money-acc ask-ger help-nmlz man ‘Man who helps people with problems by asking them for money.’ The loss of nominalization strategies discussed in Section 5.4 has resulted in the creation of relative clauses following heads. Relative clauses and heads are linked by interrogative pronouns used as relative pronouns. Consider these examples: (82) tukui llakta-kuna-pi may kay ratu puñu-ku-n all village-pl-loc where this time rest-dur-3 ‘In all the village where people are sleeping now.’ (83) tauka mamita-kuna pi-kuna-mi kay mineros-pak several mother-pl who-pl-val this miner:pl-poss warmi-kuna ka-n. women-pl be-3 ‘Several mothers who are the wifes of these miners.’
514
Jorge Gómez-Rendón
(84) tandanakuy ima-ta rurashka kay kabildu. meeting what-acc make-prf this council ‘The meeting (that) this council celebrated.’ Different from adverbial clauses with Spanish conjunctions, the foregoing clauses use IQ pronouns in the function of subjects (83)–(84) and objects (84). These pronouns may be pluralized (83) or receive case marking (84). The verb in the relative clause is finite and receives tma marking. The resulting clause is closely similar to a Spanish relative clause (Fauchois 1988: 113). Whereas these constructions are typical of the speech of bilinguals, they are not uncommon in conservative idiolects. In Section 5.1 we discuss the use of Spanish verba dicendi in quotative constructions. A parallel development in contemporary IQ is the use of Spanish relativizer que ‘that’ after the IQ verb ni- ‘say’, as illustrated below: (85) shinallatak ni-n que gallu-ta yali-shpa-ka osea yanapa-n nin. however say-3 that cock.acc pass-ger-top that.is help-3 evd ‘However they say that people help you organize the rooster festival.’ In (85) the Spanish relativizer que heads the complement clause of the finite verb form nin, not to be confused with the evidential form nin in sentencefinal position. The fact that the finite verb nin co-occurs with the Spanish relativizer suggests that it is not an evidential but a verbum dicendum whose main function is to reinforce the reportative meaning of the evidential.15
7.3. Question formation: dropped markers and Spanish intonation Yes-no questions in IQ are formed by the suffixation of interrogative -chu to the focalized constituent of the sentence, without any particular word-order or intonation contour marking the interrogation. This strategy is in the following examples: (86) a. kaya-ka pay shamu-nka-chu. tomorrow-top 3s come-3.fut-int ‘Will he/she come tomorrow?’ b. kaya-ka pay-chu shamu-nka. tomorrow-top 3s-int come-3.fut ‘Will he/she come tomorrow?’
Imbabura Quichua
515
c. kaya-chu pay-ka shamu-nka. tomorrow-int 3s-top come-3.fut ‘Will he/she come tomorrow?’ Yes-no questions in Spanish are formed by moving the main verb to sentenceinitial position and/or giving an interrogative intonation to the questioned element. These strategies have been adopted by IQ speakers. In more conservative sociolects Spanish intonation co-occurs with interrogative marker chu; in more Hispanicized ones, the interrogative marker is dropped, and declarative sentences are distinguished from their interrogative counterparts either by inverted verb–subject order with interrogative intonation (87a) or by intonation only (87b, c). (87) a. shaMU-nka pay kaya-ka. tomorrow-top 3s come-3.fut-int ‘Will he/she come tomorrow?’ b. kaya-ka PAY shamu-nka. tomorrow-top 3s/int come-3.fut ‘Will he/she come tomorrow?’ c. kaYA-ka pay shamu-nka. tomorrow-int 3s come-3.fut ‘Will he/she come tomorrow?’ Spanish also influences the formation of wh-questions in IQ. The typical order of wh-questions in IQ is WH + SOV, as shown in (88) below. This implies that interrogative and declarative sentences share the same word order: (88) ima-ta-tak paykuna-ka rura-n. what-acc-int 3pl-top make-3 ‘What do they make?’ This order is often inverted in wh-questions in contemporary IQ, with the main verb following the wh-word and followed in turn by the subject as in (89) and (90): (89) tandanakuy parti-manta rima-shpa, kikinkuna-ka imashina-ta meeting part-abl speak-ger 2.pl-top how-int winachi-shka-ngichi chay organisasion-ta-ka. create-prf-2pl that organization-acc-top ‘Concerning meetings, how did you create the organization?’
516
Jorge Gómez-Rendón
(90) kikinkuna yuyay-pi ima-shi ka-n shuk grupo ni-shka. 2.pl.poss thought-loc what-dub be-3 one group say-ptcp ‘In you opinion, what might this so-called group be?’ As shown in (90), the inverted word order occurs also in non-verbal predicative wh-questions involving a copula. The use of the copula itself is a calque from Spanish, because IQ requires no copulative verb in such cases. From the frequency of constructions like those illustrated above we conclude that IQ has calqued the syntactic pattern of Spanish, in which the verb follows the wh-word.
7.4. Adverbial clauses: Spanish subordinators and the loss of nominalization In Section 5.4 I showed that IQ nominalized constructions are being gradually replaced by subordinated clauses on the Spanish model. A related development is the use of Spanish subordinators including lo-que (cf. 5.4), relativizer que (cf. 7.2.) and conjunctions of causal relation (porque), condition (si) and concession (mas que). In this section I discuss the use of Spanish conjunctions in adverbial subordinated clauses. Consider the following example: (91) ñukanchik ishka-ndin yachachik-kuna-ka rimanchik-yarin, pero si 1pl two-com treacher-pl-top speak-1pl-aff but if tapu-nchik ñukanchik shuk-lla shimi-pi yachakuk-kuna-ta ask-1pl 1pl.poss one-lim language-loc student-pl-acc mana intindi-nga-chu porque paykuna-pa nima mana ka-n-chu. neg understand-3-neg because 3.pl-dat nothing neg be-3-neg ‘We as teachers speak [IQ] indeed, but if you ask students in our language, they do not understand, because it means nothing to them.’ In (91) the Spanish conditional si ‘if’ is used instead of the verbal suffix -kpi for non-coreferential subjects. Notice the adversative conjunction pero in the same example. The word order in the conditional sentence is SVO instead of SOV. The last clause indicates a causal relation. It is marked by Spanish porque ‘because’ and not by the IQ suffixes -manta or -rayku. The Spanish subordinator porque never co-occurs with its IQ counterpart (the suffix -manta). On the contrary, conditional como16 does co-occur with native suffixes -kpi or -shpa. Example (92) illustrates this case of doubling.
Imbabura Quichua
517
(92) chayka como yapa alpa-ta charishpaka, kay-kaman-mi then because too land-acc have-ger-top this-all-val ka-shka kan chay shuk hacienda. be-prf be-3 that one hacienda ‘As the hacienda had a lot of land, it reached up to this area.’ Another loan subordinator, for concessive adverbial clauses, is maske ‘although’, from Spanish mas que, as illustrated below: (93) maske ñuka ashta yapatalla wasi-pi rima-kpi-pash, though 1s too much-acc-lim house-loc speak-ger-adit ñuka mama wasi-pi solo kichwa rima-n 1s.poss mother house-loc only Quecuha speak-3 ‘Even though I speak too much [Spanish] at home, at my mother’s home they speak only Quechua.’ This compound conjunctive co-occurs with additive -pash on the main verb. In (93) the verb of the concessive clause carries the suffix -kpi, but this is rather uncommon. Spanish subordinating conjunctions are frequent in contemporary IQ and their use is widespread across generations and levels of bilingualism. However, their co-occurrence with native suffixes is more frequent in conservative dialects. In innovative varieties, finite verbs occur without native suffixes more often than not. The fact that subordinated conditional clauses without suffixes of coreferentiality makes innovative varieties closer to Spanish. In general, the loss of nominalization and other morphosyntactic changes associated with it is a gradual process, the stages of which can be found in different idiolects within the same speech community.
8. The lexicon The influence of Spanish on the lexicon of IQ involves all semantic fields, from kinship and household to education and administration. According to the results from a corpus of spontaneous speech collected in Imbabura, the presence of Spanish borrowings in IQ amounts to nearly one fifth of the total number of lexemes (21%). However, the contribution of Spanish borrowings to the native lexicon is not the same across idiolects, with those of older generations showing less influence than those of younger, more bilingual speak-
518
Jorge Gómez-Rendón
ers. As for the type of Spanish borrowings, all lexical classes except pronouns and adpositions are borrowed, though in different numbers. Nouns are by far the largest lexical class (55%), followed by verbs (16%), adjectives (8%) and adverbs (2%). The contribution of function words is not unimportant, with a total of 17 percent of tokens including mainly conjunctions, discourse markers, interjections, numerals and frozen borrowings. In general, frozen borrowings are distinguished from code switches on the basis of their phonological assimilation and their integration into IQ morphosyntax. A large number of these borrowings are idioms and situation-bound formulaic expressions for greeting, thanking, requesting and the like. A thorough analysis of Spanish lexical borrowing in IQ and the ways of integration of Spanish loanwords into the native system of parts of speech has been carried out elsewhere (GómezRendón 2006a).
9. Conclusion Quechua and Spanish have a history of four hundred years of contact in the Andes. The intensity of contact has substantially increased in the last century as a result of the expanding power of the nation-state and the diffusion of media in rural areas. The existence of higher levels of bilingualism in Imbabura has strengthened the influence of the dominant language on the lexicon and the grammar of IQ. The outcome is a strongly hispanicized variety of Quechua. Such variety appears to be very adaptive to the new communicative settings imposed by modern society. In fact, contemporary IQ is a living language after four centuries of contact because it succeeded in making a compromise between the communicative needs imposed by the official language and the speakers’ cultural need to preserve their linguistic identity.
Notes 1. With no linguistic census available, this is only a reasonable estimate. Ethnologue gives a number of 300,000 speakers in 1977, which is evidently an exaggeration considering that the whole population of Imbabura (i.e. Mestizos and Indians) hardly reached 250,000 people by 1982. (INEC 2001). 2. As a matter of fact, nonce borrowings are much less integrated to IQ phonology and may be considered cases of insertional code-switching (cf. Muysken 2000a: 32). Furthermore, it appears that phonetic assimilation into native patterns goes hand in hand with grammatical accommodation, as noted by Fauchois (1988: 92).
Imbabura Quichua
519
3. This is typical of Ecuadorian dialects. Peruvian and Bolivian dialects show divergent patterns. 4. Another possible candidate is the prefix la-. In Argentinean Quechua (Santiago del Estero) laya occurs before all kinds of nouns and has the meaning “type of ”. However, it is neither phonetically reduced nor cliticized. Likewise, IQ speakers use laya with all nouns except kinship terms, in which case the short form la- is used, “indicating a type of kinship following the original” (cf. CIEI 1983: LVI; my translation). Interestingly, the word laya is obsolete in Ecuadorian Spanish, except in some archaic varieties spoken in rural areas. The case of la- is all the more exceptional because no prefixes exist in IQ, nominal and verbal morphemes being all suffixes. An alternative analysis is that la- is a reduced (grammaticalized) form of the verb illa-c ‘be.missing-ag’. 5. An explanation in the frame of the model of “embedded language islands” (Myers-Scotton 2002: 139ff). 6. Notice that IQ has no articles to mark definiteness and uses the topicalizer ka for definite referents. 7. Interestingly, example (22) shows emphatic mi occupying the position typically assigned to topic marker ka. As shown in Section 7, the fact that emphatic and focus markers usually swap places in modern IQ is a result of the structural reorganization of the language under Spanish influence. 8. Cole gives the label “subjunctive” for ngapaj and chun alike. I prefer to call them ‘purposives’ because of their original meaning in IQ. 9. For a study of evidentiality in Ecuadorian Quechua in the frame of Functional Discourse Grammar, see Gómez-Rendón (2006b). 10. This morpheme has at least five different realizations in Ecuadorian dialects (CIEI 1983: XLI): [naku], [naju], [nau], [na], [nu]. The example given by Muysken comes from Lowland Ecuadorian Quechua, a group of dialects spoken in the provinces of Napo and Pastaza where the same behaviour of naku as verbal plural marker is observed. 11. My analysis here differs from the one provided by Cole (1982) who claims that “-naku does not express reciprocity but rather joint action of some kind [and] this action may be, but is not necessarily, reciprocal” (1982: 92f). Coles accepts, however, that this marker can be used as an emphatic verbal pluralizer. 12. http://usuarios.arnet.com.ar/yanasu/main.htm. Dated 21 June 2006. 13. According to CIEI (1983), “the reason for the use of this morpheme may be an assimilation to the verbal plural marker in a process of metathesis, .i.e. the inversion of syllables” (CIEI 1983: XL; my translation). 14. The IQ adverb ashtawan ‘more’ may occur in the embedded clause as in (74) or co-occur with Spanish más ‘more’ as in (75). 15. This is probably due to the semantic bleaching of the evidential in the context of a new information structure, where less emphasis is placed on evidential values, following the model of Spanish discourse. 16. orque and como have a causal meaning but only como clauses may be accompanied by IQ markers.
520
Jorge Gómez-Rendón
References Adelaar, Willem and Pieter Muysken 2004 Languages of the Andes. England: Cambridge University Press. Alderetes, Jorge R. 2001 El Quechua de Santiago del Estero. In: Jorge R. Alderetes (ed.), El Quechua en Argentina, URL: http://usuarios.arnet.com.ar/yanasu/ main.htm. Bakker, Dik, Jorge, Gómez-Rendón, and Ewald Hekking Forthc. Spanish meets Guaraní, Otomí and Quichua: A multilinguals confrontation. In: Thomas Stolz, Dik Bakker and Rosa Salas Palomo (eds.), Aspects of Language Contact. Berlin/New York: Mouton de Gruyter. Brody, Jill 1987 Particles Borrowed From Spanish as Discourse Markers in Mayan Languages. Anthropological Linguistics 29: 507521. Büttner, Thomas 1993 Uso del quichua y el castellano en la Sierra ecuatoriana. Quito: Ediciones Abya Yala. Caillavet, Chantal 2001 Etnias del Norte. Quito: Editorial Abya Yala 1987 Syntactic Change in Pipil. International Journal of American Linguistics 53 (3): 253280. 1993 On proposed universals of grammatical borrowing. In: Robert Jeffers (ed). Selected papers of the Ninth Conference on Historical Linguistics. Amsterdam: John Benjamins. Cerrón-Palomino, Rodolfo 1987 Lingüística Quechua. Cusco: Centro de Estudios rurales Andinos Bartolomé de Las Casas. CIEI 1982 Caimi Ñucanchic Shimiyuc-Panca. Quito: Ministerio de Educación y Cultura, PUCE. 1983 Ñucanchic llactapac shimi. Quito: Ministerio de Educación y Cultura, PUCE. Cole, Peter 1982 Imbabura Quechua. Amsterdam: North-Holland Publishing Company. Fauchois, Anne 1988 El quichua serrano frente a la comunicación moderna. Quito: EBIAbya Yala. Gómez-Rendón, Jorge 2001 La deixis pronominal en la media lengua de Imbabura: comunidades de Casco Valenzuela y El Topo. MA Thesis. Departament of Applied Linguistics, Pontificia Universidad Católica del Ecuador.
Imbabura Quichua 2005
521
La media lengua de Imbabura. In: Pieter Muysken and Hella Olbertz (eds.), Encuentros y conflictos. Bilingüismo y contacto de lenguas en el mundo andino, 3957. Madrid: Vervuert Iberoamericana. 2006a Condicionamientos tipológicos en los préstamos léxicos del castellano: el caso del quichua de Imbabura. Actas del XIV Congreso del ALFAL2005, Monterrey: ALFAL. 2006b Interpersonal Aspects of Evidentiality in Ecuadorian Quechua. In: Miriam van Staden and Umberto Ansaldo (eds.), ACLC Working Papers 1: 3750. Granda, German de 2001 Estudios de Lingüística Andina. Lima: Pontificia Universidad Católica del Perú. Haboud, Marleen 1998 Quichua y Castellano en los Andes Ecuatorianos: Los efectos de un contacto prolongado. Quito: Ediciones Abya Yala. Hengeveld, Kees, Jan Rijkhoff, and Anna Siewierska 2004 Parts-of-speech systems and word order. Journal of Linguistics 40 (3): 527570. Hill, Jane, and Kenneth Hill 1986 Speaking Mexicano: The Dynamics of Syncretic Language in Central Mexico. Tucson: University of Arizona Press. Instituto Nacional de Estadísticas y Censos 2003 Censo de población y vivienda 2001. Quito: INEC. Kaarhus, Randi 1989 Historias en el tiempo, historias en el espacio: Dualismo en la cultura y la lengua quichuas. Quito: Editorial Abya Yala. Matras, Yaron 1998 Utterance modifiers and universals of grammatical borrowing. Linguistics 36 (2): 281331. Muysken, Pieter 1997 Media Lengua. In: Sarah Thomason (ed.), Contact Languages: A Wider Perspective, 365426. Amsterdam: John Benjamins. 2000a Bilingual Encounters: A Typology of Code-mixing. Cambridge: Cambridge University Press. 2000b Semantic transparency in Lowland Ecuadorian Quechua morphosyntax. Linguistics 39 (5): 873988. Schachter, Paul 1985 Parts of speech systems. In: Timothy Shopen (ed.), Language Typology and Syntactic Description. Volume 1: 361. Cambridge: Cambridge University Press.
Grammatical borrowing in Paraguayan Guaraní Jorge Gómez-Rendón
1. Background Paraguayan Guaraní (henceforth PG) is a Tupi-Guaraní language spoken by five million people in Paraguay and the Argentinean Province of Corrientes. An overwhelming majority of speakers of PG are also speakers of Spanish with different degrees of bilingualism. The century-long contact between PG and Spanish has resulted in high levels of bilingualism and convergence. Contemporary Paraguayan Guaraní differs in several ways from pre-contact Guaraní1 as a result of lexical and grammatical borrowing from Spanish. Rather lately PG has been also in contact with other Indo-European languages such as German and Portuguese, but an evaluation thereof falls outside the scope of this chapter. The present contribution focuses on a one-to-one borrowing situation between Spanish and PG.2 Too often Paraguay is presented as a model bilingual society without an accurate assessment of facts. Numbers show a somewhat different scenario. According to the 2002 census,3 Guaraní monolinguals (27%) are significantly more numerous than Spanish monolinguals (6.56%), particularly in rural areas. Furthermore, the percentage of bilinguals is only 59, in other words, less than two thirds of the country’s population. Differences are also qualitative. Spanish and PG show complementary distribution across social spaces, a situation that may be qualified as diglossic, with Spanish as the dominant language (Meliá 1992). Paraguay is a unique case in Latin America, but its uniqueness is founded less on bilingualism than on the fact that PG is the only Indian language spoken by non-Indian citizens as their mother tongue. As a matter of fact, rural PG shows less Spanish influence than urban PG. However, Spanish is present in urban and rural varieties alike. Assuming a clear-cut distinction between them is therefore inaccurate. Speakers and local policy-makers often use the terms Guaraníete (‘true Guaraní’) and Jopara (‘mixed Guaraní’) to take a stand before language use and identity. Similarly, the rural-urban division represents two opposite linguistic ideologies (purism versus mixture) which are at the very heart of language policies and politics in Paraguay.4 In what follows I make reference to both
524
Jorge Gómez-Rendón
varieties grouped under the label of ‘Paraguayan Guaraní’ (PG). When contact-induced changes are attributed to only one group of the speech community, the bilingual status of speakers is mentioned accordingly.
2. Phonology Spanish loans have entered PG with or without assimilation, though unassimilated items are more numerous than assimilated ones. To the extent that core vocabulary is composed of an important number of unassimilated loans, the sounds introduced through them have become part and parcel of the phonological inventory of the language. If compared to pre-contact Guaraní, PG has six new segments, /Φ, č, ð, ř, l, λ/. With the exception of /λ/, present in Spanish but probably also introduced through an Indian language (Gregores and Suárez 1967: 89), the presence of the above-mentioned sounds in PG may be attributed to Spanish only. Apart from being present in Spanish-derived vocabulary, they occasionally occur in native items in the speech of younger bilinguals. Segments /Φ, č, ř/ show the same primary articulation as native phonemes /p, š, r/ but differ from them in their secondary articulation. Likewise, Spanish-derived /ð/ differs from its native counterpart in that the latter occurs always nasalized in the cluster /nd/. Because /l/ and /λ/ do not have native counterparts sharing place or manner of articulation, they can be considered, for all purposes, alien to the pre-contact inventory. As for consonant alternations, a significant degree of free variation is found across idiolects between sound pairs /č/-/š/ and /l/-/r/. The vowel inventory of PG has remained virtually untouched by Spanish, except for the tendency observed in some bilingual children and young adults to either relax the tensed high central vowel or pronounce it as the obstruent [γ]. This development is limited mostly to urban lects. In all, we can safely state that the six-vowel set from pre-contact Guaraní is preserved in the vast majority of speakers. Suprasegmental phonology also shows effects from Spanish contact. Some bilingual children and young adults do not (fully) nasalize affixes attached to nasal roots as required by nasal harmony rules: e.g. the non-nasal reciprocal jajo occurs on nasal roots instead of its nasal allomorph ñaño. Similarly, regressive or progressive nasalization does not always occur in the speech of some bilinguals: e.g. mitãnguera ‘children’ is realized as [mitanguera] instead of [mitãŋũẽrã]. Stress patterns also have changed as a result of contact: while assimilated loans follow the native pattern “with stress in the last syllable, no matter in
Paraguayan Guaraní
525
which syllable the stress originally fell” (Gregores and Suárez 1967: 91), the bulk of unassimilated loans preserve the primary stress in the same syllable as in Spanish: thus, for example, one may find assimilated loans such as [kesú] ‘cheese’ beside unassimilated items like [bentána] ‘window’. As regards syllable structure, unassimilated loans have introduced noncanonical onsets and codas: clusters formed by a plosive plus a flap (e.g. /tr/, /pr/); sibilant /s/ occupying coda positions in non-assimilated loans, particularly in frozen borrowings with the Spanish plural ending (e.g. kosa-s-kuéra ‘things’).
3. Typology From a typological perspective PG may be defined as polysynthetic. This statement must be taken however with caution in view of the following analysis. Consider the answer given in (1) to the question whether knowing a second language is good: (1)
a. Chéve guarã nda-i-perhudisial-ri 5 1.obj for neg-3S.prs-detrimental-neg b. re-mombarete-ve-hína pene arandu 2S-strengthen-more-prog 2poss knowledge c. a-medida-que la ñe’ẽ rei-kuaa. in.proportion.as dem speak 2S-know ‘For me it is not bad, [because] you strengthen your knowledge inasmuch as you know the language.’
The foregoing example contains three Spanish borrowings: the adjective perhudisial ‘detrimental’; the complex conjunctive a medida que ‘in proportion as’; and the singular feminine article la. Notice that Spanish adjectives and nouns are commonly used in PG in syntactic positions other than the prototypical ones. I will return to the use of lexical borrowings in Section 9. The use of the Spanish article is analyzed in detail in Section 4.3. For the moment let us focus on the linking strategies in (1). Although (1b) is semantically dependent on (1a), the causal relation made explicit in the English translation is only implicit in PG. In turn, (1c) is linked to (1b) by the Spanish conjunctive a medida que ‘in proportion as’ indicating ‘fulfilled condition’ and subordinating (1c) to (1b). Both linking strategies co-exist nowadays in PG. However, if compared to pre-contact Guaraní, which shows a strong prefer-
526
Jorge Gómez-Rendón
ence for parataxis and postpositions instead of connectives, contemporary PG has certainly lost some of its synthetic nature. In addition, if compared to an equivalent construction in pre-contact Guaraní, clause (1c) seems to be a syntactic calque of Spanish. In (1c) the predicate head kuaa ‘know’ has two arguments, the second-person subject expressed by the prefix rei- and the object la ñe’ẽ ‘the language’. The construction is fully grammatical in PG and yet syntactically different from its more synthetic equivalent reñe’ẽkuaa ‘you know how to speak it’, in which ñe’ẽ is incorporated to the verb stem while only one argument remains explicit. It is uncertain whether reduplicative structures found in PG are contactrelevant, since other Tupi-Guaraní languages use reduplication as well. Besides, reduplication is present also in Paraguayan Spanish – just like in other Spanish varieties spoken in Latin America. The most likely hypothesis here is mutual reinforcement of reduplication in a process of convergence.
4. Nominal structures Nominal structures are the category most influenced by Spanish. There are five structures that show contact-induced changes: possession, number, determiners, case marking, and local relations. I address each of them in this section.
4.1. Possession Paraguayan Guaraní has three ways to mark possession: possessor–possessed juxtaposition (2); suffixation of ablative -gui to possessor (3); the use of the Spanish preposition de between possessed and possessor (4). (2)
umi organización dirihente-kuéra ndive ro-ñe’ẽ some organization leader-pl with 1pl.excl-speak ‘We speak with some leaders of the organization.’
(3)
upépe o-ñe-monda-paite ore sy-kuéra-gui ryguasu then 3-pass-steal-all 1pl.excl mother-pl-abl hen ‘At that time our mothers’ hens were all stolen.’
Paraguayan Guaraní
(4)
527
Oĩ-há-pe guive o-je-gueraha preso padre de familia 3.be-rel-loc from 3-pass-take imprisoned parent of family ‘Since then parents of families were imprisoned.’
The construction in (3) is similar to possessed–possessor constructions followed by ablative -gua. These constructions are used as fixed expressions and show relatively low frequency of use (Guasch 1997: 62). Another less common way to mark possession in PG is through the Spanish preposition (4).6 Notice that the possessed and the possessor are Spanish loanwords. This restriction suggests that we are dealing here with phrasal units borrowed from Spanish. I return to the issue of phrasal borrowings in PG in the last section. Two other contact-induced changes concerning possession need explanation. One is the loss of consonant alternation on the possessed element in young bilingual speakers, particularly in urban areas. Accordingly, possessive constructions like (5a) occur instead of (5b): (5)
a. ko jagua hesa dem dog eye ‘The dog’s eyes.’ b. ko jagua resa dem dog eye ‘The dog’s eyes.’
The other change has to do with the distinction between alienable and inalienable possession. Although PG has no morphological marker to distinguish one kind of possession from the other, the distinction is expressed by syntactic and lexical means, for instance, through noun incorporation (6a) or the use of specific lexemes for body parts (7a). In contact varieties noun incorporation is being increasingly replaced by phrasal constructions (6b), and bodypart lexemes do not show consonant alternation, as shown in (7b). (6)
a. a-johei-ta che-juru. 1S-wash-fut my-mouth ‘I will wash my mouth.’ b. a-je-juru-hei-ta. 1S-refl-mouth-wash-fut ‘I will wash my mouth.’
528
Jorge Gómez-Rendón
(7)
a. che rague 1S.poss hair ‘my hair’ b. che tague 1S.poss hair ‘my hair’
While the loss of consonant alternation in possessive constructions like (7a) is a recent development in PG – it is ungrammatical in a number of lects – the phrasal alternative to noun incorporation in (6b) is a choice with specific connotations, as shown by some authors (Velázquez-Castillo 1995: 555). Therefore, Spanish contact has not produced the construction itself but restricted both alternatives to one.
4.2. Number Number is usually not marked in PG. The interlocutor depends mostly on contextual cues for number distinction. If necessary, the ending kuéra is added to the noun to indicate plurality (8a), except if a numeral or a quantitative adjective precedes the head noun. At the same time an incipient tendency is observed in PG to (double) plural marking on Spanish loanwords (8b), including heads modified by numerals or quantifiers (8c). (8)
a. professor-kuéra-ndi kolehio tiempo-pe. teacher-pl-com school time-loc ‘[I speak Spanish] with the teachers during the school.’ b. brasilero-s-kuéra a-ñemongeta hendi-kuéra heta vése. Brazilian-pl-pl 1S-talk 3.com-pl many time.pl ‘Brazilian people, I talk with them many times.’ c. heta-iterei kampesino-kuéra nd-o-kuaa-i la kastellano. many-very peasant-pl neg-3-know-neg dem Spanish ‘Many peasants don’t know Spanish.’
I have not found evidence of systematic use of the plural kuéra in the corpus. There is positive evidence, however, that Spanish-derived lexemes quantified by numerals or quantitative adjectives determine the obligatory use of the Guaraní plural marker.
Paraguayan Guaraní
529
Another contact-induced development in PG related to an incipient morphological distinction of number is the differential use of the Spanish article: la for plural and singular and lo for plural.7 There are few examples in which the presence of one or the other calls for a singular or plural reading. Compare the following sentences: (9) i-porã-iterei la o-ñe-mbo’e la-mitã-(kuéra)-me. 3.be-good-very pro 3-pass-teach art-child-(pl)-acc ‘It is good that he [the teacher] teaches the child/the children.’ (10) che a-segui va’ekue 1S 1S-follow pst ko edukasión rehegua lo-mitã(-kuéra*) apyté-pe dem education concerning art-people middle-loc ‘I continued to be with people on education issues.’ Example (9) is ambiguous if taken out of context: mitã may refer to a number of children or to one specific child. In (10) the homophonous lexeme mitã ‘people’ is grammatical if and only if Spanish lo is precliticized. Furthermore, the plural marker may occur in (9) without added meaning, but not in (10) where it is ungrammatical.
4.3. Determiners and pro-forms Guaraní boasts a complex system of deictics used to mark not only definiteness but also spatial relations and other referential functions. Spanish articles have been added to this system as determiners and pro-forms.
4.3.1. The Spanish article as determiner The Spanish article may be used in PG as a determiner with both native and non-native nouns. Not all Spanish articles are borrowed: only the feminine singular la and the plural masculine/neuter los. The latter has dropped the final /s/ to become simply lo. Of these two, lo is used quite rarely and with plural nouns only. On the contrary, la is used very frequently and without number distinction. Examples (9) and (10) above are illustrative of both uses. In the function of determiner, la and lo are pronounced unstressed and tend
530
Jorge Gómez-Rendón
to form one phonological word with their heads. A possessive may occur between the determiner and the head as shown in example (11): (11) ij-apyte-pe-kuéra o-u la che tio 3poss-middle-loc-pl 3S-come dem 1poss uncle ha o-henoi la iñ-ermano-kuéra. and 3-call dem 3poss-sibling-pl ‘My uncle came with them and then called his brothers and sisters.’ The order of constituents in the noun phrase in (11) cannot be explained by Spanish influence, because articles never occur before possessives in this language. The reason is to be found rather in PG grammar. The possessive in PG does not perform the same referential function as the possessive in Spanish and must be accompanied by native or Spanish-derived determiners. Consider the following example: (12) nd-ai-kuaá-i pe nde róga. neg-1S-know-neg dem 2poss house ‘I don’t know your house.’ [lit. I don’t know that your house] In a proximal-distal dimension, pe refers to an object far from the speaker but near to the addressee. Some authors classify pe as a deictic of visual reference (Trinidad Sanabria 2004: 696). An accurate identification of referents in the communicative setting defined by speaker and addressee seems to be relevant in PG. This task cannot be carried out by possessives, hence the widespread use of demonstratives, including the Spanish article in constructions like (11) above. Spanish-derived la also occurs with one of a set of tense-marked nominalizers (Gregores and Suárez 1967: 128) which include present va (13), past va’ekue (14/15) and future va’erã (16). These nominalizers make nouns out of phrases. Here are some examples: (13) la o-ñe-mbo’é-va nda-ha’e-i dem 3-pass-teach-nmlz.prs neg-3.be-neg [la misma cosa]8 la o-ñe-ñe’ẽ-va [the same thing] dem 3-refl-speak-nmlz.prs ‘What is taught is not the same as what is spoken.’
Paraguayan Guaraní
531
(14) la o-u-ypy-va’ekue peteĩ tio dem 3-come-first-nmlz.pst one uncle ‘One uncle who came first’ (15) che ru la o-mano-ma-va’ekue o-japo doce año. 1S father dem 3-die-already-nmlz.pst 3-do twelve year ‘My father, who died twelve years ago.’ (16) nd-ai-kuaa-i la ha’e-va’erã. neg-1S-know-neg dem 1S.say-nmlz.fut ‘I do not know what I would say.’ When used along with la, nominalizers make relative clauses, which may be either restrictive (14) or non-restrictive (15). These constructions are used when the nominalized clause is the subject (13) or the object (16) of the main clause. To nominalize a clause that stands in oblique relation to the main clause, the relativizer ha without la is used instead.
4.3.2. The Spanish article as a pro-form Other uses to which the Spanish article is put in PG include several types of pro-form. Pro-forms are expressed only by the Spanish article /la/, which is used either as a freestanding pronoun or as a relativizer. Freestanding forms may be used in two basic co-referential functions: cataphoric (17) and anaphoric (18)–(19). (17) nda-che-tiempo-i la a-japo hagua otra cosa. neg-1S-time-neg pro(x) 1S-do for (other thing)(x) ‘I don’t have time to do other things.’ (18) alguno-ko no-ñe’ẽ-i-ete la kastellano, some-dem neg-speak-neg-very dem Spanish(x) oi-ke-rõ eskuela-pe-nte la ña-aprende-pa. 3-come-when school-loc-only dem(x) pl-learn-compl ‘Some [of us] don’t speak Spanish, only when we go to school, we learn it.’
532
Jorge Gómez-Rendón
(19) arema rei-ko nde ko Hernandarias-pe? long.time 2S-live 2S (dem Hernandarias-loc)(x) arema ai-me-te voi la a-nace ko’ápe. long.time 1S-be-very thus dem(x) 1S-be.born here ‘Do you live here in Hernandarias for long time? – Long time I live here where I was born.’ In (17) la refers forward to the noun phrase otra cosa ‘other thing’ while in (18) the second instance of la refers back to the noun phrase la kastellano ‘the Spanish language’. As it becomes clear from both examples, la may stand not only for bare heads but also for whole phrases, regardless whether they contain native lexemes or Spanish insertions. An illustrative case of how flexible the use of Spanish-derived la may be in PG is found in (19), where la refers back to the locative phrase ko Hernandariaspe ‘here in Hernandarias’. If the second utterance stood by itself, la should be read as an adverbial equivalent of ‘where’ in the English translation. A similar reading is suitable for (20), where pro-form la refers back to the noun Brasil. Pro-nominal and adverbial readings are satisfactory in both examples. (20) che nda-se-guasu-i, Brasil-pe la a-ha. 1S neg-leave-much-neg Brasil-loc dem 1S-go ‘I don’t leave home too often, to Brazil (there) I have gone.’
4.4. Case marking Contact-induced change is also visible in case marking of objects. The patient and the recipient of a transitive verb are usually marked in PG by pe (or its allomorph me in nasal environments). Although the presence of this marker is obligatory in human objects, it is regularly dropped in direct objects with the result of seemingly ungrammatical constructions. Compare the following examples: (21) che memby-kuéra hasẽ-mba 1S child-pl cry-compl che ména-pe o-guerahá-rõ preso. 1poss husband-acc 3S-take-when imprisoned ‘My children cried to death when my husband was taken to prison.’
Paraguayan Guaraní
533
(22) pe alumno-pe o-ñe-mbo’e va’erã pe dem student-acc 3S-pass-teach oblg dem i-lengua-materna-pe. 3poss-tongue-mother-loc ‘It has to be taught to students in their own language.’ (23) ij-apyte-pe-kuéra o-u la che tio 3poss-middle-loc-pl 3S-come dem 1poss uncle ha o-henoi la iñ-ermano-kuéra. and 3-call dem 3poss-sibling-pl ‘My uncle was among them and later he called his brothers and sisters.’ The marker pe occurs in direct objects (21) and indirect objects (22). In the first case it may be dropped, as shown in (23) where the noun phrase iñermanokuéra ‘3poss.siblings’ is the patient of the verb ohenoi ‘3.call’. Spanish does not mark direct objects – except if they are proper nouns – but always marks indirect objects. Accordingly, the elision of pe in PG cannot be attributed to the contact language. A similar elision of patient markers has been observed in other languages in contact with Spanish such as Ecuadorian Quechua (Gómez-Rendón, in this vol.) There are cases in which the Spanish preposition a – marking human direct objects – co-occurs with the corresponding Guaraní marker, as illustrated in (24). This construction does not occur with native noun phrases, however. (24) nda-ha’e-i ko’ága sekundária-pe neg-3.be-neg today high.school-loc e-je-eksihi [a los alumnos]-pe. imp-pass-discipline [to art.pl student.pl]-acc ‘It is not that nowadays in high school students are disciplined.’ In my view these isolated cases are instances of code-switching rather than borrowing. This assumption is confirmed when looking at (25) where a whole Spanish noun phrase is not marked by pe: (25) o-ñe-me’ẽ-va’erã [a los padres]. 3S-refl-give-oblg [to art.pl parent.pl] ‘It will have to be given to the parents.’
534
Jorge Gómez-Rendón
4.5. Local relations Further evidence for contact-induced change in nominal structures comes from the expression of local relations. There are two relations where Spanish borrowings occur: those expressed by the Spanish equivalents of English ‘between’ and ‘side’. (26) entre mbovy pe-japó-raka’e la almacén. among some 2PL-make-dist.past art store ‘Some of you set up the store.’ (27) upéi katu a-maña óga lado. after then 1S-look house side ‘Afterwards I looked at the house’s side.’ Both loanwords are heads of their respective phrases. While the adpositional phrase entre mbovy in (26) is head-first, the noun phrase óga lado in (27) is head-last. On the other hand, all instances of entre show one argument as opposed to some uses of this preposition in Spanish (e.g. entre tu y yo ‘between you and me’). Besides, the noun lado stands in a pseudo-possessive relation to its modifier óga. The borrowing lado is grammaticalized in the expression gotyo-lado, as shown in the following examples: (28) mba’e la nde porte hína nde familia-kuéra gotyo-lado. what art 2poss live prog 2poss relative-pl to-side ‘How is your life from the side of your family?’ (29) ro-ho kuri Brasil gotyo-lado. 1PL.excl.go rem.pst Brazil to-side ‘We went to Brazil.’ Interestingly, gotyo itself is a grammaticalization of koty ‘side’ and vo ‘for’. It is not clear however whether gotyo-lado is fully equivalent to gotyo or implies a different spatial relation. It appears that the correspondence between the native structure and the mixed counterpart is not one-to-one since (28) requires an ablative reading whereas (29) is best interpreted as allative case. Arguably there exist grammatical borrowings from Spanish in locationstationary relations. Consider the following sentence:
Paraguayan Guaraní
535
(30) o-ñe-guahẽ chupe upépe [a las dos]. 3S-refl-arrive 3.acc there.loc [at the.pl two] ‘They arrived there at two.’ The fact that prepositions a ‘at’ and en ‘in’ appear only in time expressions as heads of Spanish noun phrases suggests that we are dealing here with code switches.
5. Verbal structures Verbal structures show fewer contact-induced changes. These include (1) the double marking of aspect; (2) the use of the verb ‘have’ instead of non-verbal constructions; (3) the modification of valency-changing devices; and (4) the integration of Spanish loan-verbs and the predicative use of non-verbal loans.
5.1. Double marking of aspect Influence from Spanish in the marking of aspect is clear from the co-occurrence of Spanish adverb ya ‘already’ alongside with the native perfective marker mã as in (31). (31) Kova ko masãna ya hi’ayu-pa-mã. pro dem apple already 3.ripe-compl-prf ‘This apple is already completely ripe.’ The occurrence of Spanish adverb ya in perfective constructions has resulted in double marked constructions. The loan adverb ya never occurs by itself (Gregores and Suárez 1967: 154) but always with aspect markers – including iterative jei ‘again’. (32) pero ya a-ha-jei-ta-mã-hina. but already 1S-go-again-fut-prf-pro ‘but I will be going already.’
536
Jorge Gómez-Rendón
5.2. The verb ‘have’ replacing non-verbal constructions Another contact-induced phenomenon in verbal structures is the use of the verb guereko ‘have’ instead of non-verbal constructions. Given the limited use of this verb in pre-contact Guaraní – where it expresses possession – the current use of guereko in contemporary PG may be attributed to Spanish, from which the former has calqued a number of syntactic constructions with the verb tener ‘have’. The result is twofold: the decreasing use of non-verbal alternatives and changes in the valency structure of certain verbs. Consider the following example: (33) o-porandu chéve mbovy mitã-pa a-guereko 3-ask 1S.acc how.many child-int 1S-have Ha a-guereko dies ha’e chupe. and 1S-have ten 1S.say 3S.acc ‘He asked me how many children I have, and I told him that I have ten.’ In pre-contact varieties of Guaraní sentence (33) is not expressed by a predication. The lexeme ra’y ‘son of a father’ indicates also ‘having a son’. To further specify the subject (father) and the object (son), a person marker is prefixed to the noun and followed by a modifier. The result (34) is radically different: (34) o-porandu chéve mbovy che-ra’y. 3-ask 1S.acc how.many 1S-son ha che-ra’y pa ha’e chupe and 1S-son ten 1S.say 1S.acc ‘He asked me how many children I have, and I told him that I have ten.’ In similar terms, the predicative construction in (35) differs from its nonpredicative counterpart in (36) because the latter requires only the noun h-oga ‘3.house’ and the noun inflected for third person. (35) Pa’i A. o-guereko-va’ekue mbohapy hóga Kapi’ipé-pe. priest A. 3S-have-pst three house Kapi’pé-loc ‘Father A. had three houses in Kapi’ipé.’ (36) Pa’i A. hóga-va’ekue mbohapy Kapi’ipé-pe. priest A. 3S-house-pst three Kapi’pé-loc ‘Father A. had three houses in Kapi’ipé.’
Paraguayan Guaraní
537
The verb ‘have’ also occurs in a number of Spanish expressions such as tener poder sobre ‘to have influence on’ or tener la culpa ‘to be guilty’, many of which include whole noun phrases from Spanish. Example (37) illustrates the second expression: (37) o-guereko tuicha poder ore apyté-pe. 3S-have big power 2pl middle-loc ‘He exerted big power upon us.’ A further construction involving the verb ‘have’ is a syntactic calque from the Spanish periphrastic construction of obligation, as shown in (38) below. In this case, the verb ‘have’ co-occurs in a serial verb construction even if the corresponding tense-mood marker va’erã is present also on the second verb. (38) che a-guereko a-kobra-ve-va’erã. 1S 1S-have 1S-collect-more-oblg ‘I have to collect more [money].’ The presence of guereko in serial verb constructions like (38) was uncommon in pre-contact Guaraní. The use and function of the verb ‘have’ in PG have been restructured under the influence of Spanish. 5.3. Contact-induced valency changes Contact-induced valency changes include (1) the use of the Spanish clitic se as a model for the use of the Guaraní reflexive ñe; and (2) the use of the causative marker mbo with Spanish-derived verbal lexemes. 5.3.1. Spanish clitic se modeling the use of ñe The proclitic se is used both for reflexive and impersonal passive constructions in modern Spanish. There are cases, however, where the use of se results in ambiguous constructions. On the basis of this model, the prefix ñe in PG may receive reflexive and passive interpretations. Consider the following examples: (39) Upérõ avei campesino o-ñe-mombarete. then also peasant 3S-refl-strengthen ‘Then the peasants got stronger.’/‘The peasants were strengthened.’*
538
Jorge Gómez-Rendón
(40) A-guereko gueteri upe kuatia-i 1S-have still dem book-dim upérõ o-ñe-me’ẽ-va’ekue chéve. then 3S-pass-give-pst 1S.dat ‘I still have the little book I was given.’ The Guaraní marker ñe admits passive and reflexive readings alike. Sometimes, however, only one reading is possible and contextual clues are required for disambiguation. Example (39) cannot be passive if interpreted in a broader context: peasants were not strengthened but became stronger by themselves despite the persecution they suffered. On the contrary, example (40) accepts a passive interpretation. The question is to what extent the use of ñe constructions is influenced by the inherent ambiguity of Spanish se. One way to explore the complex relations between contact-induced uses of reflexivity and passive voice is to analyze how ñe is used with Spanish loan-verbs. Consider this case of a potentially ambiguous use of ñe: (41) jamás na-ñe-komunika-mo’ãi never neg-recp-communicate-cond la ña-ñe-komunika háicha la Guaraní-me. pro 1PL-recp-communicate like pro Guaraní-loc ‘They would never communicate the way we communicate in Guaraní.’ For a correct interpretation of this example, it is necessary to analyze the meaning of the verb in question. The Spanish verb may be transitive with the meaning of ‘notify’ or intransitive with the meaning of ‘have an interchange of words’. Only the first meaning allows for reflexivity and passiveness. Both occurrences of ñekomunika in (41) mean ‘have an interchange of words’ and cannot be interpreted as passive or reflexive. As the verb in the second clause is inflected for first person plural inclusive, it becomes evident that the only possible reading of ñekomunika is reciprocal in both cases. In fact, example (41) refers to the communication between Spanish monolingual speakers and bilingual PG speakers in Paraguay. What is relevant here is that the speaker marks the reciprocal relation by means of ñe instead of the reciprocal ño. This shows that reflexivity, passiveness and reciprocity are conflated all in one verbal prefix, just like Spanish fuses the three categories in one clitic (i.e. se)
Paraguayan Guaraní
539
5.3.2. Causativization of loan-verbs A common valency-increasing strategy in PG is the causativization of transitive (42) and intransitive (43) loan-verbs by prefixing the causative marker mbo/mo to the root. (42) la autorida-kuéra o-gueraha la i-kuatia o-mo-firma. art authority-pl 3-take art 3poss-paper 3-caus-sign ‘The authorities took their paper to have them sign it.’ (43) o-ñe-ha’ã lomitã o-mo-nace peteĩ modelo pyahu. 3-refl-effort people 3-caus-be.born one model new ‘People try to create a new model.’ Person-reference markers including reflexives and reciprocals must precede the causative. In a few cases the causative occurs before personal inflection. The following sentence is an example: (44) Upe kuatia oha’ãva o-mbo-jo-topa dem paper 3-effort-nmlz 3-caus-recp-meet obispo ha dirigente-kuéra bishop and leader-pl ‘That document tried to reunite bishops and leaders.’ The order of morphemes in this case could be explained by the meaning of the loan-verb topar ‘to bump’, resemanticized in PG as ‘to meet’. This meaning is possible also in Spanish, provided the verb takes the reciprocal se. In (44) the native reciprocal jo is attached to the loanword to give the same meaning as in Spanish. PG has thus borrowed not only the phonetic form of the word but also its potential meanings, the latter realized through native morphology. A final issue related to the causativization of Spanish-derived lexemes has to do with their status in the source language. In principle only verbs are borrowed from Spanish as heads of predicate phrases. However, this is not always the case. The predicative use of non-verbal loans is discussed in the next section.
540
Jorge Gómez-Rendón
5.4. Integration of Spanish verbs and predicative uses of non-verbal loans Spanish verbs are borrowed in PG by deleting the infinitive ending /-r/, without any derivative marker (e.g. komunika ‘communicate’). The causative mbo/mo may occur on intransitive loan-verbs which are thus transitivized with a slightly different meaning from the original (e.g. mo-nace ‘create’). The causative can occur also on transitive verbs to give them a mediative meaning (e.g. mo-firma ‘to have someone sign’), in which case the prefix mbo/mo performs the same function of the suffix uka. Nouns, adjectives and whole phrases may be used also predicatively in PG through two different mechanisms: (1) the causative is prefixed to the nonverbal lexeme; (2) the non-verbal lexeme or the phrase is integrated without further marking. Whereas the first mechanism occurs more frequently with nouns (45), the second is common with adjectives and phrases (46)–(47). (45) o-moĩ i-pyti’á-re i-po, 3-put 3poss-chest-on 3poss-hand o-mbo-kurusu ha he’i chupe-kuéra. 3poss-caus-cross and 3-say 3.acc-pl ‘He put his hand on his chest, made the sign of the cross and said to them.’ (46) i-provechoso-va’erã pe i-vida-diaria-pe. 3.be-useful-oblg dem 3poss-life-daily-loc ‘It has to be useful in his daily life.’ (47) nda-i-deprovecho-mo’ãi chupe la Guaraní. neg-3.be-useful-cond 3.acc art Guaraní ‘Guaraní wouldn’t be useful for him/her.’ The loan noun kurusu in (45), from Spanish cruz ‘cross’, is transformed into a verb by the causative marker. In (46) the adjective provechoso ‘useful’ has the present third-person marker and serves as the head of the predicate phrase. The Spanish prepositional phrase de provecho ‘of benefit’ in (47) is assimilated as a single unit and receives verbal inflection just like the loan adjective in the previous example. This particular behavior of loanwords in PG is explained through the system of parts of speech of PG, in which lexemes from several word classes are used flexibly. The classification of parts of speech in PG is addressed in Section 9.
Paraguayan Guaraní
541
6. Other parts of speech Spanish grammatical borrowings include parts of speech other than nouns and verbs. This section addresses the influence of Spanish on numerals and quantifiers, pronouns, connectors, discourse markers and time adverbs. Several contact phenomena analyzed in detail in previous sections are mentioned here only in passing – e.g. the pronominal use of the Spanish article.
6.1. Numerals and quantifiers The presence of Spanish numerals in PG is massive. Pre-contact Guaraní had a five-value numerical system. Nowadays it co-exists with a Spanish-based 10-value system in PG. Recently efforts have been made to expand the system through neologisms, but their actual use by the speaking community is reduced to writing. While some Spanish numerals have been assimilated (e.g. by dropping the final /s/, dos ‘two’ > do), others are pronounced exactly as in Paraguayan Spanish. The limitations of the vernacular system resulted in the nearly total replacement thereof in Hispanicized urban sociolects or in the coexistence of both systems in rural sociolects less immersed in a market economy. These observations are valid both for cardinals and ordinals. Quantifiers include only the loanwords todo ‘all’ and poco ‘few’.
6.2. Indefinite pronouns and determiners A number of Spanish indefinite pronouns and determiners have expanded the original Guaraní inventory. Spanish pronouns and determiners are used for people and things in PG. They include known, negative, universal and other referents. Pronouns and determiners are presented in Table 1 alongside their semantic values.
Table 1. Spanish-derived pronouns and determiners in PG
Person Determiner Thing
Known
Unknown
Negative
Universal
Other
alguno alguno alguno
– – –
nipeteĩ nipeteĩ nipeteĩ
enteroveva entero enteroveva
otro otro otro
542
Jorge Gómez-Rendón
(48) alguno o-maneja-ve ko situasión. some 3-control-more dem situation ‘Some people controlled this situation better.’ (49) oi-ko alguno líder o-gusta-háicha. 3-be some leader 3-like-so ‘There were some leaders that liked it that way.’ (50) nipeteĩ na-i-ñapysẽ-i ore rendá-pe. nobody neg-3-appear-neg 2pl.poss house-loc ‘Nobody showed up by our house.’ (51) che nipeteĩ parte nd-a-juhú-i i-vai-ha. 1S no part neg-1S-find-neg 3.be-bad-rel ‘I found that no part was bad.’ (52) ro-japo reunión entero socio roi-mé-va. 2pl-do meeting all partner 2pl-be-nom ‘We had a meeting among all partners.’ (53) Piribebýi o-je-aparta-ité-voi ha o-moĩ otro téra. Piribebýi 3-ref-apart-very-AFF and 3-put other name ‘Piribebýi separated and adopted other name.’ Indefinite determiners occur often with Spanish-derived noun heads (49), although cases are attested which show such determiners alongside native PG lexemes (53). Spanish indefinite pronouns are always borrowed in their masculine form (48), (52), and (53). No indefinite pronouns are borrowed for unknown referents. The negative pronoun nipeteĩ ‘nobody/noting’ is a loan blend of Spanish ni ‘nor’ and peteĩ ‘one’ (50). The same form is used as a negative determiner in (51). The universal determiner entero ‘all’ is borrowed in masculine form but used without gender distinction. The same form used as a universal pronoun requires the nominalizer va.
6.3. Connectors I have shown in Section 3 that Spanish conjunctions are borrowed in PG as clause linkers. Spanish connectors occur both in rural and urban sociolects,
Paraguayan Guaraní
543
Table 2. Simple connectors from Spanish and their PG equivalents Coordination
Subordination
PG
Sp
PG
Sp
terã ha/katu ha
O
-rõ -re -ramo -ha
si porke aunke ke
pero y
in adult and children speech. An inventory of Spanish connectors with their PG equivalents is given in Table 2. The first criterion to classify loan connectors is whether they are compound or not. A further criterion is the type of relation they establish between clauses: connectors that link clauses at the same level (coordinators) and connectors that link subordinate clauses to main clauses (subordinators). Some examples of simple connectors follow. (54) ndo-ro-japó-i mba’eve i-cóntra-pe pero ro-torva neg-1pl.excl-do-neg nothing 3-against-loc but 1pl.excl-annoy ichupe porque ha’é-nte o-rrekohe-paité-va’erã mandyju 3.acc because 3S-only 3-collect-compl-oblg cotton ‘We did nothing to her but we annoyed her because she was the only one who had to collect cotton in the area.’ (55) ko’ãga peve livro o-ú-va castellano-pe-meme o inglés today until book 3-come-nmlz Spanish-loc-usually or English ‘Until today the books that come are usually in Spanish or English.’ (56) o-ñe’ẽ la guaraní-me 3-speak art Guaraní-loc si ha’e-kuéra oi-pota la kampesino vóto if 3-pl 3-want art peasant vote ‘They speak in Guaraní if they want to get peasant’s vote.’ (57) San-Ignacio-gua no-ĩ-ri ko tembiapo ndive, San-Ignacio-abl neg-3.be-neg dem work with aunke la juventú San-Ignacio-gua o-apoya avei although art youth San-Ignacio-abl 3-support also ‘People from s.I didn’t participate though s.I. youth did give their support.’
544
Jorge Gómez-Rendón
The most frequently used Spanish connectors are porke ‘because’ and pero ‘but’. Coordinator y is used mainly to link code switches. The difference between Spanish subordinators and Guaraní subordinators lies on the fact that the latter are postpositional. The only exception is subordinator ke, which occurs in different constructions analyzed in the next section. 6.3.1. Subordinator ke Beside the above-mentioned connectors, PG has incorporated the Spanish subordinator ke ‘that’ in four different constructions: after some Spanish prepositions used as conjunctions (58); in indirect quotations, with or without the verb ‘say’ (59); in some time expressions to link the adverbial to the clause (60); and in adjective/adverb comparison to link the compared terms (61). (58) Durante ke ha’e o-u, n-o-pená-i ore-rehé during that 3S 3-come neg-3-worry-neg 2pl-about ‘During the Father’s visit, nobody worry about us.’ (59) El dia ke pe jevy pende rape vaí-gui art day that dem again 2pl.poss way bad-abl ‘On the day [when] change your bad habit again.’ (60) Pe ka’aru katu ou jevý-ma citación, dem afternoon well 3-come again-prf notice ke karai Isaac t-o-hóje t-o-ñe-presenta-mi that mister Isaac, imp-3-go imp-3-rfl-report-mit ‘That afternoon a notice came saying Isaac has to report.’ (61) i-kuenta-vé-ta ña-ñe’ẽ inglés ke la Guaraní. 3.be-count-more-fut 1PL-speak English than art Guaraní ‘The fact that we speak English will count more than we speak Guaraní.’ Example (58) illustrates one of the few cases of Spanish prepositions in PG. Unlike the English translation – in which during is the head of the prepositional phrase – the compound connective durante qu e9 links two independent verb phrases. In (58) the speaker is using the preposition durante as equivalent to the conjunction mientras on the basis of their similar semantics. The
Paraguayan Guaraní
545
subordinator ke in (59) simply connects a relative phrase to a noun phrase. Example (60) shows the subordinator ke in a sort of quotative construction signaling indirect speech. In this case no reportative verb such as ha’e ‘to say’ or ndaje ‘it is said’ is present. Ha’e always precedes its complement in precontact PG while ndaje may occur either clause-initially or clause-finally. When the subordinator is used with the reportative, ndaje occurs only clauseinitially. Cases like (60) illustrate the ways in which the borrowing of Spanish connectors has influenced PG syntax. Another use of subordinator ke with far-reaching consequences for PG morphosyntax is the one illustrated by example (61). In this case, ke links two terms of a comparison. The use of the subordinator has induced two changes at the level of morphology and syntax: one is the drop of the ablative marker gui on the second term of the comparison (in this case, the noun phrase la Guaraní); the other is the obligatory position of the second term after the subordinator – where in pre-contact PG the second term may be located somewhere else in the sentence, provided it has the comparative marker. Similar contact-induced changes in comparative constructions have been observed in Ecuadorian Quichua (Fauchois 1988).
6.4. Discourse markers Discourse markers borrowed from Spanish are ubiquitous in PG. Some loan connectors and adverbs are used as discourse markers, for which reason it is difficult to distinguish one use from the other. The double function of some Spanish connectors and adverbs has facilitated their use as discourse markers in PG. Among the most frequent Spanish discourse markers used in PG are: the causal-consecutive entonces ‘then’; the appositives o sea ‘that means’ and por ejemplo ‘for example’; and the resumptive bueno ‘well’. The analysis of Spanish discourse markers in PG is complicated by the fact that they usually appear on code-switching boundaries.
6.5. Time deixis The main contact phenomena in time deixis in PG include the borrowing of the days of the week in unassimilated or assimilated form. Time adverbs from Spanish include ahora ‘now’, siempre ‘always’, tedia ‘today” (from este día, ‘this day”) and entonces ‘then’. As shown in the previous section, it is difficult to classify entonces and ahora as time adverbs since they are used also
546
Jorge Gómez-Rendón
as discourse markers. In such case the only way to tell one use from the other is by knowing whether they are pronounced within the intonation contour of the clause (adverbs) or independently (discourse markers).
7. Constituent order Constituent order in PG is relatively free, with SVO as the unmarked order in the clause (Gregores and Suárez 1967: 182). This feature makes it difficult to track any contact-induced change in word order and explain new syntactic developments in terms of Spanish influence. There are, however, certain structures clearly influenced by Spanish. These were addressed already in previous sections: the order of elements in possessive constructions (Section 4.1); the order of elements in the noun phrase with respect to articles and possessives (Section 4.3.1); and the use of the Spanish subordinator in comparative constructions (Section 6.3.1).
8. Syntax Apart from syntactic structures concerning coordination and subordination, both addressed in Section 6.3, there are two other contact phenomena in the field of syntax that might be attributed to Spanish influence. On the one hand, there is evidence that declarative sentences are transformed into questions by receiving Spanish interrogative intonation contours, without using the interrogative particles piko or pa. On the other hand, the first part of the negative circumfix nda-i, when attached to third-person finite verbs is often realized as no- instead of ndo-. This form resembles the Spanish negative adverb no. However, the nasal onset in both forms prevents us from being conclusive in this respect, since a replacement of the negative circumfix with the Spanish adverb would imply the elision of the second part of the circumfix, which is not the case. Spanish influence is thus limited to morpho-phonemics.
9. Lexicon Spanish lexical borrowing is pervasive in PG. All semantic fields in PG show traces from the contact language. From religion and kinship to idiomatic and context-bound expressions, Spanish words have entered PG in large num-
Paraguayan Guaraní
547
bers, with or without assimilation (cf. Section 2).10 However, there is a difference in the number and type of loanwords depending on the sociolects, with more Hispanicized lexicon in urban than rural varieties. Except for pronouns, all lexical classes contain loanwords from Spanish. Of course, the contribution of Spanish lexemes to each class is different. According to the results from a corpus of spontaneous speech collected in urban and rural areas, the presence of Spanish borrowings in PG amounts to one fifth of the total number of lexemes, excluding code switches. If these are included, the presence of Spanish in PG is even greater. An analysis of the same corpus in terms of lexical classes identified gave nouns (37.2%), verbs (18.3%), adjectives (7.4%) and manner adverbs (0.9%). The most frequent grammatical borrowings included articles (19.1%), conjunctions (7.5%), numerals (1.7%), discourse markers (0.8%), adpositions (0.5%) and non-personal pronouns (0.2%). Spanish grammatical borrowings either replace, or co-exist with, native morphosyntactic mechanisms. Spanish lexical borrowings are not used always in their prototypical functions in PG. The analysis of Spanish loanwords for syntactic position showed that even if prototypical functions are the most frequent for nouns (heads of referential phrases) and adjectives (modifier of referential phrases), these are often used in other syntactic positions. Thus, Spanish nouns are used also as verbs and adjectives whereas adjectives are used as verbs, nouns and adverbs. The following examples show the flexible use of Spanish loanwords: (62) 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.’ (63) la mbo’ehára Guaraní i-fanático. dem teacher Guaraní 3-fanatic ‘That Guaraní teacher is a fanatic.’ (64) o-ñe’ẽ atravesado la Guaraní. 3-speak crossed art Guaraní ‘They speak Guaraní confusedly.’ The Spanish noun lado in (62) is used as the head of a predicate phrase with the meaning of ‘align oneself’, without any derivation whatsoever. In similar terms, the adjective fanático in (63) is used as the head of a predicate phrase (‘to be fanatic’) without any copula or derivation. Finally, the Spanish
548
Jorge Gómez-Rendón
adjective in (64) is used as the modifier of the predicate oñe’e ‘they speak’, meaning ‘in a confused manner’. This flexible use of Spanish borrowings in PG mirrors the parts of speech system of the recipient language, where lexemes are grouped in two classes: verbs and non-verbs. The latter includes Spanish nouns, adjectives and adverbs.11
10. Conclusion Like many other Amerindian languages, Paraguayan Guaraní has been in contact with Spanish during the last 450 years. Unparalleled in other Latin American countries, the history of Spanish-Guaraní contact has produced a large bilingual population and made both languages converge to each other. Albeit not analyzed here, the influence of Guaraní on Paraguayan Spanish is not less important. In this chapter I have described the influence of Spanish on the grammar of Paraguayan Guaraní. Contact-induced changes in grammatical structures are especially noticeable in the noun phrase. All parts of speech show traces of matter or pattern borrowing. Syntactic structures have been not less affected, particularly in relation to the use of Spanish connectors in clause linking and the calque of Spanish constructions. The all-encompassing influence of Spanish has led some linguists (cf. Melià 1992) to claim that Paraguayan Guaraní is not Guaraní anymore but a third language resulting from the mixture of both languages in contact.
Notes 1. In what follows I use the term pre-contact Guaraní to refer to the Guaraní language as spoken before the Spanish conquest. For a thorough description of precontact Guaraní, see Ruiz de Montoya (1994). Arte y bocabulario de la lengua guaraní. 2. The study of the influence of Spanish on Guaraní has a seminal work in Morinigo’s Hispanismos en el Guaraní (1931). In the last two decades the study of the linguistic outcomes of Spanish-Guaraní contact has received increasing attention. Worthy of mention are a number of articles published in the two volumes of Sociedad y Lengua: Bilinguismo en el Paraguay (1982), which scrutinize the linguistic and sociolinguistic aspects of language contact in Paraguay from different perspectives. 3. For full data on the census, visit the website www.dgeec.gov.py.
Paraguayan Guaraní
549
4. While current bilingual programmes in Paraguay promote Jopara as the language of schooling, several efforts have been made in the last decade to ‘cleanse’ Guaraní by producing prescriptive grammars and dictionaries to fill lexical gaps by introducing neologisms. 5. The examples used in this chapter come from three sources: Atlas Lingüístico Guaraní-Románico, Sociología-Comentarios (2002); Kokuera Rembiasa: Experiencias campesinas 3 vols. (1992); and my own corpus, collected during a threemonth fieldwork visit to rural and urban Guaraní-speaking areas in Paraguay. 6. Other cases in which the Spanish preposition is used instead of simple juxtaposition include partitive constructions. 7. The original gender distinction in Spanish coded in feminine la and masculine/ neuter lo is absent in PG since la is used both for male and female while all cases of lo include only plural nouns regardless of gender. Both pre-contact Guaraní and modern PG lack grammatical gender. Spanish adjectives and nouns are borrowed along with their original gender endings, but this does not mean that gender is grammaticalized in PG. 8. In every example where a code switch occurs, this will be identified between square brackets. 9. This construction is ungrammatical in Spanish. To coordinate two simultaneous events the conjunction mientras is used, with or without que. 10. For a comprehensive study on Spanish loanwords in Paraguayan Guaraní, see Morínigo (1931). 11. In extensive cross-linguistic study of the use of Spanish loanwords in three typologically different Amerindian languages including PG is presented elsewhere (Bakker, Gómez-Rendón and Hekking, forthc).
References Bakker, Dik, Jorge, Gómez-Rendón, and Ewald Hekking Forthc. Spanish meets Guaraní, Otomí and Quichua: A multilinguals confrontation. In: Thomas Stolz, Dik Bakker and Rosa Salas Palomo (eds.), Aspects of Language Contact. Berlin/New York: Mouton de Gruyter. Comisión Nacional de Rescate y difusión de la Historia Campesina 1992 Kokuera Rembiasa: Experiencias Campesinas. Asunción: CEPAG Corvalan, Grazziella, and Germán de Granda (eds.) 1982 Sociedad y bilingüismo en el Paraguay. Asunción: Centro Paraguayo de Estudios Sociológicos. Fauchois, Anne 1988 El quichua serrano frente a la comunicación moderna. Quito: Editorial Abya Yala.
550
Jorge Gómez-Rendón
Gregores, E., and C. Suárez 1967 A Descriptive Grammar of Colloquial Guaraní. The Hague: Mouton. Guasch, Antonio 1997 Gramatica de la lengua guaraní y antología de prosa y verso. Asunción: CEPAG. Melià, Bartomeu 1992 La lengua guaraní del Paraguay: historia, sociedad y literatura. Madrid: MAPFRE. Morínigo, Marcos 1931 Hispanismos en el guaraní. Madrid: Editorial Espasa Calpe. Ruiz de Montoya, A. 1994 [1640] Arte y bocabulario de la lengua guaraní. Asunción: CEPAG. Thun, Harald (ed.) 2002 Atlas lingüístico guaraní-románico: Sociología. Kiel: Westensee Verlag. Trinidad Sanabria, Lino 2004 Diccionario Avañe’ẽ ilustrado. Buenos Aires: Editorial Océano. Velazquez-Castillo, Maura 1995 Noun incorporation and object placement in discourse: The case of Guaraní. In P. Downing and M. Noonan (eds). Word Order in Discourse, 555580. Amsterdam: John Benjamins. Velazquez-Castillo, Maura 2002 Grammatical relations in active systems: The case of Guaraní. Functions of Language 9 (2): 133167.
Grammatical borrowing in Hup Patience Epps
1. Background The Hup language1 (also known as Hupda or Jupda; see Epps 2005a) is spoken by approximately 1500 people in the Vaupés region, located on the border of Brazil and Colombia. Hup is a member of the small, under-described Nadahup or Makú family,2 whose four established members all live within the Upper Rio Negro region of Amazonia. These are Hup’s closest sister Yuhup (Ospina 2002), the more distant Dâw (S. Martins 2004), and the most distant relative Nadëb (Weir 1984), as measured in both genealogical (genetic) and geographic distance. The languages Kakua/Nukak and Puinavé have also frequently been classified together with the Nadahup languages (see, e.g., Loukotka 1968), but their relationship has yet to be conclusively demonstrated. Hup is currently fully viable, learned as a first language by all Hup children. However, by the time they are adults, virtually all Hupd’əh (‘personpl’, ‘Hup people’) are fluent in Tukano, the most widely spoken language of the Eastern branch of the Tukanoan family, of which some dozen or more members are spoken in the Vaupés region. This bilingualism on the part of the Hupd’əh is not unusual in the context of the Vaupés, which is well known for the striking multilingualism of its inhabitants, rooted in the local practice of linguistic exogamy. According to this practice, marriage obligatorily takes place across group or ethnic boundaries, which are defined primarily by the language its members speak (e.g. Sorensen 1963, Jackson 1983). The nearconstant contact among local languages that this practice fosters has led to the identification of the Vaupés as a linguistic area (which itself has numerous features in common with the larger Amazonian region). The effects of diffusion have been documented in detail for the Arawak language Tariana, which has undergone significant grammatical influence from Tukanoan (Aikhenvald 2002, etc.). Unlike the Tukanoan and Arawak peoples of the Vaupés, the Nadahup peoples do not practice linguistic exogamy, preferring to marry among clans within their own ethnic/linguistic groups. However, the Hupd’əh (and the Yuhup, who also live within the Vaupés) are in close socioeconomic contact
552
Patience Epps
with Tukanoan speakers. As forest-dwellers and masters of hunting and gathering, the Hupd’əh have traditionally supplied labor, meat, and other forest products to their Tukanoan neighbors, who live along the rivers and practice fishing and manioc farming for subsistence. In exchange, the Hupd’əh are given manioc products and other trade goods, but are considered socially inferior by the Tukanoans and the Tariana. Accordingly, the burden is on the Hupd’əh to learn Tukano to communicate with their neighbors, whereas the latter have in general no interest in learning Hup. This one-sided bilingualism appears to have been in place and stable for many generations. This sociolinguistic situation has led to a profound Tukanoan influence on Hup, such that Hup is clearly involved in the Vaupés linguistic area (see Epps 2007). As is the case with Tariana (Aikhenvald 2002, etc.), this influence primarily involves the borrowing of grammatical categories, rather than the actual forms of morphemes (i.e. ‘pattern’ rather than ‘matter’); this can be attributed to Hup speakers’ awareness of the local taboo against language mixing, despite the fact that they are not part of the linguistic exogamy system that fosters it. Accordingly, speakers resist the borrowing of actual forms, of which they are more aware, while less consciously rearranging the categories and organization of their grammar to conform to those of Tukano. This chapter focuses on the effects in Hup of unilateral diffusion from Tukano, with possible influence from other Tukanoan languages as well. Because this diffusion has involved mostly substance rather than form, it may be difficult to determine with complete certainty whether a given similarity is indeed due to borrowing, rather than to inheritance or independent innovation. In general, positive evidence for borrowing is taken to be the presence of a feature both in the Tukanoan languages and in Hup, and its absence in other Nadahup languages; however, features that are typologically very common are considered suspect as candidates for diffusion, since the chance that they were independently innovated in Hup and Tukano is relatively high. Furthermore, because Yuhup and to some extent Dâw have also undergone contact with Tukanoan languages, in some cases hypotheses regarding diffusion rest on the feature’s absence in Nadëb alone, and the possibility that the feature existed in the proto-language and was later lost from Nadëb may be difficult to rule out. The situation is also complicated by the fact that detailed descriptions of Nadahup languages exist only for Hup (Epps 2005a) and Dâw (Martins 2004).3 In general, Portuguese influence on Hup grammar is very limited; it is mentioned in the database where relevant, but comes up only rarely. Contact
Hup
553
between Hup speakers and non-Indians is recent (mostly limited to the past thirty years), and is still quite infrequent. Very few Hupd’əh speak more than a few words of Portuguese, and Portuguese borrowings into Hup are for the most part restricted to lexical items (especially names of culturally new objects); i.e. ‘matter’ instead of ‘pattern’. Many of these may in fact have entered the language via Tukano, which in most cases uses the same borrowed lexical forms.
2. Phonology Hup has a significantly larger inventory of vowels and consonants than do the Tukanoan languages (e.g. nine vowels in non-nasal contexts vs. Tukano’s six), but all of the segmental contrasts found in Tukano also exist in Hup. The pronunciation of word-initial /y/ as [dy] ([ñ] in nasal contexts) is a panVaupés feature and almost certainly due to contact, as is the pronunciation of intervocalic /d/ as a flap [ɾ] and the prenasalization of voiced stops. The lack of word-initial /g/ is also common to both Hup and Tukano (but not to Nadëb or Dâw). The most striking contact features in Hup’s phonology are prosodic. Nasalization is a morpheme-level prosody in both Hup and the Tukanoan languages – certainly a contact feature – and accordingly nasal consonants and voiced stops are allophones4 (whereas segments contrast for nasalization in Dâw and Nadëb; compare Hup mǔn [fully nasal] and Dâw múd [half nasal, half oral] ‘caatinga, forest with sandy soil’). Also, Hup has a system of word-accent, in which two tonal contrasts are realized on stressed syllables (example 1), which parallels the system of pitch-accent found in Tukanoan languages. (1)
Hup núh (high/falling tone) ‘head’ nǔh (rising tone) ‘tapioca’
3. Morphological typology and constituent order Like Tukanoan, Hup has relatively agglutinative morphology, with compounding of multiple verb roots and the piling up of affixes and clitics (examples 2 and 3).
554
Patience Epps
(2)
Hup d’uˇ ç höd tətəd-d’óʔ-óy=mah. timbó 3pl beat.timbó-take-dynm=rep ‘They beat timbó, it’s said.’
(3)
Tukano (Ramirez 1997a: 176) yesê-a-de pihî-dẽe-ya! pig-pl-obj call-gather.together-imp ‘Call the pigs together!’
Also like the Tukanoan languages, Hup is dependent-marking and predominantly suffixing. Nadëb, in contrast, is more isolating and is largely headmarking and prefixing. Furthermore, Hup (like its sisters Yuhup and Dâw) has nominative–accusative alignment, whereas Nadëb is ergative in its constituent order and cross-referencing pronominal forms (see Weir 1984). Finally, Hup and Tukanoan favor SOV constituent order (although both languages allow considerable flexibility governed by pragmatics), while Dâw and Nadëb are reported to favor SVO. There is no doubt that Hup fits the Vaupés typological profile, although it is not entirely clear whether some of these changes might actually have occurred in Nadëb rather than in its sisters. 4. Nominal structures One of the most striking contact effects on Hup nominal structures is in the system of case marking. Like Tukano, Hup uses a single object suffix to indicate direct objects, indirect objects, beneficiaries and recipients, and a single oblique marker for locative, instrumental, and comitative functions, as in (4). Moreover, object marking in Hup and in Tukanoan is sensitive to animacy and definiteness. The Hup case markers do not appear to be cognate with forms used for case marking functions in Hup’s sister languages, which suggests that they developed in Hup relatively recently under Tukanoan influence – although Hup apparently used its own resources for the forms of the markers, rather than borrow them directly from Tukano. (4)
Hup a. Instrumental m’aˇ c-át p´d höd bib’-ní-h, děh=teg-éh. mud-obl dist 3pl close-infr2-decl water=tree-decl ‘They would stop it up again with mud, the water tree.’
Hup
555
b. Location ʔãh yamhidʔ-h, cãwyucé-ét. 1sg sing-decl São.José-obl ‘I sang at São José Village (during a drinking party).’ c. Comitative ʔah=ʔíp-ít ʔãh ni-ʔeʔ-ní-h. 1sg=father-obl 1sg be-perf-infr2-decl ‘I lived with my father.’ Contact has also shaped the Hup system of nominal classification. As discussed in Epps (2006b), an incipient system of classifying nouns (using terms derived from plant parts, such as tat ‘fruit, round thing’) has formed in Hup under the influence of the noun classifier system in Tukanoan languages. As in Tukanoan, the Hup classifying terms organize inanimate referents on the basis of shape (round, flat and flexible, etc.), as illustrated in (5)–(6), and animates on the basis of gender (male/female). One of the most interesting features of the Hup system is that – as an incipient system currently affecting only a few nouns – it is most frequently used in creating neologisms to name new, culturally introduced items (such as soccer balls, batteries, etc.). This can be interpreted as a mechanism for resisting outright lexical borrowing (from Portuguese or Tukano), which is in keeping with the Vaupés avoidance of obvious language mixing. (5)
Hup kw=tat hot.pepper=round ‘hot pepper’ (fruit) tác=tat kick=round ‘soccer ball’
(6)
Tukano (Ramirez 1997a: 212) biâ-ga hot.pepper-round ‘hot pepper’ (fruit) kapê-ga eye-round ‘eye’
As in Tukano, possession in Hup may be either alienable or inalienable. The inalienable construction, formed via the juxtaposition of the possessor and possessed NPs, is a regionally widespread and cross-linguistically common structure, but the alienable construction in Hup likely owes its form to Tukanoan. Like Tukanoan, but unlike Nadëb (which uses special possessive classifiers for a few nouns and simple juxtaposition in all other cases), Hup indicates alienable possession by means of a possessive marker between possessor and possessed (example 7). Dâw employs a similar construction but
556
Patience Epps
an unrelated marker, suggesting that Hup’s strategy (and probably Dâw’s) was developed recently and followed the Tukanoan model (illustrated in example 8). (7) Hup pedú nˇh cug’æˇ t Pedro poss book ‘Pedro’s book’ (8) Tukano (Ramirez 1997a: 324) m’ yaá wi’i 2sg poss house ‘your house’ Number marking in Hup also resembles that found in Tukanoan languages. As with object case marking, Hup nouns normally receive an overt plural/ collective marker (which follows the noun in both Tukanoan and Hup) only when their referents are animate and definite, and obligatorily when these are human; e.g. Hup tuˇ g=d’əh (howler.monkey=pl) ‘howler monkeys’; compare Tukano emô-a (howler.monkey-pl) ‘howler monkeys’ (Ramirez 1997a: 205). Also like Tukano, Hup has a singulative marker that can occur on a small set of nouns – mostly certain kinds of insects that typically appear in groups: (9) Hup cw=d’əh (biting.ant.sp-pl) ‘biting ants (sp.)’ cw=ʔãw (biting.ant.sp-sing) ‘a (single) biting ant (sp.)’ (10) Tukano (Ramirez 1997a: 206) butu-a (termite-pl) butu-a-w˜ (termite-pl-sing)
‘termites’ ‘a (single) termite’
5. Verbal structures Several features of Hup’s tense, aspect, and mode categories probably owe their form to Tukanoan influence. While it is difficult to ascertain definitively whether contact played a role in their development, the categories of completive, inceptive, iterative, and habitual aspect are used in very similar ways in both Hup and Tukano, and are in general marked by post-stem morphology in
Hup
557
both languages. The same applies to the modal categories of counterfactuality (irrealis), optative, conditional, and epistemic modality; many of these also appear to have grammaticalized relatively recently in Hup – further evidence that they developed during the period of Tukanoan contact. Other aspectual and modal meanings are conveyed via verb compounding (see below), and compounded verb roots are a source of markers of aspect and modality through grammaticalization processes. In the case of tense, Hup appears to have developed an obligatory future specification under Tukanoan influence, as well as a distinction between recent and distant past tense, realized by means of optional contrast particles: (11) Hup a. wæd-y´ʔ-´y páh ʔa h-a h. eat-tel-dynm prx.cntr 1sg-decl ‘I ate (it) recently.’ b. wæd-y´ʔ-´y cám ʔa h-a h. eat-tel-dynm dst.cntr 1sg-decl ‘I ate (it) a while ago; before today.’ A particularly striking effect of contact between Hup and Tukanoan is found in the expression of evidentiality. While the reported evidential specification is found in Nadëb and reconstructs to Proto-Nadahup, Hup has since developed additional evidential categories of nonvisual (i.e. heard, smelled, tasted, or felt) and inferred information, using markers grammaticalized from compounded verb roots (see Epps 2005b), and almost exactly paralleling the evidential specifications found in Tukano. The fusion of evidentiality with tense (obligatory in Tukano) is also found in Hup, but is limited to the reported evidential morpheme and the optional distant past tense marker (a common combination in narrative), and then only in one dialect (examples 1213). (12) Hup j’uˇ g-út=maám töh wɔn-kot=máh-ah. forest-obl=rep.dst.cntr 3sg follow-go.in.circles=rep-dec ‘In the forest, long ago, they say, he wandered following (the tapir).’ (13) Tukano (Ramirez 1997a: 122) diay wa’î yahá-apɨ’. dog fish steal-rec.past.rep.3sg.nonfem ‘The dog stole the fish (in the recent past, it’s said).’
558
Patience Epps
Moreover, Hup has developed a second inferred evidential construction based on the borrowed verb ni- ‘be’ (one of the few examples of borrowed ‘matter’; example 14); this appears to be a calque of the identical Tukano construction, likewise built upon the verb diî [nii] ‘be’ (example 15). (14) Hup póh, děh=teg g’et-ʔeʔ-ní-h. high water=tree stand-perf-infr2-decl ‘Really high, the water-tree stood.’ (15) Tukano (Ramirez 1997: 140) yaa wecé ma’a wi’ô-’kadã dĩî-áma. poss field path obstruct-nmlz.pl.perf be-rec.past.vis.3pl ‘They’ve blocked the path to my manioc field.’ (proof: logs across the path) Hup’s strategy for verbalization almost certainly owes its form to contact with Tukano. While Tukano has a verbalizing suffix -ti, Hup employs the verb -ni- ‘be’ (discussed above) for the same purpose, usually creating ‘have N’ expressions (e.g. Hup h m-ni- ‘have a sore’, Tukano kamî-ti ‘have a sore’ [Ramirez 1997a: 353]). Several of Hup’s valency-changing structures probably also arose under Tukanoan influence. Its reflexive, reciprocal, and applicative markers are unlike those in other Nadahup languages and all appear to be recently grammaticalized; they are used in similar ways to those in Tukanoan. Like Tukano, Hup forms a passive by means of a verbal marker and an object case suffix on the semantic agent: (16) Hup yaʔám tiyıˇʔ-aˇn hup-mæ`h-æ ` y. jaguar man-obj pass-kill-dynm ‘The jaguar was killed by the man.’ (17) Tukano (Ramirez 1997b: 187) di’i-t´ wĩ’ba-g´-de bopê-dõ’o-’kado pot-cl child-masc.sg-obj break-pass-nmlz.place.perf dĩî-ap. be-rec.past.vis ‘The pot was broken by the child.’
Hup
559
Also, the reflexive marker (a verbal prefix in both Hup and Tukano) can appear as a nominal suffix/enclitic with an intensifying function in both languages (e.g. Hup ʔãh-hup, Tukano yö’ˇbasi [1sg-rflx] ‘I myself’). Finally, causative meanings are conveyed by means of verb compounding – generally a very productive process in both Hup and Tukanoan, although much less so in Hup’s sisters. In some cases the Hup compounds (causative and otherwise) appear to be calques of their Tukanoan counterparts (e.g. Hup d’oʔ-sud- and Tukano mii-sãa [take-be.inside] ‘put inside’).
6. Other parts of speech Hup’s numeral system has been profoundly influenced by contact with Tukano. While the numerals 1 to 3 are reconstructable for the Hup-Yuhup-Dâw branch of the Nadahup family, Hup 4 is clearly a calque of the Tukano form (literally ‘has a sibling/companion’), as is 5 (‘one hand’) and the numerals 6 to 20 (a base-five strategy involving the addition of fingers and toes); these forms (4 and above) are in fact found throughout the Vaupés and even beyond (Epps 2006a). In more recent years, Hup speakers have also borrowed Portuguese numerals (particularly for 6+) – both the actual forms as well as the base-ten strategy. These are currently used interchangeably with the native forms. A few other grammatical forms have been borrowed from Portuguese, probably via Tukano (which also uses these forms). These include the adverbial particle té ‘until, up to’ (from Portuguese até; example 18), and the negative emphasis particle næ` may also derive ultimately from Portuguese nem ‘neither/nor’. Interestingly, Hup has borrowed a disjunction ʔó ‘or’ (from Portuguese ou; example 19) without borrowing a conjunction (‘and’), a crosslinguistically somewhat unusual scenario (Matras 1998). (18) Hup té yawarete ʔãh ham-b´-h. until Yawarete 1sg go-hab-decl ‘I always go as far as Yawarete.’ (19) Hup ʔó cap g’`, ʔó mtaʔáp g’`, ʔãh bʔ-ni-té-h. or other year or three year 1sg work-be-fut-decl ‘Next year, or a third year, I’ll stay here to work.’
560
Patience Epps
Hup and Tukano share a few interjections, such as the expression ʔagö ‘ouch’. Hup speakers also typically use borrowed Portuguese expressions for most days of the week and times of day; these probably entered the language via Tukano.
7. Syntax Several features of Hup syntax probably owe their form to contact with Tukano. The verb ni- ‘be’ (Tukano dĩî [nii]) – one of the few examples of a shared form (‘matter’ rather than ‘pattern’) in these languages (as mentioned above) – is used in both languages as a copula with predicate nominals and adjectives (although in Hup it appears only when certain tense–aspect–mode specifications are present), as illustrated in examples (20)–(21). In addition to this copular function of ni-/niî, its simple verbal meaning ‘exist, be present’ is common to both languages, as is its use in an inferential evidential construction (see §5 above). (20) Hup náw töh ni-ʔě-h. good 3sg be-perf-decl ‘He used to be good.’ (21) Tukano (Ramirez 1997: 258) teé pũdikã ãyú-sehé dĩi-ap. those contrast good-nmzl.inan.pl be-rec.past.vis ‘Those really are good.’ Negation in Hup has several features in common with Tukanoan, also likely due to contact between the languages. As in Tukano, Hup clausal negation is expressed via a verbal suffix (non-cognate across the Nadahup family); in addition, both languages have distinct negative lexical forms meaning ‘does not exist’ and ‘I don’t know’. Another point of resemblance between these languages is the appearance of case suffixes directly on verb stems as markers of headless relative clauses in object position within the main clause, as well as on adverbial clauses relating to time, as illustrated in (22)–(23) (with Wanano, a close relative of Tukano). The presence of this feature in Hup is almost certainly due to contact.
Hup
561
(22) Hup ʔam wæd-túk-uw-aˇn d’oʔ-næ ` n-æ`h. 2sg eat-want-flr-obj bring/take-come-decl ‘(We) brought what you wanted to eat.’ (23) Wanano (Stenzel 2004: 287) ~bʉ’ʉ chʉ-dua-re ~da-ta-i 2sg eat-desid-obj bring/take-come-vis.perf.1 ‘(We) brought what you wanted to eat.’
8. Lexicon As discussed in Section 1, Hup speakers – like the other peoples of the Vaupés – tend to resist the outright borrowing of form, particularly lexical items. Nevertheless, some words in Hup clearly have a Tukanoan source – although at least as many of the identifiable lexical borrowings are calques, rather than actual borrowed forms (e.g. ‘deer’ for ‘manioc-processing tripod’; ‘medicine-house’ for ‘hospital’, etc.). An interesting characteristic of the directly borrowed Tukano words is the high proportion of verbs in relation to nouns (borrowed Tukano verbs outnumber nouns roughly two to one; see Epps forthc.), whereas nouns are cross-linguistically much more commonly borrowed. This can probably be explained by the fact that verb roots usually occur in multi-morphemic contexts, often involving several compounded stems; thus embedded in native forms within a given word, borrowed forms may be less easily noticed and therefore not censored by either speakers or listeners (as observed by Aikhenvald 2002: 224 regarding Tariana). Particularly for those nouns having a Tukanoan source, borrowings and calques seem to cluster in the semantic domains relating to agriculture and ritual/religion, suggesting that the words were borrowed together with the concepts to which they refer. Both verbs and nouns enter the language as roots, requiring no additional morphology to convert them to acceptable form. Mostly within the last generation, Hup has adopted quite a few lexical items (but almost no grammatical material) from Portuguese. These include both nouns and verbs, and for the most part refer to newly adopted cultural items (e.g. ‘cup’, ‘spoon’); nevertheless, Hup avoids many potential borrowings by creating its own neologisms from native material. Many of these Portuguese borrowings may have entered Hup via Tukano (which also uses a considerable number of them), since few Hupd’əh speak Portuguese or have
562
Patience Epps
much contact with non-Indians. Attitudes toward allowing Portuguese forms to enter one’s speech are not generally as negative in the Vaupés as they are toward other indigenous languages – possibly a reflection of the relative prestige of Portuguese in the region.
9. Conclusion Hup is an intriguing case of a language that has largely resisted the adoption of borrowed forms (MAT), while assimilating many aspects of the grammatical structures and categories found in the contact language (PAT). It is in some cases difficult to be certain whether similarities in particular structure are due to contact, rather than to inheritance or independent innovation, but there is no doubt that a considerable amount of convergence has occurred, bringing Hup firmly into line with the Vaupés regional profile. This profile has been largely determined by the Eastern Tukanoan languages – principally Tukano, which has exerted a more or less unilateral influence on the local Nadahup languages (Hup and Yuhup) and on Arawak Tariana (Aikhenvald 2002). Contact has affected Hup’s grammar at virtually all levels, including its nominal and verbal structures, syntax, and discourse. On the level of phonology, Tukanoan has had a particularly strong hand in shaping Hup’s prosodic features of nasalization and tone (word-accent) – likely due to the relatively high discourse salience of such features. Contact also appears to have restructured Hup’s typological profile in significant ways, affecting its grammatical alignment, constituent order, etc. Tukanoan has influenced the expression of number, case, noun class and gender in the Hup noun phrase; in the verb phrase, tense, aspect, mode, and evidential categories have been affected, and likewise the mechanisms for indicating changes in voice and valency. Hup’s numerals have also undergone profound contact influence, and some clausal connectors, subordinating mechanisms, and discourse markers have been borrowed as well; pronouns, on the other hand, appear to have escaped Tukanoan influence. Negation and the use of a copula may also have been shaped according to the Tukanoan model. Of all areas of Hup grammar, the lexicon appears to have been relatively resistant to change; nevertheless, considerable calquing (lexical PAT-borrowing) and some outright MAT-borrowing have taken place, including a high proportion of verbs. In conclusion, the pattern of contact between Hup and Tukanoan has favored the borrowing of PAT over MAT, a cross-linguistically unusual situ-
Hup
563
ation. This can be attributed to the sociolinguistic context in which Hup is spoken, in which the pan-Vaupés resistance to overt language mixing plays an important role – despite the fact that Hup speakers do not practice the linguistic exogamy that fuels this resistance among the other peoples of the Vaupés.
Abbreviations cl decl desid dist dst.cntr dynm flr fut hab imp inan infr2 masc nmlz
classifier declarative desiderative distal distant past contrast dynamic filler future habitual imperative inanimate inferential 2 masculine nominalizer
nonfem obj obl pass perf poss pl, pl prx.cntr rec.past rep sg, sg sing tel vis
nonfeminine object oblique passive perfective possessive plural proximate contrast recent past reported singular singulative telic visual
Notes 1. Information on Hup (aka Hupda, Jupde) was obtained via the author’s fieldwork on the Rio Tiquié, Amazonas, Brazil, conducted in 20002004. Support from Fulbright-Hays, the National Science Foundation (Grant no. 0111550), and the Max Planck Institute for Evolutionary Anthropology, Leipzig is gratefully acknowledged. Many thanks go to my Hupd’ǝh hosts and language teachers, as well as to the Museu Parense Emílio Goeldi and the Instituto Socioambiental in Brazil for practical assistance with fieldwork. I am also grateful to Alexandra Aikhenvald and to an anonymous reviewer for comments on this material. 2. The name “Nadahup” is here considered preferable to “Makú”. Not only is the latter name encountered in the literature in reference to several unrelated languages and families in Amazonia, but it is also in common use as a highly insulting ethnic slur in the Vaupés, directed toward the speakers of these
564
Patience Epps
languages. ‘Nadahup’ combines elements of the names of the four established members of this family, Nadëb, Dâw, Yuhup, and Hup. 3. Information on Tukano comes from Ramirez (1997a, b). Nadëb data are principally from Weir (1984), and Yuhup data from Ospina (2002). 4. Morpheme-level nasalization is nonetheless represented orthographically for Hup by means of nasal segments (at least one per word), for easier reading.
References Aikhenvald, Alexandra 2002 Language Contact in Amazonia. Oxford: Oxford University Press. Epps, Patience 2005a A grammar of Hup. Ph.D. diss., Deptartment of Anthropology, University of Virginia (forthc. 2008 Mouton de Gruyter). 2005b Areal diffusion and the development of evidentiality: Evidence from Hup. Studies in Language 29 (3): 617650. 2006a Growing a numeral system: The historical development of numerals in an Amazonian language family. Diachronica 23 (2): 259288. 2006b Birth of a noun classification system: The case of Hup. In: L. Wetzels, (ed.), Language Endangerment and Endangered Languages: Linguistic and Anthropological Studies with Special Emphasis on the Languages and Cultures of the Andean-Amazonian Border Area. (Indigenous Languages of Latin America series (ILLA); Publications of the Research School of Asian, African, and Amerindian Studies (CNWS)) Leiden University. 2007 The Vaupés melting pot: Tukanoan influence on Hup. In: Alexandra Aikhenvald and R. M. W. Dixon (eds.), Grammars in Contact: A Cross-Linguistic Typology, 267289. (Explorations in Linguistic Typology 4.) Oxford: Oxford University Press. Forthc. Hup. In: Martin Haspelmath and Uri Tadmor (eds.), Loanword Typology Project, Max Planck Institute for Evolutionary Anthropology, Leipzig. Jackson, Jean 1983 The Fish People: Linguistic Exogamy and Tukanoan Identity in Northwest Amazonia. Cambridge: Cambridge University Press. Loukotka, Cestmir 1968 Classification of South American Indian Languages. Los Angeles: University of California. Martins, Silvana A. 2004 Fonologia e Gramática Dâw [Dâw phonology and grammar]. Ph.D. diss., University of Amsterdam. Amsterdam: LOT.
Hup
565
Matras, Yaron 1998 Utterance modifiers and universals of grammatical borrowing. Linguistics 36: 281331. Ospina Bozzi, Ana Maria 2002 Les structures élémentaires du Yuhup Makú, langue de l’Amazonie Colombienne: Morphologie et syntaxe. Ph.D. diss., Université Paris 7 – Denis Diderot. Ramirez, Henri 1997a A Fala Tukano dos Ye’pa-Masa. Volume 1: Gramática [The Tukano Language of the Ye’pa-Masa: Grammar]. Inspetoria Salesiana Missionária da Amazônia, CEDEM: Manaus. Ramirez, Henri 1997b A Fala Tukano dos Ye’pa-Masa. Volume 2: Dicionário [The Tukano Language of the Ye’pa-Masa: Dictionary]. Inspetoria Salesiana Missionária da Amazônia, CEDEM: Manaus. Sorensen, Arthur 1967 Multilingualism in the Northwest Amazon. American Anthropologist 69: 670684. Stenzel, Kristine 2004 A reference grammar of Wanano. Ph.D. diss., Department of Linguistics, University of Colorado. Weir, E. M. Helen 1984 A negação e outros tópicos da gramática Nadëb [Negation and other topics of Nadëb grammar]. MA thesis, UNICAMP, Campinas.
Mosetén borrowing from Spanish Jeanette Sakel
1. Introduction, including genetics and sociolinguistics1 Mosetén is spoken by approximately 800 people in the foothill region of the Bolivian Andes. It belongs to the small, unaffiliated language family Mosetenan (Sakel 2004). Spanish was first introduced by missionaries in the seventeenth century and gained more in importance when permanent missions were established in the nineteenth century. It soon became a highly prestigious language that was used for communicating with strangers, and that was connected to welfare and money, while the function of Mosetén was restricted to communication among family members and friends. Still today, Mosetén is only used orally, and not all children in the ethnic group learn it. Most speakers of the language belong to the middle and old generations. Moreover, all speakers of Mosetén are bilingual in Mosetén and Spanish, and many feel more comfortable speaking Spanish. Spanish is also the language used for writing and other official purposes. Only very recently, Mosetén has been established as a written language (Sakel 1999, 2001, 2002b; cf. also Sakel 2004: 5052), but is not generally used, apart from the attempts of few individuals.2 Spanish is the only language that Mosetén is currently in contact with. Even though there are speakers of other languages in the area – mostly Aymara and Quechua settlers from the higher Andean regions, as well as some speakers of Trinitario and Yurakare – only some Mosetenes know occasional words in those languages. Certain categories in the Mosetén language, however, give clues to former contact situations. Mosetén occupies a unique position in the transition zone between the Amazon and the Andean highlands, and shows traits of an Amazonian linguistic area (such as proposed by Payne 1990), as well as loanwords – and possibly also grammatical influence – from languages of the Andean highlands. In the present chapter, I will concentrate on the one-to-one borrowing situation between Mosetén and Spanish, more specifically the influence of Spanish on the grammatical structure of Mosetén.3 The main contact phenomena in Mosetén from Spanish are found on the level of discourse organization and the integration of verbs. To my knowledge, there are no contact
568
Jeanette Sakel
phenomena from Spanish in the phonology of Mosetén, apart from sounds appearing in loanwords that have not been phonologically integrated into the system of Mosetén. The same holds for the overall typological profile of Mosetén, which does not seem to have changed through contact with Spanish. The contact phenomena are found in the areas of nominal structures, other parts of speech, constituent order, and syntax. Verbal structures are only be represented by the integration strategies of loan-verbs.4
2. Nominal structures Within nominal structures, there are only two phenomena that possibly have arisen through contact with Spanish. Both involve the remodelling of the structure (PAT), rather than direct MAT-loans. In the gender agreement system, the feminine was originally used as the unmarked gender in the language, e.g. referring to an unspecified group of people (1). (1)
Mö’-ïn yi-’-in ats-i-jo-i katyi’ äwä’-mï’. 3f-p say-f.s-p come.m.s-vi-ins-m.s eh child-3m.sg ‘They (father and mother) said that their son came.’
Most probably under the influence of Spanish, many young speaker prefer the use of the masculine gender in these instances (Sakel 2002a: 302303, Sakel 2004: 9091), i.e. mi’in ‘they masculine plural’ would be used. Another structural change in the nominal structures that could have arisen through the influence of Spanish is the tendency to use of plural marking with inanimate objects whose plurality is not in focus. Traditionally, marking of plurality by the plural marker in is only obligatory animates and in cases where the plurality is in focus. Still, many speakers use it with inanimate nouns even when their plurality is not in focus, which could possibly be due to the influence of Spanish.
3. Verbal structures (the integration of loan-verbs) In the same way as there are only very few contact phenomena in nominal structures, verbal structures are not usually remodelled or taken over from Spanish. The only phenomenon related to language contact in the verbal sys-
Mosetén
569
tem of Mosetén involves the integration of loan-verbs. Mosetén has a system of complex predicates (cf. Sakel 2007b), which means that there are only 1016 5 ‘real’ verbs – or verbness markers – in the language, while all other verbal elements are combinations of one of these and a non-finite element. Spanish verbs are treated in the same way as native non-finite elements, and accommodated into the verbal system of Mosetén by the addition of a verbness marker. The elements borrowed from Spanish are all non-finite in nature, including verb-roots (3), non-finite verbal forms (6, 7) or other parts of speech, such as nouns (2, 4, 5, 8). Only two of the native verbness markers are used to integrate loan-verbs: -yi- (examples 68) and -i- (examples 25). The latter is very frequent in established complex predicates, but is generally not productive in forming new verbs, whereas -yi- is productively used both with native non-finite forms and to integrate loan-verbs. The unproductive marker -i- only appears with 4 verbs that are borrowed from Spanish, all of which must have been borrowed at a stage when the marker could still be productively used to form verbs. This is supported by the fact that these are concepts that the Mosetenes presumably did not have prior to their contact with Spanish missionaries, but that became important in their new society and were borrowed very early on.6 (2) (3) (4) (5) (6) (7) (8)
viaje-idewe-ireso-ifieshta-ipasar-yisaludar-yisuerte-yi-
‘to travel’ ‘to owe’ ‘to pray’ ‘to party’ ‘to happen’ ‘to greet’ ‘to be lucky’
(from the Spanish noun viaje ‘journey’) (from the Spanish verb-root debe- ‘owe’) (from the Spanish noun rezo ‘prayer’) (from the Spanish noun fiesta ‘party’) (from the Spanish infinitive pasar ‘happen’) (from the Spanish infinitive saludar ‘greet’) (from the Spanish noun suerte ‘luck’)
4. Other parts of speech The bulk of Spanish loans is found in the category of ‘other parts of speech’, comprising mainly function words and discourse markers. Most loans in this category are of the type MAT, i.e. elements taken over directly from Spanish. The numeral system of Mosetén is decimal, but has probably arisen from a quintenary system (Schuller 1917; Sakel 2004: 168), though it is unclear if this remodelling has happened due to Spanish influence. Clear Spanish MAT-influence is found in the usage of numerals, since in everyday speech, Spanish numerals – especially those above 10 – are preferred. Quantifiers are
570
Jeanette Sakel
occasionally borrowed from Spanish, even though their usage is not regular and they would qualify as instances of code-switching. Turning to indefinite pronouns, the forms nunca ‘never’ and siempre ‘always’7 are frequently used in Mosetén. Other indefinite pronouns only very rarely used. There is a general tendency in Mosetén for expressions of time being taken over more frequently from Spanish than other expressions (i.e. person, thing, location or manner) (cf. also adverbializers). Particles and discourse markers make up the major part of elements borrowed from Spanish into Mosetén. All Spanish connectors are well-established loans in Mosetén, even though they differ in the environments in which they can be used. Thus, the three Spanish coordinating conjunctions y ‘addition’ (9), o ‘disjunction’ (10)8 and pero ‘contrast’ (12) can connect clauses, as well as functioning as connectors in discourse, while only o ‘disjunction’ also can be used to combine phrases (11). The connectors are borrowed together with their clause-combining construction, i.e. they appear between the two juxtaposed clauses, as in Spanish. This construction is different from native strategies in Mosetén, where addition is expressed by mere juxtaposition, while contrast is marked by a clitic, appearing on the first element of the second clause (13).9 In many instances, the borrowed Spanish markers and grammatical constructions appear together with the native strategies in double constructions (14). (9) Wën-jö-’ khö’ï mömö’ jish-yi-ti-’ y move-dj-f.s mo.f only.f comb-vy-re-f.s and.E me’-me’ shiph-ki-’ raej dyaba. so-rd leave-vk-f.s all peanut ‘She must have come, [and] she just combed herself and thus the peanuts came out [of her hair].’ (10) Me’-nä-ki jäe’mä tse’-mö’ me’ färä-yë-bän-’-yä’-wïn so-fo-co dm mother-3f.sg so fry.banana-vy-again-f.s-ad-c o jäe’mä wi-ki-’-ya’. or.E dm spin-vk-f.s-ad ‘And then [the child’s] mother was frying bananas or spinning.’ (11) Chhibin o tsiis ji’-jaem’-te penne three or.E four ca-good-vd.3m.o raft chapa-ti-k-dye-tyi’. big.raft-vt-an-b-l.m ‘Three or four rafts to make a big raft.’
Mosetén
571
(12) Me’ jïmë mö’ pero mö’ maj-jo-’ me’. so close 3f.sg but.E 3f.sg much-vj-f.s so ‘This (water-source) is closer, but that one has more (water).’ (13) Mi’-we ködy-a-j-ki-ki material, itsi-ki tata. 3m-dr beg-vi-dir-an-dk.m.s material.E nx.m.s-co father ‘They went there to ask for material, but the priest was not there.’ (14) Tyiñe-tyi’ pero-ki pen’-ki jai’ba-i.10 semi.red-l.m but.E-co side-co white-vi.m.s ‘It (the peanut) is semi-red, but one side is white.’ Pero does not only express contrast, but can also be used to mark a change in topic. This extension in function seems to have been motivated by analogy with the native marker -ki, which like pero can be used to express contrast (Sakel 2007a). In the same way as coordinating conjunctions, many subordinating conjunctions are borrowed from Spanish. These include complementizers and adverbializers. Object complements can be expressed by a marker resembling the Spanish complementizer que, often pronounced ki (cf. example 15), and most probably being borrowed directly from Spanish and then phonologically integrated.11 (15) Jam’-ki-ki rai’s-e-’ ki kasiki jam ji’-ka-te cemento. ng-co-rd want-vi-f.o that.E cacique ng ca-bring-3sg.o cement.E ‘But he did not want the cacique to make him bring the cement.’ I have only one example of the Spanish complementizer a being used in Mosetén – and the speaker later corrects it, identifying it as a “slip of the tongue”, i.e. it is not an integrated loan (example 16). (16) 1992 khan aj yäe yakchh-i-ti a karij-tya-ki 1992-in yet 1sg begin-vi-re.m.s to.E hard-vd-an.m.s jäe’mä doktor-tom hospital-khan. dm doctor.E-com hospital.E-in ‘In 1992 I began to work with the doctor at the hospital.’
572
Jeanette Sakel
Turning to adverbializers, the Spanish markers si ‘if’, pajki ‘so that’ and the temporal markers hasta, desde and cuando are borrowed into Mosetén. Again, markers of time are prominent among those borrowed. The Spanish marker si ‘if’ is used in Mosetén to introduce a conditional clause (17). This clause appears again in the same construction as in Spanish. Clauses with the borrowed marker si seem to fulfil a narrower function than in Spanish, in expressing if-clauses giving alternatives. In other cases, the native forms are used (cf. Sakel 2007a). Thus, the functions are divided between the borrowed and the native elements. There are several native ways of expressing conditional clauses in Mosetén, all of which involve clitics. Example (18) shows a hypothetical conditional clause marked by the clitic -ya’.12 (17) 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’ know-again-2sg dm cure-vi-no-p thing-rd 3F-ad 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.”’ (18) Mi’-ra’ wën-chhï-sh-än-yä’ tye-baj-te-ra’ yäe kerecha. 3m.sg-ir move-dc-ds-again.m.s-ad give-again-3m.o-ir 1sg money ‘If he comes back again, I’ll give him his money.’ Purpose and causation are expressed by the marker pajki, which seems to be a phonologically integrated marker of Spanish para que ‘so that’. The integration process has probably happened in the following way: para que is often pronounced paa-que in Bolivian Spanish, and a final aspiration was added to the first part, which is a typical phonological trait of Mosetén. The meaning of pajki in Mosetén is broader than para que in Spanish, denoting both purpose (cf. the first occurrence of pajki in example 19) and causation (cf. the second occurrence of pajki in example 19), while in Spanish it is only used as a purpose marker. The marker porque, which in Spanish expresses causation, in not borrowed into Mosetén (and pajki is used in all instances).13
Mosetén
573
(19) Khin’-ki-ra’ ti-ksi paj-ki-ra’ tsin dyaidye’-chhe now-co-ir bring-3pl.o.m.s for-co-ir 1p stranger-su bae’-ja’ paj-ki kï’-yä’-mïn na’-e-n’ live-vi.1pi.s for-co size.dim-ad-as born-vi-1.pi.s yi-nä khä jike-in. say.m.s-fo well ps-p ‘And now he will bring them [here], so that we will marry them, because we [i.e. our babies] are born [too] small, so he says.’ [situation: in old days, the Mosetenes married mainly among themselves, which lead to genetic defects and babies being born prematurely. One of the priests wanted to introduce ‘new blood’ to avoid these problems] Finally, temporal adverbial clauses introduced by ashta (a phonologically integrated form from Spanish hasta ‘until’) are very frequent in Mosetén. Again, the marker borrowed from Spanish can be used with an extended function compared to the Spanish source, denoting both an endpoint in time ‘until’ (cf. example 20) and a beginning in time ‘when’ (21).14 (20) Mi’-khan mi’ bae’-i me’-ki keo’-te-in m-in 3m.sg live-vi.m.s so-co search-vy.3m.o-p ashta tyaj-ke-te in. until.E find-dk-3m.o-p ‘They searched (for him) where he lived, until they found him.’ (21) Sokoj-ko-ye-’ sombrero-in raej-jin-in ashta wën-jö-i take.off-rd-vy-3F.o hat.E-p all-p-p until.E move-dj-m.s virjen-ya’-in. virgin.E-ad-p ‘They take off their hats when they come to the virgin (Mary).’ The marker ashta appears parallel to native ways of expressing temporal adverbial clauses, such as marking by the polyfunctional clitic -ya’ (which is, e.g., also used in hypothetical conditional clauses, cf. example 18 above). Other temporal adverbial clause markers borrowed form Spanish are cuando and desde. Cuando can, in the same way as ashta, be used to express ‘until’ and ‘when’, but is less frequent. Desde only appears in combination with ashta, expressing ‘from–until’. Apart from that, the Spanish expression cada vez ‘every time’ is occasionally used in Mosetén to express reversal and repetition, appearing together with, or replacing, native reduplication or affixal marking.
574
Jeanette Sakel
The Spanish particle pues ‘thus, well’ is sometimes used in functions similar to native cliticized focus markers. Many discourse markers are borrowed from Spanish into Mosetén. The most frequent ones are the tag question nowe, from Bolivian Spanish no ves ‘don’t you see’ (cf. example 22), and the filler awer, from Spanish aver ‘let’s see’, which can also appear at the beginning of a turn (23). All coordinating conjunctions can be used as sequential discourse markers, linking elements of discourse to the overall context (cf. examples 2425 for the introduction of turns by coordinating conjunctions and 914 above). Other discourse markers that appear – though less frequently – are osea ‘that means, so’, explaining or enquiring about what was said before, porlomenos ‘at least’, siquiera ‘at least’, pues ‘thus, then, well’, claro ‘sure’, claro pues ‘well, sure’, bueno ‘well, sure’ and eso es ‘that is it!’, used as a turn-taking device and for minimal response. (22) Mö’-nä khä Hernan tipi-ti-’ mäei’-ya’ jäe’mä 15 3F.sg-co well Hernan measure-vt-f.s first-ad dm Marcelina Duran-tom, nowe? Marcelina Duran-com right.E ‘And this Hernan, they measured the first time, together with Marcelina Duran, right?’ (23) 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. 3m.sg-fo well dm also play.game-no-vi.m.s ‘Alberto, well, let’s see, he came [hereto, performed the action, and went away again], and he was also hmmmm playing games.’ (24) a. Edy-win Edy-c ‘the dead Edy’ b. O-jam tata pariente-dyera’ khä mi’-tyi’-mi’. or.E-ng father relative.E-mo well 3m.sg-l.m-3m.sg ‘Or also the relatives of the father (i.e. from his land).’ (turn) (25) a. Aj camion-chhe’ jïj-ti tsin Rapash-khan. yet truck.E-su go-dt.m.s 1P La Paz-in ‘We went to La Paz by truck.’
Mosetén
575
b. Y ats-i-ban-ya’-nä mï’ïn chhï-tyäkä’? and.E come.m.s-vi-again.m.s-ad-fo 2P also-em ‘And when you came back you also [went by truck]?’ (turn) In some cases, the Spanish answer particle si ‘yes’ is used in Mosetén, usually in the function of a minimal response. Place deixis is often expressed by loans from Spanish, and some of the markers have both temporal and spatial meanings. Spatial prepositions include ashta (from Spanish hasta ‘until) (cf. example 26),16 desde ‘from’ (27) and rarely a ‘to’ (28). (26) Yij-ya’ ashta Karanawi jöf aj majmi jäe’mä. foot-ad until.E Caranavi yet yet road dm ‘By foot until Caranavi, there was the road already.’ (27) Jama, desde bae’edye’-khan? em from.E village-in ‘Well, from the village?’ (28) Julio-khan jiti-n-yäe a Riberalta. July.E-ad send-1sg.o-1sg to.E Riberalta ‘In July, I was sent to Riberalta.’ Temporal deixis includes the markers ashta ‘until’, desde ‘since’, nunca ‘never’ (29) and less frequently in my data en ‘in’ (30). Furthermore, ai weses (from Spanish a veces ‘sometimes’) is occasionally used in Mosetén. (29) Nunca katyi’ khä bailar-yi-’ mö’ achae Diana. never.E eh well dance.E-vy-f.s 3F.sg dog Diana ‘The dog Diana will never dance.’ (30) Me’-me’-ye-ki öi yomodye’ en mil novecientos noventa y seis … so-rd-vy-an.m.s de.f year in.E 1996.E ‘Well, in that year, in 1996, …’ Several Spanish markers expressing the opposite meaning or delimitation of some kind can be used in Mosetén, such as embesde, from Spanish en vez de ‘instead’ (31), ni ‘not even’ (32) and sin ‘without’ (33).
576
Jeanette Sakel
(31) Mi’-ya’-ki-ki mi’ kerecha-ki-ki embesde kerecha karto-ki 3m-ad-co-rd 3m.sg money-co-rd instead.E money carton.E-co mö-mö’. 3F-rd ‘And there, and instead of money, (he gave) only carton money.’ (32) Khäkï jäe’mä säem’ atsi-jo-i tsin, ni phir-ti-tsin because dm fast come-dj-vi.m.s 1p ng.E get.stuck-vt-1p jam anik me’ wën-chhïï-tsin. ng sure so come-dc-1p ‘Because we came here fast, we did not even get stuck [in the mud on the road], this way we came here.’ (33) Jike ats-i i-ya’-katyi’ tyoko’-ye-’: sin ps come-vi.m.s m-ad-eh press.stomach-vy-3f.o without.E saeks-e-dye’. eat-vi-no ‘And then he comes here and feels the stomach (of the dog): there is no food.’
5. Constituent order The constituent order of Mosetén has partially been influenced by Spanish. Most notably, while Mosetén does not traditionally make use of adpositions, it has introduced prepositions from Spanish – both as MAT loans in their constructions, as well as through remodelling of native material to fulfil the purpose (cf. e.g. the use of ashta in example 26). Furthermore, I have the impression that the constituent order of Mosetén often follows Spanish patterns, even though this is again a less clear case of contact influence.
6. Syntax There are several Spanish contact phenomena in the syntax of Mosetén. Primarily, these involve the organization of clauses, i.e. strategies for coordinating and subordinating clauses. Negation strategies are not influenced by Spanish, but a number of delimitation markers, expressing similar concepts, are borrowed from Spanish into Mosetén (cf. examples 3133).
Mosetén
577
Coordination strategies are borrowed from Spanish together with the MAT-loans of coordinating conjunctions y ‘and’, o ‘or’ and pero ‘but’. These markers appear in the same grammatical construction as in Spanish, i.e. between the clauses they coordinate (cf. examples 914). The same holds for embedding. Many subordinating conjunctions are borrowed from Spanish, together with the grammatical constructions they appear in. In this way, complementation follows Spanish patterns when the complementizer ke/ki, from Spanish que is used in Mosetén. Likewise, adverbial clauses follow Spanish patterns when Spanish adverbial clause markers are used.
7. Summary This is a typical one-to-one borrowing situation between languages of very different status and prestige. Most loans from Spanish are grammatical elements (MAT) used at the level of discourse organization, such as coordinating conjunctions, subordinating conjunctions, deictic markers (of time and space), delimitation markers, and discourse markers. Apart from that, there is a strategy for the integration of Spanish verbs, following the native patterns of forming complex predicates. Spanish also seems to have had influence on the grammatical pattern of Mosetén. Thus, MAT-loans usually appear within their Spanish grammatical construction, i.e. coordinating conjunctions appear between the clauses they combine. Other PAT-influence that is not connected to any direct MAT-borrowing, however, is difficult to establish. We do not have old data of Mosetén, which leaves it unclear whether something is the remodelling of a pattern due to Spanish influence, or whether it is a native tendency in the language (cf. word order).
Abbreviations ad an as b c ca co com dc
adessive relation antipassive associative benefactive nominal past causative contrast comitative associative motion ‘on the way there’
de dim dir dj dk dm dr
demonstrative pronoun diminutive directional marker associative motion ‘on the way here’ associative motion ‘after arrival there’ filler downriver
578 ds
Jeanette Sakel
associative motion ‘after arrival here’ dt associative motion allomorph E Spanish form eh hearsay evidential ele elicitation example em emphasis marker f feminine fo focus marker in inessive ins inceptive aspect ir irrealis l linker in noun phrase m masculine mo modal marker ng negation no nominalization
nx o p pi ps rd re rf s sg su vd vi vj vk vt vy
existential negation object plural (1st person) plural inclusive past tense reduplication reflexive and reciprocal reference marker subject singular superessive verbness marker -tyiverbness marker -iverbness marker -joverbness marker -kiverbness marker -tiverbness marker -yi-
Notes 1. This description is based on my own fieldwork on Mosetén in Bolivia, between 1999 and 2002. 2. My main consultant, Juan Huasna Bozo, is today teaching Mosetén at the newly established ‘Universidad de Monseñor Jorge Manrique en Palos Blancos’ in the Mosetén region. 3. There are a number of instances where Mosetén may have had influence on the structure of Spanish. However, there are only few data – and often the same categories appear in the Spanish spoken by other lowland peoples in Bolivia. It is thus unclear whether these are areal traits of the areas or due to substrate influence from Mosetén. I will concentrate on Spanish contact influence in Mosetén. 4. There are furthermore many lexical loans from Spanish. 5. This is based on an old system of event classification that is has become lexicalized in parts and it is difficult to established exactly how many markers were originally involved in the system (cf. Sakel 2007b). 6. Marking by -i- and -yi- does not are not applied due to phonological reasons. Rather, these are two different verbness markers with different functions in the language (cf. Sakel 2007b). 7. Siempre is often used to express hearsay-evidentiality – both in Spanish spoken by the Mosetenes, as well as when borrowed into Mosetén. The use of siempre as an evidential in Bolivian Spanish seems to have come about through PATinfluence from indigenous languages.
Mosetén
579
8. In the same way as in some varieties of Spanish, the construction o – o can be used to list alternatives, similar to ‘either or’, etc. (cf. Sakel 2007a). 9. There is no clear native way of marking disjunction in Mosetén. Most probably, the native marker was lost from Spanish marker o took over this function. In some cases, the native clitic -ki ‘discontinuous topic’ can be used in functions similar to disjunction. 10. 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. 11. There is a native discontinuous topic marker, a clitic with the form -ki, the use of which, however, is very different from a complementizer, making a native PATreanalysis less likely. 12. Apart from marking hypothetical conditional clauses, this clitic is used to mark adessive and temporal relations (cf. Sakel 2004). 13. I am not aware of this extended function being used in the Spanish spoken by the Mosetenes or among Bolivians; thus, this is probably an innovation that arose in Mosetén. 14. There is no native structure in Mosetén with such an extended function, nor – to my knowledge – in the Spanish spoken in Bolivia (cf. Sakel 2007a for further discussion). 15. A group of men and women, for which the feminine form is used in Mosetén (cf. Sakel 2002). 16. Cf. also ashta used as a subordination marker. Ashta can also express ‘even’, parallel to Spanish.
References Payne, Doris L. 1990 Morphological characteristics of Lowland South American languages. In: Doris L. Payne (ed.), Amazonian Linguistics: Studies in Lowland South American languages, 213242. Austin: University of Texas Press. Sakel, Jeanette 2002a Gender agreement in Mosetén. In: Mily Crevels, Simon van de Kerke, Sérgio Meira and Hein van der Voort (eds.), Current Studies on South American Languages, 287305. Leiden: ILLA 3. 2004 A Grammar of Mosetén. (Mouton Grammar Library 33), Berlin: Mouton de Gruyter. 2007a Language contact between Spanish and Mosetén: A study of grammatical integration. International Journal of Bilingualism 11 (1): 2553.
580
Jeanette Sakel
Sakel, Jeanette (ed.) 2007b The verbness markers of Mosetén from a typological perspective. In: Bernhard Wälchli and Matti Miestamo (eds.), New Challenges in Typology, 315338. Berlin: Mouton de Gruyter. Sakel, Jeanette (ed.) 1999 Poromasi’ Pheyak’dye’in [old stories]. Bolivia: Proyecto GRAMO (2nd rev. edn 2002). 2001 Ojtere’ [the rooster]. Bolivia: Proyecto GRAMO. 2002b Tsinsi’ kirjka [our book]. Bolivia: Proyecto GRAMO. 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.
Index of subjects ability, 69 n. 9, 33, 45 absolutive see alignment accent word accent, 553, 562 pitch accent, 553 accusative, 486, 487, 501, 502, 504, 512 nominative–accusative alignment, 388, 554 active see alignment, prefixes additive, 66, 218, 373, 374, 506507, 517 adjective, 192 active, 266, 277 adjectival predicate, 248, 547 adjective–noun collocation, 276 used as adverb, 498 attenuative, 266 adjective–noun order, 159, 253254, 528 adjective formation, 81, 87 adjective-deriving affixes, 266 noun–adjective agreement, 127, 139 used as noun, 498 used as pronoun, 504 and verb, 441, 443 xenoclitic vs. oikoclitic inflection, 265 see also affix adpositions, 42, 44, 76, 169, 170, 277, 279, 372, 518, 576 order, 86, 87, 272 tone pattern, 85 adverb, 158, 231, 270, 277, 441, 446, 454, 458 comparison, 509511 adverb-deriving affixes, 265, 266 adverbial particles, 476, 506, 559 adverbialization, 473, 490, 495, 571572
directional, 351 as discourse device, 409, 545 locative adverbial, 153, 270, 424 manner adverbs, 265, 266 modal–aspectual adverb, 250251, 333 negative, 330, 334, 339, 546 phasal, 5, 56, 57, 70, 72, 158, 270, 279 time adverbs, 170, 176, 179, 269– 270, 474, 507508, 535, 545, 573 reason/cause adverbs, 146, 293 relative clause, 412 spatial adverbs, 269270 utterance-level, 270 see also compounding affiliation, 192 adverbial clauses, 61, 66, 115, 179, 224, 408, 415, 506, 511, 514, 516517, 560, 573, 577 linkage, 292293 markers, 23, 115, 206, 273, 319, 408, 457 affix, 61, 125, 169, 203, 206, 249, 253, 308, 371, 416, 417 n. 7, 460, 491, 524, 573 adjectivizers, 253 aspect, 290 causative, 298 derivational, 249250, 251, 256, 265266, 268, 270, 315, 382 extraction, 265266, 277278 honorific, 504 inflectional, 315, 382 locative, 127 loss, 460 person-number, 82 preposition-like function, 448 pronominal, 372
582
Index of subjects
affix (cont.) and proclitics, 441 TMA 128129, 369 verb derivation, 83 verbalizer, 251 see also adverb affricates, 83, 9293, 97, 125, 152, 216, 229, 263, 417, 439, 460 agent, 44, 99, 204, 205, 369, 390, 422, 430, 467, 473, 485, 500 habitual, 311 impersonal, 203, 501 agent nominalization, 490, 492, 510 agglutination, 4041, 47, 123, 484, 553554 change to analytic structure, 132 agreement case, 271, 279 definiteness, 139, 141 dual, 445446 gender, 568 loss, 484 negation, 273, 279 noun–adjective, 127 noun classes, 126127 object, 95 person, 162163 177 person–number, 274 plural, 43, 281 n. 3, 445446 subject–object, 127 verb, 205 verb-object, 484, 506 aktionsart, 4445, 46, 66, 70, 250, 269, 281 alignment, 168, 288, 369, 484 active, 202, 204205 ergative–absolutive, 288289, 388 nominative–accusative, 554 past tense, 180 shift, 41 allomorphs, 278 animacy, 42, 404, 457, 469, 554556, 568
and gender, 79, 201 answer particles, 5758, 66, 116, 270, 575 antipassive, 67, 577 applicative 67, 430, 476, 479, 558 area see linguistic area article, 390, 519 n. 6, 529532 definite, 41, 139143, 155, 162, 201, 248, 348, 391, 448 definite–indefinite contrast, 491 indefinite, 157 proclitic, 264 articulation, 36, 37, 199, 200, 242, 286, 305, 388 secondary, 524 shift, 38 aspect, 41, 8283, 9495, 112, 128– 129, 132, 142, 155, 162, 288, 289, 290291, 293, 296, 315, 369, 455, 474, 556557, 560 double marking, 535 punctual vs. continuous, 367 see also adverb, imperfective, particle, perfect, perfective associative, 153 motion, 371, 577, 578 augmentation, 70 auxiliary-verb constructions, 370 aversive, 267 basic vocabulary, 70 n. 15, 267, 276, 286, 356 benefactive, 170, 310 bilingual(ism) 89, 19, 21, 2526, 27 n. 6, 3132, 3940, 53, 67, 68, 69 n. 2, 383 Arabic (Khuzistani)–Persian, 137 Chinese–Vietnamese, 357 Hup–Tukano, 551553 Kriol–English, 367 Kriol–Jaminjung, 383 Kurdish–Turkish, 175, 180 Manange–Nepali, 287, 297
Index of subjects Mosetén–Spanish, 567 Nahuatl–Spanish, 403 Otomi–Spanish, 435, 437, 441 Paraguayan Guaraní–Spanish, 523– 524, 527, 549 n. 4 Quichua–Spanish, 481483, 492, 494, 498, 505, 514, 517 Rapanui–Spanish, 387, 394 Romani–Hungarian, 276 Rumungro–Hungarian, 261 Rumungro–Slovak, 280 Yaqui–Spanish, 419420 body parts, 7576, 101, 256, 276, 527 calques, 16, 45, 53, 54, 55, 147, 171, 176, 192, 209, 233, 237, 278, 279, 293, 324, 355, 380, 406, 407, 414, 456, 457, 468, 469, 477, 478, 501, 516, 526, 536, 537, 548, 558, 559, 561 cardinal numerals, 51, 84, 269, 270, 271 case (marking) 4041, 42, 69, 93, 95, 153154, 162, 169170, 205, 267, 281 n. 6, 288289, 367, 374, 465, 469, 473, 486490, 514, 532535 doubling, 154 versus enclitics, 369 extension, 42, 232233 loss, 4243, 248, 271 and adposition, 44, 172, 370 causal relation, 20, 5556, 131, 219, 293, 471, 516, 519 n. 16, 525 causative, 47, 83, 98, 100101, 146, 173, 266, 268269, 296, 298 n. 7, 374, 537, 540, 559 double causative marking, 269 of transitive and intransitive loanverbs, 539, 540 see also prefix circumpositions, 169172 circumstantial, 193 classifiers, 42, 43, 44, 70, 292, 297,
583
347, 348, 357 n. 10, 358 n. 11, 360, 371, 469, 555 clitic, 70 n. 14, 110, 154, 168170, 173, 177, 202, 203, 205, 217, 220, 264, 238, 309, 325, 375, 382, 425, 441442, 445, 457, 537538, 553554, 570, 572, 573, 574, 579 n. 9, 10, 11, 12 deictic, 86 and free morpheme, 309310 intensifying, 252, 559 loss of clitic pronouns, 168 versus suffix, 369 versus verb, 178 cliticization, 178, 448, 501, 519 n. 4 clusivity, 53, 62, 206, 460, 495, 504 loss, 445446 coda, 235, 292, 307, 482, 525 code-switching, 16, 177, 180, 253, 303, 367, 396, 397, 459460, 504, 510511, 547 vs. borrowing 330340, 364, 372, 380, 381, 383, 391, 392, 393394, 397398, 399, 424, 430, 461 n. 3, 488489, 507, 518, 518 n. 2, 533, 535, 569570 and discourse markers 545 dropping case markers 487 and language loss 338 insertional 9, 331, 364, 380, 518 n. 1, 532 linking 544 comitative, 42, 119, 169, 310, 369, 430, 479, 486, 500, 554, 555, 577 comparative, 41, 59, 172, 239, 519 n. 14, 509511, 544 subject, 510 synthetic, 266 word order, 220, 429, 510 see also adjective, conjunction, negation, preposition, subordinator, suppletion complement, 95, 9798, 99, 111, 113,
584
Index of subjects
complement (cont.) 145, 168, 178179, 209, 350, 414, 457, 467, 501502 factive, 208 factual, 144, 223 factual vs. non-factual, 145, 223 of modal predicates, 274275 see also word order complementation, 71, 179, 183, 577 complementizer, 97, 98, 102, 110, 112, 114, 119, 145, 148, 162, 208, 223, 224, 275, 279, 319, 334, 337, 471, 571, 577, 579 factual, 5456, 144, 145, 158 non-factual, 274 completive, 4445, 474, 556 complex predicates, 182, 365, 370, 569, 577 compounding, 101, 174, 276, 290, 324, 354355, 440, 443, 492, 561 adverbial, 85 connective, 544 conjunction, 517 numeral, 292, 314, 424 past, 204205 perfect, 203 pronoun, 271, 279, 502 verb, 100, 142, 296, 356, 553, 557, 559 vowel, 202 concessive, 55, 56, 66, 144, 148, 208, 224, 252, 255, 517 concord, 127, 130 conditionals, 66, 9394, 142, 156, 208, 224, 330, 336, 374, 425, 516517, 573 particle (marker, subordinator) 3233, 45, 5556, 173, 219, 224, 516, 572, 579 n. 12 word order, 516 see also conditional conjunction(s) 5456, 62, 70 n. 14, 130131, 193, 176, 208, 218, 224,
295, 351352, 424, 428429, 430, 440, 502, 511, 514 adversative, 218, 273, 278279, 451, 516 causal, 146, 374, 516 comparative, 426 concessive, 144, 252 conditional, 330, 334337 contrastive, 339 coordinating, 33, 55, 179, 206, 224, 239, 240, 255, 273, 318, 559, 570– 571, 574, 577 grammaticalized, 41 phrasal, 20 subordinating, 5556, 158, 160, 177, 192, 207, 219, 222, 239, 252, 374, 397, 471472, 476, 501, 517, 577 temporal, 224, 428 see also compounding, coordinator connectors see conjunctions consonant, 37, 65, 76, 9293, 114, 152, 216, 230, 234, 236, 287, 304306, 345, 387, 420, 438, 482 alternation, 524, 527528 alveolar opposition, 286 Begadkephat, 186187 clusters, 18, 7677, 152, 167, 216, 232, 241, 247248, 307, 388, 438, 439, 482 coda, 292 deletion, 167 dental, 139 gemination, 9697, 139, 152, 188, 212, 216, 263264 gradation, 236 harmony, 439 interdental, 198 palatalization, 93, 230, 354 pharyngalized, 38, 152, 199 prenasalized, 125 retroflex, 346 constituent order see word order construct state, 41, 140141
Index of subjects converbs, 41, 61, 95, 96, 102, 173, 179, 217, 222, 293, 295 coordination, 55, 61, 85, 179, 183, 193, 254, 255, 294, 295, 409, 451, 457, 543, 546, 577, 579 see also conjunction(s) copula, 32, 46, 60, 94, 95, 102, 104, 119, 146, 156, 159, 163, 173, 177179, 202, 217, 226, 272, 273, 297, 321, 325, 393, 457, 460, 468, 511513, 516, 547, 560, 562 difference main and subordinate clause, 99, 364 word order, 220221 see also negation correlative particle, 144 counterfactual, 93, 557 see also irrealis coverbs, 48, 269, 371 cross-reference, 96, 119, 120, 505 current contact, 6, 39, 246, 247 dative 94, 95, 113114, 120, 153, 168, 170, 217, 232, 233, 248, 267, 297 days of the week, 59, 65, 157, 192, 334, 338, 348, 380, 507, 545, 560, decimal, 50, 291, 292, 338, 348, 393, 470, 569 definite article, 35, 41, 43, 139, 140– 141, 142, 148, 155, 157, 162, 201, 248, 264, 277, 390, 391, 448 definiteness, 42, 43, 44, 70, 115, 116, 138, 139142, 148, 278, 348, 391, 491, 519 n. 6, 529, 554 agreement definiteness relative clauses, 86 and specificity, 391 deictic element, 5354, 58, 66, 68, 86, 155, 158, 161, 163, 251, 252, 255, 256, 269, 271, 277, 278, 381, 473, 506507, 529, 530 development into relativizers, 130 dependent marking, 110
585
determiner, 154, 202, 490, 529532, 541542 definiteness as relative-clause marker, 155 negative, 542 diminutive, 4344, 188, 217, 233236, 249, 390391, 469, 484485, 490492 direct object, 221, 230, 442, 486, 487, 504, 512, 532533 enclitic pronouns as, 325 de-topicalization, 226 suffix, 554 topicalizing, 147 direction, 98, 127, 271, 349, 351, 356, 416 discourse procedural, 432 discourse markers/particles, 5, 20, 21, 25, 57, 58, 62, 66, 67, 70, 130133, 143, 158, 176, 238239, 241, 252253, 270, 377379, 382, 447, 451452, 455, 460, 470, 475, 506509, 541, 546, 547, 562, 569, 570, 574 emotive, 394 hesitation, 238, 410, 425, 428 resumptive, 545 sequential, 375 disjunctions, 20, 55, 114, 119, 408, 451, 457, 470, 506, 559, 570, 579 donor language, 16, 509 dual, 445446 embedding, 254, 577 ergative, 41, 288289, 290, 296297, 369, 388 past-tense inflection, 202, 204205 split ergative, 168, 288 loss, 152153 evidentials, 173, 174, 426, 496, 513, 514, 519 n. 15, 557558, 560, 578 n. 7
586
Index of subjects
exclusive see clusivity existentials, 156, 162, 171, 226, 389 verbs, 46, 142143 experiencer, 101 focus particles, 56, 57, 115, 143, 158, 270, 272, 319, 374376 frequentative, 269, 277 fricatives, 93, 124125, 186, 212, 232, 247, 263, 305306, 330, 368, 420, 438, 439 future tense/aspect, 44, 46, 66, 128, 155, 173, 189, 289, 315, 367, 391, 407, 414415, 456, 530, 557 analytic, 236237 gemination see consonants gender, 4244, 79, 99, 141, 157158, 169, 185, 189190, 200201, 239240, 249, 271, 279, 281 n. 6, 311312, 313, 390, 391, 444, 469, 542, 549 n. 7, 555, 568 genitive, 84, 138, 140, 202, 217, 233, 234, 309310 loss, 248 goal, 113, 114 habitual, 44, 46, 128, 129, 142, 155, 311, 498, 556 harmony, 3637 head marking, 110 hearsay, 578 n. 7 idioms, 100101, 147148, 218, 276, 320, 321, 469, 518, 546547 imperative, 100 imperfective, 95, 371 inceptive, 155, 556 inclusive see clusivity incorporation, 47, 527528 indicative, 155, 207, 223, 376 vs. subjunctive, 210211 indirect object, 170, 221, 226, 533, 554
inferred certainty, 128, 551, 557558 infinitive, 48, 127, 157, 161, 173, 205– 206, 217, 223, 268, 277, 406, 445 vs. nominals, 392 loss of modal ~ 217 subjective ~ 274275, 279 instrumental, 554 intention(al) 9495, 156 interrogative clause, 511 embedded, 273, 502 intonation, 458, 546 polar, 514515 see also particles interrogative pronoun, 54, 157, 252, 270, 278, 292, 318, 322, 396397, 414, 453, 454, 505 as conjunction, 224 comparison, 230 creating subordinators, 219 and indefinites, 114 and negative indefinites, 237238, 241 and relative marker, 223, 255, 264, 275, 413, 511, 513 intonation see interrogative irrealis, 128, 288289, 557 iterative, 45, 129, 155, 250, 535 kinship terms, 111, 119, 304305, 311312, 382, 415 lexicon, 24, 47, 7677, 161162, 175, 209212, 215, 225, 231, 245, 256, 275276, 295297, 344, 353, 387, 398, 415, 447, 458460, 485, 507, 517518, 562 across, 270, 351 addition, 408, 457, 470, 471, 570 around, 42 beside/next to, 153, 176 between, 176 have, 46, 95, 117, 118, 156, 171, 456, 535, 536537
Index of subjects linguistic area, 6, 16, 2224, 43, 69, 168, 175, 226, 272, 286, 288, 578 Amazonia, 567 Australia, 364, 366, 382 Balkans, 20, 226 Caucasus–Mideastern–South Asian, 47, 49 Caucasus–Anatolia–South Asia, 49 Ethiopian, 91102 Gurage, 20 Meso-America, 24, 403 Papua, 329 South Asia, 286, 288 Vaupés, 24, 551552 West Africa, 107122 Western Asia, 264 loanverbs, 250, 267268, 277, 296, 568569 locative predicate, 389 modality, 4546, 55, 56, 57, 66, 117, 156, 162, 217, 269, 315, 334, 340, 356, 367, 456 alethic, 494 complement of modal predicate, 274275 deontic, 175, 206, 288 epistemic, 46, 175, 223, 494, 557 modal complements, 161, 222223 modal verbs, 113, 296, 391, 493, 495496 non-factual complementizer, 56, 158, 274 through periphrasis, 391 see also adverb, counterfactual, irrealis, negation, particle morphological type agglutinitive, 33, 41, 110, 123, 308, 369, 467, 479, 484, 553 analytic, 479 isolating, 40, 47, 48, 49, 110, 308, 368, 554 polysynthetic, 40, 403, 414, 525
587
synthetic, 467, 479, 526 type change, 33, 40, 69 n. 7, 132, 308, 416, 444, 479 motion, 371372, 391 movement narrative, 115, 128, 293, 336, 428, 430, 557 nasal harmony, 524 necessity see modality negation, 19, 58, 117, 119120, 129, 152, 160, 206, 271, 291, 338, 341 n. 1, 350, 376377, 382, 397, 416, 560, 562, 576 agreement, 273, 279 circumfix, 546 comparison, 478 constituent negation, 376 contrastive, 339 converb, 96 double negation, 237, 254255, 339 emphasis, 20, 242 n. 4, 559 negative copula, 129 negative polarity item, 376 (inherently) negative verbs, 129, 291, 350 subordinate clause, 255 see also particles, prefix, pronouns nominalization, 314, 493 agent, 492, 510 causation, 298 n. 4 embedded propositions, 224 location loss, 516517 modality, 175 relative clause, 179, 513 vs. subordination, 501502, 510, 511 non-verbal predications, 159160, 163, 272273, 416, 511, 512513, 535, 536537 noun incorporation, 174, 527528 number, 86, 110, 111112, 143, 157– 158, 202, 312, 404, 528529
588
Index of subjects
number (cont.) absence of, 79, 312313, 499, 529 in superlative, 239 see also agreement number system quintenary, 50, 569 vigesimal, 292, 403, 408, 424, 470, 471 numerals, 20, 27 n. 4, 5053, 62, 6566, 84, 130, 157, 162, 176, 188, 191, 231, 269, 270, 271, 279, 291292, 312314, 338, 348, 381, 393, 408, 424, 487488, 502504, 528, 541, 559, 562, 569 object see direct object, indirect object obligation see modality onset, 167, 287, 307, 525 consonant clusters, 482 nasal, 546 optative, 189, 222, 226, 269, 557 palatalization, 38, 93, 231, 241, 346, 354 see also consonant participle, 8081, 142143 perfective, 174 passive, 203, 268, 277, 467 secondary, 173 source for tense, 189 particles, 35, 66, 84, 110, 206207, 218, 269, 387, 492493 agreement, 116 additive, 373 adverbial, 20, 506, 559 attributive, 139 collective, 22, 312 comparative, 172, 206, 253, 426, 440, 477 complement introducing, 414 connective, 212, 373374 contrast, 557 coordinating, 85
deictic, 251, 256, 269 discourse, 116, 175176, 225, 365, 377379, 425, 428, 430 distributive, 269 emphasis, 559 focus, 20, 5657, 115, 143, 144, 158, 252, 269270, 272, 319, 374375, 511, 512, 574 genitive, 202 imperative, 100 inclusive, 206 interrogative, 57, 176, 218, 458, 546 limitative, 426 modal, 20, 57, 175, 206, 250251, 269270, 315, 356, 494 negative, 61, 66, 160, 176, 206, 237, 376377 object, 201, 440, 448 optative, 269 phasal, 206 perfective, 315, 395, 535 possessive, 395 progressive, 45, 395 question–answer, 5758, 66 relative, 61, 86, 87, 192193, 206, 207, 239 repetition, 56 requestive, 219 sentence, 358 subordinating, 85, 207, 208, 382 superlative, 59, 172, 231, 239240, 241, 253 tense, 315 vocative, 116, 269, 311 see also topicalizer, utterance partitive, 172, 233, 450, 549 n. 6 passive, 47, 173, 191, 203, 315, 350, 354, 388389, 392, 436, 470, 479, 558 impersonal, 113, 537 and reflexive, 538 see also participle, prefix, subject patient, 203, 205, 532533
Index of subjects perfect, 142, 189, 442 compound, 203 experiential, 45, 95 past, 368 perfective, 81, 82, 9697, 128129, 174175, 281 n. 4, 288, 289, 290, 395, 535 see also particle person, 82, 241, 484, 499 double marking, 506 see also possessive pitch, 286288 pitch accent, 553 pluperfect, 142, 143, 146, 148 plural associative, 112, 266, 279, 281 n. 3 double marking, 155, 488489, 528 see also number place of articulation, 36, 38, 305, 388 possessive constructions, 41, 43, 44, 85, 100, 153, 172, 202, 217, 226, 248, 309, 310, 314315, 390, 404, 440, 450, 457, 526528, 536 alienable–inalienable, 389390, 486, 527, 555 loss of affixes, 404405, 484 possession verbs, 4747 possessive marker, 9495, 155, 348, 389, 417, 526, 555 predicative possession, 118, 171, 156 word order, 43, 60, 94, 117, 118, 154, 159, 171, 219, 253, 395, 404, 406, 455, 530 see also lexicon: have, particle, prefix, pronoun, subject potential, 371, 407 prefix, 41, 110, 112, 127, 168, 173, 264, 281 n. 5, 519 n. 4, 554 active, 308, 315 affiliation, 192 artificiality, 265 aspectual, 250 causative, 396, 539, 540
589
deictic, 53, 270271, 278 indefiniteness, 278 likeness, 319 literary, 310 locative, 127 loss, 405 mood, 369 negation, 129, 237238, 270, 291 noun class, 126127 object, 407 from particle, 45 passive, 308, 315, 537538 person, 82 possessor, 404405, 417 n. 5 potential, 371 progressive, 45, 189 pronominal, 264, 278, 536 quantifying, 271 reflexive, 537538, 559 subject, 331332, 333, 406407, 526 superlative, 59, 266, 278 verbalizer, 331, 332 preposition, 4243, 153154, 159, 160, 229230, 309311, 440441, 446, 447451, 458, 473, 476 used as conjunction, 544 comitative, 369 comparison, 172, 220 coordinative, 85 doubling, 370 emergence, 264, 267, 281 n. 8, 404406, 576 genitive, 8485 locative, 127, 176, 205, 267, 351, 369, 370, 405, 477478 oblique, 369, 533 possession, 248, 239, 315, 390, 526527 reinforcer, 224 and subordinators, 412413, 452 temporal, 426 privative, 53, 116, 449 proclitic see clitic
590
Index of subjects
pro drop, 81, 272 probability see modality progressive, 129, 189190, 205, 372, 395 emergence, 44, 45, 110, 112, 455 see also prefix pronouns, 35, 43, 5354, 62, 70 n, 14, 161, 176, 189, 277, 349, 372, 415, 547, 562 attributive demonstrative, 390 bound, 64, 349 compound, 502 composite, 113, 413, 502 demonstrative, 453 emphatic, 270, 454 genitive see possessive impersonal, 501 indefinite, 157, 251, 470, 541542, 570 interrogative, 252, 292, 413, 454 as relative, 255, 413, 453, 511, 513 negative indefinite, 237238, 270, 570 oblique case forms, 281 n. 6 optionality, 81 personal, 22, 53, 68, 111, 152, 168, 200, 205, 271, 310, 312, 389 honorific, 316317 possessive, 85, 253, 390, 440, 441, 487 quantifiers used as, 504505 reciprocal, 5354, 64, 157, 271, 279, 501, 524 reflexive, 53, 64, 84, 113, 191, 252, 270, 501 relative, 86, 130, 239, 255, 305, 398, 412413, 453, 454, 501, 502, 511, 513514 resumptive, 53, 157, 161, 331 subject, 8182, 425 see also adjective, compounding, interrogative prosody, 18, 26 n. 3, 3839, 152, 365,
368, 388, 553 purposive, 374, 493494, 519 n. 8 quantifiers, 5, 85, 157, 269, 270, 381, 404, 405, 408, 411, 470, 502504, 528, 541, 569 questions see interrogative quotative, 101, 493, 496, 514, 545 realis see modality reason, 98, 101, 119, 224, 273, 319 see also causal recipient 68, 96, 267, 354, 468, 473, 532 recipient language, 1617, 3738, 3940, 62, 67, 509 reciprocal, 53, 54, 64, 157, 191, 483, 499501, 519, 538, 539, 558 see also pronoun reduplication, 40, 110111, 312, 356, 389, 479, 489490, 526, 573 reflexive, 47, 5354, 113, 191192, 252, 277, 352, 499501, 539, 558559 intransitive interpretation, 496 and (impersonal) passive, 537538 see also pronoun relative clause, 60, 61, 81, 8687, 99, 115, 145, 161, 179, 207, 223, 230, 255, 275, 364, 398, 412414, 440441, 453, 544545 asyndetic–syndetic, 192193 causal, 98 headless, 414, 454, 560 locative, 321322 nominalizer, 531 relative concord, 130 subject, 8081, 86, 98 see also particles, pronouns, word order relativizer see pronouns repetition, 5657, 270, 456, 573
Index of subjects reportative, 493, 496, 514, 545 reversal, 573 figure–ground, 117 sampling, 14, 33, 68 semantic functions, 143, 448 serial verbs, 110, 117, 118, 537 situation-bound expressions, 320 source language see donor language spatial expressions, 85, 153, 256, 267, 269270, 279, 311, 370, 381, 403, 404, 440, 486, 473, 529, 534, 575 split ergative see ergative Sprachbund see linguistic area stative/active, 80, 82, 153, 271, 442, 443, 457 stem modification, 371 stops, 304 alveo-palatal, 368 dental, 263, 420 distinctions, 167 vs. fricative, 212 glottal, 187, 304305, 326 n. 6 420 preaspirated, 229 prenasalization, 553 velar, 420 voiced, 199, 403, 553 voiceless, 242 n. 2, 263 voicing, 403 weakening, 200 stress, 18, 8283, 187, 188, 200, 241, 247248, 264, 287288, 404, 438, 439, 482, 524, 525, 529530, 553 subject, 81, 82, 95, 96, 107, 117, 127, 202, 223, 275, 331333, 368369, 349, 377, 393, 406, 411, 493, 494, 512513, 514, 516, 536 agreement, 127, 274, 499 collective, 312 in comparative, 510 of complement, 158 impersonal, 341 n. 2 nominalized clause, 531
591
partitive, 233 in passive, 467468 in possessives, 117, 180 see also passive, prefix, pro-drop, relative clause, word order subjunctive, 17, 45, 145, 155157, 161, 173, 175, 189, 274275, 279, 472, 493494, 519 see also subordination subordination, 61, 66, 9495, 99, 144, 177, 224226, 254, 272275, 452453, 457, 493, 525526, 536, 546 conjunctive, 224 vs. nominalization, 501502 see also copula, negation, relative clause subordinator, 5556, 66, 223, 255, 269, 273274, 425, 427, 430, 543, 544– 545, 579 n. 16 in comparatives, 546 see also conjunctions, particles suffix, 23, 43, 44, 7883, 9396, 99, 102, 110, 111, 127, 130, 153, 155, 168174, 177, 181, 189192, 201, 204, 206, 233242, 248, 249, 265270, 277, 281, 290, 293, 310, 311, 355, 369, 370372, 404406, 421423, 426, 430, 440448, 454, 465, 467, 469, 473, 484, 493, 496, 505, 514519, 526, 540, 554, 558560 superlative, 232, 239240, 241 double marking, 240 see also particle, suppletion suppletion, 371 comparative, 158 gender, 79 imperative, 100 ordinals, 5253 superlative, 158 tense–aspect, 369 suprasegmental phonology, 482, 524
592
Index of subjects
syllable structure, 18, 78, 80, 82, 188, 213 n. 4, 241, 264, 292, 307308, 346, 354, 346, 387, 482, 525 tense see alignment, ergative, future, participle, particle, suppletion time expressions, 58, 535, 544, 545 see also days of the week, time of the day times of the day, 59, 65, 157, 192, 380, 507 tone, 18, 38, 77, 8283, 286288, 298 n. 6, 345 350, 354, 553 class, 83 lexical, 80 merger, 286 pretone position, 247 see also adposition topic change, 471, 571 development, 131, 221 de-topicalization, 221, 226 discontinuous, 579 n. 9, 11 fronting, 395 left dislocated, 115 marker, 238, 321, 485, 507 position, 331, 333 prominence, 485 shift, 131, 375, 378 switch, 378 topic–comment, 352, 358 n. 14 topicalization, 147, 309 transition see shift topicalizer, 490491, 508, 511, 512– 513, 519 n. 6 transitivity, 96, 202204, 218, 288289, 296297, 368, 372, 407, 500, 538 agent marking, 369 ditransitive, 96 marker, 156, 372 participant marking, 532 see also causative, ergative, reflexive, word order
utterance, 38, 48, 273 modifier, 494 planning, 20, 33, 35, 163, 221, 226 utterance-initial particle, 116 utterance-final particle, 116 see also adverb valency, 47, 48, 372 changes, 281 n. 4, 290, 499501, 535, 536, 537539, 558 differentiation, 218 marking, 49, 113, 562 see also transitivity verb compounding see compounding verb serialization see serial verbs verbalization, 48, 58, 181 n. 5, 268, 331333, 423 ideophones, 99, 100 infinitives, 422 interjections, 251 nouns, 142, 268, 558 see also prefix verbness, 49, 371, 392, 569, 578 n. 6 vocative, 200, 249, 305, 427428 see also particle voice, 47, 350, 423 marking, 113, 562 see also passive, valency voicing, 232, 247, 286287, 403, 482 vowel, 3638, 65, 78, 80, 82, 124, 152, 167, 187188, 247, 263, 306307, 330, 387, 388, 420, 436, 438, 466467 adaptation, 470 amplitude, 288 alternation, 168 compound vowel, 202 diphthongization, 445 epenthesis, 167 gliding, 305 harmony, 167, 169, 174, 263, 439 insertion, 388, 483 intensity, 288
Index of subjects length, 77, 80, 117, 124, 213 n. 4, 216, 241, 263, 403404, 420 lowering, 77 semivowels, 305 word classes, 4849, 61, 308 word order, 19, 35, 81, 8687, 99, 153, 158159, 207209, 219222, 226, 248, 254, 272, 294, 320321, 333, 335, 368, 377, 411, 454, 465, 487, 501, 504, 511517 non-predominant non-verbal predications, 512513 noun phrases, 84, 138139, 146, 253254, 321, 370, 528 change, 15, 19, 43, 60, 141, 147, 177– 179, 219, 230, 390, 429, 502, 546
593
conditionals, 516 direct and indirect object, 221 inversion, 516 pragmatic conditioning, 365, 379 in relative clause, 255 relative clause–head, 60, 193, 230, 397, 511, 513514 subordinate clauses, 177 subject–predicate, 159, 321 and transitivity, 86, 395 see also comparative, copula, possessive constructions written language, 19, 123, 161, 245, 246, 247, 302, 303, 322323, 325 n. 3, 344, 348, 354, 357 n. 2, 367, 465, 567
Index of authors Abondolo, Daniel, 280 Acharya, Jayaraj, 287, 292, 296 Adelaar, K. Alexander, 329 Adelaar, Willem, 329 Afanasjeva, Nina Jelisejevna, 229 Aikhenvald, Alexandra Y., 3, 26, 32, 36, 55, 551, 552, 561563 Albert, Ruth, 250 Alderetes, Jorge R., 499 Alidou, Ousseïna, 76, 88 Álvarez, Albert González, 430 Alves, Mark J., 345, 347, 350, 352, 358 Ameka, Felix K., 110114, 117 Andrés de Jesús, Severiano, 439442 Andrews, Henrietta, 442 Arnold, Werner, 186, 188 Aronson, Howard I., 250 Backus, Ad, 9 Bakker, Dik, 17, 437, 451, 549 Bakker, Peter, 10, 70 Barker, Milton E., 353 Bartholomew, Doris, 442 Bavin, Edith, 369 Baxter, Alan N., 315 Becker, Alton L., 309 Behnstedt, Peter, 188 Bender, M. Lionel, 75 Benedict, Paul K., 349 Bernus, Edmond, 75 Bernus, Suzanne, 75 Bickel, Balthasar, 288, 289, 297, 298 Bin-Nun, Jechiel, 256, 257 Bisang, Walter, 91, 92, 102 Blench, Roger M., 107 Blommaert, Jan, 123 Blust, Robert, 326 Bolonyai, Agnes, 16 Boretzky, Norbert, 261, 274
Bostoen, Koen, 125129 Bradley, David R., 283 Braukämper, Ulrich, 92 Brody, Jill, 489 Bryan, Margaret A., 107 Bubeník, Vít, 269 Buelna, Eustaquio, 423 Buitimea Valenzuela, Crescencio, 431 Bulut, Christiane, 166168, 172175, 179181 Burssens, Amaat, 125 Büttner, Thomas, 481 Caferoğlu, Ahmet, 215 Caillavet, Chantal, 481, 483 Campbell, Lyle, 24, 31, 169, 403, 489, 502 Canger, Una, 404 Capistran, Alejandra, 466 Carlos Silva Encinas, Manuel, 431 Carochi, Horacio, S. J. Casad, Eugene H., 422423 Cerrón-Palomino, Rodolfo, 484 Chamoreau, Claudine, 23, 466468, 473, 476 Charpentier, Jean-Michel, 366 Christiansen, Niels, 83 Christiansen, Regula, 83 Chyet, Michael, 171, 206 Cole, Peter, 482, 487490, 493493, 506, 509, 519 Comrie, Bernard, 2, 94, 431, 461 Correll, Christoph, 189, 193193 Crass, Joachim, 20, 9192, 99102 Dakubu, M. E. Kropp, 120 Đào, Duy Anh, 358 Darjowidjojo, Soenjono, 339 de Casparis, J. G., 326
Index of authors de Rhodes, Alexandre, 344, 346 De Rooij, Vincent A., 22, 131, 132 Dedrick, John M., 422423 DeFrancis, John, 344 DeLancey, Scott, 288 Diffloth, Gérard, 345 Dimmendaal, Gerrit J., 117 Dixon, R. M. W., 3, 55, 371 Dorleijn, Margreet, 166, 168, 172174 Dozier, Edward P., 431 Dryer, Matthew S., 2 Duval, R., 213 Ebert, Rolf, 254 Echegoyen, Artemisa, 442 Ecker, Lawrence, 442 Eggers, Eckhard, 249, 252254, 257 Elšík, Viktor, 3, 24, 3334, 42, 45, 5254, 58, 59, 67, 261, 262, 265, 268, 281 End′ukovskij, A. G., 235 Epps, Patience, 20, 25, 551552, 555561 Escalante, Fernando, 431 Estrada Fernández, Zarina, 420, 423, 430, 431 Everett, Daniel L., 317 Faber, Alice, 91 Fabian, Johannes, 123 Fauchois, Anne, 483, 487, 490, 491, 512514, 518, 545 Fenyvesi, Anna, 280, 281 Ferlus, Michel, 357 Fetter, Bruce, 124 Field, Fredric, 3, 33, 43, 431 Fischer, Steven Roger, 389, 399 Fleischman, Suzanne, 94 Ford, Kevin C., 120 Friedman, Victor A. Gage, William W., 345 Galand, Lionel, 86
595
Gardner-Chloros, Penelope, 330 Gemalmaz, Efrasiyap, 166, 179 Genetti, Carol, 290, 297 Gilberti, Maturino, 476 Givón, T., 55 Gołąb, Zbigniew, 16 Gómez-Rendón, Jorge, 484, 485, 498, 509, 511, 518, 519, 533, 549 Gonda, J., 311 Graber, Philip, 375 Granda, German de, 484 Grégoire, Claire, 127 Gregores, E., 524, 525, 530, 535, 546 Grijns, C. D., 326 Grotzfeld, Heinz, 193 Guasch, Antonio, 527 Guerrero, Alonso, 404, 412 Gysels, Marjolein, 133 Haboud, Marleen, 481 Haig, Geoffrey, 25, 166168, 174 Halász, Ignácz, 238 Hale, Kenneth L., 431 Harley, Matthew, 114, 116 Harris, Alice, 169 Harris, John W., 366, 367 Haspelmath, Martin, 2, 93, 298, 458, 467 Haudricourt, André G., 345 Haugen, Einar, 3, 16, 32 Hayward, Richard, 100101 Heath, Jeffrey, 15, 32, 8688, 365, 382 Heine, Berndt, 16, 17, 9293, 98, 107 Hekking, Ewald, 17, 436442, 451, 549 Hengeveld, Kees, 485 Hernández Cruz, Luis, 439 Hess, H. Harwood, 442 Hildebrandt, Kristine A., 26, 242, 285290, 293298 Hill, Jane, 502 Hill, Kenneth, 502 Hobermann, R. D.
596
Index of authors
Hopkins, S., 213 Hoshi, Michiyo, 290, 293, 298 Hübschmannová, Milena, 261, 262, 269, 280 Hudson, Joyce, 92, 93, 367 Hyman, Larry, 117 Ingham, Bruce, 137 Isaacs, Miriam, 247 Itkonen, Toivo Immanuel, 233 Jackson, Jean, 551 Jacobs, Neil, 245, 247 Jastrow, Otto, 167 Jiménez Moreno, Wigberto, 436 Johanson, Lars, 3, 16, 363 Johnson, Jean Bassett, 431 Jones, Robert B., 346 Jones, Russell, 326 Kaarhus, Randi, 486 Kabamba, Mbikay, 123 Kahn, Margaret, 167 Kamba Muzenga, 129 Kane, Thomas Leiper, 99, 101 Kapanga, André Mwamba, 123 Karttunen, Frances, 423 Kashoki, Mubanga E., 125 Katz, Dovid, 252 Kaufman, Terrence, 3, 16, 31, 32, 61, 68, 343, 382 Keesing, Roger M., 16 Kenesei, István, 280, 281 Kert, Georgij Martynovič, 231231 Khan, G., 24, 213 Klaus, Väino, 231 König, Ekkehard, 374 Korhonen, Mikko, 234 Kossmann, Maarten, 82 Kulonen, Ulla-Maija, 231 Kuruč, Rimma Dmitrijevna, 229, 237, 239 Kuteva, Tania, 16, 17, 92, 93, 98
La Polla, Randy J., 120 Lastra de Suárez, Yolanda, 439, 442, 446 Leslau, Wolf, 97 Lim, Sonny, 315 Lindenfeld, Jacqueline, 427, 431, 432 Lockhart, James, 423 Lötzsch, Ronald, 248, 253257 Loukotka, Cestmir, 551 Lowenstein, Steven, 246 Macalister, R. A. S., 151, 159 McConvell, Patrick, 367 Macdonald, R. Ross, 339 McGregor, William B., 369371 MacKenzie, David, 167, 172, 199202, 206, 207, 211 McWhorter, John H., 309 Maddieson, Ian, 107 Majtinskaja, K. E., 23, 241 Makihara, Miki, 388, 393398 Martins, Silvana A., 551, 552 Masica, Colin P., 286289 Maspero, Henri, 357 Matisoff, James A., 350 Matras, Yaron, 3, 610, 1517, 20, 21, 2427, 3134, 4146, 49, 5259, 6769, 151, 152, 163, 166, 227, 264269, 272276, 281, 289, 382, 421, 451, 489, 509, 559 Mazaudon, Martine, 292 Meakins, Felicity, 369 Mečkina, Jekatarina Ivanovna, 229 Mei, Tsu-Lin, 354 Meijering, Henk D., 250 Melià, Bartomeu, 548 Menz, Astrid, 179 Meyer, Ronny, 9192, 101 Möhlig, Wilhelm J. G. Moravcsik, Edith, 3, 33, 43, 47 Morínigo, Marcos, 549 Mougeon, Raymond, 16 Munro, Jennifer M., 366368
Index of authors Mushin, Ilana, 425 Mutzafi, H., 213 Muysken, Pieter, 3, 32, 62, 70, 364, 437, 499, 511, 518, 519 Myers-Scotton, Carol, 16, 519 Nau, Nicole, 16 Németh, Gyula, 216 Newman, Paul, 77 Nguyễn, Đình Hoà, 347, 357 Nguyễn, Phú Phong, 350 Nguyễn, Tài Cẩn, 353, 357 Nguyễn, Văn Lợi, 345 Nicolaï, Robert, 7676, 8888 Noonan, Michael, 283286, 289, 290, 298 Norman, Jerry, 356 Nugent, Paul, 107 O’Shannessy, Carmel, 369 Palancar, Enrique, 443 Payne, Doris L., 567 Polomé, Edgar C., 133 Prentice, D. J., 329 Pulleyblank, Edwin G., 358 Ramirez, Henri, 556, 558, 564 Reershemius, Gertrud, 17, 254, 256 Rhydwen, Mari, 367 Richter, Renate, 91 Rijkhoff, Jan, 2 Ring, Andrew J., 107, 108 Rogers, Clint, 285 Romaine, Suzanne, 130 Rose, Deborah, 365 Ross, Malcolm D., 3, 25, 33 Ruhlen, Merritt, 435 Ruiz de Montoya, A., 548 Sahagún, Bernardino de, 436 Sakel, Jeanette, 78, 1517, 23, 31, 41, 45, 289, 421, 567572, 578579
597
Salinas Pedraza, J., 439 Salmons, Joe, 330 Sammallahti, Pekka, 237, 239 Sandefur, John, 367 Sandefur, Joy, 367 Sasse, Hans-Jürgen, 379 Savić, Jelena M., 16 Schachter, Paul, 509 Scheller, Elisabeth, 230 Schicho, Walter, 128129, 133 Schmidt, Annette, 368369 Schrammel, Barbara, 269 Schuller, Rudolph, 569 Schultze-Berndt, Eva, 1719, 371, 376 Šebková, Hana, 261, 262 Shabibi, Maryam, 20, 137, 147 Shopen, Timothy, 369 Sidibé, Alimata, 75 Siegel, Jeff, 16 Sinclair Crawford, Donaldo, 439 Sinh, Vinh, 354 Siptár, Péter, 280 Snellgrove, David L., 283285 Socin, A., 213 Sorensen, Arthur, 551 Spitaler, Anton, 192 Stadnik, Elena, 230, 231 Stassen, Leon, 2 Stenzel, Kristine, 561 Stilo, Don, 168, 181, 213 Stolz, Christel, 3, 23, 32, 33, 382 Stolz, Thomas, 3, 23, 32, 33, 54, 55, 382 Struck, R., 107 Suárez, C., 524, 525, 530, 535, 546 Suárez, Jorge A., 439 Szabó, László, 232 Tadmor, Uri, 23, 458 Taylor, Keith W., 357 Thomason, Sarah G., 3, 10, 16, 3132, 61, 68, 317, 343, 382, 460 Thurgood, Graham, 345
598
Index of authors
Timm, Erika, 245 Tompa, József, 280 Törkenczy, Miklós, 280 Tosco, Mauro, 102 Treffers-Daller, Jeanine, 16 Trinidad Sanabria, Lino, 530 Troy, Jakelin, 366 Tryon, Darrell T., 358 Tryon, Ray, 366 Tsereteli, K. G., 213 Tufan, Şirin, 20, 32, 227 Urbano, A., 436 Vago, Robert M., 280, 281 van der Auwera, Johan, 56 van Driem, George, 283 Van den Heuvel, Wilco, 329 van Hout, Roeland, 3, 32 Van Minde, Don, 339 Van Valin, Robert D. Jr, 120 Villavicencio, Frida, 466, 469 Vitale, Anthony J., 130 Voegelin, Carl F. , 431
Voegelin, Florence M., 431 Voigtlander, Katherine, 442 Wang, Li, 354, 358 Weinreich, Uriel, 16, 65, 69, 247257, 382 Weir, E. M. Helen, 551, 554, 564 Weissberg, Joseph, 257 Westermann, Dietrich, 107 Wexler, Paul, 257 Whorf, Benjamin L., 431 Wichmann, Søren, 47, 69 Williamson, Kay, 107 Winford, Donald, 32, 36, 38, 43, 47, 70, 278 Wirasno, Umar, 309 Wohlgemuth, Jan, 47, 69 Wolff, H. Ekkehard, 76, 88 Yar-shater, E., 206 Yasugi, Yoshiho, 439 Zaborski, Andrzej, 91, 102 Zwicky, Arnold, 168