Professional Documents
Culture Documents
≥
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.
앪
앝 Printed on acid-free paper which falls within the guidelines of the
ANSI to ensure permanence and durability.
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 per-
mission in writing from the publisher.
Printed in Germany.
Contents
List of contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Yaron Matras and Jeanette Sakel
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 collec-
tion 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 rele-
vant 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 im-
pact 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.
egory, and those belonging to another. Both these issues can be expressed in
terms of hierarchies of borrowing. These in turn may contain either implica-
tional 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 distribu-
tion 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 interac-
tion). 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.
language in the past, but has little or no contemporary role in the speech com-
munity (Figure 3).
In order to be able to investigate the precise effects of contact on structural
compositions, a distinction is maintained throughout the questionnaire be-
tween the replication of linguistic matter (MAT) consisting of actual phono-
logical 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
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 limita-
tions 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 (Jamin-
jung), the Pacific (Rapanui), Central America (Nahuatl, Yaqui, Otomi, Pure-
pecha), and South America (Imbabura Quichua, Guaraní, Hup, Mosetén). It
contains languages with a tradition of native literacy (Vietnamese, Indone-
sian, 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 dif-
ferent 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 lex-
ical 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
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
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. Lin-
guistics 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 con-
vergence. Studies in Language 31 (4): 829865.
Matras, Yaron, April McMahon, and Nigel Vincent (eds.)
2006 Linguistic Areas: Convergence in Historical and Typological Perspec-
tive. 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 Mela-
nesia. In: Alexandra Aikhenvald and R. M. W. Dixon (eds.), Areal Dif-
fusion 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, Methodo-
logical 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 13
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 G. (ed.)
1997 Contact Languages: A Wider Perspective. Amsterdam: Benjamins.
Thomason, Sarah G.
2001 Language Contact: An Introduction. Edinburgh: Edinburgh Univer-
sity 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 pur-
pose 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 morpho-
logical 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 distinc-
tion 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 categoriza-
tion 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 conver-
gence”. 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 “inter-
ference without outright transfer”, which re-appears in Weinreich’s distinc-
tion between “source” and “recipient language” for MAT-borrowings, and
“model” and “replica language” for PAT-borrowings. The distinction is men-
tioned in many subsequent approaches to language contact, such as Gołąb’s
(1956, 1959) “form” versus “substance”, Johanson’s (1992) “global copy-
ing” and “partial copying”, Nau’s (1995) “material borrowing” and “loan-
meaning, loan-translation” and Treffers-Daller and Mougeon’s (2005: 95)
“borrowing, code-switching” versus “transfer”. Many other studies of lan-
guage 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 phe-
nomena of the MAT type, while more recently PAT has been integrated into
frameworks of code-switching (cf. Savić 1995, Bolonyai 1998 and Myers-
Scotton 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.
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 iden-
tify 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 mark-
ers (Reershemius, this volume).
How could the distinction between MAT and PAT-loans be employed in phon-
ology? In most contact situations, MAT-loans involve phonological changes
that can go in two directions:
Only the latter strategy, (2), involves phonological borrowing in which elem-
ents 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 phono-
logical distinctions would possibly be a case of PAT, since no new material
18 Jeanette Sakel
We will discuss the data from the sample by looking at the following cases:
(1) situations with overall MAT-borrowing; (2) situations with overall PAT-
borrowing; 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.
In the sample, Jaminjung, Biak and Vietnamese stand out in having predomi-
nantly 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 vol-
ume). 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 methodo-
logical.
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 pat-
tern 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 bilin-
gualism.
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
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 con-
tact 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 mark-
ers. 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
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 lan-
guages. 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, Chi-
nese 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 mark-
ers, 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
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 dom-
inant 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 lin-
guistic areas and situations of substrate influence display mainly PAT-loans,
MAT-loans are very frequent in the category of function words and indeed ap-
pear 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 con-
tact 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
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 at-
tempts to categorize contact situations.
We can conclude the following for the regularities behind MAT/PAT bor-
rowing: 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 Khuzista-
ni-Arabic is dominated by Persian. In the first case, the dominant languages
incorporate many MAT-loans from Arabic, while in the second case, MAT-
loans 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. With-
out 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
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 pat-
tern, 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 re-
main 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 bilin-
gualism. 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 Jezy-
koznawczego 15: 112.
1959 Some Arumanian–Macedonian isogrammatisms and the social back-
ground 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 Anthropol-
ogy 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 Gram-
maticalization 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. Lin-
guistics 36 (2): 281331.
2000 Fusion and the cognitive basis for bilingual discourse markers. Inter-
national Journal of Bilingualism 4 (4): 505528.
Matras, Yaron, and Jeanette Sakel
2007 Investigating the mechanisms of pattern replication in language con-
vergence. Studies in Language 31 (4): 829865.
Myers-Scotton, Carol
2002 Contact Linguistics: Bilingual Encounters and Grammatical Out-
comes. 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 grammat-
ical 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
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 cer-
tain 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 con-
tinuum.
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 likeli-
hood that, in respect of a particular structure which serves a particular func-
tion 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 com-
mon to both sets of communicative interactions in which they normally en-
gage: 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 clause-
combining, 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 in-
formation 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
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 con-
verge 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 seman-
tic-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 in-
terest 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 hier-
archical relationship, either (cf. Matras 1998). Where category or paradigm
values consistently show unequal or asymmetrical behaviour in contact situ-
ations, factors promoting borrowing must be sought in the inherent proper-
ties 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 particu-
lar linguistic function. The question that we ask is therefore: What is it that
makes one category (or category value) a more attractive candidate for “sys-
tem conflation” than another?
Elšík and Matras (2006: 370ff.), following Matras (1998), argue that bor-
rowing is motivated by cognitive pressure on the speaker to reduce the men-
tal processing load by allowing the structural manifestation of certain men-
tal processing operations in the two languages to merge. The need to do so
arises especially around operations that gauge the presentation of propos-
itional 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
3. Phonology
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 ar-
ticulation 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 simplifica-
tion 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:
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 Vietnam-
ese 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 lan-
guages 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
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 articula-
tory 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:
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 lan-
guage to another. The first is to maintain the complete integrity of the recipi-
ent 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 inter-
mediaries. 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
4. Typology
5. Nominal structures
“Core” relations (‘in’, ‘at’, ‘on’) are borrowed less frequently than “peripher-
al” 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 gen-
der from feminine to masculine in Mosetén, the loss of neuter gender in NE
Yiddish (as in the contact languages Lithuanian and Latvian), and the incipi-
ent 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
6. Verbal structures
Little attention has been granted in the literature to borrowing of features be-
longing 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 con-
sider the categories one by one. In the domain of tense, we see contact-in-
duced 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
(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 Rom-
ani 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 rela-
tionships. There is one case in the sample – Kildin Saami – where there ap-
pears 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 weak-
ening 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 cop-
ula, and both Indonesian and North-eastern Neo-Aramaic are reported to
have developed copula forms through imitation of forms in their contact lan-
guages. The borrowing of ‘have’ is on the whole marginal. Spanish tengo is
The borrowability of structural categories 47
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 adap-
tation 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 in-
finitive 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 gram-
matical effort in integrating verbs than for the integration of nouns. The bor-
rowing hierarchy
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 quantifica-
tion. Although it is now known that not all languages possess systems for
counting discrete entities, it is not necessarily the clash of systems of quanti-
fication 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 adopt-
ed as a result of contact in Mosetén, Indonesian adopts a Javanese tag-lexeme
indicating “teens” and re-organizes its earlier system of numeral juxtapos-
ition above 10 accordingly, Hup adopts the Tukano quintenary system for
numerals between 5 and 20, and a combination lexemes (‘ten-and-one’) re-
place single lexemes (‘eleven’) in some varieties of Kurmanji, replicating the
Turkish arrangement. This rather small group suggests the following implica-
tional hierarchy representing the likelihood of pattern-replication in numer-
als, which is yet to be confirmed by a larger sample:
More than two-thirds of the sample languages show some form of direct
matter-replication of numerals. This includes most of the languages that re-
organize their pattern of numerals, and which often employ borrowed numer-
als alongside the re-modelled “internal” or inherited system. In some cases,
numeral replication is subject to sociolinguistic constraints, with contact-lan-
guage numerals used as the preferred system for formal purposes such as cit-
ing dates and addresses and performing even simple mathematical tasks such
The borrowability of structural categories 51
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 young-
er 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
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 math-
ematical and other formal notations and transactions, but not with everyday,
casual expression of “nothingness”.
Sample languages that do not borrow numerals are Macedonian Turk-
ish, Khuzistani Arabic, Yiddish, and Manange (though incipient influence
of Nepali on the numeral system is reported). In all but the latter, we can at-
tribute the stability of numerals to a firm tradition of native-language educa-
tion, 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 lan-
guage of formal institutions and the public domain in the way of conceptual-
izing and expressing formal transactions surrounding quantification.
The hierarchies presented in (9) and (10) are fully in line with the obser-
vations 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:
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 hier-
archy, 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 ordi-
nals is a direct factor of this universal tendency to prioritize the ordinal ‘first’
The borrowability of structural categories 53
7.3. Connectors/conjunctions
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 expect-
ed 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 iden-
tical 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 disjunc-
tion; 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 bor-
rowing 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 fea-
ture that facilitates borrowing, nor of course that clause-combining is an op-
erational 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. Bor-
rowing 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, Indo-
nesian, 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
7.4. Particles
Not many languages in the sample borrow phasal adverbs, but those that do
show a clear implicational hierarchy
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, Jamin-
jung, Indonesian, Otomi), expressing an unexpected repetition of events.
Half the languages in the sample borrow focus particles, giving the impli-
cational hierarchy
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 atti-
tudes 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 per-
haps 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 fol-
lowing hierarchy (from Matras 1998) of borrowability, both frequency-based
and at least in most cases implicational, could be upheld:
vulnerable than its negative counterpart ‘no’. Rumungro and Mosetén bor-
row ‘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 hier-
archy for the borrowing of answer particles which agrees with that postulated
for Romani dialects by Elšík and Matras (2006: 343):
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:
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 bor-
rowable 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
9. Lexicon
(24) nouns, conjunctions > verbs > discourse markers > adjectives >
interjections > adverbs > other particles, adpositions > numerals >
pronouns > derivational affixes > inflectional affixes
62 Yaron Matras
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 36 11 0.305
Yaqui 6 3 0.500
Biak 7 3 0.428
Manange 10 3 0.333
K’abeena 10 3 0.333
Likpe 10 4 0.400
Mosetén 11 5 0.455
Mac. Turkish 12 3 0.250
Rapanui 13 3 0.230
Khuz. Arabic 13 4 0.307
Kildin Saami 15 5 0.333
Jaminjung 15 7 0.467
Vietnamese 17 6 0.352
Nahuatl 18 5 0.278
Tasawaq 18 7 0.389
Purepecha 19 7 0.368
Western Neo-Aramaic 20 6 0.300
Hup 21 5 0.238
Kurmanji Kurdish 21 6 0.286
Yiddish 22 7 0.318
Domari 24 9 0.375
Otomi 25 9 0.360
Quichua 26 8 0.307
Indonesian 26 9 0.346
Guaraní 31 10 0.323
Rumungro 31 10 0.323
64 Yaron Matras
evidence that borrowing begins with OPS, before continuing to other cat-
egories. 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 mark-
ers as well as in most cases unbound person markers (deixis and anaphora;
exceptions being reciprocal and reflexive pronouns, and “lexicalized” pro-
nouns 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 cat-
egories 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 bor-
rowing in the sample. A number of languages undergo considerable typo-
logical convergence: Macedonian Turkish, Domari, North-eastern Neo-
Aramaic, 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 bor-
rowed, 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 susceptibil-
ity 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 independ-
ent 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 be-
longs 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 sup-
plementing 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 hierar-
chies, may be defined as accessibility, cognitive complexity, and expected-
ness. 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
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 pat-
terns 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 boundar-
ies 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 (so-
called) 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 categor-
ies. With some, the motivation is lexical enrichment. With others, it is the fu-
sion 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 speak-
ers to dismantle the mental demarcation boundaries that separate their indi-
vidual “languages” and, around a particular selection of categories, to make
full use of their entire repertoire of linguistic structures and forms irrespec-
tive 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 Inter-
national 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 conver-
sational 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 bilingual-
ism, however rudimentary, or at least with an exposure to settings of communi-
cative 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 frequen-
cy, 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 bor-
rowing 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 Indo-
Iranian 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 obli-
gation; 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 ar-
ranged from left to right): light verbs < indirect insertions < direct insertion
< paradigm transfer. But we have no grounds on which to challenge this hier-
archy, 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 borrow-
ing 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 conjunc-
tions, phasal adverbs, focus particles, discourse markers, time deixis, adjective
comparison, constituent order, syntax, basic cultural vocabulary.
16. These are: Numerals, personal pronouns, demonstratives, indefinites, interroga-
tives, 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
Matras, Yaron
1998 Utterance modifiers and universals of grammatical borrowing. Lin-
guistics 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 con-
vergence. 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 Mela-
nesia. In: Alexandra Aikhenvald and R. M. W. Dixon (eds.), Areal Dif-
fusion 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 situ-
ations 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 Univer-
sity Press.
The borrowability of structural categories 73
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 im-
portant 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 con-
stitutes 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 Afroasi-
atic 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
2. Phonology
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 phon-
ology. 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. New-
man 2000: 399). The underlying vowel quality reappears in contexts where
the vowel in question is lengthened, e.g. Tasawaq:
In the second place, Hausa and Tasawaq have similar tonal features. Both lan-
guages have three tones, High, Low and Falling. Different from other Song-
hay 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 restric-
tions on the occurrence of Falling tones, it is simplified to a High tone in cer-
tain syllable types. The basic all-Low contour reappears in bound syntactic
contexts, e.g. in nouns when they are followed by a numeral:
3. Noun morphology
With borrowed words, on the other hand, plurality is marked on the noun ra-
ther 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 constitu-
ent of the noun phrase, e.g.
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 canon-
ical way of integrating loanwords from other languages than Tuareg, e.g.
A certain number of borrowings from Arabic and Hausa are treated as if they
were Songhay nouns, e.g.
Tuareg has a derivative gender system, in which masculine denotes male ani-
mates 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 ani-
mates. 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 fol-
lowing pairs:
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 de-
rived 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 (al-
ways 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.
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 lengthen-
ing 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 81
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, al-
though 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 Song-
hay languages, these categories are expressed by porte-manteau morphemes,
which immediately precede the verb. The positive Perfective is unmarked.
Unlike other Northern Songhay languages, the subject is not obligatorily ex-
pressed by a pronoun when a lexical subject is present. The Tasawaq con-
struction probably reflects proto-Songhay at this point, e.g.
Phrases with a lexical subject followed by a subject pronoun are quite fre-
quent, e.g.
82 Maarten Kossmann
This may represent influence from Tuareg, which has obligatory subject in-
flection, or Hausa, where the subject is obligatorily expressed in a porte-
manteau 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.
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.
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- ini-
tial verbs, on the other hand, one finds an interesting distribution (see Koss-
mann 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 Re-
sultative. 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.
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 differ-
ence 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. Tasa-
waq does not have the intricate interaction between underived Songhay-based
verbs and derived Tuareg-based verbs as found in some other Northern Song-
hay languages, such as Tadaksahak (Christiansen and Christiansen 2002).
6. Verbal nouns
Tuareg verbs are taken over together with the corresponding Tuareg verbal
noun. The great deal of variation and irregularity in Tuareg verbal noun for-
mation is reflected in Tasawaq, e.g.
84 Maarten Kossmann
In the second group, the numeral precedes the head noun and is linked to it
by means of a particle Hǹ, e.g.
This closely reflects the Tuareg pattern, in which the genitive preposition n
occurs between the higher numeral and the noun.
Tasawaq 85
(27) hááwí n gí
cow of grease
‘grease of a cow’
8. Syntax
9. Conclusion
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 speak-
ers 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 Lan-
guage 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 float-
ing High tone. In Tuareg transcriptions ă indicates a short low central vowel.
Tone is consistently marked; absence of tone marking reflects the author’s un-
certainty 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: Lin-
com.
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
1. Background
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 speak-
ing 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 un-
attested 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 sub-
area. 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 under-
taking 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 find-
ings 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 mar-
ginal 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 dis-
cussion cf. Crass 2002). K’abeena possesses four ejectives, namely the ejec-
tive 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
3. Nominal structures
The ablative case marker, which itself is not considered to be a result of lan-
guage contact, is used with verbs to form ‘since’-temporal and real condi-
tional 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 func-
tion as ablative case marker and as marker for ‘since’-temporal clauses, ex-
ample (2) the function as real conditional clause marker.
4. Verbal structures
The past tense marker is used in K’abeena and in all other investigated lan-
guages 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.
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.
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 grammat-
icalization 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 lan-
guage 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.
‘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 adjec-
tives (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 in-
ventory 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 trad-
ition, 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 Ethi-
osemitic languages to have shifted their word order because of intensive con-
tact 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, comple-
ment clauses, adverbial clauses) the fully inflected verb ’ihu ‘be’ occurs.
8. Lexicon
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 con-
sider especially interesting. The verb ’afu has the basic meaning ‘hold, catch’
and the secondary meaning ‘begin start’, the verb ’alapp’u ‘play’ the sec-
ondary 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 differ-
ent kinds (food, cloths, topics related to Islam) and some abstract nouns.
Examples for abstract nouns are K’AB suusi from AMH sus ‘mania, pas-
sion, 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 ‘pun-
ishment 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 bil-
laawa ‘(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 phe-
nomena 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 interjec-
tion to chase away cats and fillers, e.g. K’AB bal, AMH bäl ‘well’. In the case
102 Joachim Crass
9. Conclusion
Abbreviations
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 Nordostafri-
kas/Westasiens).
References
Bisang, Walter
1996 Areal typology and grammaticalization: Processes of grammatical-
ization 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
1. Background
Ega
Avikam
Alladian
Ajukru
Abidji
Abbey
Attié
Potou Ebrie
Mbatto
Krobu
Potou-Tano
Abure, Eotilé
West Tano
Akan
Tano Nzema-Ahanta
Central Tano Anyi, Baule, Anufɔ
Bia
Efutu-Awutu
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 Linguis-
tics, Literacy and Bible Translation (GILLBT) in 1987. Today there are pam-
phlets of stories, proverbs and literacy materials as well as a pamphlet contain-
ing 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 conver-
gence. 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 adap-
tation, 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 presuma-
bly inherited from Proto Niger-Congo. Features of nouns, including class and
number, are marked by prefixes on nouns. Ewe, on the other hand, is predomi-
nantly isolating with agglutinative features. Plural marking in Ewe is by an
enclitic which attaches to the element in the NP that occurs before the Intensi-
fier. 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 modi-
fiers in Sɛkpɛlé are formed by verbal reduplication similar to the pattern found
in Ewe. For instance,
Likpe (Sɛkpɛlé) 111
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 suf-
fixed with the form -m ‘pl’ to signal their plurality. This form is heterosemi-
cally related to the third-person plural pronoun.
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 categor-
ies by prefixes on the verb (including for example past progressive). How-
ever, 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).
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 lan-
guage. This may be due to areal rather than Ewe specific influence. Moreover,
the construction is one of the structures in which Likpe uses a complemen-
tizer that is borrowed from Ewe, namely b ‘quot’, as illustrated in (5).
Likpe (Sɛkpɛlé) 113
There is no special marking of verbs that are borrowed. They are only phono-
logically 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 gram-
maticalized 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-tə́ 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.’
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 disjunc-
tion 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 con-
texts 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 settle-
ment history narrative.
In addition, two adverbial connectors are borrowed from Ewe: álé bé ‘so
that’ (example 12) and tógb bé ‘although’.
(13) a. Likpe
lə-fə n bə-n-sí ko kasɔ́-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, 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 seriali-
zation 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 lan-
guages (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 lan-
guages 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 constitu-
ent 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 se-
mantic 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).
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 gram-
mar 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 mark-
ing 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 post-
positions, 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 re-
ceive 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
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 dec-
ade 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 Refer-
ence Grammar (RRG) see Van Valin and La Polla (1997: 141147). To charac-
terize 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”: Under-
goer voice constructions in Ewe and Likpe. In: M. E. Kropp Daku-
bu and E. Kweku Osam (eds.), Studies in the Languages of the Volta
Basin, Volume 3, 4362. Legon: Department of Linguistics, Univer-
sity of Ghana.
2005b Multiverb constructions on the West African littoral: Microvariation
and areal typology. In: M. Vulchanova and T. A. Åfarli (eds.), Gram-
mar and Beyond: Essays in Honour of Lars Hellan, 1542. Oslo:
Novus Press.
2007a Grammars in contact in the Volta Basin (West Africa): On contact in-
duced 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) 121
2007b 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: Ber-
keley, 2325 March 2001.
Dakubu, M. E. Kropp, and Kevin C. Ford
1988 The Central-Togo languages. In M. E. Kropp Dakubu (ed.), The lan-
guages 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 gen-
etic 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: Diet-
rich 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 Ghana-
Togo 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 Cen-
tral Volta Region. Legon: Institute of African Studies.
1995 Lɛlɛmi tone. Papers from GILLBT’s seminar week 30 January–3 Feb-
ruary 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, Vol-
ume 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 Lan-
guages: An Introduction, 1142. Cambridge: Cambridge University
Press.
Grammatical borrowing in Katanga Swahili
Vincent A. de Rooij
1. Background1
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 mi-
grants, 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 lan-
guages. 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 borrow-
ings. Many of these words denote Islamic concepts and therefore play no role
in Katanga where Islamic influence is virtually non-existent. In the few gen-
erally 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-
close-mid e o
open a
Katanga Swahili 125
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/. Further-
more, 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 prena-
salized 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 sys-
tem 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 cat-
egorically 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 noun-
class 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 Luba-Kasaï Bemba
Swahili
1 m- mu- mu- mu-
2 wa- ba- ba- ba-
3 m- mu- mu- mu-
4 mi- mi- mi- mi-
5 ji-/ri-/∅- di- li-
6 ma- ma- ma- ma-
7 ki- ki- t- t-
8 vi- bi- bi- fi-
9 N-/∅- N- niN-
10 N- niN-
10 → 6 ma-
11 u- lu- lu- lu-
12 ka- ka- ka-
13 tu- tu- tu-
14 u- bu- bu- bu-
15 ku- ku- ku- ku-
16 pa- pa- ɸa- pa-
17 ku- ku- ku- ku-
18 mu- mu- mu- mu-
Katanga Swahili 127
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 Swa-
hili 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- ex-
presses direction (toward), and mu- expresses being inside of (in).
Noun–Adjective agreement has been simplified radically: most adjec-
tives have only one generalized form that is used with nouns from differ-
ent 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 agree-
ment on the verb. In Katanga Swahili object concord markers co-indexing
non-human objects are very seldomly used, except for classes 7/8. The gener-
alized, 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 lan-
guages. 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 dummy-
elements that have to be realized for morpho-syntactic reasons. The most
frequently used preverbal tense affixes in Katanga Swahili are: -na- mark-
ing present tense, -li- or -ri- marking past tense, and -ta- marking future
tense. According to Schicho (1988: 568), in a narrative sequence time refer-
ence needs to be marked only once by a tense affix on a verb, by sentence-
initial 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])
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 dif-
ference: 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.
Numerals in dates and years are mostly in French. Discourse markers, includ-
ing items that are traditionally classified as conjunctions, constitute another
Katanga Swahili 131
group of items that occur almost exclusively in French (De Rooij 2000). Fre-
quently 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.
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 lan-
guage 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
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 ‘za-
ïrianization’. 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 lan-
guage 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 liter-
acy, 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). Ant-
werpen: De Sikkel.
Fabian, Johannes
1986 Language and Colonial Power: The Appropriation of Swahili in the
Former Belgian Congo, 18801938. Cambridge: Cambridge Univer-
sity Press.
1990 History from Below: The ‘Vocabulaire of Elisabethville’ by André Yav
(Texts, translations, and interpretive essay). Amsterdam: John Ben-
jamins.
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/tshibum-
ba3b.html (3 August 2007).
Fetter, Bruce
1976 The Creation of Elisabethville 19101940. Stanford: Hoover Institu-
tion 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 Cen-
trale.
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. Prob-
lèmes 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
1. Background
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
3. Morphological typology
A major change under Persian influence is the levelling of the status of at-
tributes. In Arabic, adjectival attributes follow the head noun, and agree with
the head noun in gender, number, as well as in definiteness:
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 conson-
ants, resulting in gemination of that consonant):
Note that in the adjectival attributive construction in (4a and 4c) overt def-
initeness 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 mark-
er of attribution, which matches the Persian (definite) Ezāfe marker -(y)e. Its
distribution now resembles that of the Persian Ezāfe attributive marker: It ap-
pears, 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 attribu-
tion, as in (4d). But in Kh. Arabic we find it in adjectival attributive construc-
tions as well, as in (5a–b); it even attaches directly to adjectives, as in aly-at
‘high.f’ in (5b):
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’
icle in both positions. Consider example (7), where the nouns are masculine,
and there is no option of using an overt Construct State marker:
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 at-
tribute 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 Per-
sian, 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 definite-
ness 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
notably in the absence of a definite article with definite head nouns of rela-
tive clauses. This too follows a Persian model (where definiteness generally
remains unmarked):
5. Verbal structures
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 auxil-
iary; the construction thus matches the Persian pluperfect:
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. Arab-
ic and are not perceived by speakers as foreign. The category that is most
obviously influenced by Persian is that of discourse markers with a primar-
ily interaction-qualifying rather than syntactic-semantic function: xō/xōb/xōš
‘well’, xōlāse ‘in sum’, albate ‘of course’, hič ‘at all, altogether’, ham ‘indeed,
well’:
These are accompanied by Persian-derived focus particles: ham ‘too’ and ham
… ham ‘both … and’.
Further Persian borrowings that are generalized in Kh. Arabic are the con-
cessive subordinating conjunctions agarče and bā īnke, both 'although/ even
though’, and the factual complementizer ke ‘that’:
Khuzistani Arabic 145
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 func-
tion of a relativizer. Nonetheless, occasionally Persian ke is also found in the
position of the relativizer:
7. Constituent order
One change in constituent order has already been mentioned above, in Sec-
tion 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 Per-
sian-type constituent order, whereby each attribute must immediately follow
its head. We repeat example (7) here:
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.’
8. Lexicon
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
Abbreviations
References
Ingham, Bruce
1982 North East Arabian Dialects. London: Kegan Paul International.
1997 Arabian Diversions: Studies on the Dialects of Arabia. Reading: Ith-
aca Press.
Shabibi, Maryam
1998 Variation in the use of Persian loan words among Iranian Arabic speak-
ers. 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 spo-
ken 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 resem-
bles 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 fam-
ily and, to a limited extent, with members of other Dom communities. The pre-
cise 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 monolin-
gual 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 situ-
ation at least in urban Dom communities elsewhere in the Middle East.
2. Phonology
The Domari sound system strongly resembles that of its contact language, Pal-
estinian Arabic, though it is not always obvious that this is due to borrowing
152 Yaron Matras
(1) n-mang-am-éʔ
neg-want-1sg-neg
‘I don’t want.’
3. Morphological typology
crete evidence linking it with the influence of Arabic, which is not ergative,
either.
4. Nominal structures
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):
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
(6) fī šare-ma
in street-loc
‘on the street’ (<Arabic fī ‘in’)
(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’)
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 pos-
sessive 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 lan-
guages 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 auxil-
iaries expressing habitual, inceptive, and iterative aspect, which retain their
Arabic tense and person inflection. Unlike in Arabic, however, these auxil-
iaries are followed by an indicative, not subjunctive, form of the main lexic-
al verb:
The conditional is formed with the Arabic auxiliary kān with the Domari an-
terior 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 bor-
rowed 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 organ-
ization 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 construc-
tions, and employs a present-subjunctive form of the embedded verb in-
stead:
(mana illi torim iyyā-h ‘the bread that you gave me [it]’; see below, on rela-
tive 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:
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 whole-
sale from Arabic, including their lexical forms, rendering all adjectives in the
language (except those whose positive forms are also Arabic-derived) supple-
tive:
7. Constituent order
Since most Indo-Aryan and Iranian languages are SOV, this can be consid-
ered 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 construc-
tion in which the noun is followed by the adjective, to which a non-verbal
predication marker is attached (19b):
8. Syntax
(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.’
The Arabic verbal negator mā is used for the negation of modal and auxil-
iary verbs borrowed from Arabic, as well as in the vicinity of Arabic-derived
prepositions and particles:
Clause combining rules in the two languages are by and large identical,
Domari drawing entirely on the pool of Arabic connectors and subordinating
conjunctions:
9. Lexicon
All semantic domains of the Domari lexicon adopt Arabic loans. The only
limitations on lexical borrowing are in the domain of grammatical vocabu-
lary, more specifically referential and deictic pronouns, and place deictics,
162 Yaron Matras
10. Conclusion
It would be useful at this stage to remind the reader that under “borrow-
ings” 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 bor-
rowing into Domari can be described as nothing less but massive. It is in-
deed 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 per-
son 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 con-
tact-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 ex-
treme 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 gram-
maticalization 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 every-
day vocabulary items, deictic and anaphoric reference tools, and the structur-
ing 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 entire-
ly 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 construc-
tion 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
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. Lin-
guistics 36: 281331.
1999 The state of present-day Domari in Jerusalem. Mediterranean Lan-
guage 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 lan-
guages, spoken across a large area of the Middle East centering at the in-
tersection 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 re-
ferred to as Kurmanji (also spelt Kurmanjî, Kurmanci, Kurmancî); the Cen-
tral 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 Azerbai-
jan, 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 var-
ieties 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 focuss-
ing 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 tradition-
ally 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 pub-
lications 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 Turk-
ish 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 fic-
tion 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, variegat-
ed, 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 Anato-
lia, Kurdish enjoyed considerable prestige as the language of many power-
ful 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 serv-
ice, 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
2. Phonology
3. Morphological typology
Central Anatolian Kurdish is both prefixing and suffixing, with a small meas-
ure of Ablaut, restricted to the past stem formation of a closed set of irregu-
lar 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 disap-
pearing, but this need not necessarily be linked to contact influence (as sug-
gested 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 pro-
nouns 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 Kurd-
ish, personal pronouns in oblique functions frequently cliticize and move to
other constituents:
such mobile pronominal clitics throughout its attested history, cf. the Anato-
lian Kurdish and Turkish versions of (1):
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 re-
sembling 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 rela-
tions 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:
The Tunceli “postposition” -ra is the functional equivalent of the Turkish da-
tive 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 bor-
rowing 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 post-
positions, göre and ait, but it does not use them as postpositions:
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 fur-
ther examples):
This type of construction, historically certainly older than (10), is now entire-
ly lacking in Central Anatolian Kurdish, and was never present in Turkish.
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 distin-
guishing voice, tense, evidentiality, negation, interrogative, mood, and as-
pect, 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 unwit-
nessed 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 follow-
ing (see Bulut 2006 for further examples and discussion):
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 nom-
inal. 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 reper-
toire 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 Ana-
tolian 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 expres-
sions 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 clause-
initially 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).
As discussed in the preceding sections, words of all major classes are bor-
rowed (nouns, verbs, adjectives), numerals, adverbs, connectors, discourse
176 Geoffrey Haig
(18) ama ‘but’, fakat ‘but (more stongly contrastive), artık ‘still, yet, al-
ready’, 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 speak-
ers 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 re-
placed wholesale by Turkish elements, but Turkish elements are used along-
side them, with the frequency and distribution of the Turkish elements vary-
ing heavily from speaker to speaker, and speech situation to speech situation.
The heaviest concentration I have seen is in the Tunceli data, where the speak-
er’s narratives are regularly interspersed with Turkish discourse markers, ad-
verbs, 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)
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. How-
ever, 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 in-
serted, 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 corres-
ponding 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 lex-
ical 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):
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:
In the processual sense, the verb precedes the copula complement. Compare
(23) with (22):
Thus most varieties of the Northern Group mark the semantic distinction be-
tween 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:
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:
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
Standard Turkish has non-finite verb forms for clause combining (converbs),
for complementization (participles, infinitives), and for relative clause for-
mation (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 consist-
ing 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 lan-
guages 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
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 distin-
guish 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 Azerbai-
jan 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 morph-
ology, 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 de-
funct.
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 dif-
ficult 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. Anka-
ra: 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 Typ-
ology and Universals 55 (1): 2548.
2003 Sprachenvielfalt und Sprachenpolitik am Rande Europas: die Minder-
heitensprachen der Türkei. In: Dieter Metzing (ed.), Sprachen in Eu-
ropa. 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 Con-
tact Areas. Historical and Linguistics Aspects, 279295. Wiesbaden:
Harrassowitz.
Kurdish (Northern Group) 183
1. Background
The modern Aramaic dialects are the remnants of a wide variety of old and mid-
dle 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 morph-
ology and vocabulary more archaic than that of Baxʿa, where gender distinc-
tion 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 dif-
ferences 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 speak-
ers. The language is a vernacular, not written, and only spoken in everyday
life, within the village and the families. The introduction of a writing sys-
tem (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 origin-
ate 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 close-
ly related languages that widely correspond in phonology and morphology,
facilitates the mutual adoption and assimilation of loans, but makes the iden-
tification 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 so-
called 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 ba-
sically after vowels. In WNA the acoustic difference between spirant and plo-
sive 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 influ-
ence 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:
The phonemes /¦̣/ [ð~] and /ž/ [ʒ] (Baxʿa /ǧ/ [ʤ]) are restricted to Arabic
loans:
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]:
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 phenom-
enon 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:
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
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).
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:
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:
Some Arabic loans receive the original Aramaic diminutive endings -ōna (m)
and -(a)nīa (f), which express diminution no longer:
Western Neo-Aramaic 189
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:
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 neigh-
bouring 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 per-
fect 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:
To express a progressive action the Arabic prefix ʿam- (< ʿammāl) with present
tense is used:
190 Werner Arnold
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’:
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:
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 fol-
lowing way:
Western Neo-Aramaic 191
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:
On the other hand the Aramaic passive stem ččaCCaC (< ettaCCaC) can also
be formed from Arabic roots:
All ordinal numerals are borrowed from Arabic and have with the excep-
tion 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:
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 in-
corporated into the Aramaic vocabulary and among young speakers have
sometimes replaced the inherited Aramaic form:
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 mor-
phological pattern ʾaCCaC:
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 fol-
lowed 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.’
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 pre-
serve 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
Abbreviations
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), Wies-
baden.
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 Arbeits-
treffen der Arbeitsgemeinschaft Semitistik in der Deutschen. Morgen-
Western Neo-Aramaic 195
ländischen Gesellschaft vom 11. bis 13. September 2000 an der Frie-
drich-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 sozial-
geographischen Einführung von Anton Escher (Semitica Viva 8),
Wiesbaden.
Correll, Christoph
1978 Untersuchungen zur Syntax der neuwestaramäischen Dialekte des Li-
banon (Maʿlūla, Baḫʿa, Ǧubbʿadīn); mit besonderer Berücksichtigung
der Auswirkungen arabischen Adstrateinfllusses; nebst zwei Anhän-
gen zum neuaramäischen Dialekt von Ǧubbʿadīn (= Abhandlungen
für die Kunde des Morgenlandes XLIV, 4), Wiesbaden.
Grotzfeld, Heinz
1965 Syrisch-arabische Grammatik (Dialekt von Damaskus) (Porta lin-
guarum 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 Neo-
Aramaic 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 south-
eastern 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, Arme-
nia 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 com-
munities 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 dia-
lects, 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 Sulemani-
yya, 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 cleav-
ages 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
2. Phonology
This shift is found in all Jewish NENA dialects that were spoken in commu-
nities 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 weaken-
ing 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 conson-
ants sometimes shifts to /r/:
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 San-
jaq and Ruwanduz (MacKenzie 1961: 4, 28).
After a vowel, /d/ and /z/ may freely alternate with each another. This pro-
cess appears to be restricted to specific lexical items and is best treated as an
alternation of phonemes rather than as allophonic alternation:
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
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 postvo-
calic 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 femi-
nine). 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:
The demonstrative adjectives in Aramaic have also lost the distinction be-
tween singular and plural in imitation of Kurdish:
North-eastern Neo-Aramaic 201
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.) ‘hedge-
hog’ 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 size-
able 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:
In a few cases the Aramaic feminine ending -ta is added to Kurdish loans that
have been assigned feminine gender:
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 art-
icle 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). Fur-
thermore, the morphological behaviour of the particle in the Aramaic dialect
is different from that of the particle in Kurdish. In the Aramaic dialect it can-
not take any further suffixes. In Kurdish, by contrast, it may take plural (-ān)
and pronominal suffixes:
202 Geoffrey Khan
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:
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:
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 pro-
nominal 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:
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 pas-
sive participle and an enclitic form of the verb ‘to be’. The same two types of
past conjugation are found in the Kurdish dialect:
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 inflec-
tion 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.’
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 tens-
es, 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 con-
jugation 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:
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.’
(21) bi-šqala-vit.
in-taking-cop:2ms
‘You are taking.’
206 Geoffrey Khan
A large number of other grammatical words have been borrowed from Kurd-
ish. These include a variety of particles and adverbials, including subordinat-
ing 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 ex-
pressing 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 ‘re-
turning back’, ‘restoring’, ‘repetition’ or ‘completion’:
(22) hiye-wa.
come:past:3ms-particle
‘He came back.’
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 Kurd-
ish. 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):
Some particles in the Aramaic dialect are borrowed from Kurdish but do not
correspond exactly to what is found in Sulemaniyya Kurdish. On some occa-
sions we seem to be dealing with doublets. This applies, for example, to the par-
ticles 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. Babylo-
nian 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 Kurd-
ish dialect (see examples 16 and 17).
Various borrowed subordinating particles are used to introduce subordin-
ate clauses in the Aramaic dialect.
The Kurdish subordinating particle ka-, which is used in the Kurdish dia-
lect of Sulemaniyya and the surrounding region (MacKenzie 1961: 131), has
been borrowed by the Aramaic dialect (pronounced either ka- or ga-) to in-
troduce 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 Kurd-
ish, the relative clause follows its head:
The Kurdish particle ka- is sometimes used in the Aramaic dialect as a subor-
dinating temporal conjunction with the sense of ‘when’:
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:
7. Lexicon
Borrowed verbs, by contrast, are inflected fully with the Aramaic verbal in-
flection. 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.
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 ex-
ample, 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 dir-
ect 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 * com-
pletely. As a result, the base of the present conjugation of the verb resembles
phonetically the corresponding Kurdish verb:
(34) Aramaic
k-e (3ms indicative) he (3ms subjunctive)
k-en (3pl indicative) hen (3pl subjunctive)
North-eastern Neo-Aramaic 211
Kurdish
e (3s indicative) b-e (3s subjunctive)
en (3pl indicative) 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 Kurd-
ish invariable verb abe (MacKenzie 1961: 106), which it resembles phonet-
ically, e.g.
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 lan-
guages 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 re-
mote 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:
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 specific-
ally 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 com-
mon 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:
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 dem-
onstrates that it is a heritage from the Aramaic dialect as it was spoken in
Kurdistan, e.g. mira < mila ‘circumcision’.
Abbreviations
Notes
1. The term was coined by Hoberman (1988: 557) to replace ‘Eastern Neo-Arama-
ic 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 lin-
guistique 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 The Jewish Neo-Aramaic Dialect of Koy Sanjaq. Wiesbaden: Harras-
sowitz.
Socin, A.
1882 Die neu-aramäischen Dialekte von Urmia bis Mosul. Texte aund Über-
setzungen. 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
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 trans-
ferred occasionally into native Turkic words: tsıs ‘shut up’ (cf. Tk. sus). Ini-
tial 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 Rume-
lian 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 Mace-
donian, and are productive with Turkish word stems: arkadaş ‘friend’ (gen-
der-neutral, and by default masculine), arkadaş-ka ‘female friend’; koyşi
Macedonian Turkish 217
(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
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 219
6. Constituent order
d. Standard Turkish
damad-ın eşya-lar-ı
groom-gen clothes-pl-3sg.poss
‘the groom’s clothes’
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 inde-
pendent copula stem in i-, which, however, occupies the position between the
subject and the predicate noun, as in the contact languages:
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.):
The default position for objects that constitute new topical information in lex-
ical predications remains, however, the pre-verbal position:
222 Yaron Matras and Şirin Tufan
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 sur-
rounding Indo-European languages. Essentially, these are based on the juxta-
position 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 embed-
ding.
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):
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 Turk-
ish infinitive. The same type of construction is used in manipulation clauses
(modal complements with different subjects):
In this manner, GT aligns itself with the other Balkan languages also in re-
spect 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 comple-
ments, and the ki complementizer (and indicative mood) in epistemic com-
plements.
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 forma-
tion 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
Thus, where Turkish already operates with finite subordinations, there ap-
pears to be no motivation to re-organize the structure of clause linking.
8. Lexicon
9. Conclusion
It is interesting to note once again that Turkish has only been a minority lan-
guage 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-speak-
ing society in the public domain, employment, technology, and so on. Never-
theless, 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
Abbreviations
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 diffi-
cult 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. Tübin-
gen: Narr.
2004 Layers of convergent syntax in Macedonian Turkish. Mediterranean
Language Review 15: 6386.
Németh, Gyula
1956. Zur Einteilung der Turkischen Mundarten Bulgariens. Sofia: Bulga-
rische 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
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 Scandi-
navia), each of which includes several languages.
One characteristic of the phonology of most Saami languages is the oc-
currence 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 lan-
guages – is the relatively large consonant inventory, which is mostly due to
the fact that almost all consonants have a phonologically distinct palatal-
ized 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 con-
sonants seems to be uncommon cross-linguistically (cf. Stadnik 2002: 31,
elsewhere).
were connected with an immense influx of manpower from Russia and other
republics of the Soviet Union. The dissolution of the traditional Saami com-
munities – 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 ori-
ginal speech communities. The former compact Saami settlements and co-
herent 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 pol-
itical decisions (for an overview on Kola Saami history and further references
see Kulonen, et al. 2005: 261265).
Along with the changes in Kola Saami culture and society, we witness a
number of contact-induced linguistic changes. These changes – almost exclu-
sively 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 contact-
induced 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 borrow-
ings which are dealt with. Only a few grammatical features are listed briefly
by the author, such as borrowed phonemes in loanwords, the borrowed su-
perlative 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, how-
ever, 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 word-
initial 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 rein-
deer's antlers’ – būrka /buÃrka/ <Ru búrka ‘leather boot’ or šūrr /ʃuÃrÃ/ ‘big’
– žoarr /ʒɒÃrÃ/ <Ru žara ‘heat’. Also the occurrence of the voiceless velar
fricative in word-initial position is restricted to loanwords, as for example in
xoz'enҍ /xozen/ <Ru xoz'áin ‘host’. Another example of a borrowed phono-
logical feature is the introduction of word-initial consonant clusters, which
do not otherwise occur in Kildin unless in loanwords such as, for example,
the word floht [flɔtÃ] <Ru flot ‘fleet’.2
Besides the examples mentioned, Kildin phonology does not seem to ex-
hibit any other borrowed features.
3. Nominal structures
In Kildin the illative case has expanded its domain to impersonal construc-
tions 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 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 suf-
fixes is on the whole very typical of Russian as well as of other Slavic lan-
guages. 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’.
The weak consonant stem in (5a) pēr̥t- (nom.sg pēr̥̥rt) is triggered by the suf-
fixation 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 suf-
fixal 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!’ (Lyt-
kin 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 ex-
istence 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.
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 re-
corded in 1975 (Sammallahti 1998: 148).3
In the Saami languages negative indefinites are usually derived from inter-
rogatives 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 par-
ticle ni (~ né); consider the examples in (10).
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 inter-
rogative 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
Note that ni in Russian is not a prefix since it can be separated from the inter-
rogative 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.
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
5.3. Connectors
Besides the Common Saami phrasal and clausal coordinator ja ‘and’ which is
a Proto Saami borrowing from Germanic, cf. Gothic -jah (Sammallahti 1998:
249), the coordinator i ‘and’ <Ru is often used by Kildin speakers. Other Rus-
sian coordinators in Kildin are ili ‘or’ <Ru and ne ‘but’ <Ru no ‘but’. Kildin
also borrowed a relative particle katóre from the Russian relative pronoun
kotóryj m ‘which’.
6. Lexical borrowings
adopted without any phonological adoptation. The word for ‘car’, mašína
(<Ru mašína ‘car’), serves as an example. The word retains both its origin-
al second-syllable stress (Saami has first-syllable stress as a rule) and its
original syllable structure (trisyllabic word stems are quite rare in Kildin).
The loanword pāss'pe /paÃsÃpe/ ‘thanks!’ (<Ru spasíbo [spa'sibɐ] ‘thanks’)
on the other hand clearly belongs to an older stratum. The first consonant
/s/ was lost since Saami does normally not allow word-initial consonant
clusters. The word stress is shifted from the second to the first syllable. The
vowel and stem consonant in the first syllable have become long while the
second syllable vowel is apocoped (the palatalization of the cluster /sÃp/ is
a remnant of the front vowel /i/ of the contracted second syllable). Finally,
the word final vowel [ɐ] of the Russian source word is further reduced to
schwa.
7. Conclusions
Abbreviations
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 Rus-
sian 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 suf-
fix -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 gram-
mar]. Nyelvtudományi közlemények 17: 145.
Itkonen, Toivo Immanuel
1958 Koltan- ja kuolanlapin sanakirja = Wörterbuch des Kolta- und Kola-
lappischen. (Lexica Societatis Fenno-Ugricae 15.) Helsinki: Suoma-
lais-ugrilainen seura.
Kert, Georgij Martynovič
1971 Saamskij jazyk (kil'dinskij dialekt): Fonetika, morfologija, sintaksis
[The Saami language (Kildin dialect): Phonetics, Morphology, Syn-
tax]. Leningrad: Nauka.
1994 Saamsko-russkie jazykovye kontakty [Saami–Russian linguistic con-
tacts]. In: Pjotr Mefodievič Zajkov (ed.), Pribaltijsko-finskoe jazykoz-
nanie, 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 fin-
no-ugorskix jazykov i literatur v ix vzaimosv'azax s jazykami i liter-
aturami 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 grammat-
ical 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.), Grenz-
gänger: 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 areal-
typologische 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 Yiddish language originates from medieval times and developed through
contact: Jewish speakers of Old High German and later Middle High Ger-
man 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 litera-
ture, 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 linguis-
tics 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 phon-
ology (Timm 1987).
246 Gertrud Reershemius
Yiddish in the West (Western Yiddish) remained in contact with spoken Ger-
man varieties and the slowly evolving German standard language. However,
migration of Jews to Slavic-speaking areas in Eastern Europe due to persecu-
tion in the West ever since the First Crusade formed Eastern Yiddish through
language contact with Slavic languages and the addition of a Slavic compon-
ent, especially to the lexicon, phonology and lexical derivation. Eastern Yid-
dish 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 coun-
tries, 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 in-
fluenced both languages. Yiddish–German contact is an extremely complex
issue due to the languages’ close structural relatedness and the various pos-
sible modes of contact. Therefore, an analysis cannot be attempted in the
framework of this chapter.
2. Phonology
ample in Dlile ‘Delilah’ or pgam ‘defect’, has been extended under the influ-
ence 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
3.3. Gender
Northeastern Yiddish lost the neuter gender, as did two of its contact lan-
guages, 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
-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
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 pro-
voke’, 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 lan-
guages: Aspect/Aktionsart, modal particles and verbal derivational patterns.
4.1. Aspect/Aktionsart
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 big-
gest’).” (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).
In the areas of pronouns, particles and discourse markers, and adjectives and
adverbs, Yiddish shows signs of contact with Slavic languages.
5.1. Pronouns
(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.’
Interrogative pronouns: the construction ver ... ver is formed according to the
example of Russian:
The intensifying enclitic -že ‘then’ after the interrogative pronoun vos, e.g.
vos-že, is borrowed from Slavic:
pean contact languages may have been in use, e.g. in code-switching or semi-
habitual code-switching.
6. Constituent 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 Ger-
manic, where it appears very rarely and only in highly stylized forms, e.g.
German: Vater unser! ‘Our Father!’
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, how-
ever, 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 con-
stituent 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 stand-
ard language. Current developments in spoken German underline this point,
since subordinate clauses introduced by weil – ‘because’ and obwohl – ‘al-
though’ tend not to follow the “finite-verb-last” rule of Standard German any
more.
7. Syntax
7.1. Negation
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, dis-
course 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 Yid-
dish from the Slavic languages. Yiddish phonology was clearly influenced
by Slavic, although some features remained regional. In the area of morpho-
syntax, many structures may or may not have evolved through contact. Closer
analysis shows that they could just as well be the result of internal develop-
ments in the Germanic component, which might, however, have been trig-
gered 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 Ger-
manic elements in Yiddish.
Yiddish 257
Abbreviations
Notes
References
Bin-Nun, Jechiel
1973 Jiddisch und die deutschen Mundarten. Unter besonderer Berücksich-
tigung 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 So-
ciology of Language 138: 930.
1999b Contentious partners: Yiddish and Hebrew in Haredi Israel. Inter-
national 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 Deut-
sch 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 259
1. Background1
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 pres-
ently spoken to only in Hungarian or Slovak, and left to acquire some com-
petence in Rumungro in adolescent and adult peer groups, if at all. Thus, Ru-
mungro 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 villa-
gers. 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 lan-
guage of Slovakia, which they use outside of the village. Also, most have ac-
quired 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 Ru-
mungro 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 sociolin-
guistic 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 Rom-
ani 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 second-
ary current L2s have contributed only a few marginal loanwords, Hungarian
has exerted, and continues to exert, a strong lexical and grammatical influ-
ence 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
3. Typology
4. Nominal structures
Nouns are commonly borrowed into Rumungro. Pre-Greek and some (pre-
sumably early) Greek noun loans show full morphological integration and are
Hungarian Rumungro 265
5. Verbal structures
Verbs are commonly borrowed into Rumungro. Pre-Greek and early Greek
loan-verbs show full morphological integration and are structurally indistin-
guishable 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 compari-
son 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 valency-
neutral, 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 morphologic-
ally adapted borrowed verbs. Instead, the participles contain the Greek-
origin participle suffix -ime, e.g. rum-ime ‘spoiled’, vič-ime ‘shouted’, téle
dógoz-ime ‘worked away’ (lit. ‘downward worked’), sledov-ál-ime ‘ob-
served, 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 nom-
inal 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 adap-
tation suffix -in-, is a productive means to derive intransitive verbs from pre-
Hungarian nouns, e.g. paramis-i ‘fairy-tale’ (< Greek) → paramis-áz-in- ‘to
tell fairy-tales’. The second extracted derivational affix is the causative suf-
fix -tat-. Rumungro allows three different structural types of causatives of
Hungarian loan-verbs: lexical borrowing and adaptation of Hungarian causa-
tives, e.g. dógoz-tat-in- ‘to make [so.] work’ (< dolgoz-tat); internal deriv-
Hungarian Rumungro 269
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, lit-
tle; 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, in-
cluding 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 ‘Wednes-
day’). 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 ‘in-
deed’), phasal adverbs (még ‘still’ and má ‘already’), focus particles (iš ‘also,
too’, čak ‘only’, ippën ‘just’, pont ‘exactly’, ëgís ‘entirely’), affirmative an-
swer 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 dis-
cussed in Section 8.
In addition to function words, Rumungro has borrowed several function-
word 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 orienta-
tion 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 Hungari-
an-origin prefixes vala-, akár-, and minden- apply to interrogative pro-words,
e.g. káj ‘where’ → ni-kháj ‘nowhere’ (negative), vala-káj ‘somewhere’ (spe-
cific indefinite), akár-káj ‘anywhere whatsoever’ (free-choice), and minden-
káj ‘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 neutral-
ization 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 stat-
ive 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-quanti-
fication prefix sa- has developed as an alternative to the borrowed universal-
quantification 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 expres-
sion 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 Hun-
garian 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 e̋šéno ‘first’, and most frac-
tion numerals), and morphologically unintegrated loanwords, which alter-
nate 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 (< Hungar-
ian) ‘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 characteris-
tic of Rumungro, numerous aspects of Rumungro clause-level order appear
to have been borrowed from Hungarian, likewise a non-configurational lan-
guage. A prominent example is the tendency to position focussed constituents
immediately before the finite verb; this frequently results in clause-final pos-
ition of the copula in non-verbal predications (4; second line).
On the other hand, linear order at the noun phrase level is syntactically
determined in Rumungro: all types of adjectival modifiers, including descrip-
tive adjectives, adnominal possessors, demonstratives, and numerals, always
precede their head nouns. While the modifier–noun order prevails in all Rom-
ani dialects (cf. Matras 2002: 165167), it has been fully grammaticalized
in Rumungro due to contact with Hungarian. The alternative noun–modifi-
er 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 pre-
Hungarian 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
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 co-
ordinators, 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).
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 in-
finitive in Rumungro does not allow any complementizer.
9. Lexicon
Out of a much larger inventory of early loanwords into Romani (as attest-
ed in different Romani dialects), Rumungro of Selice retains ca. 20 loan-
words 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 lan-
guages is rather common, the by far most important current source of loan-
words is Hungarian.
276 Viktor Elšík
10. Conclusions
Both matter borrowing and pattern replication are well attested in Ru-
mungro. Lexical matter borrowing, i.e. borrowing of syntactically free sym-
bolic 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 art-
icle. 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 lex-
ical affix borrowing. Note that affix extraction assumes not only an adoption
of an affix within loanwords from a certain L2 and its (potential) paradig-
matic identification, but also its analogical, language-internal, extension to
other etymological compartments within the L1 lexicon. There are numer-
ous 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-inflection-
al 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; de-
nominal 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 loan-
words (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. ciga-
rett-áz-in- ‘to smoke cigarettes’ ← cigarett-a ‘cigarette’ (< cigarettá-z ←
cigaretta).
Although affix extraction is the source of a greater part of borrowed Ru-
mungro 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 pronom-
inal prefixes of Hungarian and South Slavic origin (see Section 6). These af-
fixes 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 pro-
word 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. Some-
times differences in the distribution of the source word appear to be respon-
sible for selective borrowing, as in the case of the Hungarian question clit-
ic (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 unat-
tested 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 gram-
matical 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 so-
called 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 as-
pects of linear constituent order; and more. Pure replication of morphologic-
al constructions is rare, being represented especially by occasional transla-
tions 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 re-
duction of others (gender in anaphoric pronouns and feminine derivation in
nouns denoting animals). Also, the retention and productivity of some inher-
ited 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 Ru-
mungro; unintegrated Hungarian numerals tend to retain their property of re-
quiring singular head nouns; the replicated category of preverbs is enhanced
by a few lexically borrowed members; direct affix borrowing results in “sem-
icalques” 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 morpho-
phonogical rules of the model language; and so on. However, matter bor-
rowing 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
Notes
3. The Rumungro associative plurals are similar in form to nominative plural agree-
ment (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 perfec-
tive 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 sev-
eral 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 suf-
fix, 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
1. Background
Manange, also known by its endonym ŋjeshaŋ, ŋjeshaŋte, or ŋjaŋmi ‘our lan-
guage/our people,’ is a Bodish language of the Bodic subphylum of Tibeto-
Burman. 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 val-
ley 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
Manange youths “spoke Nepali willingly and fluently” (1961: 205). Snell-
grove 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 sea-
sons. In winters, women and children especially, stay in low elevation vil-
lages 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 edu-
cation (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 lan-
guages, like Nepali, Hindi and increasingly, English.
Recently, Manang has become a tourist hot-spot because the popular “An-
napurna Long Circuit” bisects the district. As a result, a tourist-driven econ-
omy 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 lan-
guage 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
2. Phonology
The phonological structure of Manange in many ways typifies that of the Bod-
ish 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 consider-
able 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 lin-
guistically 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 287
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 Kath-
mandu), 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 symp-
tom 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 envir-
onments, while Nepali is the language of necessity in public domains. Never-
theless, 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 incorporat-
ing 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 (non-
contrastive) 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
3. Nominal structures
(3) Realis
1 2
mriŋ=tse naka 2puŋ 2kol-tsi.
woman=ERG chicken egg boil-perf
‘The woman boiled the egg.’
4. Verbal structures
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 suf-
fix -āu):
Manange 291
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 mor-
phological coding of aspect on the verb is dependent on negation: negative
marked (prefixed) verbs do not show aspect marking, with the resulting dif-
ference 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).
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 de-
velopment 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 bor-
rowed 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
ŋi-t ha 1
kola
two-class child
‘two children’
4
ʃi-t ha 3
pʌle
one-class leg
‘one leg’
6. Clause combining/syntax
(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.’
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, independ-
ent 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
Even though verbs constitute a smaller amount of Nepali loans, they are
somewhat more interesting structurally, as there are different strategies em-
ployed 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.
Other loans from Nepali are the first element in a verbal compound struc-
ture 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 Nepali Meaning
1
hai 1lʌpʌ hai aaunu ‘to yawn’
3
pu 1lʌ-pʌ puknu ‘to whistle’
poke 2prim-pʌ ‘untie’ ‘to tie’
The compounding process per-se is not borrowed from Nepali; in fact, ver-
bal compounding is quite typical of languages throughout South Asia (and
beyond).
One ongoing question is what the motivation is behind the different Man-
ange 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 ergative-
marked). The verb ‘do’ does in fact function as a periphrastic causative mark-
er 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 se-
mantically related, but non-transitive patnu or kelnu). In addition, some bor-
rowed 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 Man-
ange. Some of these changes can be considered pattern borrowings, whereby
a strategy is modeled on Nepali (e.g. the lack of a negation-finiteness depend-
ency 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 bor-
rowing at all, but rather structural loss or pattern (over-) generalization in
the urban community of speakers, perhaps due to infrequent and interrupt-
ed 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 systemat-
ically 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
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 Man-
ange 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 Evolution-
ary 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. Washing-
ton, DC: Georgetown University Press.
Bickel, Balthasar
1998 Rhythm and feet in Belhare morphology. Rutgers Optimality Archive
Working Paper No. 287.
Manange 299
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 con-
vergence. Studies in Language 31 (4): 829865.
Mazaudon, Martine
2003 Les principes de construction du nombre dans les langues Tibéto-
birmanes. 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 Sino-
sphere 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 The-
ory of Language. Leiden: Brill.
Grammatical borrowing in Indonesian
Uri Tadmor
1. General1
1.1. Classification
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 Indo-
nesian are at least bilingual or bidialectal.4 Children acquire the standard lan-
guage early on from its use on television and in school. In recent years the Ja-
karta 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.
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 Malay-
Indonesian, 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 Malay-
Indonesian.
Indonesian 303
2. Phonology5
The structural domain arguably most affected by contact has been the phon-
ology. 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. Ini-
tially, 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 fea-
tures 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 re-
flect 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 loan-
words, 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 excla-
mations (tidak [tidaʔ] ‘no!, not’,7 masa [masaʔ] ‘really?!’), which contrast
with the large number of words which end in vowels. In such words the glot-
tal 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 contrast-
ed 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, how-
ever, 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 semi-
vowels 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). Unas-
similated 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 borrow-
ing, modern Indonesian also has several consonant phonemes which were
borrowed outright. All loan phonemes consist of fricatives, which is not sur-
prising, 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 en-
tered 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 rep-
resented 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 pro-
nunciations persist, but many educated speakers now realize these words as
ziarah and zaman, respectively. In European loanwords, z was initially rep-
resented by s as in bensin ‘gasoline’ (< Dutch benzine), even if the original
orthography with z was maintained as in zébra [sebra] (< Dutch zebra). In-
creasingly, z is retained unassimilated in European loanwords too (now ori-
ginating 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 per-
sists as well. Some speakers realize Arabic x in loanwords with a spelling pro-
nunciation – 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] ‘biol-
ogy’, [texəl] ‘tile’.
Finally, /v/ has been claimed as a (loan) phoneme by some authorities, for
example the official dictionary of Indonesian (KBBI 2002). However, ortho-
graphic 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 re-
constructed 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. More-
over, 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 pho-
nemicization 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 con-
tact. 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 explan-
ation 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
‘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 ‘repudi-
ate’ (< 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). Non-
homorganic 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
speakers have been abandoning their ancestral home languages and switch-
ing to Malay-Indonesian as their home language instead. This long historic-
al process received an impetus in the middle of the twentieth century, when
Malay-Indonesian became the national language of newly independent Indo-
nesia 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 Malay-
Indonesian, 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 lan-
guage 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 topi-
calization 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 Indone-
sian may be among the most important impositions of the colonial – and neo-
colonial – 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
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 mor-
phemes aku (1sg), engkau (2sg), kamu (2pl), and ia (3sg). The genitive use
of these clitics persists in modern Indonesian: rumah-ku 'my house', untuk-
ku ‘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 lan-
guage 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 Sun-
danese 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 var-
ieties of Bazaar Malay, while maintaining its original sense of ‘same’, also
developed into a preposition with a basic comitative function (‘with’), replac-
ing 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-), mean-
ing ‘(together) with’.
The function of some inherited prepositions has been extended under the
influence of similar prepositions in other languages. The benefactive prep-
osition 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 prepos-
ition 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 prep-
osition 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
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 mem-
ber (which can refer to females as well, especially colloquially). However, in
some common kinship terms borrowed from Dutch, the male–female distinc-
tion 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).
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 be-
tween singular and plural forms. Collective nouns can be formed by redupli-
cation: 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 morpho-
logical 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 trans-
port 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 (refer-
ring 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 bor-
rowed without number distinction as alim]). Dutch plurals borrowed without
number distinction include (satu) karcis ‘ticket’ (< kaartjes ‘tickets’, the sin-
Indonesian 313
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 actu-
ally 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. Ex-
amples 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 San-
skrit 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 mon-
etary 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-Indone-
sian. 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
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.
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 insert-
ed 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 archipela-
go) 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):
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 In-
donesian. 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 West-
ern 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 mean-
ing ‘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, con-
sent’), 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
6.1. Pronouns
special jargon. From there these pronouns spread to the jargon of trendy
young women, known as Bahasa Gaul (‘language of socializing’). Some-
times 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 spontane-
ously.’
“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, lin-
guists assumed that borrowing into the set would disrupt the system. (There
may have been another, more prosaic reason for this wrong assumption: pro-
nouns 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 borrow-
ing 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 fea-
tures 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
6.3.1. Conjunctions
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
Borrowed greetings include Halo! ‘Hello!’ (< Dutch Halo!), Selamat! (‘Con-
gratulations!’, 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 in-
formally. The concept of specific greeting expressions for different times of
day and situations appears to have been borrowed. Other borrowed situa-
tion-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’ (< Eng-
lish 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
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
8.1. Copulas
Early Malay did not have copulas. A nominal subject and a nominal predi-
cate 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 oc-
curs, but is more common in Malaysian and in earlier Indonesian literature.
The choice of copula depends on idiomaticity, but in (5) all three are permis-
sible:
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 Eng-
lish.
9. Lexicon
Lexical and semantic borrowing (including calquing) are often treated togeth-
er 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, with-
out 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).
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’.
Some Indonesian words of Malay (i.e., inherited) origin have changed their
meaning based on the meanings of similar-sounding words in Javanese and
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 lexi-
calized in their language, but sometimes a word is borrowed even if the lan-
guage 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 ex-
ample, 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 lan-
guages 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, espe-
cially 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
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, espe-
cially 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 Uri Tadmor
in Jones 1978; most European loanwords appear in Grijns et al. 1983; and most
Sanskrit loanwords appear in de Casparis 1997.
6. 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’.
7. The sense of ‘not’ developed from the original sense ‘no!’.
8. Similar phenomena can be found in other Austronesian languages; see Blust
1979.
9. 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].
10. 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).
11. 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.
12. Similarly, schwa phonemicized in Malaysian under the influence of loanwords,
but the process there was more complex.
13. 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.
14. Embarrassingly the present author also overlooked this paper, only becoming
aware of it just before submitting the final version of this chapter for publica-
tion.
15. The sole exception is 1sg, which has different accusative and genitive forms.
16. 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.
17. It is not clear whether Proto Malayic actually had a 3pl form (see Section 6.1).
18. 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 sin-
gular meaning.
19. According to Scott Paauw (p.c.), a similar form occurs in Chinese Pidgin Eng-
lish.
20. 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.
21. 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 Lin-
guistics B95.) Canberra: The Australian National University.
Becker, Alton L., and Umar Wirasno
1980 On the nature of syntactic change in Bahasa Indonesia. In: Paz Buena-
ventura 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 Indo-
nesia 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 In-
donesian 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 Indone-
sian 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
2. Phonology
Compared to Indonesian/LM, Biak lacks /l/, /h/, /t/ and /ŋ/, but has an add-
itional 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 Indone-
sian/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 phon-
ology 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 sub-
ject by a prefix or infix on the verb, which is optionally preceded by a coref-
erential 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-.
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 sub-
ject, 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.
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 ‘vblz-
turn’ combines with the subject prefix ko ‘1pl.inc’.
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) V<y>e-guru.
<3sg>vblz-teacher
‘He is a teacher.’
Whereas borrowed main verbs clearly function as verbs in the recipient lan-
guage, 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:
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–aspec-
tual 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)
We now return to the borrowings bisa ‘can’ and harus ‘must’, which are aux-
iliary 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 ad-
verbs. 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 pos-
ition 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).
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 nega-
tive adverb bukan ‘not’. While the discussion on bukan ‘not’ will be post-
poned 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 estab-
lished loan, while dan ‘and’ should rather be considered a code-switch. For
both conjunctions, there is no observable functional difference between the
Biak 335
In (13) the speaker describes a scene that he is watching, and gives an alterna-
tive term for the referent of ai-mun ‘wood-piece’.
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 constitu-
ent 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 con-
junctions 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
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 hypothetic-
al situation of the addressee accompanying the speaker at a journey across
the island Biak.
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.
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 con-
tact 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):
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 follow-
ing sentence:
Biak 339
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 counter-
expectational or contrastive conjunction voi ‘but’. An example of the com-
bined 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.
More than in the case of kalau ‘if’, the use of bukan ‘not’ seems to be restrict-
ed 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 sev-
eral contact-related phenomena that can be qualified as instances of gram-
matical 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 how-
ever 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
Abbreviations
an animate pc paucal
compl complementizer pl plural
cons consecutive pos marker of possession
du dual pred predicative
ex exclusive sg singular
giv marker of givenness sim simultaneous
nonSP non-specific spc specific
inan inanimate u filler
inc inclusive vblz verbalizer
loc locative
Notes
1. To be more precise, the language has an auxiliary verb pok ‘can’, which how-
ever is used in negative contexts only, as in ya-pok ya-rir aw va ‘1sg-can 1sg-
let.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
f<y>arfyáre ‘3sg-know <3sg>tell’ → ‘he can tell stories’. While the latter sen-
tence was approved in eliciation, the non-elicited examples in the corpus are
restricted to negative sentences, like sifawi siwasya va ‘3pl.an-know 3pl.an-
read 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 bor-
rowed 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 con-
junction 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
1. Overview1
2. Phonology
periods (in particular, the period during which palatalization was spread-
ing throughout Chinese in the Middle Chinese era) when Chinese vocabu-
lary 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 struc-
ture 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 lin-
guistic 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 struc-
ture, with the exception of the glide /w/ in CGVC syllables, also a character-
istic 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.
In the realm of Vietnamese syntactic structure, the only aspect which Chi-
nese 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 numer-
ous 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 lan-
guages 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
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 clas-
sifier cái (most likely the archaic Chinese form 丐 (gài)),11 which has a virtual
article-like function in Vietnamese that, when used alone without quantifica-
tion, 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 com-
bination) 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 stat-
ive 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 fol-
lowing sample noun phrase, which shows the various post-nominal modifiers
in Vietnamese, all of which precede nouns in varieties of Chinese.
It is also important to recognize that Chinese has had very little influence
on the numeral system of Vietnamese, in contrast with the substantial influ-
ence 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 main-
tained 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 particu-
lar 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 ra-
ther than primary role in the Vietnamese numeric system.
Sino-Vietnamese 349
4. Verbs
The Sino-Vietnamese word qua (SV quá) 過 (guò) ‘to cross’ has a post-
verbal directional function parallel to that in Chinese, a preposition or adverb-
like 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.
For the most part, the grammatical adverbs and conjunctions that have been
borrowed from Chinese, as listed in Table 2, were borrowed grammatical-
ized 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 phono-
logically nativized loans with standard Sino-Vietnamese counterparts with-
out those developed usages. It must be admitted that more data (e.g. written
records and both syntactic and phonological comparative research of Viet-
namese 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 ori-
ginal 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/bal-
anced’), (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 mean-
ing 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
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 ra-
ther 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. Loan-
words 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 Vietnam-
ese 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 pres-
ence 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, mean-
ing 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 borrow-
ing. 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 borrow-
ings, 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 phonologic-
al 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 through-
out China and Southeast Asia, bisyllabic compounds consisting of two syl-
lables each with distinct meanings cannot be said to be the direct influence of
Sino-Vietnamese 355
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 bor-
rowing 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 grammat-
ical 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. Nor-
man’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 (not-
ably, the verb plus two-syllable directional terms); the use of reduplication
(A-not-A) in questions; the Chinese style of reduplication of the second syl-
lable 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, com-
mands, 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 borrow-
ing 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
Notes
1. Sources of data regarding borrowing into Vietnamese are mentioned when rele-
vant. Otherwise, the data comes from the author’s research, and any discrepan-
cies 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 exten-
sively 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 discus-
sion 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 seman-
tic 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 Mon-
Khmer 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 ex-
plicit 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 de-
termined by a systematic comparison of reconstructions going back two millen-
nia 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 bor-
rowing.
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 con-
tains 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 re-
gion have sentence particles with similar unmarked phonological material, es-
sentially 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 phono-
logical 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. Gra-
ham W. Thurgood.
2005 Sino-Vietnamese grammatical vocabulary and triggers for grammat-
icalization. The 6th Pan-Asiatic International Symposium on Linguis-
tics. 315332. Hà Nội: Nhà Xuất Bản Khoa Học Xã Hội (Social Sci-
ences Publishing House).
Barker, Milton E.
1969 The phonological adaptation of French loanwords in Vietnamese. Mon-
Khmer Studies 3: 138147.
Benedict, Paul K.
1947 An analysis of Annamese kinship terms. Southwestern Journal of An-
thropology 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 Annu-
al 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 Ori-
ental 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, Viet-
nam’s demotic script. Bulletin of the Institute of History and Philol-
ogy, 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 Sym-
posium 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 histor-
ical 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
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-cen-
tral 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 Non-
Pama-Nyungan language families.
All observations are based on my own (ongoing) fieldwork on the lan-
guage since 1993. Although Jaminjung has to be considered a severely en-
dangered 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 argua-
bly not the result of borrowing or “adoption” (in the sense of Johanson 2002)
but rather of substrate influence (“imposition”) on Kriol from languages simi-
lar 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
(d) The item, when it occurs, is completely prosodically integrated (not as-
sociated with hesitation phenomena).
2. Sociolinguistic background
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 increas-
ing numbers of Aboriginal people working on the stations, the (primarily)
English-speaking pastoralists, and the non-English-speaking colonists. Ac-
cording to some authors (Munro 2005: Ch. 2), Kriol emerged, with concomi-
tant 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 lan-
guage 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 exist-
ing 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 al-
lows, as Munro (2005) does, for substrate influence, eight languages from
four different Non-Paman-Nyungan language families (Marra, Alawa, Warn-
darrang, 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 Jam-
injung 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 pro-
nominal 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) ar-
rives at the conclusion that the differences between them are slight and most-
ly 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 grammat-
ical 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 Jam-
injung 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-gener-
ation communication. Moreover, a growing number of first-language speak-
ers 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 gen-
erally 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), typo-
logical 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 Jam-
injung are very similar, this is quite certainly not due to borrowing but to sub-
strate influence. For example, like most Aboriginal languages of the area, the
basilectal variety of Kriol lacks fricatives. Thus, lexemes derived from Eng-
lish 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 observa-
tions 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 inflec-
tional morphology. TAM categories are marked periphrastically, by auxiliar-
ies 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
5. Nominal structures
No influence from Kriol has been discerned in the word order within Jam-
injung noun phrases, which is relatively free. Nominal inflectional morph-
ology 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 prepos-
itions) 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.
What was said in Section 5 for nominal structures also holds for verbal struc-
tures. No grammatical morphemes associated with the verb in Kriol are bor-
rowed into Jaminjung. Structurally, too, the complex predicate system of
Jaminjung is clearly different from the auxiliary-verb system of Kriol. Se-
mantically, there are many similarities between the categories expressed by
Jaminjung and Kriol 371
Jaminjung speakers not only borrow Kriol nominals, but also verbs quite ex-
tensively (see Section 9). Verb integration exhibits some interesting features
due to the nature of the verbal system of Jaminjung, which (like that of sur-
rounding 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 argu-
ments) at the same time as functioning as integrator, and as an indicator of
verbness itself. For example, concepts of physical or psychological manipu-
lation involve the transitive verb -angu, glossed as ‘get/handle’ in (7) (com-
pare (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
The inflecting verb also expresses valency (since the paradigm of pronominal
affixes distinguishes transitive from intransitive verbs); for example, the un-
inflecting 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 any-
thing else since there are no verbal inflections in Kriol). A small number of
derivational suffixes (the transitivity marker -im and the progressive/continu-
ous 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.
The borrowing of “matter” from Kriol into Jaminjung affects lexical items
as well as some function words; bound grammatical morphemes are not bor-
rowed, and neither are pronouns, adpositions and auxiliaries. (It should be
pointed out that the borrowing of auxiliaries has been ruled out by defin-
ition 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 gram-
matical 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
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 tradition-
al strategy of juxtaposition continues to be used frequently though, both in
Jaminjung and in Kriol.
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 cor-
responding 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.
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 associ-
ated 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’.
Kriol ani (<Engl. only) is used as a restrictive particle alongside the Jam-
injung equivalents bugu ‘just, only’ and =biji ‘only’, but, as shown in Sec-
tion 7.1, also as a contrastive connector.
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 (Schultze-
Berndt 2002), also shown in (20).
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’ par-
ticle, yawayi, common to several languages of the area, arose from contact
influence.
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 non-
indicative moods, and for constituent negation, are never used in Jaminjung
utterances in my data, with one crucial exception: there are a number of ex-
amples, including (22), where the particle nomo is used for metalinguistic
negation instead of the Jaminjung general negative particle gurrany.
Jaminjung and Ngaliwurru also have the special independent negative forms
damarlung (J) and gara (Ng). They are used on their own as a negative an-
swer 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.
Another particle, olrait (< Engl. all right), marks a transition between two
clauses which have a stronger thematic coherence, e.g. where the continua-
tion is a natural consequence of the beginning, as in (25).
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 “mat-
ter” to Jaminjung, partly replacing the equivalents. The question about the
motivation for such borrowings will be adressed in the concluding section
(Section 10).
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 transla-
tion equivalents, produced on the same occasion by the same speaker.
for which no Jaminjung equivalent exists (Section 7). Another potential area
of influence, mentioned in Section 4, is optional ergativity. The overall im-
pression 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 lex-
ical 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 expres-
sions. 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
Spatial nominals including deictics, on the other hand, are used but mostly
in those places identified as switches rather than borrowings, with a few pos-
sible exceptions such as (21) and (30).
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 Jamin-
jung equivalents, but their low frequency points to code-switching rather than
borrowing. One example is (31).
9.2. Verbs
The borrowing of Kriol verbs into Jaminjung as uninflecting verbs was al-
ready 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.
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 non-
final intonation contour, at a prosodic break.
References
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: Stud-
ies from Australia and the South Pacific, 245270. Saint-Laurent,
Quebec: Collection Champs linguistiques.
2005 Substrate language influence in Kriol: The application of transfer con-
straints 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, Vol-
ume 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 Ferguso-
nian 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.), Prag-
matic 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 categorisa-
tion 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: Syn-
chronic 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 Universalien-
forschung 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. Ber-
lin: Mouton de Gruyter.
Weinreich, Uriel
1968 Languages in Contact: Findings and Problems. The Hague: Mouton.
Grammatical borrowing in Rapanui
Steven Roger Fischer
1. Background
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 Polyne-
sian 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 Ra-
panui /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 pro-
nounce 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 community-
wide 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
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
will now sometimes be heard as to’o pōki, tu’u pōki or even tū pōki (corres-
ponding 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 demon-
strating a shift toward prepositive positioning in Rapanui, and this not only
in direct borrowings or calques. For example, Rapanui motore vaka repro-
duces 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 demonstra-
tive 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
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:
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 becom-
ing “native” Rapanui.
It is possible that Spanish ir ‘to go’ has prompted the increased use of Ra-
panui 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
When Spanish verbs are used in Rapanui, it usually entails direct integra-
tion 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 Ra-
panui 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 Island-
ers 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:
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 Span-
ish 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:
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:
In the wake of recent massive Spanish contact, the verbal semantics of flu-
ent Rapanui speakers has sometimes been re-evaluated. For example, Span-
ish 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.
In formal Rapanui speech, there are hardly any Spanish borrowings; this al-
lows us to construe that, in Rapanui, the use of Spanish entails almost ex-
clusively code-switching, not borrowing. With other parts of speech, numer-
als 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
7. Constituent order
Ante can also function, without progressives, before perfective particles and
prepositions:
8. Syntax
Yet Spanish interrogatives almost entirely occur in Rapanui only during code-
switching: 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:
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 intermin-
gle both languages, to varying degrees, throughout the day, depending on to
whom they are talking and the circumstances. Nonetheless, there is also Ra-
panui 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 lexic-
al 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 supplemen-
tation. 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 in-
volving 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 Polyne-
sian 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 Poly-
nesian-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 syn-
cretism 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 preva-
lent 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 lan-
guages, 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 gram-
matical borrowing from Spanish figures in Rapanui as well. Owing to the pau-
city 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. How-
ever, 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 re-
turn 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 for-
mally integrated into indigenous language use. The hitherto private domain
400 Steven Roger Fischer
Abbreviations
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 Wil-
liam T. Mulloy, 165168. (Oxbow Monograph 32.) Oxford: Oxbow
Books.
Rapanui 401
1. Background
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 Pue-
bla (NP) and North Guerrero Nahuatl (NG).
2. Nominal structures
The same applies to the possessor prefix in (3), which is attached to the rela-
tional noun -na:wak ‘near’, ‘at’:
Nahuatl 405
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:
In other dialects, the possessor prefix is maintained, but the whole word in-
cluding 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 move-
ment encoded by the verb -wi:¢ ‘come’:
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:
3. Verbal structures
Apart from a few exceptions, Spanish verbal structures have not been bor-
rowed, neither have Nahuatl verbal structures undergone changes. Spanish
verbs have been borrowed from an early stage of the contact between Na-
huatl speakers and Spaniards. Loan-verbs are derived from Spanish infini-
tives, which are treated like Nahuatl noun roots by means of the suffix -oa
– as shown in (8) with the Spanish verb cantar ‘sing’:
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):
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
However, its an open question, to which extent the tense category as such has
undergone contact-induced changes.
408 Una Canger and Anne Jensen
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 nu-
merals 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 intro-
ducing adverbial clauses are all Spanish. However, the syntax of adverbial
clauses remain Nahuatl. Moreover, in NG Nahuatl the three Spanish coordi-
nating connectors have been borrowed, i.e. y ‘addition’ (14), o ‘disjunction’
(15), and pero ‘contrast’ (16).
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
Apart from that, (i:)wan also coordinates phrases, i.e. like the second wan
in (18):
The first wan shows the third level of coordination – it combines two suc-
ceeding, 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 ex-
tent 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 re-
sembles the function of Spanish y, and this fact may be considered a contact-
induced 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
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
In modern Nahuatl the noun phrase always occurs immediately after noči:
The change from the originally discontinuous phrase shown in (21) to a con-
tinuous phrase like the one in (22) may be contact-induced, since a quanti-
fier and a noun (phrase) always form a continuous phrase in Spanish. How-
ever, 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
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 intro-
duced by in:
(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:
When the NP-head is +human, the pronoun a:kin ‘who’ introduces the rela-
tive clause:
Finally, the speakers use Spanish que for introducing relative clauses:
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:
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) resem-
ble Spanish constructions with the verb querer as main verb.
7. Lexicon
Spanish loanwords are adopted in many domains of the Nahuatl lexicon in-
cluding kinship terms. Presumably, the amount of loanwords correlates with
a speaker’s age, in the sense that children and younger adults use more loan-
words than older speakers. However, in the domain of grammatical vocabu-
lary – demonstratives, (emphatic) pronouns and place deictics – no borrow-
ing has taken place.
8. Conclusion
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 af-
fixes 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 poly-
synthetic as some polysynthetic structures have been substituted by perifra-
stic constructions corresponding to Spanish ones. However, a comparison
with the development of other polysynthetic languages exposed to a long con-
tact with Spanish may reveal whether the development in Nahuatl is unique
or a common feature.
Abbreviations
Notes
1. Nahuatl has had an immense influence on Mexican Spanish with regard to par-
ticular 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 sixteenth-
century Nahuatl. In (21) in Section 5, the possessor prefix i:- third-person singu-
lar 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 ex-
pressing NP-rel.
References
1. Background1
Yaqui, including Mayo, its major dialect, along with two other distinct lan-
guages of this family, Tarahumara and Guarijio, constitute the Taracahitan
branch of the Uto-Aztecan linguistic family.2 Yaqui is spoken mainly in Mex-
ico 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 Mex-
ico 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 vernacu-
lar 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 demon-
strated 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, particu-
larly 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 be-
longing to the same ethnic group, but in other everyday activities, e.g. polit-
ical, 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 Bui-
timea, 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 phono-
logical 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 combin-
ation with the glottal stop. In the adaptation of loanwords, vowel lengthening
replaces a stressed vowel from Spanish:
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 frica-
tive /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
3. Nominal structures
There are two phenomena related to nominal structures that arise in elem-
ents 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:
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:
The following examples show that Spanish mass nouns are also adapted
into Yaqui as plural nouns:
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’).
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 423
The adaptation of Spanish verbs via the Nahuatl suffix -oa, is relatively
new in Yaqui. Karttunen and Lockhart (1976: 32) mention that the -oa strat-
egy (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 pro-
vides 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 deriv-
ation through the suffix -oa, these verbs might be the case of a phonological
coincidence:
In Spanish, the comparative conjunction komo ‘than’ (< Sp. como) has
more than one function, one to express similarity, another to express com-
parison. Yaqui uses the suffix -su for the first function as in (48a), and the
postposition benasi for the second one, as in (48b):
The vocative words borrowed from Spanish and illustrated within this sec-
tion 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
tive clauses involving the Yaqui postposition benasi. The example in (54) il-
lustrates benasi before the adjective teebe ‘tall’, which is the usual position
for komo in Spanish:11
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 loan-
words 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 in-
stance of code-switching rather than of borrowing, is the usage of multimor-
phemic elements. That is, expressions (idioms) which include more than one
word. Instances of code-switching are frequently used to give communica-
tive 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:
Abbreviations
acc accusative dat dative
adjvz adjectivizer dem demonstrative
agt agentive des desiderative
appl applicative det determiner
caus causative dim diminutive
com comitative dir directional
comp comparative fut future
cond conditional imp imperative
Yaqui 431
Notes
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 An-
thropology 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 Typ-
ology. 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 Ca-
macho, 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 Au-
gust.
Yaqui 433
1. Background
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 respect-
ively. Since the Otomis had to surrender to the Nahuas in the fifteenth cen-
tury, 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 Ot-
omis “toscos e inhábiles” (coarse and unskillful). The word Otomi is prob-
ably 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 prob-
ably a derivation of the adverb hño ‘well’. The morpheme h- marks the imper-
sonal 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 Chich-
imecs, 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 con-
version. 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 be-
long 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 subsist-
ence, 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 commu-
nity 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 con-
sequence, 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; Hek-
king 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 lan-
guage 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 Sec-
tion 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
l. axo (< ajo), paxa (< paja), Xuwa (< Juan) are examples of old loanwords,
the palatals of which are pronounced as fricative palatals in modern Span-
ish
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 pos-
sible 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
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’
(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).
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 nom-
inal modification.
As we will show in Section 4, in the corpora we have collected in San-
tiago 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. How-
ever, 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 adjec-
tives 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 ad-
jective-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 borrow-
ings. Just over half of these belong to a lexical category (noun, verb, adjec-
tive 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 ma-
terial, 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 illus-
trated in example (8).
444 Ewald Hekking and Dik Bakker
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
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); es-
kohe (< 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).
Spanish verbs with vowel change and diphthongization generally keep their
diphthongs ‘ie’ and ‘ue’ (cf. kwenta in 11a), although not always (konta
in 11b).
This phenomenon has not only been found in Querétaro, but also in other dia-
lects (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 ad-
verbs, 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.
6.1. Prepositions
a. The Spanish preposition pa (< para) is often used together with or instead
of the verbal suffixes -pi or -wi to mark the beneficiary.
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 pro-
cess is under way which might eventually transform this particle, and several
others in functional prepositions.
c. In classical Otomi the particle ngu is used to mark the object of equa-
tion. Nowadays the Spanish preposition komo (< como) is often used instead.
Very common is the use of the fused double marking komo-ngu, as in ex-
ample (18).
Otomi 449
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’.
Also the Spanish prepositional construction embesde (< en vez de) is often
used in such contexts.
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 con-
cerned. In (25)–(27) we give three examples.
As shown in example (28) below, the Spanish preposition ko (< con) ‘with’ is
also used to mark the ‘made of’ function.
6.2. Coordinators
The third most important category of Spanish borrowings are discourse mark-
ers. 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
6.4. Subordinators
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 inter-
rogative pronoun, as in (34d).4
Apart from the relative pronouns, there were quite a lot of Spanish indefin-
ite 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’.
7. Constituent order
(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 com-
munal fields.’
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 pos-
sessed–possessor, so no changes may be expected here. The same applies to
the relative clause order: both languages are N–Rel.
8. Syntax
(46) Ar ’yot’u-ngo ta ar u.
def.sg dry-meat cop def.sg salt
‘The dry meat is salty.’
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
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 per-
cent 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 bor-
rowed 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.) rele-
vant 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 de-
fined 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 bor-
rowings: for over half the entries a Spanish form is used. However, it is strik-
ing 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 cor-
responding 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
Spatial relations 74 9% 4%
Moral/aesthetic 48 8% 6%
Law 26 8% 4%
Body parts 156 6% 2%
Physical world 68 3% 0%
Sense perception 48 2% 0%
Total 1413 16.3% 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 lin-
guistic levels. In the area of phonology, as a result of the adoption of unas-
similated 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, subordin-
ators, 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 charac-
terization.
In relation to constituent order we have observed that the basic Otomi or-
ders 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 situ-
ated somewhere between points 2 and 3 on the borrowing scale of Thomason
(2001).
Otomi 461
Abbreviations
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 ra-
ther 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 commu-
nities.
References
Andrews, Henrietta
1993 The function of verb prefixes in Southwestern Otomi. The Summer In-
stitute of Linguistics and the University of Texas at Arlington Publica-
tions in Linguistics 115. Dallas: Summer Institute of Linguistics and
the University of Texas at Arlington.
462 Ewald Hekking and Dik Bakker
1. Background
in the Lake Patzcuaro area, Purepecha has become SVO (Capistran 2002).
This order has been attested since the sixteenth century, and has become pro-
gressively more widespread since that time (Villavicencio 2006). Considering
that Nahuatl – and Otomi – present a verb-initial structure, this change prob-
ably has its roots in areal contact prior to the sixteenth century, with the subse-
quent influence of Spanish, an SVO language, continuing the process.
In the present chapter, I will concentrate on the influence of Spanish con-
tact on Purepecha, specifically, on the grammatical structure. related to this
contact are found in the areas of phonology, morphological typology, nom-
inal and verbal structures, other parts of speech, constituent order, and syn-
tax. 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), Pure-
pecha 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, ac-
cordingly, presents the same vowel system as Spanish (/i/, /u/, /e/, /o/, /a/).
3. Morphological typology
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).
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 (PAT-
influence), 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
In the same way, the constructions with xa ‘be there’ have adopted the val-
ues of the Spanish ‘estar’ structures: passive constructions (3), and attributive
ones (7).
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 con-
texts, the Purepecha marker may appear in a double construction (27c, 28c,
29c, 30c).
The Spanish phrasal adverb ya is used to mark temporal values with two dif-
ferent 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.
7. Constituent order
8. Syntax
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 ex-
pressions, 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
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.
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 ana-
lytic–periphrastic constructions appear, without modifying its elaborate mor-
phological system, but revealing a structural rapprochement to the Spanish pas-
sive and equational constructions: two distinct structures (a morphological one
and a periphrastic one) may simultaneously perform the same function. Pure-
pecha is exhibiting language-internal grammaticalization processes to replicate
Spanish models.
There are PAT-influences that are not connected to any direct MAT-borrow-
ing. Other PAT-influences are the comparative constructions, linked to MAT-
borrowing in some varieties, and showing only pattern reduplication in others.
Abbreviations
References
Capistran, Alejandra
2002 Variaciones de orden de constituyentes en p’orhépecha. Topicaliza-
ción y focalización. In: Paulette Levy (ed.), Del Cora al Maya Yucate-
co: Estudios lingüísticos sobre algunas lenguas indígenas mexicanas,
349402. México: UNAM..
Chamoreau, Claudine
2002a Le système phonologique du purepecha: Une étude en synchronie dy-
namique. Travaux du SELF IX: 133161.
2002b Dinámica de algunos casos en purepecha. In: Zarina Estrada Fernán-
dez and Rosa Maria Ortiz Ciscomani (eds.), VI Encuentro Internac-
ional de Lingüística en el Noreste, 271290. Hermosillo: Unison.
2005 Dialectología y dinámica: reflexión a partir del purépecha. In: Clau-
dine Chamoreau (Coord.), Trace 47: Dinámica lingüística, 6181.
México: CEMCA.
2006 En busca de un verbo “ser” en purépecha: Cadena de gramaticaliza-
ción y gramaticalización en cadena. In: Rosa Maria Ortiz Ciscoma-
ni (ed.), VIII Encuentro Internacional de Lingüística en el Noreste,
6584. Hermosillo: Unison.
2007 Looking for a new participant: Passive in Purepecha. In: Zarina Es-
trada 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
2. Phonology
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 dia-
lects. 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 dia-
lects (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 re-
ciprocal -naku-, realized as [-naju-] in Imbabura, but [naku] in the rest of the
provinces. Voiceless [t] and voiced [d] are distinct phonemes in all Ecuado-
rian 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 above-
mentioned 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 cen-
tral and Southern dialects. These facts tell us that we are before a phenom-
enon of substratum influence in IQ phonology. Recent research has shown
that a distinct cultural and linguistic group lived in the present territory of Im-
babura. 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 de-
scribed 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 (Gómez-
Rendón 2005). What makes IQ – and Ecuadorian dialects in general – differ-
ent 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:
(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’
The diminutive ending and the agentive marker occur both with Spanish lex-
emes (left column) and native lexemes (right column).
Apart from these grammatical changes, IQ shows a large number of Span-
ish 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 non-
verbs conflates nominal, verbal and adjectival functions. Spanish borrow-
ings 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. How-
ever, this new development does not imply any loss of topic prominence (see
Section 7.1).
4. Nominal structures
Contact-induced phenomena in the use and the semantics of case markers in-
clude: (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.
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 instrumen-
tality of one with respect to the other. For Kaarhus (1989) the comitative–
instrumental distinction entails a unique understanding of space-time rela-
tions 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 mean-
ings. 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.
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:
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
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 topical-
izer 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). Interest-
ingly, this use is found beyond the context of broadcasting. Consider the fol-
lowing examples:
Imbabura Quichua 491
In (26) the Spanish diminutive ending occurs on IQ wawa ‘child’. The dimin-
utive 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 in-
dicates that grammatical borrowings are used productively even though they
do not necessarily follow IQ rules. In order to form kinship terms for gen-
erations 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 mor-
phological strategies of both languages.
The Spanish agentive nominalizer dur often occurs unanalyzed on bor-
rowed 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:
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
Spanish loan-verbs, modals and particles into IQ. At the same time, the oc-
currence of pattern borrowing in valency-changing devices and, most impor-
tantly, the gradual replacement of the nominalization strategies with finite-
clause subordination are modifying the typological outline of IQ.
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 wide-
spread 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 sug-
gests 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
In the foregoing examples gulpi is used to stress the inclusivity of the first-
person 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 clusiv-
ity 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.
Examples (38), (39) and (40) are modal verb constructions in which Spanish-
derived 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-centu-
ry. 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 in-
formation including reportativity, quotativity and inference.9 Although Span-
ish 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:
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.
In the above example the verbal marker of habitual past is added to the Span-
ish 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 ad-
jectives used as nouns and adverbs. A detailed discussion of Spanish lexical
borrowings in IQ is presented elsewhere (Gómez-Rendón 2006a).
Imbabura Quichua 499
(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 in-
novative 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 recip-
rocal-plural relation is demonstrated by the trivial fact that every reciprocal
form implies several individuals and reciprocal morphemes in fusional lan-
guages 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 re-
ciprocal 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
Consider the argument structure of binda ‘bandage’ (Sp. vendar) in (53). Ori-
ginally, 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 ar-
guments 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 pro-
noun 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 im-
personal construction in (54):
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 em-
bedded 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 ad-
verbials 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.
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 ef-
fects 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.
While many IQ grammars and dictionaries boast a full set of native numer-
als 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 ex-
ample illustrates cardinal and ordinal numbers from Spanish:
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 quantifi-
ers are tuditu ‘all’ and alkunu ‘some’. The following examples illustrate their
use:
6.2. Pronouns
In the last section I showed that Spanish-derived quantifiers are used as pro-
nouns in IQ. In this section I show that the influence of Spanish on the pro-
nominal 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 ex-
change extracted from an interview:
Compare this strategy with the use of pronoun kikin in (63), where it co-
occurs 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
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
The disjunctive often co-occurs with Spanish time adverbs such as intonses
‘then’, siympre ‘always’, nunca ‘never’ or antes ‘before’. It remains to inves-
tigate whether we are dealing here with borrowings or code switches. Some
Spanish connectors have been incorporated to IQ discourse without assimila-
tion. 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 ex-
amples show Spanish adverbs used as time deictics in IQ (see (69)–(71)).
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:
phrase pero más claro ‘but clearer’. In example (75) the standard of compari-
son 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 construc-
tion 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).
The affluence of Spanish loanwords in IQ goes hand in hand with less vis-
ible 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 syn-
tactic 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 construc-
tions (instead of nominalized embedded clauses) imply the borrowing of
Spanish conjunctions. Several issues related to word order have been ad-
dressed 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 declara-
tive 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 in-
cluded here on account of their limited frequency.
512 Jorge Gómez-Rendón
Example (77) shows SVO word order but marks the direct object with accu-
sative 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. Ac-
cording to Fauchois “[the new SVO word order] is almost systematic if the
object is a nonce borrowing” (1988: 117; my translation). A further syntax-
related change induced by contact is the drop of the topicalizer and the even-
tual 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:
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 non-
verbal 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 cal-
quing. 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.
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 sentence-
final 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
Yes-no questions in Spanish are formed by moving the main verb to sentence-
initial position and/or giving an interrogative intonation to the questioned
element. These strategies have been adopted by IQ speakers. In more con-
servative 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).
As shown in (90), the inverted word order occurs also in non-verbal predica-
tive 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.
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 Span-
ish subordinator porque never co-occurs with its IQ counterpart (the suffix
-manta). On the contrary, conditional como16 does co-occur with native suf-
fixes -kpi or -shpa. Example (92) illustrates this case of doubling.
Imbabura Quichua 517
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 ra-
ther uncommon.
Spanish subordinating conjunctions are frequent in contemporary IQ and
their use is widespread across generations and levels of bilingualism. How-
ever, their co-occurrence with native suffixes is more frequent in conserva-
tive 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 asso-
ciated with it is a gradual process, the stages of which can be found in differ-
ent idiolects within the same speech community.
8. The lexicon
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 mark-
ers, interjections, numerals and frozen borrowings. In general, frozen borrow-
ings 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 (Gómez-
Rendón 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 cen-
tury 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 Im-
babura has strengthened the influence of the dominant language on the lexi-
con 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 lan-
guage after four centuries of contact because it succeeded in making a com-
promise between the communicative needs imposed by the official language
and the speakers’ cultural need to preserve their linguistic identity.
Notes
3. This is typical of Ecuadorian dialects. Peruvian and Bolivian dialects show di-
vergent 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 Span-
ish, 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 mor-
phemes being all suffixes. An alternative analysis is that la- is a reduced (gram-
maticalized) 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 re-
organization 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 spo-
ken 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 inver-
sion 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 accompa-
nied by IQ markers.
520 Jorge Gómez-Rendón
References
2005 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 castel-
lano: 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. Lin-
guistics 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: Cam-
bridge University Press.
2000b Semantic transparency in Lowland Ecuadorian Quechua morphosyn-
tax. Linguistics 39 (5): 873988.
Schachter, Paul
1985 Parts of speech systems. In: Timothy Shopen (ed.), Language Typ-
ology and Syntactic Description. Volume 1: 361. Cambridge: Cam-
bridge University Press.
Grammatical borrowing in Paraguayan Guaraní
Jorge Gómez-Rendón
1. Background
2. Phonology
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 non-
canonical onsets and codas: clusters formed by a plosive plus a flap (e.g. /tr/,
/pr/); sibilant /s/ occupying coda positions in non-assimilated loans, particu-
larly in frozen borrowings with the Spanish plural ending (e.g. kosa-s-kuéra
‘things’).
3. Typology
The foregoing example contains three Spanish borrowings: the adjective per-
hudisial ‘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 proto-
typical 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 transla-
tion is only implicit in PG. In turn, (1c) is linked to (1b) by the Spanish con-
junctive 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
4. Nominal structures
Nominal structures are the category most influenced by Spanish. There are
five structures that show contact-induced changes: possession, number, de-
terminers, case marking, and local relations. I address each of them in this
section.
4.1. Possession
The other change has to do with the distinction between alienable and inal-
ienable possession. Although PG has no morphological marker to distinguish
one kind of possession from the other, the distinction is expressed by syntac-
tic and lexical means, for instance, through noun incorporation (6a) or the use
of specific lexemes for body parts (7a). In contact varieties noun incorpora-
tion is being increasingly replaced by phrasal constructions (6b), and body-
part lexemes do not show consonant alternation, as shown in (7b).
4.2. Number
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
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. Further-
more, the plural marker may occur in (9) without added meaning, but not in
(10) where it is ungrammatical.
Guaraní boasts a complex system of deictics used to mark not only definite-
ness but also spatial relations and other referential functions. Spanish articles
have been added to this system as determiners and pro-forms.
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 be-
tween the determiner and the head as shown in example (11):
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.
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 ana-
phoric (18)–(19).
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 con-
tain 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.
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 iñer-
manokuéra ‘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 attrib-
uted to the contact language. A similar elision of patient markers has been
observed in other languages in contact with Spanish such as Ecuadorian Que-
chua (Gómez-Rendón, in this vol.)
There are cases in which the Spanish preposition a – marking human dir-
ect objects – co-occurs with the corresponding Guaraní marker, as illustrated
in (24). This construction does not occur with native noun phrases, how-
ever.
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 op-
posed 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:
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.
Influence from Spanish in the marking of aspect is clear from the co-occur-
rence of Spanish adverb ya ‘already’ alongside with the native perfective
marker mã as in (31).
In similar terms, the predicative construction in (35) differs from its non-
predicative counterpart in (36) because the latter requires only the noun h-oga
‘3.house’ and the noun inflected for third person.
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:
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.
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 causa-
tive marker mbo with Spanish-derived verbal lexemes.
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 am-
biguous constructions. On the basis of this model, the prefix ñe in PG may re-
ceive reflexive and passive interpretations. Consider the following examples:
The Guaraní marker ñe admits passive and reflexive readings alike. Some-
times, however, only one reading is possible and contextual clues are re-
quired 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:
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
Spanish verbs are borrowed in PG by deleting the infinitive ending /-r/, with-
out 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 non-
verbal 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).
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 assimi-
lated 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 ex-
plained 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
6.3. Connectors
The most frequently used Spanish connectors are porke ‘because’ and pero
‘but’. Coordinator y is used mainly to link code switches. The difference be-
tween Spanish subordinators and Guaraní subordinators lies on the fact that
the latter are postpositional. The only exception is subordinator ke, which oc-
curs in different constructions analyzed in the next section.
6.3.1. Subordinator ke
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 prepos-
itional phrase – the compound connective durante qu e9 links two independ-
ent verb phrases. In (58) the speaker is using the preposition durante as equiv-
alent to the conjunction mientras on the basis of their similar semantics. The
Paraguayan Guaraní 545
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.
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
8. Syntax
9. Lexicon
bers, with or without assimilation (cf. Section 2).10 However, there is a differ-
ence 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. Ac-
cording 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%), nu-
merals (1.7%), discourse markers (0.8%), adpositions (0.5%) and non-per-
sonal pronouns (0.2%). Spanish grammatical borrowings either replace, or
co-exist with, native morphosyntactic mechanisms. Spanish lexical borrow-
ings 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 syn-
tactic positions. Thus, Spanish nouns are used also as verbs and adjectives
whereas adjectives are used as verbs, nouns and adverbs. The following ex-
amples show the flexible use of Spanish loanwords:
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 simi-
lar 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 lex-
emes are grouped in two classes: verbs and non-verbs. The latter includes
Spanish nouns, adjectives and adverbs.11
10. Conclusion
Notes
1. In what follows I use the term pre-contact Guaraní to refer to the Guaraní lan-
guage as spoken before the Spanish conquest. For a thorough description of pre-
contact 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 Morini-
go’s Hispanismos en el Guaraní (1931). In the last two decades the study of the
linguistic outcomes of Spanish-Guaraní contact has received increasing atten-
tion. 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 dif-
ferent perspectives.
3. For full data on the census, visit the website www.dgeec.gov.py.
Paraguayan Guaraní 549
References
1. Background
The Hup language1 (also known as Hupda or Jupda; see Epps 2005a) is spo-
ken by approximately 1500 people in the Vaupés region, located on the bor-
der 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 (genet-
ic) 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 dem-
onstrated.
Hup is currently fully viable, learned as a first language by all Hup chil-
dren. However, by the time they are adults, virtually all Hupd’əh (‘person-
pl’, ‘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 near-
constant 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 diffu-
sion have been documented in detail for the Arawak language Tariana, which
has undergone significant grammatical influence from Tukanoan (Aikhen-
vald 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
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 ob-
jects); i.e. ‘matter’ instead of ‘pattern’. Many of these may in fact have en-
tered the language via Tukano, which in most cases uses the same borrowed
lexical forms.
2. Phonology
(1) Hup
núh (high/falling tone) ‘head’
nǔh (rising tone) ‘tapioca’
(2) Hup
d’ǔç höd tətəd-d’óʔ-óy=mah.
timbó 3pl beat.timbó-take-dynm=rep
‘They beat timbó, it’s said.’
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 influ-
ence – 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’ǎc-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 dis-
cussed 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’
(7) Hup
pedú nˇh cug’æ̌t
Pedro poss book
‘Pedro’s book’
(9) Hup
cw=d’əh (biting.ant.sp-pl) ‘biting ants (sp.)’
cw=ʔãw (biting.ant.sp-sing) ‘a (single) biting ant (sp.)’
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 comple-
tive, 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
(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.’
(12) Hup
j’ǔg-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).’
(14) Hup
póh, děh=teg g’et-ʔeʔ-ní-h.
high water=tree stand-perf-infr2-decl
‘Really high, the water-tree stood.’
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 un-
like those in other Nadahup languages and all appear to be recently grammat-
icalized; 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 tiyı̌ʔ-ǎn hup-mæ̀h-æ̀y.
jaguar man-obj pass-kill-dynm
‘The jaguar was killed by the man.’
Also, the reflexive marker (a verbal prefix in both Hup and Tukano) can ap-
pear as a nominal suffix/enclitic with an intensifying function in both lan-
guages (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’).
Hup’s numeral system has been profoundly influenced by contact with Tu-
kano. 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 Por-
tuguese 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 adver-
bial particle té ‘until, up to’ (from Portuguese até; example 18), and the nega-
tive emphasis particle næ̀ may also derive ultimately from Portuguese nem
‘neither/nor’. Interestingly, Hup has borrowed a disjunction ʔó ‘or’ (from Por-
tuguese ou; example 19) without borrowing a conjunction (‘and’), a cross-
linguistically 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 Tu-
kano. 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 construc-
tion (see §5 above).
(20) Hup
náw töh ni-ʔě-h.
good 3sg be-perf-decl
‘He used to be good.’
Negation in Hup has several features in common with Tukanoan, also like-
ly 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 re-
lating 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 con-
tact.
Hup 561
(22) Hup
ʔam wæd-túk-uw-aˇn d’oʔ-næ̀n-æ̀h.
2sg eat-want-flr-obj bring/take-come-decl
‘(We) brought what you wanted to eat.’
8. Lexicon
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 grammat-
ical 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 phon-
ology, Tukanoan has had a particularly strong hand in shaping Hup’s pros-
odic features of nasalization and tone (word-accent) – likely due to the rela-
tively high discourse salience of such features. Contact also appears to have
restructured Hup’s typological profile in significant ways, affecting its gram-
matical alignment, constituent order, etc. Tukanoan has influenced the ex-
pression of number, case, noun class and gender in the Hup noun phrase; in
the verb phrase, tense, aspect, mode, and evidential categories have been af-
fected, and likewise the mechanisms for indicating changes in voice and va-
lency. 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 es-
caped 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-bor-
rowing have taken place, including a high proportion of verbs.
In conclusion, the pattern of contact between Hup and Tukanoan has fa-
vored the borrowing of PAT over MAT, a cross-linguistically unusual situ-
Hup 563
Abbreviations
Notes
1. Information on Hup (aka Hupda, Jupde) was obtained via the author’s field-
work 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 grate-
fully acknowledged. Many thanks go to my Hupd’ǝh hosts and language teach-
ers, as well as to the Museu Parense Emílio Goeldi and the Instituto Socio-
ambiental 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 high-
ly insulting ethnic slur in the Vaupés, directed toward the speakers of these
564 Patience Epps
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, Uni-
versity 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: Linguis-
tic and Anthropological Studies with Special Emphasis on the Lan-
guages and Cultures of the Andean-Amazonian Border Area. (Indig-
enous 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 Typ-
ology 4.) Oxford: Oxford University Press.
Forthc. Hup. In: Martin Haspelmath and Uri Tadmor (eds.), Loanword Typ-
ology Project, Max Planck Institute for Evolutionary Anthropology,
Leipzig.
Jackson, Jean
1983 The Fish People: Linguistic Exogamy and Tukanoan Identity in North-
west 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. Lin-
guistics 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 Mis-
sionária 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 Missio-
nária 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 Linguis-
tics, 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
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).
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.
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
(2) viaje-i- ‘to travel’ (from the Spanish noun viaje ‘journey’)
(3) dewe-i- ‘to owe’ (from the Spanish verb-root debe- ‘owe’)
(4) reso-i- ‘to pray’ (from the Spanish noun rezo ‘prayer’)
(5) fieshta-i- ‘to party’ (from the Spanish noun fiesta ‘party’)
(6) pasar-yi- ‘to happen’ (from the Spanish infinitive pasar ‘happen’)
(7) saludar-yi- ‘to greet’ (from the Spanish infinitive saludar ‘greet’)
(8) suerte-yi- ‘to be lucky’ (from the Spanish noun suerte ‘luck’)
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 ‘al-
ways’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 bor-
rowed from Spanish into Mosetén. All Spanish connectors are well-estab-
lished loans in Mosetén, even though they differ in the environments in which
they can be used. Thus, the three Spanish coordinating conjunctions y ‘add-
ition’ (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 to-
gether 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 juxtapos-
ition, 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).
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 con-
junctions are borrowed from Spanish. These include complementizers and
adverbializers.
Object complements can be expressed by a marker resembling the Span-
ish complementizer que, often pronounced ki (cf. example 15), and most
probably being borrowed directly from Spanish and then phonologically in-
tegrated.11
I have only one example of the Spanish complementizer a being used in Mo-
setén – and the speaker later corrects it, identifying it as a “slip of the tongue”,
i.e. it is not an integrated loan (example 16).
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 Span-
ish. 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
Purpose and causation are expressed by the marker pajki, which seems to be
a phonologically integrated marker of Spanish para que ‘so that’. The integra-
tion 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 pur-
pose marker. The marker porque, which in Spanish expresses causation, in
not borrowed into Mosetén (and pajki is used in all instances).13
Mosetén 573
The marker ashta appears parallel to native ways of expressing temporal ad-
verbial 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 occasion-
ally 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 simi-
lar 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 mark-
ers that appear – though less frequently – are osea ‘that means, so’, explain-
ing 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 min-
imal response.
(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)
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 in-
clude ashta (from Spanish hasta ‘until) (cf. example 26),16 desde ‘from’ (27)
and rarely a ‘to’ (28).
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.
5. Constituent order
6. Syntax
There are several Spanish contact phenomena in the syntax of Mosetén. Pri-
marily, these involve the organization of clauses, i.e. strategies for coordinat-
ing and subordinating clauses.
Negation strategies are not influenced by Spanish, but a number of delimi-
tation markers, expressing similar concepts, are borrowed from Spanish into
Mosetén (cf. examples 3133).
Mosetén 577
7. Summary
Abbreviations
Notes
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: Mou-
ton de Gruyter.
2007a Language contact between Spanish and Mosetén: A study of grammat-
ical integration. International Journal of Bilingualism 11 (1): 2553.
580 Jeanette Sakel
converbs, 41, 61, 95, 96, 102, 173, 179, determiner, 154, 202, 490, 529532,
217, 222, 293, 295 541542
coordination, 55, 61, 85, 179, 183, 193, definiteness as relative-clause marker,
254, 255, 294, 295, 409, 451, 457, 155
543, 546, 577, 579 negative, 542
see also conjunction(s) diminutive, 4344, 188, 217, 233236,
copula, 32, 46, 60, 94, 95, 102, 104, 249, 390391, 469, 484485,
119, 146, 156, 159, 163, 173, 490492
177179, 202, 217, 226, 272, 273, direct object, 221, 230, 442, 486, 487,
297, 321, 325, 393, 457, 460, 468, 504, 512, 532533
511513, 516, 547, 560, 562 enclitic pronouns as, 325
difference main and subordinate de-topicalization, 226
clause, 99, 364 suffix, 554
word order, 220221 topicalizing, 147
see also negation direction, 98, 127, 271, 349, 351, 356,
correlative particle, 144 416
counterfactual, 93, 557 discourse
see also irrealis procedural, 432
coverbs, 48, 269, 371 discourse markers/particles, 5, 20,
cross-reference, 96, 119, 120, 505 21, 25, 57, 58, 62, 66, 67, 70,
current contact, 6, 39, 246, 247 130133, 143, 158, 176, 238239,
241, 252253, 270, 377379, 382,
dative 94, 95, 113114, 120, 153, 168, 447, 451452, 455, 460, 470, 475,
170, 217, 232, 233, 248, 267, 297 506509, 541, 546, 547, 562, 569,
days of the week, 59, 65, 157, 192, 334, 570, 574
338, 348, 380, 507, 545, 560, emotive, 394
decimal, 50, 291, 292, 338, 348, 393, hesitation, 238, 410, 425, 428
470, 569 resumptive, 545
definite article, 35, 41, 43, 139, 140– sequential, 375
141, 142, 148, 155, 157, 162, 201, disjunctions, 20, 55, 114, 119, 408, 451,
248, 264, 277, 390, 391, 448 457, 470, 506, 559, 570, 579
definiteness, 42, 43, 44, 70, 115, 116, donor language, 16, 509
138, 139142, 148, 278, 348, 391, dual, 445446
491, 519 n. 6, 529, 554
agreement definiteness embedding, 254, 577
relative clauses, 86 ergative, 41, 288289, 290, 296297,
and specificity, 391 369, 388
deictic element, 5354, 58, 66, 68, 86, past-tense inflection, 202, 204205
155, 158, 161, 163, 251, 252, 255, split ergative, 168, 288
256, 269, 271, 277, 278, 381, 473, loss, 152153
506507, 529, 530 evidentials, 173, 174, 426, 496, 513,
development into relativizers, 130 514, 519 n. 15, 557558, 560, 578
dependent marking, 110 n. 7
586 Index of subjects
linguistic area, 6, 16, 2224, 43, 69, synthetic, 467, 479, 526
168, 175, 226, 272, 286, 288, 578 type change, 33, 40, 69 n. 7, 132,
Amazonia, 567 308, 416, 444, 479
Australia, 364, 366, 382 motion, 371372, 391
Balkans, 20, 226 movement
Caucasus–Mideastern–South Asian,
47, 49 narrative, 115, 128, 293, 336, 428, 430,
Caucasus–Anatolia–South Asia, 49 557
Ethiopian, 91102 nasal harmony, 524
Gurage, 20 necessity see modality
Meso-America, 24, 403 negation, 19, 58, 117, 119120, 129,
Papua, 329 152, 160, 206, 271, 291, 338, 341
South Asia, 286, 288 n. 1, 350, 376377, 382, 397, 416,
Vaupés, 24, 551552 560, 562, 576
West Africa, 107122 agreement, 273, 279
Western Asia, 264 circumfix, 546
loanverbs, 250, 267268, 277, 296, comparison, 478
568569 constituent negation, 376
locative predicate, 389 contrastive, 339
converb, 96
modality, 4546, 55, 56, 57, 66, 117, double negation, 237, 254255, 339
156, 162, 217, 269, 315, 334, 340, emphasis, 20, 242 n. 4, 559
356, 367, 456 negative copula, 129
alethic, 494 negative polarity item, 376
complement of modal predicate, (inherently) negative verbs, 129, 291,
274275 350
deontic, 175, 206, 288 subordinate clause, 255
epistemic, 46, 175, 223, 494, 557 see also particles, prefix, pronouns
modal complements, 161, 222223 nominalization, 314, 493
modal verbs, 113, 296, 391, 493, agent, 492, 510
495496 causation, 298 n. 4
non-factual complementizer, 56, 158, embedded propositions, 224
274 location
through periphrasis, 391 loss, 516517
see also adverb, counterfactual, irrea- modality, 175
lis, negation, particle relative clause, 179, 513
morphological type vs. subordination, 501502, 510, 511
agglutinitive, 33, 41, 110, 123, 308, non-verbal predications, 159160, 163,
369, 467, 479, 484, 553 272273, 416, 511, 512513, 535,
analytic, 479 536537
isolating, 40, 47, 48, 49, 110, 308, noun incorporation, 174, 527528
368, 554 number, 86, 110, 111112, 143, 157–
polysynthetic, 40, 403, 414, 525 158, 202, 312, 404, 528529
588 Index of subjects
syllable structure, 18, 78, 80, 82, 188, utterance, 38, 48, 273
213 n. 4, 241, 264, 292, 307308, modifier, 494
346, 354, 346, 387, 482, 525 planning, 20, 33, 35, 163, 221, 226
utterance-initial particle, 116
tense see alignment, ergative, future, utterance-final particle, 116
participle, particle, suppletion see also adverb
time expressions, 58, 535, 544, 545
see also days of the week, time of valency, 47, 48, 372
the day changes, 281 n. 4, 290, 499501,
times of the day, 59, 65, 157, 192, 380, 535, 536, 537539, 558
507 differentiation, 218
tone, 18, 38, 77, 8283, 286288, 298 marking, 49, 113, 562
n. 6, 345 350, 354, 553 see also transitivity
class, 83 verb compounding see compounding
lexical, 80 verb serialization see serial verbs
merger, 286 verbalization, 48, 58, 181 n. 5, 268,
pretone position, 247 331333, 423
see also adposition ideophones, 99, 100
topic infinitives, 422
change, 471, 571 interjections, 251
development, 131, 221 nouns, 142, 268, 558
de-topicalization, 221, 226 see also prefix
discontinuous, 579 n. 9, 11 verbness, 49, 371, 392, 569, 578 n. 6
fronting, 395 vocative, 200, 249, 305, 427428
left dislocated, 115 see also particle
marker, 238, 321, 485, 507 voice, 47, 350, 423
position, 331, 333 marking, 113, 562
prominence, 485 see also passive, valency
shift, 131, 375, 378 voicing, 232, 247, 286287, 403, 482
switch, 378 vowel, 3638, 65, 78, 80, 82, 124, 152,
topic–comment, 352, 358 n. 14 167, 187188, 247, 263, 306307,
topicalization, 147, 309 330, 387, 388, 420, 436, 438,
transition see shift 466467
topicalizer, 490491, 508, 511, 512– adaptation, 470
513, 519 n. 6 amplitude, 288
transitivity, 96, 202204, 218, 288289, alternation, 168
296297, 368, 372, 407, 500, 538 compound vowel, 202
agent marking, 369 diphthongization, 445
ditransitive, 96 epenthesis, 167
marker, 156, 372 gliding, 305
participant marking, 532 harmony, 167, 169, 174, 263, 439
see also causative, ergative, reflexive, insertion, 388, 483
word order intensity, 288
Index of subjects 593