Publishing Partner: Cambridge University Press CUP Extra Publisher Login
amazon logo
More Info

New from Oxford University Press!


Cognitive Literary Science

Edited by Michael Burke and Emily T. Troscianko

Cognitive Literary Science "Brings together researchers in cognitive-scientific fields and with literary backgrounds for a comprehensive look at cognition and literature."

New from Cambridge University Press!


Intonation and Prosodic Structure

By Caroline Féry

Intonation and Prosodic Structure "provides a state-of-the-art survey of intonation and prosodic structure."

Review of  The Mixed Language Debate

Reviewer: Felicity Meakins
Book Title: The Mixed Language Debate
Book Author: Yaron Matras Peter Bakker
Publisher: De Gruyter Mouton
Linguistic Field(s): General Linguistics
Issue Number: 15.2736

Discuss this Review
Help on Posting

Date: Wed, 29 Sep 2004 14:55:25 +1000
From: Felicity Meakins
Subject: The Mixed Language Debate: Theoretical and Empirical Advances

EDITORS: Matras, Yaron; Bakker, Peter
TITLE: The Mixed Language Debate
SUBTITLE: Theoretical and Empirical Advances
PUBLISHER: Mouton de Gruyter
YEAR: 2003

Felicity Meakins, Department of Linguistics and Applied Linguistics,
University of Melbourne, Australia.


Mixed languages (MLs) were considered an oddity of contact linguistics until
Thomason and Kaufman (1988) revisited the challenges they posed. Before
then, debate about whether or not MLs actually existed stifled much
descriptive work or discussion of their origins. Peter Bakker's 'A Language
of Our Own' (1997) provided the first detailed description of a mixed
language, Michif. Consequently, the debates surrounding MLs have shifted
from questioning their existence to a focus on their formation and
structural features. Yaron Matras and Peter Bakker bring together many of
the authors who have written on MLs for the past decade in their new book
'The Mixed Language Debate' (2003). This book sweeps across the field of
mixed languages, containing restated positions about ML genesis, e.g. Carol
Myers-Scotton and Peter Bakker, new data for these old debates e.g. Ad Backus,
and new offerings to the ever growing list of these languages, e.g. Thomas


In their introduction, Matras and Bakker posit six types of mixed languages
based on the ML speakers' knowledge of the source languages, functionality,
typology of the structure, and various social factors - plain,
conventionalised, special lexicon of foreign origin, radical restructuring,
mixed creole and extremely heavy borrowing. They also briefly outline
various theories on the emergence of MLs. Matras and Bakker describe 'plain
MLs' as those MLs which have a high level of functionality in a language
community (ie exist as a L1 in everyday usage) and which have lost contact
with the source languages. Michif, which finds its origins in French and
Cree is the only example of a plain ML. Though only spoken now by older
generations, it has an everyday use for those speakers, and they do not know
French or Cree. English is now the main community language.

'Conventionalised MLs' include Mednyji Aleut and Media Lengua. These MLs are
usually spoken along side one of their source languages, for example Media
Lengua speakers also speak Spanish. Conventionalised MLs also emerge in
situations of full bilingualism. Ma'a , Para-Romani dialects, Lekoudesch,
Callahuaya and Abdal/Aynu comprise the 'lexicon of foreign origin'
category. Like the other categories, these languages mark an in-group
identity and arise from situations of mixed parentage, however they are
secret in nature and not in everyday use and therefore at the lower end of
the functional continuum. For example, Lekoudesch , which mixes a
Judeo-German dialect with Ashkenazic Hebrew, is the secret language of
Jewish cattle traders in Germany. 'Radical restructures' in languages such
as Javindo, which derives from Dutch lexicon and Javanese grammar, exhibit
severe genetic ambiguity and a greatly simplified morphological system.
Chavacano and Berbice Dutch are classified as 'mixed creoles' .These Spanish
and Dutch-based creoles, are also influenced by other languages in the form
of vocabulary and some syntactic features. The most marginal group of MLs
derive from 'extremely heavy borrowing'. Unlike conventionalised MLs they do
not form from situations of bilingualism, and are more easily classified

In this introduction, Matras and Bakker also outline the three major
explanations for the formation of MLs - 'relexification' (Muysken 1981),
'language intertwining' (Bakker 1997), and 'matrix language turnover'
(Myers-Scotton 1998). In this outline they raise a number of issues which
lead into the articles offered by 'The Mixed Language Debate'. Matras and
Bakker (p 15) are particularly interested in whether MLs should be
considered cases of language maintenance or shift, whether they are the
result of a gradual or rapid processes, the effect of the social situation,
whether ML formation is deliberate or conscious, the extent to which
formation is predetermined, and the role of code-switching in ML genesis. By
outlining the range of mixed languages available for analysis and theory
development, this chapter eases the reader into the issues that are
developed in the rest of the book The groupings suggested by Matras and
Bakker are based fairly broadly on a range of social, typological and
genetic features. The one criticism that could be made is that Michif is
still posited as the classic ML (reflected in the name 'plain'), the
reference point for other MLs. In fact it remains the most unusual instance
of a ML with 'conventionalised' and 'special foreign lexicon' groups
accounting for a majority of these languages.

In the first chapter, Sarah Thomason considers some of the social factors
which may have contributed to the genesis of a mixed language. Most of this
material comes from her 2001 book. At the heart of her proposal is the
conscious and deliberate nature of formation, and the role of codeswitching.
Thomason begins by restating her position on MLs which is broadly defined by
the genetic ambiguity of the ML. Under this definition, she includes pidgins
and creoles as well as bilingual MLs (referred to elsewhere as just MLs)
which are differentiated due to their "particular structural and lexical
subsystems [which] are adopted intact from each source language" (p. 22).
She then outlines seven mechanisms which she suggests contribute to the
development of MLs. Of these, only two are found to affect bilingual ML
genesis. Firstly she considers 'codeswitching', which she believes is the
most important mechanism in the genesis of Michif (p. 27). Bakker (1997)
himself objects to the idea that Michif arose from codeswitching between
French and Cree mainly due to evidence which suggests that codeswitching
does not occur where only certain categories (e.g. nouns or verbs) are taken
from another language. Thomason cites evidence from the codeswitching
practises of the Montagnais French (Drapeau 1991), which she argues looks
very similar to the structure of Michif. Bakker has also examined this data
but maintains that codeswitching was not a factor in Michif's formation. The
other mechanism which Thomason proposes as a contributor to ML genesis is
the 'deliberate decision' of a speech community. One type of deliberate
change suggested is the systematic distortion of a language's lexicon to
make it unintelligible to outsiders. Thomason suggests that small speech
communities and MLs that emerge abruptly are indicative of a deliberate
change. She also links this level of consciousness to the use of MLs as
symbols of an in-group identity, whether it be new or an already existing
threatened social entity (p. 36). As Thomason acknowledges, the problem with
any of these claims is that there is no empirical evidence to support them,
and therefore methodological grounds must be resorted to. Thomason has done
this to a certain extent, however the progression from, for e.g.,
codeswitching to an ML is still not as clearly spelt out as other writers
such as Carol Myers-Scotton, who restates her own genesis theory in the next

In "What lies beneath: Split (mixed ) languages as contact phenomena", Carol
Myers-Scotton provides a reiteration of her position on the formation of
mixed languages (which she calls split languages). Myers-Scotton's approach
differs quite radically from Bakker's on the question of language genesis.
She considers MLs to be a strong language contact phenomenon, the result of
pervasive CS practices. Myers-Scotton classifies MLs based on their
morphological structure, rather than a more general lexical/grammatical
approach, or a sociolinguistic approach. A distinction is made between early
system morphemes and late system morphemes. Both differ from content
morphemes in that they neither assign nor receive thematic roles (e.g. noun
and verb forms) (p. 77). However early system morphemes pattern with content
by specifying the meaning of the phrase head, for e.g., plural marking in
English (p. 77). Late system morphemes are structurally assigned by
constituents higher than their phrase head. Case marking by verbs in some
languages is one such instance (p. 78). These late system morphemes become
crucial for Myers-Scotton's classification of MLs. Myers-Scotton believes
that MLs are the result of an arrested Matrix Language Turnover, where the
shift from the dominance of one language to another in code-switching
situations becomes fossilised. The reason it stops is attributable to
sociological reasons (p. 90). The result takes one of three forms according
to Myers-Scotton - Type A (late system morphemes from the less dominant
language are still present), Type B (the less dominant language provides
some late system morphemes which may replace those from the dominant
language) and Type C (late system morphemes from the less dominant language
appear in the dominant language's frame but are reanalysed) (p. 92). Most
MLs are classified as Type B MLs. There are only two surprises. Firstly,
Media Lengua is not classified as a ML under this system because of the
absence of an abstract grammatical structure from both languages (p. 91).
Secondly, Michif is classified as a Type C ML and therefore quite marginal.
In fact, because Michif retains quite a strict language split between the NP
and VP structures, most of the system morphemes retained are early ones.
However French verbs integrated into the Cree VP show the dummy element from
French as well as Cree inflectional suffixes and the French infinitive form
(p. 99). Myers-Scotton suggests that "this form has been re-analysed as a
structurally-assigned element (it has a late system morpheme) to fill a
grammatical role that Cree requires in VPs" (p. 99). However by
Myers-Scotton's reckoning, Michif is quite marginal as a ML. Whilst it is
always refreshing to read a ML theory not assuming Michif as the
prototypical from, it is seems somewhat problematic that Michif is
considered so marginal, particularly when it represents the most intertwined
form of a ML presented in the literature. In general, Myers-Scotton has not
added much new material to her position on language mixing, however her
overview of her position is presented very neatly.

Peter Bakker's article "Mixed languages as autonomous systems" is nicely
juxtaposed to the Myers-Scotton article as it provides the other strong
position in the debate about the role of code-switching in ML genesis.
Bakker sets out on the same program as Myers-Scotton - to classify the
corpus of MLs and posit a theory for their formation. Unlike Myers-Scotton,
his classification is largely synchronic (based more on typological
characteristics which he doesn't relate to his theory of intertwining), and
his theory of language genesis presents specific processes for ML formation,
unlike Myers-Scotton and others (e.g. Auer 1999) who argue that MLs are the
extraordinary result of ordinary processes such as codeswitching. Bakker
argues for four types of mixed languages: 1. Intertwined languages
(lexical-grammar MLs) e.g. Angloromani and Media Lengua; 2. Converted
languages (form-semantic MLs) e.g. Modern Sri Lankan Portuguese; 3. Lexically
mixed languages (lexicon A+B ML) e.g. Russenorsk, and 4.Verb-Noun MLs e.g.
Michif. Of these, the largest category are the intertwined languages of
which there are 25 documented languages, compared with Verb-Noun MLs where
only 2 have been noted (p. 125). Perhaps the most interesting part of this
paper is his arguments against the codeswitching hypothesis. He provides
seven arguments in all, some new some old. He repeats one of his earlier
arguments against CS that the codeswitching patterns between source
languages do not look like the MLs which have formed from those languages.
He wisely glosses over his own egs of CS between Cree and French, which have
been presented fairly convincingly in opposition to Bakker's own claims.
Instead Bakker uses the example of a Turkish-Romani ML and codeswitching
between the source languages (p. 132-33). The most interesting argument
against a CS genesis that he presents is typological. Bakker cites work by
Muysken (2000). Muysken demonstrates that if an agglutinative language is
the matrix language in CS, then the resulting CS will be insertional. On the
other hand, if both languages are flectional, alternational CS will be the
result. In these cases, you will rarely find L1 stems with L2 endings.
Bakker claims that "in mixed languages it makes no difference whether that
'matrix language' is inflectional or agglutinative: the mixed language is
invariably insertional (L-G type)" (p. 131). Bakker goes on the re-present
his Intertwining Model, first suggested in 1997. He emphasises the social
factors (women/men-grammar/lexical split), and admits that ironically enough
the model does not account for Michif unless typological features are
considered. Whilst Bakker provides social factors, where Myers-Scotton does
not, the same question arises. If the language community are not
codeswitching between the source languages, what does the language situation
look like at the point of intertwining? For example, what are the children
of mixed ethnicity speaking to each other and to their parents? The other
element missing from this re-presentation of the intertwining theory is the
processes by which intertwining occurs. In his 1997 book, he does provide
some more explicit processes (p. 211). Whilst his refutation of
codeswitching has some strong arguments, more work is needed to provide a
convincing alternative.

Yaron Matras' article, "Mixed languages: Re-examining the structural
prototype", marks a shift into newer work on MLs. This article examines
typological features across seven very different MLs, concentrating on those
grammatical categories which are rarely borrowed in situations of standard
borrowing. Matras uses these to construct a typology of an ML which
differentiates it from standard borrowing. He begins by criticising the
grammar/lexicon split ML prototype (p. 152-53), then considers the types of
constraints which have been posited for standard borrowing in order to help
establish how the structures of MLs are unique. He cites the Swadesh list as
a good indicator for judging the extent of lexical borrowing, and a number
of grammatical categories such as the copula, in/definite articles, bound
TAM verbal markers and sentential negators as structures which have been
shown to be rarely borrowed in standard borrowing. Putting lexical borrowing
aside, Matras examines many of these grammatical features across a number of
mixed languages including traditional gram/lex split MLs such as Media
Lengua, NP/VP ML, Michif and secret MLs e.g. Para Romani. He posits a divide
between the INFL language and lexifier language which is more well defined
than the gram/lex split traditionally used to define MLs. The INFL language,
which provides the finite verb inflection, tends to supply rules for word
order, and anchors the predicates. It is not unlike, but more specific than
Bakker's notion of the grammar language or Myer Scotton's Matrix Language.
The lexifier language generally provides the verb roots (Michif being the
exception), and modal verbs (p. 165). This patterning is not unlike standard
borrowing restrictions, however one main difference is that MLs seem to
include "a licence to incorporate classes of grammatical items that are
otherwise rarely borrowable" (p. 166). He notes that many grammatical
categories are less 'loan proof' in the situation of MLs than in general
borrowings. For example, the copula tends to pattern with the INFL-language,
where the INFL-language is historically continued, and with the lexifier
language where there has been a shift in the INFL-language (p. 167). This
article is a renewed attempt to construct a structural prototype for MLs. It
looks more closely at the gram/lex split, defining it more rigidly and
outlines some interesting "extraordinary" processes which seem to have
contributed to MLs.

The final few chapters of this book moves away from the larger theoretical
issues of mixed languages to zero in on particular problems, specifically
that of the role of speaker intentionality in the formation of a ML. Evgeniy
Golovko begins this section with his article, "Language contact and group
identity: The role of 'folk' linguistic engineering". Golovko begins by
aligning himself with Bakker's notion of language intertwining. He
highlights one important distinction between the formation of a creole
versus an ML. Golovko suggests that MLs are not formed due to a lack of a
communication vehicle, rather ML groups always have two fully functioning
languages at their disposal (p. 191). There are other imperatives driving
language formation, the most of important Golovko considers to be a
conscious and deliberate marking of identity. Golovko supports this claim by
examining lexical re-orientation as an example of intentional language
manipulation (p.192-96). He posts language re-orientation, along with other
contact induced mechanisms which are not controlled by speakers, as an
alternative to code-switching as the main process in ML genesis.

Maarten Mous continues with the idea of 'lexical re-orientation' which he
calls 'lexical manipulation', in his article, "The linguistic properties of
lexical manipulation and its relevance for Ma'a". He also suggests that this
sort of process is quite conscious, suggesting it is "the 'conscious'
creation of lexical forms that are parallel in semantic and morphosyntactic
properties to an existing lexical item in the language" (p. 209). He
examines Ma'a as a specific instance of lexical manipulation, noting that
Ma'a's lexicon comes from a variety of languages in order to mark it as
different from its grammar language Mbugu. Some of the manipulations that
have occurred include simple borrowings (into the same morphosyntactic
categories such as gender or argument structure) (p. 210-11), the truncation
of Mbugu words, adding a dummy vowel, and substituting consonants with
marked consonants (p. 214). He goes on to look at lexical manipulation in
other language contact situations, for example respect registers, taboo,
slang and urban youth languages. He compares these situations with that of
MLs to begin constructing a set of lexical manipulation strategies which are
similar across these different language situations,. Mous concludes that
lexical manipulation "occurs not only in mixed languages but in a variety of
situations in which speakers attempt to exercise conscious control over
their language' (p. 230).

In "Can a mixed language be conventionised alternational codeswitching?", Ad
Backus sets out to answer this question. Backus reviews the arguments
against fossilised codeswitching (CS), and concludes that insertional CS
could become so conventionalised that it becomes predictable, although MLs
do not look enough like CS for this to be the only process at work. However
he concedes that alternational CS involves "unbridled variation" (p. 239),
and suggests that predictability is an important factor in CS fossilising.
He suggests that a ML borne from alternational CS would have to contain lots
of converntionalised utterances where the choice of language was predictable
(p. 256). This predictability would be on the basis of content not
discourse. However, through his own Turkish-Dutch CS data and Hill and Hill
(1986) study of Mexicano, Backus finds differences between the functional
motivations of alternational and insertional CS. He finds that
lexico-semantics motivates CS in insertional CS and pragmatics in
alternational CS. Moreover the structural configuration of insertional CS is
monolingual, whereas alternational CS requires bilingualism. He concludes by
suggesting that alternational CS serves a communicative function through the
choice of language, and an ML must contain two languages which do not index
anything separately(p. 264).

The final article in this volume, written by Thomas Stolz, considers whether
the two languages , Chamorro and Malti should be classified as MLs. Chamorro
and Malti are both languages which have been in contact with Romance
languages, and undergone significant language change as a result. Chamorro
is an agglutinative Austronesian language with features of fusional
morphology. It has a VSO word order with a split ergative case system. Malti
is an Afro-Asiatic language which is highly fusional with a relatively free
word order and a nominative /accusative structure. Stolz examines their
grammatical and lexical structures in relation to MLs. He begins with
Chamorro. Stolz finds that the grammatical systems of Chamorro and Malti are
essentially Austronesian and Afro-Asiatic respectively, with Romance words
taking on Chamorro/Malti inflections. In Chamorro he notes some Spanish
derivational morphology, and the Spanish gender distinctions. His study of
the makeup of Chamorro and Malti lexicons is perhaps more interesting.
Bakker and Mous (1994) suggest that languages which exhibit extreme
borrowing contain around45% of another language's lexicon, whereas they
suggest a figure of 90% for MLs. At the time of their proposal they noted
that no languages contained borrowings between 45 and 90%. Stolz provides
evidence that Chamorro and Malti exist in this range. He finds that Chamorro
contains 55% Spanish lexicon, and Malti, 57% non-Semitic material. However
these figures diminish when bilingual synonyms are considered, to the extent
that Stolz says "it becomes doubtful whether Chamorro and especially Malti
would even qualify as instances of massive borrowing" (p. 292).

The majority of papers draw on a wide ranges of languages identified as
mixed in order to debate some crucial issues in the field - ML genesis and
the contribution of extra/ordinary social, typological and general language
contact processes, the typology of MLs, and the role of speaker group
identity and intentionality. This book is not an introductory volume on MLs,
however it does represent a wide range of views on these ML issues, which
makes it a useful reference for those with an interest in language contact.
One trivial, but fundamental problem that arises with representing such a
wide field is nomenclature. MLs are variably referred to as mixed languages,
fused lects (Auer), split languages (Myers-Scotton) and bilingual mixed
languages (Thomason). Some uniformity in the field would avoid reader
confusion and necessary explanation at the beginnings of articles.

"The Mixed Language Debate" demonstrates an evolution in the field of mixed
language studies. Stolz ends the this 'debate' where it began a decade ago
with identification and description of potential MLs (see for e.g. Bakker and
Mous 1994). This article also gives an idea of where the field may be
heading. Stolz's study is interesting in its inability to satisfactorily
classify Malti and Chamorro. It highlights the problem of attempting to
compartmentalise and search for homogeneic features of language contact
phenomena which are the result of highly varied, and variable social
factors. The variation of these languages is probably not unusual, however
it is rare for descriptions of MLs to admit to variation of this level.
Stolz's focus on extreme variation highlights to need to deal with and
classifying this 'mess' rather than merely emphasise the patterns. As more
fundamental identification and descriptive work is done, more light will be
shed on some of these issues. Unfortunately there is still a certain amount
of elitism involved in classifying a language as mixed, such that criteria
tend to be restrictive (and sometimes somewhat arbitrary in the case of
Bakker and Mous' borrowing scale) rather than inclusive.

Auer, Peter.1999. From Code-Switching via Language-Mixing to Fused Lects:
Toward a Dynamic Typology of Bilingual Speech. International Journal of
Bilingualism 3: 309-32.

Bakker, Peter. 1987. A Language of Our Own: The Genesis of Michif - The
Mixed Cree-French Language of the Canadian Metis. New York: Oxford UP.

Bakker, Peter and Maarten Mous (Eds).1994. Mixed Languages: 15 Case Studies
in Language Intertwining. Amsterdam: IFOTT.

Drapeau, Lynn. 1991. Michif Replicated: The Emergence of a Mixed Language in
Northern Quebec. Paper presented at the 10th International Conference on
Historical Linguistics, Amsterdam.

Muysken, Pieter.1981. Halfway between Quechua and Spanish: The Case for
Relexification. In Historicity and Variation in Creole Studies. Ed. A.
Highfield and A. Valdman. Ann Arbor: Karoma. 52-78.

Myers-Scotton, Carol.1998. A Way to Dusty Death: The Matrix Language
Turnover Hypothesis. In Endangered Languages: Language Loss and Community
Response. Ed. Lenore Grenoble and Lindsey Whaley. Cambridge: Cambridge UP.

Thomason, Sarah. 2001. Language Contact: An Introduction. Edinburgh:
Edinburgh UP.

Thomason, Sarah and Terrence Kaufman. 1988. Language Contact, Creolization
and Genetic Linguistics. Berkley: California UP.

Felicity Meakins is a PhD Student at the University of Melbourne. She is a
part of the Aboriginal Child Language Acquisition (ACLA) project, and is
working on a mixed language, Gurindji Kriol, which is spoken in Northern
Australia. Meakins previously worked with a number of Ngumbin languages in
Northern Australia through the Katherine Regional Aboriginal Language

Amazon Store: