Muysken, Pieter (2000) Bilingual Speech: A Typology of
Code-Mixing, Cambridge, Cambridge University Press.
xvi+306 pages, ISBN 0521-77168-4 (HB) US$59.95
Reviewed by Madalena Cruz-Ferreira, National University of
Singapore
Linguist List book announcement at
http://linguistlist.org/issues/11/11-2080.html#1
Pieter Muysken's new book proposes a first synthesis of
previous and current work on intra-sentential code-mixing,
with particular emphasis on adult bilingualism. Pieter
Muysken (PM) puts forward a (re)analysis of the wealth of
data available in this field, from within a tripartite
typology of his own that accounts for different strategies
in code-mixing and enables the making of assumptions about
language structure in more general terms.
SYNOPSIS
Chapter 1 ("The study of code-mixing") starts by
introducing the terminology adopted in the book. The label
"code-mixing", an apt choice as a cover term for different
types of mixes, refers to "all cases where lexical items
and grammatical features from two languages appear in one
sentence" (p.1). The use of the term "bilingualism", though
not defined in the book, encompasses any use of two
languages, or dialects, by an individual or a community,
in different degrees of proficiency. The chapter then
reviews the literature on code-mixing, presenting the
different theoretical frameworks and concerns that
characterise current research in this area, and concludes
with the author's own proposal for the analysis of code-
mixing. PM takes the view that "the process of code-mixing
is not unitary, but consists of three main mixing
strategies: insertion, alternation and congruent
lexicalization " (p.32). Each of the strategies is
discussed in detail in chapters 3, 4 and 5, respectively,
where diagnostic frames are also proposed for each.
Chapter 2 ("Differences and similarities between
languages") addresses issues of language typology, from
the working hypothesis that differentiation between
languages "results from the different ways in which two
autonomous modules, the lexicon and the grammar,
interact." (pp.51-52). The chapter also gives a brief
history of theoretical approaches, particularly
generativist, to language differences.
Chapter 3 ("Insertion") deals with the inclusion of
lexical items or entire constituents of one language into
a structure of another. Insertion thus compares with the
process of syntaxis (or selection). The chapter focuses on
the behaviour of noun phrases in mixed sentences and also
provides a detailed analysis of borrowing, relating it not
only to insertion itself, but to the other two code-mixing
strategies.
Chapter 4 ("Alternation") takes up the switching over
between structures of two languages, where alternation
compares with parataxis (or adjunction) in that each of
the languages remains relatively intact. A number of
criteria, among them embedding, length and complexity of
the mixed elements, may argue for instances of alternation
as opposed to insertion.
Chapter 5 ("Congruent lexicalization") concerns mixes
where lexical material from either language occurs in a
shared grammatical structure. PM observes that there is a
cline in the degree to which congruent lexicalisation
occurs in different bilingual communities, depending on
the degree of structural similarity between the languages
involved. At one end, the process is said to be akin to
style shifting, blending into intra-linguistic variation.
Chapter 6 ("Function words") reviews criteria for
the identification and classification of function words,
encompassing research in speech production and in aphasia,
surveying both grammar-based and prosodic-based models.
The observed asymmetry in the behaviour of function words
vs. content words in mixes is discussed, namely, their
retention or omission in mixed speech.
Chapter 7 ("Bilingual verbs") discusses verbs, which
"function as the core of the clause" (p.184). Code-mixing
in the verbal system is often found to be innovative,
"leading to structures not present in either of the
languages" (p.184). Each of the three code-mixing
strategies is found in mixes involving verbs,
correlating with the syntactic type of verb or with
its morphology.
In chapter 8 ("Variation in mixing patterns"), extra-
linguistic issues are brought to bear on features of
bilingualism, in order to assess whether they determine
language mixing and, if so, of which kind. PM suggests
that different types of mixing correlate with different
sociological, temporal and geographical factors, as well
as with the (political) statuses of the languages involved
and with the extent and type of language contact. Other
significant issues affecting mixes involve topic of
conversation, level of formality, register, individual
variation in, e.g., health and age among bilinguals, and
degree of bilingual proficiency. PM's proposed typology
predicts evolution in patterns of code-mixing: for
example, as the extent of bilingual contact grows,
insertional code-mixing will predictably shift to either
alternation or congruent lexicalisation (p. 249).
Chapter 9 ("Code-mixing, bilingual speech, language
change") begins by discussing the issue of simultaneous
access to both languages by bilingual speakers, proposing
a modular simultaneous-access model where "[b]oth
languages are accessed, but different modules of each."
(p.253). The implications of the work on code-mixing for
research on language change induced by language contact
are then assessed.
EVALUATION
Code-mixing is an extremely complex area, as a cursory
look in the literature will show. Research reflects this
complexity, in that it has variously focused on different
bilingual situations and individuals, different languages
and language pairs, different types of mixable structure
and, not least, different theoretical approaches to
bilingualism. PM's look in the literature on code-mixing
is anything but cursory, and this book constitutes an
important and richly documented source of information
about the state of the art in studies on code-mixing.
Despite the disarticulate state of research in the field
(PM notes himself that many of the findings are not
comparable, e.g., p. 232), discussion is supported by
abundant exemplification and by a useful set of tables
summarising findings, that are listed, with captions, at
the beginning of the book.
The book assumes familiarity with the terminology and
methods of, especially, current generative proposals,
including a number of its notational conventions. It also
assumes some familiarity with previous proposals in the
analysis of bilingual speech, and is therefore primarily
intended for linguists.
In this evaluation, I will first consider the formal
presentation of the text, and then turn to matters of
content.
Due to the diversity and amount of examples given in the
text, a number of formal problems give rise to
difficulties in their interpretation. The problems range
from erratic notation and, at times, awkward phrasing, to
inconsistencies or obscurity in terminology. A sample of
examples follows.
The terms "switchpoint" (p.6); "point of the mix" (p.34);
"switch site" (p.82); "transition point in alternation"
(p.96); "mixing point" (p.100); "locus" (p.136) are
variously used to refer to where a change in language
occurs in a mixed utterance.
In chapter 8, the various expressions "function words,
"functional elements", "functional categories", "closed
class words", "closed class items", "grammatical
elements", "grammatical morphemes" all seem to designate
the same type of units.
Repetitions, some of them nearly literal, at times make
the text read like a draft: e.g., the second and the last
paragraphs on p.18; the second last paragraph on p.126 and
the last paragraph on pp.152-153.
On p.114, the second paragraph of the second section has
two consecutive sentences starting with "In fact"; on
p.182, two sentences of the three in the third paragraph
begin with "However", and two consecutive sentences in the
second last paragraph begin with "Thus"; on p.233, two
consecutive sentences in the last paragraph begin with "Of
course".
A few turns of phrase are either awkward or unclear:
"As to insertion, here the matrix language must be assumed
to remain active at the point of utterance of the inserted
material (...)." (p.34);
"A number of elements form a unique constituent if that
constituent contains no other elements." (p.62);
"Here I want to argue that in addition to the insertional
route (...) the alternational route words may also allow
verbs to be borrowed (...)". (p.106).
There are also a number of forward references,
presupposing familiarity with specific terminology used in
research on code-mixing, as said above. One example
concerns "flagged switching", which is mentioned on pp.
15, 30 to 32, 93, 95, and which is defined as "specially
marked" mixing only in chapter 4 (p.101).
A rather more serious formal problem concerns the
presentation of examples of mixed utterances. PM appears
to reproduce the examples as-is, together with glossing
and translations into English, from the sources on which
he draws. Since familiarity with the different conventions
used in the vast array of studies discussed in the book
cannot be taken for granted, either a standardisation of
conventions or a key to different conventions should have
been provided. Examples of erratic or unclear notation
are:
p.62: sometimes the stretch, in English, that occurs in
mixes involving this language is contemplated in the
glossing, sometimes not. Examples are (2) and (4) (5),
respectively.
p.107: in the glossing of examples (48) and (49), the use
of a question mark as opposed to the phonetic symbol for a
glottal stop is not explained. Neither is the use of
hyphens as opposed to dots. Hyphens vs. dots also occur in
the reproduction of the mixed utterances themselves, e.g.,
(55) on p.85.
p.70: examples (24) vs. (25) (26) show an erratic use of
the forward slash to indicate a change in language. In
example (50) on p. 83, however, the forward slash
indicates a choice between two alternatives, whereas the
symbol '#' is used to signal language change. The use of
these conventions is not clarified. Other unexplained
symbols include - and _ in (51) and (52) on p. 84; the
use of boldface in the tables on pp.130-131; and the use
of square brackets, e.g., in (7) and (8), pp.5-6.
A number of mixed utterances are given in what looks like
phonetic script (except for the use of dots, hyphens and
sentence-initial capitalisation), although no explanation
is provided concerning which phonetic alphabet is used,
e.g., (55) on p. 85, and (48) (49) on p.107. Taking the
symbol @ to represent the schwa vowel, the Arabic word
usually transliterated as 'wahed' is variously given as
"wahed", on pp.82-83; "wah.@d", in (55) on p.85; and
"wah.ed", in (57) on p.85.
p.159-160: the table spanning the two pages would need a
title row on the second page too. In the same table,
"DERI" is defined as "does not participate in derivational
morphology". Since the table gives binary (+ vs. -)
features for criteria distinguishing functional elements,
the phrasing in the definition implies that the negative
'*not* participating' is assigned the *positive* sign +,
and vice-versa. Also, the criterion "UNST" is defined as
"can be without stress", which raises ambiguity about what
exactly a negative sign may involve.
In other cases, the lack of accurate glossing makes
interpretation of the examples nearly impossible. Two
examples illustrate this, both from the same page (p.110)
and reproduced here verbatim, along with glossing
(morpheme-to-morpheme, separated by hyphens) and
translations (given between single quotation marks).
The first example, given as (59), exemplifies an instance
of doubling (i.e., double-language repetition) in
Spanish/Quechua mixes. According to the text, italics
indicate Spanish, and boldface the Quechua doubling.
According to the Abbreviations on pp.xiv-xvi, we have:
DUB = dubitative
O = object
TO = topic
The symbols 1, 2 are included in the Abbreviations as
indicating either noun classes or participles, whereas
here they appear to signal grammatical person.
(Grammatical person and number are listed in the
Abbreviations as "1pl" and "1sg".) The examples are:
(59) _si_-chus munawanki *chay-qa* 'if you love me'
if-DUB want-1O-2 that-TO
_si_-chus waylluwanki *chay-qa* 'if you care for me'
if-DUB want-1O-2 that-TO
The text introduces these examples by saying that the
Spanish subordinating and sentence-initial conjunction
_si_ doubles with the sentence-final Quechua conjunction
_chayqa_ (not hyphenated in the text). The text then adds
that the "conjunction _si_ always occurs with the
indefinite enclitic _-chus_." Since Quechua '-chus' is
first glossed and then defined in two apparently different
ways, one of them interpretable in terms of dubitative
uses of Spanish 'si', the examples appear as a case of
trebling instead, involving both 'si-chus' and
'chay(-)qa'. That is, it is unclear why the obligatory dubitative
'-chus' is not part of the "doubling". In addition, it is
equally unclear what causes the difference in meaning
between 'love' and 'care for', since the glossing is
identical in both cases.
The next example, (60), discusses a "potentially drastic
integration into Spanish", constituted by "the use of
Spanish elements such as sentence-introducing adverbs as
conjunctions". The Spanish adverb under discussion is
given as "_siguru_ 'certain' ", as follows:
(60) _siguru_ manan~a mamayqa kanchu
'certain that I have no mother any more?'
_sigura_ taytayqa manan~a kanchu
'certain that I have no father any more?'
The examples are not glossed, apparently in the belief
that an approximately idiomatic translation into English
will suffice to explain the point, whereas the translation
in fact suggests a use of the sentence initiators as
adverbs, not conjunctions. The lexical status of the
initiators is further obscured by the use of the form
'sigura' in the second sentence. In Spanish, 'segura' is a
feminine (adjectival) form corresponding to a masculine
'seguro'. In addition, one of these two forms (depending
on which is to be taken as marked in Quechua/Spanish
mixes) appears to trigger inversion of the two following
constituents. No explanation is given for either
morphological or syntactic alternatives.
pp. 147-148: examples (82) and (84), of Spanish/English
mixes, are given as instances of different types of mixing,
the former where a "basically English expression contains
a Spanish verb and an English complement", whereas the
latter "is the other way around, with a Spanish expression
and a Spanish verb". Both examples contain, however, the
Spanish verb "dar", translated in both cases as 'give',
both idiomatic in both languages.
In other cases still, the reader is left to wonder whether
the suggested interpretation of the mixed utterance, as
given by the English translation, actually mirrors the
speaker's intention. One example is (44) on p. 105, of an
English/Japanese mix, given as a further instance of
doubling (italics indicate Japanese):
(44) You should see his _karada kinochi warui n da_.
body appearance awful-is
'You should see his bodily appearance, it's awful.'
Since no indication is given of the intonation pattern on
which the speaker uttered the sentence (I return to
matters of intonation below), the given gloss cannot be
made to warrant the sole interpretation proposed in the
English translation, which assumes some type of pause
marked by the comma. On the strength of the gloss alone,
another, equally plausible translation of the mixed
utterance could be 'You should see, his bodily appearance
is awful', with pause after 'see'.
However, the point that PM wishes to make with this
example is that "the doubled elements are not semantically
parallel". His clarification is that "_karada kinochi_
'body appearance' functions as the object of _see_ and is
modified by _his_, but at the same time it is the subject
of the Japanese predicate phrase _warui n da_ 'is awful'."
In (44), then, the concept of doubling appears to acquire
a different meaning from the cases of doubling discussed
before and after this example. Here, doubling refers not
to the repeated occurrence of equivalent units, one in
each language, but to a dual syntactic function of the
same constituent, which is spoken in one language only.
Other difficulties arise, for example, in the
interpretation of the criteria for the assignment of
particular stretches of mixed speech to particular
languages. Two examples are (35) and (37) on pp. 259-260,
both of Moroccan Arabic/Dutch mixes. Dutch is italicised,
and the two "-ss" of "ka-yxess" and "xess-na" are dotted
under, in the text:
(35) ka-yxess [[bezzaf dyal _generaties_] _voorbijgaan_]
it must much of generations pass
'Much of a generation must pass.'
(37) xess-na [[m9a _bestuur_] _praten_]
we-must with board talk
'We must speak to the board.'
PM is discussing an "asynchrony between the syntax and the
lexicon", and states that in (35) "the Dutch plural noun
_generaties_ is modified by _bezzaf dyal_ 'much of', and
is thus part of an Arabic noun phrase", whereas in (37)
"Arabic _m9a_ 'with' occurs in an otherwise Dutch verb
phrase."
Problems such as these, besides disrupting the fluent
reading all the more necessary in a complex subject like
the one discussed in this book, make it rather trying, at
times impossible, to follow the points made and hence to
assess their validity.
Despite these difficulties, the book raises much food for
thought, and this is not the least of its merits. I would
like to take up three issues in turn, those that lie
closest to my own research interests in this field.
1) Grammar and words do not make a language.
The stance taken in the book is that a language equates
with its grammar and its lexicon. Bilinguals are said to
"dispose of two grammars and lexicons" (p.69) or, as
quoted above in the synopsis of chapter 2, differentiation
between languages is said to result from the interaction
between two modules, "the lexicon and the grammar" (pp.51-
52). However, lexicon or grammar each on its own also
appear to define a language: "Sometimes it seems more
appropriate perhaps to speak of one syntax with different
lexica attached to it (the different languages), than of
syntactic systems which have converged." (p.272). Or, in a
discussion of meaning, "social meaning is carried mostly
by the external aspects of language: *lexical choices* and
*pronunciations*." (p.249). Similar reductionist views of
language are common in other areas of linguistic study, e.g.,
(bilingual) child language. Here too, the possession of
lexical equivalents in two languages constitutes the acid
test for the recognition of a child as bilingual, and an
infant who lacks words or grammar is dubbed 'pre-
linguistic'.
PM does refer one study where "the elements that were
flagged (...) were phonologically and morphologically
fully integrated loans" (p.93), and notes that little
attention is generally paid to phonetics, phonology and
intonation in studies on code-mixing (e.g., p.250). But he
also notes that "the kind of material analysed" gives rise
to different perspectives (p.16). This has to do with the
matter of who is researching what, a matter which is
neither trivial nor can be easily dismissed, and to which
I return below. The pattern that emerges from the studies
reviewed in the book, all taking a lexico-grammatical
approach to language, is that one study, or one set of
studies, states constraints that rule out certain types of
mixing and that apply to the particular pairs of languages
addressed in the studies. Other studies, concerned with
other pairs of languages, either provide evidence for the
occurrence of the 'ruled-out' mixes or, by addressing
different domains of mixing (or by re-labelling domains)
end up, in turn, proposing equally ad-hoc constraints that
fail to be heeded in yet other studies. Although PM
frequently remarks on the scantiness, despite all
appearances, of data on bilingual use (pp.39, 117, 136,
138, 220, 227, 249), his exasperation with this situation
is patent on p.29, where he laments that "(...) we have to
work with natural speech data", or on p.34, where he
suggests that the gathering of bilingual corpora "has
reached the limits of its usefulness" and proposes a shift
of focus to, e.g., experimental techniques. I beg to
disagree: the fact that technology exists, or may one day
exist, enabling researchers to work with unnatural speech
data from a purely lexico-grammatical perspective does not
in itself guarantee any more accurate insight into the
workings of code-mixing than the one apparent in the book.
I do agree, however, that a shift of focus is acutely
needed: since no-one speaks (or babbles) without speech
sounds or melody, and since prosody plays a crucial role
in the coding and decoding of speech, it is my conviction
that attention to what a bilingual 'sounds like' cannot
but shed fresh light on the mechanisms of code-mixing.
2) Bilinguals are not monolinguals gone awry.
One of the major merits of PM's book is that it sets right
the old (but perhaps not so outdated) prejudice that
mixers are deficient language users. Theories assuming a
fundamental asymmetry between the two languages of a
bilingual may, however unwittingly, fuel such prejudice.
In virtually all literature on code-mixing, one of the
languages is taken as the core language of an utterance,
upon which the other intrudes. The task of the analyst is
then to characterise which construction or feature of the
one language is intruding upon the other.
A recurrent theme throughout the book concerns in fact the
characterisation of the core language of the utterance,
particularly the (in)validity of frameworks that have so
far been proposed to distinguish core from intruding
language, as well as elaborate criteria for the assignment
of mixed stretches of speech to one or the other of the
languages. In this spirit, it is not surprising that PM
attempts the formulation of a typology of code-mixing from
the perspective of a theory that avows its concern with an
ideal, monolingual, speaker-hearer of an ideally
homogeneous language, namely, generativism, whose
framework is acknowledged in the Preface (p.xi). Although
PM candidly discusses the problems raised by this
theoretical choice throughout the book (one example is
the application of grammatical government to mixes, on
pp.19ff.), the articulation of the theory with the factual
bilingual data reported in the book nevertheless leads to
a number of baffling statements on the nature of code-mixing.
For example, the reason PM gives for this particular research
field having "become so confusing because everyone
proposing constraints is right as well as wrong" is that
"code-mixing is impossible in principle, but (...) there
are numerous ways that this fundamental impossibility can
be circumvented." (p.30). This is tautological, in that
code-mixing is of course "impossible" from the perspective
of a monolingual theory of language. PM nevertheless
appears to suggest that the role of grammar is that of
"circumventing" (providing with "escape hatches" is
another formulation used, p.30) the facts of language
whose fundamental impossibility is assumed by the same
grammar. Another example is: "If you cannot have
equivalence, adopt another form, by bending the rules of
the systems a bit." (p.32). The "escape hatches" appear to
me suspiciously like the small print of insurance
policies, which more often than not turns out to describe
the reason why one takes out insurance in the first place.
Provided with escape hatches, "we can imagine there to be
various strategies to neutralize mixing and make it less
offensive." (p.30). Not to mention the paradox of
attempting a typology of code-mixing by "neutralising" the
object of research, the choice of words in this discussion
is, at best, unsettling: speakers offend by coming up with
uses of language that are impossible according to what can
only be interpreted as grammatical prescription. If a
grammar is a statement of possibilities and not a
Cinderella slipper to be forced on Ugly Sisters' feet for
which it was not designed, one is left to wonder which
facts of bilingual usage are, then, deemed possible at all
within the adopted theoretical framework. I return below
to the issue of a 'bilingual' grammar, but meanwhile the
answer appears to lean towards constraints that are
instead one-size-fits-all: PM is the first to admit that,
of the three proposed strategies, only insertion is
amenable to a generative account, insertion being also the
more primitive (my word, MCF) type of strategy and the one
where a clear "primacy of one language over another" may
be detected (p.249). PM also makes it quite clear that the
structural interpretations given as diagnostic for each
strategy are not watertight, and that all three jointly
contribute to the characterisation of bilingual speech
samples. He also discusses several examples of overlap or
indeterminacy among the strategies throughout the book. PM
nevertheless subsumes insertional code-mixing under "a
[very] general constraint on insertion in syntax" that
concludes chapter 3, formalised as:
(74) * [x' ... X ...Y...], where Y is a sister of X but
cannot be licensed by X.
This constraint is paraphrased as "Constituent _Y_ cannot
be licensed by _X_ if it does not have the appropriate
features" (p.95). After the rich (and challengingly
contradictory) data discussed in equally rich detail in
this chapter, this constraint appears as particularly
un-illuminating.
It is also true that a large portion of the literature on
bilingualism, accounted for in the book, concerns
immigrant bilingualism, an instance of bilingualism that
appears particularly suited to an analysis assuming a
'first' vs. a 'second' language. The assumption is
methodologically sound, but it entails the risk of taking
method for goal by taking the structures and features of
the language assumed as core for the core of bilingual
uses. Bilingual usage cannot surely be modelled from the
perspective of features apparent in one particular
language, as little as usage in any language can be
modelled on usage in another. This un-stated view of
bilingualism as essentially monolingual-based is
nevertheless pervading in the literature. One language,
that somehow should not be there, encroaches upon another:
bilingualism is not *bi*lingualism but a disruption of
*mono*lingualism. Variously, along the book, PM takes care
to dissociate himself from this stance, e.g., by stating
that "...insertional mixing is unidirectional and involves
a matrix/non-matrix asymmetry, while alternation mixing is
bi-directional." (p.99). In this sense, chapter 9 and its
discussion of simultaneous access to both languages is
probably the most illuminating. I, for one, find it quite
absurd to even think of a possibility of code-mixing if
only one language is accessed. The parallel to be drawn
seems to me to be that bilinguals have two languages like
human beings have two ears, not like we have two hands
where one comes to the rescue when the dominant one needs
help. PM discusses examples of bilingual use that appear
paradoxical "unless one admits of the possibility of
simultaneous access to components of the two languages,
something incompatible with most current analyses of code-
mixing as an off/on phenomenon" (p.254). The 'paradox'
lies of course in that the analytical assumptions are
incompatible with the data, not in the data. No theory of
human hearing assumes an on/off phenomenon, and I see no
reason why a theory of bilingualism should.
3) A language is not language.
A clear approach to bilingualism on its own, bilingual,
terms is, however, not entertained throughout the book.
This is apparent in formulations where bilingual
constructions are said to involve, e.g., "_foreign_ main
verb + _native_ helping verb" (p. 33, emphasis added,
MCF), or where bilinguals are said to "reach across into
the other language to find something equivalent" (p.35).
Rather, bilinguals reach into language, then find which
language best suits the purpose of their utterance. Lest
it appears that I am playing with words, I hasten to add
that I have to write this review in English, whose
vocabulary is particularly inadequate for a distinction
that I deem crucial for our understanding of bilingualism.
As is well known, researchers tend to report findings
that correlate with their own preferred materials
or with their own earlier research. PM is well aware of
this claim (p.10), and readily admits that his research
concerns largely Dutch, and mixes involving this language.
(Much of the available literature on bilingualism in fact
concerns mixes involving two Germanic languages, either
English or Dutch, or both - this is patent, for example,
in the whole of chapter 5.) My only disagreement with PM's
statement of this matter is that I think the claim
constitutes a healthy report on observed facts, and need
not be made "maliciously" (p.10). The claim is also true,
and I believe crucially so, of the language(s) that
researchers speak. More linguistic descriptions, as well
as more detailed ones, are available for English than for
any other language. In addition, the bulk of linguistic
thinking is published (and often worked out too) in
English, regardless of which language constitutes the
object of study. The risk here then is to mistake any
findings about English for findings about language, and
English terminology for a metalanguage in which to
encapsulate them. I believe that most of the confusion
that PM reports in the field of code-mixing stems from a
fundamental mix-up (no pun intended) in terminology,
current in other areas of linguistic study too. The
problem concerns the word 'language' which, in English,
means at least two quite different things: English is a
'language', and the capacity for speaking it is also a
capacity for 'language'. By way of a lexical idiosyncrasy,
one particular tongue (Saussure's 'langue') is often
(mis)taken for the ability to use tongues - or to speak in
them (Saussure's 'langage'). Surely the (universal)
grammar that certain linguists strive to find concerns the
capacity for 'langage' (Saussure's "faculte' de langage"),
not the ability to speak one particular manifestation of
it, be it English or Quechua.
Interestingly, PM points out just one such mix-up, in his
comment to a quotation from Chomsky: "there is no
theoretical need to adopt an unmixed perspective on
I-language, as Chomsky appears to do (perhaps unconsciously
importing thinking in terms of E-language into the
I-language domain)." (p.42). Here, PM seems to want to make
the point that particular languages correspond to
different E-languages for a common I-language. However, he
appears to immediately fall into the same trap as Chomsky
does: he starts out by commenting that "languages are seen
as fortresses", in current linguistics, "buttressed from
such opposite sides [E-language and I-language]" (pp.42-
43). PM himself would like to argue "that the fortress may
be built on quicksand, in that the two ways the fortress
is constructed do not always correspond. In fact, there
are many cases where the perfect matching between E- and
I-language breaks down." (p.43). The "perfect matching"
implies that each particular language has a E-language
*and* a I-language of its own, which is inconsistent with
his comment on the unwarranted assumption of I-language as
unmixed.
The concept of I-language is one of the most nebulous in
generative theory. As far as I am able to interpret it,
"I-language" is used to refer both to what a child
acquires (i.e., a "langue") and to knowledge of language
(i.e., "langage") internalised by speakers. For the
concept to stand up to scrutiny, it must clearly apply to
bilinguals too (including primary
bilinguals/multilinguals, who acquire several languages
from birth). The description of our knowledge of language
must then include a statement not only of the capacity for
variation among particular languages (as parametric
grammars attempt to do) but also of the variation that
transcends the tongue-bound, on/off phenomenon typical of
parametric approaches and that is present in much
bilingual speech. PM points out throughout the book that
fluent bilinguals tend to adopt the strategy of congruent
lexicalisation in their mixes (e.g., p.247), the strategy
which is least amenable to accommodation, if at all,
within proposed analytical frameworks (see chapter 5,
particularly pp.129ff.). Congruent lexicalisation is the
wild card, as it were, stepping in to account for the
innovative mixing typical of fluent bilinguals, otherwise
unaccountable for because it spills over the "fortress"
walls of particular languages.
If I-language corresponds to the (bilingual) child's
'initial state' and concerns the rules of universal
grammar (UG) from which the child draws the rules of
particular languages, then assigning tongue-labels to
I-language is a contradiction in terms. (I am deliberately
using child language analogies here, not only because
generativism grounds its theory of language on issues of
language acquisition, but also because I believe that
bilingualism, and in particular primary bilingualism,
should constitute the prime field of any investigation
into any linguistic universals worthy of that name. If
anywhere, the rules of an assumed UG are to be found in
multilingual speech, where they are particularly at
freedom to manifest themselves in human beings who are
exposed to different languages from birth.) I-language
is, by definition, mixed because UG is no tongue-grammar,
or it is not universal. Nor can UG be constrained by
idiosyncratic details of structure that characterise
individual tongues. If PM feels justified in positing a
strategy such as congruent lexicalisation where
"[b]asically, anything goes" (p.128), it is clear that
first, he is working from the perspective of one single
language and, second, he is at the same time aware that a
single-language perspective on code-mixing is patently
unsuitable. The same two conclusions can be drawn from his
mention of the "bewildering variety" (p.250) found in
bilingual usage. From within the perspective of UG, no
language uses can be "bewildering".
By the same token, if UG is innate, one is hard put to
defend the view that acquiring more than one language is
synonymous with the psychological stress or the conflict
among particular languages usually assigned to the mind of
the bilingual speaker - not to mention the fact of
assigning perpetual mental anxiety to the majority of the
world's population, which consists of bilinguals and
multilinguals. That which, from a monolingual's
perspective, surfaces as "mixing" of particular "codes" is
in my view the result of exploration of the accidental
limits within which each particular tongue happens to
vary. The act of exploring need not be fraught with
conflict, although it certainly poses a challenge, often a
pleasurable one. PM remarks, as early as on page 1, on
"the essential enrichment of having several grammars and
lexicons" participate in linguistic communication and,
further, that code-mixing is most evident in teenagers
(p.227). This is indeed the traditionally attested age for
testing and attempting to break the limits of any imposed
codes, and language-codes are no exception. Bilinguals of
any age will make use of whatever (universal) linguistic
tools are at their disposal, in whatever language, like
any normally gifted human being. From this perspective, it
may well turn out that bilinguals do not "mix" at all.
One last remark: PM's book is a most welcome assessment
of the challenges facing research in code-mixing and
bilingualism. By systematising for us the findings
about code-mixing spanning the past decade and a
half, the book makes a timely statement that we
are still toddling to capture bilingualism in a way that
does no violence to it.
About the reviewer: Madalena Cruz-Ferreira teaches
phonetics, phonology, morphology and general linguistics
at the National University of Singapore. Her research
interests include prosody, bilingual child language
acquisition and Portuguese linguistics.
|
