LINGUIST List 10.1787

Wed Nov 24 1999

Sum: Code-switching Issue 10.1759

Editor for this issue: Karen Milligan <>


  1. Sophie ALBY, Code-switching - matrix language

Message 1: Code-switching - matrix language

Date: Wed, 24 Nov 1999 09:20:20 -0200
From: Sophie ALBY <>
Subject: Code-switching - matrix language

I've received many answer to my mail (19/11/99) Issue 10.1759:
I am presently doing some doctoral work on french-kali'na (an amerindian
language spoken in French Guiana - Carib family) code-switching.

I am having trouble with the question of the 'base language' of the
sentences where I have occurences of code-mixing.
Ellen F. Prince and Susan Pintzuk (Bilingual Code-Switching and the
Open/Closed Class Distinction - University of Pennsylvania - January 1984)
states that :
"Following Joshi 1983 and others, we used the tensed verb of each tensed S
to determine the matrix language of that S".
Is there a general definition of matrix language that would fit for tensed S
and for other types of sentences ?

Thanks to :
Ellen F. Prince, Johannes Reese, Madalena Cruz-Ferre, Stephen M. Simms,
Rakesh Bhatt, Tej K. Bhatia and Nkonko Kamwangamalu.
Their informations where very useful and I was very pleased to discuss
specific problems with some of them.

Summary : about the examples

1) Ellen F. prince analyses 'a' in 'il y a' and 'man' as tensed verbs, so
according to her the following sentences are tensed S :

il y a owi �cole
exist. d�t. school
"il y a un cocotier"
"there is a school"

� c�t� owi cocotier man
prep. d�t. tree it/there is
"il y a un cocotier � c�t�"
"there is a tree next to something"

About the third example :
il y a trois cocotiers man
exist. num. N it/there is
"there is three trees"

She notices that in the dissertation on Japanes-English code-switching M.
Nishimura reported similar tokens :
S-english V-english O-english V-japanese
S-japanese O-japanese V-japanese O-english
Which she calls a 'mirror-image' form. 
The interresting thing being that the basic word order in kali'na is also SOV.

2) Johannes Reese doubts for the question of a 'matrix language' in
He also analyses 'a'in 'il y a' and 'man' as tensed verbs. So s1 is french
because of 'il y a' and s2 is kali'na because of 'man'. s3 seems to be
bilingual in any sense.

3) Madalena Cruz-Ferre
 This approach seems to imply that any codeswitched/mixed utterance will
have a tensed verb. I take 'codeswitching' and 'mixing' literally, the
former as switching between 'pure' tokens of different languages, and the
latter as a blend of (usually no more than) two languages.
 Bringing tensed verbs or other morphology into the definition doesn't
hold, for example, for the earliest production of bilingually raised
children - where each parent systematically uses only her/his language with
the child. Having faced exactly the same problem of determining the matrix
language (I call it the 'host' language, versus the 'guest' language) in
these cases, I opted for taking as host the language of the ongoing
exchange: if mother and child are speaking in language A, then A is the
host. This accounted also for the cases where prosodic patterns alien to A
were used by the child in exchanges with the same parent. If no "ongoing"
language can be established (say, the child is talking to herself), then I
take as host the language of the first word/phrase in the utterance - this
would be the language the child is currently 'thinking' in?
 So I'd take your three examples:

>1) il y a owi �cole
 2) � c�t� owi cocotier man
 3) il ya trois cocotiers man

as French host.

Summary : references

1) Stephen M. Simms
In reference to your question, C. Myers-Scotton (Duelling languages :
Grammatical structure in codeswitching, 1993) claims that determination of
the ML is not based on Infl at all (p. 66). 

2) Rakesh Bhatt
 It is sometimes insightful to use Tense (Agreement) morphology to determine
the matrix language. This is what I have used in my LINGUA paper (1997:
Code-switching, constraints, and optimal grammars)�basically, the matrix
language is the language of the Infl. Treffers-Daller (1994) also uses a
similar, if not the same, definition. And Myers-Scotton (1993) uses it too,
I believe, as one of her many definitions of Matrix language. This
definition is close to Prince & Pintzuk's definition, but considers
language of Tense (Infl), not tensed verb. The rartionale is simply that
Tense as an operator has scope over the entire proposition (VP), and
therefore must impose some structural requirements in codemixing. 

3) Tej K. Bathia
The following book chapter might interest you. 
Bhatia, Tej and Willilam C. Ritchie. 1996. Bilingual language mixing,
universal grammar, and second language acquisition. In 'Handbook of Second
Langauge Acquisition,' (Ritchie and Bhatia, eds.). San Diego: Academic
Press. Pp 627-688. The section IIA (particularly on p. 631) deals with the
question of a general framework for matrix vs. embedded language in language

4) Nkonko Kamwangamalu
The issue you are raising is an important one, and has been discussed 
in a number of studies (e.g. Joshi 1985, Nishimura 1989, Myers-Scotton
1993, etc.). Following are a couple of references in which I have
addressed this issue with respect to codeswitching in French or English with
Bantu languages or Chinese. Hope you find the references (especially the
first one) useful.

Kamwangamalu, Nkonko M & Lee Cherleng. 1991. Chinese-English 
codemixing: A case of matrix language assignment. Word Englishes 10, 
3: 247-162.

Kamwangamalu, N M 1989a. Codemixing across languages: structure, 
functions, and constraints. Doctoral Dissertation. University of 
Illinois at Urbana-Champaign. Ann Arbor, MI: University Microfilms 

- - 1989b. Some morphosyntactic aspects of French/English-Bantu 
codemixing: evidence for universal constraints. Chicago Linguistic 
Society 25: 157-170.

- 1987. French/Vernacular codemixing in Zaire: Implications for 
syntactic constraints. Chicago Linguistic Society 23: 166-180. 
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue