The Production of Code-Mixed Discourse

The production of code-mixed discourse David SANKOFF Centre de recherches math~matiques, Universit@ de Montr@al CP 6128 Succursale Centre-Ville Montr@al, Qudbec H3C 3J7 sankoff@ere.umontreal.ca Abstract We propose a comprehensive theory of code- mixed discourse, encompassing equivalence- point and insertional code-switching, palin- dromic constructions and lexical borrowing. The starting point is a production model of code-switching accounting for empirical observations about switch-point distribution (the equivalence constraint), well-formedness of monolingual fragments, conservation of con- stituent structure and lack of constraint be- tween successive switch points, without invok- ing any "code-switching grammar". Code- switched sentence production makes alternate reference to two virtual monolingual sentences, one in each language, and is based on conser- vative conditions on language labeling of con- stituents, together with a constraint against real-time "look-ahead" from one code-switch to the next. Selective weakening of model condi- tions can produce (i) the type of palindromic (or portmanteau) construction occasionally oc- curring e.g., in switches between prepositional and postpositional languages, (ii) the switch- ing by "insertion" of very specific kinds of con- stituent reported e.g., for French noun phrases in switching with Arabic and, most important, (iii) lexical borrowing. Borrowing can create ambiguity as to language membership of sen- tence items, but the model predicts where this can be resolved, and the confirmation of these predictions, based on empirical studies of inflec- tional morphology, validates key aspects of the model. Introduction Communities of bilinguals tend to evolve a con- versational mode where elements of both lan- guages appear in the same interaction and even in the same sentence despite the fact that all participants may be competent in either of the two languages. Whether this mode is used in preference to monolingual discourse depends on the type of interaction, the participants, the subject of conversation and many other fac- tors. The grammatical nature of code-mixed discourse, however, tends to be very specific to the community and varies widely among bilingual communities, even among communi- ties which share the same pair of languages. Empirical research has isolated four clearly dis- tinct processes which may be responsible for mixing to different extents in different commu- nities -- code-switching, nonce borrowing, spe- cialized incorporation and interference. None of these processes requires the deformation, al- teration or convergence of either of the two con- stituent languages at the syntactic, lexical, mor- phological, phonological, or semantic levels at the moment the mixing occurs. Except for code- switching, however, they may all lead in the long term to lexical expansion in one or both of the languages. This paper is a contribution to a coherent for- mal account of code-mixing which integrates all of these processes 1, though only code-switching and borrowing will be considered here. This is based on a series of empirical studies which now allows us to distinguish between them structurally and quantitatively. Our starting point will be a recent formal characterization of a equivalence-point code-switching (Sankoff, 1The analysis of code-mixing is a controversial sub- ject with respect to several aspects: Is all code-mixing -- borrowing, …

Content truncated to 500 words