UCL Working Papers in Linguistics 11 (1999)
Word order in German-English mixed discourse
Intrasententially code-switched data pose an interesting problem for syntactic research as two grammars interact in one utterance. Constituent-based models have been shown to have difficulties accounting for mixing between SVO and SOV languages like English and German. In an analysis of code-switched and monolingual subordinate clauses, I show that code-mixing patterns can be studied productively in terms of a dependency analysis whi c h recognises words but not phrases. That is, each word in a switched dependency satisfies the constraints imposed on it by its own language. Quantitative methodologies, in addition to the dependency analysis, are essential because some of the influences o f code-switching are probabilistic rather than absolute.