Language contact is a mechanism of language change, yet the study of bilingual speech labors under beliefs that though widely held are still to be verified. Their persistence is linked to the tolerance for isolated counter-examples as a surrogate for proof and the paucity of systematic quantitative analyses of speech corpora.

Drawing on data from spontaneous New Mexican Spanish-English interactions, I address three assumptions. One is that all unattested single other-language items are code-switches. In virtually every bilingual situation, the  majority of other-language material is made up of single words, and these tend to instantaneously assume the morpho-syntax of the language in which they are embedded, i.e. they are borrowed, if only for the nonce. A second widely held inference is that code-switching begets grammatical change. But co-activation of both grammars is not coterminous with grammatical convergence in production, which is seldom satisfactorily demonstrated, even for typologically similar languages, once appropriate benchmarks are employed. A third common belief is that bilinguals’ minority languages are unaffected by social factors operative in monolingual communities. Neither is this a foregone conclusion as, where tested, sociolinguistic patterns generally replicate those generally found, revealing the systematic character of bilingual varieties.  Bilingual borrowing and code-switching patterns point to intact and independent, rather than altered or merged, grammars.

Rena Torres Cacoullos is visiting ANU from Penn State University until 4 February. Rena's work focuses on the identification of quantitative patterns in spontaneous speech and historical texts, using variability to demonstrate grammatical similarities and differences, in bilingual communities and in diachronic grammaticalization processes. She is co-editor of Language Variation and Change.

The talk will be followed by drinks and dinner. All welcome.