Factors determining code-switching patterns in Spanish-English and Welsh-English communities

Gardner-Chloros (2009) suggests that variation in code-switching (CS) can be linked to extra-linguistic factors which are either community or speaker-specific. We aim to identify which factors account for the CS patterns found in corpora we have collected of Welsh-English and Spanish-English bilingual data. We use the choice of matrix language (ML) within an MLF (Matrix Language Framework, Myers-Scotton 2002) as the dependent variable.
The MLF model posits that one language, the matrix language (ML), is the source of morpho-syntax in bilingual clauses. Analysing bilingual clauses in the speech of six speakers from each corpus, we find that while Welsh is uniformly the ML in clauses produced by Welsh-English bilinguals, the ML in bilingual clauses produced by Spanish-English bilinguals is varied.
A multivariate analysis on the Spanish-English data was conducted to test the relationship between ML distribution and extra-linguistic variables, but no significant relationship was found. However, in comparing the questionnaire data from the two corpora we argue that contrasting community-based norms may account for the difference in uniformity (Welsh-English) vs. diversity (Spanish-English) in the choice of the ML. The uniformity in the choice of the ML in the Welsh-English data can be linked to more homogeneity in self-ascribed identity and in Welsh-oriented social networks. Conversely, the variation in ML in the Spanish-English data may be related to more heterogeneity in identity and in the language of social networks.