Proceedings of the Annual Meeting of the Cognitive Science Society

Previous code-switching research (Carter et al., 2010) demonstrates Spanish's tendency to be the matrix language and the involvement of the determiner-noun part of speech (PoS) combination in Spanish-English code-switching. This research, however, primarily uses the spoken Miami Bangor Corpus (MBC), limiting generalizability across speech communities/modalities. We examined the MBC (N=261,711), the spoken Spanish in Texas Corpus, STC (N=416,784), and the written LinCe Corpus, LC (N=278,093) to analyze language directionality and PoS effects across speech communities and modalities. Bootstrap analyses indicate that Spanish was the matrix language at a higher proportion than English for MBC and LC, but English was for STC. Logistic regression analyses show the particle-coordinating conjunction combination was the strongest PoS predictor of a code-switch. These results suggest that corpus modality and speech community both affect matrix language proportions and that previously unconsidered PoS combinations may be involved in code-switching.