From Context to Code-switching: Examining the Interplay of Language Proficiency and Multilingualism in Speech

Multilingual speakers are known to code-switch across language pairs and in association with various paralinguistic aspects of conversation. We build on prior work by asking: how is a multilingual speaker’s language proficiency, as shaped by their linguistic background, related to their code-switching (CSW) behavior? To answer this question, we examine the Bangor Miami corpus of spontaneous Spanish-English speech and analyze the linguistic and demographic profiles of its speakers alongside features of their conversational language production. We find that a speaker’s parents’ primary language, medium of secondary schooling, and self-reported ability are strongly associated with the quantity of CSW and language distribution in code-switched speech. Our work is the first to empirically show a relationship between language proficiency and multilingual speech production and calls for the inclusion of linguistic background features in future paralinguistic studies of CSW.

Recommended citation: Bhattacharya, D., Tolat, A., Hirschberg, J. (2025) From Context to Code-switching: Examining the Interplay of Language Proficiency and Multilingualism in Speech. Proc. Interspeech 2025, 4528-4532, doi: 10.21437/Interspeech.2025-165
Download Paper