Making voices count: Corpus compilation in bilingual communities

Corpus compilation is of great relevance in linguistics today, with growing appreciation of studies based on spontaneous speech, in particular for minority communities. This paper puts forward a model for corpus compilation in bilingual communities, illustrated through the New Mexican Spanish-English Bilingual corpus, in which the same speakers-from a long-standing minority community in the United States-use both Spanish and English in the same conversations, smoothly alternating between their languages. We advocate community-based fieldwork for the collection of speech data by community members; the formation of a corpus which comprises recordings of spontaneous interactions and is thus of widespread usability (rather than being tied to any particular set of research questions or elicited linguistic features); scrupulously compiling information about the demography and the linguistic history of the participants that may shape their patterns of language use; and comprehensive transcription of the data taking into account prosodic aspects and making considered decisions about how to responsibly represent the speech. This community-based approach yields linguistic data situated in its social context and amenable to systematic quantitative analysis, which allows for confronting the many claims about language contact with the facts of bilingual usage.