UCLA

This dissertation investigates the prosodic aspects of code-switching in Spanish-Basque bilinguals, which is not currently represented in the bilingualism literature. Bilinguals from the Basque Country in Spain (aged 21-31) from a Basque-dominant region (Lekeitio, n = 3) and a Spanish-dominant region (Bilbao, n = 4) participated in the present study and provided spoken data in unilingual Spanish, Spanish with Basque code-switches, and Basque with Spanish code-switches. The Matrix Language Framework (Myers-Scotton (1993)) predicts that in the presence of language-specific differences, those present in the dominant language of the interaction will override those of the less-dominant language. Spanish and Basque have several prosodic differences, including latency of peak alignment in pitch accents. The current study investigates whether peak alignment differs between unilingual contexts (Spanish only), and code-switched contexts (Spanish code-switching and Basque code-switching). Additionally, previous work on code-switching in Spanish-English communities (e.g. Olson (2012, 2016b) and Fricke et al (2016)) provide evidence that code-switched words may be hyperarticulated or perceived as more prominent than non-switched words. In order to investigate this claim, the f0 maximum (a reliable acoustic correlate of prominence in Spanish and Basque) is measured in code-switched and non-switched words in the present study.

The results of the dissertation provide weak support for the Hyperarticulation account, as higher f0 during code-switching was only found for one group (Lekeitio) during phrase-final positions of one task (Discourse Completion Task). However, there was strong support for the Matrix Language Framework, as there was evidence that the peak alignment patterns of the dominant (or Matrix) language overruled the non-dominant language in code-switching contexts. However, the Matrix Language effects also interacted with speaker language dominance and language usage. Specifically, a speaker’s dominant language was more sensitive to deviation from unilingual norms when embedded in code-switching contexts. Effects of language usage on were also seen, in which the use of Basque socially resulted in more conservative peak alignment production and lower f0 when Basque was the Matrix Language. These results suggest a complex interaction of language context, language dominance, and language switching costs which contribute to variations in prosodic realizations of code-switches.