Presentation

Madeleine Rees¹, Abbie Bradshaw², Brechtje Post¹, Matt Davis²; ¹University of Cambridge, ²MRC Cognition and Brain Sciences Unit, Cambridge

Introduction. The relationship between speech production and speech perception is highly debated. In auditory feedback perturbation experiments, the speaker learns to adjust their own feedforward speech motor commands to offset the perceived error (in reality the altered feedback). This process, termed adaptation, suggests that production and perception representations of speech sounds inform and constrain each other. Novel feedback perturbation studies of bilingual participants show that speakers with greater experience in a second language show stronger adaptation responses in that language (Shiller et al., 2021). This result suggests that the interaction between production and perception representations, previously seen in response to feedback perturbation, takes time to develop in a second language. However, it is not known with certainty how adaptation functions in bilinguals nor cross-linguistically, especially in similar (but not identical) vowels, and across dense and sparse vowel spaces. To address this gap, this experiment examines how an individual's magnitudes of adaptation in their dominant language relate to those in their non-dominant (but phonologically similar) language. Methods. Two participant groups of English-dominant (n=12) and Spanish dominant (n=11) English/Spanish bilinguals (approximately C1-level proficiency in non-dominant language) underwent altered auditory feedback tasks in both languages. In the baseline phase, speakers produced 30 unperturbed tokens of the words "head" (English) and "dedo" (Spanish). During the English Hold phase, speakers produced 30 trials of "head", while in the Spanish Hold phase, speakers produced 30 trials of "dedo". In the Hold phase, vowel productions were downshifted by 130 mels in F1 and upshifted by 130 mels in F2 in real time, using Audapter (Cai et al., 2008). Adaptation magnitudes were calculated by projecting F1 and F2 change in each trial onto the vector of perfect opposition (+130 mel in F1 and -130 mel in F2). Preliminary Findings. In the English task, English-dominant bilinguals showed significantly more adaptation to altered feedback than the Spanish-dominant bilinguals (p=0.0127). In the Spanish task, both groups followed the perturbation similarly. Unexpectedly, the Spanish-dominant group showed a significant increase in adaptation in their non-dominant language (p=0.0410). Conclusions and projections. Language dominance is associated positively with adaptation only in English. Vowel space density in the task language appears to negatively influence adaptation, as the perturbation in the sparser Spanish vowel space is less likely to encroach upon a vowel category boundary. Future work will assess production after-effects from feedback perturbation, to ascertain if bilingual speakers share phonological representations in similar vowels cross-linguistically, and can transfer effects of motor learning across languages. Further statistical analysis will investigate potential differences between compensation and adaptation processes by dividing hold phase tokens into early and late time windows. References: Cai, S., Boucek., M, Ghosh, S.S., Guenther, F.H., & Perkell, J.S. (2008). A system for online dynamic perturbation of formant frequencies and results from perturbation of the Mandarin triphthong /iau/. In Proceedings of the 8th Intl. Seminar on Speech Production, Strasbourg, France, Dec. 8-12, 2008. Shiller, D.M., Bobbit, S. & Lametti, D. R. (2021). Sensorimotor adaptation in bilingual speech. SNL 2021 Virtual Edition, 5-8 October 2021.

Topic Areas: Speech Perception, Multilingualism