Orbicularis Oris Activation in Japanese-English Bilinguals In Code-Switched Environments

Abstract

Previous research into phonetic transfer in code-switching uses acoustic measurements, showing mixed results, with many finding evidence for unidirectional transfer in the direction of an L1 [e.g., Antoniou et al., 2011, J.Phon, 558–570]. However, to our knowledge, there have been no neuromuscular measurements for code-switching. This study investigates the degree of activation of the orbicularis oris (OO) muscle in Japanese [ɯ] and English [u] in a code-switched or non-code-switched environment, with Japanese as an L1 and English as an L2. As a secondary research question, this study investigates how frequency of a word token influences OO activation. Previous literature [Uchihara et al., 2023, Lang. Learn., 84-125] suggests that high-frequency words are processed more quickly, facilitating more efficient production which could lead to less muscle activation. Surface electromyography (EMG) data of OO activation was collected using Delsys EMG Acquisition Software and synced with audio data. Mean normalized activation was then extracted for each vowel. We predicted unidirectional phonetic transfer converging towards L1 values, and less OO activation for high-frequency words. Results indicated no difference in OO activation between code-switched and monolingual environments, or between languages when frequency was not accounted for, which contradicts our hypothesis.  However, word frequency significantly impacted OO activation. High-frequency words in English had more activation than those in Japanese, but low-frequency words in Japanese had more activation than those in English. This suggests word frequency may affect L1 and L2 production differently, and highlights the importance of controlling for frequency in production studies.