Bilingual sibilant acoustics in conversational Cantonese-English speech

Abstract

Similar sibilants in different languages can differ in their trajectories, even when other cues fail to differentiate them [Reidy, J. Acoust. Soc. Am. 140, 2518 (2016)]. Whether bilinguals—whose languages influence each other in a multitude of ways—would maintain such a difference in sibilant production across their two languages remains an open question. Using the transcribed and force-aligned (female) half of a new corpus recorded in Vancouver, Canada during 2018–2019, this study compares word-initial, prevocalic Cantonese /s/ productions to English /s/ and /ʃ/ productions in the same environment, by the same set of talkers. Cantonese /s/ is of interest given its variable description in the literature, across measurements, talkers, and vowel environments [Yu, J. Acoust. Soc. Am. 139, 1672 (2016)]. This study addresses how bilinguals produce Cantonese /s/, and whether or not it is acoustically comparable to either of their English voiceless sibilants. Sibilant productions are measured in a variety of ways, for comparison with previous studies. This includes peak ERBN number trajectories, which capture spectro-temporal variation. All comparisons are within-talker. Cross-talker differences in language background are also considered.Similar sibilants in different languages can differ in their trajectories, even when other cues fail to differentiate them [Reidy, J. Acoust. Soc. Am. 140, 2518 (2016)]. Whether bilinguals—whose languages influence each other in a multitude of ways—would maintain such a difference in sibilant production across their two languages remains an open question. Using the transcribed and force-aligned (female) half of a new corpus recorded in Vancouver, Canada during 2018–2019, this study compares word-initial, prevocalic Cantonese /s/ productions to English /s/ and /ʃ/ productions in the same environment, by the same set of talkers. Cantonese /s/ is of interest given its variable description in the literature, across measurements, talkers, and vowel environments [Yu, J. Acoust. Soc. Am. 139, 1672 (2016)]. This study addresses how bilinguals produce Cantonese /s/, and whether or not it is acoustically comparable to either of their English voiceless sibilants. Sibilant productions are measured in a variety of way...