Effects of bilingualism and language mixing on the development of English and Mandarin Chinese speech perception

• Main Author: Pan, Lei
• Other Authors: Suzy Styles
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2024
• Subjects: Social Sciences; Psychology
• Online Access: https://hdl.handle.net/10356/177366
• Institution: Nanyang Technological University

Speech perception is shaped by linguistic experience from infancy. Previous research on speech perception has mainly focused on monolinguals. Given the limited knowledge about how individuals growing up in multilingual environments resolve phonemic categories in the presence of two-or-more linguistic systems, determining the interaction between early linguistic experience and phoneme perception is crucial for better understanding the patterns of sensory processing in bilinguals. This work investigates how English and Mandarin Chinese phonemic categories are perceived by bilinguals of different ages and different language backgrounds, and whether their perceptual patterns vary if there is a mismatch in the phonemic inventories of their languages, for example when a phonemic category occurs only in one language. Chapter 1 introduces the topic with an overview of the literature on speech perception. Chapter 2 presents a pre-registered phoneme identification task in two languages to investigate phoneme perception of English-Mandarin bilingual adults (pilot: N=15, main: N=66) by measuring the boundary and the steepness of the identity function when perceiving bilabial consonants /b/ and /p/ in each language. Both English and Mandarin have /b/ and /p/ in their phoneme inventories but their acoustic realizations are different, providing a controlled test of language-specific perception for mutually contrastive phoneme pairs. The perceptual boundaries for English fell significantly earlier than the perceptual boundaries for Mandarin, which was in line with previously documented acoustic distinctions in adult production for the local varieties of both languages. In addition, bilingual balance had a significant impact on the slope of the fitted identification curve. Hence, the ‘Slope Transfer Model’ proposes that bilingual children might tune their acoustic sensitivity to the statistical structure of one language, and then transfer that model of phoneme differences to a new boundary location for another language. Chapter 3 presents a study of 81 bilingual children (age 7 to 9 years) from the same speech community, tested with a shortened version of the same task, to determine whether children exhibit adult-like phoneme perception, or are still in the process of ‘slope transfer’. Contrary to the pre-registered predictions, children had a discrete phonemic representation for each language, suggesting the ‘Slope Transfer Model’ does not account for the developmental trajectory. Test-language is a significant predictor of VOT boundary and slope, with earlier boundaries and steeper slopes for phoneme identification task in English than in Mandarin. Unlike in adults, children’s bilingual balance did not have a significant impact on boundary or slope at the preregistered alpha level. Hence, children’s phoneme identification of stop consonant shows a different pattern to adults from the same speech community, suggesting perceptual development is ongoing in this age range. Previous studies do not address language-specific representations of acoustic space where phonemic categories differ between a bilingual’s languages. One target for investigation is the high front rounded vowel /y/, which occurs in only one of Singapore’s official languages (Mandarin Chinese) but not the others (English, Malay, Tamil). Chapter 4 presents an acoustic study of the high front rounded vowel /y/ in Singapore Mandarin, to establish sensory targets for use in further studies of phoneme perception. A speech elicitation tool was designed, lexical variables for target words (featuring /y/) and distractors (not featuring /y/) were collated, and a corpus of picture description in casual ‘every-day’ speech was recorded from 50 young adult bilinguals. Analysis of the dynamic formant trajectories reveals that /y/ in Singapore Mandarin is more dispersed than in other varieties, but acoustically different from the high front unrounded vowel /i/, with some variations in pronunciation by speakers of different gender and different degree of bilingualism. Based on the acoustic targets established in Chapter 4, Chapter 5 presents a study of perception of the high front rounded vowel in two groups of bilinguals in Singapore: 76 English-Mandarin bilinguals for whom /i/ and /y/ are separate phonemes, and 51 English-Malay bilinguals for whom only /i/ is phonemic. This study is a strict test of whether specific linguistic experience is necessary for native-like phoneme identification, over and above being a bilingual familiar with the speech sound – albeit in a language one does not speak. In a phoneme identification task, the two bilingual groups showed different perceptual gradients across the /i/-/y/ vowel continuum, in line with the differences in their early linguistic experience: Bilinguals with early exposure to /i/ and /y/ phonemes in Mandarin developed a stronger perceptual gradient than bilinguals who had heard these sounds spoken by others around them over development, but for whom the sounds are not phonemic in their languages. Overall, findings of the work support a model for bilingual phoneme perception in which early exposure shapes perceptual pattern of bilinguals with discrete representation for their two languages, and retrospective reports of child language exposure are powerful predictors of perception into adulthood.