Vowel hyperarticulation in Mandarin foreigner-directed speech

Abstract

When speakers modify their speech to accommodate the listener's needs, they use a special speaking style called clear speech (CS). If CS is a listener-oriented speech style, we might expect to see granularity in CS modifications based on the degree of accommodation listeners require. I tested this prediction by examining how L1 Mandarin speakers adjusted their corner vowel productions when addressing high- and low-proficiency non-native listeners. Nineteen L1 speakers completed a word-list reading task. For the analysis, speakers' productions were compared across blocks (native-directed speech, NDS; foreigner-directed speech high-proficiency, FDS-high; FDS-low) using linear mixed effects models, with a separate model for each vowel (/a/, /i/, /u/) and formant (F1, F2) pair. Across models, block was the strongest predictor of acoustic variation. From NDS to FDS, /a/ lowering and backing, /i/ fronting, and /u/ raising and backing were significant. Together, these adjustments created larger vowel spaces. Speakers also significantly reduced F1 of /u/ and F2 of /a/ and /u/ when addressing low-proficiency listeners relative to high-proficiency listeners. Analyses of two-way interactions with block revealed that prior experience interacting with non-native Mandarin interlocutors and block order influenced the degree of speech adjustments speakers made. Altogether, the findings suggest that speakers are sensitive to the different magnitudes of speech accommodation that different listeners require and adjust their speech accordingly.