The perception of Mandarin Chinese vowels by native English speakers: categorization, identification and directional asymmetry

Abstract This study investigates the perception of six Mandarin Chinese vowels (/ii, u, y, ɤ, /i/ɹ̪, /i/ɻ) in dental, retroflex, and palatal fricative and affricate contexts by adult New Zealand English native speakers, within the frameworks of both the Perceptual Assimilation Model (PAM) and the Natural Referent Vowel (NRV) model. Perceptual categorization and identification experiments revealed four PAM assimilation types: Two-Category (TC), Single-Category (SC), Uncategorized-Categorized (UC), and Uncategorized-Uncategorized (UU). These assimilation patterns subsequently influenced learners' perception of Mandarin vowels, consistent with the predictions of PAM, thereby providing an explanatory account of the specific difficulties encountered by English learners. Two novel cross-language mappings emerged: Mandarin /y/ assimilated to English /u/, whereas Mandarin /ɤ/ perceived as a distinct vowel without a close counterpart in English. Directional asymmetries are evident in the /y-u/ (SC) and /ɤ-ɹ̪/ (UU) contrasts for inexperienced learners only, with peripheral vowels favoured as NRV predicts. In experienced learners, L1-tuned boundaries overrode this universal bias. The results confirm that SC and UU assimilations trigger early phonetic reliance and NRV asymmetry, which L2 experience later suppresses, integrating universal and language-specific forces within a single predictive model. The findings are also pedagogically meaningful, highlighting the need for targeted perceptual training to address systematic perceptual biases for Mandarin vowels /y/, /i/ɹ̪, and /i/ɻ.