The perception–production link in L2 vowel imitation by Mandarin learners of English

This study investigates the perception–production relationship in second-language (L2) English vowel contrasts, /i, ɪ/ (b e a t–b i t) and /æ, ɛ/ (b a t–b e t), in Mandarin learners of English. Using vowel continua derived from natural tokens that independently manipulated spectral and duration properties, we examined group- and individual-level performance in perceptual categorization and phonetic imitation tasks, allowing us to assess how multiple phonetic dimensions are differentially weighted and linked across perception and production. In perception, Mandarin learners primarily relied on duration cues for the /i, ɪ/ contrast, but on spectral cues for the /æ, ɛ/ contrast. Correspondingly, individual differences in perceptual cue weighting predicted production distinctness along the primary phonetic dimension for each contrast, revealing a duration-based perception–production link for /i, ɪ/ and a spectral-based link for /æ, ɛ/. In the imitation of acoustically ambiguous vowels, learners showed spectral assimilation toward the /i/ category for the /i, ɪ/ contrast, indicating strong first-language (L1) influence from Mandarin /i/, whereas no reliable group-level assimilation effect was observed for /æ, ɛ/. At the individual level, assimilation of ambiguous vowels was correlated with perceptual bias for the /æ, ɛ/ contrast, but not for /i, ɪ/. Collectively, these findings suggest that L2 perception and production are linked at the representational level in a dimension- and contrast-specific manner, together shaped by phonetic cue weighting and L1 influence.