Time-based integration of Mandarin medial /j/

Abstract This study investigates how listeners integrate dynamic cues to perceive the Mandarin medial glide /j/ over time, and how this process is modulated by speech rate and second-language (L2) background. We tested 30 adults in three groups: Mandarin native speakers, intermediate Japanese learners, and beginning Japanese learners. A female Beijing Mandarin talker produced disyllabic carrier phrases of the form mao + target syllable, containing the four target syllables miao, mian, mia, and yao , at three speech rates (fast, mid, slow). For each token, the first syllable mao was kept intact, while the second syllable was progressively revealed across ten gates spanning the interval from the onset of the second syllable to the early steady-state portion of the vowel. On each trial, participants heard a two-syllable gated sequence and rated how much the second syllable sounded as if it began with an /i/-like quality on a five-point scale. Cumulative-link and linear mixed-effects models revealed robust effects of gate, speech rate, syllable structure, and group. Ratings increased steadily with gate level, were highest in slow speech, and followed a stable gradient of MNS > JP_INT > JP_BEG; CGVX syllables, especially miao , were easier and faster than the GVX syllable yao . The results highlight time-based cue integration as a locus of L2 difficulty, consistent with dynamic cue-integration accounts of speech perception and contemporary models of L2 phonology.