Yuyueze Mao, Mengchen Li, Xiao Cai
Syllable perception requires the continuous integration of segmental (consonants and vowels) and suprasegmental (lexical tones) information. While native listeners of tonal languages exhibit pre-attentive integration of these dimensions, it remains unclear whether non-tonal second language (L2) learners can develop such integration through L2 experience. The present study employed electroencephalogram (EEG) to record mismatch negativity (MMN) responses in a passive oddball paradigm, investigating the perceptual integration of consonants, vowels, and tones. Twenty-four native Mandarin listeners (L1 group) and fifteen non-tonal L2 learners of Mandarin (L2 group) were recruited. Stimuli included a standard syllable (/da1/), three single-deviants (/ga1/, /du1/, /da4/), and three double-deviants (/gu1/, /ga4/, /du4/). MMN additivity analyses compared original double-deviant MMN amplitudes against the sum of corresponding single-deviant MMN amplitudes. The L1 group exhibited robust sub-additive MMN patterns across all double-deviant combinations (consonant-vowel [CV], consonant-tone [CT], and vowel-tone [VT]), demonstrating integrated processing across all phonological dimensions. The L2 group showed delayed MMN peak latencies and reduced pre-attentive sensitivity to single consonant and single tone deviations. Crucially, they exhibited additivity for CV and CT combinations (independent processing), but a selective sub-additive trend for the VT combination, which we interpret as preliminary evidence for vowel-tone integration. These findings provide tentative support the TTRACE model and refine the dimensional transfer hypothesis, suggesting that non-tonal L1 biases may be partially overcome through L2 experience when phonological dimensions share tight acoustic proximity within the syllable nucleus (i.e., vowel-tone integration). Non-tonal L2 learners display a developmental pattern rooted in acoustic salience and bottom-up physical binding.