台大心理系

Lexical tone perception is essential for preverbal infants learning a tonal language to encode pitch variation as a lexically contrastive dimension for constructing the lexical meanings of syllables. The perception of native and non-native lexical tones undergoes rapid developmental changes between 6 and 12 months of age; however, the learning mechanisms underpinning these changes remain an open question. While distributional statistical learning supports lexical tone acquisition in infants learning non-tonal languages, and music-to-lexical tone transfer is well-documented in non-tonal language adults, it remains unclear whether early musical exposure can combine with statistical learning to facilitate learning native lexical tones in preverbal infants. Furthermore, whether the acoustic complexity of musical melodies moderates this perceptual learning has yet to be determined. Experiment 1 exposed Mandarin-learning 7- and 11-month-olds (n = 63) to a violin continuum, distributed in either a unimodal or bimodal pattern, and subsequently tested the infants on distinguishing the Mandarin Tone 2 vs. Tone 3 contrast. For both infant age groups, both distributional learning conditions induced significant and equivalent perceptual improvements, indicating that passive musical exposure is sufficient to facilitate tonal development. Experiment 2 exposed Mandarin-learning infants (n = 75) to musical melodies with a mixed-instrument continuum combining violin and erhu to increase acoustic complexity in the distributional learning context. Under high acoustic complexity, results revealed that the music exposure gain was marginally moderated by input distribution, showing an advantage of bimodal distribution over unimodal distribution. Furthermore, 7-month-olds exhibited bimodal distributional learning advantages across both single- and mixed-instrument conditions, whereas 11-month-olds improved comparably across both conditions regardless of distributional inputs. These findings demonstrate that music-to-lexical tone transfer learning is an effective mechanism for infants learning a tonal language, but this cross-domain facilitation is constrained by both the music input distribution pattern and developmental timing.