Capitalizing on musical rhythm for prosodic training in computer-aided language learning

* Corresponding author: Helen Meng ([email protected]) Abstract This paper presents a system named MusicSpeak, which strives to capitalize on musical rhythm for prosodic training in second language acquisition. The system targets for Chinese (L1) speakers learning English (L2). Their speech rhythms are considered to be syllable-timed and stress-timed respectively. Hence, language transfer creates a challenge for Chinese learners in acquiring English rhythm. We develop an automatic procedure that can be applied to any English sentence, to cast rhythmic patterns in speech (based on alternating stressed and unstressed syllables) into rhythmic patterns in music (based on musical bars and beats). We collected speech recordings from 9 speakers uttering 15 English sentences, first in natural style and then in synchrony with the generated musical rhythm. Comparison between the two styles based on rhythm metrics suggests that the latter has higher variability and better approximates stress-timed rhythm.