Impact of multimodal phonetic training on the acquisition of American English vowels by native Japanese

Abstract

This study investigated the effects of multimodal, computer-assisted vowel training on Japanese learners' recognition of American English vowels. Sixty first-year university students were assigned to either a multimodal training group using JSpace (an interactive vowel-space web tool) or a control group receiving auditory-only identification training. Both groups completed pretests and posttests measuring discrimination accuracy. Results showed significant within-group gains for both groups, although no statistically significant between-group differences were observed at posttest. The experimental group's JSpace selections demonstrated closer alignment with target formant (F1-F2) regions, particularly for the vowels in "kit", "trap", and "strut", whereas the auditory-only group showed stronger gains for high-front vowels and continued difficulty with low and back contrasts such as the vowel in "thought". These findings suggest that structured perceptual training - whether auditory only or multimodal - can support L2 vowel acquisition in computer-assisted language learning (CALL) contexts, with each approach showing distinct strengths.