Abstract:This paper explores the effect of Cantonese and Changsha dialect on English lexical stress by measuring the acoustic parameters of pitch, duration and intensity of the disyllabic words produced by the native American, Cantonese and Changsha dialect speakers.The results show that all groups of speakers could use the three acoustic parameters to produce English lexical stress, but with different ways and degree; for the general production of stress and unstressed syllables, the Cantonese group exhibited a significant deviation from the native American group, while the Changsha dialect group displayed more similarities to the native; however, for the production of trochee and iambic, the Changsha dialect group differed significantly from the native in pitch, duration and intensity, and conversely, the Cantonese group were more similar to the native.Therefore, the results indicate the effect of dialect and an obvious prosodic transfer in the production of English lexical stress.This study partially supports Altman's Stress Typology Model of and its prediction in lexical stress acquisition, while it also challenges the model's explanation and predication for production of different stress patterns.