While English word accent is linguistically almost the same as Japanese one, the word accent acoustically differs between the two languages. This fact lets us easily suppose that Japanese learners tend to generate the word accent in a manner of not English but Japanese. We propose two methods for automating the detection of the generated word accent and the evaluation of how it is generated. By using context-sensitive HMMs, stressed and unstressed syllables were modeled separately for their structure and for their position in a word. In the matching process, weighting factors were multiplied with several likelihood scores derived from different acoustic parameters. The optimal combination of the factors for the detection can be thought to reflect each speaker's own manner of the accent generation. The analysis of the optimal combination showed different tendencies between Japanese and native speakers, which mainly accorded with findings in previous studies on English teaching.
Cite as: Fujisawa, Y., Minematsu, N., Nakagawa, S. (1998) Evaluation of Japanese manners of generating word accent of English based on a stressed syllable detection technique. Proc. 5th International Conference on Spoken Language Processing (ICSLP 1998), paper 1147, doi: 10.21437/ICSLP.1998-799
@inproceedings{fujisawa98_icslp,
author={Yukiko Fujisawa and Nobuaki Minematsu and Seiichi Nakagawa},
title={{Evaluation of Japanese manners of generating word accent of English based on a stressed syllable detection technique}},
year=1998,
booktitle={Proc. 5th International Conference on Spoken Language Processing (ICSLP 1998)},
pages={paper 1147},
doi={10.21437/ICSLP.1998-799}
}