An E-Learning System of Phrase Accent and Intonation for Learners of Japanese, Based on Phonetic Segment Labels and Fundamental Frequency
音声セグメントラベルと基本周波数を用いた日本語学習者のための文節アクセント・イントネーションのe-Learning システム

Hiroshi Matsuura, Yuri Hojo, Koichi Sawasaki, Junichiro Wada, Shusuke Inukai, Masayuki Hideshima
2016 Japan e-Learning Association Journal  
A speech e-learning system of phrase accent and intonation for learners of Japanese is described in this paper. Utterances by native Japanese speakers and learners of Japanese with regard to the phrase /yubinkyokunara/ (literally, "in the case of the post office") are investigated for developing the system. With the compound word /yubinkyoku/ ("post office"), while the accent is normally placed on the mora /bi/, the current study revealed that a considerable number of informants, including many
more » ... Japanese natives, placed the accent on the syllabic nasal sound /N/ -a "special mora" that is normally not accented. Also, some learners mistakenly accentuate the mora /kyo/ or /ku/, and accentuate on two different morae within the phrase. Moreover, an inadequate degree of variation in the fundamental frequency (F0) was observed in some learners' utterances. Based on the results of the study, thresholds were formulated to serve as standards for determining accent position and the comments for utterance instruction. Finally, an experiment using the e-learning system was conducted, in which phonetic segment labels from speech are used to automatically determine morae, followed by estimates of the F0 for each mora and evaluation of accent and intonation. After instructing using the speech e-learning system, learners with an incorrect accent position were asked to repeat the phrase; in all cases, improvement was seen.
doi:10.32144/jela.16.0_29 fatcat:xxauz27fcbc3hisy6pzey4d2gu