音節言語話者における日本語潜在モーラの認識構造 : 広東語母語話者と韓国語母語話者の場合

論文(Article)

Japanese, as a mora language, has implicit moras such as geminates, moraic nasals, diphthings, long vowels and devoiced mora, which syllable-language speakers have difficulty in perceiving. This research investigates the degree of autonomy of implicit moras of Cantonese and Korean speakers and compares their way of recognition of mora with that of Japanese native speakers. Firstly, we observe recognition structure of implicit moras from the viewpoint of phonological characteristics in Cantonese and Korea. As for connection between nucleus and coda, we find that, in Japanese and Korean, the connection is weak and they are easy to disconnect because onset and nucleus are connected first. On the other hand, coda is connected with nucleus first and they are not easy to disconnect. When duration of implicit moras is too long in Japanese and Korean, the languages disconnect nucleus and coda in order that implicit moras are autonomous. Cantonese is an analytic language and syllable constituents cannot be disconnected. Therefore, a syllable is inserted after coda. But in Korean, nucleus is connected with onset, not with coda. Thus, compared with Cantonese in which another syllable has to be inserted, the degree of autonomy of implicit mora correspondent to coda is high in Korean. Secondly, we analyse phonetic characteristics of the languages and investigate a possibility in autonomy of implicit moras. Since Cantonese speakers perceive implicit moras in a word final position as longer than in a word initial or internal position, implicit moras can be easily autonomous. But moraic nasal can often be autonomous in a word internal position. This may be because of an influence of syllabicity of nasal consonants in Cantonese. There is no relationship between recognition of implicit moras and location in syllable in Korean.

Finally, in order to investigate recognition patterns of implicit moras in Japanese which is difficult for syllable-language speakers to perceive, we conducted an experiment of writing Japanese sounds with phonetic equivalent in their own languages. The result shows significance among the factors in the level of 5% an shows the order of recognition according to the first languages as in (1); (1)a. Cantonese speakers: the latter part of diphthongs > moraic nasals > long vowels ≒ geminates b. Korean speakers: the latter part of diphthongs > long voewls > moraic nasals> geminates The result also shows that implicit moras are not more likely to be autonomous in Cantonese than in Korean. Consequently we suggests that, as for recognition of implicit moras in Japanese, Cantonese speakers have more difficulty in perceiving the implicit moras than Korean speakers.