Abstract:
This paper examines the linguistic patterns of speech disfluencies in Japanese and their relationship with the phonological structure assumed in theoretical linguistics. The investigation focuses on spontaneous speech errors and stuttering produced by native speakers of Japanese in comparison with those produced by English speakers. One striking feature of Japanese speech errors and stuttering concerns the point where words are split as a result of speech disfluency. The loan word paasento ``percent,'' for example, yields pansento in speech errors. In this instance the second half of the long vowel, paa, has been replaced by the coda nasal of the following syllable, sen. This segmentation pattern is crucially different from what happens in English speech errors, where words are split between the consonant(s) and the vowel in the word-initial position: e.g., Jom and Jerry for Tom and Jerry. Essentially the same interlanguage difference in segmentation is found in stuttering: i.e., [se se se sentaa] for sentaa ``center'' in Japanese versus [s s s sent(hooked schwa)] for center in English. This paper attempts to interpret this kind of behavioral difference between Japanese and English in light of the syllable models proposed in the linguistic literature, notably the mora model and the onset-rhyme model. [Work supported by STPF.]