Abstract:

Although it is often recognized that morae per se can be defined both phonetically and phonologically in the literature, it is not well understood how they are involved with each other in Japanese speech processing. Recent off-line studies have suggested that recognition of morae in Japanese is highly involved with duration and syllable position [Otake and Yoneyama, ICSLP 94 3, 1427--1430 (1994); ICPhS 95, 2, 686--689 (1995)]. In the present study, the duration of C1V1 and V2.C3V3 segments in C1V1.C2V2.C3V3 nonsense words where C2 is a nasal with three different speech rates was manipulated in order to see whether C2, an onset nasal, can be recognized as a moraic nasal depending upon preceding and following segmental duration. The materials were presented to 15 Japanese listeners to dictate what they heard in Roman characters. The results showed that the manipulation of C1V1 duration never caused C2 to be recognized as a moraic nasal, while V2C3V3 duration did. The shorter the V2.C3V3 were, the more C2 were recognized as a maraic nasal. This suggests that a moraic duration is not absolute, but relative to following segments. [Work supported by the Japanese Ministry of Education.]