摘要：This study examines whether the Japanese consonant-glide sequence [Cj]s as in [mjakʉ] ‘pulse’ are complex onsets /CC/ or palatalized consonants /C j / on the basis of duration. We compared the controversial Japanese consonant-cluster [Cj] analysis and uncontroversial Russian palatalized /C j / analysis in duration. The results indicate that Japanese [CjV]s are significantly longer than their [CV] counterparts, whereas Russian [C j V]s are not significantly longer than their [CV] counterparts. Thus, in terms of duration, Japanese [Cj]s resemble consonant clusters /CC/ or /Cj/, and not Russian /C j V/s. On the other hand, No argument seems to exist to support that Japanese [Cj]s are palatalized single consonants. In addition to duration difference, we also assumed that if [Cj] is a consonant cluster, [j] in [C̥jʉC̥] would block [ʉ] devoicing. The results indicate that [j] blocks [ʉ] devoicing to some extent, but this may also be because infrequent morae may be less frequently devoiced. Therefore, the devoicing pattern does not support the Japanese complex onset hypothesis as strongly as the duration patterns.
其他摘要：This study examines whether the Japanese consonant-glide sequence [Cj]s as in [mjakʉ] ‘pulse’ are complex onsets /CC/ or palatalized consonants /C j / on the basis of duration. We compared the controversial Japanese consonant-cluster [Cj] analysis and uncontroversial Russian palatalized /C j / analysis in duration. The results indicate that Japanese [CjV]s are significantly longer than their [CV] counterparts, whereas Russian [C j V]s are not significantly longer than their [CV] counterparts. Thus, in terms of duration, Japanese [Cj]s resemble consonant clusters /CC/ or /Cj/, and not Russian /C j V/s. On the other hand, No argument seems to exist to support that Japanese [Cj]s are palatalized single consonants. In addition to duration difference, we also assumed that if [Cj] is a consonant cluster, [j] in [C̥jʉC̥] would block [ʉ] devoicing. The results indicate that [j] blocks [ʉ] devoicing to some extent, but this may also be because infrequent morae may be less frequently devoiced. Therefore, the devoicing pattern does not support the Japanese complex onset hypothesis as strongly as the duration patterns.
