Mariko Kondo (Waseda University): Syllable structure and its acoustic effects on 
vowels in devoicing environments

This paper examines the acoustic characteristics of the high vowels /i/ and 
/u/ in the devoicing environment in relation to Japanese syllable structure. 
Experimental results from studies in various phonetic environments found that 
devoicing rates were not always effectively triggered by phonetic factors. 
However, these results showed that devoicing rates were strongly influenced 
by the phonological environments, namely single or consecutive devoicing 
environments. Experimental results in these two devoicing environments showed 
that there were significant differences in devoicing rates depending on the 
environments. There were also considerable acoustic differences between 
devoiceable vowels that happened to be voiced despite having occurred in 
typical devoicing environments. In the single devoicing environment, voiced 
devoiceable vowels were reduced in duration and weakened in intensity. On the 
other hand, not all devoiceable vowels were shortened nor weakened 
significantly in the consecutive environment. This means that not all vowels in the 
consecutive environment are subject to the devoicing process, but some of 
them naturally remain voiced. 

The patterns of vowel devoicing in consecutive syllables suggested that high 
vowels undergo the devoicing process only when this process does not violate 
the Japanese syllable structure. When a vowel is devoiced and fails to 
maintain a syllable, the consonant preceding the voiceless vowel can 
constitute a mora, but cannot constitute a syllable on its own. In other 
words, the vowel can undergo the devoicing process if resyllabifying the 
preceding consonant to the preceding or following syllable does not violate 
the syllable structure.