7th International Conference on Spoken Language Processing
September 16-20, 2002
This study examined how differently native speakers of Japanese discriminated between the American English vowels /E/ and /æ/ in /CVd/ syllables with 23 different initial consonants, and how the differing discrimination was related to the changes of formant patterns throughout the vowels in comparison with the Japanese vowels /e/ and /a/. A perceptual test and formant analyses of the English and Japanese vowels were conducted. The results showed that there were significant differences in discrimination across the consonantal contexts: the discrimination of /E/ was significantly poorer when the initial consonant was /tS/, /T/, /dZ/, /j/, /g/, /k/ or /ð/, and that of /æ/, when it was /dZ/, /S/, /d/, /n/ or /t/. It was found that the poorer discrimination was more related to the smaller values of the formant ratios (F2/F1) of /E/ and to the larger values of the formant ratios of /æ/ in the latter part of the vowels. The acoustical closeness between the two English vowels in the latter part of the vowels could be attributed to the poorer discrimination in the particular consonantal contexts.
Bibliographic reference. Joto, Akiyo / Imaishi, Motohisa / Nagase, Yoshiki / Funatsu, Seiya (2002): "Discrimination of English vowels in consonantal contexts by native speakers of Japanese and its relations to dynamic information of formants", In ICSLP-2002, 1133-1136.