Production and perception of American English vowels by native Japanese speakers

Abstract

Native speakers of Japanese (NJ) produced American English vowels in /CVC/ contexts. Twelve native speakers of American English listened to and identified the vowels in the Japanese speakers’ productions. In general, the vowels difficult for the NJ subjects to disambiguate in production were also difficult for them to differentiate in perception. An acoustic analysis revealed that the NJ speakers’ attempted /i/ and /■/ tokens and /■/ and /■/ tokens largely overlapped in the vowel space, but significant differences in duration were observed between their /i/ and /■/ tokens. These findings agree with the results of our previous studies. In our perception experiment, /■/-/■/ was the most difficult vowel contrast for the same group of NJ speakers to discriminate, and longer /i/ tokens were more correctly identified and shorter /i/ tokens were misidentified more often as /■/. In contrast to the perception experiment, the intelligibility of American English vowels produced by the NJ subjects was not affected by the place of articulation of the preceding consonant or by the voicing status of the following consonant. However, the American English vowels produced by the NJ subjects were less intelligible when a liquid /l/ follows, which may be attributed to their inability to produce /l/ authentically.