First International Conference on Spoken Language Processing (ICSLP 90)
This study investigated how native speakers of Japanese, who are midway toward acquiring English phonemes, perceive and produce American English /r/, /I/, and /w/ sounds. From the perceptual experiments using synthesized stimuli, the following results are obtained. Japanese listeners perceive /r/ and /I/ sounds non-categorically, and they use both F2 and F3 frequencies as cues to identify /r/ and /I/. In contrast, American listeners perceive these sounds categorically using F3 as the primary cue. This result is consistent with the result obtained from the acoustic analysis of the /r/ and /I/ sounds produced by native Japanese. F3 frequency is a main feature for American subjects to produce /r/ differently from /I/. Such acoustic features, i.e. formant frequencies of the consonant part, are not well differentiated into phoneme categories when Japanese speakers produce these phonemes. However, the tendency to use both F2 and F3 frequencies, even as production cues, is observed in the sounds by Japanese speakers.
Bibliographic reference. Yamada, Reiko A. / Tohkura, Yoh'ichi (1990): "Perception and production of syllable-initial English /r/ and /l/ by native speakers of Japanese", In ICSLP-1990, 757-760.