Psychophysical boundary for categorization of voiced–voiceless stop consonants in native Japanese speakers

Purpose: The purpose of this study was to investigate the psychophysical boundary used for categorization of voiced–voiceless stop consonants in native Japanese speakers. Method: Twelve native Japanese speakers participated in the experiment. The stimuli were synthetic stop consonant–vowel stimuli varying in voice onset time (VOT) with manipulation of the amplitude of the initial noise portion and the first formant (F1) frequency of the periodic portion. There were 3 tasks, namely, speech identification to either/d/ or/t/, detection of the noise portion, and simultaneity judgment of onsets of the noise and periodic portions. Results: The VOT boundaries of/d/–/t/ were close to the shortest VOT values that allowed for detection of the noise portion but not to those for perceived nonsimultaneity of the noise and periodic portions. The slopes of noise detection functions along VOT were as sharp as those of voiced– voiceless identification functions. In addition, the effects of manipulating the amplitude of the noise portion and the F1 frequency of the periodic portion on the detection of the noise portion were similar to those on voiced–voiceless identification. Conclusion: The psychophysical boundary of perception of the initial noise portion masked by the following periodic portion may be used for voiced–voiceless categorization by Japanese speakers.