The Identification of Non-Native Speech Sounds with Psychophysical Training
Many early studies of speech perception assumed that the psychophysical properties of speech sounds were unavailable during stimulus identification. When participants are presented stimuli along an acoustic continuum and perform a 2AFC, identification performance is well-described by a logistic function suggesting two discrete phonemic categories. These phonemic category boundaries are believed to bias the classification of speech sounds from non-native languages, reducing the ability to detect acoustic differences. Using a brief period of psychophysical training and a phoneme identification task, participants were sensitized to differences along a voice-onset time (/b-p/) continuum enabling the identification of a non-native phoneme (/ph/). Finally, participant confidence reports suggested that they were generally unaware of their capacity to accomplish the task. Results demonstrate that participants could use psychophysical properties of the stimuli to identify non-native speech sounds while subjective confidence reports indicated that they had varying degrees of awareness of the psychophysical properties of the non-native speech sounds.