Phonemic contrasts in a mutiple talker task with hearing impaired listeners

Abstract

Word identification ability of mild to moderately hearing impaired listeners was measured in both spatial and nonspatial multiple talker tasks. A digital recording of the modified rhyme test was made for three male talkers. The beginning of the test word in each phrase was synchronized using Cool Edit and equalized to the same average rms power. The phrases were processed for binaural presentation using nonindividualized HRTFs with Tucker Davis Technology System 3 equipment and presented over Sennheisser 600 headphones. The phrases were presented in random order at a self-paced manner. Five mild and five moderately hearing impaired subjects, ranging in age from 24 to 54 years, participated in this study. Results were analyzed by six attributes: place, voicing, nasality, frication, duration, and stops. Phoneme identification performance was ranked based on most salient (most information transferred) to least salient (least information transferred). Spatial separation had no effect on the relative importance of the speech attributes. Duration was the most salient cue for hearing impaired listeners, followed by voicing and frication. Stops and nasality were weak cues and place was by far the least important cue.