Backwards and indirect context effects in accommodating gender differences in speech

Abstract

Listeners accommodate large amount of acoustic variability across talkers and phonetic contexts when categorizing speech sounds. This study examines accommodation of gender-related talker differences, which can occur in reverse (in sound sequence AB, sound B affects perception of sound A), suggesting more complicated mechanisms than peripheral auditory contrast enhancement alone. A continuum of fricatives “sh” and “s” was preappended to vowels whose acoustic parameters were manipulated by various properties ranging from typically feminine to masculine and vice versa. Other conditions provided matching or mis-matching visual cues. We examined the weighting of cues to talker gender, the duration of exposure to sound needed to obtain the full context effect, and the influence of hearing status, using listeners with cochlear implants (CIs). CI listeners used different cues to show a comparable gender context effect and also appear to use proxy cues for acoustic properties perceived more clearly by people with normal hearing (NH). All listeners showed implicit knowledge of how acoustic patterns are selectively reverse-compatible with different articulation sequences. NH listeners needed only 30 ms for full adjustment, while CI listeners’ data were much more variable. Collectively, these results show a complex process of auditory, visual and gesture-aware adjustment to a talker’s voice.Listeners accommodate large amount of acoustic variability across talkers and phonetic contexts when categorizing speech sounds. This study examines accommodation of gender-related talker differences, which can occur in reverse (in sound sequence AB, sound B affects perception of sound A), suggesting more complicated mechanisms than peripheral auditory contrast enhancement alone. A continuum of fricatives “sh” and “s” was preappended to vowels whose acoustic parameters were manipulated by various properties ranging from typically feminine to masculine and vice versa. Other conditions provided matching or mis-matching visual cues. We examined the weighting of cues to talker gender, the duration of exposure to sound needed to obtain the full context effect, and the influence of hearing status, using listeners with cochlear implants (CIs). CI listeners used different cues to show a comparable gender context effect and also appear to use proxy cues for acoustic properties perceived more clearly by people with...