Cue integration in the perception of fricative-vowel coarticulation in Korean

Abstract

Korean distinguishes two fricatives—fortis [s’] and non-fortis [s]. Perception of this distinction was tested in two different vowel contexts, with three types of stimuli (consonant-only, vowel-only, or consonant-vowel sequences) (Experiment 1). The relative contribution of consonantal and vocalic cues was also examined with cross-spliced stimuli (Experiment 2). Listeners’ weighting of 7 perceptual cues—spectral mean (initial 60%, final 40%), vowel duration, H1-H2* (onset, mid), and cepstral peak prominence (onset, mid)—was examined. The data demonstrate that identification performance was heavily influenced by vowel context and listener performance was more accurate in the /a/ vowel context than in the /i/ vowel context. In addition, the type of stimulus presented changed the perceptual cue weighting. When presented with conflicting cues, listener decisions were driven by the vocalic cues in the /a/ vowel context. These results suggest that perceptual cues associated with breathy phonation are the primary cues for fricative identification in Korean.