| This dissertation discusses the role of phonetic cues and warping in perceptual learning. The experiments described herein use a two dimensional stimulus set derived from the Polish alveopalatal ∼ retroflex sibilant contrast with one dimension varying in fricative noise cues and the other formant transition information. In the first set of experiments English listeners demonstrated that they largely assimilate the contrast to their native palato-alveolar fricative, but can detect some differences, primarily using vowel formant transition information. A second set of experiments confirmed that English listeners do not use the full range of cues available to make the distinction in the stimuli. However, another language group, Mandarin listeners, who have a similar contrast natively, do use both fricative noise and vocalic information to categorize these stimuli. Moreover, Mandarin listeners show evidence of cue-integration that English listeners do not.; A final set of experiments used a selective attention training paradigm to explore English listeners' perceptual space for these sounds. It was found that these listeners could indeed use the fricative noise cues for categorization if sufficiently trained and that listeners could be trained to use both cues separately to categorize the stimulus space. In these experiments listeners showed an effect of heightened sensitivity to differences in the trained dimension of categorization without any corresponding loss of sensitivity to differences in the other dimension. These results are discussed with respect to theories of perceptual learning and cue acquisition. |
