FAAVSP - The 1st Joint Conference on
Facial Analysis, Animation, and
Auditory speech is difficult to discern in degraded listening conditions, however the addition of visual speech can improve perception. The Perceptual Assimilation Model  suggests that non-native contrasts involving a native phonological difference (two-category assimilation) should be discriminated more accurately than those involving a phonetic goodness-offit difference (category-goodness assimilation), but it is not known whether auditory-visual (AV) benefit is greater for phonological than phonetic differences when the acoustic signal is degraded by speech-shaped-noise. In auditory-only (AO) and AV conditions, monolingual Australian English participants completed AXB discrimination tasks on twocategory versus category-goodness Sindhi contrasts. We also examined the relative influences of phonetic feature difference (laryngeal vs. place-of-articulation [POA]) and speaking style (clear vs. citation speech) on discrimination accuracy. AV benefit was found for POA contrasts, but no effect of speaking style, and AV benefit was larger for two-category than category-goodness contrasts. For laryngeal contrasts, AV benefit was found for the two-category contrasts (across speaking style), but for the category-goodness contrast only when it was clearly articulated. These results indicate that nonnative perceivers use visual speech to their advantage, and to a greater extent for phonological contrasts, but speaking style contributes in AV conditions only for a less salient phonetic contrast. Index Terms: speaking style, modality, cross-language speech discrimination
Bibliographic reference. Fenwick, Sarah / Davis, Chris / Best, Catherine T. / Tyler, Michael D. (2015): "The effect of modality and speaking style on the discrimination of non-native phonological and phonetic contrasts in noise", In FAAVSP-2015, 67-72.