The influence of modality and speaking style on the assimilation type and categorization consistency of non-native speech

The Perceptual Assimilation Model [1] proposes that nonnative contrast discrimination accuracy can be predicted by perceptual assimilation type. However, assimilation types have been based just on auditory-only (AO) citation speech. Since auditory-visual (AV) and clear speech can benefit nonnative speech perception [2, 3], we reasoned that modality and speaking style could influence assimilation. This was tested by presenting English monolinguals Sindhi consonants in a categorization task. Results showed that, across speaking styles, consonants were assimilated the same way in AV and AO. For consonants that were uncategorized in visual-only (VO) conditions: 1) their AO counterpart was more consistently categorized than AV; and 2) citation speech was also more consistently categorized than clear. Interestingly, this set of results was reversed for consonants that were assimilated to the same native category across modalities; participants were able to use the visual articulatory information to make more consistent categorization judgments for AV than AO. This was also the case for speaking style: clear speech was more consistently categorized than citation. Together these results demonstrate that the extent to which AV and clear speech is beneficial for cross-language perception may depend on the similarities between the articulatory characteristics of native and non-native consonants.