Non-native discrimination across speaking style, modality, and phonetic feature

Discriminating between certain non-native contrasts can be difficult. The Perceptual Assimilation Model [1] predicts that when two non-native phones are assimilated to the same native language category, as equally good or poor versions, discrimination should be poor (a single-category assimilation). However, it is not known to what extent visual and/or clearly articulated speech might assist cross-language speech perception. Monolingual Australian English listeners discriminated two single-category Sindhi consonant contrasts (/ʈ/-/t̪/, /b/-/ɓ/), across auditoryonly (AO) and auditory-visual (AV) conditions, in clear and citation speech. For /b/-/ɓ/ (a laryngeal feature difference), AV contrasts were discriminated more accurately than AO contrasts in citation speech, but not in clear speech, while for /ʈ/-/t̪/ (a place-of-articulation difference) there was AV benefit for clear, but not for citation speech. These results highlight that while perceivers attempt to utilize even subtle gestural differences, speaking style and modality differentially contribute to the success of discriminating across non-native contrasts.