Effects of visual speech information on native listener judgments of L2 speech consonants

Research on the intelligibility of non-native (L2) speech productions has focused on native listener judgments of auditorily presented L2 productions. However, little research has explored how visual information in L2 speech productions affects native listeners’ perception. In the present study, native Canadian English perceivers were asked to identify six English phonemes /b, v, s, θ, l, ɹ/ produced by native speakers of Japanese and native speakers of Canadian English as controls under three input modalities: (1) audiovisual (AV), with simultaneous presentation of speaker voice and facial/mouth movements, (2) audio-only (AO), with speaker voice only, and (3) visual-only (VO), with speaker face only. The results show facilitative effects of visual speech information on the intelligibility of non-native productions as well as deteriorative effects due to lack of visible lip-rounding in the Japanese-produced /ɹ/. These results suggest visual speech information may either positively or negatively affect the intelligibility of L2 productions.