In this study, we used eye-tracking to investigate selective visual attention paid to congruent and incongruent audiovisual speech of Japanese infants (6 to 12 months old) and adults. Infants' receptive and expressive language abilities at 12 months were measured through a questionnaire completed by their caregivers. We found that 6-month-olds looked at the mouth longer in the audiovisual congruent condition than in the incongruent condition, whereas 12-month-olds did not show any significant differences in time spent in looking at the mouth. Furthermore, the time spent looking at the mouth among 6-month-olds, in both the audiovisual congruent and incongruent conditions, was positively correlated with receptive language abilities with 12 month olds. These findings suggest that sensitivity to congruent and incongruent audiovisual speech changes during the first year, and attention to the speaker's mouth at 6 months are good predictors of later receptive language abilities.

Key words: human infants, audiovisual speech perception, language acquisition, eye-tracking