Sai (2005) investigated the role of speech in newborns' recognition of their mothers' faces. Her results revealed that, when presented with both their mother's face and that of a stranger, newborns preferred looking at their mother only if she had previously talked to them. The present study attempted to extend these findings to any other faces. By using video films, our results revealed that unfamiliar female faces were recognized in the test phase only if they had previously talked to the baby, but not if they had been silent. These results highlight the importance of an early audiovisual perception which already seems to play an important role in face processing at birth.

For several decades, many authors have claimed the existence, early in life, of a tight link between perceptual and productive systems in speech. However, the question whether this link is acquired or is already present at birth remains open. This study aimed at investigating this question by employing the paradigm of neonatal facial imitation. We compared imitative responses of newborn infants presented either visual‐only, audiovisual congruent, or audiovisual incongruent models. Our results revealed that the newborns imitated significantly more quickly the movements of the model's mouth when this model was audiovisual congruent rather than visual‐only. Moreover, when observing an audiovisual incongruent model, the newborns did not produce imitative behavior. These findings, by highlighting the influence of speech perception on newborns' imitative responses, suggest that the neural architecture for perception–production is already in place at birth. The implications of these results are discussed in terms of a link between language and neonatal imitation, which could represent a precursor of more mature forms of vocal imitation and speech development in general.

Infants demonstrate robust audiovisual (AV) perception, detecting, for example, which visual face matches auditory speech in many paradigms. For simple phonetic segments, like vowels, previous work has assumed developmental stability in AV matching. This study shows dramatic differences in matching performance for different vowels across the first year of life: 3‐, 6‐, and 9‐month‐olds were familiarized for 40 sec with a visual face articulating a vowel in synchrony with auditory presentations of that vowel, but crucially, the mouth of the face was occluded. At test, infants were shown two still photos of the same face without occlusion for 1 min in silence. One face had a static articulatory configuration matching the previously heard vowel, while the other face had a static configuration matching a different vowel. Three auditory vowels were used: /a/, /i/, and /u/. Results suggest that AV matching performance varies according to age and to the familiarized vowel. Interestingly, results are not linked to the frequency of vowels in auditory input, but may instead be related to infants' ability to produce the target vowel. A speculative hypothesis is that vowel production in infancy modulates AV vowel matching.

No abstract