Infants' attentional processes modulate audiovisual word learning

• In Event: Attention, learning and memory in infancy

Abstract

While most studies have explored word learning in highly simplified, audio-only settings, infants’ more usual experience involves complex face-to-face audiovisual (AV) environments. This suggests that infants’ attention capacities are crucial to select the relevant information from the talker’s face—both social and speech cues—to then learn novel word-object associations. Critically, the relationship between selective attention patterns during dyadic interactions (i.e., attention to a talker’s eyes or mouth and gaze following ability) and word learning performance remains to be well understood. More specifically, gaze following – i.e., learning to follow the eye-gaze/head direction towards the gazed and named object – is necessary to disambiguate the target-object, and therefore it may facilitate the subsequent word-object association. Looking at the talker’s mouth can also help better encode and retain the new-word phonemes, and thus could also enhance the new word-object association (Tenenbaum et al., 2014). To explore this question, we recorded 18-, 24- and 36-month-old toddlers’ eye gaze (N=113) while performing an AV word learning task (similar to Yurovsky & Frank, 2015). During the learning phase, a speaker looks at one of two objects (head-turn) and names it three times (object-naming). This action is performed twice. Then, at test, the two objects are presented side by side while the voice of the speaker asks for one of them; first for the named object (test) and then for a new word (mutual exclusivity test). Infants underwent this process three times, and repeated the test phase for the three new words after a 1m pause (retention phase). Group results showed evidence of word learning (i.e., greater looking to the target object in the test and mutual exclusivity phases) in 24- [t(44) = 4.31, p < .001] and 36- [t(33) = 5.83, p < .001] but not 18-month-old infants [t(33) = 1.63, p = .113] (see Figure 1). No evidence of word retention was found at any age [all ps > .1]. Interestingly, analysis of individual differences showed that the stronger predictor of word learning performance was greater looking to the target object during learning [r=.25, p < .01], and CDI receptive vocabulary scores [r=.34, p=.02]. Attention to the mouth of the speaker decreased with age but was not related to word learning nor retention capacity [p > .1]. These results confirm prior research (Yurovsky & Frank., 2015) by showing that word learning capacity from a quick AV interaction emerges between 18 and 24 months of age, and suggests that in this task, attention to the mouth AV speech cues does not help create the object sound-label association, but following gaze and attending more to the named/gazed object does. The implication of these results together with differences in finer-grained time-course analysis across development will be discussed.

Authors

• Joan Birules, University of Grenoble
  Presenting Author

• Olivier Pascalis, Centre National de la Recherche Scientifique (CNRS)

• David Méary, Université Grenoble Alpes

• Mathilde Fort, Université Claude Bernard Lyon 1