Looking Patterns Predict Expressive Language in Toddlers with ASD Only If First Words Are Acquired

Background: Infants show shifting patterns of social visual engagement to faces over the first two years of life. Specifically, after an early period of predominant eye-looking, typically developing (TD) infants show increases in mouth-looking, which seem to accelerate around the developmental time-point at which first words are spoken. Past work has provided evidence for associations between this shift and the development of speech and language abilities: in observational studies, attention to visual speech facilitated phonetic learning and language discrimination in TD infants (Teinonen et al., 2008; Weikum et al., 2007), and TD infants were more likely to look to the eyes of a talking face when learning word referents (Lewkowicz et al., 2012). In contrast to TD patterns of social visual engagement, infants later diagnosed with autism spectrum disorder (ASD) show more rapid acceleration in mouth-looking, as well as very early declines in eye looking that persist until 24 months. How patterns of visual fixation in infants with ASD relates to speech and language abilities is currently not well understood.

Objectives: The present study compares eye-tracking measures of social visual engagement with spoken language abilities in the second year of life, an important period in typical language acquisition characterized by marked inflection in word learning. We predicted fluctuation in preferential patterns of visual engagement during this period, with a significant correlation between mouth-looking and expressive language in TD children.

Methods: Chronological age-matched samples of 10-25-month-old ASD (n=56) and TD (n=28) children watched videos of actresses engaged in child-directed caregiving activities. Social visual engagement data were collected using eye-tracking technology and quantified as percentage of time spent visually fixated on regions of interest. Between-group comparisons measured levels of fixation in ASD and TD children. Within-group regression analyses tested for associations between visual fixation and concurrent expressive language levels (measured by the Mullen Scales of Early Learning), controlling for age.

Results: During the second year of life, greater mouth- than eye-looking is observed in both TD (p=0.0001) and ASD (p=0.018) children. However, the adaptive value of these looking patterns differs between groups: increased mouth-looking was positively associated with improved expressive and receptive language scores in TD children (p=0.003 and p=0.022, respectively) but was unrelated to language abilities in ASD children (both p>0.600). If analyses are constrained to ASD children for whom word-learning had begun, a weaker but positive association with expressive language is observed (p = 0.040). In ASD children who had not yet begun word acquisition, no relationship between visual fixations and expressive language was observed.

Conclusions: In the second year of life, increased visual engagement with the mouths of others is adaptive in TD children and positively predicts expressive and receptive language levels. For children with ASD however, only those children who have begun word learning show a positive association between mouth-looking and expressive language; for the remaining children, high levels of mouth-looking are unrelated to expressive or receptive language.