Social Visual Attention and Language Development in Autism: Leveraging Infant-Directed Speech and Song to Identify Multimodal Mechanisms of Communication

Language ability is a strong and consistent predictor of outcomes for individuals on the autism spectrum (ASD). ConsequenUy, understanding key predictors and mechanisms underlying positive early language development, and alterations thereof across the variability in language development in autism, is paramount for optimizing the

design, targets, and timing of interventions. In typically developing (TD) and autistic infants, beginning in the second half of the first year of life and continuing into the second year, infants shift their preferential attention from a speaker's eyes to their mouth. Increased visual attention to the mouth during infant-directed speech during

this stage is associated with language development in TD infants, likely because shifting attention to a speaker's mouth enables infants to take advantage of visual cues that complement the auditory speech signal. Multi modal signals, such as the integrated face and voice of an engaging caregiver, draw attention to and facilitate

processing of features that occur across modalities. This is especially beneficial when processing is difficult, suggesting that attention to the mouth should be particularly relevant during initial stages of language acquisition in TD, as well as for children with communication challenges, including autism. We recenUy demonstrated that

the naturally increased multi modality of infant-directed singing relative to speech potentiated TD infants' attention to the mouth of an engaging caregiver, with differential cues (e.g., rhythmic predictability, tempo, audiovisual synchrony, affect) driving mouth-looking at different developmental stages; we extended these findings to ASD

in pilot data for the current project. As well, our team demonstrated that the adaptive value of mouth-looking for expressive language in autism is moderated by children's expressive language level. Building on these findings, the current project leverages the continuum of multimodal cues across infant-directed speech and song to

investigate links across visual attention (especially mouth-looking), differential multimodal cue sensitivity, and expressive language in TD and autism. We harness a large existing dataset of well-characterized TD and autistic infants (Aim 1a), as well as infants at increased family likelihood for autism with varied outcomes (Aim 1b},

followed prospectively over the first two years of life, to quantify differential trajectories of mouth-looking during speech and song in relationship with expressive language outcomes. We combine this with new cross-sectional data collection in well-characterized cohorts of TD, autistic, and non-autistic expressive language delay (ELD)

children to determine how communicative contexts (song, speech; Aim 2) and sensitivity to specific multi modal cues (Aim 3) drive mouth-looking and are adaptive for expressive language skills across different diagnoses and expressive language levels. This research will identify basic mechanisms by which multimodal cues support

language development in specific populations, actionable targets for language intervention using diverse forms of ecologically-valid infant-directed communication, and developmental windows of opportunity for intervention development and delivery for autistic children.