Canonical Babbling in 9-Month-Old Infants Later Diagnosed with Autism Spectrum Disorder: A Naturalistic Evaluation of All-Day Recordings

Background: Canonical Babbling (CB) may serve as an early developmental marker in differentiating neurodevelopmental disorders (e.g. Belardi et al., 2017; Oller et al., 2010). However, research on vocalizations and CB in infants at high familial risk for autism spectrum disorder (ASD) is limited and conclusions are mixed (e.g. Northrup et al., 2015; Paul et al., 2011; Talbott et al., 2016). Previous studies have used short home videos to analyze CB (Patten et al., 2014; Paul et al., 2011). This may not be representative of an infant’s abilities because infant volubility fluctuates throughout the day (Warren et al., 2010). Thus, in this study we used a novel method of selecting representative segments from full-day unobtrusive naturalistic recording using the Language Environmental Analysis (LENA^TM) to analyze CB in infants at high familial risk (HR) for ASD.

Objectives: To compare CB in 9-month-old HR infants diagnosed with ASD at 24 months (HR-ASD), without ASD (HR-Neg) and low risk infants (LR), using naturalistic day-long recordings.

Methods: A total of 72 (11 HR-ASD, 35 HR-Neg, 26 LR) 9-month-old infants contributed data as part of a larger, ongoing longitudinal study, the Infant Brain Imaging Study. Infants wore a LENA recorder for at least 1 full day (>8 hours). A study-specific algorithm parsed through the LENA files and compiled a 5-minute sample from high vocalization periods spread across the day. Two coders blind to subject diagnostic status coded these samples for canonical syllables (CS; consonant-vowel combination); IRR – 0.75, p<0.001. A Canonical Babbling Ratio (CBR) was computed, dividing total CS by total child speech-like vocalizations (ChV). Infants were assessed at either 9 or 12 months of age using the Mullen Scales for Early Learning.

Results: Across groups, infants did not differ by age, sex, number of hours of LENA recording, or parental education. The groups significantly differed on Mullen Receptive Language and Fine Motor scores at 9-12 months, but not on other Mullen subscales (Table 1). Mullen Non-verbal DQ, age, sex, data collection site and maternal education were included as covariates. Group differences in CBR were not significant, although the direction of effects was consistent with previous reports (meanCBR of HR-ASD < HR-Neg < LR; F₍₅₈₎ = 1.85, p=0.16). This direction of differences was not explained by a similar pattern in ChV (Fig-1)

Conclusions: We found no significant differences between HR-ASD, HR-Neg, and LR infant CBRs at 9-12 months of age. Previously reported large effects have come from clinically ascertained samples of children diagnosed with ASD rather than from prospectively ascertained HR infants. HR infants may demonstrate more typical patterns of CB development as compared with clinically ascertained samples. We await larger samples to more fully examine these phenomena with greater power to detect potential differences. Despite the absence of statistically significant results, this study provides a framework for studying CB in infants at risk for ASD, as well as the relationship of CB to behavioral trajectories and outcomes.