A Six-Minute Measure of Canonical Babbling Status to Identify ASD in Toddlers

Background: To improve our ability to identify autism spectrum disorder (ASD) early in development, we need objective, reliable, and accessible measures. Previous work by Patten and colleagues (2014) suggests that early vocal patterns may provide a window into diagnostic status as early as 9-12 months. Patten and colleagues used retrospective home-videos to look at canonical babbling ratios (CBR), or speech-like utterances relative to all vocalizations. Here we investigated whether CBR as measured during a standardized six-minute presentation of videos on a tablet in primary care settings can be used to identify ASD status.

Objectives: (1) To determine whether CBR can be assessed during a short video presentation, and (2) To assess whether CBR can be used to discriminate diagnostic status for ASD.

Methods: Typically developing children (TD), children with language or developmental delays (DLD), and children with ASD were recorded using the camera in a tablet while they watched six minutes of videos selected to elicit behaviors associated with ASD during their 18- or 24-month well-child visits. Videos from thirty-nine children (ASD: n = 16, 3 male, M = 26.4 months, SD = 4.2; TD: n = 16, 3 male, M = 26.1 months, SD = 2.6; DLD: n = 7, 4 male, M = 24.9, SD = 2.6) were coded offline in realtime for canonical babbling based on the methods and canonical babbling status of CBR >.15 used by Patten et al. (2014).

Results: All but three participants (2 TD, 1 DLD) vocalized during this procedure, allowing for calculation of CBR for 36 children. A one-way ANOVA showed a significant effect of diagnostic status on CBR, F(2,33) = 6.96, p = .003, driven by the difference between ASD and TD participants (p = .002). Whereas 86% of TD and 83% of DLD children had achieved canonical babbling status, only 44% of children with ASD had CBR > .15. Odds ratio for the likelihood of ASD diagnosis given canonical babbling status for ASD and TD participants was 7.71, p = .002 (95% CI: 1.28 to 46.36). See Figure 1.

Conclusions: These results suggest that canonical babbling status can be measured in a short period of time in the context of a video presentation in a routine clinical setting. Further, they lend support to previous findings that CBR can help identify diagnostic status of children with ASD early in development. Given the small sample, replication will be necessary to confirm the generalizability of these findings. Future efforts will explore automatization of coding of vocalizations for increased accessibility.