Can Early Vocalizations Predict Later ASD Symptom Severity?

Saturday, May 13, 2017: 12:00 PM-1:40 PM
Golden Gate Ballroom (Marriott Marquis Hotel)
D. Garrido1, L. R. Watson2, R. Garcia-Retamero1, G. Carballo1 and E. Crais2, (1)Psychology, University of Granada, Granada, Spain, (2)Department of Allied Health Sciences, University of North Carolina at Chapel Hill, Chapel Hill, NC

Varied infant risk markers predict later diagnoses of autism spectrum disorder (ASD). Limited findings are available, however, on predicting continuous measures of ASD symptoms from behaviors of infants at-risk for ASD (Macari et al., 2012; Northrup & Iverson, 2015). This approach is consistent with the current conceptualization of ASD as a dimensional disorder (Lord & Jones, 2012), and clinically important because of the possibility that infants at highest risk for severe ASD symptoms (and most in need of early intervention) can be identified earlier than reliable diagnoses can be made. Pursuing this idea, we examined the utility of one infant ASD risk marker, a low frequency of vocalizations directed to others (Ozonoff et al., 2010; Winder et al., 2013), in predicting later ASD symptom severity.


To investigate to what extent infant vocalizations (directed/non-directed) predict later ASD symptom severity in social-affective (SA) and repetitive and restricted behaviors and interests (RRBI) domains among community-identified infants at-risk for ASD.


Our sample comprised 82 infants, identified as at-risk for ASD via community screening with the First Year Inventory (Baranek et al., 2007). Thirty-minute video samples were coded for directed and non-directed speech-like vocalizations at 14 months (range 13-15). “Directed” vocalizations were those accompanied with gestures and/or eye contact, used within an interactive context, or imitating an adult’s vocalization. We used the frequency of directed and non-directed vocalizations to predict the severity of SA and RRBI symptoms at 23 months (range 20-25), based on the calibrated severity scores derived from Autism Diagnostic Observation Schedule-2 algorithm scores (Hus et al., 2014).


First order correlations showed a non-significant negative correlation between directed vocalizations and SA severity, r = -.143, p =.199, and a significant negative correlation with RRBI severity, r = -.421, p < .001). In contrast, non-directed vocalizations correlated significantly and positively with SA (r = .287, p = .009), but minimally and non-significantly with RRBI (r = .092, p = .412). Despite patterns in the first order correlations, multiple regression analysis showed that both directed (β =-.478, t(80) =-4.65, p < .001) and non-directed (β =.217, t(80) = 2.11, p =.038) vocalizations contributed significantly to predicting RRBI at 23 months, accounting for 22.1% of the variance in RRBI severity. Similarly, both directed (β = -.235, t(80) = -2.16, p =.034) and non-directed vocalizations (β = .348, t(80) = 3.21, p =.002) made significant contributions to accounting for a total of 13.4% of the variance in SA severity.


The finding that directed and non-directed vocalizations combined to account for more variance in later RRBI severity than SA severity was partially inconsistent with our predictions, especially given the relatively strong role for directed vocalizations in predicting RRBI severity. Nevertheless, particularly if combined with other infant risk markers, vocalizations may have utility in identifying infants whose functioning will be most severely impacted by ASD. Future research should be directed at explaining the associations between directed/non-directed vocalizations and RRBI and SA severity.