Child Reciprocal Vocal Contingency Measure Using Automated Vocal Analysis with Children with Autism Spectrum Disorder

Thursday, May 11, 2017: 12:00 PM-1:40 PM
Golden Gate Ballroom (Marriott Marquis Hotel)
J. McDaniel1, A. L. Harbison1, P. J. Yoder1, A. Estes2 and S. J. Rogers3, (1)Vanderbilt University, Nashville, TN, (2)University of Washington Autism Center, Seattle, WA, (3)University of California, Davis. MIND Institute, Sacramento, CA
Background:  Child reciprocal vocal contingency (CRVC) is a new measure of a child’s participation in reciprocal vocal interactions in natural settings. Participating in reciprocal vocal interactions could facilitate speech and language development in children with autism spectrum disorder (ASD). Thus, evaluating correlations between CRVC and language development, particularly for preverbal children, might inform use of CRVC as an early indicator of treatment response. Although initial concurrent convergent validity and the stability of CRVC has been documented, CRVC’s sensitivity to change, predictive convergent validity, and divergent validity have not been evaluated.

Objectives: (a) What is the sensitivity to change of CRVC across 12 months? (b) Does initial CRVC correlate with expressive vocabulary at 4 time points spanning 18 months? (c) Does CRVC correlate with problem solving concurrently and predictively? A low correlation between CRVC and problem solving would support the divergent validity of CRVC.

Methods: Participants include 68 children aged 13 to 30 months old at study entry. Measures are displayed in Table 1. We quantified CRVC using a three-event sequential analysis to calculate the operant contingency value, which provides an index of contingency that is independent of the frequency of chance sequencing of events. Because this analysis requires long vocal samples, we used daylong Language ENvironment Analysis (LENA) audio recordings and only included participants with at least 1 hour of input near the child on the recordings. Using a computer program to classify vocalizations and compute the index of sequential association minimized human coder measurement error and vastly increased analysis efficiency.

Results: Change in CRVC across 12 months was significant (mean change = 0.037(0.077), t(25) = 2.41, p = .02). Table 2 displays correlations between CRBC, expressive vocabulary, and problem solving. Initial (Time 1) CRVC correlated significantly with expressive vocabulary at Times 2 and 4, but not Times 1 or 3. CRVC at Time 3 did not correlate with expressive vocabulary concurrently, but did 6 months later (Time 4). As predicted, correlations between CRVC and problem solving were non-significant at Times 1 and 3.

Conclusions: Overall findings suggest continued development of CRVC is warranted, despite mixed results. Positive indicators of potential clinical utility of CRVC include its sensitivity to change for initially preverbal children with ASD, correlation with expressive vocabulary but not problem solving at two time points, and its derivation using computer analysis of acoustic events without human coding, which minimizes clinician and researcher analysis time. The lack of correlation between CRVC and expressive vocabulary concurrently at some time points suggests that CRVC’s utility may be restricted to children of a particular language level or across a certain time range. Future research attending to these details is required.