ACOUSTIC AND PERCEPTUAL MEASUREMENTS OF PROSODY PRODUCTION ERRORS ON THE PEPS-C BY CHILDREN WITH AUTISM

Background: Since the first identification of the autistic syndrome, prosody has been reported to be impaired in individuals who speak (Kanner, 1943). Research has begun to identify profiles of deficits in this population across domains of prosodic function (Paul et al., 2005; Peppé et al, 2007). These studies have relied on perceptual judgments. Objective measurement of production patterns in autism is needed.

Objectives: Our objective is to examine the profile of prosodic deficits in children with autism, and to describe the acoustic differences that characterize them.Methods: Twenty-five youth (ages 8-16) with autism spectrum disorders (ASDs) and average general language functioning and 34 typical controls (ages 8-17) matched on chronological age and gender participated. Participants were given the Profiling Elements of Prosodic Systems in Children (PEPS-C), a norm-referenced measure of prosody perception and production for children 4-16. This measure provides data in the domains of affect, sentence stress, phrase structure, and discourse structure. Utterances produced by participants were scored perceptually according to PEPS-C scoring rules. They were also acoustically analyzed using PRAAT, a program for speech analysis and synthesis.

Results: Participants with ASD performed significantly worse than controls in using prosody productively for sentence stress, F(1,57)=6.19, p<.02, d=.67, and marginally worse for discourse structure, F(1,57)=3.17, p=.08, d=.48, and affect, F(1,57)=2.92, p=.09, d=.46. They also performed worse than controls on all measures of receptive prosody (p<.05), with effect sizes ranging from d=.5 to d=.9 Analysis of fundamental frequency patterns used to convey linguistic and affective cues also suggests differences between diagnostic groups. Conclusions: This study found pervasive deficits in prosody production and comprehension in ASDs that are present for both linguistic and affective domains of function. Moreover, differences between children with autism and typical controls can be measured objectively through acoustic analyses.