Acoustic-Perceptual and Quantitative Analysis of Grammatical Prosody in Adolescents with Autism

Background: Published empirical research is contradictory regarding the specific characteristics of atypical prosody among individuals with Autism Spectrum Disorder (ASD). It has been suggested that individuals with ASD have atypical pragmatic and affective prosodic abilities while grammatical prosodic abilities are relatively spared (Shriberg et al., 2001). However, there is some preliminary evidence that individuals with autism exhibit acoustic or perceptual differences in grammatical prosody (e.g., Grossman et al., 2010). Research is sparse regarding the specific features of grammatical prosodic abilities among individuals with ASD. It is important to reliably and validly identify and quantify grammatical prosodic features that contribute to the perception of “unnatural” speech of some individuals with ASD.

Objectives: This research addressed the following questions: 1) Do individuals with ASD perform significantly different on perceptual and acoustic measurements of expressive lexical stress and phrase stress tasks as compared to a TD comparison group?; 2) Do individuals with ASD perform significantly different on receptive lexical stress and phrase stress tasks as compared to TD controls?; 3) Do trained listeners who are blind to stimulus item and diagnostic category judge individuals with ASD as sounding more “unnatural” than their TD counterparts using a novel acoustic-perceptual rating scale to assess lexical and phrase stress?

Methods: A between-group study was conducted to investigate grammatical prosodic abilities among individuals with ASD (n=11; ages 13;11 to 19;11 years) in comparison to typically developing (TD) controls (n=11) matched for age, gender, and receptive language. Each group consisted of 7 males (63%) and 4 females (36%). Participants with ASD and their TD counterparts were monolingual English speakers. Four tasks from the Profiling Elements of Prosody in Speech-Communication 2015 (PEPS-C 2015) were administered to assess: 1) the perception and production of lexical stress in two-syllable words; and 2) the perception and production of a noun phrase versus a compound noun within short sentences. Receptive tasks required the participant to select a picture that corresponded with the utterance they heard from a recorded standardized test battery. Expressive tasks were rated by trained listeners using a set of predefined perceptual and acoustic measurements.

Results: The group of individuals with ASD performed significantly less accurate on the expressive Lexical Stress (p= .021) and Phrase Stress (p= .0001) subtests (PEPS-C-2015). Acoustic findings revealed significantly longer duration of utterances (p< .05) among the group with ASD. The individuals with ASD performed significantly less accurate on the receptive Lexical Stress (p= .001;) and Phrase Stress (p= .009) subtests on the PEPS-C (2015). Large effect sizes were evidenced (Cohen’s d >1). Individuals with ASD were significantly more likely to be rated as sounding “unnatural” or “atypical” as compared to TD controls.

Conclusions: The outcomes of this investigation show that there are specific differences in the production of lexical stress and phrase stress that contribute to the perception of unnatural or atypical prosody among some individuals with ASD. Atypical prosody may contribute to communication and social impairments. The impact of language on prosody, its clinical implications, and directions for future research are enumerated.