Perceptual Markers of Autism during Oral Narrative Production
Individuals with Autism Spectrum Disorder (ASD) have variable language abilities, ranging from severely impaired to above average. Verbal individuals with ASD frequently exhibit speech differences, including atypical or lack of prosody. These speech and/or vocal differences often result in difficulty conveying communicative information, such as affect, linguistic, and pragmatic information. It is not clear what the underlying mechanisms are that contribute most to the acoustic form of “speech atypicalities” among speakers with ASD compared to the speech of their Typically Developing (TD) peers. Regardless of the mechanisms (linguistic, motoric, temporal, prosodic) that contribute to the perception of speech “oddness”, it is difficult to identify precisely what sounds “different” to the human ear.
The following research questions were addressed: 1) Are there distinct acoustic-perceptual speech and language differences as perceived by human listeners that distinguish children with ASD compared to TD peers?; 2) Do experienced and non-experienced judges acoustically rate oral narratives produced by children with ASD differently than those produced by TD peers?
Oral narratives using the ENNI (2007) were collected on 31 children with ASD and 31 TD peers between the ages of 4-11 years matched for age, gender and receptive vocabulary. The 31 participants with ASD were formally diagnosed with ASD and had an IQ above 70. All acoustic samples were digitally recorded in a child-friendly sound-treated room. The “human listeners” included one group of 40 Speech Language Pathology (SLP) graduate students (trained listeners) and 40 randomly selected graduate students from different disciplines (untrained listeners) who were recruited across a university campus. All human listeners (judges) met inclusionary criteria: 1) normal hearing; 2) English as a native language; and 3) level of experience/contact time with individuals with communicative disabilities and/or ASD. The 40 SLP judges were engaged in coursework and clinical experiences working with individuals with communicative differences or disabilities and/or children with ASD. The 40 untrained judges earning degrees in different disciplines had little to no experience working with individuals with communicative disabilities and/or ASD. The judges applied a perceptual rating instrument that included 13 categories comprised of seven linguistic and six speech/acoustic variables. The judges rated each category using a three-point Likert scale.
There are distinct differences as perceived by human listeners that distinguish children with ASD as sounding “different” compared to TD peers. For the main effect diagnosis (ASD vs. TD), significant between group differences were found for four out of 13 variables: Story sequencing (p=0.008); Articulation (p=0.001); Fluency (p=0.005); and Rate (p=0.016). A trend, but nonsignificant differences at 0.05 were found for Topic organization (p=0.073) and Pitch (p=0.058). For the main effect judge (SLP vs. untrained listeners), significant within group interactions were found for three variables: Emotional language (ToM) (p=0.020); Articulation (p=0.002); and Rate (p=0.037). A trend was also found for Causal language (p=0.094). Inter-rater reliability was respectable at 0.73.
Experienced and non-experienced judges rated the acoustic/speech aspects of connected speech as the most discriminating features that distinguished the connected speech among children with ASD compared to TD peers.