Date of Award:


Document Type:


Degree Name:

Doctor of Philosophy (PhD)


Special Education and Rehabilitation Counseling

Committee Chair(s)

Sandra Gillam


Sandra Gillam


Teresa Ukrainetz


Ron Gillam


Sarah Schwartz


Stephanie Borrie


Cindy Jones


Differences in the speech prosody, or "melody" of speech, of persons with autism spectrum disorder (ASD) have long been noted by researchers. Yet, despite many studies, researchers have not identified a universal description of speech prosody in ASD. It may be flat or monotonous, not different from typical, or overly variable. However, atypical speech prosody can immediately set someone apart from their peers. This distinction could negatively social, academic, and vocational interactions. For those persons with ASD whose speech prosody is different from typical and interferes with daily functioning, valid, reliable, and efficient assessments of speech prosody are needed. Currently, there are only three validated assessments for speech prosody specific to ASD and none of them are simultaneously valid, reliable, and efficient.

The purpose of this study was to design, validate, and establish sufficient reliability of a one-item, 7-point continuous rating scale for screening the speech prosody of children with ASD. Additionally, I investigated whether a brief, online training would improve reliability. The rating scale ranged from 1 (monotonous) to 7 (overly variable). Thirty-five 30-second audio clips from previous studies were chosen from children with ASD and neurotypical development. Three expert speech-language pathologists (SLPs) selected clips for the end and mid points of the scale and developed "gold standard" ratings. A total of 42 ASHA-certified SLPs with experience in treating children with ASD rated 20 of the audio clips at two time points. Twenty of the SLPs participated the online training prior to rating.

Analyses were conducted using linear mixed-effects modeling, which were built using a research-question, theory-based modeling approach. Results indicated moderate levels of reliability, except for intra-rater reliability in the trained group, which was good (ICC = 0.76). The results also partially supported the validity of the scale; however, this prosody rating scale requires further study and development before wide use.