AbstractsPsychology

HEARING AND SPEECH SCIENCES The Stability and Validity of Automated Vocal Analysis in Preschoolers with Autism Spectrum Disorder in the Early Stages of Language Development Tiffany Woynaroski Dissertation under the direction of C. Melanie Schuele, Ph.D. and Paul Yoder, Ph.D. Learning to use spoken words to communicate during the preschool years has been repeatedly linked with long-term outcomes in individuals with autism spectrum disorders (ASD). This replicated finding has motivated much research into the predictors of spoken language in young children with ASD. Previous studies have demonstrated that measures of child vocalization complexity and adult linguistic input as derived from conventional communication samples predict spoken language in samples of preschoolers with ASD who are heterogeneous in spoken language level. Measures of child vocalization complexity and adult linguistic input may be even more useful in predicting who will use words to communicate amongst the subset of children with ASD who are preverbal or just beginning to use words to communicate. Unfortunately, the time and cost inherent to conventional communication sampling make it difficult for clinicians to measure these important predictors in everyday clinical practice. This study drew on extant data from a recent longitudinal correlational study to explore whether automated vocal analysis, a less time-consuming and costly approach, may provide a valid and reliable alternative to conventional communication sampling for measurement of child vocalization complexity and adult linguistic input in preschoolers with ASD who are still in the early stages of language development. Our selected index of child vocalization complexity as derived via automated vocal analysis was stable with a single day-long audio-recording and non-significantly different from our index of child vocalization complexity as measured in conventional communication samples. In contrast, the stability and validity of our index of adult linguistic input as derived via the novel automated vocal analysis approach was not confirmed. Post hoc analyses demonstrated that both indices of child vocalization complexity had "added value" in predicting spoken vocabulary size in our sample even when controlling for adult linguistic input as measured in conventional communication samples. Results highlight the importance of child vocalization complexity as a predictor of useful speech in this population and support the use of time-efficient and cost-effective automated vocal analysis as a viable method of measuring this construct in everyday clinical practice. Keywords: useful speech, vocabulary, language development, automated vocal analysis, LENA, autism spectrum disorder