Beyond MLU: Autistic children’s spoken language (Mathée-Scott & Venker, 2025)

<p dir="ltr"><b>Purpose: </b>Mean length of utterance (MLU) is a common measure of expressive language complexity in young children, including autistic children. However, means, by nature, obscure some information about spread and variability in data. Thus, we aimed to examine a new approach to characterizing linguistic complexity in autistic children by investigating the validity of two novel measures—range of length of utterance (RLU), standard deviation of length of utterance (SDLU)—alongside established measures: MLU and total utterances (TU). </p><p dir="ltr"><b>Method:</b> Participants were 40 autistic children (12 girls, 28 boys; <i>M</i>_age = 41.78 months). Children participated in 10-min, play-based language samples with their caregivers. Language samples were transcribed and measures (MLU, TU, SDLU, and RLU) were derived. To examine the criterion validity of these measures, we used regression analyses to examine how well each measure explained variance in children’s expressive language, as measured by the Preschool Language Scales–Fifth Edition (PLS-5). </p><p dir="ltr"><b>Results:</b> All measures (MLU, SDLU, RLU, and TU) significantly predicted PLS expressive language (<i>p</i>s < .001). Effect size comparisons revealed that all four predictors had large effect sizes (<i>R</i>² > .6). In absolute terms, MLU had the smallest effect size (<i>R</i>² = .682), followed by TU (<i>R</i>² = .72) and RLU (<i>R</i>² = .781), and SDLU had the largest effect size (<i>R</i>² = .822). </p><p dir="ltr"><b>Conclusions:</b> Findings suggest that these novel measures (SDLU and RLU) explained significant variance in children’s expressive language, as measured by the PLS-5, as did MLU and TU. SDLU had the largest explanatory power, in absolute terms, followed by RLU and TU. MLU had the smallest effect size, indicating that it had the lowest explanatory power for explaining variance in children’s expressive language as compared to the other three measures. Thus, examining spread and variability in utterance length data might provide important, previously overlooked, information about the complexity of autistic children’s spoken language. </p><p dir="ltr"><b>Supplemental Material S1. </b>R code used for data analysis. </p><p dir="ltr">Mathée-Scott, J., & Venker, C. E. (2025). Beyond mean length of utterance: Novel measures for characterizing the spoken language of autistic children.<i> American Journal of Speech-Language Pathology</i>. Advance online publication. <a href="https://doi.org/10.1044/2025_AJSLP-25-00152" rel="noreferrer" target="_blank">https://doi.org/10.1044/2025_AJSLP-25-00152</a></p>