Predictors of Speech Improvement in Minimally Verbal Children with Autism Spectrum Disorder

Background:

Of the 1 in 68 children who meet criteria for autism spectrum disorder (ASD), approximately 25% remain minimally verbal (meaning they use fewer than 20 words functionally and have no productive syntax) past the age of 5 years. Previous work has identified both child and therapy characteristics that predict growth in vocabulary and communication rate (Rogers et al. 2006; Paul et al. 2013) and expressive language (Paul et al., 2013; Yoder et al., 2015) in minimally verbal children with ASD, but to date, no research has sought to determine characteristics that predict speech acquisition in minimally verbal children with ASD.

Objectives:

In this study, we aimed to identify child characteristics that predicted response to two therapies designed to facilitate spoken language in minimally verbal children with ASD.

Methods:

Thirty minimally verbal children with ASD (four female) between the ages of 3;5 and 9;8 participated in one of two treatments. Twenty-three children (two female) received 25 sessions of Auditory-Motor Mapping Training (AMMT), an intonation-based treatment involving repetition of sung bisyllabic stimuli and simultaneous tapping on tuned electronic drums. Seven children (two female) received 25 sessions of Speech Repetition Therapy (SRT), which involves neither intonation nor drum-tapping but is otherwise matched to AMMT. Children’s Baseline responses to two sets of stimuli were transcribed, then scored for percent consonants and vowels correct and percent syllables approximately correct. Chronological age and the number of English phonemes they were able to repeat on request at Baseline were also tallied. All five Baseline measures (age, size of phonetic inventory, percent syllables approximately correct, and percent consonants and vowels correct) were then entered into a hierarchical multiple regression to predict the change score in percent syllables approximately correct.

Results:

The full model, including all five predictors, was significant, R² = .534, F(5,24) = 5.508, p = .002. In the hierarchical analysis, only Baseline phonetic inventory, Baseline percent consonants correct, and Baseline percent syllables approximately correct resulted in statistically significant increases in R². Examination of regression coefficients showed that size of phonetic inventory was significantly associated with increase in percent syllables approximately correct, while Baseline percent syllables approximately correct and Baseline percent consonants correct were significantly associated with decreases in percent syllables approximately correct.

Conclusions:

The positive moderating effect of phonetic inventory suggests that the ability to correctly imitate phonemes on request is an indicator for degree of improvement in the ability to produce word approximations seen after 25 therapy sessions. The negative moderating effect of baseline percent syllables approximately correct and baseline percent consonants correct suggests that previously learned word approximations may be challenging to improve upon in this amount of time. However, anecdotal observations and parent reports suggest that these children can continue to improve their speech output even after therapy ends.