Maternal Input Predicts Wh-Question Production In Young Children with Autism

Background:  Wh-questions play a central role in many early social interactions (e.g., Brown, 1968; Ervin-Tripp, 1970). However, these forms are often absent from the speech of young children with autism (ASD). Correlations have been found between the speech that children with ASD produce and the maternal input that they receive (Swensen et al., 2006). However, it is not known if mothers’ speech predicts children’s wh-question production, specifically. The current research is part of a longitudinal study in which young children with ASD are visited every four months across a 3-year time span to investigate their language development. This report includes maternal input data from visits 1 and 2, and children’s production data from visits 3 and 4.

Objectives:  We investigated which features of maternal speech predicted subsequent features of wh-question production in 3-year-old children with autism. 

Methods:  Mothers and children participated in 30-minute semi-structured play sessions at each visit. At onset, children averaged 32.8 months-of-age and had a mean Mullen Early Learning Composite score of 75.9 (SD = 26.8). Transcripts of the sessions were coded for maternal input features, including: number of utterances, number and proportion of wh-questions, number of different wh-words (e.g., ‘what’, ‘where’, ‘whose’) and verb diversity. Children’s productions were coded for the same features. Additionally, a series of standardized tests (e.g., MacArthur CDI) were administered at each visit, to assess children’s development.

Results:  The number of wh-questions produced by children was predicted by concurrent CDI scores at visit 3 (R² = .313, p = .037) and visit 4 (R² = .689, p < .001). The proportion of mothers’ earlier utterances that were wh-questions did not account for a significant amount of additional variance (ΔR² = .006, p = ns), but the number of different verbs used by mothers at visit 2 was a significant predictor of the number of wh-questions produced by children at visit 4 (ΔR² = .103, p = .047).  Similarly, the number of different wh-words produced by children was predicted by concurrent CDI scores at visit 3 (R² = .623, p = .001) and visit 4 (R² = .637, p = .001), and the number of different verbs produced by mothers at visit 2 accounted for a significant amount of additional variance in the number of different wh-words produced by children at visit 3 (ΔR² = .139, p = .028). Lastly, the number of wh-word types produced by mothers at visit 1 predicted the number of different wh-words produced by children at visit 4 (ΔR² = .137, p = .025).

Conclusions:  ASD children with larger vocabularies tend to produce more wh-questions and more types of wh-words than their peers with smaller vocabularies. However, maternal input also influenced the wh-question production of children with ASD. Mothers who produced a greater variety of wh-words and verbs had children who produced more, and more varied, wh-questions; even after controlling for children’s vocabulary. More diverse input may help children with ASD produce more wh-questions by modeling the use of wh-questions in a variety of forms and situations.