Maternal Input Correlates with Wh-Question Comprehension in Young Children with Autism

Background: Wh-questions play a central role in many social interactions. However, these forms are often absent from the speech of young children with autism (ASD), and are challenging for them to understand (Naigles et al., 2008). Recent studies have reported that variation in maternal linguistic input correlates significantly with variation in subsequent speech production (word use, MLU) of children with ASD (Swensen et al., 2007); the current research investigates whether such correlations might be found with children's language comprehension in general, and with their grasp of Wh-questions, in particular. The current research is part of a longitudinal language study in which children with ASD are visited every four months across a 3-year time span. This report includes maternal input from visits 1 and 2, and comprehension data from visits 3 and 4.

Objectives: We investigated the comprehension of subject- and object-Wh-questions in 3-year-old children with autism. We then correlated their comprehension scores with maternal input measures from earlier visits.

Methods: At visit 1, the children had a mean age of 33 months, had begun intensive ABA therapy, and produced 27% of the words on the MacArthur CDI checklist. At Visit 3, they averaged 45 months and produced 45% of the words on the CDI. Mothers and children participated in 30-minute semi-structured play sessions at each visit. Maternal Wh-questions during the sessions were coded for various linguistic features. At visits 3-4, children watched a Wh-question video in an Intermodal Preferential Looking paradigm. This video showed ‘hitting' events (e.g., an apple hitting a flower), followed by test trials in which the apple and flower were shown on separate screens. Three types of Wh-questions were tested: Object questions (“What did the apple hit?”), Subject questions (“What hit the flower?”) and Where questions (“Where is the apple?”). Children's eye movements were recorded and coded off-line to assess comprehension.

Results: Approximately 11% of maternal utterances at these visits were Wh-questions; 60% of these included the copula as the main verb (e.g., “What's this? Where is the cookie?”). Many significant correlations were found between maternal input at visits 1 and 2 and later comprehension by their children; we report only the ones that were still robust once maternal MLU and child vocabulary were partialled out. Maternal speech forms that correlated negatively with later child Wh-question comprehension included Wh-questions with the copula as the main verb, and Wh-questions that repeatedly used the same few verbs (e.g., want, see, have). Maternal forms of speech that correlated positively with later child Wh-question comprehension included questions with inverted auxiliaries (e.g., “What did she eat?” rather than “What I'm going to do?”) and those with content-rich verbs (e.g., build, eat).

Conclusions: Wh-question acquisition in children with ASD appears to progress more slowly when their input consists more heavily of Wh-questions with the copular “be” than Wh-questions with a wide variety of content-rich verbs. Moreover, good examples of Wh-questions (e.g., with inverted auxiliaries) also seem beneficial. Thus, some kinds of input seem especially accessible to children with ASD for learning about Wh-questions.