High-Functioning Children with Autism Flexibly Use Prosody to Parse Syntactic Ambiguity

Background: While impairments in producing prosody have been long studied in autism spectrum disorder (ASD), we know less about impairments in using prosody during comprehension. Research on the use of prosody to infer syntactic structure is particularly sparse.  Most existing research has used tasks that rely on overt judgments (Paul et al, 2005) rather than examining spontaneous differences in interpretation and the moment-to-moment processes that give rise to them. The exception to this is a recent study by Diehl and colleagues (in prep) demonstrated that 8-12 year-old high-functioning children with autism (HFA) can use prosody to parse globally-ambiguous sentences.  But this ability is fragile.  Performance is strong in the first block of trials, but drops sharply in the second block when the prosody shifts, a pattern seen in typically-developing preschoolers. Thus children with ASD show a delay in inhibiting the incorrect analysis, perhaps due to executive function (EF) impairments (Hill, 2004).

Objectives: The current study uses temporary syntactic closure ambiguities (1-4).  Temporary ambiguities may place fewer demands on EF than global ambiguities because the time period in which two syntactic analyses compete is substantially reduced.  Temporary-ambiguities are also more common in natural speech and thus provide a more ecologically valid measure of prosodic processing.  If HFA children can flexibly use prosodic cues to guide syntactic interpretation, they will succeed in this task, which places minimal demands on EF. However, if HFA children lack the flexibility to shift among syntactic analyses, then we expect to see no effects of prosody on the interpretation of these local ambiguities.

Methods: 6- to 9-year-olds with HFA (Mean age=91months, CELF=110) and typically-developing controls (Mean age=87months, CELF=114) were tested using the visual-world paradigm. Children were presented with utterances with closure ambiguities (1-4). In the early-closure conditions (EC) the second noun was the subject of the main clause. In the late-closure conditions (LC) this noun was the object of the subordinate clause.

1. When the robot baked the big postman delivered the mail (EC/neutral)

2. When the robot baked….the big postman delivered the mail (EC/cooperative)

3. When the robot baked the big muffin the postman delivered the mail (LC/neutral)

4. When the robot baked the big muffin…the postman delivered the mail (LC/cooperative)

Looks to the probable subject (postman) and the probable object (muffin) were measured in the ambiguous time window (underlined) and the noun time window (bold).

Results: In the ambiguous window controls looked at the probable object less often in the EC/cooperative condition than in the other three conditions resulting in a ProsodyXClosure interaction (p<0.01). This interaction disappeared in the noun window.  In contrast the HFA children showed no effects in the ambiguous window, but during the noun window, they displayed the same pattern as TD controls had in the ambiguous window (p<0.05).

Conclusions: We conclude that 6-9-year-old HFA and controls are capable of using prosody to sentence-parsing flexibly, unlike 5-year-old typically-developing children (Hahn&Snedeker, 2011) suggesting no developmental delay. However, they are slower at integrating prosodic information with the semantic information during real time processing.