“Story Goodness” in Adolescents with Autism Spectrum Disorders

Background: Individuals with autism spectrum disorders (ASD) have well-documented deficits in pragmatic language. Narrative elicitation is a particularly useful tool for examining pragmatic language, as it requires the speaker to provide and organize the relevant information to ensure the listener’s comprehension. Prior studies of “story grammar,” the organization of story components into logical relationships between characters and other story events, have proved it to be a quantitative measure of sociocommunicative abilities (e.g., in adults with brain injury; Coelho, 2002). Story grammar, along with story “completeness,” seems to capture what naïve listeners perceive as story “goodness” (Le et al 2011).

Objectives: Our objective was to study narrative quality in ASD as a means of understanding the pragmatic language deficits associated with the disorder. We compared narratives produced by adolescents with typical development (TD) and ASD using the quantitative methods described above (i.e., story goodness and story grammar), and through the use of real-world listener comprehension measures. 

Methods: Adolescents with ASD and TD (n= 15 per group) were matched on chronological age, gender, and full scale IQ. As part of the Autism Diagnostic Observation Schedule (ADOS), participants were shown a series of pictures of black and white drawings that depict a story about two monkeys. After these cards were removed, the adolescents were asked to tell the examiner the story from the pictures. These narratives were transcribed and broken into minimal grammatical units (T-units) for coding purposes. The narratives were evaluated for the proportion of T-units within story grammar structure, yielding a story grammar score, and the presence of six core events, yielding a story completeness score. Transcribed narrations were also rated by college students, naïve to diagnosis and study hypotheses, as a measure of story goodness, fidelity to the provided story, coherence, and oddness.

Results: Participants with ASD and TD did not differ on the measure of story grammar, indicating that narratives of both groups were similarly organized. While both groups produced a similar total number of T-units, the ASD group included significantly fewer core story events than the TD group F(1, 28) = 4.093, p = 0.053, partial η²= 0.128, such that stories were less complete. Undergraduate raters found the stories of the ASD group to be significantly less good overall (“how good a story is this?”) than those of the TD group, F(1, 28) = 15.829, p < 0.001; however, stories did not differ by group for ratings of accuracy, coherence, and oddness. This indicates that the raters detected a global difference between the narratives of the two groups, but were unable to specify the fine-grained details of this global difference.

Conclusions: Adolescents with ASD produced similarly structured, yet less complete narratives than adolescents with TD. Untrained raters were sensitive to this global difference in the ASD group’s narratives. As narratives are central in much of our daily communication, these findings suggest that this method of evaluating narratives may provide a sensitive measure of current sociocommunicative abilities and changes in response to intervention in adolescents with ASD.