Identifying Features of ASD Language Impairment in Narrative Retellings

Background: It has yet to be established whether the language difficulties observed in some children with ASD are indicative of a developmental language disorder (DLD) or are characteristic of a distinct subtype of ASD.  In this research, we employ a novel method of analyzing the results of a widely used neurocognitive assessment tool, the NEPSY Narrative Memory subtest, in order to distinguish the language features associated with DLD from those associated with ASD. In this test, a subject listens to a story and then retells the story to the examiner. The retelling is scored by counting the number of predetermined "story elements" it contains. Although these scores correlate strongly with measures of language ability in children, this procedure ignores much of the linguistic information contained in a child's retelling. The scoring method proposed here attempts to leverage that information for differential diagnosis.

Objectives: To determine whether an alternative method of scoring narrative retellings can reveal differences in language impairment between children with DLD and children with ASD who meet criteria for a language disorder.

Methods: A battery of language and neurocognitive assessment tests were administered to 74 children ages 4-8 with ASD (N=34), DLD (N=17), or typical development (N=23). The ASD group was further divided into two groups according to whether the subjects met the criteria for DLD (ASD+DLD N=21, ASD-DLD N=13). The NEPSY Narrative Memory subtest was administered to each child. The story retellings elicited in the free recall portion of the test were transcribed and scored in two ways. In the standard method, each retelling was scored according to the published guidelines in which the child earns two points for every recalled story element. In the alignment method, each retelling was aligned to the original narrative by matching the words in one to the same or similar words in the other. For instance, the word "Jim" in the original story might align with "the boy" or "he" in a retelling.  The score is the percentage of words in a retelling that could not be aligned to any matching word in the original narrative.

Results: (1) Under the standard scoring method, there was no significant difference between the DLD and ASD+DLD groups. There were significant differences between the TD and ASD-DLD groups and the DLD and ASD+DLD groups. (2) Under the alignment scoring method, the DLD group had a significantly smaller percentage of unaligned words than the ASD+DLD group. In addition, there was no significant difference between the DLD group and the ASD-DLD group.

Conclusions: We observed that many of the ASD children included off-topic content in their retellings, while the DLD children tended to report the facts of the story, albeit with great difficulty and without the required story elements. Our novel method for scoring the NEPSY Narrative Memory subtest seems to reveal these kinds of differences, offering utility for distinguishing DLD from ASD-related language impairment. Future work will focus on generating alignments automatically using existing computerized techniques and exploring other features extracted from these aligments.