Friday, December 16, 2016

The role of syntax in encoding and recall of pictoral narratives: Evidence from specific language impairment


Before an event can be recalled, it must be encoded and stored in memory. We know from previous research that the better new information is integrated with other information, the better it is encoded and the easier it is to recall later on. It is thought that in narratives, story events can be integrated by identifying temporal or causal relationships between them, or by identifying characters’ motivations (psychological causation). However, to speak about these relationships, the child telling the story must be able to use a certain level of syntactic complexity. Specifically, speaking about temporal or causal relationships often requires the use of subordination. For example, in the sentences “We got lost twice before we arrived at the hotel” and “Sheila tripped because Murray left the rake on the lawn,” two events are joined by subordinating conjunctions, “before” and “because.” In contrast, talking about characters’ motivations requires the use of mental state terms, like “think” and “try,” or complement clauses. In this context a complement is a phrase that completes the meaning of the verb, acting like an object of the verb. In the sentence, “They knew she was involved” the phrase “she was involved” is the complement of the verb “knew.” With this in mind, Bishop and Donlan hypothesized that children with poor syntactic ability would have difficulty recalling stories not necessarily because of poor understanding of causal connections, but because of limited facility with these syntactic structures.

Using two series of five pictures each, the authors collected two narratives from three groups of participants: typically developing children, children with expressive specific language impairment (SLI-E) and children with receptive and expressive SLI (SLI-R). Thirty minutes after the children generated the initial narratives, they were asked to recall both of them. Both initial and recalled narratives were coded for content, whereas initial narratives were coded further for syntax (mean length of utterance and coding for clause type) and inclusion of mental state terms. All children completed measures of nonverbal reasoning and vocabulary, and children with SLI completed additional assessments of language ability and general comprehension.

Analysis of the initial narratives showed that both SLI groups produced shorter and grammatically simpler stories than the control group, however only the SLI-R group used simpler clause structure and fewer mental state terms than controls. Examination of the recalled narratives in light of the initial narrative showed that the SLI-R group forgot disproportionately more of the story content than did the other two groups. Correlational analyses found that positive predictors of content in recalled narratives included complement clauses, subordinate clauses, mental state terms, and general comprehension. No correlation was found for nonverbal reasoning or vocabulary. The authors use these results to support the hypothesis that structural language ability to a greater extent than nonverbal reasoning contributes to encoding and therefore recall of narratives.


Blogger: Laura Pauls is a PhD student in Speech & Language Science, researching children with language impairment and/or working memory impairment.