Abstract
Nuske, H. J., & Bavin, E. L. (2010). Narrative comprehension in 4–7-year-old children with autism: Testing the weak central coherence account. International Journal of Language & Communication Disorders, 46, 108–119.
Research shows that children with autism spectrum disorders (ASD) have difficulties with comprehension, particularly with inferential processing (IP). Children with autism are characterized by a detailed focused cognitive style that has been termed weak central coherence (WCC). Little research on the relatedness of inferencing difficulties and WCC has been conducted; therefore, this study was conducted to further knowledge of language-processing difficulties in autism.
Children with high-functioning autism sometimes exhibit relatively spared structural language skills but significant comprehension deficits. They frequently have severe difficulties using contextual cues to understand the meaning of interactions. They particularly have deficits in making inferences to fill in information that is not explicitly stated in a message. Researchers (Jolliffe & Baron-Cohen, 2000; Norbury & Bishop, 2002) have found that children with high-functioning autism were less able to make inferences about spoken narratives than typically developing children despite comparable performance on factual questions. These findings indicate that persons with high-functioning autism demonstrate difficulties with IP, which appears to reflect an inability to integrate the immediate context with their own content knowledge. Message comprehension requires an integration of meanings of individual words, sentences, and paragraphs, as well as key ideas and themes. Information from various sources needs to be incorporated as a whole, using gestalt-like global processing. Instead of using global processing (the processing style in typically developing individuals), individuals with autism are biased to process information on a local level, with particular attention to detail; they have a WCC (Frith, 1989). By contrast, typically developing individuals are characterized by a strong central coherence. As comprehension requires global processing, this local processing bias may account for some comprehension problems in autism.
The processing of narratives requires three key skills: the ability to infer causal relationships between events, to distinguish the goal and internal states of the people mentioned, and to conceptually integrate different sections of the story or account (Graesser, Singer, & Trabasso, 1994). Many interrelated skills are common to communicative and narrative comprehension, including vocabulary and syntactic knowledge, and prepositional and relational concepts. In addition, short-term auditory memory and attention skills are important. Therefore, studying verbal narrative comprehension can provide information on comprehension of fictional stories and comprehension in conversational contexts. For a narrative to be taken as a coherent unit, IP skills are required to build a mental representation (Graesser et al., 1994; Skarakis-Doyle & Dempsey, 2008). Consider the following example (Charniak, 1972):
Jane was invited to Jack’s birthday party. She wondered if he would like a kite. She went to her room and shook her piggy bank. It made no sound.
Many inferences need to be made to comprehend this message: If the piggy bank made no sound then there is no money. If there is no money, a gift cannot be bought for Jack, whose birthday is coming up or has just passed since he is having a birthday party. To make an inference as to the meaning of Jane’s actions, one must draw on event schema of birthday parties and associated conventions.
Event schemas are highly organized mental representations of what happens in common real-life events. Event schemas are based on experience; thus, they involve memory processes, by which the information central to the theme or story is retained and minor details are omitted. Schemas provide a framework for understanding events. Inferences based on event schemas are script inferences. Because script inferences require global processing of event schemas, it is hypothesized that inferences of this type would be harder for individuals with autism.
Script inferences can be contrasted with propositional inferences. Propositional inferences are based on logical relationships between story statements. The underlying structure of these inferences is based on formal deduction whereby two premises are given, from which a logical conclusion is drawn. Because propositional inferences rely on local processing, not the integration of previously acquired schema knowledge, it is hypothesized that individuals with ASD would be at an advantage on these types of inferences. Scott et al. (1999) found superior logical reasoning skills in children and adolescents with autism aged 7:9–18:0, compared with verbally matched controls. An example of a propositional item is as follows:
Major premise: All cats bark.
Minor premise: Rex is a cat.
Conclusion (test question): Does Rex bark?
Typically developing children probably considered that, in the real world, cats do not bark and that Rex is a typical name for a dog. Children with autism presumably perform well on tasks like this because they do not let real-world knowledge determine their response. That is, children with autism performed well on the tasks because of a local processing bias, as predicted by the WCC account. They reacted to the information provided rather than drawing on existing knowledge. Thus, a WCC cognitive style may help to explain superior as well as impaired comprehension, depending on the task at hand.
Numerous research studies have also shown that individuals with ASD lack a theory of mind (ToM); it is argued that they are unable to comprehend the social world due to an inability or reduced ability to attribute mental states to others and to themselves, such as intentions, desires, and beliefs (Baron-Cohen, 1995). However, recent findings have shown that higher functioning individuals with ASD can pass many ToM tests (Happe, 1994b). Despite this, social interaction and communicative difficulties among this subgroup of ASD persist. By contrast, the WCC cognitive style of ASD seems to characterize individuals on the spectrum, regardless of their language or sociocognitive function (Happe, 1994a). The WCC theory does, however, help to account for the specific comprehension problems and local processing advantages experienced by individuals with autism.
The original conception of the WCC theory was that this detail processing style was at the expense of global processing or meaning extraction (Frith, 1989). However, many tasks aimed at assessing WCC have placed global and local processing in a direct trade-off. Thus, it is unclear whether results reflect poor global performance in autistic spectrum disorder (ASD), enhanced local processing, or both. Research has shown that when high-functioning individuals with autism are primed for global processing, they can perform similar to typically developing individuals (Happe & Frith, 2006).
The main aim of the present study was to assess the communicative comprehension of children with autism using short-spoken narratives. Tasks were included to examine whether the WCC cognitive style affects comprehension of spoken narratives for individuals with autism. Three hypotheses were made:
If children are primed to the event type by the title of a story, children with autism would perform comparable on questions relating to the main idea in short narratives and better than typically developing children on questions about detail.
Children with autism would perform worse than a typically developing group on script inferences, but better on propositional inferences because only local processing is required for the latter.
For children with autism, good performance on local processing tasks would not be associated with poor performance on tasks requiring global processing.
Study
Participants
In all, 14 children with autism (aged 4:6–7:11) and 14 typically developing children (aged 4:2–5:4) were matched on the Receptive Vocabulary and Picture Completion subtests from the Wechsler Preschool and Primary Scale of Intelligence – Third Edition (WPPSI-3; Wechsler 2002).
Measures
The Block Design subtest from the WPPSI (Wechsler, 2002) was used as a marker of WCC. Each child was asked to recreate block designs made by the researcher or presented in the stimulus book.
The Understanding Spoken Paragraphs subtest from the Clinical Evaluation of Language Fundamentals–4 Australian version (CELF-4; Semel, Wiig, & Secord, 2003) was used as one measure of verbal comprehension. Researchers primed children for the story by giving each story a title. Each of three stories had five related questions of which two were appropriate for this study: the main idea and detail questions. For example, in a story about a boy’s first day of primary school, the main idea question was, “Why was Marcus feeling frightened and excited?” and the detail question was “What did Marcus get when he went shopping?”
IP task. Six short stories (five to seven sentences in each) were included, each with three questions, one factual, one script and one propositional. The factual question asked children a piece of information given in the story, and was used to make sure they had listened to the story. An example of a story is as follows:
Debbie had a birthday party. She didn’t want any more dolls for her birthday. There were five candles on the cake. Debbie blew out the candles. Then Debbie opened her presents. She got some games and a doll.
The questions (and expected answers) were as follows:
Who’s birthday was it? (Debbie’s)
How old is Debbie? (Five)
Did Debbie like all her presents? (No, she didn’t want any more dolls)
Results
The scores of children with ASD did not differ from the scores of the typically developing children on main idea (inference questions) and detail questions for the Understanding Spoken Paragraphs. Priming the children by giving them story titles likely enabled children with ASD to perform similarly to the typically developing children on both types of questions.
The groups differed significantly on script inferences with the mean for the autism group being lower than the mean for the typically developing group. This is consistent with predictions of the WCC theory. Children with autism were less likely to generate the core elements of the scripts.
The groups did not differ on propositional inferences of the IP tasks. This finding was not consistent with previous research on formal reasoning of children with autism and does not support the proposed local processing advantage in autism.
Performance on Block Design did not significantly correlate with performance on the detail questions for each group; however, for the children with autism, performance on Block Design correlated significantly with performance on propositional inferences.
Inverse correlations were not found between performance on local processing tasks (Block Design, detail questions, and propositional inferences) and performance on global processing tasks (main idea questions and script inferences). Thus, high scores on the local processing tasks were not associated with low scores on the global processing tasks, which is consistent with the view of the WCC theory that a local processing bias is not at the expense of global processing.
For the typically developing group, scores on main idea questions correlated significantly with scores on propositional and script inferences and detail questions. For the autism group, performance on main idea questions did not correlate with any question or inference type.
Conclusion
In this study, two aspects of narrative processing were examined within the WCC theory: local processing (detail questions and propositional inferences) and global processing (with main idea questions and script inferences). Although scores on the local processing tasks were comparable between groups, no support was found for enhanced local processing in autism. Children with autism did not do better than the typically developing children on the detail or propositional questions. Children with autism extracted the main idea of stories when they were primed by being given titles for the paragraphs, but did not spontaneously make inferences regarding event scripts. For the typically developing children only, extracting the main idea was found to be associated with enhanced performance on other comprehension tasks. Thus, the children with autism seem to have failed to use global processing to integrate contextual clues and content knowledge into a conceptual framework of understanding. Lack of such integration leads to impaired comprehension in communicative contexts.
This study suggests that it can be useful for educators and speech/language pathologists to differentiate among the different types of questions they ask students with ASD. Math and science activities are more likely to require more propositional inferences than scriptal inferences. Because propositional inferences are easier than scriptal inferences for children with ASD, it may be advisable for SLPs to begin teaching inference making to children with ASD by using expository texts that require propositional inferences. Narrative activities, both hands-on experiences and stories, can be used to teach inferences that require scriptal knowledge.