Phonological short-term memory and relative clause comprehension in Cantonese-speaking children with specific language impairment

Abstract

Children with Specific Language Impairment (SLI) have difficulties with the comprehension of relative clauses (RCs). The literature offers three explanations for these difficulties. One is a deficit in the computation system for syntactic movement and thematic role assignment. The second explanation is a failure in detecting regularities of the RC structure from the language input. The third explanation is a deficit in phonological short-term memory (pSTM). The ability to store verbal items, and to accurately retain their serial order, is crucial for learning sentences with an RC. This project aimed to examine the relationship between pSTM and the comprehension of RCs in Cantonese-speaking children with and without SLI through three studies. Study 1 investigated children’s pSTM capacity, via tasks that specifically tapped serial order or item information pSTM. Study 2 examined children’s comprehension of subject RCs (Subj-RC) and object RCs (Obj-RC) using a picture-sentence matching task. In this task, each picture stimuli consisted of two pairs of cartoon animals, one pair depicted the target noun phrase and the other pair depicted the same action with the animals playing opposite roles. The child was required to find a character to which the RC referred from two pictures. Subj-RCs in Cantonese do not follow canonical SVO word order while Obj-RCs do. Study 3 looked at the involvement of pSTM in the comprehension of RCs using a Change Detection task. In this task, the child was asked to judge whether two RCs were identical in wordings. Each of these studies was further divided into two parts: Part 1 described the performance of 50 children with normal language development, aged between 3;10 and 6;01, and Part 2 described the performance of 18 children with SLI, aged between 4;0 and 6;01. None of the explanations in the literature adequately described the RC comprehension patterns observed in the current study. Children with normal language development showed an advantage in the comprehension of Subj-RCs over Obj-RCs, and their performance in the comprehension of Subj-RCs correlated significantly with their scores on serial order pSTM. There were no significant correlations between the comprehension of Obj-RCs and pSTM. Some children who failed to comprehend RCs interpreted the Subj-RCs and the Obj-RCs using the same strategy. They pointed at the first argument heard in the RC in the picture-sentence matching task, assuming that all RCs had a canonical word order. Adoption of this comprehension strategy could not be explained by difficulties in retaining details of RCs, as these children were at the ceiling on the Change Detection task. These results suggested that their difficulties could be a problem in understanding the syntactic relations. Children with SLI were found to have a deficit in pSTM. Yet this deficit in pSTM was not related to their performance in the comprehension of RCs. Children with SLI appeared to adopt the same strategies as their age peers with normal development in the interpretation of Subj-RCs and Obj-RCs. However, a number of them chose the last argument heard to be the character referred to in the RC. Although children with SLI were observed to have more difficulties in recalling details of the RCs than children with normal language development in the Change Detection task, there were no correlations between their scores on this task and the comprehension of RCs. Therefore, their difficulty in interpreting RCs was not the result of a failure in remembering the order of words in the structure. A plausible alternative explanation was a failure in the processing of the syntactic relations within an RC. Findings of this study have implications for clinical intervention. It is suggested that therapy on RC should focus on linguistic processing instead of a reduction of memory load.