Abstract
In this paper I discuss the relationship between thought and language in specific language-impaired children. Some central questions are at issue here. Firstly, there has been a tendency to overestimate the relationship between cognitive abilities and SLI and to exaggerate the causal relations between them. Secondly, I discuss the problems of measurement. Intelligence tests are problematic to evaluate linguistic abilities in SLI children because we don't know what kind of intelligence they are evaluating. It has been found by several authors that SLI children show a lower performance in cognitive tasks. Finally, in accordance with the works of A. Baddeley and S. Gathercole, I propose some questions concerning the relation between working memory in SLI children and their phonological deficiencies.