PURPOSE: E. Bialystok and E. B. Ryan (1985) have outlined two operations, analysis and control, which are required for grammaticality judgments. In this model, analysis is involved in determining the grammaticality of a sentence, and control is required so that irrelevant information is ignored. This study examined these processes in specific language impairment (SLI). METHOD: Sixteen children with SLI and 20 typically developing (TD) children between 8;6 (years;months) and 10;6 were presented with a grammatical judgment task. Analysis was measured by recording children's decision times in determining grammaticality. Control was assessed by examining accuracy for judgments made for semantically odd sentences. RESULTS: Relative to the TD group, it was found that the children with SLI took longer in judging sentences associated with the process of analysis. Children with SLI were also found to have more difficulty, in terms of accuracy, with items requiring control (e.g., semantically odd sentences) than did the TD group. CONCLUSION: It is argued that the longer time required for children with SLI to respond to semantically normal sentences reflects a degree of difficulty with completing analysis. The SLI group's lower level of accuracy on semantically odd sentences reflects a problem with a control and is consistent with previously reported problems with cognitive inhibition in SLI.