This project examines minimal breakdowns in understanding in which the person spoken to requests clarification from the speaker. It is directed to two sets of problems. The first deals with the various ways in which repair is initiated, the basis for selection among them, and ways in which such sequences are related to the expression or avoidance of disagreement. The second set of problems concerns the formal structure of repair sequences, such as their minimum size. The project will draw on an existing corpus of about 750 instances of repair sequences gathered from tapes and transciptions of mainly ordinary conversation with some supplementary material to be compiled during the course of the project. This is an interdisiciplinary project that is relevant to linguistics, anthropology, social psychology, and communications. Subfields of linguistics to which it will contribute include psycholinguistics, pragmatics, and discourse analysis.