Language

THE PREFERENCE FOR SELF-CORRECTION IN THE ORGANIZATION OF REPAIR IN CONVERSATION Emanuel A. SchegloffGail JeffersonHarvey Sacks University of California,Los AngelesUniversity of California, Los AngelesIrvine An Organization of repair' operates in conversation, addressed to recurrent problems in speaking, hearing, and understanding. Several features of that organization are introduced to explicate the mechanism which produces a strong empirical skewing in which self-repair predominates over other-repair, and to show the operation of a preference for self-repair in the organization of repair. Several consequences of the preference for self-repair for conversational interaction are sketched.* 1. Self- and other-correction. Among linguists and others who have at all concerned themselves with the phenomenon of 'correction' (or, as we shall refer to it, 'repair'; cf. below, §2.1), a distinction is commonly drawn between 'selfcorrection ' and 'other-correction', i.e. correction by the speaker ofthat which is being corrected vs. correction by some 'other'.1 Sociologists take an interest in such a distinction; its terms—'self' and 'other'—have long been understood as central to the study ofsocial organization and social interaction.2 For our concerns in this paper, 'self' and 'other' are two classes of participants in interactive social * We wish to acknowledge the help, through discussion and/or through bringing relevant data to our attention, of Jo Ann Goldberg, Anita Pomerantz, and Alene Terasaki at the University of California, Irvine, and of Françoise Brun-Cottan, Irene Daden, and Louise Kerr at the University of California, Los Angeles. Harvey Sacks was killed in an automobile accident while this paper was undergoing final revision. 1 Bolinger ([1953] 1965:248) writes: 'What speakers avoid doing is as important as what they do. Self-correction of speech and writing, and the correction of others in conversation ("I can't understand what you say"), in classrooms, and over editorial desks is an unending business, one that determines the outlines of our speech just as acceptances determine its mass. Correction, the border beyond which we say "no" to an expression, is to language what a seacoast is to a map. Up to now, linguistic scientists have ignored it because they could see in it nothing more than the hankerings of pedants after a standard that is arbitrary, prejudiced and personal. But it goes deeper. Its motive is intelligibility, and in spite of the occasional aberrations that have distracted investigators from the central facts, it is systematic enough to be scientifically described.' Not much has been made of the distinction—in part, perhaps, because the disciplines have used it to divide up their work, self-correction being occasionally discussed by linguists (since it regularly occurs within the sentence?), e.g. Hockett 1967 and DuBois 1974, and other-correction by psychologists, e.g. Garvey, ms. They have rarely both been in the attention of the same investigator , who might then address himself to the relation between them. a Under various guises—self/other, individual/society, ego/alter—and through various understandings of the relationship between them—opposition, complementarity etc.—this pair of notions goes back to the origins of American sociology (G. H. Mead, Cooley etc.), to the classical figures of European sociology (Marx, Weber, Durkheim), and beyond the origins of sociology as an academically specialized discipline to the origins of social and political philosophy . For one account of the development of the theme that 'external control,' i.e. control by others, will not adequately account for, or guarantee, social order, cf. Parsons 1937. 361 362LANGUAGE, VOLUME 53, NUMBER 2 (1977) organizations—in particular those which characterize the sequential organization of conversation, specifically its turn-taking system.3 Thought of in terms of the social organization of conversational interaction, self-correction and othercorrection are not to be treated as independent types of possibilities or events, nor as structurally equivalent, equipotential, or equally 'valued'. Rather (and this is a central theme of our paper), self-correction and other-correction are related organizationally , with self-correction preferred to other-correction.4 One sort of gross, prima-facie evidence bears both on the relevance of the distinction and on the preference relationship ofits components. Even casualinspection of talk in interaction finds self-correction vastly more common than other...