The filled pause and social aspects of conversations.

Selected developments in the fields of verbal and nonverbal behaviour are reviewed, with especial reference to the study of speech in social context. Among features bordering on both language studies and social psychology are hesitations and disruptions in speech. Silent pauses have been found to result from need to plan verbal sequences and a miscellaneous group of speech disruptions known as NonAhs is a sign of topical anxiety. Filled pauses ('er', 'um' and variants, also called Ahs) were first thought to belong with NonAhs but have proved unrelated to anxiety and require a separate explanation. One hypothesis, that they have an interpersonal role in apportioning the conversational floor, has fared inconclusively under test and recent writers have written it off. In two experiments, filled pause rate was measured as a dependent variable. Mutual visibility in dyadic conversations was varied from zero through intermediate levels to normal, but no changes were observed. When an interviewer's tendency to interrupt was varied, again no significant differences in Ah rate were recorded. However, the filled pause as an independent variable elicited effects supportive of the floor control hypothesis. 'Matched guise' recordings of a speaker were heard by independent groups of undergraduates and presence of Ahs yielded ratings of speaker anxiety, caution and submissiveness, consistent with either the discredited anxiety hypothesis or that of floor control. In a final experiment, naive subjects each interviewed a person whose answers varied in grammatical completion and whether they terminated with Ahs. Either grammatical incompletion, an Ah or both prolonged latencies of subjects' next question substantially. The view that conversations are competitions for the floor is rejected for a broader outlook on interpersonal regulatory cues. Ahs probably do act in floor control, but less simply than previously thought and they may have other, non-regulatory roles besides. Avenues for future research on the topic are outlined with methodological suggestions and the work is presented within a suggested systematisation of accumulating knowledge about interrelationships between linguistic features at several structural levels and various aspects of social behaviour.