Referring as a collaborative process.

In conversation, speakers and addressees work together in the making of a definite reference. In the model we propose, the speaker initiates the process by presenting or inviting a noun phrase. Before going on to the next contribution, the participants, if necessary, repair, expand on, or replace the noun phrase in an iterative process until they reach a version they mutually accept. In doing so they try to minimize their joint effort. The preferred procedure is for the speaker to present a simple noun phrase and for the addressee to accept it by allowing the next contribution to begin. We describe a communication task in which pairs of people conversed about arranging complex figures and show how the proposed model accounts for many features of the references they produced. The model follows, we suggest, from the mutual responsibility that participants in conversation bear toward the understanding of each utterance. Conversation is the fundamental site of language use. For many people, even for whole societies, it is the only site, and it is the primary one for children acquiring language. From this perspective other arenas of language usenovels, newspapers, lectures, street signs, rituals-are derivative or secondary. How, then, do speaking and understanding work in conversation? For psychologists this ought to be a central question, but surprisingly, it has not been. The main attempts to answer it have come instead from philosophy and sociology. Among philosophers the study of conversation grew out of ,an analysis of *We thank A.V. Belyaeva, E.V. Clark, E.P. Francik, R.J. Gerrig, W.J.M. Level& D. Morrow, G.L. Murphy, G. Redeker, and H. Stark for valuable counsel on this work. The project was supported by Grant MH-20021 from the National Institute of Mental Health, Grant BNS 83-20284 from the National Science Foundation, the Nederlandse Organisatie voor Zuiver-Wetenschappelijk Onderzoek, and the Center for the Study of Language and Information. Requests for reprints should be addressed to Herbert H. Clark, Department of Psychology, Jordan Hall, Stanford University, Stanford, CA 94305, U.S.A. OOlO-0277/86/$12.20 0 Elsevier Sequoia/Printed in The Netherlands 2 H.H. Clark and D. Wilkes-Gibbs what speakers mean and what listeners understand them to mean. The idea was that, when speakers utter sentences, they do so with certain intentions toward their addressees. They assert, request, promise, and perform other illocutionary acts, and their interlocutors are expected to recognize these intentions (Austin, 1962; Grice, 1957, 1968; Schiffer, 1972; Searle, 1969). In 1967 Grice argued that, for this scheme to work, people in conversation must be cooperative. Speakers must try to “make their contribution such as is required, at the stage at which it occurs, by the accepted purpose or direction of the talk exchange in which [they] are engaged (Grice, 1975, p. 45) .” Only then can their partners go beyond what is “said” to infer what is conversationally “implicated” (Grice, 1975, 1978). Among sociologists the issue has been how people direct the course of conversation and repair its inherent troubles. As this work has shown, people in conversation manage who is to talk at which times through an intricate system of turn taking (Sacks, Schegloff, & Jefferson, 1974). Further, when one person speaks, the others not only listen but let the speaker know they are understanding-with head nods, yes’s, uh huh’s, and other so-called back channel responses (Duncan, 1973; Goodwin, 1981; Schegloff, 1981; Yngve, 1970). When listeners don’t understand, or when other troubles arise, they can interrupt for correction or clarification (Schegloff, Jefferson, & Sacks, 1977). The participants also have techniques for initiating, guiding, and terminating conversations and the topics within them (Schegloff, 1968; Schegloff & Sacks, 1973). In both traditions a central issue is coordination: How do the participants in a conversation coordinate on the content and timing of what is meant and understood? The issue, however, cannot be resolved within either tradition alone. In the first tradition conversation is idealized as a succession of illocutionary actsassertions, questions, promises-each uttered and understood clearly and completely (Gazdar, 1979; Kamp, 1981; Stalnaker, 1978). Yet from the second tradition we know that many utterances remain incomplete and only partly understood until corrected or amplified in further exchanges. How are these two views to be reconciled? In this paper we propose a resolution for an essential use of language: how people in conversation coordinate in the making of a definite reference. Our concern is not with semantic reference, but with speaker’s reference-not, for example, with what the phrase the clown with the red nose means, but with what the speaker does in referring, say, to a clown as part of an assertion that the clown is funny (Donnellan, 1978; Kripke, 1977; Searle, 1969). Our premise is that making such a reference is a collaborative process requiring actions by both speakers and interlocutors. To some it may appear self-evident that the process is collaborative, but it is one thing to assume it is and Referring as a collaborative process 3 quite another to understand why it is and how it works. The goal here is important, since, if conversation is fundamental, its processes are likely to underlie or shape those in other uses of language as well. In the first section of this paper, then, we offer evidence for the premise itself and outline what we will call a collaborative model for the process of reference. In the second and third sections we describe an experiment on referring and use it to corroborate and fill in details of the model. In the final section we return to the general issue of coordination and note problems still to be resolved. Referring in conversation Traditionally, philosophers, linguists, and psychologists have presupposed what might be called a literary model of definite reference. Speakers refer as if they were writing to distant readers. When Elizabeth selects the noun phrase the clown with a red nose in talking to Sam, the assumption is that she intends it to enable him to identify the clown uniquely. She satisfies her intentions by issuing the noun phrase. Her act of referring is cotemporal with that noun phrase, beginning with the and ending with nose. Further, she retains complete responsibility and control over the course of this process. Sam hears the definite description as if he were reading it and, if successful, infers the identity of the referent. But his actions have no bearing on hers in this reference. The literary model makes these tacit idealizations. (1) The reference is expressed linguistically with one of three standard types of noun phrase-a proper noun (e.g., Napoleon, King George), a definite description (this year, the man with the moustache), or a pronoun (he, this, they). (2) The speaker uses the noun phrase intending the addressee to be able to identify the referent uniquely against their common ground. (3) The speaker satisfies her intention simply by the issuing of that noun phrase. And (4) the course of the process is controlled by the speaker alone. A conversational model of the process, however, ought to look quite different for three reasons. First, in conversation unlike writing, speakers have limited time for planning and revision. They need to overcome this limitation, and in doing so they may exploit techniques possible only in conversational settings. Second, speech is evanescent. The listener has to attend to, hear, and try to understand an utterance at virtually the same time it is being issued. That requires a type of process synchronization not found in reading. And third, listeners in conversations aren’t mute or invisible during an utterance. Speakers may alter what they say midcourse based on what addressees say and do. 4 H.H. Clark and D. Wilkes-Gibbs Indeed, once we look at actual conversations, we find that the four idealizations of the literary model are very wide of the mark. To see this, let us turn to eight types of examples that fail on one or more of these assumptions.