Changes in repeated references: Collaboration or repetition effects?

Abstract

According to the Clark & Wilkes-Gibbs' collaborative model of reference, the repetition of a referential communication task produces changes in the making of references (decreasing number of words and of turns with the passing trials, increasing number of definite references and of labels), and these changes are interpreted as revealing the outgrowth of a collaborative process aiming at mutual understanding at the lowest cost. Using a repeated referential paradigm but without any physically present interlocutor, the present study aimed at answering the question of whether these changes are attributable to the mere repetition of the referential task. In adifferent addressee (DA) condition, the subjects were told before each new trial that what they would say would be presented later to another person; in asame addressee 9SA) condition, the subjects were told before each new trial that what they would say would be presented to the same person. With the passing trials, the subjects in the DA condition produced more and more words, but did not use any definite references or labels: These observations have no common point with the ones obtained within a conversational framework. The data obtained from speakers in the SA condition were only slightly different and far from being close to the ones gained in referential dialogues: The number of words remains constant, there was only a slight increase of definite references, and there were no labels. Repetition itself clearly cannot account for the changes in referring that have been reported in recent studies of how subjects cooperate in the making of references.