Coordinating reference in conversation: The choice between linguistic conventions and linguistic precedents

Abstract

How do speakers coordinate meaning with their addresses such as choosing a referring expression that maximizes the probability that their addressee identifies its referent? Language offers a community a system of conventions for recurrent coordination problems. Furthermore, research on repeated reference involving novel and hard-to-name shapes has claimed that an initial reference sets up a precedent, a partner-specific perspective that can be used subsequently. In a referential communication task involving photos of everyday objects, the present study hypothesized that precedents are temporary solutions developed by conversational partners when community-wide conventions on how to refer to the entity are not readily available. Degree of accessibility to conventions for each object (i.e., the object’s name uncertainty) was quantified based on the distribution of words used by all participants to talk about the object for the first time. Results from two studies showed that for objects with low name uncertainty, the expressions used by a dyad on two consecutives mentions were no more similar to each other than to expressions used by other dyads, suggesting reliance to community-wide conventions. For objects with greater name uncertainty, however, a given dyad’s expressions produced on two successive mentions resembled each other more than they resembled other dyads’ expressions. Finally, the usage of bare nominals (vs. definite or indefinite noun phrases) to refer to an object on second mention decreased as the object’s name uncertainty increased, a finding that further supports the claim that conversational partners rely on precedents as temporary solutions to compensate for uncertain availability of community-wide conventions.