Perspective-taking in Conversation

Vanderbilt University, Nashville TN

Investigators

Abstract

Conversation among members of a group is a basic form of language use. It likely played a key role as languages evolved and it continues to play important roles in children's language acquisition and adult communication. Group communication is typical of a variety of high-stakes settings, including classroom education and business teams, yet surprisingly little is known about what makes communication in groups successful. One determinant of success is the ability to appreciate the perspective of the other person or people in the conversation. Basic communicative exchanges such as asking a question require calculations about what other people know. For example, if you want to know the answer to a question such as "Where is the gas meter?" it is first necessary to determine who is likely to know the answer in order to ask the appropriate person (e.g., a plumber is likely to know, but a young child is not). Keeping track of who knows what also allows us to navigate social relationships effectively. For example, if you once tell your friend "My son is getting married in June," it is usually communicating good news. If you repeat the same information to the same friend several times, it may instead signal anxiety, uncertainty, or disbelief. This project examines how people take into account the knowledge and perspective of other people in communicative settings and examines conversations among groups of three or more individuals, each of whom brings a different set of beliefs and knowledge to the conversation. Graduate and undergraduate students will have opportunities to participate in laboratory research and high school students and elderly individuals from the community will be involved in laboratory activities through community outreach programs. The results of studies on conversation will not only advance understanding of psychological and linguistic processes but can also be used to advance computer dialog systems (such as Siri). The proposed work uses methods established by the principal investigator to examine key open questions regarding language use in communicative settings. The idea is that conversational partners serve as contextual memory cues, supporting the partner-specificity of language use. This people-as-contexts (PAC) idea is used as a framework (1) to evaluate candidate hypotheses regarding the mechanisms supporting perspective-taking in multiparty conversation; (2) to test predictions concerning the binding of partners to language and generalization to new contexts; and (3) to examine how adaptation scales up to multiparty conversation. Predictions derived from the theory are tested in the domains of audience design and syntactic adaptation.