English | French

In the last few years, a growing number of studies have centered on speech patterns in conversational interaction. A major issue of interest in these studies is the tendency shown by participants in a conversation to imitate each other. Imitation seems to occurs at every level of the conversational exchange, and that includes the phonetic level. For example, it has been shown that perceived similarity in pronunciation between talkers increased over the course of a conversation and persisted beyond its conclusion.

Imitative speech behavior is a phenomenon that may be actively exploited by talkers to facilitate conversational interaction. It is consistent with the idea that listeners are sensitive to talker-dependent fine-grained phonetic characteristics, which have an influence on both the dynamics of conversational interaction, and across memory, when that interaction has ended.

This project provides a contribution that will be both novel and important in several respects. First, and although it has long been assumed that phonetic convergence effects are recurrent in conversational interaction, no study yet has investigated these phenomena in speech in a systematic and extensive manner. To undertake these investigations, we take an original, cross-disciplinary approach that aims to cover the many factors that govern imitative speech behavior, from the sensori-motor to the pragmatic level. Large-scale analyses on spontaneous speech corpora are performed in conjunction with experimental studies with the help of a wide variety of technical and methodological resources developed in our research teams over the years. The proposed research consortium is in our view in a privileged position to address these issues, given its joint expertise in conversational analysis.

Our research is likely to have empirical and theoretical implications for a number of crucial issues in spoken language production and understanding. It will shed new light on the relationships that can be established between perception and action in speech, and will in that respect contribute to the work carried out by the GIPSA group in the framework of the PACT theory (Perception-for-Action-Control, see Schwartz et al., 2007). The project will also help identify fine-grained phonetic properties that convey indexical information about the talker's identity in speech. It will have an important bearing on the current debate on the nature of the representations associated with words in the mental lexicon, and on the plasticity of these representations in mature talkers (Dufour, Nguyen & Frauenfelder, 2007). Our project will stand at the interface between phonetics and phonology in seeking to detect the formation of new generic and abstract phonological patterns that may be triggered by phonetic convergence in conversational interaction. It will also aim at studying imitation in speech as a goal-oriented behavior that performs a number of pragmatic functions central to the dynamics of the conversational exchange. In doing so, this project will endeavor to investigate speech patterns in their primary site of occurrence, namely social interaction, in the framework of the conversational analysis framework developed by the ICAR group.


Work Packages

General overview
Work Package 1
Work Package 2
Work Package 3
Work Package 4
Work Package 5
WP1: Speech databases for the study of phonetic convergence
WP2: Phonetic and phonological characteristics of imitation
WP3: Phonetic convergence and sensori-motor integration
WP4: Phonetic convergence and the sound shape of words in the mental lexicon
WP5: Role of phonetic convergence in conversational interaction
Contact the Webmaster