Listeners engaged in a dialogue face the daunting task of having to prepare a response to the speaker’s utterance before this utterance is complete. They also have to time their contribution such that its beginning is temporally aligned with the end of the current utterance. We will study this ability to anticipate both timing and content of interlocutors’ turns using Conversation Analytic methods, psycholinguistic RT experiments and neurocognitive (EEG/ERP) methods. We focus on identifying the relative role(s) of lexical, syntactic, semantic and pragmatic information in temporal alignment in dialogue through anticipation.

The principal goal of the project is to establish how listeners in a dialogue are able to accurately anticipate both the content and the timing of the turn that is currently produced.

Both the off-line experimental work by Schaffer (1983) and the on-line button-press study by De Ruiter, Mitterer and Enfield (2006) found that listeners use lexico-syntactic content to anticipate turn endings. Magyari and De Ruiter (2008) found in an off-line task that listeners are indeed able to predict the content of natural conversational turn, and that this ability correlated with the accuracy with which listeners could anticipate the end of the turn. This raises two specific research questions.

  1. Do listeners in a dialogue use syntactic information, semantic information or both for end-of-turn anticipation?
  2. Is the process of anticipating the content of a turn the same as the process of anticipating its end, as Magyari and De Ruiter (2008) hypothesized?

To find answers to these questions, we will use a combination of three well-established methods. First, Conversation Analysis (see Sacks, Schegloff & Jefferson, 1974), for grounding the phenomenon of turn-anticipation in data from natural interactions. Second, classical reaction-time like experiments to assess the accuracy with which participants are able to anticipate turn endings (De Ruiter et al. 2006), and thirdly, electrophysiological recordings (Weiss & Müller 2003) to be able to obtain differential on-line information about the semantic and syntactic processing in listeners.