Corps, R. , Crossley, A. , Gambi, C. & Pickering, M.
University of Edinburgh
During conversation, interlocutors switch between listener and speaker roles and take turns at talk. These processes are so finely coordinated that there is often little gap between utterances. To achieve such coordination, listeners must be predicting when the speaker will reach the end of their turn. Previous research has assessed turn-end prediction using a button-press paradigm, and demonstrates that listeners predict turn-ends by predicting the forthcoming words (or content) of an utterance (de Ruiter et al., 2006). However, this research has yet to consider the role of utterance length (Magyari & de Ruiter, 2012). Additionally, the button-press paradigm is not necessarily representative of a natural conversational setting, where listeners have to predict the turn-end and also prepare a verbal response. We address these issues in two Experiments, where we manipulate the content and length predictability of utterance. In Experiment 1, turn-end prediction was assessed using the button-press paradigm. In Experiment 2, turn-end prediction and response preparation was assessed using a question-answer paradigm. We find no evidence to suggest participants predict the length of an utterance. However, the role of content predictions depends on response method: content predictability is important when preparing a verbal response, but not when predicting turn-ends.