Early Anticipation Lies behind the Speed of Response in Conversation

Abstract

Abstract■ RTs in conversation, with average gaps of 200 msec and often less, beat standard RTs, despite the complexity of response and the lag in speech production (600 msec or more). This can only be achieved by anticipation of timing and content of turns in conversation, about which little is known. Using EEG and an experimental task with conversational stimuli, we show that estimation of turn durations are based on anticipating the way the turn would be completed. We found a neuronal correlate of turn-end anticipation localized in ACC and inferior parietal lobule, namely a betafrequency desynchronization as early as 1250 msec, before the end of the turn. We suggest that anticipation of the otherʼs utterance leads to accurately timed transitions in everyday conversations. ■

Coordinating Utterances During Turn-Taking: The Role of Prediction, Response Preparation, and Articulation

“…Predicting this point enables the listener to articulate their response at the appropriate moment, so the risk of extensive conversational overlap or a long gap is minimized. Most research assessing turn-end prediction has used a button-press paradigm (e.g., De Ruiter et al, 2006;Magyari et al, 2014), in which participants are presented with full conversational turns and predict (indicated via a button-press) when they expect speakers to reach the end of their utterance. Using this paradigm, De Ruiter et al found that participants could accurately predict when a speaker's turn was about to end.…”

“…Further evidence for this conclusion comes from a study by Magyari et al (2014), who manipulated the content predictability of their stimuli. In a gating paradigm, participants were auditorily presented with turns from actual conversations in fragments of increasing duration and completed these turns with the words they expected to follow the preceding context (much like a typical cloze task; Taylor, 1953).…”

“…Using the same gating method as Magyari et al (2014), Magyari and De Ruiter (2012) assessed the number of words participants expected to complete sentence fragments. They correlated these results with De Ruiter et al 's (2006) and found that turns that elicited later button-presses tended to be completed with more words in the gating paradigm.…”

“…Regardless of whether listeners launch articulation of their response on the basis of turn-final cues (Beattie et al, 1982;Gravano & Hirschberg, 2011;Hjalmarsson, 2011;Local & Walker, 2012) or a prediction of the speaker's turn-end (De Ruiter et al, 2006;Magyari et al, 2014), failure to fully comprehend the speaker's utterance because of preparing, buffering, and rehearsing a response may impact the listener's ability to respond at the appropriate moment, which may lead to more overlaps or gaps between utterances. On the other hand, the speaker's utterance generally becomes increasingly predictable as the turn unfolds.…”

“…Research has focused on how the listener can use such predictions to determine when the speaker will reach the end of their turn (e.g., De Ruiter et al, 2006;Magyari, Bastiaansen, De Ruiter, & Levinson, 2014) so they can articulate their response at the appropriate moment. However, if listeners can predict what the speaker will say before the speaker reaches the end of their turn, then the listener can also begin preparing their own response in advance (Garrod & Pickering, 2015;Levinson, 2016).…”