Seeing hand gestures’ early beginnings improves predictions of upcoming utterance meaning

In face-to-face conversation, people convey meaning multimodally, including speech and hand gestures. Previous work showed that comprehenders can use gestures to predict utterance content, but it is unknown how much of a gesture comprehenders need to see for this. Here, we test whether the gesture preparation phase – traditionally deemed to be a mere prelude to the meaning-bearing part of a gesture – does in fact contain semantic information that can be used for prediction. Participants watched videos of speakers from a conversational corpus up to the end of a gesture preparation and predicted how the speaker would continue. Crucially, the preparation was either shown or obscured. Seeing the preparations did not lead to participants more frequently predicting the exact speech of the original utterance, but it did make the meaning of predictions more accurate. This suggests that even early gestural information can contribute to shaping comprehenders’ predictions, which may facilitate rapid conversational turn-taking.