Request Correction

Multimodal Entrainment and Feedback in Online Group Meetings

This paper presents the results of a study on multimodal speaker behaviour in a corpus of online Zoom meetings. We investigate two questions: i) whether speakers display a higher degree of head movement when they exchange verbal feedback than when they don’t, as would be expected if verbal and gestural feedback reinforce one other, and ii) whether they move more or less similarly under the same conditions. Several linear mixed models were fitted to test the difference in head movement values in target and control intervals of two different durations. The results indicate that speakers indeed entrain by moving their heads more in target intervals where verbal feedback is present. This result confirms our expectations. However, speakers also appear to move in less similar ways in the same target intervals. This dissimilarity can be explained by the fact that not all speakers give the same type of gestural feedback, but also by noise created by non-communicative movements in which speakers adjust their positions or reach out for objects during the meeting.