Analysing and predicting turn changes in communication

Abstract

This dissertation deals with the analysis of the turn-taking system in natural conversation between two Maltese participants engaging in small talk. The analysis was done using a video from the MAMCO Corpus (Paggio and Vella, 2014) whereby participants interacting show multimodal cues in conversation. The data was analysed in a manner that includes head gestures, hand gestures, shoulder movements, posture shifts and intonation changes linked to turn-taking points. The analysis particularly focused on what type of gestures are attributed to turn changes, what intonation can convey regarding turn changes and what cues are specifically related to the feedback channel. The analysis includes both a quantitative and a qualitative approach. The data was annotated using the video annotation tool, ANVIL (Kipp, 2003). Features from the data were then extracted and analysed statistically using RStudio (RStudio Team, 2020) to find out how dependent the relationship is between the turn-taking system and certain gestures, and therefore, to find out how the turn-taking system can be predicted through gestures used in conversation. This analysis showed that head movements and intonation are particularly prominent in the turn-taking system and noticeable, also, in the feedback channel.