Abstract
Focusing on the gestures that people produce in co-occurrence with speech during a communicative exchange, this paper provides a useful methodological approach to the annotation and analysis of gestural data using new technologies available for multimodal research. To this end a corpus made up of authentic data retrieved from video recordings of the United States presidential election debates was compiled, in which gestures are employed strategically by candidates in order to reinforce their verbal messages for rhetorical and persuasive purposes. Our research sheds light on different issues regarding the process of compilation and preparation of the corpus, including the use of a gesture annotation scheme that can be extrapolated to other dialogic contexts with similar communicative goals. Design considerations and other major results of this corpus-based approach will also be discussed, along with relevant limitations of this experience specifically grounded on both the human and financial resources available.
