Gesture structure, insights from annotation guidelines

Abstract

This paper analyzes the internal structure of gestures, namely preparation, stroke, and recovery, by making a close reading of early (1970’s onwards) definitions. Those definitions are centered around segmenting the gesture based on saliency, creating an opposition between the stroke, the most well-defined part of the gesture, and other phases. Although this definition initially relied heavily on the association to speech, the need to operationalize the annotation forced definitions more linked to articulation features (Linguistic Annotation System for Gestures, Bressem et al., 2013). Later studies, mainly linking gestures to phonological structures of speech, started to rely on gesture kinematics to define gesture structure, which are motion-based features such as velocity profiles and movement targets (Kita et al., 1998; Loehr, 2004; Rohrer et al., 2023). The latter provides a qualitative base for the use of motion tracking technology in gesture studies, leading to the interpretation that the stroke can be better defined by its velocity profile than in connection with articulation features. In terms of theory, this impacts how gestures can be segmented and understood in language, without necessarily needing to fully integrate the phonology of the utterance it is associated to. It seems that gesture-speech association depends on the transparency it shows to its speech affiliated, moving from a tight synchrony for less representational and beat gestures, to a loose synchrony for iconic and recurrent gestures.