WhoLoDancE: Deliverable 3.2 - Report on emotion representation models

This document summarizes how we address expressivity in dance practice and teaching, in the framework of WhoLoDancE project. The aim of this document is to define the appropriate models that will be used in the Learning scenarios through an innovative but also meaningful way for dance practitioners for the different dance genres that we investigate: contemporary, ballet, Greek folk and flamenco. For this reason, we present the state-of-the-art in human movement and emotions computing, and we explain the rationale and methodology of our approach. In this context, we consider dance expressivity to be conveyed through qualitative aspects of non-verbal communication and we focus on the analysis of "movement qualities", a term which is widely accepted by both dance practitioners (especially in contemporary dance) but also in movement and gesture computing and Human-computer interaction. This term is used to describe, not the action per se or the shape of the movement, but the way the movement is performed, the dynamics of the movement which make it expressive and meaningful. Though it is true that movement can convey emotions, in none of the dance genres we examine in the context of WhoLoDancE, there exists a one-to-one relationship between emotions and movement, not even Movement Qualities. In addition, though the term “movement qualities” is accepted by both the Dance and Movement Computing communities, they do not include a specific, unique, list of terms that is accepted by all dance practitioners in a universal manner. In this context, part of our work in WhoLoDancE is to define in cooperation with the dance experts (internal and external) an acceptable subset of Movement Qualities that are both meaningful and subject to computation through analysing the multimodal recordings. This document presents the definitions of appropriate emotion models, including associations with music and expressive movement and dance. It also summarizes the outcomes of “T3.1.2 Development of emotion representation models” documenting the definition of the appropriate emotion/movement qualities models for linking, integrating and annotating the multi-modal learning material with high-level emotion-related descriptors. Finally, we explain how we plan to analyse the relationship of these features with the different modalities (music, sound, visual metaphors, different avatars) and how this affects the understanding of Movement Qualities through the Learning Scenarios.