Emotions and intentions mediated with dialogue acts

Recently, there is an increasing interest in affective computing approaches where the complex interplay of the speaker's intentions and emotions is analyzed and used to direct various human interaction processes. The presented study is an attempt to further investigate along these two social cognition- regulating dimensions, mapping dialogue acts, which are pragmatic-level units representing intentions of the speaker, to emotions. The mapping is achieved by annotating transcriptions of in-game Japanese conversations with tags of basic emotions and dialogue acts exemplifying the social context. A statistical analysis of the emotion-dialogue act co-occurrences reveals several dependency patterns, where certain dialogue acts (and, therefore, the underlying intentions) are associated or disassociated with one or more emotions. Results obtained are discussed, conclusions are drawn, and plans for future work are formulated at the end of the paper.