The Role of Affect in the Computational Modeling of Natural Language

Abstract

Expressivity is an intrinsic component of natural language. This article follows the tradition of affective computing (Picard 1997) in using affect to refer to connected concepts such as emotion, mood, feelings, personality, attitude, polarity, and related subjective phenomena. The overview clarifies the relevance of affect for linguistics and computational linguistics, summarizes useful background, outlines helpful resources, and highlights important considerations for computational modeling of affect in language and affect‐related linguistic behaviors. In addition, the article sketches unsettled debates, topics, problems, and areas in need of further exploration.

Language, emotion, and the emotions: A computational introduction

“…First of all, we concentrate on the subset of issues related to language proper, not, for example, skin conductivity or body language; even in the language realm, we will only slightly touch upon expressivity in speech, associated gestures, or patterns of interaction, in spite of their utmost importance in research today—see e.g. Alm () for those subjects. We will be mainly dealing with models of emotions inspired by computation, models of the lexicon, processing models, and linguistically driven discovery processes, emphasizing different emotion models and different languages and cultures.…”