MusicalExpression ofEmotions: ModellingListeners' Judgements ofComposed andPerformedFeatures

Abstract

ABSTRACTMusic is commonly regarded as expressive of emotions that can be perceived by listeners. Nevertheless, the specific characteristics of this perceptual process are not well understood. This study aims to investigate the relationships between various features of musical structure and the emotions perceived by listeners, with a focus on the role of interactions among such features. Eight musical features (pitch, mode, melodic progression, rhythm, tempo, sound level, articulation and timbre) were systematically manipulated in a factorial design through synthesis. Ten musically trained listeners judged the resulting 384 pieces of music on five emotion scales. The relationships between musical features and listener judgements were modelled by means of multiple regression analysis. The results (1) confirmed empirically based predictions from previouspost hocanalyses with respect to which musical features are associated with each emotion; (2) suggested that different musical features were important for different emotions; (3) indicated that some features (e.g. tempo) were more powerful than others overall; and (4) revealed that interactions made significant but small contributions to the predictive power of the regression models.