Revisiting the dimensional structure of the emotion domain

Abstract

Recent research has claimed that a novelty dimension is needed to represent the cognitive emotion structure over and above valence, power and arousal. Novelty emerged when student samples evaluated the meaning of 24 emotion terms on 142 emotion features. This claim is debatable, however, because to date novelty has never been found in similarity sorting studies. It is possible that novelty emerged because sophisticated student samples evaluated emotion terms on emotion features. The current research identified a large, representative set of emotion terms using a free-listing task in a middle childhood up to early adulthood sample (N = 5071). Children, adolescents, students and adults (N = 1184) then evaluated the similarity between these emotion terms using a similarity rating task without priming any emotion feature. Novelty robustly emerged as the fourth dimension. The existence of novelty is thus confirmed with a different method across a wide variety of participants.

Keywords:

• Free listing
• Emotion lexicon
• Similarity rating
• Dimensional structure
• Novelty

We would like to thank Dr Dimitri Van Ryckeghem for his practical support and his valuable comments on previous versions of the manuscript.

We would like to thank Dr Dimitri Van Ryckeghem for his practical support and his valuable comments on previous versions of the manuscript.

Notes

¹ It was possible that an uttering contained information about more than one category.

² No additional terms were selected because for each added term the pairwise similarities with all other terms had to be rated. Adding terms would have affected the feasibility of the pairwise similarity rating task.

³ A common base of terms was needed for technical reasons. In order to do MDS analyses with data from individual participants, some pairs had to be present for all participants.

⁴ Preliminary results of the student sample have been reported elsewhere (Fontaine & Veirman, 2013).

⁵ To investigate whether the marker terms themselves had an influence on the emotion structure, a four-dimensional structure was computed on only the 76 emotion terms. After orthogonal Procrustes rotation the dimensions of this 76-words structure correlated .999, .997, .994, and .966, respectively, with the dimensions of the 85-words structure. This implies that the reported four-dimensional structure is not affected by the inclusion of the marker terms.

⁶ Contrary to the expectation, both actief (active) and vol van energie (full of energy) showed higher coordinates on the power and valence dimension than on the arousal dimension.

⁷ The coordinates of the replicated MDS were used as the coordinates of the overall structure in the weighted MDS.

⁸ As the dimension weights generated by the weighted MDS are not independent from one another, the current findings do not mean that older participants use power, arousal and novelty less than younger participants, but that they use these dimensions less compared to valence.