Academic journal article Trames

The Structure of Estonian Concepts of Emotion: A Self-Organizational Approach

Academic journal article Trames

The Structure of Estonian Concepts of Emotion: A Self-Organizational Approach

Article excerpt

1. Introduction

Studies of the emotion lexicon with the aim of discovering the structure of emotions have enjoyed great popularity in recent decades. Most of the studies rely on a default assumption of lexical isomorphism (e.g., Church, Katigbak, Reyes, and Jensen 1998, Fontaine, Poortinga, Setiadi, and Markam 2002, Frijda, Markam, Sato, and Wiers 1995, Johnosn-Laird and Oatley 1989, Russell, Lewicka, and Niit 1989, Shaver, Schwartz, Kirson, and O'Connor 1987, Zammuner 1989). There are, however, other authors who claim that the study of words is not sufficient for discovering the structure of affect (e.g., Cacioppo and Berntson 1999, Feldmann Barrett 2004, Fillenbaum and Rapoport 1971, Ortony, Clore, and Collins 1988), or that its sufficiency cannot be taken for granted (e.g., Haslam 1995, Storm and Storm 1987). In the recent literature it is not obvious what is really under observation in the studies of the emotion lexicon, that is, whether emotions or words. Another unclear issue is whether the structure of emotional experience and emotion vocabulary should coincide or not.

In the present paper we assume that the general representations of emotions, accessible by means of different statistical methods of data analysis, are conceptual by nature. Concepts of emotion may not be independent entities but rather abstractions from experience or words. As such, they can be understood and studied both from the level of experience and from the level of verbal expression. These two alternative approaches may influence the outcome of data analysis. When emotional experience is taken as a starting point (like in emotional self-ratings), the results tend to reveal the representation of an emotional-motivational system, where positive and negative affects exist in a mutually nonexclusive way as two kinds of activation (Watson, Wiese, Vaidya, and Tellegen 1999). However, when emotion words are taken as a starting point (as in word similarity judgements), the results are best represented by an emotional circumplex of two bipolar crossing dimensions of activation and valence (Russell 1980, Russell et al. 1989).

In this paper, we focus directly on a set of Estonian concepts of emotion and study them from both these levels. Our purpose was to find out whether the level of access might affect the emergent structure of concepts of emotion as far as can be inferred from self-organising maps (SOM). Two lexical tasks were carried out which addressed the set of concepts firstly on the basis of their relation to the quality of experience, and secondly the semantic interrelations of words. The results revealed that the way the data was gathered partly affected its layout in SOM. In both cases we found two main clusters, that of positivity and negativity among the studied concepts. In addition, a cluster of alertness emerged while the concepts were accessed through the level of interrelations of emotion words. In terms of local neighbourhood relations, a partial heteromorphism of the two structures was detected.

2. A case study of Estonian concepts of emotion

Previous research into the Estonian emotion lexicon has pointed to the possibility of differences in the experiential and semantic knowledge of emotions (Vainik 2006). Some structural divergence has also been found in tasks which address emotional self-report (Allik 1997, Allik and Realo 1997, Veski 1996) and word similarity judgments (Kastik 2000, Russell et al. 1989). The results of previous research are consistent as far as the general division of the lexicon into two subcategories of positive and negative emotions is concerned. However, there are inconsistencies in the details, and no satisfactory resolution has been found for the question of whether the dimensions of positivity and negativity should be treated as mutually exclusive bipolar opposites, or as possibly independent and co-occurring measures of two unipolar dimensions.

We suspected that the level of access (lexicon-related vs. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.