Emotion terms, category structure, and the problem of translation: The case of shame and vergüenza

Abstract

We conducted three studies aimed at showing that one-to-one translations between emotion terms might be comparing independent or barely overlapping categories of emotional experience. In Study 1 we found that the speakers' most accessible features of two supposedly equivalent emotions terms (shame and vergüenza) were very different. In Study 2, American and Spanish speakers' typicality ratings of 25 out of 29 constitutive features of “shame” or “vergüenza” were significantly different. In Study 3, these important differences in the content and internal structure of “shame” and “vergüenza” were also reflected in the affective meaning of their corresponding terms. We conclude that “shame” and “vergüenza” do not share the same cluster of central features, and cannot be treated as identical categories of emotional experience. This finding raises serious questions about the one-to-one translation practices not only of most cross-cultural studies on “shame”, but also of other emotion concepts. We suggest an alternative method of translation based on Diachronic Prototype Semantics (Geeraerts, 1997).

Keywords:

• Everyday concepts of emotion
• Shame
• Translation
• Cross-cultural

Acknowledgements

This research was supported by an FPU Grant to the first author from Spanish Ministry of Education, and partially supported by the Spanish Government's grant SEJ2005–06307-PSIC. Alejandra Hurtado de Mendoza is now at the Department of Anthropology, George Washington University.

Notes

In order to clarify when we are speaking about a term of emotion or the conceptual category of the emotion we use italic characters when alluding to the term (e.g., vergüenza, shame), and double quotation marks when alluding to the conceptual category or its features (e.g., “shame”). When we talk about an emotion in a general, unspecific way we do not include special characters (e.g., shame).

²Russell et al. (1989) use emotion words to exemplify the meaning of the different areas of the grid (stress, excitement, depression, and relaxation) but Casados' version only uses pleasant/unpleasant and high/low arousal labels.

³Wierzbicka's (1997) approach based on semantic primitives would be a potential second step in the process, aimed at getting a perfect, ideal translation between two terms. Unfortunately, this solution still requires elaboration in some key theoretical and practical grounds. On theoretical grounds, Wierzbicka's definition of a concept is maybe more culture-free than others but it is still a classical definition of necessary and sufficient features (e.g., shame is defined as “when one thinks that other people see something that one has done and that should not do & when one thinks that the other persons can think something bad because of that & when one wants not to be seen by others because of this”; Wierzbicka, 1986). This definition does not include features unrelated to the presence and evaluation of an audience that have been related to “shame” (e.g., Smith et al., 2002; Tangney & Dearing, 2002). Given that our approach to concepts is probabilistic and excludes classic definitions, we cannot adopt Wierzbicka's approach in its current version. On practical grounds, the development of Wierzbicka's culture-free universal language is still confronted with formidable problems: for example, it is not clear whether such universal language can capture all the nuances of local languages, and, most important, it is still controversial whether Wierzbicka's basic terms are truly universal (see Parkinson et al., 2005, for a review of these problems).