The evidence on translatability
Although there is a very wide literature examining cross-cultural differences in subjective well-being, there is little in the published literature that specifically reports on the influence of translation. What we do know in this field raises some concern - for example, in the case of evaluative measures, Bj0rnskov (2010) states that the English word happy is “notoriously difficult to translate”, whereas the concept of satisfaction better lends itself to precise translation (p. 44). Veenhoven (2008) equally notes that perfect translation is often not possible: “If it is true that the French are more choosy about how they use the word “happy”, they might place the option “’tres heureux’ in the range 10 to 9, whereas the English raters would place “very happy’ on the range of 10 to 8” (p. 49). When Veenhoven tested this issue with Dutch and English students, the Dutch rated “very happy” as being equivalent to 9.03 on a 10-point scale, whereas the English rated it at 8.6 on average.
Factor analysis is often used in selecting the most appropriate wording for multiple- item scales to test whether all items appear to be measuring the same underlying construct. For example, Thompson (2007) used a combination of bilingual focus groups and factor analysis to develop a short version of the Positive and Negative Affect Schedule. On the basis of evidence gathered from 12 different nationalities, the words enthusiastic, strong, interested, excited, proud, scared, guilty, jittery, irritable and distressed were rejected from the final measure, either because they had ambiguous meaning for some respondents (for example, both positive and negative connotations) and/or because they performed poorly in terms of factor structure. Ultimately, the words alert, inspired, determined, attentive and active were selected to measure positive affect, and the words upset, hostile, ashamed, nervous and afraid were selected to measure negative affect.
Although precise question wording is usually helpful for communicating concepts to respondents, broad but easily-translatable descriptors may be particularly helpful to produce internationally-valid affect measures, where some emotion words may be too narrow or culturally-specific to generalise across samples. For example, Diener et al. (2009) make a case for including very broad descriptors for positive and negative feelings in affective experience measures to avoid the potential omission of feelings that are not easily translated, particularly those that might be more important in some cultures or to certain individuals. As a result, their 12-item Scale of Positive and Negative Experience (SPANE) includes quite generic items about feeling positive, negative, good, bad and pleasant and unpleasant.