RESUMO
This article provides semantic differential ratings of 1,469 concepts in Bengali, a language spoken by about 250 million individuals in eastern India and Bangladesh. These data were collected from 20 male and 20 female Calcutta respondents who rated stimuli on three culturally universal affective dimensions: evaluation-potency-activity (EPA). This study employs pan-respondent component analyses as a means of examining the respondents' usage of the standard EPA scales. The pan-respondent component analyses indicate that some respondents used the rating scales in unexpected ways, recording their feelings about one component of concepts' EPA with ratings on a scale intended to measure a different dimension. When scores were based only on respondents who used the scales appropriately, several interesting patterns were found. For respondents of both genders, potency scores have a curvilinear relation with evaluation, such that very good and very bad concepts are mostly seen as very potent, whereas evaluatively neutral concepts are seen as somewhat impotent or just slightly potent. A moderate linear correlation exists between activity and evaluation, and a modest positive relation exists between potency and activity. Gender correlations are high on evaluation, .93, but much lower for potency scores, with a correlation of .55, and even lower for activity, .30. In this article we examine several explanations for why scales denoting potency and activity were reinterpreted as indicating goodness by certain respondents, and consider the matter of including data collected from respondents who used scales in this way.