Inscribed within the framework of a larger and ambitious project addressing the relationships between affective phenomena (as units of personal experience), concepts (as units of cognitive processing) and the lexical items used in talking about affective experience (as units of verbal communication), this study focuses on levels and patterns of categorization in the domain of affectivity. Cross-cultural comparable data collected in several Romance (Spanish, Italian, Romanian) and Scandinavian (Danish, Swedish, Norwegian) languages with the help of a series of experiments inspired by Fehr and Rusell's prototype approach to emotions are analyzed from a perspective that goes beyond the interpretation of prototype theory and aligns with the theory-based approaches to categorization (cfr. the seminal paper by Murphy and Medin, 1985). Due to space limitations, only a short overview of the data obtained with the help of a supplemented version of the traditional free-listing task (a ranking and a definitional-inferential task being added) will be considered here. 10 different sets combining quantitative and qualitative data have been collected, corresponding to the various relevant superordinate categorizations in the 6 languages, with the Romance languages and Norwegian having 2, and Danish and Swedish only 1. Special attention is devoted, not as much to frequency of mention, order of mention and indexes of salience - parameters that are usually discussed in categorization and cultural domain studies - but first and foremost to the patterns of clustering and the relationships holding between the various categories mentioned by every single informant and across the lists, as made obvious by the data. Domain access point is identified as a privileged position that tends to coincide with categories promoting richest conceptual connections with other category members, while the prevailing relationships between mentioned categories appear to be either lexical (synonymic, antonymic) or conceptual (causal, metonymical).