In this study, we explore three ways in which the notion of "granularity" emerges from the study of cross-linguistic event semantics. The first interpretation of granularity has to do with event segmentation for linguistic expressions. Where humans place event boundaries varies depending on the language and cultural setting in which the event is encoded. Second, within the set boundaries of a 'single event' in time there are many elements that could all receive expression, and again languages show variation, both with respect to which elements are mentioned at all, our second interpretation of granularity, and the specificity with which these elements are characterized, our third interpretation. It remains a matter for empirical research to discover whether these differences in linguistic expression reflect or inform different representations of event boundaries and event classification for nonlinguistic purposes as well.