Since late 1960s, much evidence has been discussed in the literature that certain classes of verbs, be they morphologically simplex or derived, consist of more than one semantic component. Verbs like break, for example, can be thought of as composed of an activity of the external argument and a change of state of the internal argument. The precise content and properties of such components have long been a matter of debate in existing theories of event structure and predicate decomposition. In this presentation, I will review the recent literature, focusing on a family of theories that argue for a syntactically represented decomposition. More specifically, I will address the question of whether subevental components of an event description (e.g., the activity component or the change of state component) should be represented independently from relations between them (e.g. the causal relation). I will discuss a number of facts that can be taken to support this view: subevents and their relations are independent, since, first, their semantic properties vary independently, second, they can be spelled out by distinct morphological exponents, and third, the independence predicts correctly the observed cross-linguistic variation. This allows to eliminate the empirical problematic assumption that descriptive properties of subevent descriptions are inherently connected to characteristics of relations between subevents.