In a transitive change-of-state statement, the VP describes an event yielding to some new state. The way we identify this event very much depends on argument structure, as well as the ontological and agentive properties of the external argument if present. We argue that only in the presence of a normal, in-control and intentional agent, the VP-event is `bigger' than the change proper: it also contains the action performed by the subject. As a result, with normal, in-control agent, the VP-event can in principle start before the change itself starts. But with all other types of external arguments --- be it an inanimate agent, an inadvertent (animate) agent, a causer (denoting an eventuality or a fact) or the simple possessor of the dimension denoted by the object ('The room changed its temperature'), the VP-event is identified as a mere change. The rest of the causation event is either overtly denoted by the subject, or (softly) presupposed by the causative statement. Whether the VP-event contains more material than just the change proper or not has significant consequences on the interpretation of change-of-state statements built with an aspectual morphology with partitive semantics, such as the progressive in English.