Most works on Aktionsart and Aspect presuppose that stative verbs form a homogeneous class and that they can be easily identified by distributional tests based on (in)compatibility with the imperative (I), with the progressive periphrasis (PP) and with certain types of time adjuncts (TA). In the first sections of this paper, we analyze data from Brazilian Portuguese (BP) that goes against this view. We show that some statives are compatible with PP, or with I, or with both, and also with the kinds of TA that have been considered problematic. In the last sections of this paper, elaborating on an idea first proposed by Professor John Schmitz, we claim that, as far as BP is concerned, the class of statives must be sub-divided into sub-classes whose behavior vis-à-vis I, PP and TA can be predicted by the interaction of semantic features such as [± control ] and [± change ].