The aim of the paper is to explore the proposals as delineated in the FrameNet Project (Atkins, Fillmore and Johnson 2003; Fillmore, Johnson and Petruck 2003), allowing for Levin’s (1993) entity-specific change-of-state verbs.The author points to several drawbacks found in the examined database: (i) a limited number of entity-specific change-of¬state verbs were listed in FrameNet (eight out of twenty-one verbs); (ii) the database stresses the separation of the causative and inchoative uses of a verb into two different frames; (iii) some cases lack examples provided for the Frame Elements of a particular verb (e.g. there are no examples for the verb swell in the ‘expansion’ and ‘change of position on a scale’ frames); (iv) FrameNet usually includes literal instantiations of the predicates under consideration, but lacks cognitive motivation for the metaphorical expressions whenever figurative uses are listed; (v) frames often prove incomplete owing to the use of a small size corpus (i.e. the British National Corpus),which results in a limited number of examples for a given verb. Moreover, the article proves that metaphor and metonymy play an important role in regulating the subsumption processes between predicates and constructions.