The paper presents a study on figurative language as a serious challenge for automatic text understanding. A system capable of interpreting figurative expressions would be an invaluable addition to real-world natural language processing (NLP) applications that need to access semantics, such as machine translation, opinion mining, question answering and many others. Focusing on one type of figurative language, i.e. logical metonymy,this study presents a computational model of its interpretation bringing together statistical techniques and the insights from linguistic theory, which,compared to previous approaches, is both more informative and more accurate in the production of sense-level interpretations of metonymic phrases and the automatic organization of them into conceptual classes, or roles, discussed in the majority of linguistic literature on the phenomenon.