Abstract
Since the early 80s, metonymy has progressively gained central stage in linguistic investigations. The advent of cognitive linguistics marked a new turn in the study of this trope conceived, not as a deviation from semantic conventions (contra classical rhetorical theories), but as a phenomenon rooted in non-language-specific mechanisms of conceptualization and structuring of the world. Focusing on the particular case of whole-for-part (WP) metonymy, the general stand of this presentation will be to argue for the need to re-inject properly semantic (descriptive and formal) considerations in the study of metonymy. It will be shown that semantic typologies of meronymic relations, as well as recent results in the semantics of partial predicates and colour predicates, can be jointly exploited to attain a better understanding of the mechanism of WP-metonymy. The resulting view will in particular connect such metonymies to mechanisms of reference and property attribution to functional parts for classificatory purposes.