This article which questions whether coercion can be answered for in the natural language context, analyzes individual cases of coercion, and examines whether metonymic inferencing processes can be recapitulated in pragmatic terms, or otherwise in the light of their diachronic growth and grammaticalisation.