This paper aims to reform the notion of indirect speech acts under the weak version of the Literal Force Hypothesis. This hypothesis states that all utterances have a direct illocutionary force built into them, either through the use of an explicit performative, or by virtue of their sentence types. The weak version of this hypothesis states that every sentence type is compatible with a diverse range of illocutionary values, for instance, declaratives are considered to be the best vehicles for the performance of any type of illocutionary act. Moreover, it is claimed that the role of metonymy is crucial in providing inferential schemata that can be exploited in the performance of ISAs. For example, in the request scenario there is a limited number of semantic notions contributing to the conceptual characterization of this illocutionary category: desire, capacity, obligation, and result/consequence. Another important argument of the paper is that requestive constructions introduce the notions of illocutionary ICM and of (multiple source)-in-target metonymy. Specifically, the illocutionary ICM for requests is governed by the high-level metonymy SPECIFIC FOR GENERIC.