Pragmatics is the last and most socially embedded layer. It begins where the literal, context-independent meaning studied by semantics stops. The sentence “It is cold” can be a report, a request to close the window, a complaint about the heating, or a suggestion to go home, and the meaning depends on the speaker, hearer, and situation. Thus, pragmatics studies the context-dependent and non-literal aspects of meaning that arise in the act of utterance, and it treats speaking as a form of action.
A conversational implicature is not entailed by the sentence but can be reconstructed by contextual reasoning and embodied experience. A presupposition is a background assumption that exists outside the current situation available to the speakers.
Speech act theory analyses what an utterance does as a multi-layered approach: a locutionary act produces sounds, words, and structure, and therefore draws on phonology, morphology, and syntax. The propositional act contributes content through reference and predication, which is the domain of semantics. The illocutionary act carries the speaker’s intention, and the perlocutionary act is the effect on the hearer. Searle’s defines assertive classes: directives, commissives, expressives, and declaratives, and sorts illocutions by their point.
Grice’s maxims explain how hearers reach implicatures. Under a cooperative principle, speakers are expected to be informative, truthful, relevant and clear, and hearers infer meaning from apparent violations. Flouting a maxim is often communicative, and the critique that implicatures also arise in uncooperative settings shows the model’s limits. With phonology, morphology, syntax, semantics and pragmatics in place, the next step is to see what happens when these layers are analysed automatically at scale, which is the task of computational linguistics.