Context-update semantics views utterance meaning as how utterances change conversational context. Rather than assigning truth values relative to fixed context, it asks: what is the update effect of this utterance? This framework elegantly handles assertions, questions, and imperatives as different context-update types, explaining why presupposition is fundamentally about context change rather than truth-conditional semantics.
From your work in discourse representation theory you know that meaning extends beyond individual sentences — how we interpret an utterance depends on what has been established in the discourse so far. Context-update semantics sharpens this insight into a formal framework. Instead of asking "what proposition does this sentence express?" — the standard truth-conditional move — it asks "what does this sentence *do* to the state of the conversation?" Meaning becomes an operation on context, not an assignment of content to a fixed context.
The key technical notion is the common ground — the information mutually accepted by all participants in a conversation. Think of it as a shared whiteboard. When I assert "It's raining," I am not merely expressing a proposition; I am proposing to add that proposition to the whiteboard. If you accept the assertion, the common ground updates. If you reject it, the whiteboard stays the same. This move — treating assertion as a *proposal to update* rather than a *statement of fact* — solves a problem that classical truth-conditional semantics struggles with: it explains why assertion is a speech act, not just a display of information.
Presupposition is where context-update semantics becomes particularly powerful. Your prerequisite work in formal pragmatics introduced presupposition as information the speaker takes for granted. In the context-update framework, a presupposition is a *precondition on successful update*: an assertion can only update the common ground if its presuppositions are already part of the common ground. "The King of France is bald" presupposes there is a King of France; if that information isn't on the whiteboard, the update operation fails — not because the proposition is false, but because it cannot even be processed. This explains presupposition projection (why presuppositions survive embedding under negation and questions) without any special stipulation.
Questions and imperatives fit naturally into the framework as distinct update types. A question does not add a proposition; it transforms the common ground into an *issue* — an open set of possibilities that the discourse is now tasked with resolving. An imperative updates a different dimension of context: the set of obligations on the addressee. This multidimensional view of context — tracking facts, issues, and commitments simultaneously — unifies speech act theory with formal semantics, explaining diverse utterance types through a single principle: every utterance is a context-change instruction, differing in *which* dimension of context it targets and *how* it transforms it.