Assertion
Reference: Stalnaker, R. C. (1978). Assertion. In P. Cole (ed.), Syntax and Semantics 9: Pragmatics (pp. 315–332). New York: Academic Press. Reprinted in Stalnaker, Context and Content (Oxford UP 1999). PhilPapers record · SEP: pragmatics · SEP: presupposition
Summary
Stalnaker’s “Assertion” gives the canonical update-semantics account of what a speech act of asserting does, framed in terms of a common ground of mutually accepted propositions shared between conversational participants and a context set of possible worlds compatible with that common ground. To assert a proposition p is, prima facie, to propose that p be added to the common ground — equivalently, to propose that the context set be narrowed to its intersection with the set of worlds in which p is true. The hearer either accepts (the context set is updated) or challenges (the proposal stalls, the common ground is unchanged). The picture is symmetric, trace-checkable, and pragmatic: it tells you what an assertion does without taking a stand on the speaker’s inner mental state.
The paper develops three essential effects of assertion. (1) An assertion of p adds p to the common ground unless it is challenged. (2) The proposition asserted should be determinate — true or false at every world in the context set; if not, the context set must be re-coordinated first (anticipating the accommodation mechanism worked out in Lewis 1979). (3) The proposition asserted should partition the context set non-trivially: not be common-ground-true already, not be common-ground-false already, and not be context-incoherent. This third effect is the seed of Gricean informativeness as a built-in feature of the dynamics rather than a separately stated maxim. The framework anticipates the entire dynamic semantics tradition (Heim, Kamp, Veltman) and is the operational counterpart of Grice 1975: where Grice gives maxims governing rational conversation, Stalnaker gives the update mechanics conversation operates on.
For agent communication, Stalnaker is the philosopher who most directly anticipates trace-checkable assertion semantics. The common ground is a publicly observable object — what has been said and accepted — and assertion is a publicly observable update operation on it. There is no need to inspect any agent’s beliefs; the semantic content of an assertion is given by the update it proposes, and conformance with that semantics is checkable from a transcript. The natural pairing in this vault is with Singh 1998 on social-agency semantics, Fornara & Colombetti on commitment-based operational semantics, and ultimately CBCL — the common ground / context-set apparatus is precisely the kind of shared, monotone, trace-checkable object that monotone-verdict commitment protocols want.
Key Ideas
- Common ground: the body of propositions mutually accepted (not necessarily believed) for purposes of conversation. The mathematical reflex is the context set — the set of worlds compatible with the common ground.
- Assertion as context-set update: an assertion of
p proposes narrowing the context set to the intersection with [[p]]. If accepted, the update goes through; if challenged, it stalls.
- Three essential effects of assertion (the “essential effect” inventory): (i) the asserted proposition should be added to the common ground on acceptance; (ii) it must be determinate over the context set (true or false at each world); (iii) it must partition the context set non-trivially (informative).
- Diagonalisation / two-dimensional accommodation: when an assertion’s content varies across worlds in the context set, the hearer can use the diagonal proposition — the function picking out, at each world, the content the assertion would have at that world — to recover a determinate update. This is the engine behind much of dynamic semantics’ treatment of indexicals and modals.
- Pragmatic presupposition: a speaker presupposes
q iff, in uttering an assertion, the speaker takes q to be in the common ground. Presupposition is a property of speakers in contexts, not of sentences in models — the move that allows accommodation to be modelled cleanly.
- Context set is a public object: the framework is symmetric across participants — neither side has privileged access. Trace-checkability falls out.
Connections
Conceptual Contribution
- Claim: Assertion is best modelled as a publicly observable update operation on the common ground of mutually accepted propositions. The semantic content of an assertion is determined by which possible worlds it eliminates from the context set, with no recourse to the speaker’s inner mental state required.
- Mechanism: The common ground is the body of propositions the participants mutually accept for purposes of conversation; the context set is its set-theoretic reflex (worlds compatible with the common ground). An assertion of
p proposes intersecting the context set with [[p]]. The three essential effects (informativeness, determinacy, acceptance-by-default) govern well-formed assertion; diagonalisation handles cases where content varies across the context set.
- Concepts introduced/used: Common Ground, Context Set, Update Semantics, Diagonal Proposition, Pragmatic Presupposition, Accommodation, Essential Effect of Assertion.
- Stance: foundational philosophical paper / canonical dynamic-pragmatic theory.
- Relates to: Operational counterpart of Grice 1975 — where Grice supplies maxims of rational conversation, Stalnaker supplies the mechanics (context-set update) that conversation manipulates. Genealogically the foundation of the dynamic semantics tradition (Heim FCS, Kamp DRT, Veltman update semantics). For agent communication, the framework is the direct philosophical antecedent of public-semantics / commitment-based approaches: the common ground is a publicly observable, monotone, trace-checkable object — exactly the kind of state Singh’s social-agency semantics, Fornara & Colombetti’s operational commitment semantics, and CBCL’s ledger want. The deeper philosophical pair is Brandom: Stalnaker’s common-ground update is one of the two canonical formal models of deontic scorekeeping (the other being argumentation-style commitment stores after Hamblin).