Documentation

Linglib.Phenomena.Assertion.Studies.BringGunlogson2000

Gunlogson Felicity ↔ Contextual Evidence Bias @cite{bring-gunlogson-2000} @cite{gunlogson-2001} @cite{romero-2024}

Connects @cite{gunlogson-2001}'s felicity condition on rising declaratives to the contextual evidence framework of @cite{bring-gunlogson-2000}.

The Paper's Deepest Claim #

The Contextual Bias Condition (CBC) — that rising declaratives require the addressee to already be committed to p — is not stipulated. It follows from the uninformativeness requirement on questioning:

  1. Questioning requires uninformativeness for the addressee (Ch. 4 §4.1)
  2. A rising declarative about p is uninformative for the addressee iff the addressee's commitment set already entails p
  3. Therefore: rising declaratives can only question about p when the addressee is already committed to p (= the CBC)

This derivation is formalized in Gunlogson.cbc_from_uninformativeness.

Rising declaratives require exactly forP evidence (coarse version).

This is the coarser formulation using the ContextualEvidence type shared with polar question bias. The precise version is cbcMet, which checks the addressee's actual commitment state.

Generalization (8): Declaratives express bias.

A falling declarative always adds a self-generated commitment to the speaker — the speaker is biased toward p. This is why falling declaratives cannot be used as neutral questions.

Generalization (9): Rising declaratives do not commit the speaker.

The speaker's slate is unchanged — directly verified by definitional equality.

Generalization (10): The CBC.

Rising declaratives can only be used as questions when the addressee is already committed to p. This is the cbcMet condition, and it's derived (not stipulated) via cbc_from_uninformativeness.

Rising from empty is unstable. Confirmation does NOT restore stability (the other-generated commitment persists). Rejection is a no-op (the state stays unstable).

Rising declaratives are not questions: they do not partition the context set.

A polar question {p, ¬p} partitions the context into p-worlds and ¬p-worlds, offering both as live options. A rising declarative about p puts p on the addressee's slate as other-generated — it has a preferred answer (p), unlike a neutral question.

This is witnessed structurally: rising declaratives add a commitment (source-tagged), while questions add a partition. Gunlogson's risingDeclarative returns a GunlogsonState with a new commitment, not a set of alternative propositions.