Assertion Theories: Cross-Theory Comparison #

@cite{brandom-1994} @cite{farkas-bruce-2010} @cite{gunlogson-2001} @cite{krifka-2015} @cite{lauer-2013} @cite{stalnaker-1978}

Compares six theories of assertion along structural dimensions: Stalnaker, Farkas & Bruce, Krifka, Brandom, Gunlogson, and Lauer.

Comparison Matrix #

Theory	Commitment ≠ Belief	Retraction	Source	Entitlements	Probabilistic
Stalnaker	No	No	No	No	No
F&B	Yes	No	No	No	No
Krifka	Yes	Yes	No	No	No
Brandom	Yes	Yes	No	Yes	No
Gunlogson	Yes	Yes	Yes	No	No
Lauer	Yes	No	No	No	Yes

Key Embeddings #

Stalnaker embeds in Krifka: when commitment = belief (no lying, no hedging), Krifka's model collapses to Stalnaker's.
F&B's dcS/dcL are Krifka commitment states: dcS = speaker's commitment slate, dcL = addressee's commitment slate.
Brandom strictly richer than Stalnaker: entitlements have no Stalnaker analog.
Gunlogson models rising declaratives; Stalnaker cannot.
Lying: Krifka and Brandom handle it (commitment without belief); Stalnaker struggles (assertion = belief update).

source

theorem Comparisons.AssertionTheories.stalnaker_no_separation :

Interfaces.AssertionTheory.separatesCommitmentFromBelief Theories.Pragmatics.Assertion.Stalnaker.StalnakerTag = false

Stalnaker does NOT separate commitment from belief.

source

Stalnaker embeds in Krifka: when commitment = belief (sincere assertion), Krifka's commitment operation has the same effect as Stalnaker's CG update.

The embedding is witnessed by the fact that Stalnaker has strictly fewer structural features.

source

theorem Comparisons.AssertionTheories.brandom_strictly_richer :

Interfaces.AssertionTheory.separatesCommitmentFromBelief Theories.Pragmatics.Assertion.Brandom.BrandomTag = true ∧ Interfaces.AssertionTheory.supportsRetraction Theories.Pragmatics.Assertion.Brandom.BrandomTag = true ∧ Interfaces.AssertionTheory.separatesCommitmentFromBelief Theories.Pragmatics.Assertion.Stalnaker.StalnakerTag = false

Brandom is strictly richer than Stalnaker: entitlements have no Stalnaker analog.

In Brandom's model, an agent's normative status has TWO dimensions (commitments + entitlements). Stalnaker's CG only tracks what's mutually believed, with no notion of "being entitled to assert p without having asserted it."

source

theorem Comparisons.AssertionTheories.gunlogson_handles_rising :

Interfaces.AssertionTheory.modelsSourceMarking Theories.Pragmatics.Assertion.Gunlogson.GunlogsonTag = true ∧ Interfaces.AssertionTheory.modelsSourceMarking Theories.Pragmatics.Assertion.Stalnaker.StalnakerTag = false

Gunlogson models rising declaratives; Stalnaker cannot.

Rising declaratives require source marking (self vs other-generated commitments). Stalnaker's symmetric CG update cannot represent the asymmetry between "It's raining." (falling) and "It's raining?" (rising).

source

theorem Comparisons.AssertionTheories.lying_coverage :

Interfaces.handlesAssertion Theories.Pragmatics.Assertion.Krifka.KrifkaTag Interfaces.AssertionPhenomenon.lying = true ∧ Interfaces.handlesAssertion Theories.Pragmatics.Assertion.Brandom.BrandomTag Interfaces.AssertionPhenomenon.lying = true ∧ Interfaces.handlesAssertion Theories.Pragmatics.Assertion.Stalnaker.StalnakerTag Interfaces.AssertionPhenomenon.lying = false

Lying: commitment without belief.

Krifka and Brandom handle lying because they separate commitment from belief. An agent can be publicly committed to p without privately believing p. Stalnaker's model equates assertion with belief update, making lying incoherent as a formal operation.

source

theorem Comparisons.AssertionTheories.rising_declarative_coverage :