@cite{heim-1994}: Interrogative Semantics and Karttunen's Semantics for know #

@cite{karttunen-1977} @cite{groenendijk-stokhof-1984}

Single-paper formalisation of @cite{heim-1994} (IATL 1, pp. 128–144), "Interrogative Semantics and Karttunen's Semantics for know". The paper asks how Karttunen-style and G&S-style answer notions compare under question-embedding by know, and what minimal modification to Karttunen's semantics yields G&S-equivalent predictions.

Substrate identification #

@cite{heim-1994} introduces two answer notions:

ans₁(α, w) (eq 15) — "answer-in-the-first-sense": the intersection ∩⟦α⟧K(w) of all true Karttunen alternatives at w. This is exactly Exhaustivity.weakAnswer Q w in the substrate.
ans₂(α, w) (eq 16) — "answer-in-the-second-sense": λw'. ans₁(α, w') = ans₁(α, w) — the set of worlds whose Karttunen intersection equals w's. This is the strongly-exhaustive answer in the G&S sense.

The substrate's strongAnswer Q w := {v | ∀ p ∈ alt Q, w ∈ p ↔ v ∈ p} is one canonical formulation of the G&S strong answer; Heim's ans₂ is the reflective formulation that quotients worlds by their ans₁-class. We prove strongAnswer ⊆ heimAns2 here; the converse holds when alternatives are pairwise distinguishable (a typical empirical assumption — the Heim 1994 §7 (21)/(24) examples are counterexamples to the bare equivalence on intensional / contingent contexts).

Section coverage #

§1 Karttunen — simplifiedKarttunenKnow-style "x believes ∩q(w)" is captured by weakAnswer. The actual lexical know predicate (which involves doxastic accessibility) lives in Theories/Semantics/Attitudes/Doxastic.lean; we identify the content weakAnswer Q w here.
§2 Exhaustiveness — Karttunen's eq (5) "if q(w) = ∅ then x believes that q is empty" becomes the substrate's IsExhaustivelyResolvable (Dayal 1996 EP), already in Exhaustivity.lean.
§3 De dicto readings — requires intensional CN-meanings beyond the bare Set W substrate; deferred.
§4 Generalized Karttunen analysis — Heim's eq (8)/(9): clause (i) "x believes ∩q(w)" is redundant given clause (ii) "x believes λw'[q(w') = q(w)]". The substrate analogue is strongAnswer ⊆ weakAnswer: the strong answer entails the weak one (proved below).
§5 Groenendijk & Stokhof — their whether denotation λw'. R(w') ↔ R(w) is precisely strongAnswer (polar R) w in the substrate.
§6 ans₁/ans₂ bridge — formalised here via heimAns1/heimAns2 and strongAnswer_subset_heimAns2.
§7 Non-equivalence: Heim's (21) "John knows which students are identical with themselves" and (24) "John knows which students live with their actual spouses" — divergence cases requiring intensional CN binding; deferred.
§8 Structured propositions — requires extending the Question type with CN-meaning ↔ atomic individual structure; deferred.

Heim's two answer notions (§6 eq 15-16) #

source

def Phenomena.Questions.Studies.Heim1994.heimAns1 {W : Type u_1} (Q : Core.Question W) (w : W) :

Set W

@cite{heim-1994} (15): the answer-in-the-first-sense is the Karttunen intersection ∩⟦α⟧K(w). Identified with the substrate's weakAnswer.

Equations

Phenomena.Questions.Studies.Heim1994.heimAns1 Q w = Semantics.Questions.Exhaustivity.weakAnswer Q w

Instances For

source

@[simp]

theorem Phenomena.Questions.Studies.Heim1994.heimAns1_eq_weakAnswer {W : Type u_1} (Q : Core.Question W) (w : W) :

heimAns1 Q w = Semantics.Questions.Exhaustivity.weakAnswer Q w

source

def Phenomena.Questions.Studies.Heim1994.heimAns2 {W : Type u_1} (Q : Core.Question W) (w : W) :

Set W

@cite{heim-1994} (16): the answer-in-the-second-sense is the set of worlds whose ans₁-image equals w's. The reflective formulation of strong exhaustivity.

Equations

Phenomena.Questions.Studies.Heim1994.heimAns2 Q w = {w' : W | Phenomena.Questions.Studies.Heim1994.heimAns1 Q w' = Phenomena.Questions.Studies.Heim1994.heimAns1 Q w}

Instances For

source

@[simp]

theorem Phenomena.Questions.Studies.Heim1994.mem_heimAns2 {W : Type u_1} (Q : Core.Question W) (w v : W) :

v ∈ heimAns2 Q w ↔ Semantics.Questions.Exhaustivity.weakAnswer Q v = Semantics.Questions.Exhaustivity.weakAnswer Q w

source

theorem Phenomena.Questions.Studies.Heim1994.heimAns2_self_mem {W : Type u_1} (Q : Core.Question W) (w : W) :

w ∈ heimAns2 Q w

§6 bridge: `strongAnswer ⊆ heimAns2` #

The substrate's strongAnswer Q w := {v | ∀ p ∈ alt Q, w ∈ p ↔ v ∈ p} says v decides every alternative the same way as w. Heim's heimAns2 Q w := {v | weakAnswer Q v = weakAnswer Q w} says v and w have the same Karttunen intersection.

Same-decision-on-every-alt implies same true-alt set, hence same intersection — direct.

source

theorem Phenomena.Questions.Studies.Heim1994.strongAnswer_subset_heimAns2 {W : Type u_1} (Q : Core.Question W) (w : W) :

Semantics.Questions.Exhaustivity.strongAnswer Q w ⊆ heimAns2 Q w

Heim's §6 inclusion: if v decides every alternative the same way as w, then v and w have the same Karttunen intersection.

§4 redundancy #

Heim's eq (8)→(9) says clause (i) "x believes ∩q(w)" is redundant given clause (ii) "x believes λw'[q(w') = q(w)]". The substrate captures this as Exhaustivity.strongAnswer_subset_weakAnswer: any state contained in the strong answer is contained in the weak answer. No paper-anchored re-export — call the substrate theorem directly.

§1: simplified Karttunen content #

The simplified Karttunen meaning of know(Q)(x) at world w is "x believes ∩q(w)" — substrate-level: "x's doxastic state is contained in weakAnswer Q w". The doxastic predicate itself lives in Theories/Semantics/Attitudes/Doxastic.lean; here we expose the content as weakAnswer.

source

def Phenomena.Questions.Studies.Heim1994.simplifiedKarttunenContent {W : Type u_1} (Q : Core.Question W) (w : W) :

Set W

@cite{heim-1994} §1 (4): the simplified Karttunen content of know Q w is weakAnswer Q w — what the agent must believe.

Equations

Phenomena.Questions.Studies.Heim1994.simplifiedKarttunenContent Q w = Semantics.Questions.Exhaustivity.weakAnswer Q w

Instances For

source

@[simp]

theorem Phenomena.Questions.Studies.Heim1994.simplifiedKarttunenContent_eq_weakAnswer {W : Type u_1} (Q : Core.Question W) (w : W) :

simplifiedKarttunenContent Q w = Semantics.Questions.Exhaustivity.weakAnswer Q w

§5: G&S strong answer #

@cite{groenendijk-stokhof-1984} whether denotes λw'. R(w') ↔ R(w), which is strongAnswer (polar R) w. The substrate already provides this; we re-export under the paper's vocabulary for cross-reference.

source