Coordination of Interrogatives #
@cite{groenendijk-stokhof-1984} @cite{szabolcsi-1997}
Q1 ^ Q2 = meet (coarsest common refinement), Q1 v Q2 = join (finest common coarsening).
- @cite{groenendijk-stokhof-1984}. Studies on the Semantics of Questions. Ch. VI, Section 3.1.
- @cite{szabolcsi-1997}. Ways of Scope Taking.
Equations
Conjunction is commutative (up to equivalence).
Conjunction is associative.
The trivial question is the unit for conjunction.
A polar question about a disjunction.
"Is it the case that P₁ ∨ P₂?" has two cells: yes and no.
Equations
- Semantics.Questions.Coordination.polarDisjunction p1 p2 = Semantics.Questions.polarQuestion fun (w : W) => p1 w || p2 w
Instances For
The conjunction is the meet in the question lattice.
Conjunction preserves refinement in both arguments.
Q2 is functionally dependent on Q1 over a world set if there exist worlds in the same Q1-cell but different Q2-cells.
@cite{groenendijk-stokhof-1984}, Ch. VI: Functional dependence is what gives rise to pair-list readings. When the wh-answer varies across cells of the universal quantifier, the full answer requires listing the answer for each element.
Equations
- Semantics.Questions.Coordination.functionallyDependent q1 q2 worlds = worlds.any fun (w : W) => worlds.any fun (v : W) => q1.sameAnswer w v && !q2.sameAnswer w v
Instances For
When Q2 is functionally dependent on Q1, conjunction is strictly finer than Q1 alone: the conjunction has at least as many cells as Q1.
This uses the hypothesis: functional dependence witnesses two worlds in the same Q1-cell but different Q2-cells, which means Q1+Q2 separates them while Q1 does not. The conjunction therefore has strictly more distinctions.
Conversely, if Q2 is NOT functionally dependent on Q1, then Q1 already determines Q2 on the given worlds: any two worlds in the same Q1-cell are also in the same Q2-cell.
This means Q1 ⊑ Q2 on the given world set, so the conjunction Q1+Q2 is informationally redundant — it has the same cells as Q1 alone.
Conjunction of a list of questions via foldl.
Models embedded question coordination: "John knows [who came] and [what they brought]" denotes the conjunction of two questions.
Equations
- Semantics.Questions.Coordination.conjoin [] = Semantics.Questions.GSQuestion.trivial
- Semantics.Questions.Coordination.conjoin (q :: rest) = List.foldl (fun (x1 x2 : Semantics.Questions.GSQuestion W) => x1 + x2) q rest
Instances For
Conjunction of questions refines each conjunct.
"John knows Q1 and Q2" entails "John knows Q1": knowing the conjunction answer means knowing each individual answer.
Sluicing as question identity: the elided question Q_sluice is resolved by the same partition as the antecedent question Q_antecedent.
@cite{groenendijk-stokhof-1984}: In "Someone called, but I don't know who ⟨called⟩", the sluiced wh-phrase recovers a question Q that is identical to (or at least refined) the antecedent question.
This theorem states the core constraint: if the sluice inherits its question from the antecedent, then knowing the antecedent answer entails knowing the sluice answer.
Pair-list as conjunction: "Which X did every Y verb?" = conjunction_{y in Y} "Which X did y verb?".
Equations
- Semantics.Questions.Coordination.pairListAsConjunction [] questionFor = Semantics.Questions.GSQuestion.trivial
- Semantics.Questions.Coordination.pairListAsConjunction (e :: es) questionFor = List.foldl (fun (acc : Semantics.Questions.GSQuestion W) (e' : E) => acc + questionFor e') (questionFor e) es
Instances For
The pair-list reading refines any individual question.