Documentation

Linglib.Phenomena.Questions.Studies.Xiang2022

@cite{xiang-2022}: Relativized Exhaustivity: Mention-Some and Uniqueness #

@cite{dayal-1996} @cite{fox-2018} @cite{spector-2008} @cite{chierchia-caponigro-2013}

Single-paper formalisation of @cite{xiang-2022}, "Relativized Exhaustivity: Mention-Some and Uniqueness" (Natural Language Semantics 30:311–362). Xiang argues that mention-some (MS) is a grammatical phenomenon licensed by existential modals, and proposes Relativized Exhaustivity (RelExh) as the replacement for Dayal's EP that simultaneously (a) preserves the singular-which uniqueness effects, (b) permits MS where modally licensed, and (c) accounts for local-uniqueness in modalised singular wh-questions.

Substrate identifications #

@cite{xiang-2022}substrate
Ans_Fox (eq 20, after Fox 2013) — max-informative true answerIsStrongestTrueAnswer Q w p (without iota uniqueness)
MaxI(α, w, M, [Q]) (eq 96)substrate-relative IsStrongestRelTrueAnswer
Dayal's EP relativised, DEP(w, M, [Q]) (eq 90)IsRelativelyExhaustivelyResolvable here
Relativized Exhaustivity, REP(w, M, [Q]) (eq 91)RelExhPresupposition here
ANS^S / ANS^P (eq 27, 97)Exhaustivity.dayalAns (singular case); plural needs the topical-property layer

The Question W = LowerSet (Set W) substrate captures the propositional side of Xiang's framework. The full §4.1 topical property layer (questions as functions from short answers α : Type to propositional answers Set W) requires extending the Question type to a paired (ShortAns → Prop) × (... → Set W) shape; we defer that lift and work with the propositional projection here.

Section coverage #

What this file does NOT cover #

§4.1.2 Substrate identifications #

@[reducible, inline]
abbrev Phenomena.Questions.Studies.Xiang2022.MaxI {W : Type u_1} (Q : Core.Question W) (w : W) (p : Set W) :

@cite{xiang-2022} eq (20) (after @cite{fox-2018} eq 20): max-informative true answer. An alternative p is max-info at w iff p is true at w and not asymmetrically entailed by any other true alternative. Identified with the substrate's IsStrongestTrueAnswer (without Dayal's iota-uniqueness — Xiang follows Fox 2013 in dropping it).

Equations
Instances For
    @[reducible, inline]
    abbrev Phenomena.Questions.Studies.Xiang2022.MaxIRel {W : Type u_1} (Q : Core.Question W) (w : W) (M p : Set W) :

    @cite{xiang-2022} eq (96) modal-base-relative max-informativity. Identified with the substrate's IsStrongestRelTrueAnswer.

    Equations
    Instances For

      §6.1 Dayal's EP relativised + RelExh (eq 90, 91) #

      @cite{xiang-2022} (eq 90) Dayal's EP relativised: Q is defined in w (relative to modal base M) iff some max-info true answer exists. (Substrate: IsRelativelyExhaustivelyResolvable.)

      @cite{xiang-2022} (eq 91) RelExh: Q is defined in w iff for every singleton sub-modal-base M' ⊆ M that verifies some true answer to Q (relative to the original M), the relativised Dayal-EP holds at M'. The key idea: instead of demanding a single max-informative answer at the evaluation world, RelExh demands a max-informative answer at every accessible world (individually, treating each as its own singleton modal base).

      @cite{xiang-2022} eq (90): Dayal's EP relativised to modal base M. Substrate identification: relExh from Exhaustivity.lean.

      Equations
      Instances For

        @cite{xiang-2022} eq (91): Relativized Exhaustivity (RelExh).

        Q is defined in w (relative to modal base M) iff for every singleton M' ⊆ M (i.e., M' = {w'} for some w' ∈ M) that verifies some true short answer to Q (under M), the relativised Dayal-EP holds at M'.

        On the propositional projection: M' = {w'} makes the relative-EP at M' collapse to the standard EP at w'. So RelExh is essentially "Dayal's EP holds at every accessible world that verifies some Q-alternative true at w".

        Equations
        Instances For

          §6.2 The permitting-MS direction #

          If every accessible world has its own max-informative true answer, RelExh is trivially satisfied. This captures Xiang's §6.2.1 prediction that MS is licensed when each accessible world has a unique exhaustively-true answer relative to that world's perspective.

          When every accessible world satisfies Dayal's EP, the RelExh presupposition is trivially satisfied — the permits MS half of @cite{xiang-2022} §6.2.1.

          §5 Dilemma: Dayal-EP and MS conflict on the same question #

          @cite{xiang-2022} §5: Dayal's EP requires a unique max-info true answer; MS rejects this for can-questions. The substrate-level shadow: a question can fail Dayal's EP at w while having Resolves σ Q succeed for some non-maximal witness — same phenomenon as in @cite{fox-2018} §2.1 (already formalised in Fox2018.resolves_can_succeed_when_EP_fails). RelExh resolves the dilemma by relativising the EP to a per-accessible-world basis.

          theorem Phenomena.Questions.Studies.Xiang2022.relExh_resolves_dilemma_example {W : Type u_1} (Q : Core.Question W) (w w₁ w₂ : W) (p₁ p₂ : Set W) (hM : {w₁, w₂} Set.univ) (hSTA1 : Semantics.Questions.Exhaustivity.IsStrongestTrueAnswer Q w₁ p₁) (hSTA2 : Semantics.Questions.Exhaustivity.IsStrongestTrueAnswer Q w₂ p₂) :
          RelExhPresupposition Q w {w₁, w₂}

          The defining contrast: a non-modalised question can fail Dayal's EP at w (no unique max true answer) while satisfying RelExh relative to a modal base where each accessible world has its own max-info answer.

          Bridge to Dayal: RelExh under a singleton modal base ≡ Dayal's EP #

          @cite{xiang-2022} (90)/(91) collapse: when the modal base is the singleton {w}, RelExh reduces to Dayal's standard EP at w.