Documentation

Studies.BeckOdaSugisaki2004

Beck, Oda and Sugisaki 2004: comparison in Japanese without degree abstraction #

This file formalizes Beck, Oda and Sugisaki's analysis of Japanese yori-comparatives. A yori-constituent denotes an individual — a free relative, or a definite built from a relative clause — and fixes the standard c of a positive or contextual comparative in the main clause, in place of the than-clause degree set of the standard analysis. The authors attribute the difference to a Degree Abstraction Parameter: Japanese does not bind degree variables in the syntax. We state both repertoires of denotations and derive from them the paper's three empirical differences — variable acceptability, no subcomparatives of degree, no degree-based negative islands — and its scope argument.

Main definitions #

Main results #

Example numbers follow the journal article; the examples are typed in Data/Examples/BeckOdaSugisaki2004.json.

References #

The standard analysis #

def BeckOdaSugisaki2004.er {D : Type u_2} [LinearOrder D] (D₁ D₂ : Set D) :

The comparative morpheme on degree sets, where er D₁ D₂ holds iff max D₂ > max D₁ (8).

Equations
  • BeckOdaSugisaki2004.er D₁ D₂ = ∃ (m₁ : D) (m₂ : D), IsGreatest D₁ m₁ ∧ IsGreatest D₂ m₂ ∧ m₁ < m₂
Instances For
    theorem BeckOdaSugisaki2004.er_Iic_Iic {D : Type u_2} [LinearOrder D] (d₁ d₂ : D) :
    er (Set.Iic d₁) (Set.Iic d₂) ↔ d₁ < d₂

    With unique witnesses er compares the two maxima directly, the comparative and the subcomparative of (7′), (9′) and (10′).

    The contextual analysis #

    def BeckOdaSugisaki2004.pos {Entity : Type u_1} {D : Type u_2} [LinearOrder D] (μ : Entity → D) (c : D) (x : Entity) :

    The positive form at a contextual standard, where pos μ c x holds iff x is d-A for some d > c (20).

    Equations
    Instances For
      theorem BeckOdaSugisaki2004.pos_iff_lt {Entity : Type u_1} {D : Type u_2} [LinearOrder D] (μ : Entity → D) (c : D) (x : Entity) :
      pos μ c x ↔ c < μ x
      theorem BeckOdaSugisaki2004.pos_iff_mem_over {Entity : Type u_1} {D : Type u_2} [LinearOrder D] (μ : Entity → D) (c : D) (x : Entity) :
      pos μ c x ↔ x ∈ μ ⁻¹' Set.Ioi c

      The positive form is the preimage of the open ray above the standard.

      theorem BeckOdaSugisaki2004.maxIn_Ioi_iff_of_isGreatest {D : Type u_2} [LinearOrder D] {S : Set D} {m : D} (h : IsGreatest S m) (c : D) :
      Degree.MaxIn (Set.Ioi c) S ↔ ∃ d ∈ S, c < d

      The contextual comparative MaxIn (Set.Ioi c), max S > c (23), holds of an S with a greatest element iff some d ∈ S exceeds c (footnote 7).

      theorem BeckOdaSugisaki2004.maxIn_Ioi_Iic_iff {Entity : Type u_1} {D : Type u_2} [LinearOrder D] (μ : Entity → D) (c : D) (x : Entity) :
      Degree.MaxIn (Set.Ioi c) (Set.Iic (μ x)) ↔ pos μ c x

      On an adjective's own degree set the contextual comparative and the positive form coincide ((24d), (21b)).

      theorem BeckOdaSugisaki2004.pos_comparedTo {Entity : Type u_1} {D : Type u_2} [LinearOrder D] (μ : Entity → D) (x y : Entity) :
      pos μ (μ y) x ↔ μ y < μ x

      A context setter supplying y sets c := μ y, and the positive form becomes the comparison with y ((21), (25)).

      Yori-clauses denote individuals #

      noncomputable def BeckOdaSugisaki2004.theC {Entity : Type u_1} (C : Set Entity) (P : Entity → Prop) :
      Option Entity

      theC C P is the unique P-individual of the contextual domain C (54).

      Equations
      Instances For
        theorem BeckOdaSugisaki2004.theC_eq_some_iff {Entity : Type u_1} (C : Set Entity) (P : Entity → Prop) (y : Entity) :
        theC C P = some y ↔ (y ∈ C ∧ P y) ∧ ∀ z ∈ C, P z → z = y
        theorem BeckOdaSugisaki2004.isGreatest_card_of_isGreatest {Atom : Type u_3} {P : Finset Atom → Prop} {Y : Finset Atom} (h : IsGreatest {S : Finset Atom | P S} Y) :
        IsGreatest {n : ℕ | ∃ (S : Finset Atom), P S ∧ n ≤ S.card} Y.card

        If Y is the maximal P-plurality, then the greatest n for which some P-plurality has at least n members is the cardinality of Y (41a).

        Subcomparatives #

        theorem BeckOdaSugisaki2004.pos_ne_subcomparative :
        ∃ (μ : Fin 2 → ℕ) (ν : Fin 2 → ℕ), ν 1 < μ 0 ∧ ¬pos μ (μ 1) 0

        An individual standard lies on the matrix dimension, so the reading of the shelf–door sentence with the door as standard differs from the subcomparative ((74), (79)).

        Negative islands #

        theorem BeckOdaSugisaki2004.not_isGreatest_nobody {Entity : Type u_1} {D : Type u_2} [LinearOrder D] [NoMaxOrder D] (bought : Entity → Prop) (μ : Entity → D) :
        ¬∃ (m : D), IsGreatest (Degree.scopeDegrees (Quantifier.GQ.no bought) (Set.Iic ∘ μ)) m

        On a scale without a top the negated than-clause, the degrees d such that nobody bought a d-expensive book (11), has no greatest degree (12).

        theorem BeckOdaSugisaki2004.not_er_nobody {Entity : Type u_1} {D : Type u_2} [LinearOrder D] [NoMaxOrder D] (bought : Entity → Prop) (μ : Entity → D) (S : Set D) :
        ¬er (Degree.scopeDegrees (Quantifier.GQ.no bought) (Set.Iic ∘ μ)) S

        er over a negated than-clause never holds, the negative island ((11), (87a)).

        theorem BeckOdaSugisaki2004.yori_negative_entails_unbought {Entity : Type u_1} {D : Type u_2} [LinearOrder D] (C : Set Entity) (bought : Entity → Prop) (μ : Entity → D) (x : Entity) (h : ∃ (y : Entity), (theC C fun (z : Entity) => ¬bought z) = some y ∧ pos μ (μ y) x) :
        ∃ y ∈ C, ¬bought y

        A yori-clause built on the unbought individuals denotes the unique one of them, so the construction entails that some book in the domain went unbought ((86), (88c)).

        theorem BeckOdaSugisaki2004.er_everyone_without_shared_book :
        ∃ (read : Fin 2 → Fin 3 → Prop) (μ : Fin 3 → ℕ), er (Degree.scopeDegrees (fun (S : Fin 3 → Prop) => ∀ (b : Fin 2), Quantifier.GQ.some (read b) S) (Set.Iic ∘ μ)) (Set.Iic (μ 2)) ∧ ¬∃ (x : Fin 3), ∀ (b : Fin 2), read b x

        The than-clause of a universally quantified comparative, the degrees d such that every boy read a d-long book (95a), is the degree set of the quantifier every boy read some book; the comparative can hold when no book was read by every boy, so it lacks the uniqueness presupposition of the yori-clause ((94), (95)).

        The main clause and scope #

        def BeckOdaSugisaki2004.erJ {Entity : Type u_1} {D : Type u_2} [LinearOrder D] (c : D) (P : D → Entity → Prop) (x : Entity) :

        The comparative on adjective meanings, where erJ c P x holds iff the greatest degree to which x is P exceeds c (133).

        Equations
        Instances For
          theorem BeckOdaSugisaki2004.erJ_iff_pos {Entity : Type u_1} {D : Type u_2} [LinearOrder D] (μ : Entity → D) (c : D) (x : Entity) :
          erJ c (fun (d : D) (x : Entity) => d ≤ μ x) x ↔ pos μ c x

          For a monotone adjective, erJ and the positive form have the same truth conditions (footnote 15).

          def BeckOdaSugisaki2004.wideScope {D : Type u_2} [LinearOrder D] {W : Type u_3} (acc : Set W) (μ : W → D) (s : D) :

          The wide-scope reading of need to be exactly s long, the comparative over need, says that the greatest degree reached in every acceptable world is the target s (136b).

          Equations
          Instances For
            def BeckOdaSugisaki2004.inSitu {D : Type u_2} [LinearOrder D] {W : Type u_3} (acc : Set W) (μ : W → D) (s : D) :

            The in-situ reading, which erJ with a differential yields, says that in every acceptable world the greatest degree is the target s.

            Equations
            Instances For
              theorem BeckOdaSugisaki2004.inSitu_iff {D : Type u_2} [LinearOrder D] {W : Type u_3} (acc : Set W) (μ : W → D) (s : D) :
              inSitu acc μ s ↔ ∀ w ∈ acc, μ w = s
              theorem BeckOdaSugisaki2004.wideScope_of_inSitu {D : Type u_2} [LinearOrder D] {W : Type u_3} {acc : Set W} {μ : W → D} {s : D} (hne : acc.Nonempty) (h : inSitu acc μ s) :
              wideScope acc μ s
              theorem BeckOdaSugisaki2004.exists_wideScope_not_inSitu :
              ∃ (acc : Set (Fin 2)) (μ : Fin 2 → ℕ), wideScope acc μ (10 + 5) ∧ ¬inSitu acc μ (10 + 5)

              In a model where the acceptable versions of a 10-page draft run to 15 or 16 pages, the wide-scope reading of "needs to be exactly 5 pages longer" holds and the in-situ reading fails ((136), (138)).