Documentation

Linglib.Theories.Semantics.Classifier.Defs

Classifier Denotations à la Sudo (2016) #

@cite{sudo-2016}

A classifier denotation in Sudo's framework: at each world w, the classifier takes a numeral n and an entity x, presupposes a sortal predicate P_w(x), and asserts that x is the join of n atomic P-parts.

Concretely, Sudo (2016, eq. 4): ⟦-rin⟧ = λw. λn. λx : *flower_w(x). |{y ⊑ x : flower_w(y)}| = n

This contrasts with the Chierchia / Little-Moroney-Royer framework in Classifier.Basic, where the classifier is a noun-side predicate transformer (E → Prop) → (E → Prop) that atomizes the noun denotation without reference to a numeral. The two views are different theoretical commitments and coexist in this directory.

Key types #

Out of scope #

Non-atomic classifiers like -kumi (pair) and -daasu (dozen) per Sudo (9a/b) require explicit mereological joins of disjoint pairs and are not expressible via ofSortal. A separate constructor will be added when the non-atomic case is needed.

structure Semantics.Classifier.ClassifierDenot (W : Type u) (E : Type v) [PartialOrder E] :
Type (max u v)

A Sudo-style classifier denotation: at each world, a sortal presupposition plus a predicate whose ⊑-atomic parts of x are counted.

For atomic-sortal classifiers (the common case: -rin, -hiki, -nin, -mai, -hon, -ko, -tou), sortal = counted. For mensural classifiers like -hai (cupful), the relationship is more nuanced and handled by separate constructors.

  • sortal : Core.Intension W (EProp)

    The sortal presupposition. ⟦-rin⟧ presupposes flower; ⟦-nin⟧ presupposes human; ⟦-hiki⟧ presupposes animal ∧ small.

  • counted : Core.Intension W (EProp)

    The countable base whose atomic ⊑-parts of x are counted. For atomic-sortal classifiers, counted = sortal (see ofSortal).

Instances For
    def Semantics.Classifier.ClassifierDenot.ofSortal {W : Type u} {E : Type v} [PartialOrder E] (P : Core.Intension W (EProp)) :

    Construct an atomic-sortal classifier from a single predicate. The sortal and the counting base coincide — the standard case in Sudo (4) for -rin, (8a) for -nin, (8b) for -hiki (with the sortal being a conjunction small ∧ animal).

    Equations
    Instances For
      def Semantics.Classifier.ClassifierDenot.apply {W : Type u} {E : Type v} [PartialOrder E] (cl : ClassifierDenot W E) (w : W) (n : ) (x : E) :

      The body of Sudo's denotation (eq. 4): apply cl w n x iff sortal_w(x) AND the cardinality of {y ⊑ x : counted_w(y)} equals n.

      Uses Set.ncard so the formalization works for infinite domains (where Set.ncard returns 0 for infinite sets). For Sudo's analysis of natural-language counting, the relevant sets are always finite.

      Equations
      Instances For
        @[simp]
        theorem Semantics.Classifier.ClassifierDenot.ofSortal_sortal {W : Type u} {E : Type v} [PartialOrder E] (P : Core.Intension W (EProp)) :
        @[simp]
        theorem Semantics.Classifier.ClassifierDenot.ofSortal_counted {W : Type u} {E : Type v} [PartialOrder E] (P : Core.Intension W (EProp)) :
        theorem Semantics.Classifier.ClassifierDenot.apply_ofSortal {W : Type u} {E : Type v} [PartialOrder E] (P : Core.Intension W (EProp)) (w : W) (n : ) (x : E) :
        (ofSortal P).apply w n x P w x {y : E | y x P w y}.ncard = n

        For atomic-sortal classifiers, the body reduces to the join of the sortal presupposition and the cardinality constraint over P.