Exceptional wide scope of bare nominals #

[LBdS22]

Le Bruyn & de Swart, "Exceptional wide scope of bare nominals" (Semantics & Pragmatics 2022). Bare nominals normally take obligatory narrow scope — one of the strongest arguments for the kinds approach. But scrambled bare plurals in Dutch/German unambiguously take wide scope over negation and quantifiers, while keeping kind reference. This adjudicates two accounts:

[Chi98] (with [Day04]): bare plurals denote kinds; the existential comes from Derived Kind Predication, which is local, so scope is position-invariant and cannot be wide (NMP.chierchia_position_invariant).
[Kri04]: bare plurals are properties; the existential type shift is a local type repair at the surface position, so a scrambled (raised) bare plural scopes wide (Krifka2004.scope_follows_position).

Both accounts share one existential closure (NMP.existsClose) and differ only in where negation sits: Chierchia always keeps it outside (¬ ∃), Krifka lets it move inside under scrambling (∃ ¬). The scrambling data falsify the kinds approach's locality and confirm Krifka's position-sensitive shift. Wide scope and kind reference are dissociable.

Main declarations #

chierchia_invariant_krifka_sensitive — the theoretical crux: Chierchia's DKP is position-invariant for every model, while Krifka's ∃-shift diverges on some model
theories_agree_unscrambled / theories_diverge_on_scrambling — on a concrete two-book model the accounts agree on the unscrambled reading and disagree on the scrambled one
krifka_matches_scrambling_data — Krifka's wide-scope reading matches the Dutch boeken niet uitgelezen datum (Data/Examples/LeBruynDeSwart2022.json)
scrambled_keeps_kind_reference — scrambling affects scope, not kindhood

source

def LeBruynDeSwart2022.readingOf (row : Data.Examples.LinguisticExample) (name : String) :

Option Features.Judgment

Judgment of a named reading, if recorded.

Equations

LeBruynDeSwart2022.readingOf row name = Option.map (fun (x : String × Features.Judgment) => x.2) (List.find? (fun (x : String × Features.Judgment) => x.1 == name) row.readings)

Instances For

The theoretical crux #

Chierchia's Derived Kind Predication introduces the existential locally, so negation always scopes outside it and scrambled = unscrambled. Krifka's existential type shift sits at the bare NP's surface position, so the two can diverge.

source

theorem LeBruynDeSwart2022.chierchia_invariant_krifka_sensitive :

(∀ (Entity : Type) (kind : List Entity) (P Q : Entity → Prop), Semantics.Kinds.NMP.chierchiaDerivScrambled kind P Q = Semantics.Kinds.NMP.chierchiaDerivUnscrambled kind P Q) ∧ ∃ (Entity : Type) (dom : List Entity) (P : Entity → Prop) (Q : Entity → Prop), Krifka2004.krifkaDerivScrambled dom P Q ≠ Krifka2004.krifkaDerivUnscrambled dom P Q

The accounts make categorically different scope predictions: Chierchia's derivation is position-invariant for every domain, property, and predicate, whereas Krifka's diverges scrambled vs. unscrambled on some model.

A concrete model: two books, one finished #

Le Bruyn & de Swart's Het klopt dat ik boeken niet heb uitgelezen ("it's true that there are books I didn't finish"): with two books, one finished and one not, the scrambled bare plural is true (there is an unfinished book) but the unscrambled reading is false (it would mean "I finished no books").

source

inductive LeBruynDeSwart2022.Book :

Type