A strengthened meaning pairs a plain denotation with its strengthened version and the alternatives used to compute the strengthening.
This is the central data structure: every propositional node in a derivation carries both ‖α‖ (plain) and ‖α‖^S (strong).
- plain : Prop' World
‖α‖ — the plain semantic value
- strong : Prop' World
‖α‖^S — the strengthened semantic value
The scalar alternatives considered
Instances For
Lift a plain meaning to a trivially strengthened one (‖α‖^S = ‖α‖).
Equations
- Phenomena.ScalarImplicatures.Studies.Chierchia2004.StrengthenedMeaning.trivial φ = { plain := φ, strong := φ, alternatives := ∅ }
Instances For
Chierchia's scale axioms (99a–c) as a predicate: (a) Context activates at least 2 members of the scale (b) The uttered term is a member of the activated scale (c) The uttered term is not the strongest activated member
"Strictly stronger" means: a ⊆ₚ utt (a entails utt, true in fewer worlds) but not utt ⊆ₚ a (utt does not entail a).
Equations
- One or more equations did not get rendered due to their size.
Instances For
The Strength Condition (82): ‖α‖^S must entail ‖α‖.
If this fails, the strengthened meaning is discarded and we fall back to the plain meaning. This prevents over-generation of implicatures.
Equations
Instances For
Apply the strength condition: keep strengthened meaning if it entails plain; otherwise fall back to plain. Uses a Boolean flag for decidability.
Equations
Instances For
Krifka's Rule (cf. (81)): At a scope site, introduce a direct implicature by conjoining the plain meaning with the negation of each strictly stronger alternative.
‖S‖^S = ‖S‖ ∧ ⋀{¬‖alt‖ : alt ∈ ALT, alt strictly stronger than ‖S‖}
"Strictly stronger" = a ⊆ₚ φ ∧ ¬(φ ⊆ₚ a): the alternative entails the uttered meaning but not vice versa (true in strictly fewer worlds).
This is the source of DIRECT implicatures.
Equations
Instances For
Direct implicatures satisfy the strength condition: ‖S‖ ∧ ¬(stronger alts) entails ‖S‖.
A context function is downward-entailing (DE) over Prop' World.
f is DE iff: φ ⊆ₚ ψ → f(ψ) ⊆ₚ f(φ).
This reverses entailment: strengthening the argument weakens the result.
Note: This is the World → Prop version, paralleling IsDownwardEntailing
(Antitone) from Semantics.Entailment.Polarity which uses World → Bool.
Equations
- Phenomena.ScalarImplicatures.Studies.Chierchia2004.IsDE f = ∀ (φ ψ : Prop' World), (φ ⊆ₚ ψ) → f ψ ⊆ₚ f φ
Instances For
Negation is DE.
Strong Application (84): DE-sensitive function application.
This is the formal heart of @cite{chierchia-2004}.
Non-DE case (UE contexts): Pass strengthened meanings through. ‖[f g]‖^S = f^S(g^S)
DE case: Strip implicatures from the argument (use plain meaning), then add INDIRECT implicatures from the alternatives. ‖[f g]‖^S = f^S(g) ∧ₚ ⋀{∼(f(alt)) : alt ∈ g.alternatives, f(alt) not entailed}
The key insight: in DE contexts, direct SIs of the argument are blocked because strengthening the argument would WEAKEN the result. But indirect implicatures arise at the matrix level from the alternatives.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Classification of implicatures by their source.
- direct : ImplicatureType
Direct: from Krifka's Rule at a scope site (UE contexts)
- indirect : ImplicatureType
Indirect: from Strong Application through a DE function
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
In UE contexts, scalar items generate DIRECT implicatures. In DE contexts, the direct implicature is blocked, but the DE operator may generate INDIRECT implicatures at the matrix level.
Example:
- UE: "John ate some cookies" → direct: ¬all (from Krifka's Rule)
- DE: "John didn't eat some cookies" → direct ¬all blocked; indirect: ¬(¬all) = all may arise at matrix
Equations
- One or more equations did not get rendered due to their size.
Instances For
The SI-NPI Generalization (@cite{chierchia-2004}, (53)):
Scalar implicatures are systematically SUSPENDED in the same environments that LICENSE negative polarity items (NPIs).
Formally: If f is DE, then direct scalar implicatures of its argument are blocked. The strengthened argument g^S entails g (by the strength condition), so DE reverses this: f(g) ⊆ₚ f(g^S). Using the strengthened argument would WEAKEN the matrix meaning, violating the strength condition at that level.
This is exactly the DE property that licenses NPIs: DE contexts are precisely where scalar strengthening is blocked.
Corollary: Under a DE function, applying f to the Krifka-strengthened argument is WEAKER than applying f to the plain argument.
Chierchia's O operator (cf. (127)): exhaustification over domain alternatives.
For an indefinite with domain D, the O operator provides universal closure over the domain — the NPI "widens" the domain to the maximal set, and O yields the strengthened meaning.
O_D(∃x∈D. P(x)) = ∃x∈D. P(x) ∧ ∀D'⊂D. ¬(∃x∈D'. P(x) ∧ ∀y∈D\D'. ¬P(y))
Simplified: the assertion holds AND no subdomain alternative holds.
Equations
Instances For
NPI strengthening succeeds when the exhaustified meaning entails the plain meaning of the non-widened competitor.
(127): ‖any NP‖^S = O_D(∃x∈D.P(x)) must be stronger than ∃x∈D₀.P(x) where D₀ is the default (non-widened) domain.
Equations
- Phenomena.ScalarImplicatures.Studies.Chierchia2004.npiStrengtheningSucceeds exhaustified competitor = (exhaustified ⊆ₚ competitor)
Instances For
NPI strengthening is BLOCKED when embedding under a DE function, because the DE function reverses the strengthening relationship.
This connects NPIs to scalar implicatures: both involve DE-ness, but for NPIs, the blocking is what makes them grammatical in DE contexts (they don't need to strengthen, so domain widening is "free").
Intervention occurs when a strong scalar item (every, and, numerals) sits between an NPI licensor and the NPI.
The strong item generates an INDIRECT implicature that conflicts with the NPI's domain-widening requirement.
Weak items (some, or) do not intervene because they don't generate indirect implicatures that conflict.
- strong : ScalarStrength
Strong: top of scale (every, and, all numerals > 1). Generates indirect implicatures that can interfere with NPI licensing.
- weak : ScalarStrength
Weak: bottom of scale (some, or). No interfering indirect implicatures.
Instances For
Equations
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Whether a scalar item of given strength intervenes in NPI licensing.
Strong scalars generate indirect implicatures under DE operators; these indirect implicatures can block NPI strengthening. Weak scalars don't generate interfering implicatures.
Equations
Instances For
At a root-level scope site in a UE context, Chierchia's parallel strengthening entails Fox's exhIE.
Krifka's Rule produces: φ ∧ ⋀{¬alt : alt strictly stronger than φ} exhIE produces: φ ∧ ⋀{¬alt : alt innocently excludable}
On "flat" (linearly ordered) scales, every innocently excludable alternative
is strictly stronger, so Krifka's output negates a superset and thus entails
exhIE. The hypothesis hFlat captures this: every IE alt is strictly stronger.
Requires MC-set existence to decompose IE members into φ or ∼a forms.
Strength relation for scalar licensing.
@cite{krifka-1995a} and @cite{chierchia-2004} treat all NPIs as STRENGTHENING: the NPI makes the assertion stronger than its scalar alternatives, so under negation the negated NPI statement is informationally weaker (= more conservative), which is the hallmark of DE environments.
@cite{schwab-2022} observes that ATTENUATING NPIs (like German "so recht") work in the opposite direction: they make the assertion WEAKER than alternatives. Under negation, the negated attenuating statement is actually STRONGER — which means attenuating NPIs should NOT produce illusion effects in non-DE environments (and empirically, they don't).
- strongerThan : StrengthRelation
- weakerThan : StrengthRelation
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Unified scalar licensing parametrized by direction.
For strengthening (= Krifka's ScalAssert): Assert φ and deny all strictly stronger alternatives. φ ∧ ⋀{¬alt : alt ∈ ALT, alt ⊂ φ}
For attenuating (Schwab & Liu's condition): Assert φ and affirm the existence of a strictly stronger alternative. φ ∧ ⋁{alt : alt ∈ ALT, alt ⊂ φ} (Simplified: we record the required relationship, not the full licensing.)
Equations
- One or more equations did not get rendered due to their size.
Instances For
Bridge: scalarLicensing.strongerThan is exactly krifkaRule.
Strengthening licensing satisfies the strength condition (inherits from krifkaRule).