Attitude Verb Monotonicity #
Classifies attitude verb complement positions using EntailmentSig from
Core/NaturalLogic.lean — the same 9-element lattice used for quantifier
determiners. This makes the parallel between quantifier monotonicity and
attitude monotonicity explicit: both are instances of the same algebraic
classification.
Per-verb monotonicity data lives on VerbCore.complementSig in the
fragment lexicon (Fragments/English/Predicates/Verbal.lean), not here.
This file provides only the derivation logic: given an EntailmentSig,
what follows about conjunction distribution and neg-raising?
Key contributions #
@cite{bondarenko-elliott-2026}
Conjunction distribution derived from EntailmentSig: An attitude verb distributes over conjunction iff its complement signature refines
.mono.Neg-raising from monotonicity + EMP: Bondarenko & Elliott derive neg-raising from upward monotonicity plus an excluded-middle presupposition (EMP), NOT from veridicality (as in
NegRaising.lean). Both derivations agree on standard doxastic verbs.Mereological grounding: The conjunction distribution theorem (
mono_att_distrib_and_iffinTruthmaker.Basic) proves that the distribution follows from theSemilatticeSupstructure of states.
An attitude distributes over conjunction iff its complement signature
refines .mono (upward monotone). This includes: .all, .addMult,
.additive, .mult, and .mono itself.
Note: .mult (multiplicative) distributes over ∧ by definition —
f(A ∧ B) = f(A) ∧ f(B). The .mono refinement captures this.
Equations
Instances For
Excluded-Middle Presupposition (EMP). An attitude verb has EMP if it presupposes that the agent either holds the attitude toward p or toward ¬p. EMP(V, x, p) := V(x, p) ∨ V(x, ¬p)
Under EMP, ¬V(p) → V(¬p) is immediate (disjunctive syllogism). This is neg-raising.
The Bondarenko & Elliott insight: EMP is motivated by monotonicity. Upward-monotone verbs have a "gapless" information state — the state either contains a verifier for p or for ¬p (no undecided gap). This is a presupposition, not a semantic entailment.
Equations
Instances For
Neg-raising available via monotonicity + EMP.
This is an alternative to NegRaising.negRaisingAvailable, which
derives neg-raising from veridicality. Both are correct derivations
that agree on all standard verbs but have different theoretical
underpinnings.
Bondarenko & Elliott: monotonicity + EMP → neg-raising @cite{horn-2001}: non-veridicality → neg-raising
Equations
Instances For
The veridicality and monotonicity accounts agree on all standard doxastic verbs: both predict neg-raising for non-veridical monotone verbs and no neg-raising for veridical ones.
They could diverge for hypothetical verbs that are:
- Non-monotone but non-veridical → veridicality allows, monotonicity blocks
Currently both routes agree because all non-veridical doxastic verbs in the lexicon are also monotone.
The truthmaker conjunction distribution theorem
(Truthmaker.mono_att_distrib_and_iff) provides the semantic why
for the monotonicity classification.
This re-export makes the connection explicit: the EntailmentSig classification of attitude verbs is grounded in the SemilatticeSup structure of propositional content via truthmaker semantics.