White 2014: The Modalized Complement Analysis of forget #
@cite{white-2014}
White 2014 maintains a uniformly factive denotation for forget by positing a covert root modal in non-finite complements (the Modalized Complement Analysis, MCA). The MCA captures Williams' (1)/(4) and (2)/(5) data straightforwardly but was later shown to overgenerate: PRO-ing gerunds get a non-modal presupposition (cf. @cite{williams-2026}, §3.1.1) that the MCA cannot predict.
This file (a) verifies the MCA's coverage of finite-CP and plain-infinitive
data, (b) records the gerund overprediction as a check against the
empirical record in Studies/Williams2026.lean, and (c) confirms
Fragment-level consistency: the English Fragment's split of forget into
forget (negative implicative, infinitival) and forget_rog (factive,
finite) aligns with the typology MCA + pre-existence together predict.
The pre-existence apparatus in Theories/Semantics/Attitudes/PreExistence.lean
is post-2014 (Bondarenko 2019/2020, taken up by @cite{williams-2026});
where this file uses needsModalInsertion, treat that as the project-canonical
upgrade of White's structural prediction, with the gerund case flagged as
a known overprediction of the original MCA.
§1. MCA coverage of the canonical contrast #
White 2014 was designed to capture the finite-CP / plain-infinitive asymmetry uniformly. The MCA succeeds on these two cases.
Finite CP: MCA correctly predicts no modal insertion.
Plain infinitive: MCA correctly predicts modal insertion.
§2. The gerund overprediction #
The PRO-ing gerund is non-finite but its presupposition is non-modal
(Williams §3.1.1). The MCA's mcaPrediction is !isFinite, so it
predicts modal insertion for the gerund — wrong.
The MCA predicts a modal presupposition for the gerund, but the Williams 2026 datum is non-modal.
§3. Pre-existence as the post-2014 fix #
The pre-existence-based needsModalInsertion correctly predicts
non-modal for the gerund. This is post-2014 territory; the theorems
here document the contrast between White's MCA and its successor.
Pre-existence and MCA agree on finite CP and plain infinitive, but diverge on gerunds.
Pre-existence prediction matches all three data points.
MCA matches two of three data points (loses on the gerund).
§4. Fragment consistency #
The English Fragment splits forget into two VerbEntry records, one
for the implicative use and one for the factive/rogative use. The split
is a practical separation of entailment patterns, not a claim of lexical
ambiguity (which Williams 2026 explicitly rejects). These theorems
document that the Fragment's complement-type and factivity assignments
align with the MCA's predictions, restricted to the cases MCA gets right.