Abstract This study argues that force-endowed (FE) that -clauses form a syntactically and semantically distinct class and offers an analysis focused on cases where their licensing is not attributable to pragmatic control. I show that speech-reporting verbs employ two different mechanisms to license FE that -clauses: manner-of-speaking verbs can select them directly, whereas proffering verbs cannot, as they select propositional complements. In the latter case, a type-mismatch arises, which is repaired by internal Merge. The analysis integrates Krifka’s conception of FE clauses as index actualizers with event-like denotation and Kratzer’s view of that -clauses as predicates of content-bearing individuals and events. Within this framework, the distribution and interpretation of FE that -clauses receive a unified explanation, thereby supporting the existence of a dedicated class of speech-act–selecting verbs and illuminating the puzzling embedding of FE that -clauses under verbs that select propositional types. Empirically, the account predicts the embedded distribution of force-related main clause phenomena, such as speaker-reduced parentheticals and biscuit conditionals, and explains several characteristic properties of FE that -clauses, including their contrast with propositional that -clauses, their distinct entailment patterns across embeddings, their achievement-coercing effects, and their ability to give rise to reporting interpretations even with unergative verbs.
José María García Núñez (Wed,) studied this question.