The factive-reported distinction in English. Representational and interpersonal semantics.

Ever since Kiparsky & Kiparsky's (1970) seminal paper, it has been recognized that the English complementation system makes a distinction between factive complements (e.g. He doesn't like John's nomination for the award/that John was nominated for the award.) and reported complements (e.g. He said ?John's nomination for the award/He said that John was nominated for the award.) However, the semantico-functional motivation for this distinction remains unclear. It has been proposed that factive complements are (i) presupposed true (in the world, or by the speaker), (ii) presumed given in the common ground, or (iii) modally unqualified, i.e. representing a default commitment of epistemic certainty as expressed by the indicative. In this project, I review these different approaches and propose an alternative account for the semantic value of factive complements, as for the way it is related to the semantics of the matrix predicate and is encoded in grammar. I propose to distinguish three types of complementation constructions instead of two, viz. factive constructions, as in (1), constructions of reported speech and thought, as in (2), and a third type which I refer to as manipulative constructions, as in (3). These three complementation constructions, it is argued, can be characterized on the basis of an analysis of their representational semantics. I claim that the three complement types have a different semantic status in relation to the situation described by the main clause: factive complements are unaffected by the main clause situation, reported complements are effected by it, and manipulated complements are affected by it. This is correlated with the tendency for the three complement types to occur with specific semantic classes of predicates, which express the abstract semantics of contact/reaction, creation, and re-creation/modification respectively. (1) Julia discovered/resented that John was nominated for an award (2) Julia said/thought that John was nominated for an award (3) Julia denied/restated that John was nominated for an award On the traditional view, the proposition in a factive complement as in (1) is said to be presupposed true by the actual speaker uttering the entire complex sentence, while the proposition in a reported complement as in (2) involves commitment on the part of the represented speaker or cognizer in the main clause (i.e. Julia). To examine these commitment phenomena, I present an investigation of the interpersonal status of the different types of complement clauses. I focus on the type of modal stance (i.e. involving an assessment of degrees of certainty or desirability statuses) and the source of the modal stance (actual speaker, represented speaker or cognizer, or other) that is expressed by a modal auxiliary in the complement. In contrast to what has been proposed in the literature, I argue that all three complement types can incorporate positions of both degrees of epistemic certainty and various deontic statuses. The different complementation constructions do show differences with respect to the possible sources of the modal stance in the complement. In reporting constructions, the modal stance in the complement stems from the represented speaker or cognizer identified in the main clause. In manipulative and factive constructions, the source of the complement-internal position is more variable: the position can relate to the actual speaker, to a represented speaker or cognizer, or to a third party, i.e. an ‘echoed’ speaker. I show that this difference in terms of possible modal sources can be predicted from the representational semantics of the different complementation constructions. Taken together, the description of the three complementation constructions in terms of their representational and interpersonal semantics provides a new account of the factive-reported distinction in English, which is no longer defined in terms of (the absence of) a truth commitment on the part of the actual speaker.status: publishe