Kajsa Djärv
Department of Linguistics
University of Pennsylvania

Factive verbs: a multi-dimensional approach to the syntax and meaning of assertion and presupposition

This talk looks at sentences like (1), which involve finite that-clauses embedded under different classes of verbs.

(1) John {said, believes, doubts, resents, discovered} that Mary likes Kraftwerk.

Semantically, these classes vary in how they relate both the attitude holder and the speaker to the embedded proposition, p. For instance, discover and resent are Factive, whereas say and believe are not:

(2)

a. John {discovered, resented} that [P Mary likes Kraftwerk]. –> Mary likes Kraftwerk

b. John {said, believed} that [P Mary likes Kraftwerk]. -/-> Mary likes Kraftwerk

It’s often claimed that factive and non-factive predicates also differ syntactically, for instance:

(3)

a. I {said, *regretted} that Kraftwerk, Mary doesn’t like.

b. Who did you {say, *regret} doesn’t like Kraftwerk?

c. I said/*regret so.

A common approach to these observation argues that predicates like say and believe are “assertive”, and therefore select CPs (4a), compatible with illouctionary force, main clause syntax, extraction, so-anaphora etc., whereas factives like regret and discoverare “presuppositional”, and therefore select (potentially null) DPs (4b) in its complement; which forces p to be Common Ground, blocks main clause syntax and extraction, and requires it-anaphora.

(4) The CP/DP-hypothesis

a. John {said, believes} [CP that Mary likes Kraftwerk].

b. John {resents, discovered} [DP (the fact) [CP that Mary left town]].

In this talk I argue, based on results from a series of experimental studies, that while the CP/DP-hypothesis provides an apparently elegant picture of the syntax-meaning interface, it fails to account for both the semantics of factive verbs, and their complementation behavior. That is, factivity, as traditionally understood in the semantics literature, does not serve an explanatory role in our syntactic theory, but simultaneously over- and under-predicts the type of types of contexts that should select DP-complements. However, we also find evidence that factivity is less of a unified phenomenon than previously thought. I argue that for a theory of factivity and factive verbs to be explanatorily adequate, it must crucially disassociate the projective inference that the speaker takes the embedded proposition p to be true, from the Common Ground status of p. Once we make this distinction, we are also able to capture the complementation patterns in (3).

Snacks will be served.