There has recently been some interesting work done by Bach, Graff, Moltmann, and King that touches on the question of verbs which take CPs (e.g., that p) but do not seem to allow corresponding NPs (e.g., the proposition that p). For instance, even though each of the sentences in (1) are fully acceptable, all of the sentences in (2) are questionable:
(1a) Michael said that Bush is a criminal.
(1b) Michael proposed that Bush is a criminal.
(1c) Michael claimed that Bush is a criminal
(2a) * Michael said the proposition that Bush is a criminal.
(2b) ? Michael proposed the proposition that Bush is a criminal.
(2c) * Michael claimed the proposition that Bush is a criminal.As the question marks indicate, I am not sure that (2a) and (2b) are really incorrect. I haven't really that the issue through, but I am skeptical that no NP/DP can be found that is acceptable here. Consider:
Michael proposed, in grand fashion, the obvious truth that Bush is a criminal.
Michael claimed the obvious truth that Bush is a criminal as if it were something we didn't all know.To my ear, both of these are acceptable. (I haven't found one for "say", but I'm confident one can be found.) I strongly suspect that some acceptable def. descrips can be found for most of these sorts of verbs. The trick is to find one that supplies information not contained in the verb about the content of the expressed relation.