¹ Res Cogitans. (London: Cornell University Press, 1972Google Scholar).

² “Can One Believe What One Knows?”, Philosophical Review LXXXIV (1975). pp. 220–35Google Scholar.

³ Vendler, p. 98. Though Vendler does not tell us the framework within which he is providing “derivations” of I. and 2., the “derivation” he proposes for 2. must be incorrect. Most of the reasoning needed to show that his proposed derivation is incorrect is provided by Vendler himself (p. 104):

“…Now these (wh-nominals), themselves, are nothing but indefinite versions of that-clauses, formed, as we recall, by replacing a noun-phrase or an adverbial phrase in the sentence following that by wh plus the appropriate pro-morpheme. Consequently, whenever I claim that I know wh…‥, I guarantee that I could make another claim in which the wh-nominal is replaced by a corresponding that-clause. It makes perfect sense to say that I know what he did but will not tell you; to say, on the other hand, that I know what he did but could not possibly tell you is absurd.”

In other words, it is a consequence of Vendler's account of the derivation of know wh- sentences that if I claim that I know what he did, I guarantee that I am able to provide the “such and so” that he did because my words “what he did” derive, on Vendler's account, from some particular sentence of the form “he did such and so”.

But if this is true, it must also be true that if I say “John knows what he did”, I give the same guarantee, since on Vendler's account “what he did” derives from some particular sentence of the form “he did such and so”. But in fact I make no such guarantee when I say “John knows what he did”. I can say (sincerely) “John knows what he did” without myself being able to provide the relevant sentence of the form “He did such and so” and without giving it to be understood that I can provide that sentence. Thus it is impossible to see how any particular sentence (i.e. a sentence with something filled in for “such and so”) could be involved in the derivation of my sentence “John knows what he did”. And hence no such sentence is involved in the derivation of my sentence “I know what he did”. (Or are wh-nominals derived differently in first person and in third person know-sentences?)

⁴ Vendler, p. 99.

⁵ Vendler. p. 105.

⁶ Jones, pp. 224–25.

⁷ In effect, this is the proposal Jones is making (pp. 224–25); that is, this is what his proposal would come to if put within the framework of generative syntax.

⁸ I wish to distinguish wh-that from the complementizer wh-Q (see Baker, C.L., (1969) Indirect Questions in English, unpublished Doctoral dissertation, University of Illinois and J. BresnanGoogle Scholar, (1970), “On Complementizers: Towards a Syntactic Theory of Complement Types,” Foundations of Language, 6, pp. 297–321). I believe the latter occurs with wonder or ask but not know or discover. Though detailed arguments for such a distinction are beyond the scope of this paper, there are compelling reasons for why I need to do this. If the deep structure complementizer in e.g. I know where John lives and I know that John lives in Cleveland are the same, and if we were to take that complementizer to be wh-Q, then I know that John lives in Cleveland would contain an embedded question — an obviously unwelcome result.

⁹ There is a dialect of American English (which I call Lamod, L. A. Mod) in which this sentence is grammatical, along with such sentences as

9.1 Do you believe what that guy is wearing?

9.2 I don't believe what that guy is wearing. However, in this dialect, believe takes what appears to be a wh-complement only in questions and negative contexts:

9.3 *I believe what that guy is wearing.

It is thus puzzling how the believe-wh sentences get generated in this dialect. Perhaps in the end the wh-words should be viewed as free relatives rather than wh-complementizers in such constructions (Cf. “I don't believe the clothes that guy is wearing.”). But though this suggestion seems plausible for many examples, e.g. 9.1, I find it less appealing for a sentence like:

9.4 I can't believe where he is living.

What makes this example troublesome is that I would understand someone who uttered 9.4 to be saying not that he can't believe the house so-and-so is living in, but rather that he can't believe that so-and-so is living where he is living. Or at least 9.4 is ambiguous between these two readings. Thus the object of 9.4 is, at least under one interpretation, sentential and so seems to be a counterexample to the claim that believe does not take a wh-complementizer. But even so, I am not too troubled by this example for two reasons:

1. 9.4 seems to belong to Lamod because it too only occurs in negative or question contexts.

2. There is no parallel pairing for know:

9.5 I don't know where he is living does not mean:

9.6 *I don't know that he is living where he is living. Another apparent counterexample to my claim is:

9.7 What do you know about the applicant?

This is prima facie a counterexample because what appears to have been fronted all the way to the left of the sentence out of the complement of know. But to see that this is only an apparent counter-example, we need only note that know can take non-sentential objects, e.g.:

9.8 He knows something which he hasn't told us.

9.9 He knows a little bit more than you do.

9.10 I know something about the candidate.

It seems clear, then, that 9.7 comes from:

9.11 Wh-Q you know wh-something about the candidate, and does not not violate the rules I have proposed because 9.11 does not have a wh-that complementizer following know; or to put it another way, know does not have a sentential complement in this example.

¹⁰ Jones, pp. 232–33.