Various Language Log readers have been sending me to this BBC report about a British columnist called Simon Heffer, who has a book about the decline of proper language use coming out, and in order to promote it recently visited a school and talked to the children about his prescriptive notions. The BBC used this sentence as a hook, claiming that it is ungrammatical and Mr Heffer can tell you exactly why:
 The Prime Minister has warned that spending cuts are necessary.
Now, set on one side the issue of whether this truly exemplifies a grammatical mistake: of course it doesn't. What interests me here is the psychological question of what could possibly, even in principle, convince someone like Simon Heffer that he was wrong.
Here's the story about the alleged ungrammaticality of :
The problem, as Heffer points out, is that the verb "to warn" is transitive: it requires an object. Somebody has to be warned.
In our example, the Prime Minister had nobody to warn.
We could have said correctly: "The Prime Minister has warned the House of Commons that spending cuts are necessary".
This is an asinine claim. But my point is not its wrongness so much as what might possibly work as a way of showing Heffer that it is wrong.
Let me point out first that he cannot possibly be relying on semantics, at least if he is capable of following an argument. Nobody (not even Heffer, I will assume) wants to say that the government cannot issue a health warning. One can say The government has issued a health warning concerning the high count of fecal bacteria in the water without specifying who was being warned. A warning is just a warning: pay heed or not, it's up to you. So why are we supposed to believe that one could not or should not say The government has warned of health risks associated with the high count of fecal bacteria in the water, or The government has warned that there are health risks associated with the high count of fecal bacteria in the water, without specifying who was being warned?
If the noun warning can be used without a recipient specification and the meaning can be understood, there is no semantic case that the verb warn could not also be thus used. If it is barred from such use, it must be a syntactic matter. It must be obligatorily transitive in syntactic terms.
So how do we determine whether it is obligatorily transitive in syntactic terms? It seems reasonable enough to look at people's use of the language and see whether they typically limit themselves to using it transitively.
The word sequence "has warned that" gets over 6 million Google hits. Some are spurious, I'm sure (Google ignores punctuation as well as case, so there might be cases where one sentence ends "whom he has warned." and the next begins with "That would be silly…"). But from a brief browse most of the 6 million seem likely to be cases of intransitive warn taking a that-clause complement.
In other words, the occurrence of sentences like  is so overwhelmingly well confirmed that (as Mark recently suggested in another context) developing support for it is "like footnoting the observation that … people already wore shoes" back in the 1960s.
What if Heffer just responded that no doubt six million cases of this error had appeared on the web, but that was just a sign that in recent years people had become increasingly stupid and careless?
Well, suppose I were to point out that the great Oxford English Dictionary gives a quotation from Spenser's Faerie Queene (so well known that it is cited as "F.Q."):
1590 SPENSER F.Q. I. ii. 1 And chearefull Chaunticlere with his note shrill Had warned once, that Phoebus fiery carre In hast was climbing vp the Easterne hill.
A case from one of Engand's greatest ever poets, dated 1590, rather puts a damper on the theory that this is a matter of recent sloppiness, does it not?
Probably not. I confess I cannot figure out what could convince this man. He has written in the Daily Telegraph that writing a book on what constitutes correct English is "relatively easy to do, once one has armed oneself with the Oxford English Dictionary (OED) and some reputable grammar books by way of research materials". But he apparently didn't read the OED article on the verb warn. Would it have made a difference if he had looked at it?
His view may really be just that he personally dislikes encountering the verb warn without a noun phrase complement. He may be calling it ungrammatical merely to make his personal peeve sound more like an objective opinion.
It is probably not sensible even to try raising such matters with him. One of the opinions he has put down in writing (here) is that "English grammar shouldn’t be a matter for debate." If it is stipulated that my topic of discussion isn't a matter for debate, I guess there is no hope of an answer to my question about what would convince him that he is wrong. Simon Heffer has warned that he is not up for discussing the matter at all. Too bad, because I would have liked to know.