Posted by Neal on November 22, 2011
My brother Glen left a comment on my last post about syntactic tree diagrams, which I had illustrated with the sentences Brynn will say who stole the cookies and Who will Brynn say stole the cookies? Glen altered the sentence to make a point, writing:
Now take the sentence Brynn will say whom she gave the cookies. In a Reed-Kellogg diagram, it would be very clear why the sentence requires whom instead of who: because whom is the object of the verb gave. And that would be equally obvious for Whom will Brynn say she gave the cookies?, precisely because the diagram would be the same.
The answer to his objection, by the way, is that in tree diagrams you have a means of showing a gap in the place where the wh element would be if it didn’t have to be at the front of the sentence. Another commenter, The Ridger, pointed this out. But even as that discussion was taking place, Glen and another commenter, Ellen K., were veering off into a discussion of whether Glen’s examples were even grammatical at all. Ellen wrote:
That’s grammatical for you? It isn’t for me. It needs a to: Brynn will say whom she gave the cookies to. (or Brynn will say to whom she gave the cookies, which is awkward, but still better, for me, than without the to).
I agreed with Ellen’s grammaticality judgments, but was too busy to comment personally. Glen responded:
It’s grammatical because whom is an indirect object. She gave him the cookies does not require a to before him, which is the indirect object. Him and whom occupy the same grammatical position.
Glen is right in that whom is an indirect object, as is him in She gave him the cookies, but just because logically something should be grammatical, that doesn’t mean it is. Ellen got at this point in her next response:
[W]hen who or whom is fronted, to is required. For me. That is, for some of us. So I’m surprised to see it’s different for some people.
As it happens, at about the same time Arnold Zwicky was publishing a post on the so-called Dative Alternation (sometimes called the Dative Shift) — the availability of both give someone something and give something to someone for give and other verbs involving the transfer of something to someone. He observes:
The availability of the N[oun]P[hrase] Dative [i.e. give someone something] is apparently constrained by a huge number of factors, having to do with the semantics of the V[erb], the discourse prominence of the referents involved, the phonology of the V, the grammatical person of the NPs involved, the pronominal status of the NPs involved, and the particular V involved (with donate fine in the PP Dative but dubious in the NP Dative, for example).
Zwicky doesn’t mention whether the kind of pronoun (i.e. personal or interrogative) has an effect, but with all the other factors that do, I wouldn’t be at all surprised if this did, too. I searched COCA for strings of who or whom followed by an auxiliary verb (whether a modal, or form of do, have, or be), followed by a personal pronoun, followed by the verb give. All told there were only a couple of dozen hits at most, and every one of the relevant ones included a to, either before the whom (as in to whom did you give it?) or at the end of the VP (as in who did you give this to). I wonder if CGEL has anything to say about this…
Ah, they do! They say (pp. 248-249) that in constructions that contain an indirect object and a direct object, if you put an object before the subject (like for a question or a relative clause), it sounds much better for the direct object to be there than the indirect one. Their examples:
- The key, he gave Sue. / ?Sue he gave the key.
- The key which he lent me didn’t fit the lock. / ?The one whom I lent the key didn’t return it.
The starred or question-marked items are those that would sound better with a to. Huddleston and Pullum acknowledge that speakers vary in how bad they find the */? examples above, but “[n]evertheless, there is very widespread agreement that the [*/?] examples are significantly less acceptable than” the other examples. In other words, it’s better to use the construction that has a direct object and a to prepositional phrase if you want to front the recipient instead of the thing that’s transferred. Hence, the oddness of ?Brynn will say who(m) she gave the cookies, compared to the grammaticality of Brynn will say what she gave Fenster and Brynn will say who(m) she gave the cookies to.
In a comment on his post, Zwicky provides links to a couple of recent papers on the dative alternation — a nice introduction to a syntactic phenomenon that, as he points out, has generated a huge amount of literature.