Hilton Als, or one of his editors at the New Yorker, has opted for prescriptively correct (but now very formal and even archaic-sounding) whom in a context where I think who would be stylistically much more natural (discussion of some other cases of “Object whom“ here):
Jackie [a man] wants to make love, but Veronica has something on her mind. She’s been seeing someone else, but won’t say whom. Is it their downstairs neighbor, the motherfucker with a hat? (Hilton Als, “War Games” [review of “The Motherfucker with the Hat”], New Yorker 4/25/11, p. 86)
The ellipsis construction in won’t say whom is known in the syntactic literature as Sluicing. In its simplest form, Sluicing allows the omission of all of the material in an embedded WH-interrogative clause, leaving only the initial WH phrase:
Kim had done something illegal, and we all wondered what ___. [ellipsis understood as Kim had done ___, where the underlines indicate the position from which the WH phrase has been "extracted"]
Sandy is moving west, but I don’t know (to) where ___. [ellipsis understood as Sandy is moving ___]
Leslie could hear someone’s voice and could almost make out whose ___. [ellipsis understood as ___ voice Leslie heard]
Leslie could hear someone talking and could almost make out who ___. [ellipsis understood as ___ was talking]
In the last example, the initial WH phrase is the interrogative pronoun WHO (with forms who and whom), serving as the subject of the embedded interrogative clause and so occurring in the Nom form, who.
Unfortunately, WHO ends up here immediately following the verb (make out) in the higher clause, with nothing following it — so that it looks like it’s the direct object of that verb (actually, the whole embedded WH-clause is the direct object of that verb), and people are sometimes tempted to use the Acc form, whom, here. That gives us a case of ISOC (“in-situ subject of an object clause”; see discussion here and here):
Leslie could hear someone talking and could almost make out whom.
Prescriptively incorrect, but pretty common.
That brings us to the tricky case (which is illustrated in the Als quote): initial WH phrase WHO serving as an object (direct object or object of a preposition) within the embedded interrogative clause. There are competing systems for choosing forms for WHO. From my “Object whom” posting:
In System A, Form2 of WHO [whom] is associated with object function (though with several important complexities), while in System B, Form2 of WHO is used for an object (of a preposition or a verb) only when WHO is actually in construction with its governor.
… A fair number of speakers shift between the two systems on stylistic grounds, using System B generally but shifting to System A in contexts they believe call for an extremely formal style — yielding what I’ll call System A/B. Doing this requires some judgment as to which contexts call for System A, and different people have different takes on the matter, so the way is open for differing assessments of the results. I’m a pretty consistent System B user, so I’m likely to find some occurrences of whom (from users of System A or A/B) discordant in their contexts.
In the Als sentence, we have WHO in object function in the embedded WH-clause:
She’s been seeing someone else, but won’t say whom ___. [ellipsis understood as she's been seeing ___]
System A calls for whom, System B for who, System A/B for who in general but whom in extremely formal contexts, and I have trouble seeing the Als context as extremely formal. So it looks like the New Yorker is insisting on System A no matter what. Rules are rules.