Some Language Log readers may feel that the two rules I discuss in my latest post on Lingua Franca, "One Rule to Ring Them All," are stated too loosely for their consequences to be clear. Let me explain here just a little more carefully. The topic under discussion is whether who should be in the nominative form (who) or the accusative form (whom) in sentences with structures broadly like :
|||He's the man who(m) everyone says will one day be king.|
One informal version of the rule for deciding between who and whom in formal-style Standard English (and formal style is the only kind of English I'm talking about here; informal style would of course use who in ) would be this:
|||A preposed relative or interrogative who should be in the nominative form (who) if it is the subject of the relative or interrogative clause that it introduces, and in the accusative form (whom) otherwise.|
A different possible version (also very informal) would be this:
|||A preposed relative or interrogative who should be in the nominative form (who) if it is understood with subject function in whatever clause it is functionally associated with, and in the accusative form (whom) otherwise.|
The point I make in the Lingua Franca piece is that these have different consequences: the first says (as I interpret it, anyway) that whom is correct in , and the second rule says that who is correct.
It has occurred to me that we could make this a little more precise in an appealing way if we introduce two potentially useful terms both dependent on the same metaphor (I didn't try to do this in the limited space available on Lingua Franca).
Call a wh-word or wh-phrase a citizen of the clause where it performs a grammatical function like being Subject or Object or other complement (or being thus understood), and a resident of the clause that it actually appears in (or at the beginning of).
These two notions will coincide in a clause like Who cares about whether he agrees?, where who is both a citizen and a resident of the main clause (the one with care as its verb), and is the subject of that clause; but they pull apart in Who do you think they were looking at?, where who is a citizen of the most deeply embedded clause (the look clause, where it is understood as the complement of at) but a resident of the main clause (the do clause).
Using these terms, the rules for formal style could be restated as [2'] and [3'].
|[2']||A preposed relative or interrogative who should be in the nominative form if it is the subject of the clause in which it is a resident, and in the accusative form otherwise.|
|[3']||A preposed relative or interrogative who should be in the nominative form if it is the subject of the clause in which it is a citizen, and in the accusative form otherwise.|
In , the wh-pronoun who is a resident of the say clause but a citizen of the be clause. It's not the subject of the clause in which it is a resident, but it is the subject of the clause in which it is a citizen.
What I'm pointing out in the Lingua Franca piece (clearly enough, I hope, to be followed by anyone who knows about pronouns and relative clauses and subjects, though I think the terminology above makes it even clearer) is that to say that it is useless to stipulate that the case of who depends on whether it has subject function. That has no clear consequences at all. You have to answer the question: Subject of what?
And (I love this) The Elements of Style (4th ed., p. 11), typically for that incompetent little pamphlet of misinformation, makes a definite and dogmatic prescription to answer the question, but gives the rule in a form that entails the opposite answer! It says, "When who introduces a subordinate clause, its case depends on its function in that clause." That corresponds to  and [2'] (because the clause it introduces is the clause where it's a resident), not to  and [3'], referring to citizenship.
That is, in sentence , who clearly introduces the whole relative clause, everyone says __ will one day be king; and the subject of that clause is everyone. Since who is not the subject of the clause it introduces, the accusative form whom is predicted to be correct. That is the opposite of what was intended by Strunk and White. (Actually it was White, since he added the section I am talking about.) The rule statement was incorrect. The intention (as shown by the examples given, where the allegedly correct form is on the left and the purported correction is on the right) was to state a rule equivalent to  and [3'], making who the correct form.
The thing about all this that will give the purists the horrors is that we don't have a way to say either that  (= [2']) is the right rule or that  (= [3']) is. The corpus evidence is divided. Shakespeare seems to have favored . Dickens favored . (Invidious literary comparisons, anyone?) And whoever you pick, what will you say about Boswell, who apparently (according to Merriam-Webster's Dictionary of English Usage) vacillated between the two rules?
It's the neatest case I know of in which a dispute about what's correct turns out to have no possible resolution, because there are two perfectly defensible rules that have slightly different consequences, and the evidence from literature says that different writers make different choices of rule. It means that the dispute between two varieties of purist can go on forever, and the descriptivists can never stop them fighting. Imagine them…
A: "It should be whom, of course. The logical thing to do is to have the pronoun's case settled according to the clause the pronoun is actually in, and it's not the subject there."
B: "No, it's who. The logical thing is to have the pronoun's case settled according to the clause in which the pronoun is understood to function."
A: "That's not logical at all; it's stupid."
B: "No it's not, idiot. It makes perfectly good sense. It's you that's stupid."
A: "Oh yeah? Don't call me stupid just because I'm not buying your cockamamie, mystical, 'understood-to-function' bullshit rule!"
B: "Well you are stupid. And you're ugly, too. God, your head looks weird when you shout."
A: "Ha! You, calling me ugly, when you drive a car like that crappy old thing you turned up in!"
B: "There's nothing wrong with my Oldsmobile! You think you're cool because you drive a fucking Volvo? If you're going to insult my wheels I'm going to have to ask you to step outside!"
A: "I'd happily step outside right now and beat the crap out of you, except that I wouldn't want to get my hands all smelly."
B: "You're a fuckhead. And as a grammarian you'd make a pretty good janitor."
A: "Well you're an illiterate dork, and and as a janitor you'd make a pretty good asshole."
B: "Yeah, I bet you know all about assholes, given your rumored predilections…"
There'll be no end to it, ever. We will all die before the issue of where to use who and whom is settled in our culture. If the comments area below the Lingua Franca post doesn't provide evidence of that, then I will be mighty surprised.