January 23, 2007

Whom shall I say [ ___ is calling ]?


Commenting that "a little knowledge is a dangerous thing", Gene Buckley offered me the following example from the New York Times of 1/15/07:

(1) The answer, shaped in the National Security Council, is for the American military to make targets of Iranians whom they believe are fueling attacks, a decision that Mr. Bush made months ago that was disclosed only last week.

It's been a while since we looked at the who/whom thing here at Language Log Plaza, and Buckley's example happens to be #25 on a list I've been adding to over the years, so this might be a good time to dip once again into the murky waters of who and whom.


Here's a version of what I've said to my senior seminar at Stanford (on innovation and spread in the lexicon and syntax) this quarter.  (That's "senior" in the sense of 'fourth-year undergraduate', not in the sense of 'senior citizen'.)

1.  We're interested here in interrogative and relative WHO, with non-possessive forms who and whom; and similarly with WHOEVER, with forms whoever and whomever (references to the former should be taken as covering the second).

2.  I'll call the two forms Form1 and Form2, respectivelyThe ordinary personal pronouns have two corresponding (and roughly comparable) forms: I/me, she/her, he/him, we/us, they/them.  (For YOU and IT, Form1 and Form2 are identical.)

3.  The languages of the world have several strategies for constituent interrogative clauses and for relative clauses (sometimes more than one strategy in the same language).  Two really common ones involve special interrogative and relative pro-elements, either (a) "extracted" from the usual locations of their phrase types (in English, they appear at the beginning of their clauses, and nothing appears in the usual location; sometimes the missing constituent is referred to as a "gap"), or (b) left "in situ", in their usual locations.

English uses extraction almost entirely (extracted material is in bold face; the position of the gap is indicated by an underline; and square brackets set off embedded clauses):

main question: What did you see ___?
embedded question:
object of V: I wonder [ what they saw ___ ].
object of P: I wondered about [ what they saw ___ ].
subject in pseudocleft: [ What they saw ___ ] was a rhinoceros.
ordinary (headed) restrictive relative:
WH relative: The rhinoceros [ which they saw ___ ] was angry.
(gap without extraction: that relative: The rhinoceros [ that they saw ___ ] was angry.)
(gap without extraction: zero relative: The rhinoceros [ they saw ___ ] was angry.)
non-restrictive relative: The rhinoceros, [ which they'd just noticed ___ ], charged them.
"free relative" (headless):
object of V: I noticed [ what(ever) she had ___ in her hand ].
object of P: I looked at [ what(ever) she had ___ in her hand ].
subject: [ What(ever) she had ___ in her hand ] sparkled.

But English also has a few special types of in-situ interrogative clauses, notably in "reclamatory" and "incredulity" questions, where you're asking about the words someone just used (because you didn't quite hear them) or the substance of what they just said (because you can't believe it):
You saw WHAT?!

4.  For interrogative and relative WHO, there is a special relationship between Form2 and the syntactic function of object (the object of a "governor", V or P); for most speakers, Form2 rarely, or never, occurs as the (complete) subject of a finite clause.  On the other hand, for the ordinary personal pronouns, there is a special relationship between Form1 and the syntactic function of subject (of a finite clause); for most speakers, Form 1 rarely, or never, occurs as the (complete) object of V or P.

This asymmetry shows up in another place.  For WHO, Form1 is the all-purpose form, used when the conditions for Form2 are not satisfied.  So if someone says
I saw someone interesting yesterday.
you can ask
Who?
but even if you regularly say
Whom did you see?
you can't respond to the other person's assertion with
*Whom?

But for the ordinary personal pronouns, it goes the other way; you get Form2 all over the place.  So, for instance, if someone asks
Who did it, Brad or Janet?
you can reply with
Him.
but not with
*He.
even though you'd never say
*Him did it.
(There are many more intriguing facts like this.)

This is just a fact of life.  Though they are to some degree parallel, WHO and the ordinary personal pronouns differ in the way Form1 and Form2 are distributed.

5.  Languages that use extraction in interrogatives and relatives can differ in the details.  In particular, they can carry over case-marking from the gap to the extracted position -- the WH pronoun "inherits" the case appropriate to the gap -- or, since questioned and relativized constituents bear a kind of focus, the WH pronouns can bear a case appropriate to this focussing -- either a special case, or a defaulting to an all-purpose case.

English has (had) both systems.  Pure inheritance produces the Prescriptive System:

extracted subj: Form1
Who did you say [ ___ stole the tarts ]?
A extracted obj of V: Form2
Whom did you say [ they saw ___ ]?
B extracted obj of P, P stranded: Form2
Whom did you say [ they went to ___ ]?
C extracted obj of P, P fronted with obj: of course, Form2
To whom did you say [ they went ___ ]?

And defaulting to the all-purpose Form1 produces the Standard System:

extracted subj: Form1
Who did you say [ ___ stole the tarts ]?
A extracted obj of V: Form1
Who did you say [ they saw ___ ]?
B extracted obj of P, P stranded: Form1
Who did you say [ they went to ___ ]?
C extracted obj of P, P fronted with obj: Form2 (it's the whole PP that gets focus)
To whom did you say [ they went ___ ]?

6.  Assuming that the Prescriptive System was predominant for some time, what would move people to innovate the (now) Standard System?

Speculative answer: it has to do with the much greater frequency of main-clause subject questions (and relatives), versus all other types.  That would lead people to see Form1 as the one appropriate for questions and relatives and so to (mistakenly) extend Form1 to cases A and B (but not C).

7.  Two consequences:

First, once the Standard System had spread some, the Prescriptive System would increasingly be seen by many speakers as old-fashioned, formal, even pretentious -- and on the positive side, as serious and emphatic.

Second, in the Standard System, whom now occurs with any frequency only with fronted Ps -- a construction that is itself associated with a high level of formality.

Put these two together and you get whom itself seen as old-fashioned, very formal, serious, and emphatic.  And so available for situations in which you want those connotations.

(Note again the contrast between WHO and the ordinary personal pronouns.  For the ordinary personal pronouns, Form1 has, for many speakers, come to be seen as formal, serious, and emphatic -- a development that leads some of these speakers to prefer "between you and I" and the like in serious contexts.)

When you add in prescriptive injunctions against who in cases A and B (which began in the 18th century), which instructed writers and speakers to replace (some occurrences of) who by whom, the way was open for hypercorrection, which would produce whom in contexts not available for it in the Prescriptive System.

And so it has come to be.  See Mark Liberman's Language Log piece on

(2) [ Whomever controls language ] controls politics.

Mark refers there to a tongue-in-cheek piece by James Thurber on whom, in which Thurber recommends: "'Whom' should be used in the nominative case [i.e., for a subject -- AMZ] only when a note of dignity or austerity is desired."

Earlier, and more stunningly, Geoff Pullum displayed a photo of an American flag with this protestors' legend printed on it:

(3) Cheats Murderers Rapists Thiefs Terrorists [ Whom Captured Killed Enslaved Millions Of Africans ] [ Whom Killed More Natives Than Nazis Did Jews ] ...

Meanwhile, in e-mail from John Singler, 11/1/05, this item from a Brooklyn neighborhood website (punctuation as in the original; I forbear to indicate embedded clauses):

(4) Can anyone help me with information about Lillian Krum whom may have married a "Cunningham" ,whom had a son named Albert James Cunningham that married a Lillian Smith ,with whom they had 4 children and moved to Florida . . .

Finally, one I collected myself:

(5) Key theorists, [ whom are almost all white men ], control...
  (Stanford Humanities Center fellow, in lunchtime conversation, 10/31/05 -- possibly reaching for more formality and seriousness, in conversation with Lani Guinier).

8.  These involve straightforward subjects with no obvious factors favoring whom.  From them we shade into some cases involving predicatives, a case not in the lists above:

(6) Thank you so much [ whomever you are ___ ].
  (letter to Palo Alto Daily News, 9/17/03, p. 10, thanking a good samaritan)

(7) Who I am today is [ whom I've always wanted to be ___ ].
  (cited by Wilson Gray on ADS-L, 6/8/06)

(8) I REALLY DON'T CARE [ WHOM YOU CLAIM [ YOUR ANCESTOR WAS ___ ] ].
  (cartoon cited by Geoff Pullum here)

For predicatives, the Prescriptive System insists on Form1 for ordinary personal pronouns ("It was I"), on analogy with the system of Latin, while the Standard System is very strongly in favor of Form2 ("It was me"), as the default form, and that might influence the choice of forms for WHO.

9.  Now we come to two cases where the appearance of whom for a subject has some structural motivation.  People have been noticing examples like these for a hundred years at least (there is some discussion in MWDEU of these precedents), and it's fairly easy to find new ones.

There are two main cases. 

9.1.  In the first, we have an object clause (usually the object of a P) with WHO as its subject.  The pronoun then immediately follows the governor, and could easily be mistaken for its object (even though it's the whole clause that's the object).  In fact, I believe there are languages in which a WH pronoun in this position regularly (or optionally) has its case determined by the governor.

I'll call this case "in-situ subject of an object clause (ISOC)".  Examples:
 
(9) This is not a picture of a political tide running in one direction. It is a picture of voters venting their frustration on [ whomever [sic] happens to be in power ].
  (Wall Street Journal quoting USA Today, reported by Ron Hardin in sci.lang, 11/6/03; that's the WSJ's "[sic]", by the way)

(10) ... and works to ascertain God's leading as to [ whom should fill certain positions within our congregation ], the full congregation radifies these appointments in ...
  (here, cited in eggcorn database, 8/29/05, for "radify")

(11) This month's social has an Academy Awards theme and two prizes will be given away.  One prize will be awarded to [ whomever successfully predicts the most winners for this year ].
  (e-mail to the QUEST (Queer University Employees at Stanford) mailing list, 2/22/06)

(12) ATHENS, Ga. - Authorities are searching for [ whomever posted a long list and description of supposed sexual encounters between dozens of high school students on the online networking site MySpace.com ].
  (AP story reported by Ron Hardin on sci.lang, 10/2/06; the Washington Post version had whoever)

9.2.  The second case is a bit subtler.  Again there's an object clause, but this time its subject has been extracted and now appears at the front of a higher clause.  Still, the gap of extraction immediately follows the governor (most often, a V), so it's in a position where some languages (I believe) allow the governor to determine the case on this element; if this case is inherited by the extracted element, whom would be predicted.

I'll call this case "extracted subject of an object clause (ESOC)".  Examples, beginning with the Buckley example, repeated here:

Restrictive relative:
(1) The answer, shaped in the National Security Council, is for the American military to make targets of Iranians [ whom they believe [ ___ are fueling attacks ] ], a decision that Mr. Bush made months ago that was disclosed only last week.
  (from the New York Times of 1/15/07, here)

Main question:
(13)   [Robert Coren:] Well, I think what works best for Steph is what works best, but, much as I'd dearly love to have Sim there, I think cons work best in pleasant outdoor weather.
         [Sim Aberson, a meteorologist:] And whom do you think [ ___ would be responsible for the POW ]?  I hereby declare that if I can't go, only UOW will occur.
  (exchange on soc.motss, 1/18/05)

Non-restrictive relative (twice):
(14) Now there's antiwar Connecticut Senate candidate Ned Lamont, [ whom Moulitsas predicts [ ___ will defeat Joe Lieberman in the party primary ] ]. He'll lose. And there's Montana's senatorial candidate Jon Tester, [ whom Moulitsas predicts [ ___ will beat incumbent Senator Conrad Burns in November ] ].
  (Ben Shapiro column, reported by Ron Hardin on sci.lang, 6/14/06; see here)

Restrictive relative:
(15) Bobby Hodges, a former Texas Air National Guard general [ whom "60 Minutes" claimed  [___ had authenticated the memos ] ], says that when he was read them over the phone he assumed they were handwritten and wasn't told that CBS didn't have the originals.
  (Wall Street Journal, reported by Paul Kriha in sci.lang, 9/13/06; see here)

Non-restrictive relative:
(16) The 77-year-old Chomsky, [ whom Chavez mistakenly thought [ ___ was dead ] ], is famous as a linguist and as an opponent of U.S. foreign policy.
 ("Chomsky still best seller", Mercury News "Celebrities" section, from Chris Waigl in e-mail 9/27/06; see here)

(MWDEU has similar examples from Shakespeare -- note, from well before the age of the prescriptive grammarians.)

10.  Still more subtle (and much less frequent) examples involve subjects that are understood as denoting affected participants in some event -- that is, subjects that have some of the semantics of objects.  I have one case with the verb GET taking recipient subjects, and one with the subject in a passive:

(17) ... Hillary, or [ whomever gets the nomination ], gets a shot.
  (John Meacham, senior editor of Newsweek, on the Imus radio show, reported by Ron Hardin in sci.lang on 6/13/03)

(18) The employee, [ whom has been fired ], did not have the authority to take the equipment or the data home.
  (News editor on the Imus show, reported by Ron Hardin on sci.lang, 7/1/06)

11.  Finally, combinations of these factors. 

First, an affected subject (subject in a passive) in combination with ESOC:

(19) Currently I am reading Barry Strauss's The Trojan War.  Strauss is the type of classicist [ whom in Who Killed Homer? we once thought [ ___ were desperately needed for a dying profession ] ].
  (Victor Davis Hanson, quoted by Ron Hardin in sci.lang, 12/3/06; see here.  The wonky number agreement is a bonus.)

Then, a predicative counterpart to ISOC -- a predicative fronted in an embedded clause, where it immediately follows a governor (in this case, P):

(20) ... the courage to be open about [ whom I was ___ ].
  (James McGreevey, cited by Mark Liberman here)

12.  Some lessons.

12.1.  People struggle to discern system and meaning, on very imperfect evidence.  Yes, they thrash about and make mistakes, but mostly what we see is an attempt to find a system in what they're confronted with.  People come up with systems that are possible as languages -- they are attested in other languages -- but are not, in fact, necessarily the predominant systems of other speakers around them.

Then those systems can spread.

12.2.  The "same" grammatical category in different word classes might have quite different principles of distribution.

12.3. "Nominative" and "accusative" (or "subject case" and "object case") aren't bad names, but the labels aren't definitions, and they aren't descriptions.  We choose the labels because of the ways the forms are frequently used, not the other way around.  That's why I insisted on making the case names arbitrary and strange -- "Form1" and "Form2" -- so that you wouldn't import expectations about what "nominative" and "accusative" forms SHOULD do.

We inventory the distinct forms and list the ways they're used.  Then we pick names.  But the names are just expository icing, not anything of significance in the description.

zwicky at-sign csli period stanford period edu

Posted by Arnold Zwicky at January 23, 2007 10:31 AM