October 26, 2007

Cold comfort for whomever

Ben Zimmer forwarded to me this question from Jay Livingston, about Ben's post on the episode of The Office about whomever ( "It's a made-up word used to trick students", 10/25/2007):

I think you're wrong about whomever.  Yes, whom is disappearing, but I hear whomever all the time.  My secretary, for example, uses it as a stand-alone (I'm not a linguist, and I'm sure there's a technical term for this.  "Who's going to take these?"  "Whomever.")  And between you and I, it's used in a similar way as "between you and I," and probably for the same reason -- it sounds more sophisticated.

How can we find out if whomever is indeed on its last legs or whether its stock is still rising?  Do linguists have some way of counting the number of appearances a term makes in everyday speech?

Yes, we do. Imperfect, but good enough to show that Jay is somewhat right and somewhat wrong.

I searched the LDC Online corpus of conversational English, which I've used on some previous occasions to provide data for Language Log Breakfast Experiments™. It involves 14,137 transcribed telephone conversations, comprising a total of 26,151,602 words. Most of it was recorded in 2003 and 2004, though a small portion was recorded in 1991. The participants span a wide range of ages, regions, occupations and backgrounds.

In that collection, we find the following counts and ratios:

who/whom ratio
whoever/whomever ratio

(Note: to get the counts given above, I combined 1,008 instances of "whoever" with 28 of "who ever", and 23 instances of "whomever" with 1 of "whom ever".)

So whomever is not exactly common -- a frequency of about 1 in 1.09 million words. Not exactly "all the time", though of course Jay's acquaintances may be dealing from a different deck. And in these recordings, whom is still about seven times commoner than whomever, in absolute terms.

But in fact, these counts do support Jay's impression that whomever is holding out better, in proportion, than whom is.

How can I say this? Well, whom is 218 times rarer than who, while whomever is only 43 times rarer than whoever. Cold comfort, perhaps, but comfort nevertheless.

What about the contexts in which whomever is used? Are these mainly like "between you and I", i.e. contexts in which the objective case is not historically motivated?

Not really -- 27 of the 34 examples are like these:

yeah okay and i'm sure that you know as he gets to know them more he'll
you know consider their husbands or whomever good friends too

all you need is a letter from maybe your pastor or whomever and a letter f- b-
from you

now what what what do you usually watch like the local news or the or like world news with

There are only 7 examples out of 34 where whomever is not sanctioned by case-marking, at least by my construal:

i i mean i i see i'm not too familiar with like the laws that have been passed on something like that or if jesse jackson's trying to or whomever is is trying to
incorporate some law to congress federal law to congress about you know hiring you know about having your quotas and so forth you know

uh and certain things it's it it well it whoev- whomever is uh uh is holding the the highest positions of controlling

situations involved which would you know kind of allow for for people to think um you know like america's the bad guys or even the way you know they're you know whomever is r- ruling the country can spin it off
um to make to make america look like the bad people [laughter]

you know just uh how could the government or the Pentagon uh whom- whomever's in charge there been so careless

but if these terrorists or whomever knows that it's done randomly

and you know not lose yourself in someone else and whatever the words are it's just perfectly said
so whomever's listening from fisher they should go play that song and that's my ideal man [laughter] whomever she's singing about

right and i was like what happened what why didn't your um your investor analyst or your or whomever say 'here it comes'

It's possible that Jay's experience is quantitatively very different from this, but I'm skeptical. I think that it's more likely that he's the victim of a form of the Frequency Illusion, where the relative frequency of striking or characteristic patterns of usage is overestimated.

[Note: in the LDC Online transcripts, the sequence that I've rendered "Jesse Jackson's" was transcribed as "Jesse's actions". That's an example of what can happen when you hire transcriptionists from New Zealand...)

Posted by Mark Liberman at October 26, 2007 06:37 AM