Body and One: Corpus count fail
« previous post | next post »
This morning we're continuing to explore the difference between somebody and someone, which all started on 11/10/2009, when David Landfair wrote to Arnold Zwicky ("Ask Language Log: someone, somebody") to ask for protection against the bizarre idea that someone is nominative (like he/she/who) while somebody is accusative (like him/her/whom).
Arnold offered first aid ("Yes, it is preposterous"), quoted MWDEU ("In the 20th century, … someone has come on strong … But both, of course, are equally standard…"), and suggested checking out corpus studies. I offered some corpus-based evidence that -one remains more formal than -body, and (nevertheless?) is gradually superceding it, even in conversational speech ("Less body in your lexicon", 11/11/2009; "Body loses Supreme Court appeal", 11/13/2009).
In the comments on the second of those posts, Geoff Nunberg suggested that "the forms in -one are marked for closeness to the speaker and individuation, and thus preserve some part of the sense of one", citing Dwight Bolinger, and offered as evidence the Google counts for "some{body|one} special" vs. "some{body|one} or other". Along similar lines, David Cantor quoted his wife's opinion that "Someone is used when referring to a specific and defined group 'Someone in this room committed the murder,' while Somebody fit better with an undefined group: 'Somebody left a mitten on the sidewalk.'" And marie-lucie agreed.
I thought that I'd start by checking Geoff's test in a number of corpora where the counts are more believable than Google counts are. (The numbers in parentheses are rates per 100 million words).
| somebody special |
someone special |
body% | somebody or other |
someone or other |
body% | |
| TIME | 7 (6.4) | 6 (5.5) | 54% | 6 (5.5) | 0 (0) | 100% |
| COCA | 18 (4.4) | 98 (24) | 16% | 10 (2.5) | 14 (3.4) | 42% |
| BNC | 5 (5) | 21 (21) | 19% | 11 (11) | 2 (2) | 84% |
| LDC News | 33 (1.3) | 228 (8.9) | 13% | 24 (0.94) | 24 (0.94) | 50% |
Although the actual rates are all over the place — probably due to differences in the mixture of registers and time-periods covered — there's a consistent pattern. As Geoff predicts, the proportional use of somebody is greater, in every source, for the context "__ or other" compared to the context "__ special".
However, we're looking at subtle proportional differences rather than categorical ones. And when we look at the hits in detail, the whole enterprise seems to fall apart. Geoff (as I understand him) wants "some{body|one} special" to represent the case where the writer or speaker has a particular individual in mind, and "some{body|one} or other" to represent the case where the reference is open. But when we check, we find that that the facts are almost exactly the opposite of this.
Cases of "some{body|one} special" are almost always either not referential at all, or entirely generic. Here's a representative sample of the COCA cases:
Make you feel like you were somebody special.
You act like you're somebody special.
I knew that it would take somebody special…
it's just a great place to be with somebody special
Got to have a chance to move out, to grow, to be somebody special.
Find someone special in six months, guaranteed.
… that her baby girl would turn out to be someone special.
Mia was the last of the trio to find someone special in her life.
The itch to travel is strong, but go with someone special …
Sandee had never given up on finding someone special to share her life with
And in contrast, cases of "some{body|one} or other" usually refer to some specific person whose name the speaker doesn't know, or at least to a variable taking on one of a notionally specific series of values:
he remembered somebody or other saying it was simplistic
One by one, somebody or other stopped me and asked with whom I had business
Does a Speed somebody or other got an office in this building?
I was reading somebody or other today who said …
the Senate have now signed a letter saying they'll hold up the nomination of somebody or other
I only speak to my father once or twice a year, and even then it's usually to argue about money that I owe somebody or other
Nearly every week, someone or other from the construction site headed to the emergency room.
…a market in which everything of value should be owned by someone or other.
my mom went visiting someone or other and probably isn't coming back
He has his mother's large eyes and a flat mouth from Someone or other.
what he was telling Dino, about someone or other who refused to fight in Argelia
A man was holding forth. He was introduced as someone or other, art critic, veteran of the War, …
On their face, these observations could be taken to suggest that the Bolinger/Nunberg/Cantor/marie-lucie theory is backwards: perhaps somebody really suggests specificity, given its greater proportional usage in with specific referential intent in the "__ or other" frame? But then again, maybe the proportional differences just reflect a difference in the frequencies of the two frames across registers, as Geoff suggested — you could get the same result if "__ or other" is simply commoner in colloquial contexts.
A difference in referential properties between someone and somebody, if it exists, must be a relatively subtle or contextually limited one, since all sorts of referentiality/specificity conditions seem to be associated with both words in the examples given above.
But in any case, the distribution of "__ special" and "__ or other" doesn't give much leverage on the question. To explore this question effectively, it seems, we may not be able to find a convenient corpus-count proxy, but instead need to annotate a larger list of examples with a better-specified set of features.
Craig Russell said,
November 14, 2009 @ 12:21 pm
Hmmm… it seems like everyone has some kind of personal theory about -one/-body based on their own gut instinct. Mine (if anyone cares to hear it) is that the -body forms are less common in formal contexts because the word "body" in itself evokes the physical and the personal, and lacks the distance necessary for formality.
Bodies have ingrown hairs and sweat glands and odors; ones are hypothetical and philosophical. Even if people don't consciously go through this thought process every time they choose one or the other forms, I suggest that these considerations subconsciously steer us away from "-body" forms in certain contexts.
If that's true, then it's easy to imagine that once the habit of using "-one" forms is developed, it could easily spread over to non-formal writing as well; after all, the two forms are interchangeable, and there's no reason NOT to use "-one" forms in casual writing/speaking.
One thing I do know is that when I'm writing I'm more likely to use "nobody" than "no one", because I'm always unsure of how to write the latter. Noone? No-one? No one? Noöne? (if you're writing for the New Yorker)
Anyway, that's just my opinion on the matter; as I said, with this question, it seems like everybody has a different one.
John Lawler said,
November 14, 2009 @ 1:01 pm
My own personal theory is that *one words simply falute higher than *body words (where * represents the Kleene closure, rather than ungrammaticality or reconstructed status).
When I'm going to use one of them, I don't use a *one word unless I'm consciously attempting to speak more formally, like when I state a generalization in a lecture, or in some context in which somebody might be taking notes. Or (and this is important) when I prefer one final syllable to two, for euphonic reasons; I often do.
Once again, this is a personal theory. (See http://www.umich.edu/~jlawler/disclaimers.html for disclaimers). However, since I believe that individual usage variation is orders of magnitude more important than other variation, that's what I'd expect anyway.
Robert Coren said,
November 14, 2009 @ 1:17 pm
As far as I'm aware, I have no preference for either form in any context, but I just wanted to stand back and admire John Lawler's elegant coinage "simply falute higher".
mdl said,
November 14, 2009 @ 1:57 pm
I'm coming late to this discussion, but I would have assumed preference is linked to regional dialect. Have you tried corpus checks to correlate with words of known regional preference, like "ain't" or "yonder"?
The perception of -one as more formal than -body may just be a side effect of the higher social esteem of Northern accents.
Sili said,
November 14, 2009 @ 2:26 pm
I have to agree with Robert Coren that John Lawler does indeed backformate very well.
That first example makes me think that I would most likely say "Have you found that special someone yet?" (note the inversion) rather the *body version. Google seems to agree.
Geoff Nunberg said,
November 14, 2009 @ 3:12 pm
I don't have any strong commitment to Bolinger's account, though I've learned as most linguists have that it's never a bad idea to give the benefit of the doubt to his remarkable semantic Sprachgefühl (Bolinger could hear the subtle difference between a marlinspike). And in his defense, it's worth nothing that all the examples Mark gives of "someone special," but only the fourth of the examples of "somebody special," seem to denote someone "close to X," where X is, not the speaker but the subject with whom the speaker is apparently in empathy. As I said, I don't have Meaning and Form to hand so can't say exactly how he puts this, but these at least seem in line with his overall approach. In this connection, Google suggests that "my special someone"/"your special someone" are dramatically more frequent than the alternatives with somebody. Though again, this could all be an epiphenomenon of the register difference.
empty said,
November 14, 2009 @ 3:17 pm
It happened again: I thought of a song containing either "someone" or "somebody" (in this case "Someone to Watch over Me"), and upon investigation it proved to contain both. How many examples do we have now? Five or six? And I've only come across one song that has only one of the two.
I'd say there's no doubt that for songwriters there's only difference between these words that matters enough to matter.
Geoff Nunberg said: Well, you wouldn't want to leave out the greatest "someone" song ever:
John Lawler said,
November 14, 2009 @ 3:42 pm
But it's not only songwriters — it's anybody with an ear for phrasal rhythms and a desire to make use of them in speech. As I pointed out above, that's important: sometimes you want one syllable, others two, for your own purposes. This is part of what pronouns are for — disappearing into the background noise while carrying the necessary sense forward in some way or other.
Tim said,
November 14, 2009 @ 5:37 pm
I don't think I've read all the comments to all the posts on this topic, so maybe someone (!) has already suggested this, but perhaps the numbers should be run for "anyone/anybody" and "no-one/nobody", just to see if they pattern up the same way as "some-" and "every-".
Franz Bebop said,
November 14, 2009 @ 7:06 pm
I agree Craig Russel.
It's sensible to compare usage of "someone" and "somebody" with the corresponding pairs "anyone/anybody," "no one / nobody," and even just "one" and "a body." That last pair of pronouns (pseudo-pronouns?) would probably shed a lot of reflected light.
It's also worth comparing with constructs like "someplace" vs. "somewhere", as well as "sometime" (but not *somewhen) and "somehow" (but not *someway, not as an adverb) and "for some reason" (but not *somewhy).
Dan T. said,
November 14, 2009 @ 9:17 pm
The Partridge Family:
Doesn't somebody want to be wanted like me Where are you
Faith said,
November 15, 2009 @ 12:09 am
I am so taken with the idea of linguistic first aid that I can't take in any of the rest of this post. I am reminded that Ruth Hubbard, Stephen Jay Gould, and Richard Lewontin used to call themselves the biology fire brigade, attempting to correct misapprehensions as they sprung up in popular culture.
D.O. said,
November 15, 2009 @ 4:49 am
@John Lawler: I liked your Disclaimers List. Would you consider to put Organic or Kosher on the top?
John Lawler said,
November 15, 2009 @ 10:45 am
@D.O.: Thank you, but it's not mine. I found it somewhere a long time ago and when the Web came along, it was easy to put it on. I am responsible for the formatting, however. I think I get the joke, but neither Organic nor Kosher officially disclaims anything.
J. Goard said,
November 15, 2009 @ 11:10 am
I mostly study nominal reference, crosslinguistically and in L2, and one central observation is that it varies a lot and moves quickly, across constructions and dialects. Using an accessibility scale like Ariel (1990) or Gundel et al. (1993 inter alia), it is easy to understand why:
(i) Frequency is very high.
(ii) Speakers are inclined to stretch the limits in either direction. (Will ya look at that moon! vs the moon; Love the potatoes, honey. vs these potatoes)
(iii) Reanalysis of a situational exaggeration as a plain description is a common source of semantic change.
So it doesn't surprise me that our intuitions are so diverse. I, for one, find somebody natural mostly in nonspecific reference or in some constructions where it's stressed, like Somebody's had a little too much to drink. I would guess I say someone a large majority of the time. Several of the examples in the post with -body sound really odd to me.
Especially in this area, though, the corpus data probably obscure interesting dialectal/idolectal variation, don't you think?
Craig Russell said,
November 15, 2009 @ 11:48 am
More evidence from song lyrics (hope nobody's brought this one up):
The new song by Kings of Leon "Use Somebody" uses both "somebody" and "someone" very frequently, and alternates freely between both. It's very clear from this song that the alternation is entirely based on sound.
"Somebody" has three syllables; "someone" has two. In parts where the song moves slowly and lingers on every syllable, "somebody" is preferred. In parts where the song speeds up, "someone" is preferred. Often the two words come one after another:
You know that I could use somebody,
Someone like you,
Give it a listen:
http://www.youtube.com/watch?v=JCZfJ5ai07U