September 17, 2004

Sidney Goldberg on NYT grammar: zero for three

Sidney Goldberg at The National Review claims that the reputed 150 copy editors over at The New York Times are either illiterate or asleep. He fulminates; he positively foams at the mouth about it. Naturally, Language Log felt it had to investigate. And having had my rabies shots, I was handed this plum assignment. So let's take a look.

The article begins with grumbles that are entirely about spelling. The Times twice misspelled lectern as lecturn; once misspelled "took effect" as "took affect"; and often misspells the preterite form of the verb lead as lead. (It should be led. Nasty little point, that: the metal known as lead has the sound of led but the spelling of lead; and meanwhile the verb read has a preterite that rhymes with led but has the spelling read, which looks like lead, only led is not spelled lead... Are you confused? Then it shouldn't be you that casts the first stone.) I'm with Goldberg all the way on these: these are spelling errors, and you've just got to get your spelling right.

So at this point I was hoping for some grammar examples to get us into more serious territory, but instead Goldberg wanders off for a while into a strange tirade against the The New York Times for ridiculing Dan Quayle, who long ago misspelled potato as potatoe when MC-ing a spelling bee (his flashcard was wrong), but not writing any jokey stories about how Chief Justice Warren Burger used to misspell homicide as "homocide" and Associate Justice Harry Blackmun (whose papers were recently released) used to circle the misspelling angrily when commenting on the Chief Justice's draft opinions.

However, Goldberg finally pulls himself out of this bitter rumination on political bias: "All of this concerns orthographic ignorance," he says; "But the Times commits innumerable errors in syntax and style as well. "Innumerable" you say? Aha! I'm all ears: I'm waiting for a long, juicy list of errors of syntax and style. Unfortunately, only three are supplied, and only one is illustrated from The Times itself.

1. That and which. The first charge is that the Times "consistently proves that it does not know the difference between ‘that’ and ‘which,’ greatly favoring the latter." There's only one thing he could be alluding to here: he's one of those people who believe the old nonsense about which being disallowed in what The Cambridge Grammar calls integrated relative clauses (the old-fashioned term is "restrictive" or "defining" relative clauses). Strunk and White perpetuate that myth. I've discussed it elsewhere. The notion that phrases like any book which you would want to read are ungrammatical is so utterly in conflict with the facts that you can refute it by looking in... well, any book which you would want to read. As I said before about which in integrated relatives:

As a check on just how common it is in excellent writing, I searched electronic copies of a few classic novels to find the line on which they first use which to introduce an integrated relative with which, to tell us how much of the book you would need to read before you ran into an instance:

  • A Christmas Carol (Dickens): 1,921 lines, first occurrence on line 217 = 11% of the way through;
  • Alice in Wonderland (Carroll): 1,618 lines, line 143 = 8%;
  • Dracula (Stoker): 9,824 lines, line 8 = less than 1%;
  • Lord Jim (Conrad): 8,045 lines, line 15 = 1%;
  • Moby Dick (Melville): 10,263 lines, line 103 = 1%;
  • Wuthering Heights (Bronte): 7,599 lines, line 56 = 0.736%...

Do I need to go on? No. The point is clear. On average, by the time you've read about 3% of a book by an author who knows how to write you will already have encountered an integrated relative clause beginning with which. They are fully grammatical for everyone. The copy editors are enforcing a rule which has no support at all in the literature that defines what counts as good use of the English language. Their which hunts are pointless time-wasting nonsense.

But it's nonsense that Goldberg firmly believes in, you see. There will be no talking him out of it. He'll be about 3% into his copy The New York Times and he'll see something like "the idea which they considered" and he'll spit coffee out into his muesli and splutter for his wife to bring him his red pen and he'l circle it furiously like Justice Harry Blackmun circling "homocide"; only the difference is that Blackmun was right, "homocide" is an error. Using which in an integrated relative clause is not, and nobody who has carefully studied the English language would think that it was.

2. What and which. The second of the three syntax points is that the Times "also repeatedly confuses ‘what’ with ‘which’: ‘What movie are you going to see tonight?’" Is there really a confusion here? This case is interesting (there's a beautiful discussion by Rodney Huddleston in The Cambridge Grammar, pages 903–904), but again Goldberg doesn't really know his stuff. The example he gives (which I think is made up) is grammatical. You see, there are differences between which and what, but I'd bet the mortgage money Goldberg couldn't characterize them.

The relevant difference here is semantic. Which is selective: it asks for a pick from a defined list. What doesn't care, and leaves a wide-open field of things to pick from. As a result, you need which in what is called the partitive construction, which makes the set to be picked from explicit: you say Which of these jackets is yours?, not *What of these jackets is yours?. Nobody gets that wrong, including The New York Times. Another consequence is that if you use a cardinal numeral, you'll need which rather than what: you say Which three people in this group photo have spent time in jail?, not *What three people... etc.

But when no range is made explicit, it's just common sense that tells you what the range must be, both are OK: Which movie are you going to see tonight? is normal; the range to pick from isn't specified, but you can get it from the local paper. What movie are you going to see tonight? is also fine: it leaves the field of movies wide open, but again, the practical possibilities are limited to what's on this week. The difference between which and what doesn't matter in those contexts, and both are common.

Unless Goldberg has caught the Times saying something truly ungrammatical like *What of the candidates will win? (which seems unlikely), is getting his underpants in a bunch over nothing at all.

3. Had to have been. Goldberg only has one other case. He caught an editorial saying: "By late 2002, you'd have had to have been vacationing on Mars not to know...". He harrumphs that this a "monstrous construction". And once again he's wrong. He presumably thinks that the last occurrence of have is redundant, on grounds that You would have had to be vacationing on Mars not to know could be used instead (the "Omit needless words" mantra from Strunk and White's toxic little book of crap is doubtless ringing in his ears).

But unfortunately that would change the meaning. To say that in order to be ignorant you would have had to be vacationing on Mars is to say that it would have been necessary for you to be on Mars enjoying your vacation right at that point, the point of ignorance. Whereas to say that you would have had to have been vacationing on Mars is to say, in effect, that it would have been necessary for you to be recently back from a recent Martian vacation.

That is, to be vacationing on Mars (call that being in condition A) is to be there right now, hence out of the office and unavailable for comment. To have been vacationing on Mars (call that being in condition B) is to be back in New York after your two-year flight home showing photos of the Martian desert around at the office. The sentence Goldberg complains about was saying you would have had to be in condition B not to know: to have been vacationing on Mars in the past few months.

So Goldberg is fairly clearly mistaken on all three of the grammatical sins he mentions (only one of them actually illustrated). He's fairly clearly howling at the moon on two of the areas he alludes to, and the syntax employed by The New York Times is right in the only case where he gives a quote.

It is so often like that. The amateur language pontificators (Sidney Goldberg is a retired senior vice president for syndication who used to work at United Media) know very little of the subject they're pontificating about. They don't look anything up in serious grammars or dictionaries. They just shoot their mouths off. And of course (let's face it, politics is involved here), if they're criticizing the (reputedly way too liberal) New York Times, then the (thoroughly and angrily conservative) National Review will publish them without any fact-checking.

Nobody does fact-checking on stuff about language. You may recall the two spectacular cases (here and here where Mark Liberman caught journalists (Cullen Murphy and John Powers) inveighing in print against "mistaken" word uses, and being wrong, in both cases, on all three cases out of three that they cited. "Can't anybody use a dictionary anymore?", asks Mark. It looks like the answer is no. And they don't know even about the existence of The Cambridge Grammar. Everyone just assumes that whenever a stern grey-haired male professional says somebody's grammar is wrong, the charge must automatically be correct and the accused guilty, and no facts need to be checked. Well, it's not so.

Mr Goldberg, now that you're retired, you can educate yourself. My elementary course on Modern English Grammar starts in just over two weeks, on September 27th; you have time to get your butt out here to California and sign up as a concurrent-enrollment student (it's filling up, but I'll save you a seat). It's not the editorial staff at The New York Times who need syntax lessons, it's you.

Posted by Geoffrey K. Pullum at September 17, 2004 09:25 PM