September 22, 2004

Facts, theories, fetishes

On phonoloblog, Marc van Oostendorp quotes Stephen Anderson:

If a paper on ‘the morphosyntax of medial suffixes in Kickapoo’, bursting with unfamiliar forms and descriptive difficulties, is typical of American linguistics, its European counterpart is likely to be a paper on l’arbitraire du signe’ whose factual basis is limited to the observation that tree means ‘tree’ in English, while arbre has essentially the same meaning in French.

Marc adds that "[t]his is obviously a caricature (of the way things were in the 1930s), and a funny one at that, but it is also accurate even to describe the current situation. [...] It is a mystery to me what explains this different academic and intellectual culture, especially since it seems to have been true for such a long time".

For discussion of some related issues, see Marc's extensive comment on my post on (some) Europeans' self-identified difficulty in "getting" the internet. And as a contribution to understanding the stubbornness of this difference between American and European intellectuals (maybe this should be Anglophone and Continental intellectuals?) , I'll match Marc's Steve Anderson quote with one from Adam Gopnik's collection of essays Paris to the Moon (pp. 94-97):

My favorite bit of evidence of the French habit of pervasive, permanent abstraction lies in the difficulties of telling people about fact checking. (I use the English word usually; there doesn't seem to be a simple French equivalent.) "Thank you so much for your help," I will say after interviewing a man of letters or politician. "I'm going to write this up, and you'll probably be hearing from what we call une fact checker in a couple of weeks." (I make it feminine since the fact checker usually is.)

"What do you mean, une fact checker?"

"Oh, it's someone to make sure that I've got all the facts right, reported them correctly."

Annoyed: "No, no, I've told you everything I know."

I, soothing: "Oh, I know you have."

Suspicious: "You mean your editor double-checks?"

"No, no, it's just a way of making sure that we haven't made a mistake in facts."

More wary and curious: "This is a way of maintaining an ideological line?"

"No, no -- well, in a sense I suppose ... " (For positivism, of which New Yorker fact checking is the last redoubt, is an ideological line: I've lived long enough in France to see that move coming ...)

"But really," I go on, "it's just to make sure that your dates and what we have you quoted as saying are accurate. Just to be sure."

Dubious look: there is More Here Than Meets the Eye. On occasion I even get a helpful, warning call from the subject after the fact checker has called. "You know, someone, another reporter, called me from the magazine. They were checking up on you." ("No, no, really checking on you," I want to say, offended, but don't -- and then think he's right: They are checking up on me too: never thought of it that way, though.) There is a certainty in France that what assumes the guise of transparent positivism, "fact checking" is in fact a complicated plot of one kind or another, a way of enforcing ideological coherence. That there might really be facts worth checking is an obvious and annoying absurdity: it would be naive to think otherwise.

I was baffled and exasperated by this until it occurred to me that you get exactly the same incomprehension and suspicion if you told American intellectuals and politicians, post-interview, that a theory checker would be calling them. "It's been a pleasure speaking to you," you'd say to Al Gore or Mayor Giuliani. "And I'm going to write this up; probably in a couple of weeks a theory checker will be in touch with you."

Alarmed, suspicious: "A what?"

"You know, a theory checker. Just someone to make sure that all your premises agree with your conclusions, that there aren't any obvious errors of logic in your argument, that all your allusions flow together in a coherent stream -- that kind of thing."

"What do you mean?" the American would say, alarmed. "Of course they do. I don't need to talk to a theory checker."

"Oh, no, you don't need to. It's for your protection, really. They just want to make sure that the theory hangs together... "

The American subject would be exactly as startled and annoyed at the idea of being investigated by a theory checker as the French are by being harassed by a fact checker, since this process would claim some special status, some "privileged" place for theory. A theory checker? What an absurd waste of time, since it's apparent (to us Americans) that people don't speak in theories, that the theories they employ change, flexibly, and of necessity, from moment to moment in conversation, that the notion of limiting conversation to a rigid rule of theoretical constancy is an absurd denial of what conversation is.

Well, replace fact (and factual) for theory in that last sentence, and you hav ethe common French view of fact checking. People don't speak in straight facts; the facts they employ to enforce their truths change, flexibly and with varying emphasis, as the conversation changes, and the notion of limiting conversation to a rigid rule of pure factual consistency is an absurd denial of what conversation ought to be. Not, of course, that the French intellectual doesn't use and respect facts, up to a useful point, any more than the even the last remaining American positivist doesn't use and respect theory, up to a point. It's simply the fetishizing of one term in the game of conversation that strikes the French funny. Conversation is an organic, improvised web of fact and theory, and to pick out one bit of it for microscopic overexamination is typically American overearnest comedy.

Gopnik ignores one crucial asymmetry -- the French don't actually have theory checkers, as far as I know. And lord knows, some continental intellectuals could use one. (Though in fairness, I have to admit that many American news organizations seem to have delegated fact-checking to the pajamahadeen in the blogosphere...) But I think Gopnik makes a valid point, all the same. WCFCYA is a characteristically American boast.

[Update: Bill Poser emailed:

I heard a story once about [famous French linguist X]. He gave a talk on autosegmental phonology at Harvard, speaking in very abstract terms. At length, someone in the audience intervened and asked him if he could give an example. X was discomfited. He paced back and forth, scratched his head, and finally strode to the board and wrote: CVCVC.

After reading this anecdote, David Nash emailed:

That gave me a chuckle Bill, and makes me recall how I (eyewitness!) was in an MIT seminar by [famous French linguist X] (where X is undoubtedly the same X as above!), who said something like "Take a word of Mongolian (or whatever)" and wrote up "CVCVCV" and said stuff about it; then "Take a word of Tubatulabal (this name I do recall)" whereupon he erased the "CVCVCV" and wrote up "CVCVCV".


[Update: Hubert Truckenbrodt's advice about how to write papers is also relevant:

American-style (caricature)

Right after the introduction, the author makes it clear, what the claim of the paper is.

"In this paper, I present a solution to the old puzzle why eggs play no role in the reproduction of whales. I will show that whales lay eggs, which dissolve in salt water about one minute after they were laid. And that is the reason eggs are not used for reproduction with whales. ..."

After that, the arguments for the claim are presented, if possible comparing the new theory with earlier claims. State the strong arguments that you have. After that, conclusion, and shut up. You say what you have to say, and that's that.

Traditional German-style (caricature)

Did the Ancient Greeks have something to say about the topic? What has been written about it since? Extra credit if you find someone who has commented on it a few centuries ago, and has since been overlooked.

From there, the author builds up slowly and steadily. After about half of the paper, the specific question of the paper comes into view. First in a vague, general way, then somewhat more concretely. One starts to suspect that there may also be a claim made later on. A preference of the author becomes noticeable. At some point, however, the paper ends. Sorry, we are out of space. More next time.

(via Kai von Fintel at semantics etc.)

The trouble with this kind of cultural stereotyping is that it can seem accurate as well as funny, while in the end offering such a varied and diffuse set of stereotypes for overlapping groups that even quite incompatible sorts of behavior on the part of a denigrated or admired type can be accepted as evidence that "yes, they're all like that". Thus in commenting on Kai's post, Tony Marmo offers a stereotype of the Dutch as fact-ridden and uninterested in general points other than negative ones. This may strike some as valid -- I don't see it, myself -- but in any case it's the precise opposite of the characterization of Europeans in general that Marc van Oostendorp started with.


[Jeff Erickson at Ernie's 3D Pancakes writes:

We have another word for "theory checkers" in my line of work. We call them referees. On the other hand, we're a bit short (and distrustful) of fact checkers. Far too many algorithms papers claim "in practice, people do X" when in practice, people don't do X, or even anything remotely similar to X. Conversely, in the more practical parts of computer science, fact checkers are endemic, but theory checkers are rare.

So does that make les algorithmistes the French of computer science? Quelle horreur!

While you're visiting Jeff's blog, take a look at his related post on "'Applied' Papers at SoCG, which mutatis mutandis describes the situation in several other technical areas known to me. ]


Posted by Mark Liberman at September 22, 2004 09:46 PM