Chomsky, Popper, et al

Scott Martens smartens at moncourrier.com
Fri Mar 31 11:42:44 PST 2000


Okay, now for another long post to answer yesterdays posts. If I missed anybody, let me know.

For Carrol:

Thank you for your complement. I'm going to try to keep my arguments to digest posts like this one, since I seem to be posting a lot to this thread. I think the three post limit is a pretty good idea all in, all. It looks like it keeps my mailbox from getting too crowded, and provides me with a good excuse for ducking arguments I can't answer. :^)

For Bill Lear:

I look forward to seeing the results of your research. I am unaware of a generativist response to those issues, but that doesn't mean there isn't one.

Too many linguists are dismissive of the arguments of their opponents. I think that's one reason the generativists don't publish rebuttals often. However, that doesn't excuse me for being dismissive. I promise to consider the arguments you bring forward seriously, and you and everyone else who reads this is allowed to remind me that I said so if I don't.

For Yoshie:


> One of the disputes -- Chomskyans versus dependency syntax
> people -- that you described seems to have a political
> dimension. The idea that syntax is independent of semantics
> strikes me as akin to a liberal notion of justice: justice
> is procedural; free speech as form can & should be defended
> independent of content; etc. On the other hand, the extreme
> position that goes into the opposite direction -- there are
> only words & their properties and is no such thing as
> grammar at all -- may have an affinity with postmodern
> pragmatism & (implicit) empiricism.

That's kind of interesting, and I'm going to have to give it some thought. Both groups are occaisionally dismissive of postmodernism. I recall Chomsky saying some awful things about Lacan once ("a self-conscious charlatan" were his words I believe) while dependency syntax people often cite Anna Wierzbicka, who is very much a part of cultural anthropology (which I tend to associate with the extremes of postmoderism.) Yet, I find postmodernists citing Chomsky (often out of context) while dependency syntax people view themselves as highly structuralist, and the cognitive scientists are, to quote a friend of mine, "not post anything."

In the end, I suspect the network of affinities and influences in linguistics is much too complicated to make it too useful to attribute them to the many philosophical camps.

Shifting the subject slightly, deeply conservative academics often quote Chomsky on linguistics, while others use the arguments of anti-Chomskyan thinkers. Geoffrey Sampson at the University of Essex claims that politics motivates the agendas of all the schools of linguistics, and implies that Chomskyanism is a Marxist plot. Sir John Lyons, who ought to know better, hints at the same sort of convictions in his last few books.

Ironically, nearly everyone in linguistics and cognitive science who has well known political tendencies places themselves on the left. Only a few old-timers in the artificial intelligence community stand firmly behind conservative politics. (John McCarthy is the first example to come to mind.) Some conservatives tar all of the schools of thought in this area as tinted with the evil stain of Marxism, while others are quite happy to take the ideas of cognitive science, Chomskyan linguistics, and other schools, and happily use them to justify capitalism, imperialism, even racism.

Steve Sailor is a particularly horrendous example of this. Take a look at his articles at http://www.isteve.com/.

For Michael Pollack:


> Actually this example, as well, I think, as all of your others,
> support the Kuhn/Lakotos consensus against Popper. When Kuhn
> and Lakotos are cited, somewhat misleading, as saying that
> scientific theories are not falsifiable, they mean not
> falsifiable in the original Popperian sense, where, as he put
> it famously and often, a single bad predication is enough by
> itself to falsify a theory (whereas no number of positive
> results can ever prove one.) Kuhn & Lakotos argue that if you
> look at the scientific record, every decisive falsification
> of one theory is at the same time the verification of a
> competing and better theory -- and both moments are
> indispensible. They argue that no working scientist throws out
> an established theory on the basis of on bad result alone --
> if there is no good explanation of the discrepency, it becomes
> an "anamolous result."

I agree with you (and by extension Kuhn and Lakotos) that a single falsifying event does not kill well established theories. My biggest complaint about Popper is that he doesn't really investigate the sociology of science in any depth. Now, rather than tossing Popperian falsification out the window, I prefer to view it as a sort of "idealised science" that we have to keep in mind when doing actual science, but which is insufficient by itself to doing science.

The underlying truth about science is that is it not practised by people who are seeking some ideal knowledge for its own sake. Scientists care about their reputations and they work for money, power, prestige, respect, sex and all the other base motives that make the rest of us get out of bed in the morning.

It is therefore irrational to expect scientists to practise ideal Popperian logic. When a scientist has tied his career and reputation to a hypothesis, it is only human that he will resist falsifying reports. And to be sure, many experiments are the result of poor methods, errors, and unrepeatable flukes. Popper misses out in understanding the operation of the cultural mechanisms that handle those situations.

However, the beginning of the end of well-established theories is repeated, consistent anomalous results. Logically, a single falsifying event ought to be enough to topple a theory, but the existence of error, and the possibility of misunderstanding, and the necessary existence of human pride, stubbornness and even stupidity, insure that this process takes more time and effort than that.

As for why theories aren't rejected until an alternative exists, this fits Popper pretty well, I think. He believes that knowledge is conjectural, and I agree with him. Even if we know that some theory is incomplete, we still have to have some conjecture to explain the world. One that works most of the time is better than none at all.


> After all, a theory only gets established because it accounts
> for a large set of predications; it can always be kept just by
> limiting its scope. The anamolous results might be ignored, or
> they might become central to further research. But they won't
> become a falsification until there is a comprehensive account
> that can explain them and (ideally) all of what the old theory
> could as well. When that happens, we have a "falsification"
> precisely because we have a verification. Verification was of
> course a concept Popper was trying desparately to get free of.
> He wanted a world where you could just appeal to facts.

That is not how I read Popper. He contends that even facts are conjectural. You can't ever be sure your senses aren't lying.


> What you get instead is one where facts are of central
> importance, but finally inexplicable apart from the (competing)
> ideas that give them significance. The end result is that
> relation between ideas and facts becomes considerably more
> complicated and lifelike than it was in Popper's original
> formulation.

Rather, I see Popper as claiming that we are never able to have facts. We can only have conjectures, and that scientific conjectures are the ones that can have consequences that we ought to be able to observe, e.g. that are falsifiable. I can never know positively that I'm actually writing e-mail to real people, but if I wasn't, I ought not to be getting so much e-mail back. So, I think that there are other people on the 'Net, and you're not all just figments of my imagination. It's a good conjecture, but it isn't fact in an absolute, positive sense.


> It is not an argument that falsification is unimportant or
> impossible. It's simply that its is more complex than Popper
> stated it originally.

Yes, I can agree with that. Where I can't agree with in Kuhn is distinctions between "revolutionary" and "regular" science in terms of methods and criteria for accepting hypotheses. I also disagree with his use of "paradigms" to suggest that the very methods of science change as we discover more. On the whole, they don't. The final arbiter remains repeatable, experimental falsification, in more or less the manner of Popper.

In linguistics, Kuhn abuse is so widespread that I'm convinced the social sciences would be better off reading Popper, even with his flaws, and not reading Kuhn at all.

Scott Martens

---

Envoyé par Moncourrier.com Vous aussi obtenez gratuitement votre adresse électronique sur Moncourrier.com



More information about the lbo-talk mailing list