Chomsky -- Put up or blah blah

Ken Hanly khanly at mb.sympatico.ca
Wed Mar 29 18:51:45 PST 2000


I thought that the view that scientific theories had to be falsifiable was no longer held by most philosophers of science. The Lakatosian position as I recall is that there is a core of theory that is not regarded as falsifiable. The theory of relativity would be a present-day example. An earlier paradigm would be Newtonian mechanics. As Kuhn points out theories are not typically rejected by being falsified but through paradigm shifts--as with the Ptolemaic and Copernican theoryes. Another problem with the Popperian falsification viewpoint is that there can be many falsifiable but competing theories but the theory does not give any positive reason for choosing one unfalsified theory over another. The theory can tell us when a theory is false and to be rejected but not its degree of confirmation. Some type of confirmation theory such as that of Bayes is needed. Duhem and Quine have shown as well that any theory can be saved from falsification by adjusting auxilliary hypotheses etc. There is no such thing as a crucial experiment that makes or breaks a theory. I am not expert in this area but the difficulties with the view that scientific hypotheses must be falsifiable are well established. The "naturalized" approach to the philosophy of science recognises such charactertistics as simplicity, predictive success, and heuristic fertility, as characteristics of good scientific theories.

Theories in neo-classical economics would surely fail the falsification test. But I suppose they are supposed to have some of the virtues listed above and of course best of all the theories justify free market capitalism, the real test of the acceptability of economic theories within capitalism.

Then there are the Feyerabendians who claim that even observations are theory-laden and so there is not even the conceptual possibility of confronting a theory with a simple falsifying observation.

Cheers, Ken Hanly

Scott Martens wrote:
>
w
>
> I don't think the problem is that linguistics isn't "hard" enough, it that it isn't done in a very scientific manner at all. I came into linguistics after washing out of graduate school in high energy physics, so I knew what the scientific method was and I knew that people weren't using in linguistics.
>
> They could use it, but it takes more than just the scientific method to make that happen, it takes a culture of science. People need to know that the very first thing that will happen when they publish a paper is that someone will criticise it on the grounds that it can't be falsified, and revise what they write accordingly. They need to understand that results have to be repeated, and that studies have to have controls in order to be meaningful, and that they have to report data that is contrary to their hypotheses and try to analyse the amount of error in their work. In physics, this is something you get enculturated to - it's part of a sort of scientific boot camp training - but linguistics programmes generally don't teach these kinds of things. No matter how much math you have or how much formalism you use, without these falsification you just aren't doing science.
>
> A physicist with a theory has to show that some kind of experiment could falsify it. A physicist must, therefore, consider the possibility that he or she is wrong. When you realise that your ideas may be completely discredited, you tend to treat your opponents with much more respect. In linguistics, no one ever has to concede that they were wrong, because they don't really try to falsify their hypotheses and they can fiddle them endlessly in order to evade falsification, and so they have no reason to fear their opponents.
>
> This is especially true because a lot of linguistics is done in journals with a particular slant. There is relatively little impartial peer review. People working in G&B publish in G&B oriented journals, reviewed by G&B people and are only read by G&B people. Other schools of linguistics are unfortunately similar in behaviour.
>
> Until there is a scientific culture in linguistics - something that is growing slowly, but is growing - all that will distinguish one theory from another will be the largely subjective judgement of how "intuitive" and "neat" it is, and as long as results are judged on such a subjective basis, the disputes will never get quieter.
>
> I have some hope that cognitive scientists, who are enculturated into scientific culture and do publish in common journals rather than setting up a journal for each sect, will help change some of this in linguistics.
>
> Scott Martens
>
> ---
>
> Envoyé par Moncourrier.com
> Vous aussi obtenez gratuitement votre adresse électronique sur Moncourrier.com



More information about the lbo-talk mailing list