Chomsky vs Pater

Theories- like everything else- are either useful or useless. If they are useful, they are likely to be part of a system which yields more and more functional information. But if a thing is useful in improving predictions or performance or can otherwise 'pay its way', it stops being 'theory' or the province of pedants. It is just part and parcel of a profession or an industry which is subject to competitive pressure.

The law of increasing functional information, in any useful field- or, indeed, one under selection pressure- does not militate for a unique structural causal model or one and only one way to 'carve reality up along its joints' at least when it comes to co-evolved systems where there is likely to be 'plasticity' or 'robustness' such that correlation rather than causation is what is observed or what gives rise to 'surprisal'- i.e. is taken greater notice of. 

Chomsky is not interested in utility. He wants magic. For him, a theory is 'adequate' only if it can distinguish between what is possible and impossible in advance. Ptolemaic Astronomical theory is adequate because it could say in advance that it was impossible for man to land on the moon. NASA's astronomy is inadequate because it can't assure us even that faster than light is impossible. 

To be fair, Chomsky wasn't speaking of Astronomical theories. His field is linguistics or cognitive science. He insists that a minimal condition of adequacy for a theory of human language is that it distinguish possible from impossible languages. Sadly this involves having an a priori method of discriminating what is possible and impossible for human beings since, so far as we know, anything possible for us can be communicated by us in some sort of language. If that were not the case thought would not have a representation as 'internal language'. 

Thus, what Chomsky thinks 'adequate' for his theory is nothing less than magic! But with magic, anything is possible. 

It is reasonable is to say a theory of human language must also be a theory of language for non-humans. Otherwise, the theory is empty. It just returns 'because humans are super-special' to any and every question. This isn't compatible with the theory of Evolution. Humans simply aren't super-special. 

At one time there was a shibboleth that a proper Science could solve its own demarcation problem- i.e. say what was or wasn't a scientific statement. But this was foolish. A guy might be as crazy as a bed-bug but might still be doing good science. Another guy might gas on endlessly about Scientific Method but produce nothing useful. 

Turning aside from theories of language- which we can all come up with if we need to- what about 'a theory of Universal Grammar?' If what it amounts to is 'humans have it innately. Non-humans don't.' then it isn't a theory. It is merely bigotry. It doesn't explain anything. It doesn't predict anything. It doesn't improve the performance of anything. It is a useless piece of shit. 

Consider the demand for 

Cognitive Plausibility: A theory of language should reflect constraints imposed by the human mind.

In which case, human theories are constrained. Try to do without them. Just look for statistical regularities. Since language and thought and theories are constrained in various ways, give priority to facts or techniques of a useful kind. In other words, ignore what people say. Just look at what they do and decide if it is worthwhile to imitate them.  But this means ignoring people like Chomsky. 

If it predicts that wildly unnatural or unlearnable systems are possible languages, it fails to model human linguistic capacity.

Human beings have a capacity to suffer cancer though  the thing is highly undesirable. This does not mean modelling human biology has failed if it predicts a very bad outcome- e.g. getting cancer. One may say 'if we didn't predict cancer, there would be no cancer. People would enjoy better health.' But this is magical thinking.

Typological Insight: Linguists observe recurring patterns across languages. A theory should explain why certain structures are universal or common, and why others never occur.

This can be done very quickly. It isn't rocket science. After that you can have a theory of why there is a lot of commonality in the way people walk or cough. But who will listen to you gas on about it?

Learnability & Acquisition: Children acquire language rapidly and robustly.

Machine learning may do so too after the preliminary spadework is done. Indeed that is what seems to have happened over the last five years.  

A theory must account for this by ruling out systems that are too complex or unstructured to be learned.

Again, this can be very quickly done. We don't need to devote resources to a theory of human language which explains that humans have the sort of language humans have because they are human. Also, they walk like humans because they are humans.  

Chomsky wasn't stupid. He did know one or two languages quite well. But he didn't know how language works or that it could say things which were suggestive rather dogmatic. This is because he wrote dreary shite. Walter Pater, too, was a pedagogue. But, in his essay on style he wrote this-

SINCE all progress of mind consists for the most part in differentiation, in the resolution of an obscure and complex object into its component aspects, it is surely the stupidest of losses to confuse things which right reason has put asunder, to lose the sense of achieved distinctions, the distinction between poetry and prose, for instance, or, to speak more exactly, between the laws and characteristic excellences of verse and prose composition.

This is like Chomsky clinging to stupid shite he came up with sixty years ago. After all, once a particular position is seen as 'progressive' it must remain so for all time, right? It isn't fair that other people do amazing things which were previously thought impossible thus making you look like a stick in the mud. 

On the other hand, those who have dwelt most emphatically on the distinction between prose and verse, prose and poetry, may sometimes have been tempted to limit the proper functions of prose too narrowly; and this again is at least false economy, as being, in effect, the renunciation of a certain means or faculty, in a world where after all we must needs make the most of things.

In other words, utility is all that matters in a world of scarce resources and an unforgiving fitness landscape.  

Critical efforts to limit art a priori, by anticipations regarding the natural incapacity of the material with which this or that artist works, as the sculptor with solid form, or the prose-writer with the ordinary  language of men, are always liable to be discredited by the facts of artistic production;

Chomskian shite has been discredited by the facts regarding A.I.  

and while prose is actually found to be a coloured thing with Bacon, picturesque with Livy and Carlyle, musical with Cicero and Newman, mystical and intimate with Plato and Michelet and Sir Thomas Browne, exalted or florid, it may be, with Milton and Taylor, it will be useless to protest that it can be nothing at all, except something very tamely and narrowly confined to mainly practical ends—a kind of "good round-hand;" as useless as the protest that poetry might not touch prosaic subjects as with Wordsworth, or an abstruse matter as with Browning, or treat contemporary life nobly as with Tennyson.

In other words, language will never be reducible to a theory about it.  

In subordination to one essential beauty in all good literary style, in all literature as a fine art, as there are many beauties of poetry so the beauties of prose are many, and it is the business of criticism to estimate them as such; as it is good in the criticism of verse to look for those hard, logical, and quasi-prosaic excellences which that too has, or needs. To find in the poem, amid the flowers, the allusions, the mixed perspectives, of Lycidas for instance, the thought, the logical structure:—how wholesome! how delightful! as to identify in prose what we call the poetry, the imaginative power, not treating it as out of place and a kind of vagrant intruder, but by way of an estimate of its rights, that is, of its achieved powers, there.

There was a time when Linguistics was about learning different languages so as to be more cultured, creative and open minded, or- at the very least- more useful or productive in some quotidian occupation. Chomsky, foolishly, decided to turn his subject to shit for some paranoid reason to do with defeating the machinations of those evil bastards who ensured his life was prosperous and secure.