© Rae West 1997, 2012, 2013

Summary: What follows is quite a long post to the group (apologies) intended to show that Chomskyan linguistics is a superficial study, looking at artefacts, while leaving the problems of language unresolved. In short, a stairway to nothing, except perhaps a fall. The only critic of Chomsky I've found is George Steiner, who however seems firmly to belong to the Jewish windbag school. Since this is the Russell newsgroup, let me start with a 40-year old quotation from Bertie Russell: "I think that philosophy, like science, should aim at agreement with fact and cannot treat language as autonomous." The title was suggested by another Russell comment, that during the disaster of the First World War he had difficulty not becoming a complete antinomian—Rae West

Note added November 2013: Chomsky was born in 1928; Stalin's last published booklet, in 1950, was on linguistics. There may be some connection—Rae West

I have to say I haven't intensively studied the various Chomskyesque schools; however, I have read 'Syntactic Structures', and comments in NC's other books, which in passing include remarks on language. Other sources include: the Canadian film 'Manufacturing Consent', various journalistic & Internet articles, a taped talk which I've put on this website, John Lyons's book (3rd edn), and occasional discussions with linguists.

Something which encourages me is that, as with relativity, I've found theoretically-qualified people are often unable to argue their case, at least in my opinion. Several examples: after a lecture by Chomsky, speaking to several people, one was researching into the most effective way to teach language. I asked her whether with Latin they couldn't try transposing English words into Latin, inventing approximate equivalents ('postpositions' and so on) where necessary, to give the feel of the way the language is structured. She had never even thought of that. Others, when I tried to explain why I had doubts about the idea of 'structure', spent their time writing down non-sentences (beginning with asterisks, in the style of '*Cat are mat'), unable to answer. The attitude behind Chomsky's well-known 'colorless green ideas sleep furiously' may well have been copied from Quine's 'pretty little girls' camp' of 1941, or Russell's 'present King of France is bald'. I might also comment on Uri Strauss' account of the resemblance of linguistic theory to a computer program, generating all acceptable sentences: Lyons (chap 4, p43) warns against this view ('.. suggests.. that the grammatical structure .. is being described from the point of view of the speaker rather than the listener; that the grammar describes the production, and not the reception, of speech..')

Before looking at structure, I want to get across the general idea of the randomness of ordinary language and the high percentage of 'slang'; most people don't realise this, as, naturally, they stay within their own little frames of reference. They get used to a phrase like 'used to', and never stop to think how meaningless it is. They say 'get it?' and it never occurs to them that 'catch it?' would be just as good, if not better. Suppose I type 'I'm sitting at home feeling fed up.' Remarkable as it seems, it's only a series of minor accidents that led me not to put 'I be chairing in home touching choked up.' If you search for such things, you find them everywhere; a huge proportion of ordinary language consists of such phrases; and each one has to be learned separately, as a sort of special creation. It's sometimes possible to remember first encountering a new phrase, like 'rip off' or 'a spot of bother' or 'cognitive dissonance' or 'pay someone out'. There's no possible general rule that can invent or generate them. Language translation programs, and human translators too, must in effect have libraries of phrases.

John Lennon said 'any road' for what other parts of Britain say 'anyway'; presumably the root of each was conveying the idea that the speaker is at a mental crossroads, with no reason to choose one particular path. Similar examples occur where English has evolved in different countries; 'right away', 'over there', 'right on' in date order illustrate. It's also clear in other languages: 'ceci' is something like this-here, 'peut-être' something like ought-to-be.

To look at this another way, consider English which is regarded as 'pure' or 'classical'. I've seen Walter Pater's prose described like this. Of course, on reading any such specimen, the unfamiliar words and usages seem ludicrously obvious.

This suggests that language can evolve in a purely individual way, as a heap of words and phrases, with words being somewhat interchangeable with phrases (e.g. 'red giant'—the clear distinction of words, and sentences too, needs written language). There must be a Darwinian-type mechanism (combined with fiats) which, after centuries of use, keeps sounds sufficiently different from each other, so that 'yes' sounds different from 'no', and common words (is, sun, grass, house..) short. If there's a shift of pronunciation, say in Grimm, one lot of vowel changes naturally have a knock-on effect on the others. When languages clash, Darwinian processes can be seen—e.g. the word 'coney' rhyming with honey dropped out of use in favour of rabbit, presumably because it resembled another word which I won't mention either. This is why Greek or Latin prefixes (ortho-, meta-, para-; sub-, super-; intra-, inter; de-, cum-, ex-, ad-) are so useful—they are pre-sorted for distinctive similarity. 'Man' and 'woman' it seems evolved from different roots, but may well have been settled on because of their distinctive similarity. Typesetters talked of 'mutton' and 'nut' spaces, one wider than the other; the words are in a similar mental framework. Anglo-Saxons used prefix a- as an opposite—a-down as not down, hence the modern use of 'downs', where the 'a' is lost, to mean 'hills'. Almost the whole modern medical vocabulary is built on classical words and their prefixes, and suffixes (-itis, -osis...) in this way.

The number of sounds which is used (plus intonation, body language, gesture) presumably also develops in a Darwinian way; too many taking too much effort, and too few being restrictive. Thus e.g. Chinese uses a tonal system; contrary to what might be thought, it makes perfect sense for the words or ideas pronounced at different pitches to be very different, as—once they are individually remembered—this makes for less ambiguity. And it makes sense for adjectives to be next to nouns; some languages have them first ('blue sky') and others second ('ciel bleu') but it's unsurprising they aren't separated by (say) four words.

I think the idea of structure probably came in with writing, and was invented or formalised by grammarians; one of the first being (I've read) Panini with Sanskrit. But it's noteworthy that these people come on the scene very many years after the language has existed; what claim do they really have for the primacy of their work? It's -also notable that all the commonest things tend to be irregular! All the most-used structures in European languages (I think not Arabic) have to be individually learnt, as anyone starting (say) German soon finds, with long lists of irregular verbs. Grammarians tidy up things, and give rules which can be used in new cases not yet covered, so languages are a bit of a patchwork, with traditional stuff and then new officialese tacked on. If they are successful, and form a sort of priestcraft, a special formal 'high' language may develop, helped no doubt by the applications side, e.g. bureaucracies. I visualise Noam influenced by his father, poring over ancient Hebrew, a good example of a bureaucratised dead language.

So far this seems to have to do with structure, except to suggest it too may be fluid, according to this model of the development of language as a chaotic improvised process. The problem is that 'structure' in linguistics is very elastic; it seems to be stretchable, in a way of course corresponding to 'intuition', which I suppose means one's own collected memories. In short, these people seem to cheat. For example, it seems obvious that 'the dawn came' means the same as 'came the dawn', so obvious that nobody asks how this is known to the user. If a linguist says, well, they obviously mean the same, then he/she can go on to say they have the same 'structure', modified 'in an acceptable way'. The problem is, it's only known to be acceptable because that's what you've been told. This perhaps is clearer in foreign languages: Il y'a can be converted into a questioning form, y-a-t'il. Presumably, if you're French, you get used to juggling the words round and putting in a t. But what objective evidence is there that it's a 'modified structure' and not just a new one? Or consider the parts of a sentence: Chomsky conservatively retains nouns and verbs, symbolised as N (or noun phrase, NP) and verbs. But how do you know what a 'noun' is? The only way is the circular argument that it's something found in a noun-like bit of the sentence. It could include an elephant, a 'unicorn', a round square, an event that will never happen, dishonesty, fruit. There seems no reason why a language shouldn't specially separate out special categories, as I believe some are supposed to do—e.g. things which are frightening, big & strong, small, feminine, or whatever. But the use of NP etc. just cuts out all this and imposes its own preconceptions based on English. (And there are borderline cases: what about gerunds? Is 'walking' a noun or a verb?) To quote Bertie Russell on Mill: 'He states for example that propositions are formed by putting together two names, one of which is a subject and the other the predicate. This I am sure appeared to him an innocuous truism; but it had been in fact the source of two thousand years of important error.'

To try to get a better sense of 'structure' in the linguistic sense, let's play with the idea that it's a pseudo-science. Thus 'deep structure' sounds profound; in fact if Lyons is to be believed (chapter 7) all it means is (e.g.) that 'John was persuaded by Harry to take up golf' has a 'deep structure' of 'Harry persuade John' with embedded sentence 'John take up golf'. Not 'deep' in the usual sense. Similarly there's a lot of discussion of 'infinite' languages. This sounds mysterious and scientific; but one presumes there's some practical limit, and therefore the word is selected more to sound impressive than for a serious reason. 'Universal grammar' sounds profound as: '.. view that the structure of particular natural languages, such as English, Japanese, Arabic, Navaho, etc., is constrained by the principles and categories of universal grammar.' But it seems to have an elusive, or perhaps no, meaning; at any rate, I've been unable to find any account of what 'universal grammar' actually means in a concrete sense. Lyons' appendix has some (evolutionarily useless?) material: e.g. frivolous use of set theory: '.. given that the vocabulary, V, of the language, L, is the two-member set {a,b}, L might have as its members not only a, b, ab and ba but also aa, aba, bba, etc.' Elsewhere I was depressed to see 'Theta Theory'; it turns out this name is based on 'thematic roles'—I can recall students proudly using the Greek psi as an abbreviation for 'psychology'. There's discussion of children and language; how do they do the trick of learning language—supposing in fact that children do? Why, with L.A.D.—'language acquisition device.' I suspect 'anaphore' to be a similar case. Another trick is to use the phrase 'natural language'. The fact is of course that there have now been generations of compulsorily-educated children who've had official English stuck down their throats; in the usual sense this is not 'natural'. Indeed, without some such theory as mine, it's difficult to see why children should need to be taught their own language at all (except for immigrants). I suggest the vocabulary and approach is modelled on aspects of science, or what looks like science, but has in fact little substance; like mathematical economics, or Piaget's psychology theory, or 'the other cognitive sciences' (Lyons blurb) it takes from relatively well-defined parts of knowledge and attempts to cover its nakedness with them. Uri Strauss and others also lay stress on the supposed unconscious following of rules. This seems not very accurate. Children readily produce their own constructions like 'an astronomist' or 'please stop noising', in which they're obviously following some rule. It's true that one forgets much of this stuff, but nevertheless it's presumably tucked away in there. One might compare this with learning to walk: an adult makes all kinds of fine muscular adjustments to stand and move around, and there's constant modification e.g. with new shoes or carrying heavy loads, and I can't see that language can't be explained by similar processes—and admittedly the full process in each case is mysterious.

To take a few examples of the supposed laws, or 'rules', discovered by linguistics -which 'restrict the range of possible sentences in all possible languages (this list from Uri Strauss):

Binding Theory attempts to account for the distribution of anaphors (myself, yourself, etc.) vs. pronouns (me, you, etc.). Take a sentence with an anaphor in it—usually you won't be able to replace the anaphor with a pronoun. The converse is also true. The theory posits a specific structural relation ("binding") that must exist between the anaphor or pronoun and its antecedent which determines whether an anaphor or a pronoun is appropriate.

Burzio's generalization: A verb that does not have an external argument (subject argument) fails to assign accusative case.

The Case Filter: Every overt noun phrase must be assigned case by a governing head that of a case assigning category.

The Empty Category Principle: Traces must be properly governed.

I would suggest the idea of a profound overall 'structure' is a myth imposed on the chaos of natural language, and reinforced of course by learning—there's no doubt most people develop over the years and years they spend learning language a feeling that certain pronunciations or word-orders or phrases are right, and this suggests something like a magnetic field of a part-sentence—it needs something to finish it. As Churchill said of Latin sentences, 'they fit together like polished mahogany.'

Let's look at English in particular (and despite pretensions this is a pidgin language, bodged together from many sources), and try to take account of spoken language. John Lyons at the start of his book says the object of linguistics is to study spoken language; this seems completely untrue, as there's no suggestion in his book that anyone has studied taped voices; all the examples are of rather formal written English, with a bias to artificial rather trivial stuff of the 'cat sat on the mat' type. Such usages as "murder is a terrible crime because you get hanged for it" or "I was diagnosed arthritis of the neck" or "quite a sophisticated machine to what it was" or "I was feeling inadequate to help her" (all from live TV) are cautiously avoided, or (Lyons 44) explained away by 'linguistic incompetence'. Written language usually has more of an official seal on it. But is a sign outside a building saying 'now open' English? I *think* this is dealt with by saying, well, it is really, because it's an abbreviation of 'this building is now open.' Is this English: 'Hume was an idiot. Not.'? Or consider the forcing of new structures into use; e.g. the American, or US TV, use, 'Who is this?' on phone; in UK English this is nonsense, to many people, because 'this' is near, and 'that' is far; so it should be 'who is that?' or 'who are you?' There's a Darwinian mechanism involved in phrase constructions, just as in words; e.g. tiny French kids asking what something is learn 'qu'est ce que c'est [something]'; the adults have learnt to make sense of this, but from the kids' point of view might as well just be (say) Q, so it seems absurd to impute any 'structure' to the kids' use.

[Note added Oct 9 2013] Most modern languages, such as English, are quite remote from primitive, unwritten languages. In the same way that modern cities with powered vehicles, electricity, water supplies and so on are different from ancient towns; and in the same way that war with rented experts and high-tech machinery is different from tribes of men with primitive weapons; and modern ships are different from canoes. There's a huge supporting network: dictionaries, reference books, commentaries, experts, and so on—worlds away form the simplicities of a memory-only language. In Chomsky's lecture, on this site, Minimimalist Explorations, he specifically brackets English and Swahili together, as specimens of 'language'. This alone is sufficient to cast doubt on his competence.

So it seems possible that a language may be made of a collection of separate structures, e.g. a question, a reply 'Yes it is' as confirmation, a plea, etc. Possibly there is a ragbag of such structures, all having to be individually learned, and all evolved roughly in a way that keeps them distinct in normal use. This might explain why feral children (scanty evidence suggests) never make up their language deficit, how kids learn—without the assumption they have the entire structure of every possible language throughout all time in their heads, and what professional students of literature do—expand their repertoire of tricks: as Bertie wrote, to a 15 year old, 'The best way to learn to write is to read and to write considerably.' It also explains how elaborate literary material may, on examination, prove disappointingly banal. It also explains why, when you try to learn a foreign language, you start with the simple stuff.

Taking an example, guaranteed a non-sentence by Uri Strauss, let's look at '*came dawn the.' Strauss says correctly—at present—that this is meaningless. But it wouldn't take much to force such a construction into circulation; it's easy to imagine a rap song repeating this over and over, though the song would probably need some specially meaningful dawn. As in 'Came dawn THE'. No doubt in due course it could diversify into "Came dawn A". I can't see why such a thing couldn't be generated; there seem to be plenty of similar examples, like 'keep your cool' or 'enjoy' or 'I'm studying film'. (These are rather tame examples; bigger changes take longer & I don't want to insert Chaucerian or Anglo-Saxon stuff). I've just opened a book, in what must be regarded as English, with a heading Who is Israel?

On the problems of structure, may I suggest an analogy from mathematics?:- when analysts were trying to define 'continuity', no sooner would someone produce a definition, then someone else would design an oddly-shaped curve which was an exception - culminating in the snowflake curve, which was supposed at its limit to change slope at every single point.

Looking only at the finished appearance of language therefore is superficial, or what I've called an artefact. For people who like concrete metaphors, I extend my interior design comparison: these linguists are like people eagerly collecting interior design examples from their own neighborhood. (Note that Uri Strauss thinks every aspect of interior design is conscious and deliberate). But they're completely baffled when someone invents bean bags, or open planning, or sunken beds, or light pipes. Still less do they understand the force of tradition, or the effects of legal coercion, or of subsidies on some materials, or scarcity of other materials, on the interiors they look at. As Greg Marks pointed out, the way to try to understand these things is to grasp the laws—natural and man-made—determining what's happening and how much scope there is for variation, not just try to classify the outcomes. If you look at the results only, you have the cart before the horse. (But, faced with practitioners such as 'Shakespeare', I'm not sure I'd give much for their chances of success).

Let me give an analogy: imagine someone set out to study interior design. They'd soon find that carpets tend to be on the floor, that furniture doesn't overlap, that fragile stuff is out of the way etc. Language might be visualised like this, a jumble of things which get ideas across, most of the time, and made of components of different types. But the idea there's an underlying structure in furniture arrangement might be simply untrue. It might be an 'artefact' in the scientific sense, i.e. something that appears to be there as a result of processing the thing being looked at.

Looking at language, there are so many exceptions and oddities that any theory of structure seems vulnerable, in the sense anyone can just break the rules to produce a new effect, as in 'came the dawn' say, or 'enjoy!' However you try to pin down a structure or (as it were) sliding puzzle of interlocking bits of language, someone can do something a bit different if they want which doesn't fit.

Although this is a different, pragmatic, line of argument, I've read that Chomskyan structures don't actually work in attempting e.g. computer translation; and I think this must be true, since machine translation seems not to be very successful—I don't see routine interconversion between foreign language Internet sites. There are some translation sites: admittedly I've only tried the free versions, but I tried, with Babelfish "The spirit is willing but the flesh is weak' into Spanish and back, and received, just as a newspaper had alleged in about 1965, 'the alcohol is arranged but the meat is weak'.

Nor other applications: for example, a scientifically-designed new Esperanto would presumably establish as much commonalty as possible between 'important' languages and then minimise the evolutionary changes needed for each; I can't see anything remotely like this. Similarly I can't see the slightest progress in finding the most efficient ways to use language in teaching.

Can I also warn about 'neurolinguistics' and such things? My biological friend Harold Hillman has tipped me off about the extent to which the brain isn't understood; synapses for example cannot be the way they appear in textbooks. The confident statements about the location of certain facilities ought to be viewed with some scepticism. E.g. PET scans look for oxygen combination; unfortunately because of the electrical stuff in the brain this is like looking at a battery—the actual work may be going on somewhere else. The information from people with damaged brains must be viewed with some doubt too; how can researchers know for sure which parts are damaged? There are other examples; however I'd just suggest that sceptically-minded people be more aware than they are of pitfalls in science. I think probably Chomsky is somewhat naive about science, not having generalised his doubts on social scientists into this field. For example, an article he wrote on AIDS (or 'AIDS', in quotation marks) a year or so ago struck me as uninformative and contentless, and unaware of the frauds in this field.

All this raises the odd question as to why Noam should be highly praised by journalists and others who clearly are far too dim-witted to have much grasp of what he's saying. It's easy to see why they condemn his politics; but when even the most careful student, soaked for years in Chomsky's thought, has great difficulty in discovering his meaning, the reader with the intelligence of the average supporter of the U.S. Empire must find his writings for all practical purposes unintelligible. Perhaps these people just need an American intellectual hero. But perhaps, in view of the achievements of the US in philosophy, the arts, and science perhaps this adulation is understandable too.

My guess is that in a century or so linguistics will appear as a divertissement, a form of scholastic distractionism, '.. this kind of degenerate learning' (Bacon). I have an uneasy feeling that there's a parallel with Russell (and Whitehead).

It occurs to me that Noam might be approaching retirement age (I don't know MIT's policy on this); perhaps he'll make a retirement speech, renouncing his work and denouncing all his followers as derivative imitators with no spark of brain-power. This would be a fine anarchic thing to do, and spiffing fun for everyone (except other linguists).

Back to Noam Chomsky Subsite

Back to Main Index of Rae West's Site

HTML etc Rae West. First uploaded 99-02-12. A few things added 12-09-25. Swahili note 2013-10-09.