Language & Communicafion, Vol. 8, No. I, pp. 69-84, 1988 Printed in Great Britain.

0271.5309/88 $3.00 + .OO Pergamon Press plc

IDEAL LINGUISTICS

NIGEL LOVE

R. P. Botha, The Generative Garden Game: Challenging Chomsky at Conceptual Combat (= Stellenbosch Papers in Linguistics 16), University of Stellenbosch, 1987

The Generative Garden Game consists of the exposition seriatim of 96 conceptual distinctions relevant to a proper grasp of the theoretical underpinnings of Chomsky’s (not Chomskyan) linguistics, interspersed with italicised passages of relentlessly alliterative word- play in which enigmatic fun is poked at Chomsky’s academic opponents, Chomsky and, occasionally, the author himself. A wide variety of attempts to ‘challenge Chomsky at conceptual combat’ are analysed and, for the most part, treated as having foundered because of the challenger’s ignorance or misunderstanding of one or another ‘conceptual fork’- although occasionally a battle is represented as inconclusive, or as having been only apparently ‘won’ by Chomsky, by such tactics as tacitly changing the subject or disappearing behind a rhetorical smokescreen. In this manner Botha skilfnlly guides his readers through a large number of theoretical debates within and about generativism (although the book would have been even more useful in this role had an index been provided), bearing particularly on Chomsky’s ideas about language acquisition and his work on syntax (there is nothing on generative phonology, for instance). The Master’s (sic) theoretical positions are presented largely via extensive quotation from mostly recent writings (especially Chomsky, 1986), so the book also provides a succinct account of relatively new developments such as the supersession of syntactic ‘rules’ by a system of ‘principles and parameters’.

The underlying story that Botha tells in this somewhat fragmented and episodic way is a familiar one. In essence it runs as follows. Chomsky’s linguistics is one specific version of the psycho-biological interpretation of generativism. Broadly speaking, to interpret generativist linguistics psycho-biologically is to suppose that the writing of generative grammars, and the study of generative grammatical theory, are relevant to an attempt to distinguish those aspects of the general phenomenon called ‘language’ which fall within the province of the natural sciences from those which do not. Equipping us with the capacity for language is perhaps biology’s chief contribution to human culture, and it would therefore be interesting to understand just how, in what respects, and to what extent language is determined by our biological nature.

More narrowly, Chomsky is held to be particularly concerned with psycho-biological generativism as an approach to the question of how children acquire their native language. The starting-point for this inquiry is the proposition that no matter what linguistic community they are brought up in, all physically and psychologically normal children develop unconscious knowledge of, and hence the ability to utter and comprehend, the

Correspondence relating to this paper should be addressed to Dr N. L. Love, Department of Linguistics, University of Cape Town, 7700 Rondebosch, South Africa.

69

70 NIGEL LOVE

infinite set of spoken sentences which constitutes the language in question. This ability is achieved despite the fact that the experiential basis for doing so (the ‘stimulus’) is in various respects deficient. Deficiencies of the stimulus fall into two main categories: ‘degeneracy’ and ‘poverty’. According to Botha (p. 19), Chomsky’s account of native- language acquisition ‘is based on (his perception of) the poverty of the stimulus and not on (his assessment of) the degeneracy of the stimulus’. Chomsky’s discounting of degeneracy will call for discussion later: let us for the moment focus on poverty.

The stimulus is impoverished in that although a child comes eventually to be able to deal with an infinite range of sentences of his language, he directly encounters only a finite sample of them. Yet as a mature speaker he has no more difficulty in producing and understanding sentences he has never heard before than he has in dealing with those that have actually figured in his past linguistic experience. Moreover, it is not just that he deals effortlessly with new sentences as such. He will also somehow come to know that well- formed sentences of previously unencountered grammatical types are legitimate and acceptable. Conversely, children are not systematically informed as to the illegitimacy of certain logically possible but non-occurrent sentence-types, but nonetheless never produce sentences of such types. In recent generativist writings illustrations of this latter phenomenon tend to turn on the grammatical vagaries of either the English phrase each other or else of the formation of wh-questions. Berwick and Weinberg, for instance, offer (1984, pp. 18-19) the following example. In a sentence such as Ronald Reagan finally issued a statement without contradicting it, the final pronoun it, they observe, ‘co-refers with’ the object of the verb, a statement. One can formulate a similar sentence where the final pronoun stands for the subject: Ronald Reagan finally issued a statement without contradicting himself. Both subject and object of the first part of these sentences may be questioned, as in who finally issued a statement? and which statement did Ronald Reagan finally issue? But if similar questions are formed corresponding to the original full sentences, a divergence arises between the treatment of subject and object: although which statement did Ronald Reagan finally issue without contradicting? (where it is omitted at the end) is grammatical, *who finally issued a statement without contradicting? (where himself is omitted at the end) is not. The point is that there is no obvious reason for this lacuna in the inventory of possible English sentence-types; and its impossibility is not explicitly taught to apprentice English- speakers. But all mature English-speakers unconsciously know that sentences of that kind are ungrammatical. Likewise, speakers know, without ever having had it overtly pointed out to them, that a given sentence of their language is ambiguous, or that certain sets of sentences are paraphrases of one another. In short, speakers come to have unconscious knowledge of the structure and organisation of their language, for at least some parts of which there is no direct evidence available in the data they are exposed to as children.

The poverty of the stimulus, in Chomsky’s view, reveals the kind of phenomenon first- language acquisition is. How can the mature speaker eventually have linguistic knowledge for which his direct experience as a child provides no warrant? Moreover, despite variation among individuals as to background and intelligence, their mature linguistic capacity emerges in a fairly uniform fashion, in just a few years, without much apparent effort, conscious thought, or difficulty; and during its development only a narrow range of the logically possible ‘errors’ are ever made. Children do not test random hypotheses, gradually discarding those leading to ‘incorrect’ results. On the contrary, the illformed sentences produced by young children seem to be few in type and rather uniform from one child to another. Despite the poverty of the stimulus, normal children attain a rich system of

IDEAL LINGUISTICS 71

linguistic knowledge by five or six years of age, and a mature system by puberty. Such a feat can only be possible on the basis that certain things about the language they are learning are known apriori, in the sense that they are somehow available to them irrespective of their idiosyncratically deficient linguistic experience. Furthermore, this apriori knowledge cannot be knowledge pertaining to the structure of particular languages, since children master whatever language they happen to be exposed to in infancy.

What is proposed, therefore, is that certain abstract principles pertaining to the grammatical structure of all languages (that is, of human language in general), are genetically encoded in the brain. Children come into the world already primed to acquire a language of a certain logically rather arbitrarily delimited type. It will be a particular species of the genus ‘human language’, whose general organisational principles are universal, determined by our common genetic inheritance. The unfailing ease and accuracy with which children acquire their language is to be explained by supposing that the inadequate and fragmentary evidence as to its structure available to them empirically is supplemented by a system of innately known universal organisational principles.

Chomsky insists that the gap between the empirically available evidence and the linguistic knowledge of the mature speaker cannot be bridged by (appeal to) any general mechanisms or processes of learning. This view he bases on considerations such as the following:

There are, in fact, striking and obvious differences between language learning and the learning (or discovery) of physics. In the first case, a rich and complex system of rules is attained in a uniform way, rapidly, effortlessly, on the basis of limited and rather degenerate evidence.’ In the second case, we are forced to proceed on the basis of consciously articulated principles subjected to careful verification with the intervention of individual insight and often genius. It is clear enough that the cognitive domains in question are quite different. Humans are designed to learn language, which is nothing other than what their minds construct when placed in appropriate conditions: they are not designed in anything like the same way to learn physics. Gross observations suffice to suggest that very different principles of ‘learning’ are involved (Chomsky, 1983, p. 320).

Indeed, it is questionable whether ‘learning’ is a useful notion at all where language acquisition is concerned: we are dealing, rather, with the growth and eventual maturation, under environmental stimulus, of a ‘mental organ’. In this view, we no more ‘learn’ our native language than ovaries ‘learn’ to shed eggs. In fact, ‘it is possible that the notion “learning” may go the way of the rising and setting of the sun’ (Chomsky, 1980, p. 139).

So human beings must be equipped with a distinct ‘language faculty’, providing them from the outset with what is sometimes referred to as a ‘general theory of grammar’. This represents the genetic equipment that facilitates language acquisition and delimits the linguistic knowledge that will eventually be attained, by the interaction of the language faculty with the evidence available to the child as to the forms and structures of the particular language he is acquiring.

As far as the syntactic part of a language is concerned, Chomsky nowadays characterises the initial state of the language faculty as a system of ‘fundamental principles’, many of which have ‘open parameters’ associated with them. ‘Subjacency’, for instance, is the principle that, roughly speaking, a phrase cannot be moved ‘too far’ within a sentence, where ‘too far’ means ‘beyond the limits of two bounding categories’. As a fundamental principle, subjacency is in Chomsky’s view genetically encoded. However, a principle such as subjacency may have one or more open parameters whose values are fixed by the child’s linguistic experience. For instance, English and Italian both exhibit subjacency, but they differ in regard to the way in which the choice of bounding category is fixed. The important point here is how this distinction between fundamental principles and open parameters

72 NIGEL LOVE

relates to the general question of language acquisition. It provides the basis for an account of how it is possible for the initial state of the language faculty to satisfy two apparently conflicting conditions. On the one hand, the initial state must be sufficently accommodating for it to be possible in principle for the child to acquire any one of the wide variety of human languages. On the other hand, it must be sufficiently restrictive for it to be possible for the child to acquire, on the basis of limited evidence, the specific language of the speech community to which he belongs. The genetically determined fundamental principles provide for the plasticity to acquire any language, while the open parameters make it possible to acquire some specific language, by a process of ‘fixing the parameters’ on the basis of linguistic experience.

There are various reasons why it is questionable to what extent these ideas add up to a theory of language acquisition. These have to do with various idealisations on which the theorising is based. But before discussing these idealisations it is necessary to draw attention to Chomsky’s concept of a language.

Chomsky is not-at least, not in theory-concerned with languages as institutionalised cultural products: that is, with entities such as those we call ‘English’ or ‘French’ or ‘German’, as discussed, inventoried and analysed by philologists, lexicographers, grammarians etc. In the latest terminology such entities are known as ‘E[xternalised]- languages’, and contrasted with ‘I[nternalised]-languages’. The essence of the distinction is that whereas, as Botha puts it (p. 76) ‘an E-language is an object that exists outside the mind of a speaker-listener’, an I-language ‘is some element of the mind of the person who knows the language, acquired by the learner, and used by the speaker-hearer’ (Chomsky, 1986, p. 22). That is, an I-language is a native speaker’s linguistic knowledge, or internalised grammar.*

So it is not native-speaker acquisition of ‘English’, ‘French’ etc. as the layman knows them that Chomsky is concerned with. Languages, in this sense, are no more than by- products of the use of an I-language; and the use of an I-language in any case involves the interaction of the language faculty with many other departments of the mind.

But there is a strange twist here. It is not just that Chomsky’s theory is uninterested in accounting for the acquisition of what most people are likely to think of as ‘languages’. His theory may not be directly concerned with such languages (E-languages), but many of them are at least allowed for by the theory, in the sense that they are seen as a sort of collective precipitation thrown up the use of the I-languages of the speaker-listeners in a particular speech-community. But there are some E-languages which do not appear to have even this status. That is, there are languages which are apparently not be envisaged as the product of a large number of different but similar ‘underlying’ I-languages, because they are ‘impure’; and Chomsky’s theory is exclusively concerned with what he calls ‘pure’ languages (Chomsky, 1986, p. 17). So speakers of impure E-languages are, presumably, either held not to have I-languages, or else their I-languages and the acquisition of them lie outside the domain of Chomsky’s investigation. And the twist is that the boundary between pure and impure E-languages appears to correspond roughly to the boundary between languages which do, and languages which for one or another reason do not, fall within the purview of the traditional grammarian, philologist, etc.

Chomsky’s example of an impure language is the mixture of Russian and French spoken by some members of the nineteenth-century Russian aristocracy. It is not entirely clear what general idea this example is intended to illustrate,3 but it is clear enough that such

IDEAL LINGUISTICS 73

a mixture would not be the object of a traditional grammarian’s attention. Roughly speaking, the traditional grammarian is concerned, first, with the literary languages of classical antiquity; secondly, with the standardised ‘official’ languages of the nation-states of post-Renaissance Europe; and thirdly, with any other languages similar to these in being culturally prestigious and having given rise to a more or less stable written form (preferably enshrined in literary works of acknowledged merit), so that ‘speaking the language correctly’ may be represented, for all practical (e.g. pedagogical) purposes, as implementing that form (or something very close to it) in the oral-aural dimension. If this is what a pure language is then clearly Russian and French, for instance, would qualify as pure languages, while a mixture of Russian and French would not. Nor, presumably, would most pidgins and creoles, which are, at least in the traditional view, highly unstable ‘jargons’ arising in anomalous circumstances and for the most part lacking cultural authority.

Of more equivocal status are dying languages, such as Welsh. In its time, Welsh has of course been an object of traditional grammatical inquiry. But it is questionable whether the contemporary spoken language can be viewed, in the usual way, as an oral-aural manifestation of that object. The first problem with Welsh, in this context, is to decide who counts as a ‘speaker’. Monolingual Welsh-speakers are thin on the ground these days, and tend to be either under four years old or over 80. Most Welsh-speakers are bilingual in English, and their Welsh is to varying degrees contaminated by English. Welsh contains an enormous number of lexical borrowings from English. Many of these are long standing, and established as part of any Welsh-speaker’s vocabulary. But the fact that the native Welsh vocabulary is radically deficient in the words required for talking about virtually any aspect of life specifically characteristic of the twentieth century gives rise to a penumbral area where it is difficult to decide whether we are dealing with the use of a non-native but nonetheless Welsh word or construction, or a mere lapse into English. Welshmen, when speaking Welsh, use words like wage-freeze, hamburger, fridge and in writing they may spell them according to Welsh conventions (e.g. ffrb], but it is unclear whether we should count them as Welsh words.

There are also more complex kinds of English influence, affecting grammar as well as vocabulary. Here are some, not at all far-fetched,4 examples of contemporary ‘Welsh’ expressions: mae’r rhain wedi catchio on ‘these have caught on’, mae hi wedi committio suicide ‘she has committed suicide’, mi na‘th o chickenio allan ‘he chickened out’, mae’r ducks yn swimo ar y water ‘the ducks are swimming on the water’ (-io is an infinitive verb ending; allan means ‘out’; all the obviously English words and phrases here, such as catch on, swim, water etc. have native Welsh equivalents). Such ad hoc improvisations are characteristic of ‘semi-speakers’ of a dying language: those for whom interference from language B has reached the point where their grasp of language A, even if it is their mother tongue, is no better than that of a not very efficient second-language learner. Not all Welsh speakers would say such things, and not all of those who say them sometimes would always do so: some also control what Chomsky might be inclined to see as a ‘purer’ Welsh for use on more formal occasions. But there are certain lexical subsystems where substitution of English expressions is quite general. Many Welsh speakers invariably use the English words for the months of the year (but not, for some reason, the days of the week). Common, too, is the use of left and right, and English numerals: mae hi’n byw yma ers twenty-two years ‘she has lived here for twenty-two years’, mae’r car ‘ma yn gneud thirty miles i’r galwyn ‘this car does thirty miles to the gallon’. This leads to the time being told in English:

74 NIGEL LOVE

mae’r bws yn mynd o’rpentre twenty-five to six ‘the bus goes from the village at twenty- five to six’.

Welsh-speakers (whoever we decide they are exactly) can be presumed to differ in respect

of their attitude to such ‘sentences’ of their language. At one end of the spectrum we have those who would be disinclined to count any of the expressions quoted here as Welsh at all. At the other end, there are those for whom an unstable mixture of English and Welsh phrases embedded in a basically anglicised syntactic structure is all the Welsh they know. In between there are those who have, to varying degrees, the ability to shift among more or less decomposed forms of the language. And almost all speakers, whatever the state of their Welsh, are fluent in English. In so far as this complex state of affairs suffices to render Welsh ‘impure’, (many) Welsh-speakers, we must assume, are simply not persons to whom Chomsky’s theory of language acquisition is intended to apply. And this is very peculiar. For no general reason is offered for the exclusion of speakers of impure languages from the domain of the theory.

A different idealisation of the acquisition process arises from a distinction between what are known as the ‘logical’ and the ‘psychological’ problems of language acquisition. The logical problem is essentially that of explaining in general terms how it can in principle be possible for children, on the basis of limited and partial exposure to their language, to acquire the complex and rich system that constitutes their mature linguistic knowledge. The ‘psychological’ problem, in contrast, is the problem of ‘real-time’ acquisition: what does the actual process of acquiring a language consist in? Chomsky does not seek an answer to this latter question; and concentration on the ‘logical’ problem to the exclusion of the ‘psychological’ problem implies that certain kinds of evidence which might prima facie be thought to have a bearing on the question of acquisition are automatically discounted as irrelevant. For instance, the data constituted by the various kinds of distorted, simplified or otherwise ungrammatical utterances produced by a child at various stages in the process of acquiring his language are presumably available to children themselves, and are, no doubt, ‘input’ to a subsequent stage of the ‘real-time’ acquisition process. But since the real-time acquisition process is not what is at issue, such utterances are excluded from consideration. In other words, Chomsky assumes for theoretical purposes that acquisition is instantaneous: he is concerned with the general question of what in principle is required if the gap between available linguistic evidence and mature linguistic knowledge is to be bridged, and not with the question of how in practice the bridge is actually built.

This is a curious theoretical compromise. We can, however, see how it is forced on Chomsky. On the one hand, real-world speakers and speech-events must come into the story somewhere, otherwise the ‘logical problem’, as posed, would vanish. It is a gap that the initial state of the language faculty fills, not a void. But on the other hand, real-world speakers and real-time speech-events must not be given a role so large that they might conceivably abolish the gap altogether. Acquisition must not be assimilable to learning. This is why the whole question of ‘degeneracy’ is such a hot potato. The degeneracy of the stimulus is held to lie in the fact that the speech the child hears does not consist uniformly of complete, well-formed sentences. Unscripted speech is notoriously full of false starts, hesitations, grammatically unanalysable fragmentary utterances and other ‘performance errors’ which on the face of it ought to be an obstacle to acquiring unconscious knowledge of (the principles governing) an infinite set of well-formed sentences, and hence grist to the mill of anyone who seeks to establish a gap between available linguistic evidence and

IDEAL LINGUISTICS 75

mature linguistic knowledge. The trouble is, as has frequently been pointed out, that generativists have exaggerated the extent to which the evidence available to the child is degenerate, in that adults to some extent make a deliberate effort to teach their language to children, and by way of doing so are careful to address to them well-formed, simplified and carefully enunciated utterances. Chomsky’s response to this, as reported by Botha (pp. 19-26), is to deny that such ‘motherese’ actually forms part of the evidential stimulus to acquisition. Presumably the implied argument is that the language faculty primes the child to ignore motherese and to attend exclusively to the non-simplified utterances that he encounters. If this is indeed the argument, not much contemplation of it is required to see that, from Chomsky’s point of view, wisdom might lie in refusing to entangle himself in the degeneracy issue altogether. Hence the compromise whereby the child must be envisaged as acquiring his language in some version of a ‘real world’, but a real world which offers to him as evidence for the language he is acquiring an impoverished (but not degenerate) sample of its sentences.

Some may find it is difficult to see why any eventual solution to the ‘logical problem’, as thus conceived, should be treated as having any bearing on language acquisition in the real ‘real world’. Not only is the compromise between completely idealised and genuinely real-world acquisition theoretically unsatisfactory, but the questions which the theory is designed to answer in any case seem largely spurious.

For instance, the query ‘how do children come to master an infinitely large set of spoken sentences?’ is vulnerable to two observations. First, it is far from clear what a ‘spoken sentence’ is, and hence even less clear that a spoken language is properly envisaged as an infinite set of them.5 Secondly, even if these conceptual problems were resolved, it would remain to demonstrate that native speakers are actually in possession of such mastery. Thus the proposition which forms the starting-point for the whole inquiry is far from securely established.

On the other hand, the claim that there is a gap between our linguistic knowledge and what we have an evidential basis for knowing is prima facie more promising, in that here we are undoubtedly faced with facts requiring explanation. But the interpretation imposed on those facts is curious. It is a fact, for instance, that most English-speakers (that is, most speakers of I-languages grouped under the heading ‘English’) are unlikely to produce utterances like ‘who finally issued a statement without contradicting?‘, and a further fact that there is nothing in what English-speakers do say from which an apprentice English- speaker could readily infer the relevant grammatical proscription. But what the relevant grammatical proscription is, in any particular case, is not evident from mere inspection of the linguistic expression (or non-expression) held to exemplify it. As far as this particular example is concerned, we are invited to see the tip of an iceberg of language-faculty- embedded principles governing the formation of w/z-questions. But there is at least as much reason to see no more than one or more of a number of idiosyncratic facts about the verb contradict. One is that contradict, unlike its Germanic counterpart gainsay, is peculiar in allowing a reflexive personal object at all (compare Ronald Reagan contradicted himself with *Ronald Reagan gainsaid himselfi. Or the relevant fact could be conceived as being that contradict, unlike demur, dissent and other words in the same general semantic field, is currently transitive. In this view no more would be required to make who finally issued a statement without contradicting? grammatical than the kind of change that has made which policies were the demonstrators protesting? grammatical in American, but not British,

16 NIGEL LOVE

English. Or, given that contradict is transitive, the peculiarity might be said to be that it has so far failed to develop the kind of formally intransitive use with object implied or ‘understood’ which characterises many basically transitive verbs (he drinks, for instance, is as grammatical as he drinks whisky). There is nothing ungrammatical about the sentence formed by omitting himself from who finally left the bathroom without washing himself?; and it is hard to believe that there is some deep reason of principle why contradict should not develop a similar construction. It could start happening tomorrow.

There are two points here. First, it is not that the non-sentence in question cannot be seen as exemplifying the grammatical point it is held to exemplify. But whether it does so depends on the point of view adopted. And the point of view from which the non- occurrence of *who finally issued a statement without contradicting? demonstrates a restriction on the possible types of well-formed wh-question is that of the linguist concerned, precisely, to establish the general grammatical fact in question. But its non- occurrence cannot necessarily be expected to have the same significance for the speaker- listener. The fact, if it is a fact, that he ‘knows’ this sentence to be ungrammatical is not in itself evidence for his ‘knowledge’ of the underlying explanatory principle. Botha (pp. 22-23) is scornful of those who criticise generativists for assuming that the language-acquirer is a ‘little linguist’. But it seems that some such assumption is inherent in the very attempt to demonstrate the existence of grammatical principles given by the initial state of the language faculty by citing what the native speaker ‘cannot say’. For it is only if he adopts the linguist’s point of view that the reason he cannot say it has to be seen as being the one offered.

The second point is that what requires explanation here is misinterpreted by the generativist. Granted that there is a stability of usage which allows it to be correctly said, for example, that in current English *who finally issued a statement without contradicting? is ungrammatical, the interesting question is how such stabilities can come about, given the circumstances in which language is actually used. What are the processes by which, out of the incessant flux of utterance, usages come to be, or at any rate to seem, at least temporarily fixed and codifiable? A satisfactory answer to this question, clarifying as it would at least one dimension of the fundamental distinction between ‘language’ and ‘a language’, would clearly be relevant to an understanding of language acquisition. But it is not a question that the generativist considers. His question is, granted that a language is a fixed system of grammatical sentences for which there is a paucity of evidence in the utterances to which the learner is exposed, how is the system nonetheless acquired? His answer to this question leads him to seek to explain, not how and why there should be stabilities at all, but the much less significant fact that the stabilities currently happen to be what they are. In short, what requires explanation is misrepresented from the outset by a priori theoretical fiat.

For these various reasons it is hard to see that there is very much to Chomsky’s claim to have a theory of language acquisition. In fact, it is clear enough that the preoccupation with acquisition is superimposed on the real business of generativism, which is the writing of generative grammars.6 The mature linguistic knowledge eventually attained by a speaker is represented or modelled by the linguist as a ‘grammar’ which generates all and only the sentences of the speaker’s language, specifying form and meaning for any particular sentence. Since speakers can use and understand indefinitely many sentences, a generative grammar must be a finite system that can characterise an infinite number of sentences.

IDEAL LINGUISTICS ?I

Such a grammar contains explicit and formal statements (‘rules’ under the earlier dispensation) whereby sound and meaning are correlated. In practice, generative grammarians have usually given priority to syntax: the syntactic rules of the grammar assign to the sentences of the language correct descriptions of their syntactic structure, and associate the syntactic structure of sentences with phonological and semantic interpretations (that is, say how they are pronounced and what they mean). The grammar also incorporates a ‘lexicon’, which is a repository for all the information about the basic elements of linguistic structure (‘formatives’) which cannot be derived by phonological, syntactic or semantic rules. The correct grammar will generate the infinite set of well-formed sentences of somebody’s language, and none of the deviant ones. Writing such grammars is, at least in principle, the substantive task of generative grammarians. In practice, however, the task seems to have been postponed sine die. Why should this be?

The conceptual background sketched in earlier is mainly set out by Botha in what is by far the longest of his five chapters, under the heading ‘the maze of mentalism’. The questions which now natur~ly arise: namely, ‘how does one set about specifying someone’s internalised grammar?’ ‘how does one determine the nature and content of the linguistic knowledge it is supposed to characterise?-more generally, ‘what does the activity or inquiry to which this is the background actually consist in?‘-are dealt with much more perfunctorily. In fact, all that Botha has to say here is compressed into nine of the 235 pages of his main text. Having started by reminding us that ‘a generative grammar of a particular language is a theory that has to give an answer to the question “What does knowIedge of this Ianguage consist of”, he proceeds, as before, by drawing attention to a series of conceptual distinctions. For instance, we must be careful to note the difference between a grammar as ‘a property of the (ideal) speaker-listener’s mind/brain’ (p. 155) and a grammar as a linguist’s theory of that property. A generative grammar must, further, be distinguished from a ‘traditional/pedagogical’ grammar (pp. 156-157). Moreover, we must beware of supposing that, because the term ‘generate’ is used, a generative grammar explains how sentences are actually produced and interpreted (pp. 157-158). Then there is the distinction between ‘weak’ and ‘strong’ generative capacity (pp. 157-161). Finally, we are invited to attend to Chomsky’s recent ‘conceptual shift’, and mark the difference between a grammar as a set of rules generating sentences and a grammar as a system of principles with initially open parameters that are eventually fixed (pp. 161-162). Botha concludes this section by observing that:

grammars are no longer formal systems of explicitly formulated rules for the generation of sentences. Grammars of the post-shift sort must contain devices specifying how the values of open parameters are fixed in the language. In doing this, such devices, in terms of a recent statement by Chomsky (1986, p. 93) are to perform a sort of ‘licensing’, the assumption being that ‘every element that appears in well- formed structure must be licensed in a small number of available ways’. It is not clear in what format (m~irn~ly explicit, of course), such ‘p~ame&er-fixing’ or ‘licensing’ devices are to be represented (p. 162).

From Botha’s point of view the significance of this is that here we find ourselves in an area of the ‘generative garden’ in which ‘purposeful play’ (that is, a convincing challenge in the second-order game of arguing about Chomsky’s principles) is not yet possible. But many readers wiI1 find this Iess important than the implication that it is not yet possible to play the first-order game of actually writing a grammar to Chomsky’s principles either. In so far as we take the ‘shift’ seriously, the entire discussion of conceptual distinctions turns out to be purely programmatic prolegomenon (the style is catching) to an enterprise which nobody as yet quite knows how to embark on.

Here we come upon a fundamental difficulty with psycho-biological generativism, whether

78 NIGEL LOVE

it be Chornsky’s linguistics, Chomskyan linguistics, or, as Botha puts it (pp. 7-S), ‘radical Chomsky-like linguistics’ that we are concerned with. In fact, the doctrinal divergences thus alluded to are themselves symptomatic of the difficulty. Rather than write generative grammars, generativists have from the outset concentrated on arguing about the preliminaries to writing them: hence the rapid development of various kinds of theoretical volte-face, known either as heresies or conceptual shifts, depending on whether you happen to approve of them. What this preference for second-order theorising conceals, or fails to conceal, is the difficulty of grasping how in practice the first-order activity is actually to be pursued.

It is difficult for reasons which cut across the differences between competing sub-schools of theorists. These reasons have to do with yet another-fundamental-idealisation on which the theorising is based. This is the well-known statement that ‘linguistic theory is

concerned primarily with an ideal speaker-listener, in a completely homogeneous speech community, who knows its language perfectly, and is unaffected by such grammatically irrelevant conditions as memory limitations, distractions, shifts of attention and interest, and errors (random or characteristic) in applying his knowledge of the language in actual performance’ (Chomsky, 1965, p. 3).

Leaving aside the problem of seeing precisely what acceptance of this idealisation commits us to (what exactly do we understand, for instance, by a ‘totally homogenous’ speech community?), we can see that there are, in general, two ways in which it might be taken. The difference between them is essentially a matter of whether we take as ‘given’ (a) languages, or (b) speaker-listeners.

Under interpretation (a), languages are E-languages: entities such as ‘English’, ‘French’ and ‘German’, familiar to the layman and overtly teachable to him by a pedagogical tradition which treats them, essentially, as fixed networks of form-meaning correspondences enshrined in a dictionary and combinable into larger units according to the rules laid out in a grammar-book. If, for purposes of generativist analysis, a language is considered as the content or output of a traditional grammar plus a dictionary, a generative grammar of such a language represents the linguistic knowledge of a speaker-listener who is ‘ideal’ in that his linguistic knowledge, consisting as it does of all and only the content of grammar- cum-dictionary, both exceeds and falls short of the knowledge of any real speaker-listener. This is perhaps the obvious interpretation of the ‘ideal speaker-listener’, and is manifestly the one that underlies much published generativist work. Studies such as Lees (1960) Rosenbaum (1967), Kayne (1975), Awbery (1976), Radford (1977) and many others, combine discussion of theoretical principles with presentation (often merely by way of illustration) of partial or fragmentary generative grammars of E-languages. In effect such grammar-fragments amount to a reformulation in generativist terms of the traditional descriptions from which their data are drawn. In the main, no reference to real speaker- listeners is contemplated, or indeed required [which is why there is nothing absurd about attempting to write a generative grammer of a dead language, as does Lakoff (1968)]: for these are grammars tacitly known only by an ideal speaker-listener equipped with perfect knowledge of the E-language as codified in the grammar-cum-dictionary which provides

the data-base.

Now interpretation (a) is, of course, quite incorrect (although precisely when it became incorrect is a fascinating problem in the historiography of generativism). It is not that real speaker-listeners fail to measure up to an ideal linguistic knowledge embodied in traditional

IDEAL LINGUISTICS 19

descriptions. It is, rather, that traditional descriptions fail to measure up to (are inaccurate descriptions of) the linguistic knowledge possessed by any real speaker-listener. That is everyone is an ideal speaker-listener, but of his own idiosyncratic idiolect (I-language). This is interpretation (b).

Now it must be admitted that although interpretation (b) offers a more accurate account of the import of Chomsky’s idealisation (at any rate, as currently understood), it has scarcely been as fruitful as a basis for writing even fragments of generative grammars. Published generativist accounts of idiolects (as distinct from languages or parts of languages) are rare. Why?

The fundamental underlying assumption in the network of doctrines that constitutes psycho-biological generativism is that what makes language-use possible is unconscious knowledge of a determinate set of rules generating an abstract system of correspondences between forms and meanings. For this assumption to be tenable, it must be possible to decide where the line shall be drawn between those features of language-use which depend on linguistic knowledge and those which involve non-linguistic knowledge of one kind or another. Clearly, linguistic knowledge itself cannot be the sole contributory factor to the success or failure of a communicational exchange. Other things must be involved as well. For instance, knowledge of the meaning of a linguistic expression must be distinguished from knowledge of facts about what that expression refers to.

In this connection it should be observed that it is by no means the case that being a native speaker of a language guarantees the ability to understand utterances in it. Consider the following extract from a newspaper column:

When dummy went down, I realised that 6S would make if suits broke well and the diamond honours were favourably placed. I therefore decided to assume that twelve tricks were not readily available, hoping thereby to beat all the pairs going down in slam contracts and to make more tricks than those who had lingered in game. I won the opening heart lead in dummy, drew trumps in two rounds and cashed the ace of clubs. I then ruffed a heart in the closed hand, paving the way for an elimination, and cashed the king of clubs, discarding a diamond from dummy. When the ten of clubs appeared from East, I did not need to look any further for twelve tricks: I ran the jack of clubs, throwing another diamond from dummy, and subsequently discarded a third diamond on the established nine of clubs. I was then able to ruff two diamonds in dummy, thereby collecting twelve tricks by way of five spade tricks and two ruffs, one heart, one diamond and three clubs (Markus, 1982).

One imagines that many speakers of English will fail to make much of this. And yet the linguist would be hard pressed to argue that this is because it is not English. It is clearly English, and yet speakers of English may have difficulty in understanding it. What is the source of the difficulty? It is not, for the most part, the obscurity of the vocabulary. Apart from ruff, dummy, slam contract and a few others, not many of the words in the passage are in themselves obscure to anyone with at least a slight acquaintance with the practice of card-playing. And, apart from the occasional oddity (when dummy went down, those who had lingered in game, where common nouns are used without an article, as though they were proper names), the syntax is quite straightforward too. Nor does the problem arise from the fact that the words are used metaphorically or in some other non-literal way: what we have here is a perfectly sober and prosaically literal account of a game of bridge. The conclusion must be that those who find it incomprehensible do so because

of their lack of familiarity, not with the English language, but with bridge.

This conclusion has an important bearing on the distinction between linguistic knowledge and other knowledge that might be brought to bear when using language. Lack of the relevant extra-linguistic knowledge may hinder understanding of the linguistic expressions

80 NIGEL LOVE

used to describe that piece of the world. But the theorist’s position must presumably be that although we understand the langauge of the passage as such (because we are native speakers of the language of which it is a sample), where we fall down is in our grasp of the situation being described. But this seems to be a distinction without a difference.

This point is confirmed if we inspect one writer’s illustration of the distinction (Leech, 1974, pp. 7-8). My uncle always sleeps standing on one toe is said to be factually, rather than linguistically (or contingently, rather than necessarily) absurd. It just happens to be the case that human beings cannot sleep in that posture. But if the world had been ordered differently they might have been able to, and if they were, there would be nothing absurd about such a sentence. Whereas no conceivable reordering of the world could eliminate the absurdity of a sentence like my uncle always sleeps awake, for if my uncle is asleep then, by definition, he is not awake, and if he is awake, then he is not asleep. But there is no very obvious reason why these explanations of the two kinds of absurdity should not be transposed. Instead of saying that it is the nature of standing on one toe that makes talk of sleeping in that position absurd, why should we not locate the absurdity in the conjunction of the meanings of the expressions sleeps and standing on one toe? Conversely, it might be argued that what makes sleeping awake inconceivable is not the incompatibility of the meanings of sleep and awake, but the incompatibility of the states of consciousness referred to by those words. Our understanding of sleep and wakefulness is such that we have no (non-absurd) use for sentences about creatures sleeping awake. But it might have been the case that there were animals which exhibit a state of consciousness that bears some of the characteristics of both. If so, talk of them sleeping awake might make perfectly good sense, The non-existence of such creatures, if it is a fact, is as much a fact about ‘the world’ as is the non-existence of uncles who sleep standing on one toe.

So it seems that there are problems in determining just what part of a speaker’s capacity to use his language in successful acts of communication is to be accounted for the semantic part of his postulated linguistic knowledge. But if this cannot be determined, a grammar which characterises and reflects that linguistic knowledge cannot be written.

If it is difficult to distinguish linguistic knowledge from knowledge of the world, it is no less difficult to distinguish linguistic knowledge from knowledge of language.’ Here is how this issue is broached in a recent generativist textbook:

. it may be that the informant is influenced by prescriptive notions learned at school: thus, some English- speakers asked about a sentence like

(16) who did you meet at the party? would reply that such a sentence is unacceptable because it is ‘bad grammar’, and should be ‘corrected’ to

(17) whom did you meet at the party? But of course this is nonsense: a sentence like (16) is perfectly well-formed, and is characteristic of everyday conversation. In this case of performance factor (prescriptive education) is interfering with the natural competence of the native speaker, with the result that the acceptability judgements which he gives are not an accurate reflection of the well-formedness or otherwise of sentences in his language (Radford, 1981, p. 14).

What is difficult to understand here is the confidence with which the conclusion is reached

that who did you meet at the party., 7 despite the hypothetical informant’s protest, is nonetheless to be acounted grammatical. It is true that there may be many English-speakers to whom a rule proscribing who as an object form has been overtly taught. But there may well be other speakers for whom it might count as part of their tacit grammatical knowledge (if there were such a thing)-perhaps because they were exposed at a crucial stage to speakers to whom it was taught with sufficient thoroughness to make it an invariable feature of their speech. In fact, one may question whether it was ever a practical proposition, in a

IDEAL LINGUISTICS 81

society where discussion of the native language is a fundamental part of the formal educational programme, to contemplate distinguishing systematically between grammatical rules which belong to the speaker’s unconscious knowledge and those which do not-or, more generally, between the speaker’s linguistic knowledge and his knowledge of language. Once again, if this distinction cannot be drawn, a grammar cannot be written.

Any speaker of a dialectally and idiolectally diverse language, such as English, is likely to come into contact with speakers who use forms and constructions different from his own; and he is likely to be in doubt where to draw the line between those linguistic structures generated by his own grammar (that is, belong to his own idiolect), and those with which he merely happens to be acquainted. This is most obvious as regards vocabulary. English is particularly rich in lexical borrowings from other languages; and such borrowings differ in the extent to which they have been assimilated to the native word-stock. The average English-speaker may be uncertain whether such words as b&charnel or Lebensraum should figure in his internalised lexicon or not. It is not that he will be uncertain about whether he uses them or recognises them in the speech of others. It is that he will not know whether to count them as ‘his’. If the fact that he uses and recognises them were taken as deciding the issue in favour of their inclusion in his lexicon, then there would seem to be no reason not to include within his idiolect all the lexical items known to him belonging to any foreign languages that he happens to be acquainted with. Similar indeterminacies can be adduced in respect of any other aspect of linguistic structure. He may be acquainted with variant pronunciations of certain words, without necessarily using both indiscriminately. Economics can be either [i:kannmrks] or [ekannmtks], but most people probably use one or the other consistently. There may be syntactic constructions which, although in some sense he ‘knows’ them, are not clearly part of his own usage. Purists fight against the recent tendency to treat the verb heZp as a quasi-auxiliary, followed by a dependent infinitive without to (as in helps fight tooth decay, as opposed to helps to fight tooth decay). Speakers of a form of English in which the normal way of framing negative yes/no questions is to attach the contracted form of not to the auxiliary and prepose auxiliary-plus-contraction to the subject NP (as in haven’t you seen the film?) may or may not be aware of, or use as a variant, the alternative construction which involves preposing the auxiliary but leaving not uncontracted and to the right of the subject NP (as in have you not seen the film?). How intimate does their relationship with this alternative have to be for it to figure in their grammar? To take a semantic example, does aggravate, for a given English-speaker, mean ‘make worse’, or ‘annoy’ or both? Suppose he is someone who thinks it ought to mean ‘make worse’, and uses it exclusively in that sense himself, but nonetheless acknowledges that the word has an alternative or additional meaning. How does his knowledge pertaining to aggravate divide between that which is provided by the semantic part of his internalised grammar, and that which is merely known from acquaintance with the usage of others? Where are the answers to such questions supposed to come from?

These various problems seem formidable enough. But Botha caps them by offering yet a third interpretation of the ideal speaker-listener. Under interpretation (a) we start with a given E-language, and treat real speakers of this language as non-ideal; the grammars we write are grammars of E-languages. Under interpretation (b) we start with real speakers as given, and write grammars of their I-languages. While in practice it may be difficult to delimit the scope of such a grammar, it is at any rate supposed to be a property of real speaker-listeners. Or so one might have thought. But Botha makes it clear (e.g. p. 142) that, at least in his view, there is a second layer of idealisation involved. On Botha’s

82 NIGEL LOVE

interpretation (c) a generative grammar characterises the linguistic knowledge not of an ideal speaker of an E-language, nor of a real speaker of an I-language, but of an idealised speaker of an I-language. But this double idealisation makes it in principle impossible to write generative grammars. To contemplate such an enterprise at all, one must have at least one foot in the real world. Interpretation (a) offers an implementable recipe for grammar-writing: take the forms and constructions of a language (an E-language) and state them in generativist terms. Interpretation (b) offers a theoretically conceivable recipe for grammar-writing: take a real live speaker-listener, attempt to determine (e.g. by eliciting from him his ‘intuitions’) his ‘linguistic knowledge’, and state that knowledge in generativist terms. But interpretation (c) simply leaves us with nowhere to begin, even in theory.

What are we to make of the claims and pretentions of psycho-biological generativism? ‘Noam Chomsky’, asserts one recent commentator (Gellner, 1985, p. lOl), is ‘a thinker second only to Freud in his impact on our views on the mind’. This might conceivably be true, but only because it is possible to come second and still lag a long way behind. Freud ranks with Marx as one of the great modern mythologists of the human condition. Singlehandedly he has equipped Western man with his preferred idiom for articulating and apprehending the soul’s afflictions. Pace Gellner, Chomsky has wrought no comparable transformation. Despite all the huffing and puffing in the 30 years since Chomsky published his first book, the average person’s thinking about language and mind has obstinately refused to bear out the large claims that have persistently been made for the importance of his ‘impact’. Perhaps that impact has been poorly mediated by those who have taken on the task of Chomskyan exegesis. Perhaps Chomsky’s subject-matter is inherently less interesting, less vital, than Freud’s or Marx’s. All the same, by this time, one could reasonably expect something more to have emerged by way of a captivating account of the new vision in linguistics than the sort of defensively programmatic statements of intent

(e.g. Lightfoot, 1982; Newmeyer, 1983) that are all that have been offered. Might it not be that something more captivating is inherently impossible?

An important reason for thinking so can be deduced from the foregoing discussion. One might start by observing that it is virtually impossible even to state Chomsky’s theory of language acquisition clearly without getting ensnared in a conceptual incoherence. What a child acquires, according to the theory, is an I-language. His I-language is his own, unique idiolect. Nonetheless, what is seen as requiring explanation about the process of acquiring this I-language is the child’s ability, given the inadequacy of the ‘evidence’, to achieve a swift and error-free mastery of ‘the language’ of his community. But what is this community language? For the community, presumably, consists of nothing but particular individuals, whose linguistic output reflects their particular idiosyncratic idiolects. On the face of it, the theory has no room for a community language (an E-language) over and above the sum total of its members’ idiolects, except as a cultural construct erected via retrospective analysis of I-language-controlled linguistic output. Having taken this in, however, we have to understand that the community language somehow comes into the picture after all. For it sets the target at which the child is held to be aiming with the inadequate weapons provided by real-world experience; and it is into the gap between the power of the weapons and what they are called upon to achieve that Chomsky rhetorically inserts the need to postulate a distinct language faculty. One can only tell that the child has succeeded in learning his language, let alone be amazed at how rapidly and efficiently he has done so, if one has some yardstick against which to measure his achievement. That yardstick is the language of his community. In other words, the Chomskyan concept of an internal I-language seems

IDEAL LINGUISTICS 83

to be inextricably tangled up with ‘some pretheoretical (or alternative theoretical) sense [of the notion of “a language”] . . . unavoidably present but not theoretically allowed for’ (Pateman, 1987, p. 44). Hence the baffling uncertainty as to whether generative grammars are supposed to describe languages or idiolects, as well as the mysterious insistence that the language or idiolect under description should be ‘pure’. In the end, E-languages are apparently too precious to be given up. The reason that Chomsky’s much-trumpeted impact on our ideas about the mind has been so curiously insubstantial in practice is that Chomsky’s linguistics is not really about the mind at all. It is about languages, in the traditional grammarian’s sense. Far from being a serious attempt to investigate the psycho- biological phenomena associated with language, it is an attempt to provide a scientistic vindication of the traditional concept of languages, by projecting languages on to minds as the basis for individuals’ linguistic behaviour. Languages, in this sense, were always idealised constructs, and the idealisation has inevitably been incorporated-albeit in an incoherently re-theorised guise-into the Chomskyan vision. The trouble is that nobody’s understanding of himself and his relation to his own linguistic behaviour is going to be radically affected by hypothetical consideration of what life might be like for ideal speaker- listeners in an ideal world.

And what are we to make of Botha’s anguishedly ambivalent third-order anatomising of second-order generativist controversies? It is hard to imagine such a book being written by someone who thought that there was useful substantive work to be done in the field. So should it be construed as an admission that generativism has nothing to offer-that there is nothing more worthwhile to do than settle down to playing the ‘garden game’? Certainly Botha’s concluding account of his intentions seems to support such an interpretation: he says merely (p. 235) that he has tried ‘to assist in populating The Garden with people of “the right stuff” ‘, thereby contributing ‘to raising the general level of The Game’. What he does not say, however, is why anyone should find the game worth playing.

NOTES

‘The reference here to ‘degenerate evidence’ appears to contradict Botha’s claim that Chomsky dismisses degeneracy of the evidence as irrelevant.

‘Alternatively, one might characterise an I-language as ‘what a native speaker’s linguistic knowledge is knowledge of’, or ‘the total output of his internalised grammar’. It is hard to see that it makes any difference whether one speaks in these terms or in those used in the text.

3’The language of such a speech community (e.g. the nineteenth-century Russian aristocracy) would not be “pure” in the relevant sense, because it would not represent a single set of choices among the options permitted by U[niversal] G[rammar] but rather would include contradictory choices for certain of these options’ (Chomsky, 1986, p. 17). Chomsky says that the sense of sense of ‘pure’ here ‘must be made precise’, and promises to return to the issue later. But he does not do so.

4Fetched, in fact, from Jones (1979).

‘Not only does spontaneous speech, on the whole, not consist of discrete stretches recognisable as the phonic counterparts of written sentences (it is notoriously difficult to analyse speech grammatically without a great deal of preliminary editing), but also the phonic medium permits the differentiation of spoken utterances in ways that have no parallel in writing. This latter point vitiates much of what generativists have to say about, for instance, ambiguity. One of the things that native speakers are held to know without being told is that certain sentences of their language are ambiguous. One of Lightfoot’s examples (1982, p. 19) of an ambiguous sentence is John kept the car in the garage. The two different readings can be brought out be expanding the sentence in different ways: (i) John kept the car in the garage, but sold the one in the street, (ii) John kept the cur in the garage, but left the bicycle iri the porch. In (i) in the garage is an adjectival phrase qualifying the car, in (ii) it is an adverbial phrase modifying kept. Learners of English, says Lightfoot, not only assimilate the fact that this sentence

84 NIGEL LOVE

is ambiguous, but also know without being told that the question what did John keep in the garage? corresponds exclusively to the second reading of the statement-sentence. But what requires explanation is why we should suppose that spoken English has a unit identified by citing the sequence of written forms John kept the cur in thegarage. For in speech not only will the two readings have different patterns of stress and intonation, but also in actual spoken use they will be embedded in a communicational context which will clarify the intended sense. As far as speech is concerned, that is, there is no reason to suppose that we have to do with a single ambiguous item in the first place.

6Botha inconclusively discusses (pp. 77-81) the question whether language acquisition was retrospectively invoked as something for generative grammatical theorising to be about.

‘This terminological distinction between ‘linguistic knowledge’ (tacit knowledge of one’s I-language as modelled by a generative grammar) and ‘knowledge of language’ (overt knowledge pertaining to linguistic phenomena considered as objects of contemplation) is borrowed from Smith and Wilson (1979). But this useful usage is by no means firmly established: Botha, for instance, regularly uses ‘knowledge of language’ to mean ‘linguistic knowledge’ in this sense.

REFERENCES

AWBERY, G. M. 1976 The Syntax of Welsh: A Transformational Study of the Passive. Cambridge University Press, Cambridge.

BERWICK, R. C. and WEINBERG, A. S. 1984 The Grammatical Basis of Linguistic Performance: Language Use and Acquisition. M.I.T. Press, Cambridge, MA.

CHOMSKY, N. 1965 Aspects of the Theory of Syntax. M.I.T. Press, Cambridge, MA.

CHOMSKY, N. 1980 Rules and Representations. Blackwell, Oxford.

CHOMSKY, N. 1983 On cognitive structures and their development: a reply to Piaget. In Piatteli-Palmarini, M. (Ed.) Lunguuge and Learning: The Debate between Jean Piuget and Noam Chomsky. Routledge & Kegan Paul, London.

CHOMSKY, N. 1986 Knowledge of Language: Its Nature, Origin and Use. Prager Publishers, New York.

GELLNER, E. 1985 The Psychoanalytic Movement. Granada Publishing, London.

JONES, M. 1979 The present condition of the Welsh language. In Stephens, M. (Ed.) The Welsh Language Today, 2nd edn. Comer Press, Llandysul.

KAYNE, R. S. 1975 French Syntax: The Transformational Cycle. M.I.T. Press, Cambridge, MA.

LAKOFF, R. 1968 Abstract Syntax and Latin Complementalion. M.I.T. Press, Cambridge, MA.

LEECH, G. 1974 Semantics. Penguin, Harmondsworth.

LEES, R. B. 1960 The Grammar of English Nominulisutions. Research Centre in Anthropology, Folklore and Linguistics, Bloomington.

LIGHTFOOT, D. 1982 The Language Lottery: Toward a Biology of Grammars. M.I.T. Press, Cambridge, MA.

MARKUS, R. 1982 Bridge. The Guardian Week/y, 5 December 1982, p. 23.

NEWMEYER, F. J. 1983 Grammatical Theory: Its Limits and Possibilities. University of Chicago Press, Chicago.

PATEMAN, T. 1987 Language in Mind and Language in Society. Clarendon Press, Oxford.

RADFORD, A. 1977 Italian Syntax: Transformational and Relational Grammar. Cambridge University Press, Cambridge.

RADFORD, A. 1981 Transformational Syntax: A Student’s Guide to Chomsky’s Extended Standard Theory. Cambridge University Press, Cambridge.

ROSENBAUM, P. S. 1967 The Grammar of English Predicate Complement Constructions. M.I.T. Press, Cambridge, MA.

SMITH, N. and WILSON, D. 1979 Modern Linguistics: The Results of Chomsky’s Revolution. Penguin, Harmondsworth.