by Noam Chomsky
93 See Postal, Constituent Structure (The Hague: Mouton, 1964), for discussion of contemporary approaches to syntax that accept this limitation. Many modern methodological discussions actually imply, further, that linguistic investigation should be restricted to the surface structure of the given utterances of a fixed corpus; thus Sahlin reflects modern attitudes in criticizing Du Marsais (p. 36) for the “inexcusable fault on the part of a grammarian” of using invented examples instead of restricting himself to utterances actually observed in living speech, as though a rational alternative were conceivable.
For further discussion of the problem of analyzing deep and surface structure see Chomsky, Syntactic Structures (The Hague: Mouton, 1957), Current Issues in Linguistic Theory, Aspects of the Theory of Syntax; Lees, Grammar of English Nominalizations (The Hague: Mouton, 1960); Postal, “Underlying and Superficial Linguistic Structures,” Harvard Educational Review 34 (1964); Katz and Postal, An Integrated Theory of Linguistic Descriptions; Katz, The Philosophy of Language, and many other publications.
94 To mention just one example, consider Harnois’s introductory statement in his discussion of “philosophical grammar” (“Les théories,” p. 18; it should be emphasized that this discussion is unusual in that it at least pays attention to the actual doctrines that were held by philosophical grammarians, instead of attributing to them absurd beliefs that were completely counter to their actual work). He points out that participants in this work felt themselves to be contributing “a science which had already produced one fundamental work [viz., the Port-Royal Grammar], namely by enriching an existing tradition and adding to the numerous results already attained. This opinion may appear ridiculous to a modern linguist, but it was really held.”
It should be mentioned that the modern disparagement of traditional linguistic theory develops, not only from the decision to restrict attention to surface structure, but also, quite often, from the uncritical acceptance of a “behaviorist” account of language use and acquisition, common in its essentials to several fields – an account that seems to me to be pure mythology. [Note too the quotation from the reprint of Chomsky’s review of B. F. Skinner’s Verbal Behavior in the editor’s introduction’s note 20.]
95 Véritables principes de la grammaire (1729), quoted by Sahlin, César Chesnau, pp. 29–30. The dating of this is discussed by Sahlin in the Introduction, p. ix. Much earlier, Arnauld had pointed out that “one has not usually treated as matters of particular grammars what is common to every language” (1669, cited by Sainte-Beuve, Port-Royal, p. 538), and the distinction between general and particular grammar is implicit, though not expressed, in the Port-Royal Grammar. Wilkins also distinguishes between “natural” (that is, “philosophical,” “rational,” or “universal”) grammar, which deals with the “ground and rules as do necessarily belong to philosophy of letters and speech,” and “instituted” or “particular” grammar, which deals with the “rules which are particular to a given language” (Essay, p. 297).
96 Beauzée, Grammaire générale, Preface, pp. v–vi.
97 Quoted by Sahlin, César Chesneau, p. 21. Note that there is a difference in emphasis in the remarks of Beauzée and D’Alembert on the relation between particular facts and general principles. The two views, however, are not inconsistent.
98 Cf. Sainte-Beuve, Port-Royal, pp. 538f.; Harnois, “Les théories,” p. 20.
99 There is, to be sure, an implicit element of so-called “prescriptivism” in his choice of “cultivated usage” (that is, the usage of the best authors, but, particularly, “the usage of spoken language” in the Court) as the object of description.
100 Note that a restriction of linguistic study to description without explanation does not entail a corresponding restriction to the investigation of surface structure. The latter is a further and independent limitation. [Compare the restriction to “pure description” to Wittgenstein’s Blue Book and Philosophical Investigations. It is interesting that Chomsky, like Wittgenstein, holds that it is very likely impossible to construct a science (serious theory) of language use – and for parallel reasons having to do with language’s creative use (although Wittgenstein did not use this terminology). Chomsky, of course, unlike Wittgenstein, holds that it is possible to construct a science of language (Universal Grammar). For some discussion, see the editor’s introduction and McGilvray 1999, 2005.]
101 Vaugelas is by no means the first to insist on the primacy of usage. A century before, in one of the earliest French grammars, Meigret insists that “we must speak in the way that we do speak” and that one may not “make any law against the way French is usually pronounced” (quoted by Ch.-L. Livet, La grammaire française et les grammairiens du XVIe siècle).
It is interesting to note that the reaction of the Cartesian linguists against pure descriptivism recapitulates the evolution of speculative grammar in the thirteenth century, as an attempt to provide rational explanation in place of a mere record of usage. Speculative grammar also distinguished universal from particular grammar; for example, Roger Bacon assumes that “with respect to its substance grammar is one and the same in all languages, although it does vary accidentally (Grammatica Graeca, ed. Charles, p. 278, cited in N. Kretzmann, “History of Semantics,” in Encyclopedia of Philosophy, ed. P Edwards (New York: Macmillan, 1967)).
102 Quoted by Sahlin, César Chesneau, p. 26, from the article “Datif” in the Encyclopedia. Sahlin also gives (p. 45) a much earlier quote from the Véritables principes (see note 95): “Grammar does not come before languages. There is no language that has been based on grammar; the rules [observations] of grammarians must be based on usage, and are not laws that have preceded usage.” This quote is followed by the comment that Du Marsais did not adhere to this principle, but, though there is much to criticize in his work, I find little evidence to support this charge.
103 This is, of course, consistent with Cartesian methodology, which insists on the necessity of observation and of crucial experiment for choice among competing explanations. See Discourse on the Method, part VI. The Cartesian origins of the concern for a “general (universal) grammar” [grammaire générale] (expressing what is a common human possession) and an “explanatory grammar” [grammaire raisonnée] (which will explain facts instead of merely listing them) are too obvious to require discussion. Similarly, it was the newly rediscovered Aristotelian concept of rational science that led to the speculative grammar of the thirteenth century. Cf. Kretzmann, op. cit.
104 This discussion is due to Arnauld and appears in his correspondence a year before the publication of the Grammar. Cf. Sainte-Beuve, op. cit., pp. 536f.
The Grammar is, incidentally, not entirely fair to Vaugelas in tacitly implying that he was unaware of counter-examples. In fact, Vaugelas himself mentions one of the cited counter-examples (namely, the vocative, for which he proposes an understood, deleted article). Furthermore, Vaugelas does in fact offer a tentative explanation, rather apologetically, to be sure, for the rule as he formulates it.
105 For further discussion of the matter of explanation in linguistics, see Chomsky, Syntactic Structures; “Explanatory Models in Linguistics,” in Logic, Methodology and Philosophy of Science, ed. E. Nagel, P. Suppes, A. Tarski (Stanford: Stanford University Press, 1962); Current Issues in Linguistic Theory, and J. Katz, “Mentalism in Linguistics,” Language 40 (1964), pp. 124–137. [See also Chomsky 1975a, 1975b, 1980, 1981, 1986, 1988a, 1992, 1995b, 2000, 2005, 2007. In a sense, Chomsky’s most recent work has gone “beyond explanation” for – unlike the state of linguistics in 1966 – he can now plausibly hold that the issue of explanatory adequacy as he originally posed it in Aspects (in effect, solving “Plato’s problem”) has been sufficiently settled to move on to other matters. Now he focuses on other explanatory issues – matters of computational economy and biological embodiment, some of which bear on the evolution of language and on ‘accommodating’ language to biology. See the editor’s introduction. Incidentally, it is illuminating to compare Chomsky�
�s view on innateness with Humboldt’s.]
One of the most striking features of American descriptivism in the 1940s was its insistence on justification in terms of precisely specified procedures of analysis. The emphasis on precision and on the necessity for justification of descriptive statements in some language-independent terms constitutes a major contribution. But the requirements that were placed on justification (namely, that it be “procedural,” in the sense of the methodological discussions of the 1940s) were so strong as to make the enterprise unfeasible, and some of the reactions to this stringency (in particular, the view that any clearly specified procedure of analysis is as good as any other) detracted substantially from its potential significance.
106 Observe, however, that the discussion in the Port-Royal Grammar, if interpreted quite literally, does not identify the underlying structures with actual sentences. Cf. pp. 83–84 above, and note 73. It is thus quite close, in conception, to transformational generative grammar of the sort developed in the references of note 93, which has also been based on the assumption that the structures to which transformational rules apply are abstract underlying forms, not actual sentences. Notice, incidentally, that the theory of transformations as originally developed by Harris, outside of the framework of generative grammar, does regard transformations as relations among actual sentences and is, in fact, much closer to the conception of Du Marsais and others, in this respect (see Z. S. Harris, “Co-occurrence and Transformation in Linguistic Structure,” Language 33 (1957), pp. 283–340, and many other references). See Chomsky, Current Issues in Linguistic Theory, p. 62n., for some discussion bearing on this point.
107 Humboldt’s picture was, however, a good bit more complex. Cf. pp. 69–73 above.
108 Notice that, when described in these terms, linguistic universals need not be found in every language. Thus, for example, when a certain set of phonetic features is claimed to constitute a universal phonetics, it is not proposed that each of these features functions in every language, but rather that every language makes its particular choice from among this system of features. Cf. Beauzée, op. cit., p. ix: “the necessary elements of language. . . are in fact present in all languages, and their necessity is indispensable for the analytic and metaphysical exposition of thought. But I do not intend to speak of an individual necessity, which does not leave anyone free to reject any idiom; I mean to indicate only a specific necessity [une nécessité d’espèce], which sets the limits of the choices that one can make.” [This view of a mind’s ‘choice’ among phonetic features anticipates Chomsky’s later (early 1980s) principles and parameters approach to the ‘choices’ a child’s mind makes in acquiring a language. For discussion, see the editor’s introduction.]
109 Translated by M. H. Carré (1937), University of Bristol Studies, No. 6.
110 These developments are familiar except, perhaps, for seventeenth-century English Platonism. See A. O. Lovejoy, “Kant and the English Platonists,” in Essays Philosophical and Psychological in Honor of William James (New York: Longmans, Green, 1908), for some discussion of English Platonism, in particular, of its interest in the “ideas and categories which enter into every presentation of objects and make possible the unity and interconnectedness of rational experience.” Lovejoy’s account, in turn, is based heavily on G. Lyons, L’idéalisme en Angleterre au XVIIIe siècle (Paris, 1888). See also J. Passmore, Ralph Cudworth (Cambridge: Cambridge University Press, 1951); L. Gysi, Platonism and Cartesianism in the Philosophy of Ralph Cudworth (Bern: Herbert Lang, 1962). Some relevant quotes from Descartes, Leibniz, and others are given in Chomsky, Aspects of the Theory of Syntax, chap. 1, §8, where the relevance of this position to current issues is also briefly discussed. [Chomsky notes in CL’s conclusion (p. 107 below) that some figures have been omitted from his survey of ‘Cartesian linguists’, or have been inadequately discussed. He mentions Immanuel Kant in particular. It is perhaps significant that the Cambridge Platonists had more to say about the scientific issues of language acquisition that Chomsky discusses than Kant, who was primarily interested in epistemological issues and had little to say that could be seen as anticipating Chomsky’s ‘biologizing’ of language.]
See also Chomsky, Explanatory Models in Linguistics, and Katz, Philosophy of Language, for discussion of an essentially rationalist approach to the problem of language acquisition and of the inadequacy of empiricist alternatives. In the same connection, see Lenneberg, “Biological Perspective,” and Biological Foundations of Language (New York: John Wiley, 1967), and §VI of The Structure of Language: Readings in the Philosophy of Language, ed. J. Fodor and J. Katz (Englewood Cliffs: Prentice-Hall, 1964). [See also Jenkins 1999 and several of the chapters of McGilvray 2005, plus references in both. Biolinguistics remains a burgeoning field of study. A particularly good starting point is Hauser, Chomsky, Fitch 2002; a very useful overview is found in Chomsky (2007 and forthcoming); some more technical discussion is in Chomsky 2005.]
111 Leibniz, Discourse on Metaphysics. The quotations here are from the English translation in Leibniz: Discourse on Metaphysics and Correspondence with Arnauld, trans. G. Montgomery (La Salle: Open Court, 1902). With reference to Plato’s theory, Leibniz insists only that it be “purged of the error of pre-existence.” Similarly, Cudworth accepts the theory of reminiscence without the doctrine of preexistence that Plato suggests as an explanation for the facts he describes: “And this is the only true and allowable sense of that old assertion, that knowledge is reminiscence; not that it is the rememberance of something which the soul had some time before actually known in a pre-existent state, but because it is the mind’s comprehending of things by some inward anticipations of its own, something native and domestic to it, or something actively exerted from within itself” (Treatise concerning Eternal and Immutable Morality, p. 424; page references, here and below, are to the first American edition of works of Cudworth, vol. II, T. Birch, ed., 1838). [The quotation is found on p. 74 of Sarah Hutton’s recent edition of the Treatise (Cudworth 1996). References to this text have been added to Chomsky’s references below and in the text.]
Leibniz’s view (Discourse on Metaphysics, §26) that “the mind at every moment expresses all its future thoughts and already thinks confusedly of all that of which it will ever think distinctly” might be regarded as suggesting the fundamental insight regarding language (and thought) that we discussed in §2.
112 Cf. Beauzée, Grammaire générale, pp. xv–xvi. He defines “grammatical metaphysics” [la Métaphysique grammaticale] as being nothing but “the nature of language brought into the open, established in its own terms, and reduced to general notions”:
The fine points that this metaphysics discovers in language. . .come from eternal reason, which unconsciously directs us. . . It would be vain to claim that those who speak the best are not aware of these delicate principles. How could they put them into practice so well unless they were somehow aware of them? I admit that they would perhaps not be ready to use all the rules in their reasoning, because they have not studied them systematically [l’ensemble et le système]; but in the end, since they apply these principles, they are conscious of them within themselves; they cannot escape from the imprints of this natural logic which, covertly but irresistibly, directs honest minds in all their operations. But general grammar is simply the rational exposition of the procedures of this natural logic.
113 But cf. p. 97 above. The typical Cartesian view would apparently have been that, although these principles may function unconsciously, they can be brought to consciousness by introspection.
114
But whatever trouble we take to teach them certain things, we often find that they know the names of a thousand others that we did not intend to show them; and, what is more surprising, we see that when they are two or three years old they are capable, through the mere force of their attention, to distinguish the name we give to a thing in all the constructions we use in speaking about it.
(pp. 47–48)
H
e also points out that children learn their native language more easily than an adult can learn a new language.
It is interesting to compare these quite commonplace but perfectly correct observations with the picture of language learning that one generally finds among many modern writers, whose conclusions are, in fact, based not on observation but on a priori assumptions about what they believe must take place. Cf., e.g., the speculation on how all language “habits” are built up by training, instruction, conditioning, and reinforcement in Bloomfield, op. cit., pp. 29–31; Wittgenstein, Blue Book (Oxford: Blackwell, 1958), pp. 1, 12–13, 77; Skinner, Verbal Behavior (New York: Appleton-Century-Crofts, 1957); Quine, Word and Object (Cambridge and New York: M.I.T. Press and John Wiley, 1960); etc. [See also the editor’s introduction.]
Occasionally, modern discussions invoke some process of “generalization” or “abstraction” that functions along with association and conditioning, but it must be emphasized that there is no known process of this sort that will begin to overcome the inadequacy of empiricist accounts of language acquisition. For discussion, see the references in note 110. In considering this problem, one must, in particular, bear in mind the criticism advanced by Cudworth (Treatise, p. 462; Cudworth 1996:114–116) against the attempt to show how general ideas might arise from sensory images (phantasms) by “abstraction” and thus require no postulation of innate mental structure. As he points out, the intellectus agens either “doth know what he is to do with these phantasms beforehand, what he is to make of them, and unto what shape to bring them,” in which case the question is begged, an “intelligible idea” being presupposed; or, if he has no such plan, “he must needs be a bungling workman,” that is, the act of “abstracting” can lead to any arbitrary and absurd result.
