by Noam Chomsky
8 Obviously, the properties of being unbounded and being stimulus-free are independent. An automaton may have only two responses that are produced randomly. A tape recorder or a person whose knowledge of a language extends only to the ability to take dictation has an unbounded output that is not stimulus-free in the intended sense. Animal behavior is typically regarded by the Cartesians as unbounded, but not stimulus-free, and hence not “creative” in the sense of human speech. Cf., for example, Francois Bayle, The General System of the Cartesian Philosophy (1669) (English translation 1670, p. 63): “And because there may be an infinite variety in the impressions made by the objects upon the senses, there may also be an innumerable variety in the determination of the Spirits to flow into the Muscles, and by consequence, an infinite variety in the Motions of Animals; and that the more, because there is a greater variety of parts, and more contrivance and art in the structure.” The unboundedness of human speech, as an expression of limitless thought, is an entirely different matter, because of the freedom from stimulus control and the appropriateness to new situations.
It is important to distinguish “appropriateness of behavior to situations” from “control of behavior by stimuli.” The latter is characteristic of automata; it is the former that is held to be beyond the bounds of mechanical explanation, in its full human variety.
Modern studies of animal communication so far offer no counterevidence to the Cartesian assumption that human language is based on an entirely distinct principle. Each known animal communication system either consists of a fixed number of signals, each associated with a specific range of eliciting conditions or internal states, or a fixed number of “linguistic dimensions,” each associated with a nonlinguistic dimension in the sense that selection of a point along one indicates a corresponding point along the other. In neither case is there any significant similarity to human language. Human and animal communication fall together only at a level of generality that includes almost all other behavior as well. [Studies since 1966 continue to indicate that there is no counterevidence. Studies also show that unless humans are given at least a minimal amount of experience of the relevant kind (hearing or seeing language spoken or signed by others, for example) before a certain critical stage, they cannot acquire full linguistic competence. See, among others, the study of Genie in Curtiss 1976.
I (the ed.) emphasize the special status of appropriateness in the creative aspect of language use in the new introduction to this third edition. Recursion can make sense of innovation or the unbounded characteristics of language, and – at a stretch – a randomizing element could deal with stimulus freedom. So a computer program might allow for both. But there is no obvious way to make sense of appropriateness while meeting the other two. It has proven impossible so far – and may always prove so – to meet all three conditions of being linguistically creative.]
9 In general, then, “although machines can perform certain things as well as or perhaps better than any of us can do, they infallibly fall short in others, by the which means we may discover that they did not act from knowledge, but only from the disposition of their organs.” There are, then, two “very certain tests” by which we can determine whether a device is really human, the one provided by the creative aspect of language use, the other, by the diversity of human action. “It is virtually impossible” (in the Haldane-Ross translation, “morally impossible”) “that there should be sufficient diversity in any machine to allow it to act in all the events of life in the same way as our reason causes us to act.” In taking this position, Descartes expands on his conception of the “cognitive power” as a faculty which is not purely passive and which is properly called “native intelligence [ingenium]” when it “ forms new ideas in the corporeal imagination, or concentrates on those already formed,” acting in a way that is not completely under the control of sense or imagination or memory (Rules for the Direction of the Mind (1628); CSM I, 42). Still earlier, Descartes remarks that “the high degree of perfection displayed in some of their actions makes us suspect that animals do not have free will” (“Olympian Matters” c.1620; CSM I, 5).
The idea that the “cognitive power” is properly called “mind” only when it is in some sense creative has earlier origins. One source that might well have been familiar to Descartes is Juan Huarte’s Examen de Ingenios (1575), which was widely translated and circulated (I quote from the English translation by Bellamy, 1698). Huarte understands the word Ingenio to have the root meaning “engender,” “generate” – he relates it to gigno, genero, ingenero (p. 2). Thus “one may discover two generative Powers in Man, one common with the Beasts and Plants, and the other Participating of Spiritual Substances, God and the Angels” (p. 3). “Wit [Ingenio] is a generative power . . . the Understanding is a Generative Faculty” (p. 3). As distinct from divine “Genius,” the human “rational soul” and “spiritual substances” do not have “sufficient Force and Power in their Generation to give real being to what they Ingender” but only “to produce an accident in the Memory,” “an Idea and Image of what we know and understand” that must be given concrete existence by work and art (pp. 4–5). Similarly the arts and sciences are “a sort of Images, and Figures, begotten by [men’s] Minds in their Memory, which represent to the Life the Posture and natural Composition of the Subject relating to the intended Science” (p. 6). One who learns some subject must “Engender within himself an entire and true Figure” that represents its principles and structure (p. 6). Truly active minds will be “such, that assisted by the subject only, [they will] without the help of any Body, produce a thousand Conceits they never heard spoke of ” (p. 7). The empiricist maxim, “That there is nothing in the Understanding, but what has past through the Sense,” attributed to Aristotle, applies only to “docile wits” that lack this capacity. Although the “perfect wit” is only an ideal case, “yet it must be granted, we have observ’d many Persons approach very near it, inventing and saying such things as they never heard from their Masters, nor any Mouth” (p. 16). There is even a third kind of wit “by means of which, some have without Art or Study spoke such subtle and surprizing things, and yet true, that were never before seen, heard, or writ, no nor ever so much as thought of ” and which may involve “a mixture of Madness” (p. 17); these three types of wit involve the memory, understanding, and imagination, respectively. In general, “all [man’s] Honour and Nobility, as Cicero observed, consists in his being favour’d with, and having an Eloquent Tongue: As Wit is the Ornament of a Man, so Eloquence is the Light and Beauty of Wit. In this alone he distinguishes himself from the Brutes, and approaches near to God, as being the greatest Glory which is possible to be obtained in Nature” (p. 22). The most severe “disability of wit,” under which men “differ not at all from Brute Beasts,” is the disability, which “very much resembles that of Eunuchs . . . unable for Generation,” that prevents the rational faculty from arriving at “the first Principles of all Arts implanted in the Scholar’s Mind, before he begin to learn, for which the Wit can give no other proofs of itself, than to receive them as things already known; and if he be not able to form an Idea of them in his Mind, we may strongly conclude him wholly incapable of the Sciences.” In this case, “neither the Lash of the Rod, nor his Cries, nor Method, nor Examples, nor Time, nor Experience, nor any thing in Nature can sufficiently Excite him to bring forth any thing” (pp. 27–28).
See K. Gunderson, “Descartes, La Mettrie, Language and Machines,” Philosophy 39 (1964), pp. 193–222, for an interesting discussion of Descartes’s arguments as related to contemporary discussions of “intelligence” of automata. For general background on the development and critique of Descartes’s theory of the extent and limits of mechanical explanation, see Rosenfield, op. cit., and H. Kirkinen, “Les origines de la conception moderne de l’homme-machine,” Annales Academiae Scientiarum Fennicae, ser. B, vol. 22, Helsinki (1961).
10 Translated (in part) in H. A. R. Torrey, The Philosophy of Descartes (New York: Holt, 1892), pp. 2
81–284. [The translation that appears here, and in all subsequent quotations from Descartes’s correspondence, is from The Philosophical Writings of Descartes, vol. III: The Correspondence, trans. J. Cottingham, R. Stoothoff, D. Murdoch and A. Kenny (Cambridge: Cambridge University Press, 1991) (abbreviated CSMK).]
11 That is, by conditioning. When animals are taught “by art,” their actions are produced with reference to a passion, in the sense that this behavior is associated with the “stir of expectation of something to eat” or the “motions of their fear, their hope, or their joy” that constitute the original contingency for the teaching. Descartes is therefore pointing out that, just as in its normal use “verbal behavior” is free of identifiable external stimuli or internal physiological states, so it is evidently not developed in the individual by conditioning. He does not elaborate on this, regarding it perhaps as too obvious to merit discussion. It is noteworthy that modern behaviorist speculation about human learning denies these truisms. For some discussion, see Chomsky, “Review of Skinner, ‘Verbal Behavior,’” Language 35 (1935), pp. 26–58; Aspects of the Theory of Syntax, chap. I, §8; J. Katz, Philosophy of Language (New York: Harper & Row, 1966); J. Fodor, “Could Meaning be an ‘rm’’” Journal of Verbal Learning and Verbal Behavior 4 (1965), pp. 73–81. [For useful contemporary discussion of organisms’ modular learning and its explanation, see Gallistel 1990, 2002. Chomsky has in recent years referred approvingly to Gallistel’s work. For a fascinating study of linguistic modularity in a polyglot savant, see Smith and Tsimpli 1995.]
12 The Descartes–More correspondence, in so far as it relates to animal automatism, is translated in full by L. C. Rosenfield (L. Cohen) in the Annals of Science 1 (1936) [and in CSMK].
13 Descartes goes on to explain that he does not deny to animals life or sensation or even feeling, in so far as it depends only on the bodily organs.
14 Discours physique de la parole (1666). Page references are to the second edition, 1677. There is an English translation, dated 1668. Rosenfield remarks that Cordemoy develops Descartes’s argument involving lack of true speech among animals so fully that after him “the point was given very little attention, as if subsequent authors considered this the last word on the subject” (From Beast-Machine to Man-Machine, p. 40).
15 There is no problem, for Cordemoy (as for Descartes), in determining whether he himself possesses a soul, since it is evident to him, by introspection, “that certain thoughts always accompany in me most of the movements of my organs” (p. 3).
16 La Mettrie, L’homme machine (1747). A critical edition with notes and background material is La Mettrie’s L’homme machine: A Study in the Origins of an Idea, ed. A. Vartanian (Princeton: Princeton University Press, 1960). [The translations given here are from La Mettrie 1996, but reference is also given to the translation in Man A Machine (La Salle, Ill.: Open Court, 1953) (abbreviated MaM), which contains the French text.]
17 Père G. H. Bougeant, Amusement philosophique sur le langage des bestes (1739).
18 This is not to deny that the method of explanation suggested by La Mettrie may be in principle correct. What concerns me here is not the adequacy of the proposed explanations of Descartes and others, but the observations on human language that elicited these attempts. [The prospect of being unable ever to scientifically explain the creative aspect of language use plays an important role in Chomsky’s post-1966 discussions, where it illustrates general claims about the (biological) limitations of human intelligence. See the editor’s introduction for discussion and references.]
19 G. Ryle, The Concept of Mind (London: Hutchinson, 1949). See J. Fodor, “Is Psychology Possible?” chap. I of Psychological Explanation (New York: Random House, 1968), for a critique of the views of Ryle and others regarding psychological explanation.
20 These are described in terms of “powers,” “propensities,” and “dispositions,” which are characterized only through scattered examples. These constitute a new “myth” as mysterious and poorly understood as Descartes’s “mental substance.”
21 L. Bloomfield, Language (New York: Holt, 1933), p. 275. When a speaker produces speech forms that he has not heard, “we say that he utters them on the analogy of similar forms which he has heard.” For Bloomfield, human language differs from animal communication systems in no fundamental way, but only by its “great differentiation.” Otherwise, its function is similar. “Man utters many kinds of vocal noise and makes use of the variety: under certain types of stimuli he produces certain vocal sounds, and his fellows, hearing these same sounds, make the appropriate response” (p. 27). He holds that “language is a matter of training and habit” (p. 34) and that with careful statistical investigation “we should doubtless be able to foretell how many times any given utterance . . . would be spoken within a fixed number of days” (p. 37) (a conclusion that is certainly correct, since for almost all normal utterances the predicted number would be zero).
22 C. F. Hockett, A Course in Modern Linguistics (New York: Macmillan, 1958), §36, p. 50. He remarks that “it has been said that whenever a person speaks, he is either mimicking or analogizing,” and he accepts this view, stating that “when we hear a fairly long and involved utterance which is evidently not a direct quotation, we can be reasonably certain that analogy is at work” (p. 425). Among modern linguists, Hockett is unusual in that he has at least noticed that a problem exists. In discussing innovation, Hockett seems to imply that novel expressions can be understood only through reference to context (p. 303). In fact, failure to consider the linguistic mechanisms that determine the meaning of the ordinary, generally quite novel sentences of everyday life is typical of modern linguistics.
23 Modern discussions of the difference between human language and animal communication systems occasionally recapture some of the Cartesian insights. See, for example, L. Carmichael, “The Early Growth of Language Capacity in the Individual,” in New Directions in the Study of Language, ed. E. H. Lenneberg.
24 J. G. Herder, Abhandlung über den Ursprung der Sprache (1772). This is now available in part, in Herder’s Sprachphilosophie, ed. E. Heintel (Hamburg: Felix Meiner Verlag, 1960), pp. 1–87. Page references are to this volume. [Translations from this work are by Susan-Judith Hoffmann; references remain as they were in the original edition. There is a modern translation of the work in Herder 1966.]
25 This is true as well of the development of language in the individual. Study of the “origin of language” is essentially a study of the “essence of language,” in this period, and the growth of language in the individual and its growth in the nation are often taken to be parallel in their general characteristics. Cf. A. W. Schlegel, Die Kunstlehre (1801) (Stuttgart: W. Kohlhammer Verlag, 1963), p. 234: in the discovery of language by children, “that which takes place in the invention of language by the human race in general finds itself ever repeated, albeit in fainter traces”; in general, “in the acquisition of language, we find the same ability at work, which is present in the invention of language, in a higher degree” (p. 235). Under the influence of Humboldt, H. Steinthal goes even further and states, “There is no distinction between the primordial creation of language and its daily re-creation” (Grammatik, Logik und Psychologie. [Berlin, 1855], p. 232).
26 Discourse on the Method, CSM I, 140.
27 Descartes does not restrict language to purely intellectual function in a narrow sense. See, for example, Principles of Philosophy, pt. IV, art. 197 (CSM I, 284):
For we see that spoken or written words excite all sorts of thoughts and emotions in our minds. With the same paper, pen and ink, if the tip of the pen is pushed across the paper in a certain way it will form letters which excite in the mind of the reader thoughts of battles, storms and violence, and emotions of indignation and sorrow; but if the movements of the pen are just slightly different they will produce quite different thoughts of tranquillity, peace and pleasure, and quite opposite emotions of love and joy.
28 Treatise the Third: Concerning Happiness,
a Dialogue (1741). In Harris’s Works, ed. the Earl of Malmesbury (London: F. Wingrove, 1801), vol. I, p. 94.
