by Noam Chomsky
For further discussion see J. Veitch, The Method, Meditations, and Selections from the Principles of Descartes (Edinburgh: Blackwood and Sons, 1880), note II, pp. 276–285.
71 In the French original, the cited sentence is: “La doctrine qui met le souverain bien dans la volupté du corps, laquelle a été enseignée par Epicure, est indigne d’un Philosophe.” The Dickoff-James translation, which I have followed elsewhere, translates this as: “The doctrine which identifies the sovereign good with the sensual pleasure of the body and which was taught by Epicurus is unworthy of a philosopher.” But in this translation the explicative relative “which was taught by Epicurus” would naturally be taken as a determinative clause conjoined with the first determinative clause “which identifies. . .” in which case the point of the example is lost. [In Arnauld and Nicole 1996, the sentence is translated: “The doctrine that places the highest good in bodily pleasure, which was taught by Epicurus, is unworthy of a philosopher” (90).]
72 Notice, incidentally, that adjective–noun constructions in the surface structure may derive by grammatical transformations of the type proposed in the Port-Royal Grammar from either type of relative, as is evident from the examples given there and, more strikingly, in such ambiguous examples as Jespersen’s “The industrious Japanese will conquer in the long run” (Philosophy, p. 112).
73 Notice that, in such cases, it is not true that each of the elementary abstract objects constituting the deep structure itself underlies a possible sentence; thus “je vous dis,” for example, is not a sentence in itself. In current terminology, it is not the case that each item generated by the underlying base (phrase structure) rules underlies a possible kernel sentence. Similarly, in all work in transformational generative grammar of the last ten years or more, it has been taken for granted that the phrase-structure rules can introduce “dummy symbols” that receive a representation in terms of morpheme strings only as a result of application of embedding rules of one sort or another (as, for example, in verb–complement constructions in English), and the elementary strings in which these dummy symbols appear will not underlie kernel sentences. Various related ideas that have been explored during this period are summarized and discussed in Chomsky, Aspects of the Theory of Syntax, chap. III. [See also the bracketed discussion and references for note 80.]
74 A rather different analysis of these structures is presented by Beauzée, op. cit. He regards them as based on relative clauses with the antecedent transformationally deleted. Thus the sentences “L’état présent des Juifs prouve que notre religion est divine,” “Ich glaube dass ich liebe,” and “I think (that) I love,” derive, respectively, from “L’état présent des Juifs prouve une vérité qui est, notre religion est divine,” “Ich glaube ein Ding dass ist, ich liebe,” and “I think a thing that is, I love” (p. 405).
75 For further discussion, see Chomsky, Aspects of the Theory of Syntax. It is worth mentioning that the theory of transformational generative grammar has in many respects moved toward a point of view like that implicit in the Port-Royal theory, as new evidence and insights have accumulated during the few years in which it has, once again, become an object of fairly intensive investigation.
76 Some earlier notions are reviewed by Sahlin, César Chesneau, pp. 97f. The idea that a sentence can be regarded simply as a sequence of words or word categories, with no further structure, is frequently expressed (whether or not it is actually believed) by many later writers.
77 Notice that this is referred to as the principal, not the unique role of verbs. They are also used “to indicate other movements of our minds, as in to desire, to ask, to command, etc.” (p. 90). These matters are taken up again in chap. XV, where the grammatical means by which these mental states and processes are realized in various languages are briefly discussed. See p. 79 above.
78 The Grammar goes on to observe that it would be a mistake to assume, with certain earlier grammarians, that verbs necessarily express actions or passions or something that is taking place, and it offers as counterexamples such verbs as “existit,” “quiescit,” “friget,” “alget,” “tepet,” “calet,” “albet,” “viret,” “claret” (p. 94).
79 As noted earlier (p. 117): “it is often necessary to transform such a sentence from the active to the passive voice in order to put the argument into its most natural form and to express explicitly that which is to be proved.”
80 It is hardly just to attribute this insight to twentieth-century British philosophy, as its “central and fundamental discovery” (cf. Flew, Introduction to Logic and Language, First series (Oxford: Blackwell, 1952), p. 7; or Wittgenstein, Tractatus Logico-Philosophicus (1922), 4.0031, where it is attributed to Russell). Nor is the observation that “grammatical resemblances and dissimilarities may be logically misleading” (Flew, p. 8) quite as novel an insight as Flew suggests. See, for example, p. 91 below.
The general assumption of Cartesian linguistics is that the surface organization of a sentence may not give a true and full representation of the grammatical relations that play a role in determining its semantic content, and, as we have noted, a theory of grammar is sketched in which actual sentences are derived from underlying “deep structures” in which these relations are grammatically represented. The extent to which “logical form” is actually represented by the syntactically defined deep structures, in the technical modern sense or the related sense suggested in Cartesian linguistics, is a further and in many respects open question. See J. Katz, The Philosophy of Language (New York: Harper & Row, 1966), for discussion.
[Chomsky entertained the view that ‘semantic intepretation’ takes place at deep structure in his 1965 Aspects. He was to abandon this idea soon after in favor of increasingly refined versions of a view he had adopted earlier in his Logical Structure of Linguistic Theory (published as Chomsky 1975b) and in Syntactic Structures, that semantic interpretation takes place at an ‘output’ level of a derivation, where “conceptual–intentional” systems use the product of a derivation as a “tool” (Syntactic Structures) for whatever operations they perform. By the 1970s, that output level of a derivation came to be called “LF” (for “logical form”) or, later in the 1990s, SEM (for “semantic interface”). Deep structure – but not as the ‘place’ where semantic interpretation takes place – remained until the early 1990s as the place where basic “thematic assignments” are made, but it was abandoned as the “minimalist program” developed and more and more of what used to be thought of as irreducible linguistic structure came to be seen as ‘epiphenomena’ of primitive operations. Quite recently (2001), even LF is abandoned as a ‘level’ of a derivation, and SEM comes to indicate simply an “interface” with other mental/internal systems. For relevant – but quite often technical – reading, see Syntactic Structures and Chomsky 1975b, 1965,1975a and b, 1980,1986,1992,1995b, 2000, 2001, 2005, 2007.
It is quite likely that Chomsky’s reading of the Cartesian linguists in the late 1950s and early 1960s influenced his decision to endorse – provisionally and temporarily – Katz and Postal’s suggestion that semantic interpetation takes place at deep structure, rather than at some ‘output’ level. For some comment on the influence of this reading from Chomsky, see his Current Issues in Linguistic Theory and Aspects of the Theory of Syntax, especially the latter.]
81 Referred to, typically, as the “natural order.” See pp. 75–76 above.
82 Many of Du Marsais’s published and unpublished works on language are printed posthumously in Logique et Principes de Grammaire (1769). Page references here are to this volume. The correlation between freedom of word order and inflection is noted by many other writers, e.g., Adam Smith in his Considerations concerning the First Formation of Languages.
83 When Bloomfield (along with many others) criticizes premodern linguistics for obscuring the structural difference between languages “by forcing their descriptions into the scheme of Latin grammar” (Language, p. 8), he is presumably referring to such claims as this, which he re
gards as having been disproven. If so, then it must be observed that his book contains no evidence to support either the conclusion that philosophical grammar was wedded to a Latin model, or the conclusion that its actual hypothesis concerning the uniformity of underlying grammatical relations has been brought into question by modern work.
In general, it should be noted that Bloomfield’s account of premodern linguistics is not reliable. His historical survey consists of a few haphazard remarks that, he asserts, summarize “what eighteenth century scholars knew about language.” These remarks are not always accurate (as, for example, his astonishing assertion that prior to the nineteenth century linguists “had not observed the sounds of speech, and confused them with the written symbols of the alphabet” or that the writers of general grammars regarded Latin as supreme in embodying the “universal canons of logic”); and, where accurate, they give little indication of the character of what was done in this period.
The manner in which the sounds of speech were analyzed in this period deserves a separate discussion; it is quite arbitrary to exclude this topic from the present survey, as I have done. Most of the works discussed here, and many others, contain discussions of phonetics, and the Aristotelian dictum that “spoken words are the symbols of mental experience and written words are the symbols of spoken words” (De Interpretation, 1) is apparently accepted with no discussion. There are a few modern references to the phonetics of this period. For example, M Grammont comments on the phonetics in Cordemoy, op. cit., in the following terms: “. . .the articulations of a certain number of French phonemes are described with remarkable clarity and precision” (Traité de phonétique (Paris: Librairie Delagrave, 1933), 4th edn.(1950), p. 13n.; he goes on to observe that: “These are the descriptions that Molière reproduced word for word in Le Bourgeois gentilhomme, acte II, scène 6 (1670)”). [Chomsky developed his views of phonology and phonetics during the late 1950s and early 1960s with his colleague Morris Halle; see Chomsky and Halle 1968. Like his view of ‘meanings’ (LFs or semantic representations), Chomsky’s view of linguistic sounds is that they are “in the head.” See in this regard Chomsky 2000, which is a collection of his more recent works on language and mind.]
84 Grammaire générale, pp. 340f. Bentham suggests a similar analysis (Works, p. 356).
85 A distinction between the “ideas principally expressed” by a linguistic form and the “accessory ideas” associated with it is developed in the Port-Royal Logic, chaps. XIV, XV. His principal idea is what is stated by the “lexical definition,” which attempts to formulate in a precise way the “truth of usage.” But the lexical definition cannot “reflect the whole impression the defined word makes on the mind,” and “it often happens that a word excites in our minds, besides the principal idea which we regard as the proper meaning of the word, other ideas – ideas which we may call accessory ideas and to which though we receive their impression we do not explicitly attend” (p. 90). For example, the principal meaning of you lied is that you knew that the opposite of what you said is true. “But in addition to this principal meaning, these words convey an idea of contempt and outrage which suggest that the speaker would not hesitate to harm you – a suggestion which renders his words both offensive and injurious.” Similarly, Virgil’s line To die, is that such a wretched thing? (Usque adeone mori miserum est?) has the same principal meaning as It is not so very wretched to die (Non est usque adeo mori miserum), but the original “expresses not only the bare thought that death is not so bad a thing as one supposes but suggests as well the image of a man who challenges death and looks it fearless in the face” (pp. 91–92). Accessory ideas may be “permanently attached to words,” as in the cases just mentioned, or they may be attached only in a particular utterance, for example, by gesture or tone of voice (p. 90). The association may, in other words, be a matter either of langue or parole.
The distinction is rather like that of cognitive and emotive meaning. Also relevant to contemporary issues is the example (p. 91) of how certain grammatical processes may change the accessory ideas expressed, without modification of principal meaning; thus, so it is claimed, to accuse someone of ignorance or deceit is different from calling him ignorant and deceitful, since the adjectival forms “express, in addition to the idea of particular shortcomings, an idea of contempt, whereas the nouns mean only the particular lack with no accompanying condemnation.”
86 C. Buffier, Grammaire françoise sur un plan nouveau (1709), cited by Sahlin, César Chesneau, pp. 121–122, with typical modern disparagement based, once again, on the assumption that surface structure alone is a proper object of study. See J. Katz and P. Postal, An Integrated Theory of Linguistic Descriptions, §§4.2.3, 4.2.4, for development and justification of a very similar idea.
87 “De la construction grammaticale,” Logique et Principes de Grammaire, p. 229.
88 The Latin example suggests a variety of problems, however. For some remarks on the phenomenon of so-called “free word order,” within the present context, see Chomsky, Aspects of the Theory of Syntax, chap. II, §4.4. [The phenomenon of “case marking” in different languages has proven to be a particularly interesting issue for the linguist constructing a Universal Grammar. See Chomsky, 1986, 1995b, and references provided there.]
89 It is not entirely clear from the context whether these conditions on transformations are regarded as matters of langue or parole, as conditions on a grammar or on the usage of a grammar; nor is it clear whether, within the framework that Du Marsais accepts, this question can be sensibly raised.
The account of sentence interpretation given by Du Marsais can be profitably compared with that proposed by Katz, Fodor, and Postal in recent work. See Katz and Postal, op. cit., and references cited there. [See also the references in the bracketed addition to Chomsky’s note 80.]
90 The examples that I give here are cited by Sahlin as indicative of the ridiculous character of Du Marsais’s theory, concerning which “it would be unjust to confront it with modern science so as to reveal the altogether obvious errors in it” (Sahlin, César Chesneau, p. 84).
91 T. Reid, Essays on the Intellectual Powers of Man (1785). For some remarks and quotations, see Chomsky, Aspects of the Theory of Syntax, pp. 199–200.
92 Except to the extent indicated by the final example, the analysis of indefinite articles. Such attempts to go beyond surface form are tolerated by modern linguistic theory and have been the subject of much methodological discussion during the 1940s, particularly in the United States.
