Creative aspect of language use

We have observed that the study of the creative aspect of language use develops from the assumption that linguistic and mental processes are virtually identical, language providing the primary means for free expression of thought and feeling, as well as for the functioning of the creative imagination. Similarly, much of the substantive discussion of grammar, throughout the development of what we have been calling “Cartesian linguistics,” derives from this assump- tion. The Port-Royal Grammar, for example, begins the discussion of syntax with the observation that there are “three operations of our minds: conceiving, judging, and reasoning” (p. 27), of which the third is irrelevant to grammar (it is taken up in the Port-Royal Logic, which appeared two years later, in 1662). From the manner in which concepts are combined in judgments, the Grammar deduces what it takes to be the general form of any possible grammar, and it proceeds to elaborate this universal underlying structure from a consideration of “the natural manner in which we express our thoughts” (p. 30).64 Most sub- sequent attempts to develop a schema of universal grammar proceed along the same lines. James Harris’s Hermes, which does not bear the imprint of the Port-Royal Grammar to the extent usual in eighteenth-century work, also reasons from the structure of mental processes to the structure of language, but in a somewhat different way. In general, he maintains, when a man speaks, “his Speech or Discourse is a publishing of some Energie or Motion of his soul” (p. 223).65 The “powers of the soul” are of two general types: perception (involving the senses and the intellect) and volition (the will, passions, appetites –” all that moves to Action whether rational or irrational” (p. 224). It follows that there are two kinds of linguistic acts: to assert, that is, “to publish some Perception either of the Senses or the Intellect”; or to “publish volitions,” that is, to interrogate, com- mand, pray, or wish (p. 224). The first type of sentence serves “to declare ourselves to others”; the second, to induce others to fulfill a need. Continuing in this way, we can analyze the volitional sentences in terms of whether the need is “to have some perception informed” or “some volition gratified” (the inter- rogative and requisitive modes, respectively); the requisitive is further analyzed as imperative or precative, depending on whether the sentence is addressed to

inferiors or non-inferiors). Since both interrogatives and requisitives serve “to answer to a need,” both types “require a return” – a return in words or deeds, to the requisitive, and in words alone, to the interrogative (p. 293f.).66 Thus the framework for the analysis of types of sentences is provided by a certain analysis of mental processes. Pursuing the fundamental distinction between body and mind, Cartesian linguistics characteristically assumes that language has two aspects. In partic- ular, one may study a linguistic sign from the point of view of the sounds that constitute it and the characters that represent these signs or from the point of view of their “signification,” that is, “the manner in which men use them for signifying their thoughts” (Port-Royal Grammar, p. 41). Cordemoy announces his goal in similar terms: “in this discourse I make a precise survey of everything that speech [la Parole] derives from the soul and everything it borrows from the body” (Discours physique de la parole, Preface). Similarly, Lamy begins his rhetoric by distinguishing between “the soul of words” (that is, “their mental [spiritual] aspect,” “what is particular to us” – the capacity of expressing “our ideas”) from “their body” – “their corporeal aspect,” “what the birds that imitate the voices of men have in common with us,” namely, “the sounds, which are signs of their ideas” (De L’art de parler). In short, language has an inner and an outer aspect. A sentence can be studied from the point of view of how it expresses a thought or from the point of view of its physical shape, that is, from the point of view of either semantic interpreta- tion or phonetic interpretation. Using some recent terminology, we can distinguish the “deep structure” of a sentence from its “surface structure.” The former is the underlying abstract struc- ture that determines its semantic interpretation; the latter, the superficial organiza- tion of units which determines the phonetic interpretation and which relates to the physical form of the actual utterance, to its perceived or intended form. In these terms, we can formulate a second fundamental conclusion of Cartesian linguistics, namely, that deep and surface structures need not be identical. The underlying organization of a sentence relevant to semantic interpretation is not necessarily revealed by the actual arrangement and phrasing of its given components. This point is brought out with particular clarity in the Port-Royal Grammar, in which a Cartesian approach to language is developed for the first time, with considerable insight and subtlety.67 The principal form of thought (but not the only one – cf. p. 79 below) is the judgment, in which something is affirmed of something else. Its linguistic expression is the proposition, the two terms of which are the “subject, which is that of which one affirms” and the “predicate, which is that which is predicated” (p. 29; PRG 67). The subject and the attribute may be simple, as in “Earth is round”, or complex [composé], as in “An able magistrate is a man useful to the republic” or “Invisible God created the visible world.” Furthermore, in such cases as these, the propositions

contain, at least in our mind, several judgments, from which one can make as many propositions. Thus, for example, when I say “Invisible God created the visible world” three judgments that pass through my mind are included in this proposition. For I judge: (1) that God is invisible; (2) that He created the world; (3) that the world is visible; and of these three propositions, the second is the principal and essential one of the original proposition. But the first and the third are only subordinate, and comprise only part of the principal proposition – the first composing its subject, the third its predicate. (p. 68; PRG 99–100) In other words, the deep structure underlying the proposition “Invisible God created the visible world” consists of three abstract propositions, each express- ing a certain simple judgment, although its surface form expresses only the subject-attribute structure. Of course, this deep structure is implicit only; it is not expressed but is only represented in the mind: Now these subordinate propositions are often in our mind, without being expressed in words, as in the example cited [viz. “Invisible God created the visible world”]. (p. 68; PRG 100) It is sometimes possible to express the deep structure in a more explicit way, in the surface form, “as when I reduce the above example to these terms: ‘God WHO is invisible created the world WHICH is visible’” (pp. 68–69; PRG 100). But it constitutes an underlying mental reality – a mental accompaniment to the utterance – whether or not the surface form of the utterance that is produced corresponds to it in a simple, point-by-point manner. In general, constructions of a noun with a noun in apposition, an adjective, or a participle are based on a deep structure containing a relative clause: “all these modes of speech include the relative pronoun in their meaning, and may be resolved by using it” (p. 69; PRG 100). The same deep structure may be realized differently in different languages, as when Latin has “video canem currentem,” and French “Je voy un chien qui court” (pp. 69–70; PRG 100). The position of the relative pronoun in the “subordinate proposition” [proposition incidente] is determined by a rule that converts deep structure to surface structure. We see this, for example, in such phrases as “God whom I love” and “God by whom the world has been created.” In such cases, the relative pronoun is always placed at the beginning of the proposition (although, according to the meaning it ought to be only at the end), unless it is governed by a preposition, for the preposition comes first, at least ordinarily. (p. 71; PRG 101) In the case of each of the sentences just discussed, the deep structure consists of a system of propositions, and it does not receive a direct, point- by-point expression in the actual physical object that is produced. To form an actual sentence from such an underlying system of elementary propositions, we

apply certain rules (in modern terms, grammatical transformations). In these examples, we apply the rule preposing the relative pronoun that takes the place of the noun of the incident proposition (along with the preposition that precedes it, if there is one). We may then, optionally, go on to delete the relative pronoun, at the same time deleting the copula (as in “Dieu invisible”) or changing the form of the verb (as in “canis currens”). Finally, we must, in certain cases, interchange the order of the noun and the adjective (as in “un habile magistrat”).68 The deep structure that expresses the meaning is common to all languages, so it is claimed, being a simple reflection of the forms of thought. The transforma- tional rules that convert deep to surface structure may differ from language to language. The surface structure resulting from these transformations does not directly express the meaning relations of the words, of course, except in the simplest cases. It is the deep structure underlying the actual utterance, a structure that is purely mental, that conveys the semantic content of the sen- tence. This deep structure is, nevertheless, related to actual sentences in that each of its component abstract propositions (in the cases just discussed) could be directly realized as a simple propositional judgment. The theory of essential and incident propositions as constituent elements of deep structure is extended in the Port-Royal Logic69 with a more detailed analysis of relative clauses. There, a distinction is developed between explica- tive (nonrestrictive or appositive) and determinative (restrictive) relative clauses. The distinction is based on a prior analysis of the “comprehension” and “extension” of “universal ideas,”70 in modern terms, an analysis of meaning and reference. The comprehension of an idea is the set of essential attributes that define it, together with whatever can be deduced from them; its extension is the set of objects that it denotes: The comprehension of an idea is the constituent parts which make up the idea, none of which can be removed without destroying the idea. For example, the idea of a triangle is made up of the idea of having three sides, the idea of having three angles, and the idea of having angles whose sum is equal to two right angles, and so on. The extension of an idea is the objects to which the word expressing the idea can be applied. The objects which belong to the extension of an idea are called the inferiors of that idea, which with respect to them is called the superior. Thus, the general idea of triangle has in its extension triangles of all kinds whatsoever. (p. 51; PRL 39–40) In terms of these notions, we can distinguish such “explications” as “Paris, which is the largest city in Europe” and “man, who is mortal” from “determi- nations” such as “transparent bodies, wise men” or “a body which is trans- parent, men who are pious” (pp. 59–60, 118; PRL 44–45, 89): A complex expression is a mere explication if either (1) the idea expressed by the complex expression is already contained in the comprehension of the idea expressed

by the principal word of the complex expression, or (2) the idea expressed by the complex expression is the idea of some accidental characteristic of all the inferiors of an idea expressed by the principal word. (pp. 59–60; PRL 45) A complex expression is a determination if the extension of the idea expressed by the complex term is less than the extension of the idea expressed by the principal word. (p. 60; PRL 45) In the case of an explicative relative clause, the underlying deep structure actually implies the judgment expressed by this clause, when its relative pronoun is replaced by its antecedent. For example, the sentence “men, who were created to know and love God, …” implies that men were created to know and love God. Thus an explicative relative clause has the essential properties of conjunction. But in the case of a restrictive relative clause (a determination), this is obviously not true. Thus in saying “Men who are pious are charitable,” we do not affirm either that men are pious or that men are charitable. In stating this proposition, we form a complex idea by joining together two simple ideas – the idea of man and the idea of piety – and we judge that the attribute of being charitable is part of this complex idea. Thus the subordinate clause asserts nothing more than that the idea of piety is not incompatible with the idea of man. Having made this judgment we then consider what idea can be affirmed of this complex idea of pious man. (p. 119; PRL 89–90) Similarly, consider the expression “The doctrine which identifies the sovereign good with the sensual pleasure of the body, which was taught by Epicurus, is unworthy of a philosopher.”71 This contains the subject “The doctrine which … taught by Epicurus” and the predicate “unworthy of a philosopher.” The subject is complex, containing the restrictive relative clause “which identifies the sovereign good with the sensual pleasure of the body” and the explicative relative clause “which was taught by Epicurus.” The relative pronoun in the latter has as its antecedent the complex expression “the doctrine which identifies the sovereign good with the sensual pleasure of the body.” Since the clause “which was taught by Epicurus” is explicative, the original sentence does imply that the doctrine in question was taught by Epicurus. But the relative pronoun of the restrictive clause cannot be replaced by its antecedent, “the doctrine,” to form an assertion implied by the full sentence. Once again, the complex phrase containing the restrictive relative clause and its antecedent expresses a single complex idea formed from the two ideas of a doctrine and of identifying the sovereign good with the sensual pleasure of the body. All this information must be represented in the deep structure of the original sentence, according to the Port-Royal theory, and the semantic interpretation of this sentence must proceed in the manner just indicated, utilizing this information (pp. 119–120; PRL 90). A restrictive relative clause is based on a proposition, according to the Port-Royal theory, even though this proposition is not affirmed when the

relative clause is used in a complex expression. What is affirmed in an expres- sion such as men who are pious, as noted above, is no more than the compat- ibility of the constituent ideas. Hence in the expression “minds which are square are more solid than those which are round,” we may correctly say that the relative clause is “false,” in a certain sense, since “the idea of being square” is not compatible with “the idea of mind understood as the principle of thought” (p. 124; PRL 93). Thus sentences containing explicative as well as restrictive relative clauses are based on systems of propositions (that is, abstract objects constituting the meanings of sentences);72 but the manner of interconnection is different in the case of an explicative clause, in which the underlying judgment is actually affirmed, and a determinative clause, in which the proposition formed by replacing the relative pronoun by its antecedent is not affirmed but rather constitutes a single complex idea together with this noun. These observations are surely correct, in essence, and must be accommodated in any syntactic theory that attempts to make the notion “deep structure” precise and to formulate and investigate the principles that relate deep structure to surface organization. In short, these observations must be accommodated in some fashion in any theory of transformational generative grammar. Such a theory is concerned precisely with the rules that specify deep structures and relate them to surface structures and with the rules of semantic and phonological interpretation that apply to deep and surface structures respectively. It is, in other words, in large measure an elaboration and formalization of notions that are implicit and in part expressly formulated in such passages as those just discussed. In many respects it seems to me quite accurate, then, to regard the theory of transformational generative grammar, as it is developing in current work, as essentially a modern and more explicit version of the Port-Royal theory. In the Port-Royal theory, the relative pronoun that occurs in the surface form does not always have the dual function of standing for a noun and connecting propositions. It may be “shorn of its pronominal nature” and may thus serve only the latter role. For example, in such sentences as “I suppose that you will be wise” and “I tell you that you are wrong” we find that, in the deep structure, “these propositions, ‘you will be wise’, ‘you are wrong’, make up only part of the whole propositions ‘I suppose …’ and ‘I tell you …’” (Grammaire, p. 73; PRG 104–105).73 The Grammar goes on to argue that infinitival constructions play the same role in the verbal system that relative clauses play in the nominal system, providing a means for extending the verbal system through the incorporation of whole propositions: “the infinitive is, among the other moods of the verb, what the relative is among the other pronouns” (pp. 111–112; PRG 139); like the relative pronoun, “the infinitive has, over and above the affirmation of the verb,

the power to join the proposition in which it appears to another proposition” (p. 112; PRG 139). Thus the meaning of “scio malum esse fugiendum” is conveyed by a deep structure based on the two propositions expressed by the sentences “scio” and “malum est fugiendum.” The transformational rule (in modern terms) that forms the surface structure of the sentence replaces “est” by “esse”, just as the transformations that form such sentences as “Dieu (qui est) invisible a créé le monde (qui est) visible” perform various operations of substitution, reordering, and deletion on the underlying systems of propositions. “And from this has come the fact that in French we almost always render the infinitive by the indicative of the verb and the particle ‘que’: ‘Je sais que le mal est à fuir’” (p. 112; PRG 140). In this case, the identity of deep structure in Latin and French may be somewhat obscured by the fact that the two languages use slightly different transformational operations to derive the surface forms. The Grammar goes on to point out that indirect discourse can be analyzed in a similar way.74 If the underlying embedded proposition is interrogative, it is the particle “if” rather than “that” that is introduced by the transformational rule, as in “They asked me if I could do that,” where the “discourse which is reported” is “Can you do that?” Sometimes, in fact, no particle need be added, a change of person being sufficient, as in “He asked me: Who are you?” as compared with “He asked me who I was” (p. 113; PRG 140–141). Summarizing the Port-Royal theory in its major outlines, a sentence has an inner mental aspect (a deep structure that conveys its meaning) and an outer, physical aspect as a sound sequence. Its surface analysis into phrases may not indicate the significant connections of the deep structure by any formal mark or by the actual arrangement of words. The deep structure is, however, represented in the mind as the physical utterance is produced. The deep structure consists of a system of propositions, organized in various ways. The elementary proposi- tions that constitute the deep structure are of the subject-predicate form, with simple subjects and predicates (i.e., categories instead of more complex phrases). Many of these elementary objects can be independently realized as sentences. It is not true, in general, that the elementary judgments constituting the deep structure are affirmed when the sentence that it underlies is produced; explicative and determinative relatives, for example, differ in this respect. To actually produce a sentence from the deep structure that conveys the thought that it expresses, it is necessary to apply rules of transformation that rearrange, replace, or delete items of the sentence. Some of these are obligatory, further ones optional. Thus “God, who is invisible, created the world, which is visible” is distinguished from its paraphrase, “Invisible God created the visible world,” by an optional deletion operation, but the transformation that substitutes a relative pronoun for the noun and then preposes the pronoun is obligatory. This account covers only the sentences based exclusively on judgments. But these, although the principal form of thought, do not exhaust the “operations of

our minds,” and “one must still relate to what occurs in our mind the conjunc- tions, disjunctions, and other similar operations of our minds, and all the other movements of our souls, such as desires, commands, questions, etc.” (p. 29; PRG 67). In part, these other “forms of thought” are signified by special particles such as “and,” “not,” “or,” “if,” “therefore,” etc. (pp. 137–138; PRG 168). But with respect to these sentence types as well, an identity of deep structure may be masked through divergence of the transformational means whereby actual sentences are formed, corresponding to intended meanings. A case in point is interrogation. In Latin, the interrogative particle ne “has no object outside the mind, but only marks the movement of the soul, by which we wish to know a thing” (p. 138; PRG 168). As for the interrogative pronoun, “it is nothing more than a pronoun to which the signification of ‘ne’ is added; that is to say, which, beyond taking the place of a noun like the other pronouns, further marks this movement of the soul which desires to know something and which demands to be instructed about it” (p. 138; PRG 168). But this “movement of the soul” can be signified in various ways other than by the addition of a particle, for example, by vocal inflection or inversion of word order, as in French, where the pronominal subject is “transported” to the position following the person marker of the verb (preserving the agreement of the underlying form). These are all devices for realizing the same deep structure (pp. 138–139; PRG 168–169). Notice that the theory of deep and surface structure as developed in the Port-Royal linguistic studies implicitly contains recursive devices and thus provides for infinite use of the finite means that it disposes, as any adequate theory of language must. We see, moreover, that, in the examples given, the recursive devices meet certain formal conditions that have no a priori necessity. In both the trivial cases (e.g., conjunction, disjunction, etc.) and the more interesting ones discussed in connection with relatives and infinitives, the only method for extending deep structures is by adding full propositions of a basic subject-predicate form. The transformational rules of deletion, rearrange- ment, etc., do not play a role in the creation of new structures. The extent to which the Port-Royal grammarians may have been aware of or interested in these properties of their theory is, of course, an open question. In modern terms, we may formalize this view by describing the syntax of a language in terms of two systems of rules: a base system that generates deep structures and a transformational system that maps these into surface structures. The base system consists of rules that generate the underlying grammatical relations with an abstract order (the rewriting rules of a phrase-structure gram- mar); the transformational system consists of rules of deletion, rearrangement, adjunction, and so on. The base rules allow for the introduction of new propositions (that is, there are rewriting rules of the form A → …S…, where S is the initial symbol of the phrase-structure grammar that constitutes the base);

there are no other recursive devices. Among the transformations are those which form questions, imperatives, etc., when the deep structure so indicates (i.e., when the deep structure represents the corresponding “mental act” in an appro- priate notation).75 The Port-Royal grammar is apparently the first to develop the notion of phrase structure in any fairly clear way.76 It is interesting, therefore, to notice that it also states quite clearly the inadequacy of phrase-structure description for the representation of syntactic structure and that it hints at a form of transforma- tional grammar in many respects akin to that which is being actively studied today. Turning from the general conception of grammatical structure to specific cases of grammatical analysis, we find many other attempts in the Port-Royal Grammar to develop the theory of deep and surface structure. Thus adverbs are analyzed as (for the most part) arising from “the desire that men have to abbreviate discourse,” thus as being elliptical forms of preposition-noun constructions, for example, “wisely” for “with wisdom” or “today” for “on this day” (p. 88; PRG 121). Similarly, verbs are analyzed as containing implicitly an underlying copula that expresses affirmation; thus, once again, as arising from the desire to abbreviate the actual expression of thought. The verb, then, is “a word whose principal use77 is to signify affirma- tion or assertion, that is, to indicate that the discourse where this word is employed is the discourse of a man who not only conceives things, but who judges and affirms them” (p. 90; PRG 122). To use a verb, then, is to perform the act of affirming, not simply to refer to affirmation as an “object of our thought,” as in the use of “a number of nouns which also mean affirmation, such as ‘affirmans’ and ‘affirmatio’” (p. 90; PRG 122). Thus the Latin sentence “Petrus vivit” has the meaning “Peter is living” (p. 90; PRG 123), and in the sentence “Petrus affirmat” “‘affirmat’ is the same as ‘est affirmans’” (p. 98; PRG 128). It follows, then, that in the sentence “Affirmo” (in which subject, copula, and attribute are all abbreviated in a single word), two affirmations are expressed: one regarding the act of the speaker in affirming, the other the affirmation that he attributes (to himself, in this case). Similarly, “the verb ‘nego’ … contains an affirmation and a negation” (p. 98; PRG 128).78 Formulating these observations in the framework outlined above, what the Port-Royal grammarians are maintaining is that the deep structure underlying a sentence such as “Peter lives” or “God loves mankind” (Logic, p. 108; PRL 83) contains a copula, expressing the affirmation, and a predicate (“living,” “loving mankind”) attributed to the subject of the proposition. Verbs constitute a sub- category of predicates; they are subject to a transformation that causes them to coalesce with the copula into a single word. The analysis of verbs is extended in the Logic, where it is maintained (p. 117) that, despite surface appearances, a sentence with a transitive verb and its object

“expresses a complex proposition and in one sense two propositions.” Thus we can contradict the sentence “Brutus killed a tyrant” by saying that Brutus did not kill anyone or that the person whom Brutus killed was not a tyrant. It follows that the sentence expresses the proposition that Brutus killed someone who was a tyrant, and the deep structure must reflect this fact. It seems that this analysis would also apply, in the view of the Logic, if the object is a singular term; e.g., “Brutus killed Caesar.” This analysis plays a role in the theory of reasoning developed later on in the Logic. It is used to develop what is in effect a partial theory of relations, permitting the theory of the syllogism to be extended to arguments to which it would otherwise not apply. Thus it is pointed out (pp. 206–207; PRL 159–160) that the inference from “The divine law commands us to honor kings” and “Louis XIV is a king” to “The divine law commands us to honor Louis XIV” is obviously valid, though it does not exemplify any valid figure as it stands, superficially. By regarding “kings” as “the subject of a sentence contained implicitly in the original sentence,” using the passive transformation79 and otherwise decomposing the original sentence into its underlying prepositional constituents, we can finally reduce the argument to the valid figure Barbara. Reduction of sentences to underlying deep structure is resorted to elsewhere in the Logic, for the same purpose. For example, Arnauld observes (p. 208; PRL 160) that the sentence There are few pastors nowadays ready to give their lives for their sheep, though superficially affirmative in form, actually “contains implicitly the negative sentence ‘Many pastors nowadays are not ready to give their lives for their sheep.” In general, he points out repeatedly that what is affirmative or negative “in appearance” may or may not be in meaning, that is, in deep structure. In short, the real “logical form” of a sentence may be quite different from its surface grammatical form.80 The identity of deep structure underlying a variety of surface forms in different languages is frequently stressed, throughout this period, in connection with the problem of how the significant semantic connections among the elements of speech are expressed. Chapter VI of the Port-Royal Grammar considers the expression of these relations in case systems, as in the classical languages, or by internal modification, as in the construct state in Hebrew, or by particles, as in the vernacular languages, or simply by a fixed word order,81 as in the case of the subject–verb and verb–object relations in French. These are regarded as all being manifestations of an underlying structure common to all these languages and mirroring the structure of thought. Similarly, Lamy com- ments in his rhetoric on the diverse means used by various languages to express the “relations, and the consequence and interconnexion between all the ideas that the consideration of things excites in our mind” (De L’Art de Parler, pp. 10–11). The encyclopedist Du Marsais also stresses the fact that case systems express relations among the elements of discourse that are, in other

languages, expressed by word order or specific particles, and he points out the correlation between freedom to transpose and wealth of inflection.82 Notice that what is assumed is the existence of a uniform set of relations into which words can enter, in any language, these corresponding to the exigencies of thought. The philosophical grammarians do not try to show that all languages literally have case systems, that they use inflectional devices to express these relations. On the contrary, they repeatedly stress that a case system is only one device for expressing these relations. Occasionally, they point out that case names can be assigned to these relations as a pedagogic device; they also argue that considerations of simplicity sometimes may lead to a distinction of cases even where there is no difference in form. The fact that French has no case system is in fact noted in the earliest grammars. Cf. Sahlin, p. 212. It is important to realize that the use of the names of classical cases for languages with no inflections implies only a belief in the uniformity of the grammatical relations involved, a belief that deep structures are fundamentally the same across languages, although the means for their expression may be quite diverse. This claim is not obviously true – it is, in other words, a nontrivial hypothesis. So far as I know, however, modern linguistics offers no data that challenge it in any serious way.83 As noted above, the Port-Royal theory of grammar holds that for the most part, adverbs do not, properly speaking, constitute a category of deep structure but function only “for signifying in a single word what could otherwise be indicated only by a preposition and a noun” (p. 88; PRG 121). Later grammar- ians simply drop the qualification to “for the most part.” Thus for Du Marsais, “what distinguishes adverbs from other kinds of words is that adverbs have the value of a preposition and a noun, or a preposition with its complement: they are words which abbreviate” (p. 660). This is an unqualified characterization, and he goes on to analyze a large class of items in this way – in our paraphrase, as deriving from a deep structure of the form: preposition–complement. This analysis is carried still further by Beauzée.84 He, incidentally, maintains that, although an “adverbial phrase” such as “with wisdom” does not differ from the corresponding adverb “wisely” in its “signification,” it may differ in the “accessory ideas” associated with it: “when it is a matter of contrasting an action with a habit, the adverb is more appropriate for indicating the habit and the adverbial phrase for indicating the action; thus I would say ‘A man who conducts himself wisely cannot promise that all his actions will be performed with wisdom’” (p. 342).85 This distinction is a particular case of “the antipathy that all languages naturally show towards a total synonymity, which would enrich an idiom only with sounds that do not subserve accuracy and clarity of expression.” Earlier grammarians provide additional instances of analysis in terms of deep structure, as, for example, when imperatives and interrogatives are analyzed as,

in effect, elliptical transforms of underlying expressions with such supplemen- tary terms as “I order you …” or “I request…”86 Thus “Come see me” has the deep structure “I order/beg you to come see me”; “Who found it?” has the meaning of “I ask who found it?” etc. Still another example that might be cited is the transformational derivation of expressions with conjoined terms from underlying sentences, in the obvious way; for example, in Beauzée, op. cit., pp. 399f. Beauzée’s discussion of conjunctions also provides somewhat more interesting cases, as, for example, when he analyzes “how” [comment] as based on an underlying form with “manner” [manière] and a relative clause, so that the sentence “I know how it happened” has the meaning of “I know the manner in which it happened”; or when he analyzes “the house which I acquired.” In this way, the underlying deep structure with its essential and incident propositions is revealed. An interesting further development, along these lines, is carried out by Du Marsais in his theory of construction and syntax.87 He proposes that the term “construction” be applied to “the arrangement of words in discourse,” and the term “syntax,” to “the relations which words bear to one another.” For example, the three sentences “accepi litteras tuas,” “tuas accepi litteras,” and “litteras accepi tuas” exhibit three different constructions, but they have the same syntax; the relations among the constituent elements are the same in all three cases. “Thus, each of these three arrangements produces the same meaning [sens] in the mind: ‘I have received your letter’.” He goes on to define “syntax” as “what brings it about, in every language, that words produce the meaning we wish to arise in the minds of those who know the language … the part of grammar that provides knowledge of the signs established in a language to produce understanding in the mind” (pp. 229–231). The syntax of an expression is thus essentially what we have called its deep structure; its construction is what we have called its surface structure.88 The general framework within which this distinction is developed is the following. An act of the mind is a single unit. For a child, the “idea” [sentiment] that sugar is sweet is at first an unanalyzed, single experience (p. 181); for the adult, the meaning of the sentence “Sugar is sweet,” the thought that it expresses, is also a single entity. Language provides an indispensable means for the analysis of these otherwise undifferentiated objects. It provides a means of clothing our thought, so to speak, of rendering it perceptible, of dividing it, of analyzing it – in a word, of making it such that it is communicable to others with more precision and detail. Thus, particular thoughts are each an ensemble, so to speak, a whole that the usage of language divides, analyzes and distributes into parts by means of different articulations of the speech organs which form the words. (p. 184) Similarly, the perception of speech is a matter of determining the unified and undifferentiated thought from the succession of words. “[The words] work together to produce the whole sense or the thought we wish to arise in the

minds of those who read or hear them” (p. 185). To determine this thought, the mind must first discover the relations among the words of the sentence, that is, its syntax; it must then determine the meaning, given a full account of this deep structure. The method of analysis used by the mind is to bring together those words that are related, thus establishing a “meaningful order” [ordre significatif] in which related elements are successive. The actual sentence may, in itself, have this “meaningful order,” in which case it is called a “simple construction (natural, necessary, meaningful, assertive)” (p. 232). Where it does not, this “meaningful order” must be reconstructed by some procedure of analysis – it must be “re-established by the mind, which grasps the meaning [sens] only by this order” (pp. 191–192). To understand a sentence of Latin, for example, you must reconstruct the “natural order” that the speaker has in his mind (p. 196). You must not only understand the meanings of each word, but, furthermore, you would not understand anything in it except by putting together in your mind the words in their relation to one another, and you can do this only after you have heard the whole sentence. (pp. 198–199) In Latin, for example, it is the “relative word-endings which makes us consider the words in the completed proposition in accordance with the order of their interrelations, and hence in accordance with the order of the simple, necessary and meaningful construction” (pp. 241–242). This “simple construction” is an “order which is always indicated, but rarely observed in the usual construction of languages whose nouns have cases” (p. 251). Reduction to the “simple construction” is an essential first step in speech perception: The words form a whole that has parts. The simple perception of the relations between these parts makes us conceive the whole of them, and comes to us solely from the simple construction. Setting forth the words in accordance with the order of succession of their relations, this presents them in a manner that is best fitted to make us recognize these relations and to make the whole thought arise. (pp. 287–288) Constructions other than the “simple constructions” (namely, “figurative con- structions” [constructions figurées]) are understood only because the mind corrects their irregularities, with the help of accessory ideas which make us conceive what we read and hear as if the sense were expressed in the order of the simple construction … (p. 292) In short, in the “simple construction” the relations of “syntax” are represented directly in the associations among successive words, and the undifferentiated thought expressed by the sentence is derived directly from this underlying representation, which is regarded, throughout, as common to all languages (and, typically, as corresponding to the usual order of French – cf., e.g., p. 193). The transformations which form a “figurative construction” effect reordering and ellipsis. The “fundamental principle of all syntax” (p. 218) is that reordering

and ellipsis must be recoverable by the mind of the hearer (cf. pp. 202, 210ff., 277); that is, they can be applied only when it is possible to recover uniquely “the strict metaphysical order” of the “simple construction.”89 Many examples of reduction to simple constructions are presented to illus- trate this theory.90 Thus the sentence “Who said it?” is reduced to the simple construction “The one who said it is which person?” (Sahlin, p. 93); the sentence “Being loved as much as lovers, you are not forced to shed tears” is reduced to “Since you are loved as much as you are lovers, …”; the sentence “It is better to be just than to be rich, to be reasonable than to be wise” is reduced to four underlying propositions, two negative, two positive, in the obvious way (p. 109), etc. A rather different sort of example of the distinction between deep and surface structure is provided by Du Marsais in his analysis (pp. 179–180) of such expressions as “I have an idea/fear/doubt,” etc. These, he says, should not be interpreted as analogous to the superficially similar expressions “I have a book/ diamond/watch,” in which the nouns are “names of real objects that exist independently of our thought [manière de penser].” In contrast, the verb in “I have an idea” is “a borrowed [empruntée] expression,” produced only “by imitation.” The meaning of “I have an idea” is simply “I am thinking” or “I am conceiving something in such-and-such a way.” Thus the grammar gives no license for supposing that such words as “idea,” “concept,” “image” stand for “real objects,” let alone “perceptible objects.” From this grammatical observa- tion it is only a short step to a criticism of the theory of ideas, in its Cartesian and empiricist forms, as based on a false grammatical analogy. This step is taken by Thomas Reid, shortly after.91 As Du Marsais indicates with abundant references, his theory of construction and syntax is foreshadowed in scholastic and renaissance grammar (see note 67). But he follows the Port-Royal grammarians in regarding the theory of deep and surface structure as, in essence, a psychological theory, not merely a means for the elucidation of given forms or for analysis of texts. As indicated above, it plays a role in his hypothetical account of the perception and production of speech, just as, in the Port-Royal Grammar, the deep structure is said to be represented “in the mind” as the utterance is heard or produced. As a final example of the attempt to discover the hidden regularities under- lying surface variety, we may mention the analysis of French indefinite articles in Chapter VII of the Port-Royal Grammar, where it is argued, on grounds of symmetry of patterning, that de and des play the role of the plural of un, as in Un crime si horrible mérite la mort, Des crimes si horribles méritent la mort, De si horribles crimes méritent la mort, etc. To handle the apparent exception, Il est coupable de crimes horribles (d’horribles crimes), they propose the “rule of cacophony” that a de de sequence is replaced by de. They also note the use of des as a realization of the definite article, and other uses of these forms.

Perhaps these comments and examples are sufficient to suggest something of the range and character of the grammatical theories of the “philosophical grammarians.” As noted above, their theory of deep and surface structure relates directly to the problem of creativity of language use, discussed in the first part of the present work. From the standpoint of modern linguistic theory, this attempt to discover and characterize deep structure and to study the transformational rules that relate it to surface form is something of an absurdity;92 it indicates lack of respect for the “real language” (i.e., the surface form) and lack of concern for “linguistic fact.” Such criticism is based on a restriction of the domain of “linguistic fact” to physically identifiable subparts of actual utterances and their formally marked relations.93 Restricted in this way, linguistics studies the use of language for the expression of thought only incidentally, to the quite limited extent to which deep and surface structure coincide; in particular, it studies “sound-meaning corre- spondences” only in so far as they are representable in terms of surface structure. From this limitation follows the general disparagement of Cartesian and earlier linguistics,94 which attempted to give a full account of deep structure even where it is not correlated in strict point-by-point fashion to observable features of speech. These traditional attempts to deal with the organization of semantic content as well as the organization of sound were defective in many ways, but modern critique generally rejects them more for their scope than for their failures.