THE RENAISSANCE AND ENLIGHTENMENT

Semantics, Logic, and Epistemology

As the Middle Ages gave way to the Renaissance in the late fifteenth century, logic (on which semantics had been centered) first lost its medieval attainments and then subsided into inactivity until the middle of the nineteenth century. What little there was in the way of logical inquiry from about 1450 to about 1850 was carried on under the view of logic as the art (or science) of reason, the idea of scientia sermocinalis having been ridiculed into oblivion by the Renaissance humanists. Aside from the work of late Scholastics, such as the Ars Logica of John of St. Thomas (1589–1644), and an occasional deliberate attempt at revival, such as the Logica Fundamentis Suis a Quibus Hactenus Collapsa Fuerat Restituta (1662) of Arnold Geulincx (1624–1669), there were no further developments of the logicosemantic theories of the logica moderna.

Philosophers retained their interest in semantics, however, after losing interest in and even all knowledge of the kind of logic with which it had been associated. Epistemology dominated the philosophy of the Renaissance and the Enlightenment (for present purposes, roughly 1500–1800) as logic had dominated medieval philosophy, and the development of semantics during this period centered on epistemology. As a consequence, much of the development took place in the context of discussions of nonlinguistic signs, such as representative ideas, and will not be directly considered here.

Perhaps partly because logic had lost its identity as an inquiry into language, the interest of philosophers in language was more intense and diversified during this period (and especially in the eighteenth century) than at any earlier time. Some of this interest was manifested in widespread speculation about the origin of language and in projects for a universal language or a "real characteristic." Although works on these subjects are typical of the period and often contain material of value for the history of semantics, they can be considered here only as they bear directly on a theory of meaning or philosophy of language selected for discussion.

Bacon

Francis Bacon (1561–1626) produced comparatively little that can be described as philosophy of language, but the occasional novel insights and the programmatic character of what he did produce helped to give it a considerable influence over philosophy of language in the Enlightenment. Almost everything of his that is relevant to the history of semantics is to be found in the "Art of Elocution or Tradition" (in the Advancement of Learning and the De Augmentis Scientiarum ) and the doctrine of the "Idols of the Market Place" (in the Novum Organum and the De Augmentis Scientiarum ).

The first of these is plainly Bacon's revised version of the medieval trivium —grammar, logic, rhetoric—although he nowhere says so. In the later Middle Ages these subjects had sometimes been designated the artes sermocinales, and in the De Augmentis Scientiarum Bacon said that the subject matter of the ars tradendi was sermo. This inquiry into "tradition"—that is, discourse or communication—had three branches, concerning "the organ," "the method," and "the illustration" of tradition; and most of the work of the three branches was explicitly associated with grammar, logic, and rhetoric, respectively. For present purposes the first of these three branches is much more important than the other two.

In his scheme of "Human Philosophy" the Art of Tradition occurred as "the fourth kind of Rational Knowledge" (Spedding, Ellis, and Heath, eds., 3.383–4), because reason was "as it were the soul of discourse," according to Bacon. "Nevertheless, in treating of them reason and discourse ought to be separated, no less than soul and body" (1.651). He began his separate treatment of discourse by identifying speech and writing as the most familiar organs of discourse and stressing their connection with reason by citing with approval the traditional version of Aristotle's doctrine: "Words are the images of cogitations, and letters are the images of words" (3.399; but cf. 3.284, 3.85–86). But his interest in less familiar organs of discourse prompted him to frame a set of general conditions for an organ of discourse: "Whatever can be broken down into differences sufficiently numerous for explicating the variety of notions (provided those differences are perceptible to sense) can become a vehicle of cogitations from one man to another" (1.651; cf. 3.399). An organ of discourse can be used to communicate nothing but notions, but it will contain elements that express not only notions but also things.

In the most familiar arrangement of organs of discourse, words (by which Bacon meant only articulate sounds [2.411–412]) are expressed by letters—that is, phonograms. Letters, in turn, may be expressed by ciphers—that is, cryptograms—and both letters and ciphers may be designated "nominal characters." But he recognized another kind of "notes of things, which signify things without the aid or intervention of words," either "on the basis of congruity" or "arbitrarily." As examples of the former sort he cited hieroglyphics and gestures, gestures being "transitory hieroglyphics," the "words" for which hieroglyphics may be the "letters," and he classified them together as "emblems"—that is, sensible images to which intellectual conceptions could be reduced by analogy (1.652–653; 649). As examples of the latter sort he cited "real characters" such as Chinese ideograms, which "have nothing emblematic in them, but are simply surds, no less than the elements of letters themselves; … there ought to be as many of them as there are radical words" (1.653).

Despite that disadvantage, real characters could and, Bacon thought, did function as an organ of discourse beyond the limits of a single natural language just because they signified "things and notions" without the intervention of words (1.652). Although he was convinced that there were no more convenient organs of discourse than words and letters, Bacon listed the study of the notes of things among his desiderata (1.653). Acting on this suggestion, the Royal Society commissioned some of its members to look into the project of a universal real character, the eventual result being John Wilkins's Essay towards a Real Character and a Philosophical Language (1668), one of many such attempts during this period.

As another part of the inquiry into organs of discourse Bacon proposed a "philosophical grammar," and this desideratum likewise had an extensive but un-Baconian influence. Some of what he had to say about philosophical grammar was reminiscent of the medieval speculative grammar—for instance, it was to be "a kind of grammar that would carefully inquire not into the analogy of words to one another, but into the analogy between words and things or words and reason" (1.654)—and this is what seems to have caught the imagination of his many successors in the Enlightenment who produced works in philosophical or "universal" grammar. What Bacon really had in mind was probably something more nearly like the comparative philology characteristic of the nineteenth century: "But the noblest kind of grammar would, I think, result if someone well taught in many languages, learned as well as vulgar, would treat of the various properties of languages, showing in what respects each excels and in what respects it is deficient" (ibid.; cf. 3.230, 3.401). He did, however, go on to suggest that one might combine all the best properties uncovered in that analysis into "a very finely formed image and remarkable model of speech itself for expressing the mind's meanings aright" (1.654).

In his sketch of a philosophical grammar Bacon emphatically disapproved of what he believed Plato had been attempting in the Cratylus, an inquiry into "the imposition and original etymology of names" (ibid.; cf. 3.531), but his own concern in the doctrine of the "Idols of the Market Place" closely parallels Plato's real concern in the Cratylus, that is, distinguishing between correct and incorrect names. "The idols imposed on the understanding through words are of two kinds. Either they are names of things that are not (for just as there are things that lack a name because they have not been observed, so there are names that lack things, resulting from a fantastic supposition); or they are names of things that are, but confused, ill-defined, and rashly and irregularly abstracted from the things" (1.171). As an example of the first he gave "prime mover"; his example of the second kind was "humid," which, as his discussion of it shows, is less objectionable on these grounds now than it was in seventeenth-century English.

Hobbes

Thomas Hobbes (1588–1679) conceived of his systematic philosophy as beginning with an investigation into language and produced different versions of the investigation in Human Nature (1650), Chapters 5 and 13; Leviathan (1651), Chapters 4–7; and Elementa Philosophiae Sectio Prima: De Corpore (1655; English 1656), Part I, "Computatio Sive Logica." (The latest of those versions is also in most respects the fullest and is used as the basis of the following account.)

Philosophy, Hobbes observed, depends on ratiocination, or "computation" (Molesworth edition, 1.3). In reasoning regarding particular things "we add and subtract in our silent thoughts, without the use of words" (ibid.; see 3.32); but in most instances, and certainly in philosophizing, "men owe all their true ratiocination to the right understanding of speech " (1.36), such ratiocination being "nothing but reckoning, that is adding and subtracting, of the consequences of general names " (3.30). In the second chapter of his Logic, devoted specifically to "names," Hobbes produced a novel combination of several elements in the Aristotelian-Scholastic account of the semantics of names. Ratiocination of every kind depends on memory, and the intelligent use of memory requires what Hobbes called "marks, namely, sensible things taken at pleasure, that, by the sense of them, such thoughts may be recalled to our mind as are like those thoughts for which we took them" (1.14).

It is possible, Hobbes thought, for a man to "spend all his time partly in reasoning and partly in inventing marks for the help of his memory, and advancing himself in learning"—that is, to devise and profitably use a private language—but if science and philosophy are to develop, there must be "certain signs by which what one man finds out may be manifested and made known to others" (1.14). Signs that do "signify the cogitations and motions of our mind" are "words so and so connected," or what Hobbes called "speech, of which every part is a name " (1.15). The use of names as marks, he held, was logically prior to their use as signs, since "names, though standing singly by themselves, are marks …; but they cannot be signs otherwise than by being disposed and ordered in speech" (1.15). He recognized the syntactic disposition of names in speech as necessary but not sufficient for "declaring our conceptions to others." Speech cannot "perform that office alone without the help of many circumstances," such as "time, place, countenance, gesture, the counsel of the speaker" (2.274). We must "consider the drift, and occasion, and contexture of the speech, as well as the words themselves" (4.23).

When names are ordered in speech so as to be signs rather than marks, "it is manifest they are not signs of the things themselves" but signs only of our conceptions (1.17). Hobbes seems to have been following Aristotle's lead here, but more faithfully than most, since he went on to say, "That the sound of this word 'stone' should be the sign of a stone, cannot be understood in any sense but this, that he that hears it collects that he that pronounces it thinks of a stone" (1.17). Thus, even though indirectly and only in virtue of signifying that the speaker is thinking of a stone, the name "stone" ordered in speech is a sign of a stone. At any event, Hobbes nowhere suggested that "stone" occurring in speech was a name of some mental entity. On the contrary, in going on to show that "it is not at all necessary that every name should be a name of some thing," Hobbes began by pointing out that " 'man,' 'tree,' 'stone' are names of things themselves" (1.17), though they may be used as signs of our conceptions of men, trees, and stones and as names of "fictions and phantasms of things," such as images in dreams. "Moreover, that which neither is, nor has been, nor ever shall, or ever can be, has a name, namely, 'that which neither is, nor has been,' &c.; or more briefly this, 'impossible'" (1.17). For "a name is not taken in philosophy, as in grammar, for one single word, but for any number of words put together to signify one thing" (1.23), Hobbes having decided "to apply the word 'thing' to whatsoever we name; as if it were all one whether that thing be truly existent, or be only feigned" (1.18).

Much of Hobbes's investigation of names was presented in the form of discussions of traditional classifications of names. His treatment of them sometimes presents the half-understood remnants of complex medieval theories—for instance, his treatment of names of first and second intention (1.20–21)—but there are occasional interesting novelties as well. In his discussion of common and proper names he put forward his strict nominalism: "this word 'universal' is never the name of any thing existent in nature, nor of any idea or phantasm formed in the mind, but always the name of some word or name" (1.20); at another point he remarked that the univocal-equivocal distinction "belongs not so much to names as to those that use names" (1.23); and he based the distinction between simple and compound names not on appearances but on considerations of analyzability, so that in the context of a discussion of man "body" is a simple name while "man" is a "more compounded name," being equivalent to "animated rational body" (1.23–24).

Hobbes encountered important difficulties in his discussion of names of "certain and determined" and of "uncertain and undetermined" signification (1.21–23), which is evidently a badly distorted remnant of supposition theory. In the course of that discussion Hobbes was led to claim, for example, that particular names—such as "some man"—"are of uncertain signification, because the hearer knows not what thing it is the speaker would have him conceive" (1.22), as if the "uncertainty" in, say, "some man will marry my daughter" were the sort that could always be resolved by asking the speaker "which man?" Even worse confusion resulted from his attempt to show that such quantifiers as "every" and "some" were unnecessary for purposes of reasoning. Such words, he maintained, "which denote universality and particularity, are not names; so that 'every man' and 'that man which the hearer conceives in his mind' are all one; and 'some man' and 'that man which the speaker thought of' signify the same. From whence it is evident, that the use of signs of this kind, is not for a man's … getting of knowledge by his own private meditation (for every man has his own thoughts sufficiently determined without such helps as these) but … for the teaching and signifying our conceptions to others" (1.22).

In his treatment of propositions Hobbes sometimes spoke as if only such propositions as "Cicero is Tully" were true—for instance, "that proposition only is true in which are copulated two names of one and the same thing" (1.57)—but usually his description of a true proposition was more moderately and more accurately expressed along such lines as these: "A true proposition is that, whose predicate contains, or comprehends its subject, or whose predicate is a name of every thing, of which the subject is a name" (1.35; cf. 4.23–24). He produced a detailed analysis of falsity as reducible to combinations of names of different sorts of entities (1.57–62). His truth theory was, however, quite radical in other respects. The "first truths," he claimed, "were arbitrarily made by those that first of all imposed names upon things, or received them from the imposition of others. For it is true [for example] that man is a living creature, but it is for this reason, that it pleased men to impose both those names on the same thing" (1.36). This suggests an identification of the proposition with a particular sequence of words, but Hobbes elsewhere gave the impression of having been on the point of drawing a clear distinction between propositions and the vehicles of their expression—for instance, "every proposition may be, and uses to be, pronounced and written in many forms.… And therefore, whensoever they [students of philosophy] meet with any obscure proposition, they ought to reduce it to its most simple and categorical form" (1.39).

Hobbes rejected the analysis of contingent categorical propositions into their corresponding hypothetical forms, pointing out that while this analysis was allowable for necessary categoricals, "in contingent propositions, though this be true, 'every crow is black,' yet this, 'if any thing be a crow, the same is black' [i.e., '(x )(Cx ⊃ Bx )'], is false" (ibid.). In several places Hobbes discussed the various uses of speech—for example, Human Nature, Chapter 13—and at one point argued against the notion that a promise simply by its form of words creates an obligation (2.18–20).

The Port-Royal Logic

René Descartes (1596–1650) had very little to say about language, but Antoine Arnauld (1612–1694) took an avowedly Cartesian approach to semantic questions in the Port-Royal Grammar (Grammaire générale et raisonnée, with Claude Lancelot, 1660) and the Port-Royal Logic (Logique ou l'art de penser, with Nicole, 1662). The latter book had a tremendous influence; it marks, better than any other, the abandonment of the medieval doctrine of an essential connection between logic and semantics. Disdain for medieval theories was emphatically expressed in it—"No one, thank God, is interested in … second intentions" (Premier Discours )—but at several points the theories were still employed (for instance, in 1.2 and 2.10) and elsewhere in the book they were supplanted by innovations that sometimes obscured what had been clear in the logica moderna.

In words reminiscent of Hobbes's on this point, Arnauld remarked that if logic considered only an individual's reflections on his ideas, the investigation of language would form no part of it. But we must use "exterior signs" for communication, "and since this custom is so strong that even when we think by ourselves things are presented to our mind only together with the words with which we are accustomed to adorn them in speaking to others, it is necessary in logic to consider the ideas joined to words and the words joined to ideas" (introduction; cf. Descartes, Principles, Part I, Principle 74). Arnauld of course argued (1.1) against Hobbes's anti-Cartesian suggestion that reasoning might be "nothing more than the uniting and stringing together of names or designations by the word 'is,'" so that all we could ever conclude is "whether or not there is a convention (arbitrarily made about their meanings) according to which we join these names together" (Objections to Descartes's Meditations, 3.4).

Signs and signification were frequently discussed in the Port-Royal Logic, sometimes with interesting results; the most fundamental questions were, however, treated with the kind of inattention to detail that came to characterize most of the many semantic theories of the Enlightenment. "The sign," said Arnauld, "comprises two ideas—one of the thing that represents, the other of the thing represented—and its nature consists in exciting the second by means of the first" (1.4). In the case of words, the "thing represented" was identified as a "thought" or an "idea." Even proper names were defined as those "that serve to mark … the ideas that represent only one single thing," and "general words" were said to be those "that are joined to universal and general ideas" (1.6). The doctrine is so far consistent and recognizably Cartesian, even if crude. But it is complicated, no doubt inadvertently, by many suggestions of a different sort of signification for words. Thus, on a single page Arnauld began by calling words "sounds that are intended to signify ideas," went on to speak of "things and modes" as "the objects of our thoughts," and ended by defining names as "the words intended to signify both things and modes " (2.1); and he nowhere provided an account that might justify this extended use of "signify." He may have been assuming a transitivity of signification—words signifying ideas representing things—but John Locke was the first to attempt to spell out such a theory.

Arnauld warned against the "great equivocation in the word 'arbitrary' when we say that the signification of words is arbitrary," pointing out that while "it is purely arbitrary to join one idea to one sound rather than to another," nevertheless the ideas, "at least those that are clear and distinct," are not arbitrary. The result of correct reasoning is "a solid, effective judgment regarding the nature of things based on the consideration of the ideas of them a man has in mind, which ideas it has pleased men to mark by means of certain names" (1.1). But, on the other hand, one of his reasons for rejecting Aristotle's categories was that they were "arbitrary names that form no clear and distinct idea in the mind" (1.3).

Arnauld also explicitly rejected Aristotle's definition of a verb, putting in its place one that not only captured the essence that Aristotle missed but was also much simpler: "a word that signifies affirmation" (2.2). Evidently he did not mean that it signified the idea of affirmation (as would the noun "affirmation"), but he did not work out the definition in a way that tied it to the rest of his signification theory.

Like so many other philosophers of the period, Arnauld believed that "the best means of avoiding the confusion of words to be found in ordinary languages is to make a new language and new words that would be attached only to the ideas we want them to represent." He differed from most, however, in suggesting that this be accomplished simply by a conscientious, systematic use of precise nominal definitions attached to already extant vocables of ordinary languages (1.12).

One of the more interesting notions in Arnauld's doctrine of signification was introduced in his observation that "it often happens that besides the principal idea (which is regarded as its proper signification) a word excites several other ideas that may be called accessory." Sometimes the accessory ideas are attached to the words "as the result of common usage," as in "you lied," the proper signification of which is the idea that you knew the contrary of what you said, the ideas of contempt and outrage being accessory (1.14). In some respects this is reminiscent of Augustine's doctrine in Principia Dialecticae, especially when Arnauld uses it to argue (against Cicero) that certain words may, in virtue of their accessory ideas, be described as unchaste (1.14). Accessory ideas may also be attached for the purpose of a single use of a word, and on that basis Arnauld attempted an explanation of the varying signification of the demonstrative pronoun "this," here as elsewhere in the book applying his semantic doctrines to the elucidation of the formula of transubstantiation—"this is my body" (1.15).

Arnauld's notion of accessory ideas might have been (but was not) used to advantage in his discussion of problems of identity of reference, where he argued that when the mind frames the proposition "that Rome, which was of brick before the time of Augustus, was of marble when he died, the word 'Rome,' which appears as only one subject, nevertheless marks two subjects that are really distinct but reunited under a confused idea of Rome that prevents the mind from perceiving the distinction of subjects" (2.12). The suggestion is that the proposition should be rejected by anyone having a clear and distinct idea of Rome, which is preposterous. Even if the proper signification of "Rome" was taken to be only the idea of buildings, surely such accessory ideas as location, population, and institutions could have been invoked to warrant the continuing use of the single proper name.

Like Hobbes, Arnauld recognized complex terms expressed in a single word, but instead of Hobbes's criterion of analyzability Arnauld employed the notion of accessory ideas attaching to the word under certain circumstances. Thus, the term king, which is simple "in expression," was "a term complex in sense " when uttered in seventeenth-century France, "because in pronouncing the word 'King' we not only have in mind the general idea corresponding to that word; we also mentally join to it the idea of Louis XIV, who is now King of France. There is an infinity of terms in ordinary human discourse that are complex in this respect" (1.8).

Arnauld's analysis of the semantics of sentences clearly illustrates the importance of the loss of supposition theory. In one badly confused but typical passage he claimed that "when one says that men are animals, the word 'animal' no longer signifies all animals, but only the animals that are men" (2.17). Not only does this transform predication into identity, it also violates his own doctrine of signification. Again, in discussing "some man is just" Arnauld maintained that "just" there "signifies only the justice that is in some man," the result being that "some man is identified with some just [thing]" (2.18).

A complete chapter of the Port-Royal Logic is devoted to the discussion of propositions such as "this is Alexander" (pointing to his portrait), which he described as "expressions in which one uses the name of the thing to mark the sign," seldom if ever causing any difficulty in actual use, and propositions such as Joseph's explanation of Pharaoh's dream—"the seven full sheaves are seven full years of plenty"—"expressions in which, the sign being marked by its own name or by a pronoun, one affirms of it the signified thing." One result of this novel approach to metaphor is his formulation of a rule governing the appropriateness of the second sort of proposition: "the mind of those to whom one speaks must already regard the sign as a sign and be concerned to know of what it is a sign" (2.14).

"comprehension" and "extension."

Certainly the most influential semantic doctrine of the Port-Royal Logic was Arnauld's introduction of a distinction between the "comprehension" and the "extension" of a term. (In the nineteenth century Sir William Hamilton renamed the former "intension.") The principle of such a distinction had been employed in the medieval distinction between simple and personal supposition, and even the distinction itself had occasionally been anticipated in one form or another (for instance, by Cajetan), but Arnauld's introduction of it seems to have been original and certainly was the first instance of a systematic use of it.

It is difficult, however, to say exactly what Arnauld intended by the distinction, for its exposition is obscured by his generally confused account of signification. He first advanced the distinction as one pertaining to "universal ideas" (or terms). "I call the comprehension of the idea the attributes it comprises in itself that cannot be removed from it without destroying it, as the comprehension of the idea of the triangle comprises extension, figure, three lines, three angles, the equality of those three angles to two right angles, etc. I call the extension of the idea the subjects with which that idea agrees, … as the idea of triangle is extended to all the various species of triangle." And Arnauld went on to say that the idea could be restricted in its extension by "applying it to only some of the subjects with which it agrees, without thereby destroying it," for example, by attaching to it "an indistinct and indeterminate idea of a part, as when I say 'some triangle'" (1.6). If, however, the extension consists of species and not of the individuals, which is what Arnauld maintained, then such a device for restricting extension is always to be read as "some (species of) triangle," which produces an absurdity. Because of his theory of signification there would be theoretical difficulties for Arnauld in simply identifying the term's extension with the individuals in question, but for the most part he seems to have had that identification in mind rather than the one he laid down.

Individual terms, too, were said to have comprehension and extension. In the phrase "Julius Caesar, the greatest commander the world has ever seen," the comprehension of that individual term is "explicated" in one of countless possible ways. But the extension of an individual term cannot be restricted, Arnauld maintained, and thus every singular proposition is universal (1.8, 2.3).

Locke

In the third book of his Essay Human Understanding (1690) John Locke (1632–1704) produced the first modern treatise devoted specifically to philosophy of language. No work had a greater influence over the development of semantics during the Enlightenment than did Book III of this work, "Of Words"; yet its semantic theories were neither novel in principle nor clearly and thoroughly developed. To go no further back, many of its principles had been anticipated in Kenelm Digby's Two Treatises (1664), in Richard Burthogge's Organum Vetus et Novum (1678), and in Hobbes's works. Of course Locke's "Of Words" acquired importance simply by being a part of the enormously influential Essay, but the source of its special influence lay in the fact that Locke had expressly connected semantic inquiry with theory of knowledge. He had set out to investigate "our knowledge," and along the way he found himself unexpectedly compelled to investigate "the force and manner of signification" of words (3.9.21), having discovered that "there is so close a connexion between ideas and words … that it is impossible to speak clearly and distinctly of our knowledge, which all consists in propositions, without considering, first, the nature, use, and signification of Language" (2.33.19). The new epistemological orientation of semantic inquiries, apparent even in the logic books of the period, was first explicitly established in Locke's Essay.

Locke evidently thought of the material of Book III as serving two purposes in his philosophy. On the one hand, he characterized his new "way of ideas" as nothing more than "the old way of speaking intelligibly" (third letter to Stillingfleet), which he reduced to a few commonsensical maxims for the avoidance of "jargon," very much in the spirit of Bacon's treatment of the "Idols of the Market Place." The semantic theory in Book III was developed in part as a support for these "remedies of the … imperfections and abuses of words" (3.11), and Locke's preoccupation with that practical aim may help to explain some of the imprecision and inconsistency in his theoretical statements. He did, however, clearly recognize a more strictly theoretical purpose in the semantic inquiries of Book III, one which he summarized in his description of the third branch of science—"Σημειωτικὴ, or the doctrine of signs " (4.21.4), the consideration of ideas as the signs of things and of words as the signs of ideas.

Locke's account of words as the signs of ideas shows little of the sensitivity to the complexities of language that had characterized the work of many of his predecessors, including Hobbes. Except for one very short, cryptic chapter on "particles" (by which he evidently meant syncategorematic words but perhaps also verbs), the semantics of words in Book III is exclusively a semantics of "names"—names of "simple ideas," of "mixed modes," and of "natural substances"—with no suggestion that anything has been left out of consideration (3.4.1).

The development of his fundamental thesis regarding the signification of these names begins with his observing that "words being voluntary signs, they cannot be voluntary signs imposed by [a man] … on things he knows not." Now what a man knows is in his mind, but all that is in a man's mind is his own ideas. Therefore, "words, in their primary or immediate signification, stand for nothing but the ideas in the mind of him that uses them " (3.2.2). This is not markedly different from the starting point of many earlier semantic theories, but Locke's initially uncompromising development of it led to some extreme consequences. Men, he observed, "suppose their words to be marks of the ideas also of other men" or "to stand also for the reality of things." Faithful to his fundamental thesis, Locke nevertheless insisted that "it is a perverting the use of words, and brings unavoidable obscurity and confusion into their signification, whenever we make them stand for anything but those ideas we have in our own minds" (3.2.4–5).

Thus, the basic semantic relation in Locke's account of language is that of a word used by some speaker as a proper name for some idea in that speaker's mind. It seems to follow from this doctrine that as long as one does use words in this (the only approved) way, one cannot misuse them; and Locke does sometimes suggest that in the early chapters of Book III (see, for instance, 3.2.3). Those chapters indeed present a classic formulation of what Wittgenstein was later to criticize as the notion of a "private language."

Establishing words as proper names of ideas in the speaker's mind fulfills the first of Locke's two principal conditions "for the perfection of language" (3.1.3). The second was the devising of "general words," which he thought men accomplished by using words "for signs of general ideas." It was evident, Locke observed, that general words "do not signify barely one particular thing; for then they would not be general terms but proper names." His account of the signification of general words is, however, severely damaged by the inclusion in those same passages of his declaration of a thoroughgoing nominalism: "things themselves … are all of them particular in their existence, even those words and ideas which in their signification are general" (3.3.11–12).

Although many of his most careful theoretical statements ruled out any extension of the signification of a word beyond an idea in the speaker's mind, Locke here (and frequently in the later chapters of Book III) was apparently assuming that by virtue of signifying an idea, a word also (secondarily and indirectly, perhaps) signified whatever the idea signified. However, he never examined that assumption or even recognized it to be one. When he came to apply his theory to the discussion of various sorts of names, he often relaxed or ignored the strictures laid down in the general theory developed in the first three chapters. Thus, in his chapter on the "names of our ideas of substances" he found it convenient to say "By the word gold here, I must be understood to design [that is, designate] a particular piece of matter; v.g., the last guinea that was coined. For if it should stand here in its ordinary signification, for that complex idea which I or anyone else calls gold, i.e. for the nominal essence of gold, it would be jargon" (3.6.19).

When, on the other hand, Locke did apply his semantic theory strictly, he was likely to produce such surprising results as his doctrine that every generalization about a substance, such as "all gold is fixed," means either "that fixedness is a part of the definition, i.e., part of the nominal essence the word gold stands for; and so this affirmation 'all gold is fixed,' contains nothing but the signification of the term gold. Or else it means, that fixedness, not being a part of the definition of the gold, is a property of the substance itself, in which case it is plain that the word gold stands in the place of a substance.… In which way of substitution it has so confused and uncertain a signification that, though this proposition—'gold is fixed'—be in that sense an affirmation of something real, yet it is a truth will always fail us in its particular application [since we know only our idea of gold and not 'the real essence' of gold], and so is of no real use or certainty" (3.6.50; compare his interesting treatment of "trifling propositions" in 4.8).

Locke's strictly subjectivist, nominalist theory of signification in the opening chapters of Book III, which gave him so much trouble in its application, may represent nothing more than his overzealous attempt to state precisely such characteristically commonsensical observations as can be found in his Conduct of the Understanding, Section 29, where he advised "those who would conduct their understanding right, not to take any term … to stand for anything, till they have an idea of it. A word may be … used as if it stood for some real being; but yet if he that reads cannot frame any distinct idea of that being, it is certain to him a mere empty sound without a meaning."

(Locke's influence is frequently discussed in the remainder of this entry. See, for instance, the sections on Gottfried Wilhelm Leibniz, George Berkeley, and Étienne Bonnot de Condillac and on universal grammar.)

Leibniz

Gottfried Wilhelm Leibniz (1646–1716) developed some of his views on language specifically as criticisms of Locke in his Nouveaux Essais sur l'entendement (finished after 1709; first published 1765). One example of this ad hoc development is his rejection of Locke's account of "general words" as no more than devices for avoiding the proliferation of proper names. Leibniz argued that they were necessary ingredients in the "essential constitution" of languages and went so far as to claim, in an exact reversal of Locke's position, that "it is certain that all proper or individual names were originally appellative or general" (3.1.3; see 3.3.5). Even in the Nouveaux Essais, however, most of Leibniz's views on language can be traced to considerations that lie at the center of his own philosophy.

universal characteristic

Perhaps the most important of the central considerations of Leibniz's philosophy is his lifelong preoccupation with the idea of a "universal characteristic," which cannot be examined here except as it bears directly on his philosophy of language. Leibniz's earlier doctrine of the characteristic (c. 1679) was "that a kind of alphabet of human thoughts can be worked out and that everything can be discovered and judged by a comparison of the letters of this alphabet and an analysis of the words made from them " (Gerhardt edition 7.185). Descartes, by contrast, had maintained that such a language (he never knew of Leibniz's scheme, of course) depended on the prior establishment of "the true philosophy" (letter to Marin Mersenne [1629], in Adam and Tannery edition 1.76).

Leibniz's initial response was that while the establishment of the characteristic "does depend on the true philosophy, it does not depend on its completion"; for as long as we have the true "alphabet of human thought" to begin with, we can complete the true philosophy simply by correctly manipulating the characteristic (Couturat edition, pp. 27–28). (The many artificial languages projected during the Enlightenment may be classified as "Cartesian" or "Leibnizian," depending on whether they were put forward solely as devices for recording and communicating knowledge or also as heuristic devices. It is the Leibnizian rather than the Cartesian projects that bear a significant resemblance in principle to the formalized languages for logic developed after the middle of the nineteenth century.) Writing some years later (1697) and in a context where the issue between his own and Descartes's views was not explicit, Leibniz did nevertheless acknowledge that "genuinely real, philosophic characters must correspond to the analysis of thoughts. It is true that such characters would presuppose the true philosophy, and it is only now [when he believed himself to have discovered the principles of the true philosophy] that I should dare to undertake the construction of them" (Gerhardt edition 3.216).

By a "real" characteristic Leibniz meant a symbolism that was in some important respect naturally (rather than conventionally) associated with what it symbolized. Although a thoroughly real characteristic could be developed only in an artificial language, Leibniz observed that natural languages were in certain respects real characteristics. It was on the basis of that observation that he became the first major philosopher after Epicurus to suggest an appeal to ordinary language as a philosophical technique. His general attitude is expressed in the Nouveaux Essais : "I truly think that languages are the best mirror of the human mind and that an exact analysis of the signification of words would make known the operations of the understanding better than would anything else" (3.7.6). Part of what he meant by "exact analysis" closely resembled Plato's use of etymology in the Cratylus.

In his preface to a 1670 edition of Nizolius (in which he has a great deal to say about language) Leibniz argued that "the good grammarian and the philosopher too can, so to speak, deduce the use of a word from its origin by means of an unbroken sorites of metaphors" (Gerhardt edition 4.140). But he also viewed ordinary language in its unanalyzed state as having a special philosophic value: "Whatever cannot be explicated by means of popular terms (unless like many kinds of colors, odors, and tastes, it consists in immediate sensation) is nothing, and should be kept away from philosophy as if by a kind of purifying incantation" (4.143). Not every ordinary language was equally valuable as a touchstone for philosophy: "No language in Europe is better suited than German for this certifying trial and examination of philosophical doctrines by means of a living language, for German is richest and most nearly complete in real characters [in realibus ], to the envy of all other languages.… On the other hand, the German language is easily the least well suited for expressing fabrications [commentitia ]" (4.144; cf. Duclos edition 6.2.10 ff.). Leibniz's praise of German for its high proportion of real characters was very likely based simply on the fact that it contains words of Germanic origin where English and the Romance languages are likely to have words of Greek and Roman origin—for instance, Unabhängigkeit and "independence"—a feature of the German language which no doubt does provide its native speakers with comparatively easy access to many abstract notions.

Leibniz also recognized a more pervasive kind of "realness" in natural languages that might be called syntactic, in contrast with the historically more familiar kind just discussed. It constituted the essential ingredient in his doctrine of "expression" and thus formed part of his metaphysics (monads express the universe) as well as of his philosophy of language. In What Is an Idea? (1678) he offered this account:

That is said to express a given thing in which there are relations [habitudines ] that correspond to the relations belonging to the thing expressed. But these expressions are of different kinds—e.g., the model of a machine expresses the machine itself; the projective delineation of a thing in a plane expresses the solid; discourse expresses thoughts and truths; characters express numbers; an algebraic equation expresses the circle (or some other figure)—and what is common to these expressions is the fact that we can pass from the mere consideration of the expressed relations to a knowledge of the corresponding properties of the thing being expressed.

Leibniz drew the conclusion that "it is clearly not necessary that that which expresses be similar to that which is expressed as long as a certain analogy of [internal] relations is preserved" (Gerhardt edition 7.263–264). What he was proposing, however, was clearly a novel approach to resemblance as a basis for semantic relations, suggesting for the first time that in complex signs the "realness" of the symbolism may consist in the resemblance between the schemata of the expression and of what is expressed rather than in a resemblance between the elements of those two schemata. This was brought out most clearly in his Dialogue (1677)—for example, in the observation that "even if the characters are arbitrary, still the use and interconnection of them has something that is not arbitrary—viz. a certain proportion between the characters and the things, and the relations among different characters expressing the same things. This proportion or relation is the foundation of truth" (7.192).

In describing this schematic resemblance as the foundation of truth, Leibniz stated the principal thesis in his novel doctrine of propositions as extralinguistic, extramental schemata. Although such a notion of propositions had been hinted at by Hobbes, Leibniz was evidently the first to make it explicit; and, as it happened, he developed his doctrine in conscious opposition to Hobbes's view of truth as dependent on words and hence arbitrary. It had been standard philosophical usage from the beginning of the Middle Ages to use the word proposition for whatever was either true or false, and the principal refinement of this usage before Leibniz had been the medieval distinction of "mental" propositions from propositions spoken or written. Leibniz's first objection against what he called Hobbes's "super-nominalism" might be interpreted as going no further than that, as in his observation that "truths remain the same even if the notations vary" (Preface to Nizolius [1670], in Gerhardt edition 4.158). He subsequently recognized, however, that those "truths" could not be identified with true propositions that had been, were, or would be actually in someone's mind—for instance, in the Dialogue of 1677: "A. … Do you think that all propositions are thought? / B. I do not. / A. You see, therefore, that truth does belong to propositions or thoughts, but to possible [propositions or thoughts], so that this at least is certain, viz. that if anyone should think in this [or a contrary] way, his thought would be true [or false]" (7.190).

Once Leibniz had distinguished propositions from actual thoughts and from combinations of words, he was in a position to reject the traditional account of truth as "the conjunction or separation of signs according as the things themselves agree or disagree among themselves," in which account "by 'the conjunction or separation of signs' one must understand what is otherwise called a proposition." Leibniz's attack on this tradition contrasted its technical terminology with ordinary usage in order to show that it concealed rather than resolved problems:

An epithet—e.g., "the wise man"—does not make a proposition, and yet it is a conjunction of two terms. Negation, moreover, is something different from separation, for saying "the man" and after an interval pronouncing "wise" is not to deny. Finally, agreement [or disagreement ] is not, strictly speaking, what one expresses by means of a proposition; two eggs have agreement and two enemies have disagreement. The manner of agreeing [or disagreeing] at issue here is quite extraordinary [toute particulière ]. Thus I think this definition completely fails to explicate the point at issue. (Nouveaux Essais 4.5.2)

And Leibniz went on from this criticism of the traditional doctrine of propositions to present once again his own view of them as entities distinguishable both from words and from actual ideas (Nouveaux Essais 4.5.2).

"leibniz's law."

Leibniz's famous principle of substitutivity, known in recent literature as Leibniz's Law, was frequently used as a starting point by twentieth-century writers on semantics. Leibniz employed the principle as part of the primitive basis of his logical calculus and put forward several versions of it in papers written from 1679 through the early 1690s. The various versions may be accurately synthesized as follows: Those entities are the same, one of which may be everywhere substituted for the other, preserving the truth(-value) (see 7.219; 7.228; 7.236).

Although Leibniz did not identify the entities in question and sometimes discussed the principle as if it applied, for example, to geometrical figures, the context generally makes it plain that its principal intended application was to terms in propositions actually expressed in some notation. His discussion of the principle in the papers in which he applied it took no account of contexts in which the principle does not apply, but at least one passage in his later writings shows that he had by then recognized that cases of what the medievals had called material supposition did not fall under the principle. "Indeed, one sometimes speaks of words materially without being able in that context [cet endroit-là ] to substitute in place of a word its signification, or its relation to ideas or to things. This occurs not only when one speaks as a grammarian but also when one speaks as a lexicographer, in giving the explication of a name" (Nouveaux Essais 3.2.6). Recent criticism of Leibniz's Law has often begun with the complaint that he failed to notice just such exceptions.

Berkeley

Locke had argued that a word was significant solely in virtue of standing for an idea in the mind of the user of the word. When George Berkeley (1685–1753) began philosophizing, he accepted that doctrine as axiomatic. In several early entries in his private Philosophical Commentaries (1707–1708) he presented it as part of the basis of his otherwise anti-Lockean position—for instance, "All significant words stand for Ideas" (Luce and Jessop edition 1.45; see 1.39, 1.43, 1.53). Even before ending the Commentaries, however, Berkeley had rejected Locke's semantics too and had begun to replace it with a doctrine of great importance in the development of his own philosophy and in the history of semantics.

The actual turning point was apparently reached in his discovery that some words that should have been paradigm cases for Locke's semantics had no precisely correspondent ideas. "Qu: How can all words be said to stand for ideas? The word Blue stands for a Colour without any extension, or abstract from extension. But we have not an idea of Colour without extension; we cannot imagine Colour without extension" (1.62). In this passage Berkeley questioned for the first time not only Locke's semantics but also (indirectly) his doctrine of abstract ideas. He very soon saw that the connection between the two was essential. Given Locke's semantics, together with the facts that a general word was significant and that no concrete particular idea corresponded to it, one was forced to introduce a Lockean abstract idea simply in order to give a general word something to stand for. (As Berkeley pointed out [2.36], Locke had virtually admitted as much in the Essay [3.6.39].) Berkeley's alternative account in the Introduction to The Principles of Human Knowledge (1710), Section 11, was that "a word becomes general by being made the sign, not of an abstract general idea but, of several particular ideas, any one of which it indifferently suggests to the mind" (2.31; see 2.127). Berkeley's account thus involved abandoning Locke's semantic principle that there be a single idea to serve as the name-bearer for each significant word.

In the history of semantics, however, as in the history of philosophy in general, Berkeley's rejection of abstract ideas is more important than his alternative account of the signification of general words. The rejection was based not only on the well-known exposition of the internal inconsistency—as in Principles, Introduction, Section 13 (2.32–33)—but also on his many and varied attacks on their semantic foundation. Since Locke's commitment to the view that each word had to stand for one idea in order to be significant was what had compelled him to introduce abstract ideas, Berkeley set out to show, by means of various sorts of counterinstances, "that words may be significant, although they do not stand for ideas. The contrary whereof having been presumed seems to have produced the doctrine of abstract ideas" (3.292–293; see 1.70).

He seems to have found at least four sorts of counterinstances, the first and most obvious consisting of words that stand for something other than ideas. Words such as "volition," "I," "person," and the "particles" (or syncategorematic words) are significant in virtue of standing for "spirits" or their activities (the particles standing for "the operations of the mind") (1.65, 1.80, 1.81; see 3.292). But in Berkeley's immaterialism there were no entities other than spirits and ideas for which words could stand, and so there could be no other counterinstances consisting simply of words that stood for nonideas. He was thus led to investigate the relation "stands for" more closely than its relata. In a move reminiscent of supposition theory he attacked Locke's account of "understanding propositions by perceiving the agreement or disagreement of the ideas marked by their terms," claiming that when he asserted of a particular dog "Melampus is an animal" he had not two ideas but "only one naked and bare idea, viz. that particular one to which I gave the name Melampus." Nor does "animal" in that proposition "stand for any idea at all. All that I intend to signify being only this, that the particular thing I call Melampus has a right to be called by the name animal." But it would not do, he pointed out, to say that "animal" here stood for the same idea as did "Melampus," since that would make the proposition a tautology (2.136–137; cf. 1.69, 8).

The principal effect of this second sort of counterinstance was to raise some serious doubt regarding the nature of the relation "stands for," and Berkeley's remaining counterinstances took the almost unprecedented step of suggesting that that relation was not always an essential ingredient in significance. Words that might in certain occurrences be said to stand for ideas are very often used in reasoning and in ordinary conversation as uninterpreted (but interpretable) "counters." A word used in that way does not in each of its occurrences stand for an idea in the mind of the user or, for that matter, raise a corresponding idea in the mind of the hearer or reader (2.37, 3.291–292, 8.25, 8.27).

Finally, a word sometimes occurs in a context such that one would miss rather than grasp its significance by taking it to stand for the idea to which it is customarily attached. "For example, when a Schoolman tells me Aristotle hath said it, all I conceive he means by it, is to dispose me to embrace his opinion with the deference and submission which custom has annexed to that name" (2.38). What is more, a word may occur in a context that precludes the possibility of taking it to stand for an idea without thereby being rendered insignificant—for example, the subject term in "the good things which God hath prepared for them that love him are such as eye hath not seen nor ear heard nor hath it entered into the heart of man to conceive."

It was Berkeley's view that the significance of propositions such as these last two was to be found not in the ideas the words might otherwise be said to stand for but in the purpose, or "design," of the proposition. The design of this last example cannot be "to raise in the minds of men the abstract ideas of thing or good nor yet the particular ideas of the joys of the blessed. The design is to make them more cheerful and fervent in their duty " (2.137 [italics added]; see 2.293, 3.292). Words, he held, "have other uses besides barely standing for and exhibiting ideas, such as raising proper emotions, producing certain dispositions or habits of mind, and directing our actions" (3.307). Thus, in his attacks on the semantic foundation of Locke's doctrine of abstract ideas Berkeley came nearer than anyone since the Stoics to abandoning, or at least supplementing, the attempt to account for all linguistic meaning in terms of the relation between names and their bearers.

As for Locke's semantics, Berkeley had reduced it to the unexceptionable principle of common sense that had no doubt prompted Locke's theoretical claims, namely, we ought not to use words without knowing their meaning (1.78; 2.76). But he took Locke's call for a new "doctrine of signs" quite seriously, summarizing his own (mostly anti-Lockean) semantic theory under that heading in Alciphron (1732), Dialogue VII, Section 14 (3.307). Like Bacon, Hobbes, and Locke before him, Berkeley thought of himself as providing philosophical remedies for the abuse of words, but he differed from them in making this the core of his philosophy, announcing that "the chief thing I do or pretend to do is only to remove the mist or veil of Words" (1.78; see 2.40). He set out explicitly to do just that at many points throughout his writings, but nowhere in a more concentrated form than in the introduction to the Principles.

In keeping with that aim Berkeley frequently urged his readers to contemplate ideas apart from words, maintaining that "if men would lay aside words in thinking 'tis impossible they should ever mistake save only in Matters of Fact" (1.84; see 2.40). He felt, therefore, that it was "absurd to use words for the recording our thoughts to ourselves: or in our private meditations" (1.62) and introduced his "Solitary Man" for the purpose of examining that pristine state of mind in a concrete example, "to see how after long experience he would know without words" (1.71; see 2.141–142). Such passages taken together suggest an anticipation of Wittgenstein's attack on the notion of a private language, but Berkeley had second thoughts about the absurdity of the private use of words and seems to have concluded that "the Solitary Man would … find it necessary to make use of words to record his Ideas if not in memory or meditation yet, at least, in writing without which he could scarce retain his knowledge" (1.75).

Berkeley's ingenious linguistic analogy in his account of sense experience, the "Universal Language of Nature," was first put forward in his New Theory of Vision (1709) and developed in several later works. Speaking strictly, it belongs to his theory of knowledge rather than to his philosophy of language, but in the course of developing the analogy he often made interesting observations about language conceived in the ordinary sense (see, for instance, 1.228–233, 1.264–265).

Maupertuis and his Critics

In the latter half of the eighteenth century philosophical interest in language was concentrated among French philosophers. Under the influence of Condillac and the British empiricists they eventually came to consider the analysis of signification their most important task. Among the earliest figures in this development was Pierre-Louis Moreau de Maupertuis (1698–1759), who first published his brief Réflexions philosophiques sur l'origine des langues et la signification des mots in 1748. To some extent his position resembled those taken by Berkeley and by Condillac in his first book, Essai sur l'origine des connaissances humaines (1746); but Maupertuis seems to have written his Réflexions before he knew their work. Partly because of the author's fame as a scientist, the Réflexions attracted considerable attention and was commented on by Baron de L'Aulne Turgot (1750), Condillac (1752), Denis Diderot (1753), Voltaire (1753), and Maine de Biran (1815), among others.

Maupertuis conceived of the question of the origin of language very much as philosophers since Descartes had been conceiving of the question of the origin of knowledge. It was intended to give rise not to speculations about prehistoric man but rather to an analysis of the hypothetical circumstances of a man with fully developed faculties who has suddenly been deprived of all his memories and of all human society. Would such an individual frame a language at all? If he did so, what would be the stages of its development? By asking and answering such questions as these within the framework of his "metaphysical experiment," Maupertuis expected to gain insight into the nature of language and its relation to the acquisition of knowledge. He began by imagining himself in the condition of the adult newborn. As soon as he had had two perceptions,

I should see that the one was not the other, and I should try to distinguish between them. And since I should have no ready-made language, I should distinguish between them by means of any marks whatever and might be satisfied with the expressions "A" and "B" as standing for the same things I now mean when I say "I see a tree," "I see a horse." Receiving new perceptions afterwards, I could designate them all in that way. (Réflexions, Sec. 7)

It is not clear whether his saying "A" to himself in this protolinguistic context is really separable from his act of individuating the perceptual event of his seeing a tree, but Maupertuis did consider "A" and "B" as signs of his perceptions and thus presented a nearly classic case of what Wittgenstein later described as a "private language." The first development beyond those initial "signs" was recognized by Maupertuis as sufficiently radical to be described as "another language." In Section 8 he wrote:

For example, in the preceding perceptions I should recognize that each of the first two had certain characteristics that were the same in both and that I could designate those by a single sign. Thus I should change my first simple expressions "A" and "B" into these: "CD" and "CE," which would differ from the first only in that new convention, and which would correspond to the perceptions I now have when I say "I see a tree," "I see a horse."

Maupertuis's analysis proceeded in this way until he had introduced devices for discriminating kinds of perception, numbers of objects perceived, remembered and anticipated perceptions, and so on. His purpose in doing so, however, was to provide the background for a new philosophic method, which he applied most notably in his analysis of "the force of the proposition 'there is ….'" Although in saying "there is a tree" I may seem to be making a claim that goes beyond the evidence of my perceptions, once language has been reconstructed on the basis of my perceptions alone, I am in a position to see that "there is a tree" is no more than an abbreviation for "I shall see a tree every time I go to that place." This latter proposition in turn is reducible to the sequence "I was in a certain place," "I saw a tree," "I returned to that place," "I saw that same tree again," and so on (Secs. 24–28).

Eight years after writing the Réflexions and having meanwhile read the French translation of Berkeley's Dialogues (1750), Maupertuis readily admitted the similarity of his metaphysics to Berkeley's and rested his claims of independent importance on having introduced an analysis of language as the means to that end. "The point is that this philosopher [Berkeley] attacks the system of our errors only by parts. He demolishes the structure at the top, and we undermine its foundations. This is a structure quite different from that famous tower the erection of which on the Plains of Shinar was prevented by the confusion of tongues; this one is not erected except by abusing or forgetting the meaning of words" (Reply to Boindin, Sec. II, in Oeuvres, 1756 ed.). Berkeley's own attitude was, of course, much the same; but the Dialogues alone among his major works fails to bring that out.

Several of Maupertuis's critics, most notably Turgot, attacked the hypothesis on which he rested his inquiry: "A solitary man such as Maupertuis imagines … would not try to find marks with which to designate his perceptions. It is only when confronted with other people that one looks for such marks. From this there follows what is obvious in any case, that the first purpose and first step of language are to express objects and not perceptions" (Remarques critiques, Sec. 7). Only Maine de Biran among Maupertuis's critics defended his use of the private-language hypothesis (Note sur les Réflexions, Sec. V).

Condillac

Étienne Bonnot de Condillac (1715–1780) wrote his first book, Essai sur l'origine des connaissances humaines (1746), in an effort to do what he felt Locke might have done if he had not "realized too late" the importance for epistemology of the material in Book III of his Essay, "Of Words" (Locke's Essay 3.9.21). Locke had "treated only in his third Book what should have been the subject matter of the second" (Essai, edited by G. Le Roy, 1.5a). Condillac acknowledged its historical value: Locke seemed to him to have been "the first to have written on this material as a genuine philosopher. I felt, nevertheless, that it had to form a considerable portion of my own work, both because it can be viewed in a novel, more extended way and because I am convinced that the use of signs is the principle that discloses the source [développe le germe ] of all our ideas " (1.5b; italics added). Condillac thus became the first modern philosopher to found his theory of knowledge, and consequently his entire philosophy, on considerations of signification and language, considerations that occupied him throughout his career and that shaped French philosophy for at least fifty years afterward.

Like Locke, Condillac denied that the ideas produced in sensation alone constitute a kind of knowledge, but he began his divergence from Locke in his account of the acquisition of knowledge on the basis of such ideas. "The sole means of acquiring knowledge is to trace our ideas back to their origin, to observe their generation, and to compare them under all possible relations. This is what I call analysis " (1.27a). Analysis consists in discriminating and ordering elements that are presented confusedly and simultaneously and thus requires the introduction of interrelatable signs for those elements. On these observations Condillac based his leading principle that "every language is an analytic method and every analytic method is a language" (2.419a).

This has the look of a vicious circle. Analysis is said to be a necessary condition of knowledge, and language to be necessary for analysis; but surely knowledge is also necessary for the formation of a language. Condillac attempted to break this circle by introducing the notion of an innate language, which he called the language of action. "The elements of the language of action are born with man, and those elements are the organs given us by the author of our nature. Thus there is an innate language although there are no innate ideas. Indeed, it was necessary that the elements of some sort of language, prepared in advance, should precede our ideas, since without signs of some kind it would be impossible to analyze our thoughts" (2.396b).

In its most rudimentary form this "language" consists simply in overt reactions: "our external conformation is set to represent everything that takes place in our soul" (ibid.). Involuntary expressions of fear, pain, desire, and so on are not elements of analysis for the individual producing them, but observers of his responses can, as a result of observing the order of events making up his responses, see analyzed for them what is simply gross experience for the respondent. "Men begin to 'speak' the language of actions as soon as they feel anything, and they speak it then without having any plan of communicating their thoughts. They form the plan of speaking it in order to make themselves understood only when they notice that they have been understood" (2.397a). The usefulness of results gained by this means stimulates a natural feedback process of development on "the principle of analogy," and the language of action is made more effective by the gradual transformation of "natural" and "accidental" signs into "signs of institution," the most convenient of which are articulate sounds (1.60b–62a). The origin of language, discussed as an independent topic by many of his contemporaries and successors, is thus an essential consideration in Condillac's epistemology.

Signs of institution, including, of course, words, are themselves natural in the sense that as a language develops, they are framed on analogy with more primitive elements in that same language (and ultimately with elements of the language of action). The principle of analogy is in fact a necessary ingredient in any usable language (compare Bacon's doctrine of "emblems"). "Imagine an absolutely arbitrary language, such that analogy had determined neither the choice of words nor their various senses. That language would be an ununderstandable gibberish" (2.471a). If the principle of analogy remained unimpaired in ordinary languages, "we would reason as nature teaches us to reason, moving effortlessly from discovery to discovery." But every ordinary language has been impaired to some extent by the intrusion of words that have the roots of their analogy in other languages. (A similar line of reasoning had led Leibniz to praise the German language as a natural "philosophical characteristic.") Perhaps, then, the principle of analogy can be retained in a perfectly unadulterated form only in a highly artificial language, such as algebra, which Condillac describes as "the language of mathematics."

Since language of some sort is a necessary condition of knowledge, it is a mistake to maintain, as Locke had done, that the primary purpose of language is to communicate knowledge. "The primary purpose of language is to analyze thought. In fact we cannot exhibit the ideas that coexist in our mind successively to others except in so far as we know how to exhibit them successively to ourselves. That is to say, we know how to speak to others only in so far as we know how to speak to ourselves" (1.442a). It is a consequence of this view that the art of thinking, or logic, reduces to the art of speaking.

Although a thought is not a succession in the mind, it has a succession in discourse, where it is decomposed into as many parts as there are ideas making it up. Then we can observe what it is we do when we think, we can give an account of it, we can, consequently, learn how to conduct our reflective thought. In this way thinking becomes an art, and that art is the art of speaking. (1.403b)

Condillac's view of the connection of thought and language was reinforced by his observations on "abstract general ideas." "When, for example, I think about man, I cannot consider anything in that word except a common denomination, in which case it is perfectly plain that my idea is in some way circumscribed in that name, that it extends to nothing beyond the name, and that, consequently, it is only that name itself" (2.401b). Thus the clarity and precision of abstract ideas "depends entirely on the order in which we have produced the denominations of classes. Therefore, there is only one means of determining ideas of this sort, and that is to produce a well-made language" (ibid.). Abstract general ideas, however, are the principal ingredients of reasoning, and Condillac was even ready to say that "to speak, to reason, to produce abstract or general ideas for oneself, are at bottom one and the same thing" (2.402a). His consideration of abstract ideas, then, was one more "proof that we reason well or badly only because our language is well or badly made" (ibid.).

All intellectual progress, on Condillac's view, depended on and in part consisted in establishing a "well-made language," and "a science, properly treated, is nothing other than a well-made language" (1.216a). The one perfectly well-made language so far established, he thought, was mathematics, which he examined from this point of view in his last book, La langue des calculs (1798). One reason why Condillac was prepared to identify a science with a well-made language is to be found in his doctrine of propositions. All that remains to be done in a science once the appropriate language has been established is the mechanical exposition of the truths proper to that science. The exposition is mechanical because "a proposition is only the unfolding of a complex idea in whole or in part," and since a proposition "in which one and the same idea is affirmed of itself" is an identical proposition, "every truth is an identical proposition" (2.748a). An identical proposition may, however, be instructive for some persons, namely, those who observe "for the first time the relation of the terms out of which it is formed.… Thus a proposition may be identical for you and instructive for me" (2.748b). Nevertheless, "if in all the sciences we could equally trace the generation of the ideas and everywhere apprehend the true system of the things, we should see one truth give birth to all the rest, and we should find the abridged expression of all we know in this identical proposition: the same is the same " (2.749b).

Condillac's influence extended not only to philosophers but also to the great chemist Antoine Lavoisier (1743–1794), who in his Méthode de nomenclature chimique (with Louis-Bernard Guyton de Morveau, 1787) and Traité élémentaire de chimie (1789) wholeheartedly adopted Condillac's notion of a science as a well-made language. Operating under this notion, Lavoisier introduced such technical terms as "phosphoric acid" and "sulphuric acid" in a successful attempt to initiate the development of the language of modern chemistry on Condillac's principle of analogy.

Lambert, Hamann, and Herder

In the century between Leibniz and Wilhelm von Humboldt, philosophy of language in Germany was concentrated in the writings of three men: Johann Heinrich Lambert (1728–1777), Johann Georg Hamann (1730–1788), and Johann Gottfried Herder (1744–1803).

Lambert was a distinguished mathematician and the first man to follow Leibniz's lead in his contributions to logic, the most important of which was the earliest attempt at a calculus of relations. His work in the philosophy of language appeared in his Neues Organon (1764), especially Part III, "Semiotik, oder die Lehre von der Bezeichnung der Gedanken und Dinge."

In philosophy of language, as in logic, the principal influence on Lambert was that of Leibniz, as may be seen in his preoccupation with the effect of language on thought and knowledge and with the possibility of controlling and improving that effect. Various natural languages impose various structures on our knowledge, but every natural language is fundamentally the product of prephilosophical, prescientific humankind. When we attempt to use such a language in advanced intellectual activities, we must submit our thought to the tyranny of usage (III, 1). We are thus led to seek an artificial language that from its inception could be entirely subjected to the needs of the intellect.

Lambert's attitude toward such artificial languages differed, however, from Leibniz's and constitutes a significant development in this line of thought. Great men, he observed, have worked at the project of a simple, perfectly regular and precise rational language, but without notable success. In any case, the adoption of such a language would be practically impossible (III, 2, 330). If we then revert to natural languages, however, we find that, strictly speaking, we cannot adopt any single one of them as a foundation for knowledge. There are, in the first place, conflicting usages even within a single natural language, some of which would have to be more or less arbitrarily ruled out; and, in the second place, any set of usages finally adopted would inevitably continue to undergo changes within the natural language of which they were a part.

Once we recognize that we do thus necessarily deviate to some extent from any given language in adapting it to intellectual purposes, it is apparent that we ought to do so consciously and under the guidance of preexamined criteria. The criteria developed and employed by Lambert were, he observed, the sort that might have served as the operative rules of a philosopher's artificial language. In fact, he seems to have elevated Leibniz's projected "universal characteristic" to the status of an ideal language, the principles of which are approximable to varying degrees but never fully realizable. He described his detailed examination of language as one that made a point of not distinguishing sharply between "actual and possible languages," meaning thereby that his approach to natural language was a mixture of description and prescription in which he attempted to point out those aspects of the actual language which were already accommodated to certain requirements of the ideal and to suggest ways in which those aspects might be enhanced and extended without introducing radical reforms that had little chance of acceptance.

The fundamental criterion employed by Lambert in his evaluation of sign systems in general and of natural languages in particular was the interchangeability of "the theory of the signs" and "the theory of the objects" signified, the degree of interchangeability marking the extent to which the signs approximated the fundamental ideal of being "scientific" (III, 23–24)—he cited musical notation as an example of a particularly close approximation. It seems evident that this fundamental criterion, which with its many corollaries pervades Lambert's philosophy of language, constituted his adaptation of Leibniz's doctrine of "expression." Besides systematic general chapters on various aspects of language, Lambert's "Semiotik" includes specific examinations of the character and function of various parts of speech and of the philosophical significance of etymological and syntactic interrelations among words.

Lambert and Hamann shared the conviction that the character of language was a topic of the greatest importance for philosophy, but they differed in almost every other important respect. Hamann's writings are undisciplined, obscure, and strongly colored by religious mysticism. Philosophically he was a forerunner of romanticism and existentialism, consciously rejecting most of the attitudes of the Enlightenment.

To the extent to which Hamann's philosophy exhibits a structure, it centers on his views on language, so much so that he himself called it verbalism (Schriften, edited by C. H. Gildemeister, 5.493–495). In almost everything he had to say about language, however, he opposed his contemporaries—Lambert (and the Leibnizian tradition) implicitly, Herder explicitly. The fundamental thesis of Hamann's verbalism is that ordinary natural language does and should take philosophical precedence over all technical or abstract language. Occasionally he wrote as if his basis for this claim was that God had employed such language as the instrument of revelation (Schriften, edited by F. Roth and G. A. Wiener, 1.85–86, 1.99), but he seems to have had more generally evaluable reasons for it as well. He evidently felt that the opposition between the rationalists and the empiricists of the Enlightenment was irresolvable largely because of the reliance of both parties on introspection. The special importance of ordinary language in this connection was that it constituted a medium in which the operations of reason and experience were united and made publicly accessible. The operations of reason, indeed, consisted entirely in linguistic operations (Roth and Wiener, 6.15 and 6.25; Gildemeister, 5.515, 7.9). Philosophy, however, had traditionally adulterated what should have served as its principal source and instrument. Hamann brought this out in a characteristic attack on Immanuel Kant's abstract, technical language in the first Critique :

While geometry fixes and fictionalizes the ideality of its concepts of points without parts, of lines and planes conforming to ideally divided dimensions, by means of empirical signs and figures, metaphysics misuses all the word-signs and figures of speech of our empirical knowledge as mere hieroglyphs and types of ideal relations and as a result of this learned mischief transforms the straightforwardness [Biederkeit ] of language into such a senseless, whirling, unsteady, indeterminable something (= x ), that nothing remains but a windy murmuring, a magical shadow-play, at best … the talisman and rosary of a transcendental superstition regarding entia rationis, their empty sacks and slogan. (Roth and Wiener, 7.8)

It was not only Kant's misuse of language that attracted Hamann's criticism but more especially his utter neglect of language as a topic for inquiry, which from Hamann's point of view vitiated Kant's claim to have provided a critique of reason. To point up the folly of such neglect, he tried to show that at various crucial junctures in his argument (as in the deduction of the categories) Kant had uncritically relied on certain linguistic conventions and that what he had called paralogisms and antinomies of reason really had their roots in the misuse of language.

Hamann based his doctrine of the preeminence of ordinary language not only on its value as a subject for philosophical inquiry but also on the fact that it alone among types of language was "objectively given." As such it served as the "womb" of reason and of all specialized, abstract languages designed to aid the operations of reason. Moreover, since ordinary language thus constituted the ultimate link between language and reality, all such abstract languages must be held finally accountable to it, that is, translatable into it.

As a philosopher of language, Herder is best known for his prize essay on the origin of language (1771), a topic with which this entry is generally not concerned. Herder's essay, however, occupies a position of special importance among hundreds of similar productions by eighteenth-century philosophers, for it began the trend away from the speculative problem of origin and toward the scientifically more accessible problems of the development of language. (It was praised for that reason by several of the great linguists of the nineteenth and twentieth centuries, such as Grimm, Theodor Benfey, Edward Sapir, and Otto Jespersen.)

Ostensibly Herder was adjudicating between two rival accounts of the origin of language, but his real purpose was to dismiss the problem as senseless. The two theories at issue were those of special divine creation and of deliberate human invention of language, the former as represented in J. P. Süszmilch's work and the latter associated primarily with Jean-Jacques Rousseau's second Discours. Herder took reason to be the defining characteristic of man and argued in support of Hamann's position that the operation of reason and the use of language were inseparable. He then drew the obvious conclusion that if God had created what was genuinely man, He had created a language-using animal and no special divine creation of human language was conceivable. By the same token, animals correctly describable as men could not conceivably have invented language. Thus the question of how and when humans came to use language was misconceived, although the question of how primitive human languages developed was well worth considering. Hamann ridiculed Herder's argument, with justification. It seems probable, however, that the argument was intended as irony, to deflate the pretensions of the theorists rather than to refute the theories.

The "IdÉologues."

Antoine Louis Claude Destutt de Tracy (1754–1836) devised the name idéologie for the new "science of the analysis of sensations and ideas." One section of the Second Class of the Institut National (founded in 1795) was devoted to that science in lieu of the prescientific inquiry known as metaphysics, and Destutt de Tracy and other philosophers associated with the work of that section became known as idéologues. For about eight years, until Napoleon Bonaparte abolished the Second Class of the Institut in the reactionary atmosphere of the First Empire, the idéologues were the dominant philosophical group in France. They thought of themselves as working in a field that had been opened by Locke and first thoroughly explored by Condillac, whose most original contribution was considered to be his discovery that language was as essential to the more fundamental processes of thought and analysis as it was to communication. Although part of at least Destutt de Tracy's interest in language was directed specifically to grammaire générale (see below), the idéologues followed Condillac in considering language a topic of importance in every area of philosophical inquiry.

The idéologues resented being thought of as disciples of Condillac, however, and while they did, professedly, share some of his broad philosophical convictions, much of what they had to say about language (as about other topics) involved substantial revision or outright rejection of Condillac's specific doctrines. There seems to have been some tendency for the revisions to take the form of a generalization of Condillac's doctrines, notably from a concern with language to a concern with signs of all sorts. Thus, in a representative passage of his Rapports du physique et du moral (delivered in 1796), Pierre-Jean-Georges Cabanis (1757–1808) purported to be defending and explicating Condillac's central claim that every language was an analytic method by arguing that "one distinguishes among sensations only by attaching to them signs that represent and characterize them; one compares them only in so far as one represents by signs either their resemblances or their differences." Of course, Cabanis pointed out, taking this account as explicative of Condillac's claim required taking "language" in "the broadest sense," as meaning "the methodological system by means of which one pins down [fixe ] one's own sensations" (Oeuvres, edited by Claude Lehec and Jean Cazeneuve, p. 157).

The question of the nature and epistemological function of signs (including linguistic signs) took on critical importance for the idéologues as a result of Destutt de Tracy's Mémoire sur la faculté de pensée (delivered in 1796), prompting them to set "the influence of signs on the faculty of thought" as the subject for the first essay competition sponsored by the section on the analysis of sensations and ideas. The best entries were Des Signes envisagés relativement à leur influence sur la formation des ideés by Pierre Prévost, Introduction à l'analyse des sciences, ou de la génération, des fondements, et des instruments de nos connaissances by P.-F. Lancelin, and Des Signes et de l'art de penser, considérés dans leurs rapports mutuels by Marie-Joseph Degérando (1772–1842). Degérando's essay won the prize and was published in an expanded four-volume version in 1800. In it some of the principal issues in the philosophy of language of the seventeenth and eighteenth centuries were subjected to a final scrutiny, partly as a result of the historically apt questions provided by the idéologues as a guide for the essayists:

(1) Is it really the case that sensations can be transformed into ideas only by means of signs? Or, what comes to the same thing, do our first ideas depend essentially on signs? (2) Would the art of thinking be perfect if the art of signs were perfected? (3) In those sciences in which there is general agreement as to the truth, is this because of the perfection of the signs employed in them? (4) In those branches of knowledge that provide inexhaustible fuel for dispute, is the division of opinion a necessary effect of the inexactitude of the signs employed in them? (5) Is there any means of correcting badly made signs and of rendering all sciences equally susceptible of demonstration? (Mémoires de l'Institut National des Sciences et Arts. Sciences morales et politiques, 1.i–ii)

Question (1) was on a thesis of idéologie itself, as may be seen in the passage from Cabanis quoted above. Degérando's answer was complex, but it was sufficiently affirmative to mark him an idéologue. On the one hand he felt that the mind needed no signs but merely an act of attention in order to pin down its sensations. On the other hand, "I shall give the name ['sign'] to every sensation that excites an idea in us in virtue of the association obtaining between them. Note carefully that it is not the sensation as such to which the name is given; it gets the name only in respect of the function it performs. Thus I shall say, for example, that the smell of a rose is the sign [not of the rose but] of the ideas of color and of form that the smell excites" (1.62–63). He distinguished between such prelinguistic signs and linguistic signs by pointing out that while the former "excite" ideas in us but attract attention to themselves, the latter lead our attention away from themselves to the ideas they have been made to signify, a formulation that constituted a refinement of the traditional distinction between natural and conventional signs.

In his detailed answers to questions (2) and (5) Degérando carefully criticized the many attempts at universal characteristics, calculi of reason, and philosophical languages that had been made by philosophers of the Renaissance and the Enlightenment. He laid down five criteria for such systems: (a ) unambiguous relations between signs and signified ideas; (b ) relations among signs exactly analogous to relations among signified ideas; (c ) simplicity, that is, minimum number of primitives (conditions premières ), each sign as abbreviated as possible, perspicuity of the sign system as a whole; (d ) distinctness among signs of different sorts and among syntactic relations of different sorts; (e ) as many distinct sorts of signs as distinct sorts of ideas to be signified (4.353–355). The only hopes of satisfying such criteria, he maintained, lay along four different lines, or "systems for philosophical language." Having examined each in detail, he concluded that all were in some respects unacceptable. Like Lambert, however, he suggested that a judicious application of the principles of such systems might produce some improvement in natural languages for philosophical purposes (4.355–415).

Perhaps even more important historically than his arguments against the feasibility of such artificial languages was his attack on the attitudes underlying them. It had usually been assumed that such a language would be international, and, indeed, if it were not, it would fail to achieve a good part of its purpose. But, Degérando maintained, there was no feasible means by which to establish it internationally, and even if it were established, it would soon be modified into separate dialects in various localities (3.557). Worst of all, the notion is pernicious, for such a language could at best be the instrument of communication exclusively among the learned and would thereby tend to separate them further from those they ought to instruct (3.572).

The fundamental mistake giving rise to all such schemes, according to Degérando, is the confusion of "the method of reasoning employed by the mathematicians with the mechanical processes of their calculations. Their method, as I have shown …, they do have in common with the metaphysicians," but the mechanical processes of their perfectly satisfactory artificial languages are the result of "the relative simplicity of the ideas on which they operate" (4.447–451). Other idéologues, particularly Destutt de Tracy, joined in this thoroughgoing repudiation of artificial languages for philosophy (see, for instance, Mémoires de l'Institut, Vol. III).

Questions (3) and (4) together called for an examination of Condillac's contention that a science was to be identified with a well-made language. Cabanis, himself a scientist, and Destutt de Tracy had frequently made significant use of this doctrine, but Degérando rejected it in a way that seems symptomatic of the end of the Enlightenment conception of a science. Some of the basis for his answers to these questions is evident in his answers to questions (2) and (5). "A well-made language," he maintained, "proclaims and presupposes a science that is already well advanced," thus adopting the Cartesian position on this issue rather than the Leibnizian. "We shall say that the great art of perfecting a science consists above all in making better observations and only then adopting a better language—i.e., one that is better suited to the observations that have been assimilated" (3.150–151). "The nomenclature of a science is related to the science itself as monuments are related to history: it preserves what is, but it can neither predict what is not yet nor unfold the future" (3.199).

Degérando resembled other idéologues more closely, however, in his view that improvements in philosophy—that is, in the analysis of sensations and ideas—did depend on a thorough examination of the natural language in which it was carried out. His own rather novel, never-realized scheme for accomplishing this was the construction of a philosophical dictionary.

It has been recognized that we can have clear ideas only in possessing a well-made language, and that a language can be well-made only in so far as we have reformed the most familiar operations of the mind from the very outset, only in so far as we have grasped the relation that interconnects them all. That being the case, we have felt the need of remaking the language in its entirety and, in some sense, recommencing the education of the human mind. The surest and perhaps the most truly efficacious means of accomplishing this great project would be, I think, the formation of a philosophical dictionary truly worthy of the name—one, that is to say, that would in some sense be a genealogical tree of our ideas and of the signs we use. Such a dictionary would be a sequence of definitions strictly bound to one another. Each notion would be defined in it by showing how it was acquired, or at least how it should have been acquired. The mind would find itself naturally led to create the words rather than seeking merely to explain them to itself.… The dictionary I propose would have an aim altogether different [from that of ordinary dictionaries]. In it one would seek to explain not so much how we speak as how we think; the conventions of the language would be presented in it as results, not as principles. … [This dictionary would not be arranged in alphabetical order but] it would be a book, a history. The order of facts would be the only order observed in it. It would not … be designed to be consulted occasionally, but it would have to be the object of a connected reading.… A definition would never be offered in it except in accordance with one general rule—that of determining an idea by means of tracing it back to the ideas that must have preceded it in the age when language was instituted among men.… The dictionary would thus in some sense embrace the history of mankind and would serve as a natural introduction to all the sciences. The study of it would be necessary for all who wished to think well, and its formation would be one of the noblest undertakings of philosophy. (4.80ff.)

What little influence the idéologues had is somewhat more noticeable in British than in Continental philosophy of the nineteenth century, partly because of the interest of some of the Scottish commonsense philosophers in their work.

Universal Grammar

Among the most important and distinctive influences on the philosophy of language of the Enlightenment was the development of universal grammar. In the broadest sense of the term it was, as defined by James Harris (1709–1780), "that grammar which without regarding the several idioms of particular languages only respects those principles that are essential to them all" (Hermes, or a Philosophical Inquiry concerning Language and Universal Grammar, 1751, Book I, Ch. 2). Although it resembles the speculative grammar of the Middle Ages in some of its basic assumptions, universal grammar seems to have had an independent origin that may with reasonable accuracy be dated 1660, the year in which Arnauld and Claude Lancelot published the Port-Royal Grammar—Grammaire générale et raisonnée.

Lancelot was a grammarian in the scholastic (rather than humanist) tradition who provided the subject matter that Arnauld presented in accordance with Descartes's method, believing that he was thereby "developing in grammar a branch of Cartesianism" (p. 137). In grammar, as in every subject, the method consisted fundamentally in "beginning with the most general and simplest matters in order to proceed to the least general and most complex," and in the study of language one therefore had to begin with principles and elements common to all languages in order to proceed to the study of one's own and other particular languages. Thus, universal grammar, as Arnauld (and many of his successors) conceived of it, was an investigation of language (langage ) designed as a propaedeutic to the study of languages (langues ). In practice, however, the elements and principles of universal grammar tended to be those of traditional Latin grammar, and thus many of the so-called universal grammars, at least before Condillac, are of little value either to linguists or to philosophers.

More important than the content of the early treatises on universal grammar, however, is the connection they established between grammar and philosophy, especially in France, where universal grammar dominated linguistic studies for 150 years following the Port-Royal Grammar. César Chesneau Dumarsais (1676–1756), the foremost of the universal grammarians between Arnauld and Condillac, maintained that "grammarians who are not philosophers are not even grammarians" (Véritables Principes de la grammaire, 1729, 1.201), and men engaged in the inquiry at that time styled themselves grammairiens-philosophes. Dumarsais seems often to have thought of "philosophy" in connection with grammar as no more than a certain scientific attitude (see, for instance, his article "Grammairien" in the Encyclopédie ), but he also held some of the views that were to serve as the basis for a more strictly philosophical grammar, as can be seen in his explanation that "grammar has a necessary connection with the science of ideas and reasoning because grammar treats of words and their uses and words are nothing but the signs of our ideas and our judgments" (1.201).

The grammairiens-philosophes began to be more markedly philosophers than grammarians beginning with the articles on grammatical topics in the Encyclopédie. Although Dumarsais was in general charge of them, several were written by philosophers such as Voltaire, Diderot, and Turgot; and the articles as a group contain less information on the announced topics than they do discussions of philosophical questions more or less vaguely associated with those topics. Nicolas Beauzée (1717–1789), one of the authors of those articles, defended the new approach to grammar in his Grammaire générale (1767): "Why should one think metaphysics out of place in a book on universal grammar? Grammar ought to expose the foundations—the general resources and the common rules—of language, and language is the exposition of the analysis of thought by means of speech. No aim is more metaphysical or abstract than that" (Préface, p. xvii).

In the works of Condillac, the emphasis was no longer on the propriety of taking a philosophical approach to grammar but rather on the fundamental importance of universal grammar as an inquiry serving the purposes of philosophy itself. In the "Motif des études" introducing his course of studies for the prince of Parma, Condillac described grammar as "a system of words that represents the system of ideas in the mind when we wish to communicate them in the order and with the interrelations we apperceive." Consequently, as he remarked in the "Objet de cet ouvrage" preceding his Grammaire (1775), he regarded grammar "as the primary division of the art of thinking. In order to discern the principles of language we must observe how we think; we must seek those principles in the analysis of thought. But the analysis of thought is quite complete in discourse, with more or less precision depending on the greater or less perfection of languages and the greater or less exactness of mind on the part of those who speak them. This is what makes me think of languages as so many analytic methods."

Condillac's elevation of universal grammar to the status of a fundamental philosophical inquiry marked the beginning of a new phase in the development of universal grammar, but his view of it was no more than a natural consequence of the then well-established belief that the construction of an absolutely universal grammar for all languages was a feasible undertaking. If language depicted thought and all languages shared a set of elements and principles, then the study of those common elements and principles would provide a science of human thought.

For Condillac's successors, the idéologues, who took the analysis of sensations and ideas to be the whole of philosophy, universal grammar became the philosophical method. As Destutt de Tracy put it, "this science may be called idéologie if one attends only to the subject-matter, universal grammar if one has reference only to the method, or logic if one considers only the goal" (Élémens d'idéologie, 1801, 1.5). It was in accordance with this conception of philosophy that the traditional chairs of logic and metaphysics in the écoles centrales of France were replaced in 1795 with chairs of universal grammar, which "by offering instruction in the philosophy of language would serve as an introduction to the course in private and public morality" (Élémens, Préface, p. xxiii).

Destutt de Tracy devoted the second part of his Élémens d'idéologie, some 450 pages, to a presentation of universal grammar suited to the purposes of the new course. Although he did occasionally cite parallel examples in other European languages and stress the value of knowing several languages, his principal interest was in what might fairly be described as the analysis of ordinary French. As an idéologue he was committed to provide analyses that would in every case disclose the signified idea (or sensation). His first step in establishing the conditions for such an analysis was to insist that the unit of signification was not the word or phrase, no matter how complex, but only the proposition, the linguistic device expressive of a judgment. If one simply utters the words "Peter," "to be not tall," we say that it means nothing, it makes no sense, although if one merely changes the form of the verb so that one says "Peter is not tall," thereby expressing a judgment, we can discern in what he says signs of his having an idea of Peter and an idea of his height (2.29–33).

Thus Destutt de Tracy committed himself to providing ideological analyses only if the linguistic entity to be analyzed occurs within a proposition. Even so, locating the signified idea sometimes required considerable ingenuity, as in his attempt to analyze all "conjunctions" in such a way as to show not only two propositions related by each conjunction but also the idea signified by each.

One can say as much regarding the conjunctions we use in asking questions, even though they might at first seem not to connect two propositions, because the first is suppressed. Thus when I say "how did you get in again?," "why did you leave?," I am really expressing these ideas: "I want to know [Je demande ] how you got in again," "I want to know why you left." And when we unfold the sense of those conjunctions the result is: "I want to know a thing that is the manner in which you got in again," "I want to know a thing that is the reason for which you left." (2.136)

In Destutt de Tracy considerations of grammar were entirely subject to the demands of ideological analysis, as is plain not only in the structure of his final analyses above but also in his readiness, quite unusual even among grammairiens-philosophes, to revise the classifications of traditional grammar (treating the "adverbs" comment and pourquoi as "conjunctions") when philosophical considerations seemed to call for their revision. When his volume devoted to universal grammar appeared in 1803, the experimental substitution of universal grammar for logic and metaphysics in the schools had already been abandoned, along with the idéologues ' highest hopes for revolutionizing philosophy.

In Germany, Christian Wolff and Lambert had taken notice of universal grammar, but the movement had no appreciable impact on philosophy. In England it affected mainly the work of Harris, of James Beattie (1735–1803), and of John Home Tooke (1736–1812), all of whom developed universal grammar far less as a philosophical than as a philological inquiry. By the beginning of the nineteenth century, universal grammar was rapidly going out of fashion as a branch of philosophy, even in France, despite the last efforts of the idéologues.