Since the earliest Greek philosophers were primarily cosmologists, their views on language are not the most fully developed (or best preserved) of their doctrines. Sources very late in antiquity attributed to Pythagoras (fl. 530 BCE) the view that although the soul assigned names to things, it did so not arbitrarily but on the basis of a natural connection between them, somehow like that between mental images and their originals. Modern historians sometimes credit Heraclitus (fl. 500 BCE) with having thought a great deal about language, but most of the fragments offered in evidence have to do with the logos, which surely is to be interpreted as the guiding principle of nature rather than as word or language. While we have nothing of his explicitly on language, it seems likely that Heraclitus did attach philosophical significance to the puns or contradictions in terms on which some of his paradoxical remarks depend.
Semantic theory seems to have made its first definite appearance in philosophy in the monism of Parmenides (fl. 475 BCE), who maintained that only what was true was expressible. He evidently based this remarkable doctrine on the argument that a statement is false if and only if it contains a false name, but a false name is by definition a name lacking a real bearer and hence a name that names or expresses nothing. (His monism of course entailed that there was only one real name-bearer.) Thus he described several words, such as "becoming" and "perishing," as "mere names that mortals have established, believing them to be true"—that is, believing that there really are such processes, which Parmenides denied.
Language first became a subject of specialized inquiry among the Sophists, who, unlike their philosophical predecessors, were more interested in man than in the cosmos. That orientation alone would probably have drawn them to the study of language, but there was also the fact that they earned their livings teaching people to speak well. Economic as well as philosophical considerations therefore probably played a part in leading them to include at least grammar as an important part of their work. Protagoras (fl. 445 BCE), the first of the Sophists, may also be considered the first grammarian. He distinguished the tenses and something like grammatical moods (classifying sentences as answers, questions, commands, and wishes), and he classified nouns as masculine, feminine, and "inanimate" (a division based on semantical rather than syntactic considerations, since it depended on the particular sex or lack of sex in the things the nouns were used to name). Grammar developed rapidly among the Sophists. Among the more philosophically interesting parts of grammatical theory to be found in Plato, who doubtless learned much of it from the Sophists, are distinctions between subject and predicate, between substantive and adjective, between the active and passive voices, and among types of discourse—political, rhetorical, conversational, dialectical, and technical.
The Sophists originated semantical as well as grammatical inquiries. Prodicus (fl. 435 BCE), who Plato thought was the best of the Sophists on language, seems to have operated on the hypothesis that there were no genuine synonyms, that where there were two words, there were two meanings. In Plato's dialogues Prodicus is depicted drawing instructive distinctions between "enjoyment" and "pleasure," "esteem" and "praise," "fearlessness" and "courage," for example; and he insisted on the study of "the right use of words" as the beginning of education. Protagoras, Prodicus, and Hippias (fl. 435 BCE) are all credited with treatises on "the correctness of names," and Socrates (d. 399 BCE) is depicted discoursing on that subject in Xenophon's Memorabilia (III, xiv, 2–7).
Semantics may have become a theoretical issue for the first time in the paradoxical arguments propounded by Gorgias (fl. 435 BCE) in support of his third nihilistic thesis. The three theses were (1) nothing exists; (2) even if something existed, it would be unknowable; (3) even if something existed and were knowable, it would be incommunicable. Gorgias gave four arguments for thesis (3) along the following lines. Suppose there really are things and they can be perceived by our senses. Then (a ) some of those things will be perceivable by one sense only and others by another sense only; and since one sense cannot perceive objects proper to another sense, a system of audible signs will not permit communication regarding things perceivable only by sight, and so on for the other senses. In any case, (b ) those supposed things are not identical with any signs one might use to communicate about them, and so one could never convey the things themselves to another person but only the signs.
Moreover, (c ) even if one could produce signs exactly representing those supposed things, he could not communicate those signs to another person, for the signs themselves are things, and no one can have in his mind the same things that someone else has in his mind at the same time. Finally, (d ) any signs we might use would have to be formed as a result of our perception of those supposed things, but since genuine knowledge of a cause cannot be gained from its effect, no knowledge of those things could be communicated by means of any signs. Occasionally in arguments (a ), (b ), and (c ) Gorgias seems, like Jonathan Swift's Laputans, to have sophistically confused talking about things with handing them around; but not all his paradoxes of communication are transparent, and some passages in Plato and Aristotle suggest that Gorgias's arguments may have helped to shape their semantic theories.
Conventionalism and Naturalism
The oldest surviving arguments in support of a particular semantic theory may be those attributed very late in antiquity to Democritus (fl. 420 BCE), perhaps presented originally in his book On Words. He is supposed to have offered the following four considerations in support of his position that the relation between names and things named is conventional (θέσει ) rather than natural (ϕύσει ): (a ) the occurrence of homonyms, that is, one and the same name for things different in nature; (b ) the occurrence of synonyms, that is, different names for one and the same thing; (c ) the occurrence of name-changes while the thing named remains the same in nature; (d ) the nonoccurrence of verbal analogies corresponding with real analogies, for instance, there is a verb analogous to the noun "understanding" but none analogous to "justice."
In all probability no philosopher ever held a thoroughgoing semantic naturalism, although there are traces of tendencies in that direction in the doctrine attributed to Pythagoras and in the assumptions that appear to underlie the work of Prodicus and Gorgias. The opposition of naturalism and conventionalism as semantic theories forms the point of departure for the development of Plato's semantics of names in the Cratylus. Much of the significance of the Cratylus and of ancient philosophy of language generally has been obscured, from antiquity onward, by the confusion of this semantic issue with a dispute over the origin of language in which "naturalism" and "conventionalism" were the principal doctrines. In that dispute, however, it was not the naturalist but the conventionalist position that was preposterous, conventionalism in that context being the claim that language first arose as a result of agreements among men or because some especially powerful individual compelled those around him to use his names for things. There are, of course, implications for semantics in theories about the origin of language, but neither Plato nor any other ancient philosopher of the first rank failed to distinguish between the two inquiries.
The oldest surviving work of any kind on language is Plato's Cratylus (probably written about 388 BCE). The main topic of this dialogue is the nature of the relation between names and things named.
At the beginning of the Cratylus a kind of semantic naturalism is attributed to Cratylus and a kind of semantic conventionalism to Hermogenes. All that is said about naturalism at the outset is that it seems unintelligible, and the first serious undertaking is a discussion of the conventionalism advanced by Hermogenes in these words:
I cannot be persuaded that there is any correctness of name other than convention [ξυνθήκη ] and agreement [δμολογία ]. For it seems to me that whatever name anyone gives to a thing is the correct one, and if someone changes that name for another, the later one is no less correct than the earlier—just as when we change the names of our slaves. For no name has been generated by nature for any particular thing, but rather by the custom [νόμῳ ] and usage [ἔθει ] of those who use the name and call things by it. (384 c–d)
There is nothing in this conventionalism we have not already seen in the Democritean arguments except the claim that "whatever name anyone gives to a thing is the correct one," and Socrates immediately asks whether this claim is intended to apply to private persons as well as to nations (385a). Hermogenes fails to appreciate the difference, and when, as a result, Socrates is on the point of showing that this subjectivist claim destroys the possibility of distinguishing between true and false statements, Hermogenes tries to salvage it by suggesting an analogy between arbitrary individual name-giving and different natural languages (385d–e). The picture presented is that of a conventionalist who recognized that the existence of different autonomous natural languages was strong confirmation of his position and was then so carried away as to produce a doctrine of autonomous idiolects, evidently reasoning as follows: Just as the Greek word for horse is no more and no less correct than the Persian word for horse, so there is no basis for correcting a Greek who should decide to use the Greek word anthropos where other Greeks use hippos and vice versa.
The conventionalism presented as a basis for discussion in the Cratylus is entirely plausible except for the obviously untenable doctrine of autonomous idiolects. One consequence of the doctrine is that at any given time a given thing (or type of thing) has just as many correct names as there are people who name it differently (385d). This suggests some sort of Protagorean skepticism in its author, but Hermogenes is ready to agree that "things have some fixed reality of their own, not relative to us or caused by us" (386d). Socrates uses this admission to show the necessity of objectively correct names. There are real things, he says, and real things are not subject to our whims. We recognize that we cannot do certain jobs involving real things simply by fiat. We must make the correct moves, using the correct tools, and the correct tools for a given job cannot be generally described as the first ones anyone may choose (386e–387b). Now in the use of language, names are our tools, and we employ those tools in doing two essential jobs plainly involving real things: "teaching" (communicating the truth) and "classifying things according to their natures" (387b–388b). If "whatever name anyone gives to a thing is the correct one," we clearly have no chance of succeeding in communicating the truth to one another or in developing classification schemes that will "carve reality at the joints."
The destruction of the doctrine of autonomous idiolects leaves a gap in conventionalism, a gap that was there in any case but that would not have been so easily seen if Plato had not thus deliberately marred this conventionalism in order to call attention to it. Not just anyone is an arbiter of the correctness of names; but then "who does provide us with the names we use?" (388d). The answer is derived from the sounder portions of Hermogenes's conventionalism, in which he claimed that custom or law generates our names for things. This suggests that the arbiter of custom, or the lawgiver (νομοθέτης ), may be identified as the name-maker (ὀνοματουργός ) (388d–389a). The "law-giver" is Plato's personification of a recognized stipulative linguistic authority, more nearly like the French Academy or the Oxford English Dictionary than like an individual—Solon, for instance.
This refurbished conventionalism is adequate as far as questions of pronunciation, word order, and usage are concerned; these can be settled by having recourse to the recognized authority. The question raised by the criticism of autonomous idiolects, however, was not, "how do we determine which names are accepted? " but, rather, "how do we determine which names are correct? " Plato took the two questions to be distinct and made his most important contribution to the semantics of names in answering the second of them. The development of his answer may be traced out as follows.
If the refurbished conventionalism is to do any more than offer an account of the phenomena of a language, it must be augmented by part of Cratylus's naturalism, which was originally stated in these three claims: "(a ) for each of the things that really exist there is a correctness of name that has been produced by nature; (b ) that is not a name which some people agreeing together to give as a name do give as a name, uttering a bit of their voice in accordance therewith; but (c ) there is a kind of correctness of names that is the same for all, both Greeks and barbarians" (383a–b).
At the beginning of the dialogue this position was taken to be unintelligible because it was thought to be in competition with conventionalism as an account of the phenomena of a language. Claim (b ) does seem to justify the view that the theory is just a wrongheaded account of that kind. Temporarily ignoring claim (b ), Socrates proceeds to show that this naturalism makes sense as an account of the conceptual underpinnings of all languages. The fact that the word for horse in Greek is "hippos" and in Latin "equus" shows that different linguistic authorities are operative in different natural languages. Both those words are perfectly acceptable, intertranslatable names for horse; and what makes them so is the fact that each of them embodies in different marks (or sounds) a single "ideal name," which belongs to horse "by nature," whose correctness has been produced by nature, and which is the same for all, both Greeks, who say "hippos," and "barbarians," who say "equus" (389c–390a). That single ideal name cannot be the type of which occurrences of "hippos" or of "equus" are the tokens, since it is "the same for all." Nor can it be identified with what Plato called the form of horse, for although the form of horse may be the ideal horse, there is nothing of which it could conceivably be a name. Instead, the ideal name embodied as well in "equus" as in "hippos" is the correctly framed concept horse, and the difference between the two words is merely the difference between two equally good notations. To say that the concept is framed correctly is to say that it is the concept of the form rather than of individuals participating in that form; to say that its correctness has been produced by nature is to say that it somehow resembles the form. The correctly framed concept horse is a logically proper name of the form of horse; it is the ideal name for which all the words correctly translatable into English as "horse" are various notations.
Plato goes on to develop and apply this theory along the following lines. If we should come across a natural language the speakers of which owned horses and cows but had only one name for both species, or had no single name for horse, using instead an indifferently ordered string of names for legs, head, tail, and so on, then we should have a genuine case of incorrect names. The speakers of that language would be laboring under the influence of incorrectly framed concepts, concepts that fail to carve reality at the joints. Thus, we avoid incorrect names (such as "phlogiston") to the extent to which philosophy and science (personified by Plato as "the dialectician") have provided us with a correct conceptual schema (390c–e).
But the embodiment even of correctly framed concepts in the evolving phenomena of a natural language will sooner or later lead to the development of homonyms and synonyms, which, although not incorrect, are infelicitous for the purposes of science and philosophy. Such infelicities could be avoided if we were to construct a precise, consciously designed concept-notation for the use of philosophers and scientists (421e–423e, 424d–425a). And even if we do not or cannot actually construct it, the notion of a perfectly systematic embodiment of correctly framed concepts may serve as an ideal against which to measure the adequacy of technical language (435c). Thus, the frequently recurring project of an ideal language is to be found for the first time in the very first extant treatise on language.
Perhaps the single most unusual feature of this remarkable semantic theory is the doctrine of the ideal name. Within the Cratylus itself the identification of the ideal name with the correctly framed concept is not explicit, although it is clearly implied. That implication is strengthened by the many passages in other dialogues in which Plato did treat concepts as a kind of name—for instance, Theaetetus 189e, 206d; Sophist 263e; Philebus 38e–39a; Phaedrus 276a.
Cratylus's naturalism and Hermogenes's conventionalism are so expressed in the dialogue as to give every appearance of being simply Plato's devices for raising semantic questions. Each of them contains an obvious, completely gratuitous overextension. (Later in the dialogue [428a ff.] Cratylus's claim [b ] goes the way of Hermogenes's autonomous idiolects.) Neither position alone is remotely plausible or likely to have been actually held by any philosopher, but each of them contains an essential ingredient of Plato's own semantics of names.
the parmenides and the sophist
Plato's other major contributions to semantics occur in the later dialogues Parmenides and Sophist, in which he goes beyond the doctrine of the Cratylus in undertaking the connected tasks of (1) giving an account of the semantics of such names as lack existent bearers, (2) refuting the Parmenidean doctrine that false statements express nothing, and (3) giving an account of the semantics of simple statements.
(1) In Parmenides 160b–161a there is an attempt to state three necessary conditions for the meaningfulness of a denial of existence. (The example actually employed is the hypothesis "if a One does not exist," which is eminently generalizable.) If we are meaningfully to say of x that it does not exist, then (a ) "there is knowledge of" x (since "otherwise the very meaning of … '[x ] does not exist' would be unknown"); (b ) x is "something different from other things"; (c ) "this non-existent [x ] has the characters of being that, and something, and of being related to this, or to these, and all other such characters.… If it does not exist, there is nothing against its having many characters; indeed it must [have many characters] if it is this [x ], and not another, that does not exist. If what is [said] not to exist is neither the [x ], nor this, and the statement is about something else, we ought not so much as to open our lips." These three interdependent conditions do not seem inconsistent with the semantics in the Cratylus, and much of what was to be brought out later in the Sophist is already implicit in them—for instance, the distinction between existential and predicational occurrences of "is" ("if it does not exist, there is nothing against its having many characters").
(2) When Parmenides or Plato speaks of expressing nothing, he means saying nothing meaningful, rather than saying nothing at all. This is implied in Parmenides' fragments and is quite plain in Plato, when he says, for example, "Must we not assert that [a man] is not even expressing anything when he sets about uttering the words 'a thing that is not'?" (Sophist 237e; Cratylus 429e). Those words constitute what Parmenides called a false name. A true, or meaningful, name is one having an identifiable existent bearer, a name that signifies something real; and there is no sharp semantical distinction between true names and true statements—"If we are speaking the truth, evidently the things we are speaking of must be" (Parmenides 161e).
Thus the Parmenidean doctrine is that false statements are meaningless, or that truth and meaningfulness are indistinguishable. Although its scope was never restricted, the doctrine makes most sense when applied to statements of the form "x exists," with which Parmenides was preoccupied. Such a statement, he would say, either is true or expresses nothing. In order to preserve the possibility of falsity, even in the limiting case of such statements, Plato had to question the Parmenidean dictum and establish that what is not has being in some respect (Sophist 241d), which he does in the complex, important doctrine of the interweaving of the Forms (252e–259c). However, his most direct answer to the Parmenidean doctrine is developed in his semantics of statements, an account based directly on the ontological theory just mentioned, since "any discourse we can have owes its existence to the weaving together of Forms" (260a).
(3) "Now, remembering what we said about Forms, … let us consider words in the same way.… Words that when spoken in succession signify something, do fit together, while those that mean nothing when they are strung together do not" (261d–e). "Now a statement never consists solely of names spoken in succession, nor yet of verbs apart from names" (262a). Thus "the simplest and shortest possible kind" of statement is exemplified in "Theaetetus sits" or "Theaetetus flies," "because … it gives information about facts or events; … it does not merely name something but gets you somewhere by weaving together verbs and names. Hence we say it states something" (262d). "Whenever there is a statement, it must be about something" (262e). Both the examples above are about one and the same existent thing, the bearer of the name "Theaetetus," but the second is a combination of name and verb in which "what is different is stated as the same or [as is actually the case in this example] what is not as what is," and anything "answering to that description finally seems to be really and truly a false statement" (263d).
In the Parmenides and Sophist, then, Plato not only extended semantics for the first time beyond the consideration of names to that of statements but, in doing so, also distinguished between meaningfulness and truth, showing for the first time that truth depends not merely on names but on certain syntactically regular combinations of verbs and names. It should be noted, however, that he does seem to have taken meaningfulness as the necessary and sufficient condition of grammaticalness.
Plato's semantics of statements may be better appreciated against the background of the semantical doctrines of his contemporaries Antisthenes the Cynic (fl. 390 BCE) and Stilpo the Megarian (fl. 340 BCE). Beginning with the familiar "two names, two bearers" view, Antisthenes managed to reject all predication, on the grounds that what the subject named was one thing and what the predicate named was quite another, and to accept only identity statements of the form "x is x " or analogies of the form "x is like y." Stilpo, too, rejected predication, perhaps on ontological grounds, since he insisted on "the unity of being" and may have thought that this could be expressed only in strict identity statements.
Aristotle's primary interest in language was naturally that of a logician, and while his writings contain many passages on semantic questions, there is relatively little developed theory. His semantics of words (he treats of more than just names) is like Plato's in many respects and is to be found mainly in De Interpretatione, Chapters 1–3. There he presents, with little or no argument, the following account of signification.
Although there are different natural languages, the people who use them are confronted with the same extramental things. The mental modifications arising from that confrontation are likenesses (ὁμοιώματα ) of the things, and they are thus the same for all men too. Within a given natural language, written words are conventional symbols (σύμβολα ) of spoken words. (Aristotle was no doubt unaware of ideographic notations.) The spoken words are, in turn, related to the mental modifications, first of all as symptoms, or natural signs (σημει̑α ), of them—that is, of the presence of mental modifications in the speaker. More important, the spoken words are related to the mental modifications in the same way that written words are related to spoken words, as symbols of them. Just as written words constitute a conventional notation for (or embodiment of) spoken words, so do spoken words for mental modifications. Discussions of these passages have almost invariably failed to recognize the first of the two relations between spoken words and mental modifications as distinct and have confused the second relation with that of name to bearer.
It seems that, according to this account, words signify things in virtue of serving as symbols of mental modifications resembling those things. What sorts of "things" can words thus be made to signify? Not much is said on that question in De Interpretatione, but in Categories (Ch. 5) and Sophistical Refutations (Ch. 22), for example, various words are said to signify (σημαίινειν ) "a certain this," "a qualification," "a substance of a certain qualification," "passivity," "a certain relation to something else," "a quantity," and so on. More important, "'man' and every common name signifies not a certain this, but a quality or a relation or a mode (or something of the sort)" (Sophistical Refutations 178b38).
Ambiguity, Aristotle maintained, is theoretically unavoidable, for since "names and the sum-total of formulas [λόγοι ] are finite while things are infinite in number … the same formula and a single name must necessarily signify a number of things." This will, however, give us no trouble unless "we think that what happens in the case of the names happens also in the case of the things, as people who are counting think of their counters," which are in a one-to-one correspondence with the things counted (Sophistical Refutations 165a5). Although this passage is part of a warning against sophisms of ambiguity, when taken together with the preceding passage it seems to constitute an injunction against seeking the bearer of a common name, as Plato and so many of Aristotle's successors did. A single individual is the bearer of many names in that they are all correctly predicable of it, but "we do not identify having one meaning with being predicable of one thing, since on that assumption even 'musical' and 'white' and 'man' [all of which are predicable of Socrates] would have one [and the same] meaning" (Metaphysics 1006b15).
The principal kinds of words recognized by Aristotle were the name (ὄυομα ) and the verb (ῥη̑μα —"predicate" is possibly a more accurate translation). He described them both as the smallest conventionally significant units, incapable of being true or false independently. A name without a bearer, such as "unicorn," is neither "false" (as some of his predecessors had claimed) nor nonsignificant; and a name combined with "is," "was," or "will be" always produces something true or false. A verb uttered by itself is a name, but it additionally signifies time and "some combination, which cannot be thought of without the components." Because of the latter additional signification, a verb "is always a sign (σημει̑ον ) of things being said of something else" (De Interpretatione 16b24, 16b7).
"Non-man" names nothing definite and so is not strictly a name; analogously, "does not walk" holds indifferently of all sorts of existents and nonexistents. These negated words Aristotle put into the separate categories of "indefinite names" and "indefinite verbs." "Inflections," such as "man's," are not names either, since they produce nothing true or false when combined with "is," "was," or "will be"—nor is "walked" a verb; it is an "inflection," because it signifies additionally "a time outside the present." In "complex names," such as "lifeboat," the parts are significant, but not independently, since, for example, "life" in this occurrence cannot be given an ordinary interpretation (De Interpretatione, Chs. 2 and 3). Finally, there are "connections" (σύνδεσμοι ), words and phrases that "make many things one" (Rhetoric III, 12; 4), which seem to include particles, conjunctions, prepositions, and idiomatic phrases of several sorts and which in one passage of doubtful authorship are said to be nonsignificant (ἄσημοι ) (Poetics, Ch. 20). (The "connections" are almost certainly the direct ancestors of Priscian's "syncategoremata," which figured prominently in medieval semantics.) This loosely organized classification, vaguely consistent at best, is based on a tangle of semantic and syntactic considerations, but it does contain important advances—for instance, in the treatment of names without bearers and complex names.
Aristotle's semantics of sentences is concentrated in but by no means confined to De Interpretatione, Chapters 4–8. Names have no significant parts and complex names no independently significant parts, but a sentence (λόγος ) must have independently significant parts. (This is surprising in view of the fact that in a highly inflected language such as Greek there are frequent occurrences of one-word sentences—"I-walk," "he-walks," and so on.) "Every sentence is significant—not as a tool but … by convention," he maintained (16b33), apparently dissociating his view from Plato's in Cratylus 386d ff. Plato, however, was talking about names, not sentences, and Aristotle here seems to have gratuitously set aside an insight into the semantics of sentences that was later to be developed by the Stoics. Some sentences, such as "prayers" and future contingents, are neither true nor false according to Aristotle, and he set the pattern for nearly all logicians thereafter when he put such sentences aside and attended solely to the always true or false "statement" (λόγος ἀποϕαντικός ).
Aristotle maintained that among the independently significant parts of a statement there must be either a name or an indefinite name and a verb or an inflection of a verb arranged in such a way that the whole "signifies something about something." It is only in such a combination that there is truth or falsity, and, as Aristotle put it in the early chapters of De Interpretatione, it looks as if he took the combination in question to be one of words. In Metaphysics 1027b23, however, he said that "falsity and truth are not in things … but in thought ; while with regard to simple concepts and essences falsity and truth do not exist even in thought … but … the combination and the separation are in thought and not in things," and he suggested something similar in De Interpretatione, Chapter 14 (and elsewhere), as well (cf. Plato, Republic 382b). There is no evidence that Aristotle distinguished consistently or clearly between sentences and what later philosophers called propositions or judgments, but such passages indicate at least his sense of the difficulty in locating truth in strings of words, or in a direct relation between strings of words and arrangements of things.
Aristotle seems sometimes to have considered the communicative capacity or public character of a locution as a criterion of its having independent significance. Thus in Metaphysics 1006a21 he remarked that if a man "really is to say anything," he must "say something that is significant both for himself and another "; and in support of his claim that when a verb is uttered by itself it is really a name and signifies something, he noted that on such an occasion "the speaker arrests his thought and the hearer pauses."
The nature of the Stoics' philosophy of language is the most tantalizing problem in the history of semantics. We know enough of it to say that it was by far the most intricate and probably the most insightful theory of its kind in antiquity and for centuries afterward; but we cannot be certain what its details were, and even its leading principles are sometimes obscured by vague or conflicting testimony. Those Stoics who had most to say about language were, naturally, the logicians, and the difficulty of determining the exact character of what they had to say stems from the fact that none of the many works of the Stoic logicians is extant. The best surviving sources (which date from almost five hundred years after the period of greatest development in Stoic logic and semantics) are Sextus Empiricus, Outlines of Pyrrhonism, Book II, and Adversus Mathematicos, Book VIII; and Diogenes Laërtius, Book VII. Under these circumstances it is seldom possible to assign a particular doctrine to a particular Stoic, but much of the best of their logic and semantics is very likely to be the work of Chrysippus (c. 280–206 BCE).
Under the Stoic division of philosophy into physics, ethics, and logic, logic was divided into rhetoric and dialectic, and dialectic further divided into an account of language (περὶ τη̑ς ϕωνη̑ς ) and an account of things signified (περὶ τω̑ν σημαιυόμενων ). Both these subdivisions contain material relevant to semantics. In their account of language the Stoics distinguished vocal sound generally, "which may include mere noise," from the sort that is articulate (ἔναρθρος ), that is, capable of being embodied in written symbols (ἐγγράμματος ). Articulate sound, in turn, may be nonsignificant—for instance, "blityri"—or significant (σημαντική ); but for any articulate sound to be considered a sentence (λόγος ) it must be significant and a product of someone's reason (Diogenes Laërtius 7.55–57).
Within that same branch of their dialectic the Stoics recognized five kinds of words and distinguished their semantic or syntactic functions. They were the first who clearly separated (1) names, such as "Socrates," from (2) appellatives (προσηγορίαι ), such as "man." (Cf. Aristotle's similar but significantly different distinction in De Interpretatione, Ch. 7.) A name "points out a kind proper to an individual," while an appellative "signifies a common kind." (3) A verb "signifies a predicate"; (4) a conjunction "binds together the parts of a sentence"; (5) an article (possibly also what would now be called a relative pronoun) serves to "distinguish the gender and number of nouns" (Diogenes Laërtius 7.58). Thus the function of conjunctions and articles is purely syntactic, the semantic function of (proper) names is different from that of appellatives (or common names), and the appellative and the verb—the standard ingredients of the simplest kind of logicians' sentence—have one and the same kind of semantic function. The appellative occurring in a sentence signifies a subject and the verb a predicate or "something attachable (συντακτόν ) to the one or more subjects."
Obviously the division between the accounts of language and of things signified was not exclusive, but the transition from the one account to the other as the Stoics conceived of them may be seen in the claim that all we utter (προϕέρειν ) is sounds, while what we express (λέγειν ) is matters of discourse (πράγματα ), or lekta —"expressibles" (Diogenes Laërtius 7.57). It is the doctrine of the lekton around which the Stoics organized their account of things signified. In its novelty, importance, and difficulty that doctrine overshadows all the considerable remainder of their philosophy of language.
Probably the clearest introduction of the notion of the lekton is the one to be found in these passages from Sextus:
The Stoics … said that three things are linked together: (1) what is conveyed by the linguistic sign [τό σημαινόμενον ], (2) the linguistic sign itself [τὸ σημαι̑νον ], and (3) the object or event [τὸ τυγχάνον ]. Of these the linguistic sign is the sound—e.g., "Dion"; what is conveyed by the sign is the matter of discourse indicated thereby, which we apprehend over against and corresponding to our thought (while the barbarians do not understand, although they do hear the sound); and the object or event is the extra-mental entity—e.g., Dion himself. Two of these are corporeal—viz. the sound and the object or event—and one is incorporeal—viz. the matter of discourse conveyed by the linguistic sign, the lekton. (Adversus Mathematicos 8.11–12)
They also say that the lekton comes into being as corresponding to a rational presentation [λογικὴν θαντασάν ], and that a rational presentation is one presenting something that can be set forth in a sentence. (Adversus Mathematicos 8.70)
The kind of lekton associated with the name "Dion" was said to "stand in need of completion," and the only categories cited for such completable lekta were subjects and predicates. In order to be "set forth in a sentence," the completable lekta must enter into the composition of a lekton "complete in itself." The kind of complete lekton regularly associated with a standard subject-predicate sentence was called a statement (ἀξίωμα ), and truth or falsity was ascribed to it, not to the sentence. Statements naturally received most attention from the Stoic logicians, but they recognized many other varieties of complete lekta as well. The fact that they did so strongly suggests that they had developed other categories of completable lekta too, for most of the other complete lekta cannot be analyzed into subject and predicate. Among the other varieties were commands, prohibitions, yes-no questions, questions requiring more than "yes" or "no," curses, prayers, doubts ("Can it be that life and pain are akin?"), and quasi statements ("How like to Priam's son the cowherd is!") (Adversus Mathematicos 8.71–73; Diogenes Laërtius 7.65–68).
Since these are categories of incorporeal lekta rather than of sentences, they cannot be identified with strictly grammatical categories. Moreover, although some of the distinct lekta do correspond to grammatically distinct sentences—for instance, the two kinds of questions—many of them do not. The Stoics' own example of the kind of lekton called a doubt was expressed in what is grammatically a yes-no question; commands and prohibitions get expressed in declarative as well as in imperative sentences, and occasionally both may be expressed in one and the same sentence, for example, "Abstain from strong drink." Thus Plutarch reports, in his attack on the Stoics, that "they themselves maintain that those who forbid say one thing, forbid another, and command a third. For he who says 'you ought not to steal' forbids stealing and commands not stealing at the same time as he says you ought not to steal" (On the Contradictions of the Stoics 1037d).
As many as three different complete lekta may, then, be associated with a single sentence, and those lekta are obviously not to be identified as thoughts or intentions on the part of the speaker or hearer. Nor does it seem likely, despite Plutarch's way of presenting the doctrine, that all the complete lekta associated with a given sentence must be expressed whenever the sentence is uttered. Besides being far-fetched, that requirement would ignore the sense of expressibility built into the Stoics' technical term "lekton." Instead, the Stoic doctrine seems to be that a number of distinct linguistic jobs—such as stating, commanding, prohibiting—can be performed by means of a single sentence, depending on which of the complete lekta associated with that sentence is actually communicated on a given occasion of its use. Thus the three lekta associated with the example given by Plutarch may be presented as (1) the statement that one ought not to steal, (2) the command not to steal, and (3) the prohibition of stealing. It seems to be a discovery of the Stoics (and their greatest contribution to semantics) that the explication of meaning involves not only the things we talk about and the thoughts we express but also the jobs we do by means of language alone.
Of the Stoic semantic triad—linguistic sign, what is conveyed thereby (the lekton), and external object or event—the Epicureans accepted only the first and third, ascribing truth and falsity directly to spoken sentences (Sextus Empiricus, Adversus Mathematicos 8.13). This rejection of the lekton is typical of the Epicureans' mistrust of any doctrine that went beyond the evidence of the senses. Plutarch describes them as "completely doing away with the category of lekta, leaving only words and objects and claiming that the intermediate things conveyed by the signs simply do not exist" (Adversus Coloten 1119F), but there is also a vague suggestion that they may have found it convenient to provide "lekta" as dummy referents in one important kind of case. "They deprive many important things of the title of 'existent,' such as space, time, and location—indeed, the whole category of lekta (in which all truth resides); for these, they say, are not existents [ὄντα ], although they are something [τινά ]" (Adversus Coloten 1116B).
In stating their atomist metaphysics, the Epicureans were of course obliged to use such words as "space" and "time," and it looks as if they may have clumsily attempted to provide referents for them by associating only such words, or sentences containing such words, with lekta. However, even if they did maintain that there are two kinds of referents for words, real things and lekta, the latter to be invoked only in case the former are unavailable and the words are indispensable, there is nothing of the Stoic lekton in their doctrine.
Aside from this putative special use of special lekta, the Epicureans' philosophy of language seems to have remained remarkably faithful to their fundamental sensationalism. Epicurus (341–270 BCE) had originally stressed the importance of beginning the study of physics (one of the main branches of Epicurean, as of Stoic, philosophy) by ascertaining the ultimate referents (ὑποτεταγμ́ενα ) of words, "so that our proofs may not run on untested indefinitely nor the terms we use be empty. The primary intent (ἐννόημα ) of every term employed must be clearly seen and ought to need no explication" (Diogenes Laërtius 10.37–38); and he went on to claim that these ultimate referents must then be "our sensations," "present impressions," "actual feelings." These are always veridical since their immediate causes are the eidola, and thus "the agent productive of each of them is always entirely presented and, as being presented, it is incapable of being productive of the presentation without being in very truth as it appears.… Thus the visible object not only appears but actually is as it appears.… The presentations that occur are, then, all true" (Sextus Empiricus, Adversus Mathematicos 7.203–204). The square tower in the distance appears round, but its round appearance is itself a physical object, an eidolon detached from the tower and impinging on the apparatus of sight. If I say, then, "the tower is round," I may (and in this case I shall) be mistaken, since the tower is not the immediately presented object. But if I say "the appearance (or presentation) of the tower is round," I cannot be mistaken (at least not in that same way). Although this is a move in the direction of protocol sentences, it does not rest on a distinction between sense-datum and physical-object sentences, since for the Epicureans the protocol sentence was only a more correctly framed physical-object sentence.
No full account of this Epicurean reductivism is extant, but its principle is clearly operative not only in their physics but in their ethics as well, where one pervasive maxim for the avoidance of fear is to reduce the mysterious (for example, in natural phenomena) to what is actually presented and to describe it in terms precisely associated with the features of the actual presentation (see, for instance, Diogenes Laërtius 10.78 ff.).
Epicurus's followers evidently took the nature of the relation between words and sensations as a major topic for psychological theory. The notion of prolepsis is at the center of the Epicurean psychology, and in at least one of its many guises prolepsis seems to be the act of associating a word with a typos, or outline left in the mind as the result of repeated similar presentations. One example of prolepsis is the identification "such and such a thing is a man"—"for no sooner is the word 'man' uttered than we think of the typos of man in accordance with the prolepsis, the senses having led the way. As a result, the immediate referent of every name is apparent.… Nor would we have given a name to anything if we had not first come to know the typos of it in accordance with prolepsis" (Diogenes Laërtius 10.33).
The typos, then, is the immediate referent (τὸ πρώτως ὑποτεταγμένον ) of every name. When a name is used and understood, an act of prolepsis at once brings the corresponding typos to mind. (Since nothing but sensation can produce a typos, the need for some other sort of referent in the case of words such as "space" and "time" is apparent.) If this was indeed the core of the Epicurean semantics of words, it must be judged inferior to many other theories of its kind in antiquity.
Epicurus himself and the Epicureans generally had a good deal to say about the origin of language, and what they said usually makes better sense than most such accounts in antiquity. Lucretius (99–55 BCE) is especially good on this topic, which he treated at some length in Book V of his poem. Among his more novel and interesting achievements is an extended series of arguments against the theoretical possibility that language (as distinguished from a language) might have been invented (De Rerum Natura 5.1041 ff.).
See also 190. GREECE and GREEKS ; 207. HISTORY .
- a person who lived before the Flood. —antediluvian, adj.
- an interest in the customs, art, and social structure of earlier peoples and civilizations. —antiquarian, n., adj.
- the field of description of antiquities. —archaeographical, adj. —archaeographer, n.
- archaeology, archeology
- the scientific study of human remains and artifacts. —archaeologist, archeologist, n. —archeologie, archaeologic, archeological, archaeological, adj.
- the study of the language and culture of ancient Assyria. —Assyriologist, n. —Assyriological, adj.
- the principles or style of classic art or literature. —classicist, n.
- the study of ancient Egyptian language, history, and culture. —Egyptologist, n. —Egyptological, adj.
- the deciphering and interpreting of ancient inscriptions. —epigraphist, epigrapher, n. —epigraphic, epigraphical, adj.
- the study of Etruscan civilization, especially its artifacts and language. —Etruscologist, n.
- Ancient Greek culture and ideals. —Hellenist, n.
- Rare. the research and composition of treatises about relics. —lipsanographer, n.
- the study of mummies.
- paleography, palaeography
- the study of ancient writings, including inscriptions. —paleographer, palaeographer, n. —paleographic, palaeographic, adj. papyrology the study of ancient writings on papyrus. —papyrologist, n.
- a person who lived after the Flood. —post-diluvian, adj.
- the policies and actions distinctive of ancient Rome.
an·tiq·ui·ty / anˈtikwitē/ • n. (pl. -ties) 1. the ancient past, esp. the period before the Middle Ages: the great civilizations of antiquity. ∎ a specified historical period during the ancient past: cameos dating from classical antiquity. ∎ (usu. antiquities) an object, building, or work of art from the ancient past. 2. great age: a church of great antiquity.