On the surface, the Cratylus is about two competing theories concerning the “correctness of names” (orthotes onomaton), a naturalist and a conventionalist one. For the most of the dialogue, the reader has the impression that the objective of Socrates’ debate with Hermogenes and Cratylus – the conventionalist and the naturalist respectively – is to weigh the pros and cons of each of these two theories in order to reach a somewhat intermediary position by taking both of them into account. Thus, in his discussion with Hermogenes, Socrates reinforces naturalist claims whereas, in discussing with Cratylus, he points out that conventionalism cannot be altogether abandoned. He is seemingly heading towards a basically naturalist theory (names imitate the things as they are in themselves) with a conventionalist ingredient added in (for some linguistic phenomena there is no other explanation than usage, i.e. convention).

Nevertheless, the outcome of the discussion as a whole is a different one. Socrates ends up disapproving of Cratylus’ take on the correctness of names, and more particularly of his claim that names provide knowledge of things. We cannot rely on names, Socrates concludes, if we are searching for knowledge of things. As a matter of fact, the analysis of compound names into primitive ones, and of the latter into their elements, eventually demonstrates that, without an extra-linguistic criterion of truth, we cannot decide which names have been formed correctly, i.e. so as to imitate the things they refer to, and which have not. This outcome does not dismiss the naturalist principle as such – it still may be the case that names somehow imitate the nature of what they refer to – but it undermines the epistemological claim that we can draw on names in order to acquire knowledge of the nature of things. On the contrary, to establish whether names are correct or not, we have to know the things “from themselves rather than from their names” (439b7–8).¹ To put it in more general terms, parts of speech as such, no less than their elements and compounds, are not the place to learn the truth, either about names or about the things they denote. It thus turns out that the dialogue as a whole does not aim at finding a middle way between the two competing theories, but, much more radically, at attacking the very idea (not unfamiliar to modern readers) that the study of language is the privileged or even the only way of acquiring knowledge. What is at stake is the epistemological value of words.

Despite the fact that Cratylus still declares himself unshaken in his stance, the last section of the dialogue (437a–440e) casts a long shadow on the series of etymologies (393a–427d) that Socrates had previously provided in order to substantiate the naturalist theory, much to Cratylus’ satisfaction (428c). This does not come totally as a surprise. There were warning signs from the very beginning of Socrates’ astonishing performance, most notably his tracing the inspiration that befell him back to his having listened to Euthyphro earlier that day, as well as his declaring that, tomorrow, he and his interlocutors will need to purify themselves from it (396d–e). The important thing about these etymologies is that, on the whole and particularly up to the turning point in 437a, they fit a conception of reality as being in motion and of things as fundamentally differentiating themselves from each other in the way they move. While this point may not be obvious from the start, it gradually becomes plain until it is expressed most fully in the etymology of the words for “truth” (aletheia) and “being” (on, ousia) in terms of motion – “truth” signifying “divine roaming” (aletheia = theia ale) and “being” meaning “going” (on = ion, 421b-c) – and generalised in the following statement: “We say that names signify for us being on the assumption that everything moves and is carried about and flows” (436e2–4).² The turning point consists in the analysis of words for “knowledge” (episteme), “certain” (bebaion), “enquiry” (historia), “convincing” (piston) and “memory” (mneme) in terms of rest (stasis), instead of motion (kinesis), with the general conclusion that, in such examples, names signify things at rest (menonta ta pragmata, 437c6–8). The two alternatives, one based on the assumption that things are in motion, the other on the assumption that they are at rest, create a dilemma, which leads to Socrates’ above-mentioned conclusion, according to which knowledge of the nature of things cannot be acquired from their names. More precisely, and more detrimentally to the naturalist concept of the correctness of names, the two accounts show that those who first established the names may have done so based on a false assumption, so that the names that we analyse deceive us (439c). As a matter of fact, the assumption of a general flux of things, which seems to underlie the signification of words previously analysed by Socrates, does not allow for the possibility of knowledge (gnosis) since the latter requires stability (439d–440e). Should knowledge exist, affirms Socrates, there must be rest on the side of the act of knowing, as well as on the side of the object known. This is why Socrates prefers to maintain that the things themselves – the beautiful itself, the good itself, etc. – are not in motion but at rest, i.e. that they always remain what they are. Having arrived at this conclusion, which Cratylus refuses to accept, the basic question is no longer about the epistemological value of words. It is now about the ontology allowing for a theory of knowledge. Can there be knowledge at all on the Heraclitean assumption of the flux of things? Must we not rather assume, for the sake of knowledge, that things in themselves are at rest?

This is a question which is given great prominence in other, presumably later,³ dialogues such as the Theaetetus and the Sophist. The Theaetetus contains an in-depth discussion of a flux ontology and its consequences for epistemology. It offers a broader but also a much more fine-grained picture of such an ontology than what we have learnt about it from the Cratylus. The points of contact, however, are of interest. The Cratylus attributes the view that everything is in flux to the “Heracliteans and many others” (440c2). The Theaetetus is more specific in attributing it to “all the wise men (sophoi) except Parmenides,” more particularly to Protagoras, Heraclitus and Empedocles, and tracing it back to the founders of comic and tragic poetry, that is Epicharmus and Homer respectively (152e). Homer is presented as the commander-in-chief of this whole army (153a) and Protagoras as the most prominent theorist of this crowd (152a ff.). The secret of his theory consists in the claim that there does not exist a single thing itself in itself because everything is in a process of becoming (aei gignetai) instead of a state of being (esti, 152d–e). A deeper analysis of becoming, attributed to the more refined among the adherents of a flux theory, based on the principle that everything is motion and nothing else (156a), analyses motion into two concomitant powers, a power to act (dynamis poiein) and a power to be acted upon (dynamis paschein), so that nothing can be one thing in itself, but everything always results from the interaction of these two powers, neither of which allows the other to stop changing (156a–157c).⁴ Since this account applies also to sense-perception and all kinds of judgement, cognitive acts are always changing processes, too, with no criterion available which would make it possible to differentiate between them in terms of truth and falsehood (157d–162a). The more elaborate picture of the ontology of flux in the Theaetetus thus serves a purpose similar to the rougher, if by no means less playfully presented, version of it in the Cratylus: the ontology of flux does not offer a firm ground for knowledge that distinguishes between truth and falsehood.

In the Sophist, an explicit sequel to the Theaetetus, the same theory of flux resurfaces. A class of refined flux-ontologists appears on stage – or rather the Eleatic Stranger imagines that there might be such a kind of reformed corporealists (246d) – who hold that being consists in nothing else than the capacity (dynamis) to act (poiein) and to be acted upon (pathein). The former is the power to make another thing different, the latter is an aptitude of something weaker to suffer from something more powerful (247d–e). These reformed corporealists – or “dynamists” as we may call them – are contrasted with the friends of Forms who distinguish between being (ousia) and becoming (genesis) and hold that intelligible and incorporeal Forms alone constitute true being, whereas bodies are nothing but a flux of becoming (246b–c). These two ontologies, the dynamist and the idealist, as presented by the Eleatic Stranger, have in common the theory of flux conceived of as an interaction between conjoined capacities of acting and being acted upon. Yet whereas the dynamists take such an interaction to be the only being there is, the idealists relegate it to the realm of becoming, which is opposed to that of true being from which the capacities to act and to be acted upon are banished (248c). Both theories are tested for their capacity to account for knowledge and both fail this test. The friends of Forms are the first to be brought on trial (248d–249b). If they hold that our body is in communion (koinonei) with becoming by means of sense-perception, whereas our soul is in communion (koinonei) with true being by means of reasoning, and if they concede that, in both cases, to be in communion or communicate (koinonein) consists in acting and being acted upon, they will be forced to acknowledge that the act of knowing (gignoskein), by which the soul is in communion with true being, is acting (poiein) and that, correspondingly, in being known (gignoskesthai), true being is acted upon and thus subject to motion. Since this would be in contradiction with the theory of being the idealists profess, they will prefer to reject altogether this kind of relationship as an explanatory pattern for knowledge. But in so doing – so at least the Eleatic Stranger puts it – they will withhold, together with motion, also life, soul and thinking (phronesis) or intellection (nous) from true being. There will be an unmoved being at rest (on … akinēton hestos), in splendid isolation as it were, but there will be no intellection (nous) of anything whatsoever by anyone whomever, if intellection occurs in the soul and the soul moves and lives. On the other hand, and this is the verdict on the dynamists (249b–c), if we allow that everything is moved and changed (pheromena kai kinoumena panta einai), we will deprive the realm of being of the very character of identity (tauton touto, to kata tauta kai hosautos kai peri to auto) which – so at least the Eleatic Stranger suggests – seems to suppose that there is rest. The problem is that where there is no identity there is no intellection. This is why such a theory must be fought by all rational means, the Eleatic Stranger says, since it would make all knowledge, thought and intellection vanish (249c).

As in the Cratylus and the Theaetetus, we also learn in the Sophist that the ontology of flux is no good basis for epistemology. But unlike the accounts of the former dialogues, and more disturbingly, the discussion in the Sophist seems to suggest that the dream Socrates spoke about at the end of the Cratylus – the dream about the beautiful itself, the good itself and each one of ta onta (439c–d) – may not be a viable alternative either. In this respect, the aporia in which the gigantomachia of the central part of the Sophist results resembles the aporiai in which the old Parmenides, in the eponymous dialogue, involves the young Socrates. The latter too, like Socrates in the Cratylus and the friends of Forms in the Sophist, puts forward a very elementary theory of Forms (Parm. 128e–129e) which posits their existence, each one in itself, without specifying the nature of the relationship that exists between them and sensible objects, without asking whether there are relations among the Forms themselves and without raising the question of the relation between the soul and the Forms. In the Parmenides, the character Parmenides does raise these questions (130a–135b) but gives no satisfactory answer to them. Instead he engages in a dialectical performance (135b–166c) the epistemic value of which is as ambiguous as that of Socrates’ etymological performance in the Cratylus. Each performance seems to constitute a deliberate concoction of science and trickery, of truth and falsehood.

In contrast to this, in the Sophist the Eleatic Stranger offers a way out of the perplexity concerning the fact that knowledge appears impossible both on the assumption that everything is in motion and on the assumption that being is at rest. The solution, to put it briefly, consists in the thesis that there are different kinds of relationships between the Forms themselves and that these relations allow for bridging the gap between being and becoming. They do so in so far as they allow for the being of things other than being itself, i.e. for participation in being, so that not only Forms other than the Form of being, but also things or phenomena other than intelligible Forms participate in being. The Forms of being, identity and difference, by participating mutually in each other, play a major role in the distribution of being among everything other than the Form of being itself. Motion and rest, on the other hand, being genuine opposites one to another, do not accede to this lofty all-pervasive status so that neither of them can claim to constitute a fundamental characteristic of being. On this account, not only must the ontology of flux be dismissed, but so too must an ontology that divides the whole of reality between rest and motion as two opposites. Rather, motion and rest have to be re-evaluated in their respective roles within the structured whole dominated by the Forms of being, identity and difference.⁵

This structured ontology, as we may call it,⁶ provides a new basis for a theory of language which will differ from the attempted theory of the correctness of names in the Cratylus. There are two main points that differentiate the one from the other. The first consists in the fact that the Sophist does not focus on the name (onoma), but on the sentence (logos) whose basic elements are the name (onoma), in the function of the grammatical subject, and the verb (rhema), in the function of the grammatical predicate: the rhema indicates an action, the onoma signifies the agent (262a–c).⁷ The simplest combination (symploke) of onoma and rhēma in this sense is “so to say the first and smallest of utterances” (σχεδὸν τῶν λόγων πρῶτός τε καὶ σμικρότατος, 262c6–7). Not every combination of words is a meaningful utterance (logos), only a sentence gives rise to such speech. But even among meaningful utterances, not every combination of onoma and rhema is right. Sentences – and compounds of them – have a specific quality. They are true or false with respect to the things they refer to. They are true if they say (legein) about the things that are that they are (τὰ ὄντα ὡς ἔστιν) and they are false if they say about the things that are not that they are (τὰ μὴ ὄντα ὡς ὄντα, 262e–d).⁸ The important idea is that the truth or falsehood of a logos results from a right or wrong combination of elements suited to form it, the smallest elements of such a kind being the grammatical subject (onoma) and predicate (rhema) and the smallest combination of them being the sentence (logos). This implies that words outside sentences are neither true nor false. This implication is, of course, fatal for Cratylus’ view on the correctness of names, as well as for the kind of etymologies Socrates produces in the Cratylus.

According to this theory, logos, as a combination of nouns and verbs, has a mimetic nature, but what it imitates are not directly the things (pragmata), as Cratylus’ naturalist theory has it, but thinking (dianoia) or judgment (doxa), the latter being nothing other than a completed act of thinking (dianoias apotelesma) which constitutes either an affirmation or a negation and which is either true or false (Soph. 263b–264b). The analysis of speech in terms of a combination (sumploke, plegma) of components functioning as grammatical subject and predicate¹² must apply to thinking as well, which constitutes such a plegma or symploke prior to speech. One may ask: If the logos is a combination of onoma and rhema, both of them being words formed by articulated sounds, what are the components combined by dianoia and doxa? This we are not told. Certainly, the components of thought which are mirrored in speech by onoma and rhema are not themselves words in the sense of articulated sounds. Nor are they mental representations (phantasia) since representation, the Eleatic Stranger says, is a combination of sense-perception and judgement (symmeixis aistheseos kai doxes, Soph. 264b). One might think that they are the intelligible Forms (eide or gene) which themselves constitute a plegma or symploke prior to thinking. After all, does not the Eleatic Stranger say that “logos came about through the combination of Forms one with another”?¹³ Forms themselves may be what the dialectical science (dialektike episteme) knows in so far as it grasps them correctly in their mutual relationships (253b–e),¹⁴ but they will probably not be reducible to the contents of knowledge.¹⁵ Be that as it may, thinking (dianoia), judgment (doxa) and representation (phantasia), no less than speech (logos), admit of falsehood (263d–264d, see also 241e). This is what makes them imitations of things (mimemata ton onton, 264d4). If the logos is an image, whether true or false, of dianoia/doxa, the latter is likely to be an image, whether true or false, of things (ta onta), an image constituted by the combination of certain elements of thought – we may call them concepts, but Plato does not – which imitate the Forms and are themselves imitated by onomata kai rhemata. Thus in the last resort, the logos does imitate ta onta, but it does so through the intermediary of dianoia/doxa, so that, to put it more accurately, speech is an image of an image of ta onta.

We are not given more details about the way in which dianoia/doxa relates to ta onta and logos to dianoia/doxa. In particular, we are not told what exactly the imitation of ta onta by dianoia/doxa and of the latter by speech consists in. Nor are we told how exactly the logos relates to the phantasia as a combination of doxa and aisthesis, and more precisely how this very combination comes about. To know this is important in order to understand how logos relates not only to the objects of thought, but also to the objects of sense-perception.¹⁶ But, from the discussion in the Sophist, there emerges what I would like to call, for the sake of convenience, the principle of structural analogy. Let me formulate this principle as follows: If the Forms that constitute being in its complexity (the symploke ton eidon) are terms in a network of combinatory relationships, as the Eleatic Stranger’s structured ontology has it (259a–b), so are the corresponding components of dianoia/doxa (concepts in the soul) and so are the correspondent components of logos (onomata kai rhemata uttered by voice). If nothing else, at least this structural analogy provides the mimetic link between the three strata of being, thinking and speaking, the first of which is paradigmatic for the subsequent two. Even the composition of words made of vowels and consonants – though no word in itself is either true or false – repeats this pattern, since the connection of vowels with consonants to form syllables is not arbitrary, but governed by combinatory rules (253a). The formal analogy between the network of interrelated terms at the level of logos with the one at the level of dianoia and, again, the one at the level of eide is a version of the mimetic principle which lies at the heart of Cratylus’ naturalist theory. In this formalised version, transferred from the onoma to the logos as the minimal unit which can be true or false, the mimetic principle is to some extent even compatible with the conventionalist theory. The basic elements of speech (grammata, stoicheia) and the rules governing their combinations may, in fact, be a matter of social convention. The fact that different societies use different languages with different sets of such elements and different rules for their combination testifies to this fact. But the capacity of these languages to produce statements, i.e. affirmations and negations, which are either true or false, does not stem from convention. It is based on the mimetic principle which applies, first, to the relation between thinking and being and, second, to the relation between speaking and thinking.¹⁷ Since the object of thought is itself constituted by definite relations between definite items (Forms), thinking assimilates them by putting analogous items (concepts) in analogous relations and speaking imitates thinking in producing again analogous relations between analogous items (words). It is through their analogous structures that speech imitates thought and thought imitates the combination of Forms.

In comparison with the Cratylus, the theory of language put forward in the Theaetetus and the Sophist marks a double shift: from onoma to logos and from the relation onoma–pragma to the relation logos–dianoia–eide. The alternative between an ontology of flux and an ontology of unchanging Forms which underlies the debate about the correctness of names in the Cratylus resurfaces with vehemence in the Theaetetus and the Sophist. The Sophist tries to resolve this conflict by assigning motion and rest their proper places among the greatest genera, thus replacing the primitive ontology of single Forms with a structured ontology of relationships among them. Most importantly, this ontology provides a basis for the theory of falsehood of judgement and speech and, more generally, for the theory of imitation in which, unlike in Cratylus’ view, imitation can be true or false. On this new basis, the conception of speech as a kind of mimesis can be reformulated. Since speaking and judging consist in a kind of imitation and since imitation, as such, allows for falsehood, neither speech nor judgement can be the criterion of truth. Only what is not itself an imitation can provide this criterion.