Plato's Cratylus: Language, Reality, and the Limits of Naming

Plato’s Cratylus is a dialogue about whether names are correct by nature or correct only by convention: whether “dog” is somehow the right sound for a dog because of what a dog is, or whether it is right simply because Greek speakers agreed to use it that way. It is one of Plato’s works that gets less attention than the Republic or the Symposium, but it is the closest thing Plato wrote to a philosophy of language, and it ends in a place many readers find surprising: Socrates spends most of the dialogue arguing for the natural-correctness side, then turns around near the end and undermines it.

This essay works through the dialogue’s structure: the two positions and the two men who hold them, Socrates’ long and often playful etymologies, his eventual argument that names can be wrong and that things must be known through themselves rather than through words, and what all this has to do with Plato’s theory of Forms and with Heraclitus.

What question does the Cratylus ask?

The dialogue opens with Hermogenes and Cratylus already arguing, and Socrates is brought in to settle the dispute. Cratylus holds that there is a natural correctness to names, built into the nature of the things named, that holds for Greeks and non-Greeks alike. Hermogenes finds this absurd. He thinks names are fixed by convention and agreement, and could just as easily have been assigned differently, since whatever a community agrees to call a thing is what it is correctly called (383a–384d). This is the ancient version of a live linguistic debate: is meaning grounded in the world, or is it a social contract that could have gone another way?

Who are Cratylus and Hermogenes?

Hermogenes is identified in the dialogue as the brother of the wealthy Callias, and he appears elsewhere in Plato’s circle as an associate of Socrates. Cratylus is more significant for the history of philosophy: Aristotle reports that Plato, in his youth, kept company with Cratylus and adopted Heraclitean views from him, before Socrates (Metaphysics I.6, 987a32). If that report is accurate, and most scholars treat it as at least broadly reliable, then the character arguing for a fixed, natural correctness of names in this dialogue was one of Plato’s own early teachers, which gives the dialogue’s staged debate an unusual personal edge.

Does Socrates think names are natural or conventional?

For a long stretch of the dialogue, Socrates appears to defend something like Cratylus’s position, but by way of a third option neither Hermogenes nor Cratylus proposed outright: that names are made correctly or incorrectly by a specialist craftsman, the name-giver or nomothetes, who fashions each name to fit the thing it names, rather in the way a weaver fashions a shuttle to fit its purpose, guided by someone who knows how to judge the work, the dialectician (388e–389a). This lets Socrates hold that correctness in naming is real and not arbitrary, without agreeing with Cratylus that it happens automatically or that every existing name is therefore correct.

Who judges whether a name is correctly made?

Socrates extends the craftsman analogy in a direction that matters for the rest of the dialogue. A shuttle-maker fashions a shuttle out of wood, but he does not decide, purely on his own judgement, what shape the shuttle should take. He works to a pattern supplied by the weaver, the person who will actually use the tool, because the weaver understands weaving and therefore knows what the shuttle needs to do. In the same way, the name-giver fashions names, but the person qualified to judge whether a name has been made well is not the name-giver himself. It is the dialectician, the person skilled in asking and answering questions, since using names correctly in argument is the real test of whether they have been made to fit their purpose (389a–390e).

This has a quiet but important consequence. It means correctness in naming was never, even on Socrates’ more sympathetic construction of Cratylus’s view, a property that names simply have by themselves, the way a natural kind has its properties. It is relative to a skilled user’s judgement, which already opens the door to the possibility Socrates presses hard later on: that a name-giver, however skilled, can get the job wrong, and that only someone who already knows the thing named is in a position to say so.

Why does Socrates offer so many etymologies?

The dialogue’s long middle section (397c–421c) is Socrates working through a run of etymologies, mostly of gods’ names and words for virtues and natural phenomena, tracing each one back to some descriptive root that supposedly explains why it fits its referent. Some of these are genuinely thoughtful. Others are stretched, punning, or frankly absurd, and Socrates himself repeatedly signals that he is improvising, crediting his sudden inspiration to Euthyphro and disclaiming full confidence in the results. Scholars disagree about how seriously Plato means this section to be taken: some read it as a real if playful attempt to illustrate how a natural-correctness theory might work in practice, others as an extended parody designed to show, by its own increasingly strained examples, that the whole approach cannot be trusted as a route to knowledge.

Socrates does offer a more general theory underneath the specific etymologies: individual letters can imitate qualities of the things they help name, so that rho, formed with a quick motion of the tongue, is well suited to words involving motion and force, and lambda suits smoothness and gliding (426c–427c). This is Plato’s closest approach to a theory of phonetic iconism, the idea that the sound of a word can directly mirror what it describes, rather than being purely arbitrary.

Why does Socrates turn against Cratylus at the end?

Having spent most of the dialogue building a case for natural correctness, Socrates pivots decisively from 428d onward. He gets Cratylus to admit that a name can, in principle, be given wrongly, and that a wrongly given name is still a name, just an inaccurate one. If that is possible, then names cannot be a perfectly reliable guide to the nature of things, because a false or badly made name would mislead exactly the person trying to learn from it. Socrates draws the conclusion directly: things must be learned and investigated through themselves, or through whatever is most akin to them, not through names, which are at best a derivative and potentially unreliable image of the thing (438d–439b).

This is the dialogue’s real turn, and it matters more than the etymological display that precedes it. Cratylus’s confidence that a name-giver, knowing the true natures of things, built that knowledge permanently into language, and that anyone can therefore recover the truth about the world simply by studying words closely enough, is exactly what Socrates dismantles. Language, on the dialogue’s final view, is a tool that can be used well or badly, and can go wrong, which means it can never substitute for direct knowledge of the things themselves.

What does the Cratylus have to do with the theory of Forms?

The dialogue’s closing pages connect the naming question to two of Plato’s largest commitments. First, if the objects named by words are always changing, as Cratylus’s own Heraclitean background would suggest, then no name could ever be stably correct for them, since the thing itself would have moved on by the time the name was fixed to it (439c–440d). Second, and this is where Socrates’ own view surfaces, if there is to be any knowledge at all, there must be some stable objects for that knowledge to be about, which points toward the unchanging Forms that anchor Plato’s theory of Forms elsewhere in the dialogues. Heraclitus’s picture of universal flux is treated here as a serious problem for knowledge, not just for naming: if everything really is in constant change, nothing can be truly known, named or otherwise, and Socrates leaves the question open rather than resolving it inside this particular dialogue.

What happened to Cratylus’s view afterward?

Aristotle reports, in a passage often read alongside the dialogue, that Cratylus later pushed Heraclitean flux to an extreme, becoming convinced that since everything is constantly changing, nothing can truly be said of it at all, so that in the end he thought one should say nothing and merely moved his finger, on the grounds that even to speak a single word claims more stability than reality can support (Metaphysics IV.5, 1010a). This is a report about Cratylus, not a claim Plato makes inside the Cratylus itself, and it should be labelled as such, but it fits neatly with the dialogue’s concern that Heraclitean flux, taken seriously enough, threatens to make language, and communication generally, impossible.

Hermogenes, Cratylus and Socrates compared

Position	Held by	Core claim
Conventionalism	Hermogenes	Names are correct by agreement alone; a community could have named things differently
Naturalism	Cratylus	Names are correct by nature, fitting the thing named, and hold true across all languages
Craftsman theory, then its limits	Socrates	A skilled name-giver aims at natural fit, but can succeed or fail; names can be wrong, so knowledge cannot rely on them alone

Why does the Cratylus still matter?

The debate this dialogue stages, nature versus convention in language, is not settled and was not meant to be. What the dialogue does settle, through Socrates’ argument against Cratylus, is a narrower but more important point: whatever the origin of words, they are not a shortcut to knowledge of the world. That conclusion sits comfortably with the wider Socratic method practised throughout Plato’s works, and with the intellectual world the dialogue depicts, one shaped by the Sophists’ interest in the power of words and by Parmenides’ insistence that being and thought are tightly bound together. How Socrates mentored Plato is visible here too, in the dialogue’s method: test a claim, follow it to an uncomfortable place, and be willing to abandon a position you spent pages defending once the argument demands it.

Centuries later, the linguist Ferdinand de Saussure would argue for the arbitrariness of the sign, that the link between a word and its meaning is a matter of convention within a language community, not a natural fit. The parallel to Hermogenes’ side of this dialogue is suggestive, though the two debates sit in very different intellectual contexts and should not be treated as the same argument in different dress. What the Cratylus offers is not a settled answer to how language works, but an unusually clear demonstration, from inside Plato’s own thought, of why he did not think language could ever be trusted to supply it on its own.