As the overview showed, “what is language” was born between a teacher and a student. This chapter rereads how each answered it, and where they diverged.
The two positions Plato opened
In the fourth century BC, Plato’s dialogue Cratylus gives two speakers, Hermogenes and Cratylus, opposite positions. For Hermogenes, a name’s correctness is nothing but “convention and agreement.” For Cratylus, names are fixed “by nature.” Socrates takes neither side outright, and instead rebuilds the argument between them: a name is a tool (organon) for dividing things, made by a “lawgiver” and judged by a “dialectician” (article 038). By the dialogue’s end, Socrates rejects the strong form of the natural-correctness view and concludes that “knowledge should be sought from things themselves, not from names” (article 041).
So the Cratylus never actually settles the nature-versus-convention question. Instead, it closes by offering a third path: treat the name as a tool.
Aristotle’s answer — turning convention into an axiom
After Plato’s death, Aristotle spent roughly twenty years studying at his Academy. In the opening lines of his logic textbook, De Interpretatione, defining the noun, he settles the question his teacher left open:
A noun is a spoken sound significant by convention, without time.
No argument accompanies it. It’s simply set down as a premise (article 043) — as if the question his teacher’s dialogue had spent decades probing were already closed. But Aristotle doesn’t simply inherit Plato’s answer. What he rejects is the very tool model Plato’s dialogue arrived at in the end.
A sentence signifies by convention, not as a tool.
Words are not instruments that carve up the world; they are symbols (symbolon) that carry meaning by convention. Where Plato built up an elaborate case for “the name is a tool,” Aristotle replaces it with a single line. He keeps the conclusion — convention — while discarding the reasoning that got Plato there.
Shifting the unit — from name to sentence
But Aristotle’s real move lies before that substitution. Plato’s Cratylus had, from start to finish, been about the correctness of names (onoma, single words) — is “anthropos” (human) the right word, is a name well-made like a weaver’s shuttle, and so on.
Aristotle shifts the very unit of the question. Neither a noun (onoma) nor a verb (rhēma) carries truth or falsity on its own. Truth and falsity live only in the sentence (logos) formed by combining them — specifically, in the “assertion” (apophansis) (article 045). Neither “human” nor “white” is true or false by itself. Only “the human is white,” as a sentence, can be evaluated for truth.
This is a quiet but enormous turn. Plato asks “is this name correct?” Aristotle asks “is this sentence true?” The first question — as the Cratylus itself demonstrates — leads back, however far you trace etymologies and sound-mimicry, to a shrug at convention. The second can be built into a system. And in fact, starting from this classification of sentences, Aristotle constructs affirmation, negation, contradiction, and the square of opposition (article 046) — the skeleton of logic that would still be in use over two thousand years later.
What was carried forward, and what was cut away
What Aristotle inherited was the conclusion itself — that language works by convention. What he cut away was the path that got Plato there: examining a name as a tool for dividing up the world. In its place, he confines the sign’s arbitrariness to sound and letter alone, resting on the strong premise that the “affections of the soul” (pathēmata tēs psychēs) behind them are the same for everyone, natural and universal (article 043). Arbitrary only at the surface of the sign, universal at the concept beneath it — this is exactly the premise that Humboldt, and later Sapir and Whorf, would challenge more than two thousand years on.
So the question Plato had carefully left open hardens, in Aristotle’s hands, into an unexamined axiom. “Language works by convention” would go essentially unquestioned for over two thousand years — until Saussure rebuilds it, once more, as a first principle under the name “the arbitrariness of the sign” (Chapter 4). But before that, others would try to build something entirely different on top of this same axiom — in a monastery outside seventeenth-century Paris.