Cratylis as a first foray into philosophy of language.
The question: how do words have meaning? Note that random syllables do not have a meaning. But "Frege" or "philosophy" do. What's the difference?
One theory: stipulation--e.g. naming. E.g. baby: two people dub an arbitrary thing a string of syllables.
Plato considers the stipulation theory--has one of the characters
Recall that Cratylus was one of the (radical) Heracliteans ('can't step in the same river once')
Cratylus disagrees: words have meanings by nature.
Note this a familiar Sophistic distinction: nature and convention (as with morality).
So what is it to have a meaning by nature? Cratylus says that things have true names. Getting them right is saying something true; getting them wrong is not naming it at all (i.e. not saying anything at all). This good news for Plato, since he agrees with the relativist picture on which you can't say anything false.
But note this different from the Theaetetus reasons for absence of falsehood: it's a semantic reason why there can't be falsehood.
Socrates' reply to the stipulation theory: names, as man-made, have functions. So not just any words can be names. But what words are suitable names or not? Socrates' answer: the word that reveals something of the referent's nature. His evidence: Greek words whose etymology reveal the nature of what they refer to.
Objection: the etymologies are far-fetched. Interpretive dispute about whether we're supposed to take Socrates' etymologies seriously.
Note: proposal to remove letters to reveal "real" etymology?
Regress objection: if we etymologize, e.g., "god", to show how it derives from "run", then what about "run" itself?
Socrates' reply: some things have a 'natural' likeness to their referent--i.e., onomatopoeia. That explains all the roots of names. Get a theory of words originating in nascent Heraclitean ideas: people used sounds to signify different kinds of change.
This not an affirmation of Heraclitean theory; just an analysis of semantics of language that appeals to an alleged sociological fact about early prevalence of Heraclitean thought.
Socrates' reply to Cratylus (the actual Heraclitean): language can't be entirely natural. Saw, e.g., change of letters/sounds in the earlier reply. But word still functions after this. So rejects Cratylus' extreme nature-based account of meaning.
So words are representations of their referents, in Socrates' theory. That opens up, against Cratylus, the possibility of mismatch between the word and what it represents. So we get the possibility of mismatch again (note similarity of this to the way falsehood shown possible in the Theaetetus: there, a mismatch of wax imprints).
So Socrates trying to vindicate two uses of language: to communicate intentions (a la naming a baby) and to reveal the true nature of things (a la getting things right and wrong).