Socrates: Perhaps it will seem absurd, Hermogenes, to think that things become clear by being imitated in letters and syllables, but it is absolutely unavoidable. For we have nothing better on which to base the truth of primary names. Unless you want us to behave like tragic poets, who introduce a deus ex machina whenever they’re perplexed. For we, too, could escape our difficulties by saying that the primary names are correct because they were given by the gods. But is that the best account we can give? Or is it this one: that we got them from foreigners, who are more ancient than we are? Or this: that just as it is impossible to investigate foreign names, so it is impossible to investigate the primary ones because they are too ancient? Aren’t all these merely the clever excuses of people who have no account to offer of how primary names are correctly given? And yet regardless of what kind of excuse one offers, if one doesn’t know about the correctness of primary names, one cannot know about the correctness of derivative ones, which can only express something by means of those others about which one knows nothing. Clearly, then, anyone who claims to have a scientific understanding of derivative names must first and foremost be able to explain the primary ones with perfect clarity. Otherwise he can be certain that what he says about the others will be worthless. Or do you disagree?
Hermogenes: No, Socrates, not in the least.

Cratylus 425d1-426b2