Anyone who has spent enough time reading Plato - especially the shorter dialogues, often thought to present Socrates in a more-or-less historically accurate light - will notice that the conversations tend to wind up stymied and inconclusive. Be the topic piety, friendship, knowledge, or color, Socrates' notorious "what is [x]?" questioning style strongly tends toward rejecting every single answer that gets offered. The reason is that no straightforward account we give of anything really seems to capture what that thing is.
Plato seems to have taken the implication to be that what things are is better captured by a kind of intellectual intuition than by linguistic attempts at offering formulas. This implication is, however, hard to find openly expressed as following from the breakdown of discourse. It's almost as though Plato (or Socrates) didn't want to let us in on the secret to baffling and seducing one interlocutor after another: the game is rigged, and so long as we use language we can't answer "what is [x]?" to full satisfaction. (We have to, as it were, point upward.)
Isn't this a kind of trick? A Wittgensteinian response to Socratic questioning might run like this: What Socrates demonstrates, over and over, is the fact that in many cases a linguistic expression (like "knowledge" or "the good") cannot be replaced in all contexts by another one (like "justified true belief" or "that at which all things aim"). The expressions asked about simply play by different rules than any of the expressions used in trying to answer. On this reading, what Socrates succeeds in showing us is not the elusiveness of realities that outstrip our linguistic grasp, but rather the futility of trying to use one bit of language to substitute for another while preserving all our semantic intuitions.