τί οὖν, ἦ δ᾽ ὅς, οὐκ ἐρωτᾷς;
ἀλλ᾽ ἐρήσομαι, ἦν δ᾽ ἐγώ.
Then why, he said, do you not ask?
But I am asking, said I. What is the function of δε here?
transitional or copulative: and he said… and I said…
or it may be a fixed expression that is not used without δε
Yes ἦν δ᾽ ἐγώ is a fixed phrase, where the δ’ has no copulative function. Similarly with fossilized 3rd person ἦ δ᾽ ὅς, not confined to Plato.
It is certainly fossilized, but it seems clear to me that the δέ is still doing something.
Probably the easiest illustration for this difference is that while you see ὁ … ἔφη, you don’t see ὁ … ἦ δ᾽ ὅς, because the δ᾿ in the phrase marks speaker contrast. If there is a name specified, it’s always after: ἦ δ’ ὃς ὁ Σωκράτης.
What’s happening is that Plato deploys ἦ δ᾽ ὅς and ἦν δ’ ἐγώ (and ἦ δ᾽ ἥ) to mark the change of speaker and make the dialogues easy to follow. He doesn’t always mark the change, and sometimes he signals it by ἔφη ὁ or otherwise, but you can see that overall ἦ δ᾽ ὅς is deployed in a subtly different way than plain ἔφη.
To demonstrate how this goes with an example from the Apology:
“Ὦ Καλλία,” ἦν δ’ ἐγώ, “εἰ μέν σου τὼ ὑεῖ πώλω ἢ μόσχω ἐγενέσθην, εἴχομεν ἂν αὐτοῖν ἐπιστάτην λαβεῖν καὶ μισθώσασθαι ὃς ἔμελλεν αὐτὼ καλώ τε κἀγαθὼ ποιήσειν τὴν προσήκουσαν ἀρετήν, ἦν δ’ ἂν οὗτος ἢ τῶν ἱππικῶν τις ἢ τῶν γεωργικῶν· νῦν δ’ ἐπειδὴ ἀνθρώπω ἐστόν, τίνα αὐτοῖν ἐν νῷ ἔχεις ἐπιστάτην λαβεῖν; τίς τῆς τοιαύτης ἀρετῆς, τῆς ἀνθρωπίνης τε καὶ πολιτικῆς, ἐπιστήμων ἐστίν; οἶμαι γάρ σε ἐσκέφθαι διὰ τὴν τῶν ὑέων κτῆσιν. ἔστιν τις,” ἔφην ἐγώ, “ἢ οὔ;”
“Πάνυ γε,” ἦ δ’ ὅς.
“Τίς,” ἦν δ’ ἐγώ, “καὶ ποδαπός, καὶ πόσου διδάσκει;”
“Εὔηνος,” ἔφη, “ὦ Σώκρατες, Πάριος, πέντε μνῶν.”
We start with the notification that he has switched from his defense speech to reporting a dialogue where he is the first speaker, ἦν δ’ ἐγώ.
At the end of the first section, he reminds the listeners that he is still the one speaking, but as contrast would be inappropriate, he just says: ἔφην ἐγώ.
Speaker switches: ἦ δ’ ὅς
Speaker switches: ἦν δ’ ἐγώ
And finally, the back and forth is established, and the hearer does not need so much emphasis on who is speaking, and so we go back to: ἔφη
That’s how it seems to me, anyway. But it’s plain enough if you compare usages of ἔφη and ἦ δ’ ὅς in general that they are being deployed slightly differently (even if the difference generally disappears in translation).