This question is about the Greek equivalent of sentences like
I do not fear the Greeks, but I do fear the Romans.
Socrates didn't write dialogues, but Plato did.
These sentences use or imply the same verb in both parts of a parallel construction, of which the first is negative and the second is positive.
These sentences can be translated literally as
τοὺς μὲν Ἕλληνας οὐ φοβοῦμαι, τοὺς δὲ Ῥωμαίους φοβοῦμαι.
Σωκράτης μὲν διαλόγους οὐκ ἔγραψεν, Πλάτων δὲ ἔγραψεν.
What I'm wondering is whether the second verb can be left out, to give
τοὺς μὲν Ἕλληνας οὐ φοβοῦμαι, τοὺς δὲ Ῥωμαίους.
Σωκράτης μὲν διαλόγους οὐκ ἔγραψεν, Πλάτων δέ.
It seems that this is possible at least sometimes. Searching the TLG for "δέ." at the end of a sentence yields some relevant examples:
ἔστιν δὲ τοῖς ἁμαρτάνουσι περὶ μυθολογίαν καθαρμὸς ἀρχαῖος, ὃν Ὅμηρος μὲν οὐκ ᾔσθετο, Στησίχορος δέ (Plato Phaedrus 243) "for those who have sinned in matters of mythology there is an ancient purification, unknown to Homer, but known to Stesichorus" (tr. Fowler)
εἰσὶν οὐ πόρρω τῶν εὐελπίδων, χείρους δ’ ὅσῳ ἀξίωμα οὐδὲν ἔχουσιν, ἐκεῖνοι δέ (Arist. Nic. Eth. 1117a) "they come very near to those whose bravery rests on a sanguine temperament, though inferior to them inasmuch as they lack self-confidence, which the sanguine possess" (tr. Rackham)
(The last example isn't a μέν ... δέ, but it's the same idea.)
These examples make me think that at least 4 above is good Greek. I'm still not sure about 3, though, because in that sentence the δέ is not final, and it's much harder to search for cases of this construction with non-final δέ.
So, are there examples like 3, with non-final δέ? And more broadly, how productive, restricted, or idiomatic is the use of verb elision as in 3-4, as opposed to repeating the verb as in 1-2? Are there pragmatic or other constraints on this construction? Since this is a somewhat difficult question to get a handle on, I'd also be interested in answers that simply give potentially relevant examples.