I encountered this bit of interesting Greek in 4 Baruch 8.2-3:
καὶ ἐρεῖς τῷ λαῷ· Ὁ θέλων τὸν κύριον καταλειψάτω τὰ ἔργα τῆς Βαβυλῶνος. 3 καὶ τοὺς ἄρρενας τοὺς λαβόντας ἐξ αὐτῶν γυναῖκας, καὶ τὰς γυναῖκας τὰς λαβούσας ἐξ αὐτῶν ἄνδρας, διαπεράσωσιν οἱ ἀκούοντές σου, καὶ ἆρον αὐτοὺς εἰς Ἰερουσαλήμ· τοὺς δὲ μὴ ἀκούοντάς σου, μὴ εἰσαγάγῃς αὐτοὺς ἐκεῖ.
and you will say to the people, ‘Let him who desires the LORD leave the works of Babylon behind.’ 3 And (of) the men who took wives from them, and the women who took husbands from them, let those who hear you cross over, and take them up to Jerusalem; but (as for) those who do not hear you, you must not lead them there (translation from Charlesworth).
Note what we have here. It’s striking, first of all, that the translator here chose a translation more suitable to a third person imperative than a third person subjunctive for διαπεράσωσιν οἱ ἀκούοντές σου. Such a translation implies more volitionality on the part of the audience (the returning exiles), but the Greek actually gives agency to the spouses–οἱ ἀκούοντές σου.
Also then notice the person and semantic role shift on the part of the speaker. The left dislocation introduces a change in topic from those who heard (I would prefer the translation ‘obey’ personally) to those who did not hear. The accusative case of the left dislocation cross-references it with an argument in the clause itself: αὐτοὺς. But the actor and subject of the second subjunctive is now explicitly the returning Jews. The spouses are relegated to undergoer status in the clause.
In terms of negation scope, this clause is rather interesting. Usually, there’s a fairly strong correlation between (1) negation scope and (2) focal information. But in this case, the scope of the negation is over the pronoun αὐτοὺς, which cannot be treated as focal because of its status as a resumptive pronoun for the left-dislocation. Thus this clause functions as linguistic evidence that negation scope and information structure and distinct grammatical phenomenon and cannot be conflated or assumed to be identical.
Finding clauses like this is the means by which we are able to provide an empirical basis for our grammatical claims for ancient languages without (consistent) access to native speakers.