In his book A Stylometric Study of the New Testament, Anthony Kenny writes:

The Fribergs divide the Greek conjunctions of the New Testament into three classes. The two clauses or propositions of structures which are joined by a conjunction may be intended to have equal prominence in their context, or one may be given greater weight than another. Accordingly, the conjunctions which link them may be classed as co-ordinating (giving equal weight), as subordinating (introducing a clause less prominent than that to which it is linked), or as hyperordinating (introducing a clause more prominent than that to which it is linked). Thus every conjunction will be tagged either CC, CS or CH. (Kenny, 32).

Based on what I've learned about αλλα, my understanding is that αλλα should always be, in the Fribergs terminology, "hyperordinating". But an examination of their analysis shows that of the 638 NT instances, the Fribergs tag 553 of them hyperordinating (aka "superordinating"), 84 of them as coordinating, and one instance as subordinating. This post examines Rev 2.9, the lone "subordinating" αλλα in the NT.

Οἶδά σου τὴν θλῖψιν καὶ τὴν πτωχείαν, ἀλλὰ πλούσιος εἶ, καὶ τὴν βλασφημίαν ἐκ τῶν λεγόντων Ἰουδαίους εἶναι ἑαυτοὺς καὶ οὐκ εἰσὶν ἀλλὰ συναγωγὴ τοῦ Σατανᾶ. (Rev 2.9, UBS4)

I know your tribulation and your poverty (but you are rich) and the slander of those who say that they are Jews and are not, but are a synagogue of Satan. (Rev 2.9, ESV)

There are two αλλα in this verse; according to the Fribergs, the "subordinating" αλλα is the first, τὴν πτωχείαν, ἀλλὰ πλούσιος εἶ ("and your poverty (but you are rich)"). The second is "hyperordinating" (what would be expected).

So, what is it about the first instance that is different? Is it because the αλλα clause is seen as secondary to the primary clause; an in-stream parenthetical comment that doesn't seem to add much to the larger structure? The larger point seems to be built around the comparison between "your tribulation and your poverty and the slander ...", with the party doing the slandering further qualified as not Jews (though they confess to be Jews) but instead a "synagogue of Satan". In this latter instance, "the synagogue of Satan" has the prominence (indeed, Fribergs mark it as hyperordinating, thus it being the "more important" of the conjoined items). This is fairly standard with αλλα, the following statement offering correction to the first one and highlighting the correction.

[Note: The following paragraph has been added subsequent to the original post]

Upon further reflection, I believe the compared clauses are "I know your affliction and poverty" and "but you are rich". The corrective response is not simply to the note of poverty; "affliction" and "poverty" are one unit, joined by και (and perhaps too the genitive phrase following the αλλα?); the correction is to that unit, not simply to being poor.

[Back to the original post]

I'd argue the same thing for the earlier instance. In the context of the two conjoined items, "your poverty, but you are rich" it is the encouragement of the author to his audience. This is the letter "to the angel of the church in Ephesus", thus these words are from Christ to that church. While they find themselves temporally poor, they are to be encouraged that in fact they are rich in what matters. While their circumstances are tough, those circumstances will change—indeed, they already have begun to change. To me, this as well seems to be the basic "corrective" use of αλλα, correcting the first item and marking the correction as the important, salient bit in the comparison of elements.

I'd have to say that, at least with the first instance of αλλα in Rev 2.9, the Friberg's morphology should mark it as "conjunction, superordinating (hyperordinating)" instead of "conjunction, subordinating".

This as well serves as a case to show once again why I don't like such morpho-syntactic labels applied at the word level; it leads many who use such data to think there is something about αλλα itself in this instance that is "subordinating" or "hyperordinating". In reality, the conjunction morphology (part-of-speech) is just a convienent place to hang this item when it rightly belongs at a higher level of the annotation. But since "morphologies" only consider words as data tokens, they only have words to hang such data on—whether it rightly belongs on the word (as several "morphological" criteria do) or whether it rightly belongs at a higher level of the discourse (marking phrasal relations, clausal relations, or discourse-level relations).

While I am fairly sure that the Fribergs don't intend to mark αλλα itself as somehow morphologically producing a "hyperordinating", "coordinating" or "subordinating" result, less-informed use of such resources could easily make (and attempt to defend) such a conclusion. This is a common problem, and it is visible everywhere in everything. Calvin would (rightly) dispute against many who claim to be "Calvinists" as having misrepresented his thought; Darwin would also (rightly) dispute many who claim to be "Darwinists".

Anyway, enough from me. I don't know that I'll work through the 84 "coordinating" instances of αλλα to show how I would instead consider them to be "hyperordinating". But you never know. Maybe. In case you want to peek at them, here are the references:

Mt 24:6; Mk 3:27; 4:22; 6:9; 11:32; 13:7, 24; 14:28, 49; 16:7; Lk 6:27; 7:25, 7:26; 11:42; 16:21; 21:9; 23:15; 24:21, 22; Jn 1:31; 3:28; 5:42; 6:22, 36, 64; 8:26; 11:11; 14:31; 15:21, 25; 16:2, 4, 6, 7, 20; Ac 10:20; 19:2; 26:16; Ro 4:2; 5:15; 6:5; 10:2, 16, 18, 19; 11:4; 1Co 2:9; 3:2; 4:3, 4; 6:6, 11(3x); 1 Co 6:12(2x); 7:7; 8:7; 9:12; 10:5; 12:24; 15:35, 40, 46; 2Co 1:9; 7:11(6x); 8:7; 11:1; Ga 4:8; 4:23; Eph 5:24; Php 1:18; 2:17; 1Ti 1:16; Heb 3:16; Jas 2:18; 1Pe 3:16; Re 2:6; 10:7.