As I mentioned earlier, a blog post I wrote a few years back was footnoted on p. 151 in the proceedings from the 2006 LIABG symposium.
The article is titled "Conjunctions and Levels of Discourse", by Stanley E. Porter and Matthew Brook O'Donnell, running from p. 145 to p. 156 and is basically the same paper that was presented to the European Associate of Biblical Studies Annual meeting in Budapest, Hungary, from Aug. 6-9, 2006.
I never thought I'd be footnoted in a serious grammatical discussion; I suppose this is proof that the blogosphere can have some sort of influence/impact on current discussions in all areas of Biblical studies. Even blog posts by a motivated autodidact (read: no graduate degree held) such as myself.
The context is in the section of the article titled "3. The Greek Conjunction System", speaking of "Vertical Axis of Levels of Discourse". This has to do with the level of discourse on which the conjunction may function. Is it just a joiner of words or word groups, or is the joining further up (vertically) the ladder of discourse (join words, join word groups, join clauses, join clause complexes, paragraphs, discourses or whatever).
The post that Porter & O'Donnell refer to is about 1Th 5.15. In that post, I don't really make any statements, I just ask a lot of questions about how αλλα functions based on what happens with constituent order before and after αλλα. I was asking if αλλα might be doing something more in that context.
Porter & O'Donnell's point on p. 151 is that αλλα functions "only at the lower levels (but perhaps not the lowest level of the word). This conjunction joins word groups in Mt 9.13 ... clauses in Mk 4.17 ... and clause complexes in 1Th 5.15." (Porter & O'Donnell, Proceedings, 151).
Now, I would say that what I suggested in 2006 concerning αλλα in that instance was wrong (specifically the section breakdown near the top of the post); v. 15 is a standard μη .. αλλα instance, where v. 15a is a foil that serves to make 15b more prominent in the discourse. On the segmentation of units, I think Porter & O'Donnell's suggestion in the footnote (vv. 14-15a, 15b-18a, 18b-21ff) is fine. I have no problem using the term "clause complex" instead of "subparagraph" or "paragraph"; those are all strange terms that mean what one wants them to mean anyway. The idea that αλλα is somehow indicating a relationship between two separate clause complexes (vv. 14-15a & 15b-18) is the important bit for my concerns with αλλα at present.
My confusion today stems in the sentence that runs from p. 151 through 152. In the above-quoted portion of p. 151, they limit αλλα to word group, clause, and clause complex connections. But on the sentence running from pp. 151-152, they note:
An initial analysis indicates that there are a limited number of conjunctions that function at all of the levels of discourse. These include only και, δε, αλλα, and some of the negative conjunctions, such as ουδε and μηδε. All of the rest of the conjunctions are more circumscribed in the linguistic levels at which they may be used (Porter & O'Donnell, Proceedings, 151-152)
Porter & O'Donnell's only level above clause complex is paragraph (I think, they don't seem to explicitly list them but they mention paragraph above clause on p. 151); and the only level below word group is word.
[Corrected, 2008-06-04] Porter & O'Donnell list the following discourse levels along the vertical axis: word, word group, clause, clause complex, paragraph and discourse (p. 151)
I would say that there are instances of αλλα that join paragraphs, and Porter & O'Donnell seem to acknowledge this as well (based on p. 152). That's good, because I plan to have examples of αλλα functioning at the paragraph level in my ETS paper (unless my preliminary analysis changes between now and then, which it may). But this discussion gives me some more support in positing αλλα as a paragraph conjoiner in certain contexts.