Just in case anybody missed this thread the first time round, all of Levinsohn's discourse annotations of the whole NT are still up for free access as PDFs at:http://www-01.sil.org/~levinsohns/BART.html
These are enormously helpful. For those who have access to Runge's analyses via Logos these works by Levinsohn are a great complement. On the whole their theoretical assumptions (in so far as I understand them) are very very similar, and the results are very similar too. However, there are differences - a small number of which are evaluative, but a number of differences represent their choices of what to highlight via their annotation schemes. I'm pretty sure users of Runge will be able to 'translate' Levinsohn's nomenclature by reference to Runge's DGGNT.
These works are very similar and in some ways mutually supporting, but here's a rough-and-ready list of *some* general differences I've noticed:
Levinsohn marks "over-encoding" of active participants, Runge does not.
Levinsohn marks "Dominant Focal Elements", Runge does not (though sometimes these work out as "main clause emphasis" in LDGNT)
Levinsohn marks more right dislocations than Runge, who sometimes marks appropriate examples as "forward pointing targets" of "forward pointing references".
Runge marks Nominitive circumstantial frames, Levinsohn does not
Runge marks overspecification, Levinsohn does not (NB - over specification =/= over-encoding)
Runge marks Thematic Address, Levinsohn does not.
I'm sure that there are more.
This is a recommendation of both these BART annotations and LDGNT etc...