I believe we've been through this once before not terribly long ago, but before you joined the list. You might find it convenient to consult the archives, of which there are two--the first might be easier for finding the thread as listings there are searchable by thread, date, and author of the message:
B-Greek Archive http://www.entmp.org/cgi-bin/lwgate/B-GREEK/archives/
B-Greek digest http://www.gramcord.org/bgreek/index.htm
Briefly (and I'm sure I'll be corrected if and when found wrong!), Paul cited the LXX text which, had it followed "standard" practice to mean what Paul read it to mean, would have had EK PISTEWS sandwiched between hO and DIKAIOS or else the article should have been repeated before EK PISTEWS. Quite simply put, you're right about what the citation OUGHT to mean in terms of its word-order; the translations offered are in terms of HOW PAUL READ AND UNDERSTOOD THE TEXT in order to use it as a proof text for a proposition which (I would guess) he did NOT derive from that text but rather, independently, perhaps, as in Galatians, where he works it out ealier, from Genesis 15:6).
I think this must be a jarring discovery for everyone who has carefully learned the distinction between attributive and predicate positioning of the article and then comes to Paul's use of this LXX citation. I can well remember being thrown for a loop when I first confronted it in reading Romans 1: "How can he DO that? That text CAN'T mean what he takes it to mean!" I do think Paul's reasoning from OT texts--with the masterful exception of Romans 4 on Abraham--calls repeatedly for defenses in terms of, "But that sort of argumentation was standard practice then." I think if it were offered today it would be called eisegesis.
Carl W. Conrad
Department of Classics, Washington University
One Brookings Drive, St. Louis, MO, USA 63130
(314) 935-4018
cwconrad@artsci.wustl.edu OR cwc@oui.com
WWW: http://www.artsci.wustl.edu/~cwconrad/