Attraction of Relative Pronoun

>I would like to know what evidence there is of a caseless relative marker in
>ancient Greek.  There are all kinds of (pardon the hyperbole) uninflected
>particles in ancient Greek.  Could any of these have been derived from a
>caseless relative marker?  Are there any other syntactic explanations for
>the so-called case attraction?  I am aware of the exegetical/hermeneutical
>(basically, semantic) explanations, but what about syntactic perspectives
>which allow for the application of GB Case Theory?

     From the standpoint of New Testament Greek, which is the area of Greek
studies I am most familiar with, the semantic explanation of this phenomenon
seems sufficient.  A. T. Robertson, for instance, in _A Grammar of the Greek
New Testament_, says "The agreement of the relative with antecedent in
person, number, gender, and sometives case, is just the natural effort to
relate more exactly the two clauses with each other" (p. 711).  This
explanation would seem to be borne out by the also-often-occuring inverse
attraction of the antecedent to the relative (J. H. Moulton,._Grammar of New
Testament Greek_, Vol. III, _Syntax_, by Nigel Turner, p.324), and by certain
exceptions to the rule of attraction according to semantic considerations
(Blass, DeBrunner, _A Greek Grammar of the New Testament_, #294, 1).

     If, however, I were looking for a possible origin for the attraction of
the relative in Ur-Greek syntax and the possibility of a "caseless relative
marker in ancient Greek," I would probably take up the clue of the original
identity of the relative and demonstrative.  Sanscrit might give some further
clues, if you are on the right track.  Dana and Mantey, _A Manual Greek
Grammar of the New Testament_, have some comments along this line  and give
some bibliographic references (probably somewhat dated by now) in their
treatment of the relative pronoun.  My copy of their work is in Spanish, so I
can't give you a page number for the English edition, but the section number
is 135.

     I hope this may be helpful.

David L. Moore