Ronald's analysis is right on target; I'd only add a couple notes on how
this construction appears to me to have developed historically.
(1) Although the classical relative pronoun hOS, hH, hO is still being used
in the NT era and is well exemplified in the GNT, there is abundant
evidence within the GNT that what previously had been a QUANTITATIVE
relative adjective, hOSOS, hOSH, hOSON ("however much/many") used wih a
correlative DEMONSTRATIVE quantitative adjective, TOSOS, TOSH, TOSON ("so
much/many") has already become virtually even if not solely the working
relative pronoun that it is in modern Greek.
(2) The older modal particle AN had been used to relativize protases of
general and future conditions calling for a subjunctive and also with
potential optatives frequently used as result clauses in future remote
conditions. Since in the subjunctive clauses its most common associated
element was EI ("if"), the combination EI + AN merged ("crasis") in a
single new form EAN. By the NT period of Koine, the EI element of EAN had
been lost more or less completely and EAN was used with subjunctives in any
of the conditional protases where earlier AN had been used.
(3) Yet another development is the structural simplification of what had
been in earlier classical Attic a more complex distinction between
temporal, relative, and generalizing conditions, all requiring an AN when
used with a subjunctive, into a common type of conditional construction
more or less identical with what earlier had been distinctly a relative
conditional construction. E.g.:
Earlier: hOS (or hOSTIS) AN TAUTA POIHSHi, ... (relative)
EI AN (or EAN) TIS (or a noun) TAUTA POIHSHI (general/future)
hOTE AN (or hOTAN) TIS (or a noun) TAUTA POIHSHi (temporal)
Later: hOS/hOSTIS AN TAUTA POIHSHi (all-purpose nominative formula)
hOSA AN POIHSHTE (subjunctive, even fut. indic.: this
becomes the all-purpose object formula.
You may still find the earlier, more sharply-distinguished formulations in
the NT and contemporary secular Greek literature, (e.g. Mt 18:19 hOU EAN
AITHSWNTAI, with EAN = older AN), but what I've listed as the "later" type
has become more or less standard.
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/