Gweinic Gender Marking

I have a problem with Lahabic, in that I don't know whether there are two or
three genders. On the assumption that there are three, I shall call them
unmarked, animate, and inanimate.

A verbal root or particle such as hiyik- 'go' or toog 'after' is unmarked.
hiyikra- 'the one who goes' (hiyik- + animate present active suffix -r(a))
is animate, while hiyika- 'the act of going' (hiyik- + inanimate present
active suffix -a) is inanimate. In a word trasparently composed of a verbal
root plus suffix, such as hiyiken- 'road' (hiyik- +-en, animate instrumental
suffix), the animacy is clear. The geographical suffixes, such as -auan,
'large straight (usually horizontal) thing', cause unmarked roots to become
animate. Thus one derives medweelauan- 'hall' from medweel- 'live'. The
diminutives and augmentatives are unmarked for animacy. Thus one gets
telemniya- 'sapling' and telemniya- 'small support beam' from telembra- and
telema- respectively. These potential homnyms are differentiated by the
addition of the present active suffixes to produce telemniyar- and
telemniyan-. In the case of a semantically transparent word such as
hiyikenniya- 'path', however, this is not necessary because the animate
instrumental suffix -en- provides the inanimacy. In words such as
bhweteusin- 'publican', animate, from bhwetes- 'drink', unmarked, + -win,
animate instrumental suffix, it becomes less clear. bhweteusin- spawns the
words bhweteusinouon- 'pub', which becomes inanimate, and bhweteusinniya-
'barmaid (not necessarily female in Dhabra culture)', which remains animate.
From bhweteusinouon- comes bhweteusinouondra- 'pub-crawler', which is once
again animate. In principle any part of speech (including seel 'and', which
yields seerra- 'one who adds, toady, yes-man') can undergo this process,
which could result in deep burial of the 'animating' suffix. I am not
certain whether the unmarked gender counts as a gender in the technical
sense (since it includes so much, but is quickly buried)  and if so what to
call it. Do any natlangs have this process?

Also, the change from Lahabic to Labic, since it involves contraction, would
obscure this mechnism. Would it then become a system of apparently arbitrary
gender?

On a different front, the stress system of Lahabic almost always results in
the following pattern for singular, dual, and plural nominative.
Animate
      ghau-'id-ra-khe, ghau-'id-ra-ghe, 'ghau-id-ra
Inanimate
     ,ghau-id-'a-lei, ,ghau-id-'a-see, ,ghau-id-'a-see
The problem arises from the contraction of Lahbic into Labic forms. Now the
animate singular and dual would probably just refose to contract. The other
forms could just become ghaud-, but I have reasons for wishing to avoid that
solution with certain roots. Instead, ghau.id.a.see [g<h>au.id.aze:] could
become ['g<h>aujd.a.ze:]/ Yours truly, however, cannot pronounce this
triphthong, so perhaps [g<h>aujd.a.ze:] could become [g<h>o:jd.a.ze:]. The
fact that this change would produce the series ('ghauidrakh, 'ghauidragh,
'ghoidra) and (ghoid'alee, ghoid'asee, ghoid'asee) is not a problem, since
it gives a 'modern' and archaic form for related meanings (ghauidrakh,
'guard, warden, watchman', ghoidrakh, 'soldier, border guard', although
these meaning are not yet certain).
What I would like to know is whether this process makes sense, given
its contextual limitation
Proposed Rule:
If first syllable=stressed
aw+i -> o:j
aj+u -> e:j
ew+i -> u:i/wi:
ej+u -> i:u/ju:
oj+u -> e:j
ou+i -> u:i/wi
________________________________________________________________________
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com