Piotr:
> Right, but still *-i seems to have been some kind of optional enclitic
> rather than an obligatory case ending.
Perhaps. One idea that I had, influenced a little by a reinterpretation
of Etruscan grammar, was that there is not only an animate accusative
and inanimate accusative but also underlyingly a contrast of animacy in
the genitives and locatives. I'm speaking of course of an IndoTyrrhenian
stage that would have been something like this:
ANIM INANIM
NOM *-sa --
ACC *-m unmarked
LOC *-i unmarked / *-dHi
GEN *-ase *-ala
> I only wondered whether the *-oi- part of *-oisu should be analysed in
> the same way.
Ah. Interestingly, Etruscan has a STEM-PLURAL-CASE order like you
suspect for pre-IE. However, considering that *-es is in my mind
from earlier *-it (with softening of final *-t), I can't accept this
and am unconvinced by the small evidence for *s/*t alternations based
solely on *so & *to-, these roots being seperate in origin.
If STEM-PLURAL-CASE was the original state, then I'd expect to see
alternations of plurals in nominative *s and oblique *t. Of course
I've suggested earlier such an alternation once in *-kmt- (which is
only found in medial position in IE itself before more number
markers, *-ih and *-x).
> THEMATIC DUAL (ANIMATE)
>
> nom. *-o-G-z > *-o:(h) > *-o:
> acc. *-o-G-m > *-oGw > *-o:(w) (falling together with the nom.)
> gen. *-o-G-s > ... > *-o:s
If I accept the dual marker in pre-IE, it would have to be because
there was a marker *-ak in Proto-Steppe which becomes *-a? in
early IndoTyrrhenian because of automatic uvularization of *[k]
to [q] when neighbouring a [+low] vowel, *a. This would yield *-h
after Syncope in early Late IE, and if vocalized, *-e. Since we
know that *?/*h has no colouring, this would appear an optimal solution
over and above deriving it from *hW or *G.
We'd have thematic stems behaving like this in my theory:
MIE nom *-a-h (No nom. *-sa in non-singular stems)
=> eLIE *-ah (Phonotactic Resistance of Syncope)
=> lLIE *-oh (Vowel Shift)
MIE acc *-a-h-am
=> eLIE *-ahm (Syncope)
=> mLIE *-ahw (devoicing assimilation of *hm > *w)
=> lLIE *-oh(w) (Vowel Shift)
MIE *-a-h-sa
=> eLIE *-ahs (Syncope)
=> lLIE *-ohs (Vowel Shift)
However, why would a dual stem in *-h have an athematic genitive
*-sa in MIE when this is reserved for semivocalic stems? We would
have to say then that stems in *-h also received *-sa (> *-s)
instead of accent-stealing *-ása (> *-os) which was found elsewhere.
> ATHEMATIC DUAL (ANIMATE)
>
> nom. *-C-G-z > *-C-e-(h) > *-Ce, with a prop vowel
Quite simple. eLIE *-h vocalizes to *-e after consonant right after
Syncope. This of *-e here as nothing more than *-h with a pretty
circle under it as we see for *m, *n, *l and *r. Similarly, we
expect *x to be sometimes vocalized to *a or *hW to *o at that time.
Its merely the voicing of *h, which sounds awefully like a vowel,
doesn't it.
> Inanimate duals are similar but with *-i- added before the duality
> marker:
>
> athem. nom./acc. *-C-i-G > *-Cih > *-Ci:
> thematic nom./acc. *-C-o-i-G > *-Coi(h1) > *-Coi
I don't think this is related. I see *-ih being the synthesis of
an *i-stem, often inanimate in itself, and dual *-h.
> THEMATIC PLURAL
>
> nom. *-o-D-z > *-o:D > -o:s (beside *-o:s-es and *-oi)
> acc. *-o-D-m > *-o:m > -o:n-s (with added *-s)
> voc. *-o-D > *-oi
The nominative is *-o- + *-es. The end. The accusative follows suit
and adopts a long vowel too. So *-o:ns < *-ons < eLIE *-am-s < MIE
*-am-es.
> ATHEMATIC PLURAL
>
> nom. *-C-D-z > *-C-&-D > -es (with a prop vowel)
> acc. *-C-D-m > *-C-m > -Cn.-s (with added *-s)
The addition of *-es, or irregularly syncoped *-s in the oblique.
> Non-sg. locatives show a final *-u, so perhaps
>
> loc.du. *-o-G-u > *-ou
This is by analogy with *dwo:u and *okto:u, themselves being
given long ago plural endings related to *-x which eroded by
automatic uvularization.
> loc.pl. (thematic) *-o-D ~ *-o-D-u > *-oi ~ *-osu > *-oi-s-u
> (athematic) *-C-D-u > *-C-s-u
I'm now thinking *-i + *-su 'among' without further splicing.
= gLeN