FYI (pre-IE reconstructions)

[Miguel Carrasquer Vidal]

On Sun, 27 May 2001 17:29:43 +0200, "Mate Kapovif"
<jozo.kapovic at zg.tel.hr> wrote:

>Just a brief note. In the MCV's text on pre-Indo-European there is a claim
>that Hittite ablative -az comes from PIE *-od-s. Considering the Luwian form
>of ablative -ati it can safely be assumed that the protoform of these two is
>*-oti.

>*o > a in both lgs, *t is palatalized to z in Hittite in front of *-i which
>is later omitted.

This interpretation is of course the generally accepted one, but I
have a few problems with it.  Luwian does not distinguish Abl. and
Ins., and the ending -ati (Lydian -(e)di) serves both functions.  It
should be compared to both Hitt. Abl. -az and Ins. -it.  While it is
true that Hitt. palatalizes *ti to (z)zi (e.g. the 3sg. vb. ending
-(z)zi), there is no rule that Hittite drops final -i (e.g. the 3sg.
vb. ending -(z)zi).  Within Hittite, it is preferrable to derive -az
regularly from *-od-s or *-ot-s, rather than irregularly from *-ot-i.
The comparison with Luwian strengthens the case for *-oti, but there
are alternatively explanations which are equally likely.  The evidence
from non-Anatolian Indo-European points to an Abl. *-od ~ *-ed (or:
*-ot ~ *-et, one cannot tell) and an Ins. *-eh1, with plural forms
*-[bhi-]os / **-[oi-]os and *-[bhi-](h1)s / *-[oi-](h1)s.  This
suggests to me an original distinction between Abl. **-od and Ins.
**-et, which was maintained in Hittite (albeit with Abl. -az instead
of expected -at, possibly from the plural form **-ods, which gives
*-os elsewhere [merging with Dat.pl. *-os]) and in "non-Anatolian"
(albeit with a proposed regular development *-t > *-h1 in the Ins.).
In the Luwian group, Abl. and Ins. added what is probably the same *-i
as we find in the Dat. (*-o-i) and Loc. (*-0-i), giving **-od-i and
*-et-i, which then merged (Luw. -ati, Lyd. -edi).  Within this
analysis (where the Abl. must have a different consonant from the
Ins.), deriving Hitt. -az < **-odi becomes doubly irregular: not only
is the loss of -i inexplicable, but *-di should palatalize to **-si,
not *-z.

=======================
Miguel Carrasquer Vidal
mcv at wxs.nl