Just to clear up one little detail:
We had seen that the acc.pl. of (j)a:- and (j)o-stems could
be explained as follows:
n-raising h-raising shorten. length. Umlaut =
*-a:Nh > *-o:Nh > *-u:Nh > *-uNh > *-u:N -- -y
*-o:Nh > *-u:Nh > *-u:Nh > *-uNh > *-u:N -- -y
*-ja:Nh > *-jo:Nh > *-ju:Nh > *-juNh > *-ju:N > *-ji:N -jeN
*-jo:Nh > *-ju:Nh > *-ju:Nh > *-juNh > *-ju:N > *-ji:N -jeN
That leaves only one paradox, the i-stem acc.pl:
*-iNh > *-iNh > *-iNh > *-iNh > *-i:N -- -i
Umlauted *-ju:N > *-ji:N gives -jeN/-jê, but original *-i:N
gives -i: > -i.
How to explain this?
After all the raisings, shortenings and lengthenings, we had
the following nasal combinations in the Auslaut:
-iN (> -I, e.g. acc.sg. i-stems)
-eN (> -eN, e.g. NA.sg.n. n-stems)
-oN/-joN (> -oN/-joN, e.g. Asg. a:-stems)
-uN/-juN (> -U/-jI, e.g. acc.sg. o-stems)
-i:N (> -i, e.g. acc.pl. i-stems)
-e:N (> -eN, e.g. nom.sg. nt-stems)
-u:N/-ju:N (> -y/-jeN ~ -jê, e.g. acc.pl. o-stems)
We can classify these as follows:
1) nasalization not lost
*-eN, *-(j)oN, *-e:N
2) nasalization (partially) lost _after_ J-umlaut:
*-u:N/*-ju:N
3) nasalization lost _before_ J-umlaut:
*-i:N
4) nasalization lost before or after J-umlaut:
*-iN, *-uN/*-juN
For symmetry, we can classify *-iN with *-i:N, and *-(j)uN
with *-(j)u:N.
So the chronology was:
*-i(:)N > *-i(:)
*-ju:N > *-ji:N
*-u(:)N > *-u(:)
*-jiN > *-ji
*-ji:N > *-jêN
*-jêN > *-jê / *-jeN
The changes are for the most part quite natural. Nasality
is lost earlier by high vowels (i/u) than low (e/o/a), and
earlier by front vowels (i) than back (u). Lowering of high
nasal vowels (iN > eN) is also a natural development.
The only slightly puzzling question remains why back -u(:)N
lost its nasalization, while front -ji:N didn't
(everywhere). This can of course be readily explained by
the lowering of *-ji:N, but wy didn't *-u:N lower, or for
that matter, *-jiN?
=======================
Miguel Carrasquer Vidal
mcv@...