Apart from a relatively small number of i- and u-stems with
a divergent paradigm (genitive sg. in -jés/-jós, -wés/-wós),
some of which may be old n-stems (e.g. the u-stem neuters
*doru(r), *g^onu(r), *pok^u(r) ~ *pek^u(r), *medhu(r) ,
etc., *potis if from *pótyo:n, *poti(n)m), the vast majority
of i- and u-stems (the "standard" type), are characterized
by a gen.sg. in *-eis / *-eus (*-ous?).
In the past I have tried to explain the different types by
an appeal to proterodynamic vs. histerodynamic accent types.
For instance me in 2002:
>Firstly, we turn to the i- and
>u-stems, which I reconstruct as stems ending in Pre-PIE **i and **u.
>I will distinguish between initial-stressed (proterodynamic) words
>with light root syllable (PD-L), proterodynamic words with heavy root
>syllable (PD-H)
[I would just call that amphidynamic now.]
>and end-stressed (hysterodynamic) words (HD). At the
>first stage, we have:
>
> PD-L PD-H HD
>N '-î-z '-i-z -í-z
>V '-îy '-iy -íy
>G -íyâs -íyas -iyás
>
>The circumflexed vowels have been lengthened by the svarita-rule ("in
>final syllables, the vowel is lengthened if the preceding syllable is
>stressed and short"). In the vocative and the oblique cases (but not
>in the nominative and accusative), the *i is followed by a glide
>(linking the *i with the next vowel, which in the vocative had already
>been lost). Notice that at the stage of svarita-lengthening, the
>glide did not count for purposes of syllabification (otherwise we
>would have had *-í-yâs in the PD-H form, as in the PD-L declension).
Now this has never convinced me.
We would expect:
PD AD HD
N '-î-z '-i-z -í-z
V '-îj '-ij -íj
G -íjâs -íjâs -ijás
which should have given:
PD AD HD
N '-jes/'-is '-is -jés/ís
V '-ei '-i -éi
G -éjos -jós -jés
Instead of attested:
-is -ís
-ei
-eis -éis (also: -jés, -jós)
So, the standard i-stem genitive in *-eis does not follow
from my 2002 analysis (not without special pleading).
Another problem is that the standard i- and u-stems (with G.
in -eis, -(e/o)us) do not show any kind of Ablaut in the
root. We would expect *CéRC-is, CR.C-éis if the origin was
really a C-stem PD/AD paradigm.
A different solution. What if the "standard" i- and u-stems
were really thematic stems, with theme vowel *i and *u
instead of *a (o-stems)?
As in the o-stems, we would have an originally stressed
thematic vowel, quite possibly preceded by a lengthened
grade root. A long pretonic *a: would subsequently attract
the accent and cause the paradigm to become barytone. Where
the stress was not retracted the paradigm remained oxytone.
N. -í-z(V)
A -í-m(V)
V. -í-(V) => -íj
n. -í
G. -í-as(V) => -íjas
D -í-a(V) => -íja
Now with lengthening of the thematic vowel before a voiced
segment (as in the o-stems!):
N. -í:-z
A -í:-m
V. -í:-j
n. -í
G. -í:-jas
D -í:-ja
After zero-grade and optional retraction of the accent, we
would have:
barytone
N. *-je(:)s => *-is
A *-je(:)m => *-im
V. *-e(:)i => *-ei
n. *-i
G. *-e(:)is => *-eis
L. *-e(:)i-i => *-e:i
Which is exactly the attested pattern.
For the oxytone forms, we would expect:
N. *-jé:s
A *-jé:m
G. *-é:is
L. *-é:i
Tempting as it is to connect the oxytone NAsg. with the
Latin e:-stem declension (-ie:s, -ie:m), I believe a much
more hopeful connection can be set up with the Greek
i-stems, which Sihler (p. 313) gives as Proto-Greek:
N. -is
A. -in
V. -i
G. -e:yos
D. -e:yi
pl.
N. -e:yes
A. -ins, -e:yas, -iyas
G. -iyo:n
D. -isi
The -e:y- is usually traced to the Lsg. *-e:i (< *-ei-i),
which, as Sihler notes, is not entirely satisfactory. If
there existed an oxytone i-stem paradigm with */é:i/
throughout the oblique, that would offer a more convincing
explanation for the Greek i-stems.
Gothic G.sg -ais, which forces Sihler to reconstruct an
incongruent *-oys for the i-stem genitive, can, if I'm not
mistaken, also be explained as coming from the oxytone
variant *-é:is.
Baltic and Slavic can show no distincton between *ei and
*e:i in final syllables. Sanskrit can, but offers no support
for the long vowel.
If we submit the u-stems to the same treatment as the
i-stems, we run into some difficulties. Lengthening of **u
would produce *wo (*o before u/w), parallel to **i: > *je:
(*e: before i/j). This nicely explains the apparent cases
of *-ou- in the oblique (again: Gothic -au, -aus), but *-wo-
when unstressed in the barytone forms (let alone when
stressed in the oxytone forms) would not have been reduced
regularly to *-u-. There are two possible solutions: (1)
*-us/*-um were restored analogically after *-is/*-im, or (2)
the lengthening of the thematic vowel took place at a later
stage than other vowel lengthenings (e.g. those in root
nouns [**pad- > **pa:d- > *pod-] or those caused by the
svarita-rule [**h2ák^-man > *h2ák^-ma:n- > *h2ák^mon-]). The
result of lengthened thematic *u (> *[w]e?) may then have
been *we: rather than *wo, and a reduction of unstressed
*we: to *we to *u is [more or less] regular (as in the
i-stems). However, thematic **a: _does_ give *o when
lengthened --perforce at the same time-- before a voiced
segment. Perhaps the two solutions are best combined:
thematic **u gave *(w)e: instead of *(w)o when lengthened
under the influence of i-stem *(y)e:.
=======================
Miguel Carrasquer Vidal
miguelc@...