On p. 82 and further, there's a discussion of the Ablaut of
the PIE root aorist. Jasanoff claims that the indicative (as
well as the 2pl. imper.) was characterized by e-grade
everywhere, except in the 3pl.. "This pattern is normal in
Indo-Iranian (cf. Ved. 1pl. (á)karma, (á)ganma, (á)da:ma,
etc., 2 pl. (á)karta, (á)ganta, (á)da:ta, but 3pl. (á)kran,
(á)gman. (á)duh.) [...]".
The description in Macdonell is slightly different: the root
aorist has strong forms in the ind. sg., and weak forms
elsewhere (same as the present, or the imperfect), _except_
that roots ending in a "vowel" (-y, -w, -r/-l, -H) [and this
of course also applies to roots ending in -n, -m and -C]
maintain the strong forms, except in the 3pl.
This would mean that roots ending in -RC should show weak
forms in the 1/2pl. and 1/2/3 du. root aorist, and indeed
some examples can be given (|dr.s'|- > ádr.s'ma besides
ádars'ma, |muc| -> ámuktam, but |chid| -> chedma).
A not unimportant factor is obviously the syllabic shape of
the endings: the 1/2pl. and du. endings are of the shape
-CV.. (-ma, -ta, -va, -tam, -ta:m), but the 3pl. is -VC r ta
least *-VC (-an/-ur < *-r.s or *-é:rs).
Even so, that does not fully explain why the imperfect has
weak 1/2 plural and 1/2/3 dual forms (except in the case of
-C(C) verbs), while the root aorist has so many full grade
forms there.
Jasanoff is rather vague in general about the connection
between Ablaut, stress, vowel length and syllable structure.
As I basically accept (with some modifications) Rasmussen's
"Morphophonemik" explanation, if I were to accept e-grade in
the root-aorist 1/2 pl., that would inevitably have
consequences for the original structure of the forms. If we
take the verb *dheh1- as an example, I would have expected
the root aorist (similarly to the imperfect) to have
developed as follows:
*dhéh1-m > *dhé:m > ádha:m
*dhéh1-s > *dhé:s > ádha:s
*dhéh1-t > *dhé:t > ádha:t
*dh(e)h1-mé > *dh&mé > *ádhima (actually: ádha:ma)
*dh(e)h1-té > *dh&té > *ádhita (actually: ádha:ta)
*dh(e)h1-érs > *dhé:r > ádhur,
with {analogical|syllabic structure} explanation for 1/2pl.
On the other hand, if ádha:ma and ádha:ta are original, then
there is no other solution than to either reconstruct a long
vowel in the root:
*dha:h1-má > *dhéh1-me > ádha:ma
*dha:h1-tá > *dhéh1-té > ádha:ta
*dha:h1-án > *dhéh1r(s) > ádhur,
which begs the question of why we seemingly have no long root
vowel in the singular, or to conclude that the _endings_ of
the 1/2pl. (perhaps also the dual, if that's not analogical)
were originally _asyllabic_:
*dhéh1-m > *dhé:m(e) > ádha:ma
*dhéh1-t > *dhé:t(e) > ádha:ta.
But this would divorce the aorist endings from the imperfect
endings.
For the h2e-conjugation aorists, Jasanoff claims o/e-Ablaut
(I've read that part only superficially for now, but that's
my impression), so that would take care of lengthening (i.e.
o-grade) in the singular for those verbs (e-grade in the
plural spreading analogically to the mi-conjugation aorists
[but why not the the 3pl.?]).
The other possibility, asyllabic endings, can also work.
There are good reasons to think that the endings of the
1/2pl. perfect/hi-conjugation (and *not* that of the 3pl.)
originally *were* asyllabic: reduction of the original
endings in the absolute Auslaut is the only way I can explain
the discrepancy between perfect *-me (or *-meH?), *-te (or
*-He?), and the historically related middle endings *-modhH-
(or *-mwo?) and *-dhwo-[*]. So if part of what we know as
root aorists were originally conjugated (and in Anatolian
still were) with "h2e"-endings (e.g. 1pl. *dhéh1-mH, 2pl.
*dhéh1-dH), that would explain the aberrant Ablaut of the
1/2pl. in the root aorist (having spread analogically to the
mi-conjugation root aorists, but not the mi-conjugation
imperfect).
[*] All the endings of the middle/perfect (stative) complex
show signs of having stood in the Auslaut originally.
Starting from a Proto-Nostratic prototype:
*-(a)-k(V) *-at-m(V)-k(u)
*-t(a)k(V) *-at-t(V)-k(u)
*-(a) *-an,
we get:
*-k *-tmkW
*-tk *-ttkW
*-0 *-án,
giving:
*-h2 *-(d)mGW (?)
*-th2 *-dGW (?)
*-0 *-ér
The middle endings diversified at this point:
1. *-h2-a-r(i), *-h2-a(i), *-h2-a-dh(i)
2. *-th2-a-m, *-th2-a(i), *-th2-a-dh(i)
3. *-o-m, *-o-r(i), *-o(i), *-o-dh(i)
*-t-o-m, *-t-o-r(i), *-t-o(i), *-t-o-dh(i)
1. ?
2. *-dhw-o-m, *-dhw-o(i), *-dhw-o-dh(i)
3. *-ro-m, *-ro-n (!), *-ro(i), *-ro-dh(i)
*-nt-o-m, *-nt-o-r(i), *-nt-o(i), *-nt-o-dh(i)
[
for 1pl., either:
1. *-mw-o-r(i), *-mw-o(i), *-mw-o-dh(i)
And *-mwodhi taken as the basis for:
*-mwodhar(i), *-mwodho(i)
or directly (metathesis *-dmGw > *-m(w)dG ?):
1. *-mwodhh2-r(i), *-mwodhh2-a(i)
]
In the perfect/hi-conjugation, 1/2pl. were further reduced in
the Auslaut:
*-m(H)
*-t(H) ~ *-H
Then, -e was added (except to the 3pl.), giving the attested
endings:
*-h2-e
*-th2-e
*-e
*-m-é (?)
*-t-é (?)
*-é:r, *-r.s
The exact Ablaut depends on the presence and accentuation of
the reduplication. So, for the "classical perfect", we have:
*(mi-)má:n-h2 > *(me)món-h2e
*(mi-)má:n-th2 > *(me)món-th2e
*(mi-)má:n- > *(me)món-e
*mí:-man-mH > *mé:-m(e)n-m(H)e > *me-mn.-mé
*mí:-man-dH > *mé:-m(e)n-t(H)e > *me-mn.-té
*mí:-man-an > *mé:-m(e)n-ers > *mé:-mn-r.s / *me-mn-é:r
=======================
Miguel Carrasquer Vidal
mcv@...