On Wed, 01 Oct 2003 03:18:34 -0700 (PDT), "Alwin K." <
alwinmail@...>
wrote:
>Beekes in his Introduction (p 182) seems to
>reconstruct the words as a *-(e)h2- stem, and it looks
>like he explains the absence of Brugmann by assuming
>that originally *-(e)h2-stems have had zero-grade
>suffixes as well (e.g. gen.sg. *Hros-h2-os). Perhaps
>absence of Brugmann was generalized because of these
>cases?
There is no evidence for Beekes' *-h2os. The ending was generally [but see
below] *-ah2os (giving post-laryngeal *-a:s).
Beekes says:
"The genitive, dative and locative of Sanskrit must have borrowed their -y-
from the ih2-stems. The reason for this was that in -h2-os etc. the
laryngeal was lost, so that the complete suffix disappeared. Old Irish
also has -e < -ih2-os. Lat. -ae has taken over the -i: of the o-stems.
Another ending has been preserved in expressions such as <pater familia:s>.
OCS -y is perhaps analogical after the nom.pl. Elsewhere -h2-os was
replaced by -a:s. Dat. -h2ei > -ai was replaced by -a:i. [...] Ins. Skt.
-aya: and OCS -ojoN have been taken over from the pronouns."
The Latin and OCS a:-stem gen.sg. play no significant part in the argument.
The older Latin ending was -a:s, while the OCS ending can only be
analogical after the nom.pl. if it was also -a:s (and then both the gen.sg.
and nom.pl. -a:s -> -a:ns > -y are analogical after the acc.pl. -a:ns).
Old Irish indeed shows paradigmatic influence from the ih2- (devi: and
vrki:s types) and/or yah2-stems (also in the acc.sg., for instance), but
this can hardly be used as an argument against *-a:s (*-ah2os) as the
original gen.sg. ending. Old Irish mná (*gWná:s), gen. of ben "woman"
shows that *-a:s was also present in Irish. Confusion between paradigms is
not uncommon in Old Irish, and understandable given the phonetic wear of
final syllables in that language.
That leaves only the peculiar Sanskrit (Indo-Iranian, actually) forms to
somehow support Beekes' argument. The traditional explanation for Skt.
-a:ya:s etc. is that they consist of the suffix *-ah2 + the endings of the
devi:-type *ih2-stems (*-ya:s). Beekes seems to accept that, and explains
the substitution (Sanskrit has suffered none of the phonetic wear that Old
Irish has) by his assumption of an ending *-h2os > *-os, which by being too
undercharacterized as an a:-stem, yielded to the new construction
*-a:-ya:s.
My own opinion is very different. In the first place, I think the a:-stems
are thematic, i.e. the endings contain the thematic vowel (plus something
else). That excludes the possiblity of and ending *-h2os, with loss of the
thematic vowel: the main characteristic of the thematic vowel is precisely
that it is not affected by the zero-grade rule.
The feminine suffix following the thematic vowel is *ih2, as it appears in
the athematic ih2-stems of the de:vi: and vrki:s types. To explain the
nominative sg. *-ah2 out of *-a-ih2 (thematic vowel + *-ih2), we require a
soundlaw that deletes *i after the thematic vowel under certain
circumstances (specifically, when the thematic vowel was stressed, i.e.
*-á-ih2 > *-á-h2). The standard paradigm, with columnar accent, then
becomes:
nom. -á-(i)h2 > -ah2 > -a:
acc. -á-(i)h2-m > -ah2m > -a:m
voc. -á-(i)h2 > -a(h2) > -a
gen. -á-(i)h2-a:s > -ah2os > -a:s
loc. -á-(i)h2-ai > -ah2i > -a:i
abl. -á-(i)h2-a:t > -ah2ot --
dat. -ah2ái > -a:i
ins. -ah2áh1 > -a:
nom. -á-(i)h2-asW > -a:s
acc. -á-(i)h2-m-sW > -a:ns
gen. -á-(i)h2-a(:)m > -a:m
dat. -á-(i)h2-bhi-a:sW > -a:bhios
loc. -á-(i)h2-sW-i > -a:su
ins. -á-(i)h2-bhi-sW > -a:bhis
Besides this, there was also a paradigm with strong and weak forms, i.e.
with mobile accent (it has only left traces in the singular):
nom. -á-(i)h2 > -ah2 > -a:
acc. -á-(i)h2-m > -ah2m > -a:m
voc. '-a-(i)h2 > -oi(h2) > -oi
gen. -a-íh2-a:s > -oyéh2os > -oya:s
loc. -a-íh2-ai > -oyéh2i > -oya:i
abl. -a-íh2-a:t > -oyéh2ot --
dat. -a-ih2-ái > -oyh2ái > -oyyái
ins. -a-ih2-át > -oyh2áh1 > -oyyá:
This corresponds almost exactly with the Sanskrit paradigm:
N -a:
A -a:m
V -e:
G -a:ya:s
L -a:ya:m (*-oyah2-m)
D -a:ya:i (< loc.)
I -aya:
The instrumental *-oyh2áh1 (with additional *-m, as in the Skt. loc.) also
explains OCS -ojoN, as well as Armenian -oj^ (G/D/L in a few archaic nouns
such as knoj^ (kin "woman") or mioj^ (mi < *smih2 "one"), and in the
Armenian (i)ja:-stems, which I have reconstructed as follows:
N *-íya: > *-íya > -í
A *-íya:m > *-íya > -í
G *-iyóya:s > *-iwóya > -wóy
DL *-iyóya:y > *-iwóya(i) > -wóy
I1 *-iyóyya:(m) > *-iwój^a > -wój^
I2 *-iyá:-bhi > *-iwáBi > -eáw
The paradigm of the PIE ih2-stems of the de:vi:-type is of course exactly
that of the "a:-stems with mobile accent", but without the thematic vowel:
nom. '-ih2 > -ih2 > -i:
acc. '-ih2-m > -ih2m > -i:m
voc. '-ih2 > -i(h2) > -i
gen. -íh2-a:s > -yéh2os > -ya:s
loc. -íh2-ai > -yéh2i > -ya:i
abl. -íh2-a:t > -yéh2ot --
dat. -ih2-ái > -yh2ái --
ins. -ih2-át > -yh2áh1 > -yá:
=======================
Miguel Carrasquer Vidal
mcv@...