On Sun, 16 May 2004 13:15:09 +0000, elmeras2000
<
jer@...> wrote:
>The idea is this:
>
>The stem of the IE verb reflects an agent noun: Ved. hánmi 'I kill'
>contains the same stem as vrtra-hán- 'killer of Vrtra', so *gWhen-
>m+i is properly "a killer (am) I".
>
>With some root structures the agent noun of this short type has an
>extension in *-t-, as Ved. soma-kr-t- 'make of soma'. The 3sg form
>of the aorist ákar from *kWer-t will then have meant "a maker (was
>he)". The passive participle in *-to- must be an adjectival
>derivative from this: *kWr-t-ó- "belonging to a maker, what a maker
>has, what a maker has made, made".
>
>Other roots (or root structures?) form the agent noun as an n-stem.
>Thus Avest. spasan-, OHG spehho 'scout', OLat. as-se:do:, -
>o:nis 'assessor', OHG man-ezzo 'cannibal'. Some of these form ppp in
>*-no-, Ved. ptc. sanná-, sbst. ánna-m 'food'.
OK.
>The joint testimony of this leads to the conclusion that -t- and -(e)
>n- are not only allomorphs, but originally phonetic variants.
>Martinet toyed with a pre-nasalized [Nd] on many occasions, and it
>is hard to see where else the solution to the riddle may lie.
This is not an IE phenomenon, we find the same thing in
other Nostratic languages as well (e.g. Semitic plural
-a:n(u/i) or -a:tu/-a:ti, Kartvelian plural -n, -t-, etc.).
I already gave my solution: in the Nostratic proto-language,
final *-p, *-t, *-k/*-q gave *-m, *-n, *-N when no vowel
followed. A parallel of this same soundlaw is found
(independently, of course) in Samoyed (Selqup) and in
Eskimo-Aleut. I see no need for pre-nasalized anything.
Given an original, already Nostratic, variation *-n, *-tV (>
*-n, *-t/*-s/*-y after loss of final vowels), it's not
surprising to find different grammaticalizations of
*-t/*-tó, *-n(>*-r)/*-nó, etc., although it would be nice to
know the exact pathway that led to the establishment of
*n-variants in some roots and/or functions, *t-variants in
others.
>We
>need a special element anyway if final *-t goes to *-s in the 2sg
>marker (and the s-stems, I'd say, and of course in the few remaining
>s/t-stems).
>
>It may then also be assumed that the active nt-participle is in
>origin basically this morpheme followed by case-endings.
No, that would be *-t.
I don't know where the ptc. in *-Vnt comes from. It is
strange that its neuter NA form also ends in *-nts.
Szemerényi wants to derive it from *-mt- on very dubious
grounds, which are not worth mentioning, but *-mt(t) _would_
approach it to the middle participle in *-mtn-ó- (>
*-mh1nó-).
>I would like to add myself that I would assume that the 3pl active
>of the verb /-ent/ is then ultimately composed of this same morpheme
>followed by a plural marker. There seems to be no way of singling
>out a pluralizing "/-(e)n-/" in the pair 3sg *-t : 3pl *-ent.
It's easy: *-0-t [sg-3p] ~ *-én-t [pl-3p]. The system is
regular, provided the ending *-t was added at a later stage
than the 1/2 endings:
*-m(u) > *-m(W) > *-m
*-t(u) > *-s(W) > *-s
*-0 > *-0 *-t
*-mu-án > *-mWén > *-men/*-wen
*-tu-án > *-tWén > *-ter
*-0-án > *-en *-en-t
The ending *-s in the preterit was added even later:
*-m > *-m > *-m
*-s > *-s > *-s
*-0 > *-0 > *-s
*-m-en > *-m-en > *-men/*-wen
*-t-en > *-t-en > *-ter
*-0-en > *-er > *-ers
Note that in the 1/2pl., we also have the variants *-més (<
*-mu-átu) and *-té (< *-tu-átu), but not in the 3pl., where
*-t > *-n was always final (cf. Samoyed 1pl. -mV-t, 2pl.
-tV-t, 3pl. -to-n).
The dual forms were made with the suffix *-ik > *-iN and
*-iku, giving:
1du. *-mu-íN > *-mén/*-wén (merges with 1pl.)
*-mu-íku > *-mWéxW > *-mWéx > *-wah2/*-wh2-ás
2du. *-tu-íN > *-tén (> *-tér) (merges with 2pl.)
*-tu-íku > *-tWéxW > *-tWex > *-tah2/*th2-ás
3du. *-íN > *-én (> *-ér, *-én-t, merges with 3pl.)
(note that Greek, which has 1pl. -mén [besides -més]
therefore does not have a separate 1du. verbal ending).
=======================
Miguel Carrasquer Vidal
mcv@...