On Wed, 29 Jan 2003 12:15:14 +0100, Miguel Carrasquer <
mcv@...>
wrote:
>On Wed, 29 Jan 2003 09:04:24 +0000, "Glen Gordon"
><glengordon01@...> wrote:
>
>>> >However, that being so, there were instances where the loss of
>>> >unstressed vowels was resisted
>>>
>>>So what are the rules?
>>
>>Therefore, I'm correct that you don't understand what I said.
>
>"there were instances where the loss of unstressed vowels was
>resisted": which instances? What are the rules?
>
>>>There is no reason in your theory why this should not have given
>>>*udnos. Cf. the collective root *udó:r, *udéns.
>>
>>The reason is painfully simple to any sensible IEist:
>>*u is the normal zero-grade of *eu/*au/*ou, not of *we/*wa/*wo.
>
>Samprasa:ran.a is the term used by the old Indic grammarians.
I now realize what the source of the misundertanding was. Glen stated
that zero grade was blocked if an asyllabic stem in the paradigm would
have resulted, using *ped- and *wed- as examples of blocking, *kun- as
an example of normal zero-grade. My insistence on knowing the exact
_rules_ was based on my understanding that the zero grade of *wed- is
*ud-, which is syllabic, while Glen thought he had given all the rules
there were to give ("no asyllabic stems"), based on his
(mis)understanding that *wed- cannot have *ud- as a zero grade.
I compounded the misundertanding further by bringing up the oblique
stem *pk^w-, which is in fact irrelevant to the whole question, given
that *pek^u- is not a root noun but a compound (*pek^- + *-u (<
**-un)), and we all know that the _root_ of compound nouns can easily
be asyllabic (*dhg^h-o:m, *k^w-o:n (*pk^-wo:n), *ph2-wó:r (*ph2w-ó:r),
*dr-ú-h2, *h1d-ónt, etc.) even if the _stem_ is not.
The only relevant word mentioned so far is *sem-, obl. *sm- which
happens to be one of the few root nouns I'm aware of with e-vocalism,
and certainly the only one with only two consonants (we have a handful
more with three consonants, such as *krep-s, *k.rp-és "body", and
perhaps *sh2a:l(-s), *s&2l-és "salt", (**mwés-s >) *mu:s, *musés
"mouse").
The question is: why do so many root nouns have o- or sometimes
e:-vocalism (*k^é:rd, *mé:ms) and so few normal e-grade? In the
context of my theory (where *o < **a:/**u:, and *e: < *i:) that
translates to: under what circumstances was the vowel of
[monosyllabic] root nouns *not* lengthened?
A few words about Samprasa:ran.a.
Pa:n.ini and the other Sanskrit grammarians worked in terms of a root,
which could be strengthened to gun.a-grade, and further strengthened
to vr.ddhi grade. E.g. a root /cit/ has gun.a grade /cet/ [*cait] and
vr.ddhi grade /cait/ [*ca:it]. Modern IEists use a different
paradigm: the fundamental form of the root is given in "gun.a" grade
(*kWeit or o-grade *kWoit), which can be weakened to zero grade
(*kWit) and strengthened to lengthened grade (*kWe:it/*kWo:it)
[o-grade culd not have been foreseen by Pa:n.ini, and due to
Brugmann's Law many Pa:ninian vr.ddhi's are in fact "gun.a" o-grades].
The difference between the two viewpoints becomes clear when we take a
root such as *swep- "sleep". From the modern point of view, there is
no problem: the zero grade is *sup-, the lengthened grade *swe:p- or
*swo:p-. For Pa:nin.i, these roots were irregular, because the gun.a
of /u/ was supposed to be */au/ (=/o/), not /va/. The solution was to
give the root in its gun.a form (svap), and assign the alternation
/svap/ ~ /sup/ (and similar roots with /ya/ ~ /i/, /va/ ~ /u/, /ra/ ~
/r./) to a special "samprasa:ran.a" gradation class.
=======================
Miguel Carrasquer Vidal
mcv@...