From: Patrick Ryan
Message: 42089
Date: 2005-11-13
----- Original Message -----
From: "Richard Wordingham" <richard.wordingham@...>
To: <cybalist@yahoogroups.com>
Sent: Saturday, November 12, 2005 4:52 PM
Subject: Re: [tied] IIr 2nd Palatalisation
--- In cybalist@yahoogroups.com, "Patrick Ryan" <proto-language@...>
wrote:
Patrick:
In addition, every one of the other nine Old Indian derivations of this
*kel-
have <k> (<kalká). Nine *o- or zero-grades?
Richard:
Look to Iranian for the e-grades: "npers. c^arma, kurd. c^erme `weis|' (:
schweiz. helm)"
Patrick:
But in this case, I am convinced that Pokorny's 4. *kel- really contains
two unrelated roots: *kel-, 'reddish' (parallel to 4. *ker-); and **k^el-,
'grayish' (parallel to 6. k^er-). The MPers form *c^arma, 'mold', could as
easily relate to *k^er- as to **kel- (better **k^el-).
***
Richard:
No. PIE k^ Iranian s (at least, Avestan _sat&m_, Modern Persian _sad_
'100').
Patrick:
Well, let me then suggest *c^arma from *sk^er-mo-.
*********************
Patrick:
PIE roots are *CVC. Thus the root for cRtáti is *ker-; this is
acknowledged (almost) by Pokorny when he refers to 3. *(s)ker- under the
*kert- listing. In view of an attested kRNátti under this root, I suggest
a prototype for cRtáti would probably be better reconstructed as *skRtéti.
Richard:
Except that unsoftened PIE *sk gives Sanskrit /sk/ (or /k/) e.g. the
derivatives of PIE *ska(m)bH (e.g. skabhna:ti, skabhno:ti 'support') and
softened *sk gives Sanskrit /ch/ e.g. _chyati_ 'cut' from Pokorny's *ske:i.
Patrick:
Of course, I should have written *sk^Rtéti as the source of cRtáti. As for
the difference between <c> and <ch>, I am as nearly positive as one can be
in these matters that *ske:i- should be emended to **sk(h)o:i-. Here, the
long vowel is caused by the absorption of the aspiration from
*k(h), which is always associated with 'cutting'. *k^(h), on the other
hand, is always associated with 'speed, running'. I think we see the form
without s-mobile in Pokorny's *k^o:(i)-. It looks to me as if an earlier
*k(h)o[:]y- has become *kyo- then *k^o- through metathesis while some
derivatives maintained *kyo-.
Richard :
But *sk^ seems to have had two outcomes, /ç/ and /ch/. It's also
possible that the outcome /ç/ is simply evidence that the *s was omitted.
Patrick:
I do not see an Old Indian word with <ç> for *sk^ that I can accept;
moreover, it is rather a hopeless task to try to prove it for an initial
since we have no guarantee that s-mobile was present for any specific root
beginning with *k^.
Now, we do see <c> and <ch> as reflexes of *sk^ and, I would suggest, *sk^h.
*********************
Patrick:
I know this will not be satisfactory for you but I am beginning to suspect
that cópati may be derived from a root of the form *kWéupeti or *kwéupeti.
Richard:
**kwéupeti would have yielded *kvópati. Deriving _cópati_ from *kWéupeti and
kopáyati from *koupéyeti strains credulity.
Patrick:
I have omitted ^ again. I meant *k^Wéupeti (or *k^wéupeti).
*********************
Patrick:
I think we can both agree that *kew6p- is a highly unusual root form, and
the reflexes are more than a little confusing.
Richard:
If we exclude the derivatives of **kap, don't we simply have a case of
schwebeablaut?
Patrick:
Perhaps. Schwebeablaut is not my favorite explanation for anything.
***********************
Patrick:
We do have <kópa-, which is as likely to have been an *e-grade as an
*o-grade. Would you agree?
Richard:
No. The simple thematic verb from **keup is PIE **kéupeti, which happily
yields _cópati_. In the thematic present, e-grade is much, much commoner
than o-grade. Conversely, the *moneye-ti type derivative (one of the
Rasmussen infix forms) is also a common verbal stem, which in this case
would yield *koupeyeti, whence _kopáyati_.
Patrick:
Pokorny's <kópa- is not a verbal form, is it?
Richard:
Sorry, I was talking of _kopáyati_. However, I think *koupo- is more likely
than *keupo-, but I would welcome an expert opinion on the patterns of
derived noun formation. If e-grade and o-grade are equally likely, then
that word (_kópa_) simply provides no evidence.
Patrick:
As we see, OI <c> can, apparently, also be a response to PIE *ske- (I
would say *sk^V). I am beginning to wonder if <ch> does not tell us that the
PIE form should be *sk^(h)-?
Richard:
I haven't seen any evidence for *sk or *sk^ yielding Sanskrit /c/, though I
can't yet preclude it arising through Grassman's law. What you need is a
case where cognate words in Sanskrit yield /sk/ and /c/, but not /k/. It
would be better if /sk/ rather than /k/ always occurred in the cognates in
other languages, i.e. if we could demonstrate 's immobile'.
***
Patrick:
What makes this so infernally problematical is that we can never really be
sure about a satem reflex of *k^ vs. *sk^.
I would suggest códati as a possible example of *sk^ = Old Indian <c>; and I
would emend (s)keud- to *(s)k^(h)eu-d- (from *k^(h)e with association to
'fast'); yes, I am aware that by the conjecture I made above, I should
predict *chódati.
Of course, we also have skundate. This is obviously zero-grade *sk^undetoi.
I would have to emend my proposed *sk^ to <c> to be excepted when followed
by <u>.
**************************
(new topic)
Patrick:
I feel as if I am arguing with one hand tied behind my back because I must
reconstruct PIE forms with *e and *o for satem when I do not believe that
satem had them, except as resolutions of diphthongs.
How would it be if we put the shoe on the other foot?
I will propose my 'rules', and you show me why, in your opinion, they do not
work?