From: Piotr Gasiorowski
Message: 31014
Date: 2004-02-13
----- Original Message -----
From: "elmeras2000" <jer@...>
To: <cybalist@yahoogroups.com>
Sent: Wednesday, February 11, 2004 11:20 PM
Subject: [tied] Re: Eggs from birds and swift horses (was: the palatal sham)
> To posit new types we need solid evidence, not just theoretical
possibilities. Avestan pouru- is a perfect match for Skt. puru- from
*pl.Hu-. And Av. vouru- matches Skt. uru- completely as IE *wr.Hu-.
The variant Av. uru- may be /vru-/ (or /ru-/) strongly reminiscent
of Ved. gru-mus.t.i- with compositional gru- for free guru- (*gWr.Hu-
, Gk. bar�s).
You're right. I warned you it was just a wild guess :-)
> I do not normally see the infix -o- occupying the
wrong position: Gk. trop�o: and trop�:, not *torp-; phlogm�s not
*pholg-; ON rakr 'upright' (*H3rog^-o-s) not *ark-, etc. There may
be a case in Slavic rota from *wrot(h)aH2, if that has anything to
do with the rooot *werH1- 'speak', but that's the other way around.
OK, but you could expect to find it in the wrong place occasionally _if you
know where to look_. The motivation behind metathesis is the need to
optimise the phonetic string (with regard to language-specific phonotactic
preferences). The *o reflex of the infixed *R will be attracted into the
wrong position only if the resulting structure is phonotactically superior
to other segment orderings. In a root of the canonical form *CLeh (where "C"
is an obstruent, "L" a lateral or rhotic and "h" a laryngeal) Rasmussen
infixation can be expected to result in *CRL (with the Saussurean loss of
the laryngeal), a simpler structure in terms of both syllabification and
segmental count than the next-best candidate *CLRh (which is still better
than the input structrure *RCL(h)). *CLeh roots are relatively rare, but
*pleh1- has just the right shape, hence my hypothesis that <poll�-> is a
valid example -- a thematised and *R-infixed derivative of *pleh1-mn.;
honestly, I can't see what else it could be (I don't insist on a similar
analysis of pol�-, which may be a slightly irregular reflex of *pl.h1�-).
And you _can_ have it "the other way round" too. Consider *h2�ug-es-
'growth, power' and its derivatives. In the consonantal skeleton *{h2wgs}
the optimal place for *R insertion is right in the middle: *(h2)wRgs -->
*woks-, and sure enough that's what we get in *woks-eje-, Gmc. *waxsan-,
etc., vs. *h2aug-, *h2ug-r�-, *h2�ug-jos-, *h2�ug-mon-. (I know Gk.
a(w)ekso: is difficult to account for, but since it's a variant of <aukso:>
I'm not losing hope yet).
Piotr