Of course *-rl- can be found in some
branches. Slavic has deverbal adjectives ("past participles") in *-l- (<
*-lo-), in which the suffix attaches to all kinds of verbal stems, also
consonantal ones in *-r and *-l, e.g. *mIr-lU 'dead', dIr-lU 'having torn',
mIl-lU 'having ground'. This doesn't mean that each particular form dates back
to PIE (*mr-lo-, *dr-lo-, *ml(x)-lo-, etc.) even if the roots in question do and
if derivation involving *-lo- is PIE as well.
*ker-l-/*kar-l- is Germanic but has no
cognates with *-rl- elsewhere. I'm not saying that PIE *-rl- could not arise at
morpheme boundaries, but I don't know any convincing examples, and the fact that
the distribution of liquids in PIE was subject to phonotactic restrictions makes
me wary of any postulated liquid clusters.
Sanskrit -ll- usually appears in expressive
and dialectal words. It could derive from several Indo-Aryan clusters, even from
-dr-, as in Vedic kSullaka 'small' < kSudra-. The term bhalla-/bhalluka- is
obscure (bhalla- can also mean 'arrow'), and its form strongly suggests
dialectal origin. Most geminates (except -tt-, -dd- and -nn-, which could be
produced by common morphological processes) were clearly regarded as
characterising 'uncultivated' or 'barbarous' speech and avoided with some
consistency until the Epic Sanskrit literature (which had absorbed much
substrate and colloquial Middle Indian influence) I'd hesitate to connect
bhalla- directly with Germanic bear, let alone proposing a pre-Indo-Iranian
*bHer-lo- without any corroborative data to bridge the gap between the two
branches.
Piotr
----- Original Message -----
Sent: Sunday, October 22, 2000 9:09 PM
Subject: Re: [tied] IE -rl- > Indo-Iranian ?
I was thinking in BHALLA < *BHERLOS (?)
"bear"
And Germanic KERLA-/KARLA- , churl, Karl, Charles,
etc.