On 2007-03-15 16:53, Sean Whalen wrote:
> And hr- isn't permitted, of course, either.
True, but the question is whether we are dealing with vowel epentehsis
in <haru-> and <varus> or with vowel _loss_ in <prae>, <trans> and a few
other words, e.g. <glo:s> (Gk. gálo:s, Sl. *zUly). Raimo Anttila points
out, in his Schwebeablaut book, that even _full_ vowels seem to have
been lost sporadically in the first syllable Latin if the result was an
acceptable stop-plus-liquid cluster, as in <gru:s> (cf. Slavic
*z^erav(j)I), derivable from something like *gerh2o-h2wi-, i.e.
'shriek-bird' (cf. Gk. gere:n < *gerh2-e:n and *gér&2-n-o- in Gk.
géranos, Celtic *garanos). There are also examples of a retained prop
vowel in the type that normally has none, e.g. caro: 'meat, flesh' <
*kr.r-o:n (a Lindeman form?) from *(s)ker- 'cut'. It seems to me, at any
rate, that the phenomenon is restricted to the environment mentoned
above, and not entirely regular.
Piotr