29-06-03 10:36, P&G wrote:
> *sleh1b welh1bh leh1d bheh1dh kleh3dh bhreh1g' wreh1g' bheh1g'h
> leh1g'h
> treh3g bheh3g sleh1g h2reh1g etc
I was especially interested in very securely attested roots and words;
items like *bHerh2g^o- 'birch' and *swah2du- 'sweet' certainly qualify
as such. Their structure does look odd. Richard's idea that they might
represent original two-stop clusters is interesting, but still doesn't
explain the root structure. Hypothetical *bHerdg^- looks even odder than
*bHerh2g^-. The (impressionistically) high frequency of *-h2g^- may
suggest that some of those "roots" are obscured compounds with *h2ag^-
'lead, drive' as the second element, but does **bHer-h2g^-o- make any
sense as a tree name? And what about *swah2du- : **swe-h2d-u-? What the
hell might *h2ad- mean here? The only potentially esculent root that
springs to mind is that which underlies Lat. ador 'spelt' < *h2ad-es-.
Piotr