On 2006-02-21 16:51, alexandru_mg3 wrote:
> Why to exclude a PIE *melk(') with k(') for the Germanic words (see
> Slavic *melk-o) ?
Because of Grimm's Law. PIE *k(^) becomes Germanic *x (h), which may
then become *g if originally preceded by an unaccented syllable. For
example, something like *mélko- would have become PGmc. *melxa-, and
something like *ml.k(^)ó- would have become PGmc. *mulga-. The actually
attested Germanic forms (*melk-/*mulk-) perfectly match Lat. mulgeo:,
Gk. amelgo: and Skt. {mr.j-}, as well as the Slavic root *melz- and
Lith. mélz^-/mi`lz^- 'to milk'; it's the Slavic noun *melko that is the
odd man out, most plausibly explained as a loan from a non-satem dialect
with *k from *g^. Only Germanic meets that description, and what's more,
Germanic has just the right noun with the right meaning. That
practically clinches the case.
Piotr