[tied] Re: Dog

> I guess that Pokorny was wrong, because we have attested place

names

> in Illyria with second part -via: Dom-a-via > Argentaria >
> Srebernica, Set-o-via (today Studenci). So, my segmentation looks
> more plausible: Cand-a-via. This kind of composition is very common
> in Albanian too: mal-a-zog 'bird of forest', gur-a-kuq 'red

stonte',

> etc.

Latin canis obviously doesn't fit with *kWn.- either (loan from the
Old European 'a-language'), so how would the assumption of a proto-
Albanian *kan(d)- be in violation of Occam? That would not add any
extra root to the ones we already have to assume.

Torsten