>But in Slavic there was no gYH>z but gH>g , suggesting gYHG > gHG at least, a theory helped by the presence of colored a. Words for 'goose' and 'duck' both contain a and are either C- or i-stems, so the possibility of a compound word is high.
> Of course I'm aware of "Lubotsky's Law", but have my reservations about
> it. What looks like *a: ~ *a ablaut is by no means restricted to roots
> ending in a media. For example, I find myself unable to accept the
> Leiden analyses of 'nose' or 'goose' as plausible real-world patterns
> and much prefer straightforward reconstructions like *(h)na:s-, *g^Ha:ns-.