Re: Dissimilation of gW/kWVw to gVw/kVw

> >And, in connection with the vowel-glides, do you assume as I do,

that PIE *k^- is a result of pre-PIE *ke?

>
> That would be *ki in my scheme. No. The PIE *k^-series is
> many times more frequent than the *k-series, which leads me
> to believe that it stands for unmarked /k/ (/g/, /gh/).
> That would make the *k-series marked back velars (uvulars)
> in pre-PIE (*q, *G, *Gh). There *is* a connection with
> vowel quality, as follows:
>
> **-ká- > *-k^é-
> **-kí- > *-k^é-
> **-kú- > *-kWé-
>
> **-qá- > *-ká-
> **-qí- > *-ké-
> **-qú- > *-kWá-
>
> So after velars/uvulars, it's _almost_ possible to
> reconstruct the original vowel quality, which is impossible
> after other consonants. For instance, *t:
>
> **-tá- > *-té-
> **-tí- > *-té-
> **-tú- > *-té-
>
> (but when *t follows:
>
> **-át- > *-ét-
> **-ít- > *-ét-
> **-út- > *-és-
> )
>

You propose that the plain *k-series are derived from pre-PIE
uvulars *q etc. I believe they occur only in loanwords. Can you
propose a way to decide which alternative is true?


Torsten