On Sun, 14 Sep 2003, Mate Kapovic wrote:
>
> [Jens:]
> >I suppose this also explains the modest number
> > of roots with other vowels than /e/.
>
> Would you be so kind to elaborate a bit on this last statement?
Certainly! I would like to bridge the gap between belligerant groups none
of whom are telling lies. Some say the vowel is /e/ in practically all
IE roots, and some insist that there *are* roots with /a/ or even /o/
underlyingly. I think both statements are true:
If there was a Sanskrit-type collapse of a number of vowels into (the
prestage of) /e/ in some prestage of the IE protolanguage, we would find
all the old vowels as /e/. But if there was some time between that event
and the spread of IE, the language could have had time to absorb or invent
new words with other vocalisms. This would give the picture we find: IE
root vowels are most often /e/ (and its later variants produced by the
events we summarize as ablaut), but some have /a/ or /o/. The deviating
items do not all have to be younger words, though most probably are; there
could also have been restrictions to the vowel collapse rule which left
the old vowel qualities unchanged (or differently changed) under special
conditions. Thus, if 'nose' and 'salt' are *na:s-, *sa:l- with old long
/a:/ and no laryngeals as I have supposed (rightly or wrongly), they may
have simply retained their original vowels because the vowel collapse rule
only applied to short vowels.
Jens