21-04-2004 01:21, Miguel Carrasquer wrote:
>>Sorry, but these "superlong" everything just don't cut it for me...
>
> And this is based on what? You don't believe in
> compensatory lengthening? A form like *swéso:r had a long
> vowel (because of Szemerényi lengthening). The sandhi-form
> *swésõ: then lost the -r with compensatory lengthening,
> resulting in "super-long" (circumflex) -õ:. I don't see
> what the problem is, especially since sesuõ: and dukt~e: is
> exactly what we find in Lithuanian.
Superlong vowels are fairly rare cross-linguistically, but to the extent
that they do occur, compensatory lengthening in final syllables (where
quantity is universally less constrained than in other positons) is what
seems to produce them most frequently; cf. the overlength produced by
loss of final /-e/ in Low Saxon and Limburgish. From the point of view
of phonetic typology, the kind of overlengthening suggested in this
thread is fairly natural.
I have speculated before that in Slavic *r-stems we also see traces of
*o:-raising:
*kW(e)twó:r > *kW(e)two:: > *c^itu: (affecting analogically the reflex
of *kWétwores);
*(nekWto-)pto:r > *-p(t)o:: > *-pu: (hence, by analogy, acc.sg. *-pu:ri,
adopted as the base of the remodelled paradigm).
Piotr