On 2007-06-08 23:41, tgpedersen wrote:
> In the adjective, yes. But...
> I argued there was an alternation.
It my be useful to round up examples, actual or apparent, of *tt > *st
(1) As far as I can see, *-t-tro- yields Gmc. *-stra-, which is
reminiscent of the Latin treatment of the same sequence:
*Blo:stra- 'offering' from *Blo:t-a/i- '(kill for a) sacrifice'
*Gelstra- 'tax' from *GelD-a/i- 'yield, pay'
(2) 2sg. pret. & pret.-pres.:
*waist(a) < *woid-th2a
*warst(a) < *(we-)wort-th2a
This is in all likelihood analogical, since the *t, expected after all
non-coronal obstruents and PIE *s, is generalised also elsewhere, as in
*kWamt(a) < *(gWe-)gWom-th2a, where it _must_ be analogical.
(3) In isolated words:
*ru(:)sta- 'rust' (if from *h1rudH-to-), and a few i-stems like *xlasti-
'load' (the verb is *xlaT-a/i-). Only *ru(:)sta- goes back to a securely
reconstructed PIE root, but the actual derivation is unclear, the vowel
quantity is puzzling, and since *h1reudH- appears quite often with an
*-s- extension, the reconstruction *h1rudH-to- is not all that compelling.