>>I was wondering if the original pattern wasn't something like **sab(a)túm
>
>>*septm.', without a stress shift.
>No. ...Stress on the
>last syllable was possible in Semitic if the vowel was long, but not
>otherwise.
This is not the case in Classical Hebrew, where an accented syllable may
have a short or long vowel, in closed or open syllable. The pattern you
suggest is commonest, but accented short final vowels do occur, eg
/te:ra:e'/ (Gen 1:9), /thye'/ (Ezek 23:32) /ama'r/ (passim) /&Se'r/
(passim).
Admittedly, this is attested somewhat later than the period of a
borrowing from Semitic into IE. So just how sure are we of your rule in
earlier Semitic?
Peter