Jens:
> Thus, if you leave it up to the metrical indications supplied by the
> Rigveda to decide whether the first syllable of pada's is longer or
> shorter than that of apa's 'from the water', you can only find that
> they are both short.
Yes, they are, but then they both would have closed first syllables
because we'd be talking about */ped.dos/ and */xep.pos/.
> If that is reason enough to speak of the first syllable of pada's as
> closed then the language just has no open syllables with a short vowel,
> in which case the whole question loses its meaning.
No, it has nothing to do with *t:, the ancestor of IE *d. It has nothing
to do with any phoneme. I was thinking about this and I came up with an
idea.
What if there is a constraint on _every_ syllable that there is at least
one consonant in the onset. A syllable can be either light (CV) or heavy
but it must have an onset.
Now, each word has a root in the word. We know that there is a restriction
on a root syllable such that there must be a coda. (I'm speaking more
specifically of noun, verb and adjective roots, of course.) With *pedos,
we may syllabify the word as */pe.'dos/ but this destroys constraints on
the root syllable. We might have */ped.'as/ but this also violates the
ban against onsetless syllables. So */ped.'das/ is the result, by
splitting the coda of the root syllable in half to be shared with the
onset of the next syllable. This then would work for *gWénex */'gWen.nex/.
However for reduplicated presents, we'd see */bHi.'bHer.ti/ from
eLIE *bebérti /bHe.'bHer.ti/. This is because the first syllable is not
the root syllable and therefore is allowed to be CV. This leaves a longer
pretonic *e to rise to *i while the *e in the root syllable of *pedás
remains as is because it is in a closed syllable.
This would also work on thematic stems in compounds which occur
pretonically because the thematic vowel is not part of the root syllable,
hence it is free to be CV and thus open. Since it is in an open syllable,
it rises to *i without problems.
= gLeN