On 2007-02-14 16:51, tgpedersen wrote:
> A small point: you've been taken hostage by your own notation, where
> you consistently write *-ah2 etc, with the vowel already colored by
> the laryngeal, but *-em; if you'd assume h2 preserves, rather than
> creates the /a/ the root would be *gWam-/*gWax-, or traditionally,
> *gWem-/*gWex- which means the alternating suffix is *-m/*-x, not
> *-em/*-ax, which is close to *-n,W/*-xW and the reduplication you
> argue for.
As it's just a matter of notation (of the kind everybody in this line of
trade knows how to handle), I don't worry much about it. The PIE vowel
is predictable from the context, so one might as well write the roots as
*gWVh2-, *gWVm-, etc. The division *gWe-m- and divisions like *dre-m- in
parallel cases are hardly satisfactory, so the actual suffix (if my
analysis is correct) was probably of the *-VC- type (like most IE
suffixes), which means that the "ultimate root" is just *gW-. Whether
this means that pre-PIE once allowed uniconsonantal roots, or that
something more complex had shrunk down to *gW- by PIE times is a moot
question. The latter option seems preferable to me, as other roots which
I suspect might contain similar suffixes (like *drem ~ *drah2- ~ *dreu-,
*pleu-, *sreu-) are typically triconsonantal.
Piotr