On Thu, 21 Feb 2008 14:46:04 -0600, "Patrick Ryan"
<
proto-language@...> wrote:
>Miguel, that is not what I was getting at.
>
>Why is it *dwo- rather than *dwé-?
>
>Before endings added.
Well, without endings added, the form appears to be *dwi-.
The form *dwóh3 can equally be written *dwéh3, as /e/ > /o/
is automatic before *h3. However, given that the thematic
vowel is lengthened (i.e. becomes *o) before voiced
segments, I believe the proper rendition is *dwóh3: *h3 was
voiced. Before *-ih1-, the thematic vowel must also appear
as *o, because *i is voiced.
The word declines as a normal o-stem dual:
NA *d(u)wo-h3 => Skt. d(u)vá:(u), Greek duó:, etc.
n./f. *d(u)wo-ih1 => Skt. d(u)vé:, Lith dvì, OCS dvê, OE
twa:
G *d(u)wo-ih1-h3ous => Skt. d(u)váyo:s
L *d(u)wo-ih1-h3u(m) => Grk. duoîn
DIAb *d(u)wo-ih1-bh(i)/m-oh3(m) => Av. dvae:ibya, Lith.
dviem~
On a more speculative note, I reconstruct the dual ending
for some pre-stage of PIE (as well as Uralic, Eskimo-Aleut
and others) as *-ik- (construct/pausal *-ik > *-iN,
nominative *-iku, accusative/genitive *-iki). If the root
was plain *du- (c.q. *t.u-), that would have given PIE:
C **du-íN > *dwéj
N **du-íku > *dwéh3 > *dwóh3
A **du-íki > +dwéh1
*dwóh3 might then have been reinterpreted as a thematic
form, yielding neuter (< acc.) *dwoih1 instead of expected
+dwéh1. The construct form appears as PIE *dwi-, *dwéj-.
Alternatively, the form may have been thematic to begin
with:
**du-á-(i)ku > *dwóh3
**du-a-íki > *dwoíh1 > *dwóih1,
but that would perhaps have given a construct form
*du-á-(i)N > +dwór- ~ +dur-, which is AFAIK completely
unattested.
If we leave the root as *du-, one attractive possibility
that opens up is explaining *dék^m.t "10" as something like
**dú-kamt- "two hands".
=======================
Miguel Carrasquer Vidal
miguelc@...