Miguel:
> I don't see why. The Southern forms (Arabic sitt-, Sabaic sdt_, Ge'ez
> s&d&s-) unequivocally point to *s-d-t_. Outside of Semitic, we have
> Egyptian s-r-s- ~ s-j-s- and Berber <sd.is>, which confirm the
> reconstruction *sidc^-.
Okay, that's convincing enough for me. Thanks a lot, Miguel. But now on to
more specifics about this *sidc^-. You mentioned Diakonoff's system which
seems to reinterpret those pesky palatal sibilants as laterals, correct?
If we follow his idea, this gives phonemic elbowroom for potential
allophony between *s and *s^ if *s were to become palatalized, no?
As a result, the numeral could have been pronounced as *[S(W)Id(t)S-]
because of *s- neighbouring *i. As well, one would expect that it would be
hard to distinguish between *sidc^- and *sids^-. The latter sibilant could
end up sounding like *c^ merely because of the former *d, don't you think?
So if we can make *s allophonic with *s^, this would explain *sw- in IE
(at least in some dialects of IE) for "6" while there is only ever *s- for
"seven".
Personally, the idea of specially reconstructing Mid IE **sWetse is overly
indulgeant. We need only reconstruct what is expected to have been the
previous form based on regular sound rules: *sWeksa.
I know what you're thinking. gLeN changed something again.
Before anyone harps about why I'm saying *sWeksa and not *sWekse as I used
to endorse, you may recall that I briefly mentioned a sudden change in
thinking I had a bit of a while back concerning what I thought was my
indestructible rule called "Mid IE Vocalic Constraint". In that rule, Mid
IE had restrictions on what vowels could be placed in unstressed
syllables: Only *e (pronounced as schwa) could exist in unstressed
position.
Well it turns out that there is some evidence (such as the origin of the
thematic verbs and the relationship between durative and aorist
conjugation) which forces me to consider that allophones of the stressed
full vowels *e and *a had _contrasted_ in unstressed position, albeit as
non-peripherals /I/ and /@/ respectively. We even see unstressed MIE *e
/I/ eventually rising to /i/ when preceding stress accent, hence
explaining the use of *i between compounded elements (nothing other than
the prestress allomorph of the thematic vowel which was normally *e at the
time). At that, we now must treat MIE *a as the sneaky "disappearing
schwa" while MIE *e continues on in unstressed positions as the new answer
to a schwa in Late IE... and it is _that_ vowel that becomes the *e/*o
alternating thematic vowel we all know and love.
AAAAAAAnyways, before I get carried away, I now reconstruct MIE *sWeksa
(pronounced /'sWeks@/) with unstressed *a regularly disappearing in Late
IE. And this perfectly accounts for everything, based on the internal
facts of IE as well as the external comparison of the numeral with the
Semitic "construct form" counterpart in its feminine form, *sidc^u,
especially if it was pronounced ['S(W)IdSU] as I suggest. It would then be
most probably the combination of voiced stop *d and voiceless *c^ that led
to the odd resolution of *ks in IE.
= gLeN