On Fri, 12 Oct 2001 22:05:12, "Glen Gordon" <
glengordon01@...>
wrote:
>Me to Miguel:
>>Of course, you may agree that *-r is a suffix but will avoid the
>>idea that *-r < *-n. Alright. However, is there truely such
>>a suffix *-r < *-r?
>
>I still got no response so lemme prod further...
There is a "suffix" -r, that does not alternate with -n-, and which
has both animate and inanimate forms. For instance, g^hes-r "hand".
>If *kWetwores is somehow from **kWet- + *-wr, this *-wr ending
>is known to alternate with *-n-. Afterall, *-wr is a composite
>suffix of *-w- and our pesky inanimate *-r (which derives from *-n).
>What other *-r suffixes are there in IE except for *-e:r? However,
>*-e:r doesn't belong here since the semantics would be too
>problematic.
What's problematic are the semantics of *-wr (which makes verbal
nouns, and *kwet(w)- is no verbal root).
>I don't see what your position is, Miguel. Do you think that
>*kWetwor- is completely indivisable (in which case, we still have
>your *n/*r-less *kWetesor to make sense of) or do you think that
>*-r was a suffix, and if so, which one other than the very
>alternating suffix that terminates *wodr, *?edwr and *yekWr?
See above, there are pure r-stems in IE (why shouldn't there be?).
My position is that I don't know whether *kwetwor(-es) is segmentable
or not, and if so, how.
As you know, I think the initial consonant in this word was originally
*pw, as evidenced by the f- in Germanic, and the whole may derive from
**put(u)- through *pwat(w)- > *kwet(w)-. This in turn can be related
to Afro-Asiatic *(?a)p.ut.u- "4" (Chadic <fud.u>, Egyptian <?ift.aw>,
Somali <afar>, Beja <fad.ig> and Semitic <?arba3u> < *<?a-p.t.a-3u>),
possibly to Etruscan <huth> "4" and more remotely to e.g. Basque
<laur> if from *l-aputV or even (to make Torsten happy)
Proto-Austronesian *<xepate>. As such, it seems to be a very ancient
numeral stem (like *trei- < **tilati-, cf. P-Sem. *c^ala:c^-, Basque
<hirur> < *tilut-, PAN *telu-), but in competition with other words,
such as *me(i)w- "4" (which I have connected with Etr. <mach> "5" <
*m(a)wa-kwe [with the same *-kwe "and" as in PIE *pen-kwe "...and
five"], <muv-alch> "50") and *ok^t- "4" (*ok^t-oh3 "8" = 2x4), which,
unless it's related to *kwetwor-, as suggested in EIEC (through
*o-kw(e)t- cf. vocalic prefixes also in Afro-Asiatic forms such as
Som. <a-far>, Eg. <?i-ft.aw>, Sem. <?a-rba-3>), may be related to
Uralic *kutti "6", or, again, Etr. <huth>, depending on whether Etr.
h- < *p- or from *k- (or both!). Despite all this, it's still not
clear to me whether the final part of *kwetwor- is part of the root
(comparable somehow with Semitic -a3u, Beja -ig: I'm not particularly
impressed by the chances of a development `ayn > /r/ here, although
it's phonetically plausible), or whether it's some agglutinated form
of the root *wer- (*kwet-wor < "4 times/turns", like *tris-wor "3
times"), maybe suggesting a phase where the unmarked word for "4" was
supplied by one of the other roots (*me(i)w- or *ok^t-), or whether it
is the noun *wi(:)r "man" after all, or yet something completely
different (maybe related to the -r in *<g^hes-r>, <fing-er> <
*penkw-ro-).
Let me give my speculations on the numbers to compare them with Glen's
(I can't find that post right now).
1. *sem- < **sam-
2. *duoh3 < **du(w)-aku
3. *treies < **tilati-atu
4. *kWetwores (*pWetwores) < *putu-(w)a:r(?)-atu
5. *penkWe (< *kem(t)kwe ?) < *kam(a)t- + -kua (?)
10. *dek^m(t) < *du + kam(a)t- (?)