On 2007-01-30 23:28, Miguel Carrasquer Vidal wrote:
> My explanation is slightly different:
I know. I am just trying to explore alternative possibilities. The PIE
personal pronouns are so quirky and so unlike anything else even in the
pronomonal system that it's hard to make any safe assumptions about
their forms (like "such-and-such an ending should be expected in
such-and-such a case form"). The whole system is probably extremely
archaic and can't be fully understood in terms of the "standard" PIE
case system. So even if an *m appears in the accusative, I'm wary of
"obvious conclusions", especially if the *m alternates with something
else. Reduplicated personal pronouns are nothing particularly exotic,
they can be found e.g. in Latin and Welsh.
> I think that the
> element *-mé/-wé in the acc/gen of personal pronouns is
> cognate with the normal accusative ending *-m. The
> difference is one of accentuation. Ultimately I would
> reconstruct the form as *mwa (the accusative in *-a of a
> demonstrative/relative pronoun *mu).
>
> **mu-mwá > *m&mWé > *mé (acc), *méme (> *méne) (gen)
> **tu-mwá > *t&mWé > *twé (acc), *téwe (gen)
> **su-mwá > *s&mWé > *swé (acc), *séwe (gen)
>
> The genitive forms appear to be derived from the accusative
> forms through some kind of vrddhi.
In my tentative proposal, the clitic accusative is the bare oblique stem
without any endings, while the accented accusative and the form labelled
"genitive" are emphatic counterparts thereof.
> The structure of the plural/dual oblique is similar:
>
> pl.
> *mu-atu-mwá > n.smé ~ nó(:)s
> *tu-atu-mwá > usmé ~ wó(:)s
> *su-atu-mwá > [smé]
> du.
> *mu-iku-mwá > n.h3wé ~ nóh3
> *tu-iku-mwá > uh3wé ~ wóh3
> *su-iku-mwá > [sh3wé]
>
> Again, the variants nó(:)s etc. seem to be built by some
> kind of vrddhi (n&smé > n&:s(me) > nós), although different
> from the one in the singular.
The distribution of *-mé and *-wé in the dual and plural forms is far
less clear than in the singular. For example, Katz argues for 2pl. *uswé
rather than *usmé, and the actual evidence hardly supports either form
beyond reasonable doubt. Here is what I imagine might have happened:
pl. *mwe-D-mwe, *twe-D-twe > *me-s-me, *(t)we-s-we > *n.s-mé, *us-wé
du. *mwe-G-mwe, *twe-G-twe > *me-h3-me, *(t)we-h3-we > *N.h3-mé, *uh3-wé
If one accepts Jens's suggestion that *-h3m- > *-h3w- at some pre-PIE
stage, we get 1du. *N.h3-wé, and analogy turns that into *n.h3-wé, while
*uswé is perhaps replaced by *usmé. The regularised system is:
pl. *n.s-mé, *us-mé
du. *n.h3-wé, *uh3-wé
Hence the new enclitic forms, *nVs, *wVs and *noh3, *woh3 (further
transformed into pseudothematic *no:s, *wo:s, *no:, *wo:?), back-derived
from the emphatic pronouns. The original unreduplicated forms are
perhaps partly preserved as subject pronouns: 1pl. *wei(s), 1du. *we <
*(m)we-D, *(m)we-G (assuming the loss of *G word-finally and the change
of word-final *D into a palatal glide, cf. *to-i). The 2pl. *ju:(s) (<
*ju-i-?) and 2du. *ju are absolutely puzzling, as far as I'm concerned.
Piotr