On Thu, 11 Dec 2003 16:56:31 +0100 (MET), Jens Elmegaard Rasmussen
<
jer@...> wrote:
>On Wed, 10 Dec 2003, Miguel Carrasquer wrote:
>>
>> It's the same principle at work as I posit for the PIE plural/dual forms:
>>
>> 1. *mu-áti > *wéy < acc. (less commonly *més < nom.)
>> *mu-íki > *wéh1 < acc. (no evidence for **móh3)
>>
>> 2. *tu-átu > *yús < nom. (less commonly *swéy < acc.)
>> *tu-íku > *yúh3 < nom. (no evidence for **swéh1)
>
>I completely fail to see by what discovery procedure one gets from IE
>*wéy, *yús to pre-PIE *mu-áti, *tu-átu, and from IE *wéh1, *yúh3 to
>pre-PIE *mu-íki, *tu-íku. I see no way of retrieving such a system from
>any traces left in IE. As I have shown the IE system may be analysed in a
>much simpler and more consistent way.
>
>You may have your reasons to posit your original system, but the reasons
>appear to be alien to Indo-European. Am I right?
No.
The "discovery procedure", if I remember correctly, started with the thorny
problem of Armenian pl. -k`. After much thinking, I came to the conclusion
that the soundlaw to explain this development was already known: Armenian
*sw > k` (as in *swesor > k`oyr).
Is there more evidence that the plural morpheme was originally *-esW? I
looked, and found the loc.pl. *-su ~ *-si (this assumes, not unreasonably,
that the morpheme *-s of the plural oblique cases (*-bhio-s, *-bhi-s,
*-s-u) is cognate with the nom.pl. *-es. The loc.sg. has an added element
*-i, so the loc.pl. logically had to be *-sW + *-i > *-su, with umlaut of
the high vowel. So far, so good. The plural morpheme was *-(e)sW. Now
*sW gives Slavic *s^ (e.g. s^estI < *sWek^s-ti), so the next step was
hypothesizing that the 2sg. verbal ending was also *sW (explaining Slavic
thematic *-es^I < *-e-sW + -i).
The ending now makes perfect sense as simple agglutination of the personal
pronoun *-tu. So final *-tu > *-sW, as already suggested by the well-known
PIE alternation of *t / *s in *meh1not-/*meh1nes- and the ptc.med.
*-wot-/*-us- This implies that the plural *-esW can be **-Vtu as well as
**-Vsu.
If the 2sg. ending is agglutinated *tu, then the 1st person verbal ending
is *-mu, which explains the alternation *m ~ *w (Luw. -wi, du. *-wh2a(s),
pl. Hitt. *-wen, and thematic 1sg. *-o: < *-o:w (= Toch -ew) < *-owu <
*-o-mW-i.
Medial *-tW- remains as *-t-, as can be seen from the verbal ending *-te
(*-tes, *-ten, *-ter), from *-tu + something else. But can initial *tW-
also give *sW-?
Here the first association is with another PIE mystery, the alternation
*to- / *so- in the demonstrative pronoun. What if *so is from *tw-o (*tu-
+ something) and the oblique/neuter from *t-o (*ta- + something)? The
second thing that came to mind were the 2pl./2du. pronoun forms in *sw-.
If what can be expected in the 2nd. person plural is *sw-, on the model of
1pl. *mw-, I had no longer any need to explain away these forms by ad-hoc
metatheses, loss of u- and whatnot... Hitt. sume:s, Celtic *sWi:, Greek
spho:, etc. directly continue PIE *sWey-, which is structured exactly like
*wey- "we".
This alternation *-sW / *-y had already caught my eye in the o-stems, where
besides plural *-o:s(es) we have an oblique stem *-oy-. The personal
pronouns show the same pattern (but without thematic vowel): *-(e)s in *mes
and *ju:s, *-ey in *wey- and *swey-. The forms in *-ey must then be old
oblique stems.
But how to explain a nom.pl. in *-Vtu versus an oblique in *-Vy? Well,
here I must admit that external comparisons did lead the way. We have the
same thing in Uralic (nom. *-t, obl. *-j) and we have the same thing in
Semitic (nom.pl. *-atu, acc/gen.pl. *-ati). This cannot be a coincidence.
In the same way that PIE had a soundlaw *-tu > *-sW, it must also have had
a soundlaw *-ti > *-y (this one shared with Uralic). Additional evidence
comes from the numeral *trey-, with ordinal *tr.t(i)yos, pointing to
**tréti > *tréy, **treti-ós > *tr.t(i)yós.
Having established a paradigm *-atu/*-ati for the plural, we can expect the
dual to be structured in a similar way. Based again on Uralic *-k, and the
lack of *-k in PIE (certainly to be explained by fricativization of *-k >
*-h2, as in the 1p. stative/perfect), I conjectured that the dual forms may
have been in origin *-aku, obl. *-aki, with *-ku > *-h3 and *-ki > *-h1.
As it happens, there are reasons (the dual forms *-ih1 and *-ye(h1)) to
think there was also a difference in vocalism, so I changed that to
**-iku/**-iki subsequently.
Combining all of this with Jens' insights into the pronominal system (1.
*m-u-, 2. *t-u-, 3. *s-u-), I arrive at:
>> 1. *mu-áti > *wéy < acc. (less commonly *més < nom.)
>> *mu-íki > *wéh1 < acc. (no evidence for **móh3)
>>
>> 2. *tu-átu > *yús < nom. (less commonly *swéy < acc.)
>> *tu-íku > *yúh3 < nom. (no evidence for **swéh1)
In sum, the whole thing is solidly based on internal reconstruction of PIE,
with only some extra-IE (Semitic, Uralic) considerations thrown in, in
exactly the same way that Jens arrives at the reconstructions du. **-G (>
-h1/-h3) and pl. **-D (> -y/-s) by pointing to parallels in Uralic and
Eskimo-Aleut.
=======================
Miguel Carrasquer Vidal
mcv@...