Re: [tied] Re: PIE conjugations

On Thu, 11 Jan 2001 23:34:51 +0100, "Piotr Gasiorowski"
<gpiotr@...> wrote:

>> The argument rests entirely on the transfer of *-s to the 2nd p. sg. form (*-ei > -eis), and the subsequent analogy -es / -eis :: -e / -XXX. The original shape of the 3sg. form is irrelevant to the argument (which is not mine, BTW: one can find it in e.g. Szemerényi and I suppose in not a few others).
>
>I'm sceptical, no matter who supports it. In athematic verbs we also have analogical -s in the 2sg. (phe:is < *pha:-i < *bHa:-si, Homeric eis for ei ~ essi 'thou art') but no restructuring in the 3sg. (tithe:si ~ tithe:ti). It doesn't seem as if the 2sg. pattern had automatically been generalised.

Good point. Maybe Cowgill's explanation (the *-t- in *-eti was lost
by regular sound change) offers a better solution, but I haven't seen
it.

>> Of course it's conjectural, but at least it's a real possibility, unlike deriving Grk. -eis from something like *-eth2a(i).
>
>Far be it from me to derive anything in this manner. That would be pure folly :). I doubt if *-e-th2a-i should be reconstructed at all (except possibly -- here I agree -- in a certain kind of subjunctive).

Don't tell me you accept my explanation of <esi>, <audiam>/<audie:s>
etcetera! :-)

>>The total loss of *-u after *-o: is a problem, though not as great as losing an *-i in most of the alternative theories. At least the *-o(:) itself is naturally explained, not analogical as in *-oh2[i]. I used to be a strong believer in some kind of laryngeal in the 1sg. thematic, but I've lost the faith completely: the *o is wrong, the loss of *i is inexplicable, and the *h2 is undemonstrable. The only other possibility I can think of besides the one I gave (*-omwi > *-o:[w]) is a derivation from the thematic conjunctive (**-o-om), where you get the *o: for free, and the *i was never there. But you have to lose the nasal, which is much easier to do (and attested in Luwian!) within my preferred theory. Since the thematic conjunctive is attested as *-o: everywhere, here too it seems preferrable to posit original **-o-omw (> *-o:[w]) to account for that.
>
>Losing the *i is no problem in your approach? Do you mean *-wi > *-wu is quite regular?

No it isn't. I have identified it in the locative singular of the
u-stems (if *-o:u/*-e:u is from *-owu/*-ewu) and in the locative
plural (*-su < *-swi). In the verb, however, *-i generally remains
even after what I take to have been labialized consonants (Slavic 2sg.
is *-es^I, not *-exU) [*].

>Your interpretation of Luw. -wi is far-fetched. You derive it from *-mwi via phonetic development, but this depends crucially on accepting your "rounded labials", which few if any people would be prepared to do at present: extraordinary phonemes require extraordinarily good justification. Proto-Anatolian would have to have *-mwi, otherwise Hittite -mi would not be derivable. Where else can we see Hit. -m- : Luw. -w-? A more down-to-earth alternative is the replacement of *-mi by Luw -wi motivated by the generalisation of *-wani in the 1pl.

But *-wan(i)/*-wen(i) are themselves undoubtedly connected to
non-Anatolian *-men (and both in turn, one would expect, connected to
singular *-m(i)/*-w(i)). The correspondences are irregular, but to me
that suggests that we're dealing with an unstable proto-phoneme **mw.

> Finally, if I may quote you,
>
>"[*-th2a-] might explain the Tocharian 2sg. ending in *-t, but that is not an exclusively thematic form, which makes it less relevant to the discussion of the PIE thematic present."
>
>The same holds a fortiori with regard to Luwian -wi, since (1) Luwian has no other known 1sg. present ending,

There is awimi "I come".

>(2) the traditional thematic conjugation is essentially a "non-Anatolian" concept.

Why? What are Hittite verbs like iiami, iasi "to do" or zinnami,
zinnisi "to stop" if not thematic verbs?


[*] The fact that many of the phenomena associated with my postulated
pPIE palatalized and labialized phonemes show irregularities is in
itself not surprising (though annoying): it seems to be rather common
in similar and better documented cases. In the case of Old Irish,
which had neutral, palatalized and "u-coloured" consonant qualities,
the "u-quality" is unstable (see Thurneysen, 170-171). In Tocharian,
*i and *u became *yä and *wä [something similar may have happened in
pre-PIE], but the palatalization caused by *i is sporadically, the
labialization caused by *u mostly lost (except initially, and
"erratically after PTch. *k" (Adams, Tocharian, pp. 15-17). We saw
that /nem-/, /jem-/ and /n^em-/ as reflexes of PIE *n^em- "to take"
coexist in a single language, Latvian. Why does Luwian have -wi, but
Hittite -mi, and why do both have -wen, but Greek -men? The evidence
for former labialized and palatalized consonants consists mostly of
this kind of irregular remnants (only the *kw-series was stable enough
to attain official status). I think these irregularities tell us
something about a very remote stage of pre-PIE, but proving it is
obviously a difficult task.

=======================
Miguel Carrasquer Vidal
mcv@...