[Jens wrote:]
>--- In cybalist@yahoogroups.com, Miguel Carrasquer
><mcv@...> wrote:
>> On Wed, 11 Aug 2004 00:14:40 +0200 (MET DST), Jens Elmegaard
>> Rasmussen <jer@...> wrote:
>>
>> >The effect of analogy could just as well be like this: *-to-syo-
>> *-toyo- > *-teyyo- (after *eyyo-, the prestage of eius) > *-teiyo-
>> > *-ti:yo-. In that case it works fine, why mustn't it?
>>
>> The analogical steps are completely unnecessary. It works
>> just fine by regular processes:
>>
>> *-tosyo > *-toyyo > *-teyyo > *-ti:(y)o > -ti:us
>
>That is phonetically regular only if there is no regular
>simplification of *-yy- after non-first vowel (mora). This
>traditional belief may be right, in which case there is no
>possibility of accounting for Lat. ferat in the same way as
>OIr. beraid although both are subjunctives of thematic
>verbs. You are writing as if this were a new thought, it was
>in fact Anders Jørgensen's point of departure, i.e. general
>opinion before he got started. The new observation is that
>to combine the Italo-Celtic morphological colleagues it
>would take the introduction of a rule simplifying *-yy-
>after non-first vowel to *-y- before the loss of
>intervocalic *-y- in Italic. That works fine, except that a
>few forms will then have to be analogical.
If I'm being unfair, you're being the opposite here. It's
not a few forms: it's _all_ of them. Apart from the
a:-subjunctive itself, no independent evidence for -yy- >
-y- > -0- in non-first syllable has been put forward. That
makes the argument circular: the a:-subjunctive comes from
-asyV- > -ayyV- > -ayV- > -aV- > -a:- because of a soundlaw
-yy- > -y- in non-initial syllable, and the soundlaw is
justified by the a:-subjunctive, and nothing else (all
counterexamples being analogical).
>The analogy to be
>assumed if isti:us, u:ni:us are to avoid reflecting
>undisturbed *-esyo is not strange by any standard; you tried
>to make it look that way, but did not succeed. The fact that
>*non-assumption* of a morphological solution that will
>necessitate an analogy makes the analogy superfluous is a
>pure tautology. You know that already of course,
>I'm saying this to show others the tendentious character
>your protests have.
>
>> >The derivation of Oscan -eis from
>> >*-esyo is itself very far from being certain. If the classical
>> >explanation
>> >as the ending of i-stems is correct, there are no real problems.
>>
>> That doesn't explain G. O. eiseis, U. erer (< *esyo).
>
>Oh no? What's wrong with *eis-eis, repeated i-stem genitive?
>Since -eis is also the genitive of o-stems in O-U, the stem
>was then treated as *eiso-. What is so strange about having
>a word whose nominative is *is (Osc. iz-ik) inflect as an
>i-stem? Why do you not even mention this obvious analysis?
Because I find it unbelievable. The stem of *is is not zero
(it's *i-, obl. *e-, pl. *ei-), nor is there any reason to
think that it was bizarrely reinterpreted as such in O-U.
Silvestri, in Ramat & Ramat, gives the following forms (I'll
just quote the masc. sg. forms):
Osc. Umbr.
nom. izik ere, erek/erec
acc. ionc --
gen. eiseis, eizeis erer, irer, ererek
dat. -- esmei, esmik
loc. eíseí, eizeic esme
abl. eísúd, eizuc eru(ku), eru(com)
The Umbrian paradigm is quite close to the original PIE one,
with conservation of the stem *e-sm- in the dat. and loc.
sg. In Oscan, the whole oblique has been reshaped
analogically after the genitive *eis- (no doubt with some
help from the G.pl. *ey-s-o(:)m > U. eisunk), in Umbrian
this is limited to the ablative. It's obvious that in the
O-U genitive *eis-eis, the first *eis- is the oblique stem
(derived from the original genitive), while the second *-eis
is a secondarily added ending (analogical after o-stem
-eis). If we compare *eis-eis with Latin eiu-s and PIE
*esyo, we see that where Latin has lost the sibilant (*esy-
> eyy-), O-U has retained it (*esy- > eys-), with metathesis
of the two elements (as in Lepontic o-stem G. -oiso <
*-osyo).
The OU o-stem G.sg. -eis itself can also be derived from
*-osyo > *-oiso, perhaps directly (loc.sg. *-oi gives -ei,
and posttonic vowels are sometimes deleted, as in U. húrz <
*hortos), perhaps indirectly (*-esyo instead of *-osyo,
analogical after *is and other pronouns; i-stem genitive
-eis, analogical after *yo-stems with nom./acc.sg. -is <
*-yos, im < *-yom). In any case, the o-stem G.sg. offers no
proof that *-sy- gave *-yy- or *-y- in Osco-Umbrian, if
anything it offers additional proof for a development *-sy-
> *-ys-.
In sum, I see no independent evidence that a verbal form
*-&syV- should have given -a:- in Latin and Osco-Umbrian.
Using the evidence from the Latin and O-U. genitive forms, I
would expect *-&syV- to have yielded Latin -i:(V)- and O-U.
*-ais(V)-.
The most attractive aspect of Jørgensen's theory is of
course that it would allow a re-unification of the Italic
a:-subjunctive with the Old Irish a:-subjunctive, two forms
which had been divorced by the Sihler/McCone (possibly
others) theory linking the Old Irish a:-subjunctive (as
*-&sV-) with the Old Irish s-subjunctive (*-s(V)-).
I haven't seen Sihler, and McCone's theory only through a
short description in Russell's "Celtic Languages", where it
is stated that the s-subjunctive continues the old
subjunctive of the s-aorist, while the a:-subjunctive is
analogical after the s-aorist subjunctive of seT-roots
(TeRH-s-e/o- > TeRase/o- > TeRa:-). This explanation does
not justify a variant with *-sy- instead of *-s-, so I'm
assuming Jørgensen has a different explanation for the
sigmatic form. Something to do with the Lithuanian future?
And, since Jens gave the prototype with *h1 (*-&1se-) only,
if I remember correctly, does that mean the laryngeal is
also to be interpreted otherwise?
How do Sihler and McCone explain the incongruence between
a:-subjunctive and s-subjunctive in the 3sg., where Old
Irish has -a (< *-a:(t), which must be thematic _if_ from
*-H-se-t) vs. -0 (< *-s-t, which must be athematic)?
=======================
Miguel Carrasquer Vidal
mcv@...