--- In
cybalist@yahoogroups.com, Miguel Carrasquer <mcv@...> wrote:
> I'm just trying to clean up some of the
> remaining problems (which both Winter and Narten undeniably
> have). A solution with Ablaut /e:/ ~ /0/ (besides /o/ ~ /e/
> and /o/ ~ /0/) just fits the facts better than a solution
> with /e:/ ~ /e/ ablaut.
I do not know a phonetic rule yielding /e:/ ~ zero. I observe an
alternation /e:/ ~ /e/ which I can understand as parallel to /V/ ~
zero in terms of phonetic rules giverned by the accent. I also
observe that the alternant /e:/ is sometimes replaced by /o/
or /o:/. I then handle that by the rules I believe I do understand.
> I now have an elegant solution for /e:/ ~ /0/ and
> /o/ ~ /0/ ablaut in nouns (*k^é:r(d), *k^r.dés, *pé:r(r),
> *pr.nés; *póntoh2s, *pn.th2ós), an explanation for the zero
> grade in the plural of Narten presents, possibly also an
> explanation for *tudéti/*yugóm thematics, etc.
Skt. há:rdi indicates to me that the /e:/ of 'heart' has been caused
by the lengthening effect of the collective marker *-H2; that makes
the weak stem *k^r.d- the regular weak form of *k^erd-. Nothing
shows that Hitt. pe-ir ever had a long vowel; *per-/pr.- may be
fully regular. Also *pónt-/*pn.t- may be simply V/zero, only this
time with /o/. I don't find anything to explain of a phonetic nature
in tudáti and yugám: the unaccented root segments have each lost
their old vowel.
> >The
> >core of the matter is the controversy of the basis of the hi-
> >conjugation which just will never end. I think it is simple: The
hi-
> >conjugation continues the perfect and is made up of all verbal
> >lexemes that preserved the IE perfect, if only (originally) as a
> >preterite, AND all the many other verbs whose vocalism was also -
o-
> >or so close to -o- that they were given the same inflection as
the
> >descendant of the perfect. The second part is often forgotten in
> >accounts, in which case the account loses all credibility.
>
> I've always felt that account to be completely backwards.
> After reading Jasanoff, I'm surer than ever. There is no
> way that the Hittite hi-conjugation could have evolved out
> of the perfect. I can't motivate it in a few lines here
> better than Jasanoff has in 200 pages, so I'll leave it at
> that.
But you demand that I motivate, explain and prove everything in a
one-liner? The endings of the hi-conjugation *are* basically those
of the perfect (of the other branches). And like the perfect, the hi-
conjugation regularly reflects o-vocalism. Other verbal stems are
either present stems or aorist stems (in terms of the other
branches). Thus the preterite of a Hittite verb generally has a
correspondence in the other branches, be it imperfect, aorist or
perfect. The perfect type is found where the root of the stem had o-
vocalism; that seems to be the ticket into the hi-conjugation. I
really don't understand the fuss.
> I don't see why the infinitive (or rather its pre-form)
> should have been exempt from the initial accent rule. An
> end-stressed form *molh2-téi would have given *mélh2-t(e)i
> regularly (cf. *pod-, dat. *ped-(e)i), which quite naturally
> would have been given the vocalism of the present tense,
> resulting in *mólh2tei > málti, as attested.
The Balto-Slavic infinitive does not have to reflect the weak
alternant of either the verbal stem or the root. Many verbs with
normal ablaut have infinitives with full grade. As you know I do not
accept a rule whereby the /é/ is the regular weakened ablaut variant
of /ó/. I explain this from reduplicated forms.
It needs an explanation that infinitives of the type of Lith. gérti
do not retract the accent by Hirt's law, while those of the type
kálti do. The phenomenon is accorded a natural explanation in terms
of known rules if kálti contained a nonvocalized laryngeal, while
gérti had schwa from interconsonantal laryngeal. It is well known
that reduplicated forms do not show schwas, while unreduplicated
forms do. Intensives were reduplicated and had o-vocalism in their
strong forms. Therefore, the structure of intensives will explain
all of this.
I would not like to exclude that an intensive like *w(e)r-wórt-mi,
3pl *wér-wr.t-n.ti could be reduced to *wórt-mi, *wért-n.ti, i.e.
pretty much the paradigm Jasanoff assumes for his showpiece hi-
conjugation forms, by a process of "sqashing". One would just have
to add the endings of the perfect-based conjugation. Still, I do not
know how such forms could develop into the reduplicated structure of
Indo-Iranian. If we could assume PIE variants, i.e. squashed and
unsquashed forms side by side, it might be easier. But I don't
really see the need.
Jens