Re: [tied] Re: Latin barba in disaccord with Grimm's Law?

On Tue, 27 Jun 2006 17:52:39 -0700 (PDT), Sean Whalen
<stlatos@...> wrote:

>--- Miguel Carrasquer <mcv@...> wrote:
>
>> On Mon, 26 Jun 2006 22:23:04 -0700 (PDT), Sean
>> Whalen
>> <stlatos@...> wrote:
>>
>> >--- Miguel Carrasquer <mcv@...> wrote:
>> >
>> >> On Mon, 26 Jun 2006 12:02:42 -0700 (PDT), Sean
>> >> Whalen
>> >> <stlatos@...> wrote:
>> >>
>> >> >--- Miguel Carrasquer <mcv@...> wrote:
>> >> >
>> >> >> No, I'm saying there is no o > a in Armenian.
>> >>
>> >> What there is in Armenian is a marked tendency to
>> >> favour the
>> >> oblique (weak) stem (cf. for instance barjr <
>> >> *bherg^h- ~
>> >> *bhr.g^h-, etc.):
>> >
>> > I think e>o after p/b/bh; for many IE languages.
>> >This is too common for me to accept any other
>> >explanation.
>>
>> There is no such development in Armenian (berem <
>> *bher-,
>> bekanem < *bheg-, etc.).
>
> It's not regular in any IE language.

I take it you're not a follower of August Leskien's.

>> Yes, I should have checked. In any case, the root
>> is
>> *h2eid- (Arm. ayt, aytnum, Lat. aemidus) ~ *h2oid-
>> (Grk.
>> oidos, oideo:, oidma) ~ *h2id- (Latv. idra).
>>
>> >> The Armenian verb is
>> >> based on
>> >> the e-grade root (or zero-grade) *h2eid- ~ *h2id-
>> (>
>> >> ayt-),
>> >> LIV 258.
>> >
>> > I don't believe *h2o remains; instead *h2o > h2a.
>>
>> Certainly not.
>
> There are forms that should have o-grade which show
>a; so (I'm using x for h2):
>
>*-oxor > *-axar (or -0, -i ending) > -ar L, -mai/-me:n
>Gk

I suppose you're referring to the middle endings *-h2ai
(*-h2ar) and *-th2ai (*-th2ar). That's the exception to the
rule, although in fact there is no o-grade there, just a
vowel which is basically equivalent to the thematic vowel
(except that it follows the personal morphs instead of
preceding them). The thematic vowel is /e/ before voiceless
segments, /o/ before voiced (middle 3rd. person *-(t)o,
*-nto / *-ro is analogical after *-(t)oi / *-ntoi ~ *-roi,
*-(t)or / *-ntor ~ *-ron, cf. perfect 3sg. *-e), and _can_
be coloured by *h2.

>*xoiteye- > aitéo: "ask for" Gk
>*xonx^.mo- > ánemos "wind" Gk

No reason to expect o-grade here, in my opinion.

What you should be looking for are cases of Ablaut variation
*h2e ~ *h2o (a ~ o) and *eh2 ~ *oh2 (a: ~ o:). Such cases
are abundant, (*h2ost()- / *h2est()- "bone" (Celtic *ast-),
*h2owi- ~ *h2ewi- "sheep" (Skt. avi-), to give just two off
the top of my head).

> Since the changes in Latin for *oi can vary
>depending on its environment I can't be sure if
>aemidus would indicate *ai or *oi.
>
> Do you think analogy or a new formation created
>oidos not *ai-?

There are several possibilities. The root may show heavy
Ablaut *h2oid- (strong) ~ *h2eid- (weak), but that doesn't
fit in well with *h2id- (or *h2i-n-d-). We may have
middleweight Ablaut *h2oid- ~ *h2id-, with secondary
analogical *h2eid-. If we have regular Ablaut, then *h3eid-
~ *h3id- is a possibility, if *h3id- can somehow give *&3i-

> ai- in Armenian and Latin (can it?).

>> > Are there any cases of PIE *oi you think made it
>> >into Armenian?
>>
>> Of course: me:z "urine" < *moig^hos, for instance.
>
> But e: (in me:j "middle", -e:r vs -ayr, -oyr < *uyr
>third singular imperfect) comes from e or ei; there's
>no guarantee there was o-grade here.

*moig^hós had o-grade because the initial laryngeal is
deleted (cf. Grk. omeikh- vs. moikhós) (de Saussure,
Rasmussen).

>> >> >atamn vs odont- Gk
>> >>
>> >> Zero grade *&1dn.(t)-mn.t-ós > atamán (G). Cf.
>> >Olsen
>> >> 505.
>> >
>> > Some Mycenean forms may indicate *down.t- (I
>> >reconstruct v not w) without original connection to
>> >"eat". Why would h1 appear as a here?
>>
>> That's the standard Armenian reflex of a vocalized
>> laryngeal.
>
> But h1 (x^) in
>
>* x^regWo-s > érebos, erek

Olsen derives this from *regWos, with "unetymological" e- in
Armenian and Greek to avoid initial r-. Cf. for Armenian
erku < rku < dWu: < dwo: and similar cases.

>* x^n^ewn. > ennéa, inn

Well that's a can of worms. In general, IE languages
reflect */newn./, which could be from */h1newn./, except
that does not explain either the Greek form (with /nn/) or
the Armenian form (with */e-/ > /i-/, not /a-/). Personally,
I believe there is a prefixed (and optional) element
*(h1)edh- "1" here (as there often is in words for "9"), for
which cf. Slavic jed-inU "1", Armenian ez "only", and
perhaps the obscure Anatolian morpheme -ed(an)- wich appears
in Hittite pronouns in places where elsewhere in IE we have
*-sm-. */h1edh-newn./ could have given Greek *ennewa >
ennea and Armenian *enun > inn, as well as go a long way to
explain Baltic and Slavic *dewn.- > deveNtI.

>and the Greek forms for "name" and "tooth" both have
>variants with e- not o- (which seems like assimilation
>to the following o).

Exactly. Armenian has a- in both (anun and atamn).

>> >> *&1noh3mn > anu(w)n (/m/ > /w/ before or after
>> /o:/,
>> >> Olsen
>> >> 132). Or perhaps zero-grade *&1nh3mn. (= Slavic
>> >> *inmin >
>> >> jImeN)
>> >
>> > Where is this form from?
>>
>> Oblique (e.g. genitive *h1nh3mnós). The original
>> paradigm
>> was NA *h1nóh3mn., G *h1nh3mnós, L *h1nh3méni.
>
> I meant jImeN; I've never seen that before (I just
>checked to make sure the OCS form was imeN).

*jI- gives i- in OCS. The initial (unlike the final) is
more faithfully reflected in Czech jméno.

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