Re: [tied] The phonetic value of PIE *h3 and the 'drink' root.

From: Jens Elmegaard Rasmussen
Message: 14103
Date: 2002-07-21

On Sat, 20 Jul 2002, Miguel Carrasquer wrote:

> [...] The ending -oj^ is a Locative in the
> ea-stems (plus Ablative -oj^e^ < -oj^+ *eti), but a general oblique
> (Gen./Dat./Loc. and Abl.) in kin ~ knoj^(e^) and mi ~ mioj^(e^).
> I was torn between associating Armenian -oj^ with the Sanskrit/Slavic
> instrumental in *-oyh2ah1 > *-oyya:, or with the Sanskrit
> Gen./Dat./Loc. in *-oyeh2- > *-oya:-.  The former option is more
> attractive phonetically (Arm. -yy- > -j^-), while the latter fits
> better with the function of -oj^ in Armenian.  Phonetically, come to
> think of it, I'm not aware (yet) of anything that might be
> counterevidence to a development *-oy > -oj^ in the (secondary)
> Auslaut, so perhaps Armenian -oj^ is best compared to the Sanskrit
> oblique forms (Instrumental excepted).
> >The "ea-stems" alternate between io-stem inflection in the sg. and
> ea-stem
> >inflection in the pl., with the exception of the instr. sg. in -eaw
> which
> >is rather obviously back-formed from the in -eawk'. Therefore,
> >the type is basically o-stem in the sg., a-stem in the pl., that must
> >old thematic neuters. Therefore, if they have -oj^, the -o- is
> not
> >at all unexpected and there is little basis for diagnosing it as a
> >feminine sign.
> The fact remains that -oj^ is *only* found in former feminines (or
> words secondarily attracted to a feminine declension, such as neuter
> io-stems), which is hard to explain what we're dealing with is a
> suffix *-dhi.

In the wo/ea-inflection, -oj^ is not a "general oblique", but locative
only. The genitive has -woy and is rhus an o-stem. I refuse to see that as
a sign of old feminine gender.

> >I kept knoj^ and mioj^ out of the discussion because their relevance is
> >unclear. Birgit Olsen, The Noun in Biblical Armenian (Mouton de Gruyter
> >1999) 172, arrives at an analysis by bringing in the word aloj^ 'lamb'
> >which she derives from *H1lm.bhiH2, taking -amy-/-any- to yield
> regularly
> >-oj^- (via a nasal o).
> In view of Gmc. *h1lombh-es-, could it be *h1lombh + -ih2?

I can't say, and neither can Birgit. The comparison is made with Gk.
élaphos and of course lamb, so there is no special priority of one
vocalism over another. A note: I am told the meaning of the Arm. word is
'young goat', sorry. But it still is Kleinvieh. There are apparently very
few examples of Ny > Nj^, even where one expects to find most, viz. in the
books by Dz^aukjan.

> >In like fashion she then derives *oj^ from gen.
> >*sm.-yaH2-s (or dat. *sm.-yaH2-ay, etc), taking it to have been
> >secondarily adjusted to the nom. mi, the result being mioj^. For 'woman'
> >she departs from a extension also underlying Gk. gunaik-
> >(whose -k- she gets from a nom. with *-iH2-s > *-ik-s before), this
> giving
> >gen. *gWn.H2íH2os > *kany- > *koj^ -> *kinoj^ > knoj^.
> For <mi>, I suggest a transfer to the (j)a:-stems, based on N. *smih2
> > smia (reinterpreted as *sm-ya:), oblique forms *smyoya:s/*smyoya:i >
> mioj^.

That's a thing I really don't understand. If the word was transferred to
the ja:-stems, why did it not stick to the genitive it had all along,
the one seen in Gk. miâs, at least for the ending, which was already of
the ja:-type?
In your other mail you suggest *-oyV- > *-oy > -oj^, i.e. hardening of
/y/ to /j^/ when it came to be word-final upon loss of endings, perhaps
no direct evidence, since I frankly know on no good examples of IE *-oyV-
at all. After /e/, we don't have it in erek' from *treyes, but that may
not be relevant. I would, however, find it unlikely in the extreme if
*-osyo turns into -oy without further strengthening, and simple *-oyV is
then taken to have undergone strengthening. Does less yield more? I know
of a few odd cases where that is in fact so, but why invent it just to
defend it where there is no evidence to disprove it, it's one of those
things that need the strongest kind of proof in their favour. I see no
such strong support.

> As to <kin>, the word has several distinct stems in the Armenian
> paradigm (NA kin, GDLAb knoj^(-e^), I. kna-w, pl. NA kanay-, pl. obl.
> kanan-).  The NA can be either from *gWe:nh2 or *gWen-ah2 or even
> *gWenh2-ah2, but on the basis of the instrumental (kina:- + -bhi), it
> seems likely that it's one of the latter two (*gWen(h2)ah2-).  The
> plural NA has the same stem as the Greek oblique gunaik-, which I
> prefer to reconstruct as *gWn.h2-a-ih2, with (thematic?) vowel
> inserted between -h2 and -ih2.  The plural oblique is an n-stem
> (analogical).
> If the singular stem (NA and I) is *gWéneh2, the oblique knoj^ is in
> my analysis comparable to the Sanskrit oblique stem of the compound
> a:-stems (i.e. *gWen-o-yeh2-os > *gWenoya:s > kinoy > knoj^ /
> *gWen-o-yeh2-i > *gWenoya:i > kinoy > knoj^).

I do not think there was such a thing as "compound a:-stems" in *-o-yeH2-.

> Which is why in my previous message I erroneously referred to *gwenh2-
> as an eh2-stem (it is only in some languages, e.g. Slavic).  The
> original paradigm must have been *gWén&2 ~ *gWé:nh2 (with "Szemerényi
> lengthening" due to *-Rh2; = Skt. ja(:)ni), G. *gWn.h2ás (= Skt.
> radical a:-stem gnás[*]).
> [*] Thus Macdonell (for já: "child"), but Pokornmy gives G. gná:s for
> "divine woman"?

I'm sure 'woman' *is* an eH2-stem, but one that ablauts, i.e. one not
containing the "thematic vowel": *gWén-&2(-s), *gWn-áH2-s,
*gWn.-H2-ó:m. The ablaut explains the OIr. paradigm ben, mná, ban, and
also the alternants of Vedic jáni-/gna:-, and Germanic *kweno:(n-),
*kuno:(n-), and even Toch.B s´ana, obl. s´ano. I do not know if the nom.
was sigmatic. An s-less nom. *gWén-H2 (prior to syllabification of the
laryngeal) should not lengthen the root vowel, for that rule worked before
the Schwundablaut, i.e. at a time when the form was *gWén-oH2 with a
(reduced, later-to-be-lost) vowel between the /n/ and the /H2/. The
lengthening rule which was revived (with some patently wrong, but
nevertheless influential details) by Szemerényi, works only after
consonants. I am not sure about the origin of Goth. qens and Skt. já:ni-,
but they look like vrddhi formations with an extra mora and i-stem
addition (i.e. old-style thematicization), but I fail to see a functional
justification. In the RV, -ja:ni- appears to be the compositional form
(second part) of jáni-, but I do not understand that either.