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

On Mon, 22 Jul 2002, Miguel Carrasquer wrote:

> On Sun, 21 Jul 2002 23:55:56 +0200 (MET DST), Jens Elmegaard Rasmussen
> <jer@...> wrote:
>
> >> As to the Luwian and Lycian i-Motion, I raise you the Hittite i-stem
> >> adjectives (oblique in -aya- = Sanskrit -a:ya:- = PIE *-oyah2- ?).
> >
> >In Hittite intervocalic -y- is lost, so gen. -ayas (beside -as) is
> >analogical on -awas of adjectival u-stems.
>
> I'm not familiar with a soundlaw by wich intervocalic -y- is always
> lost in Hittite.

You can read about it in Melchert's Anatolian Historical Phonology p. 130
or Kimball's Hittite Historical Phonology p. 364, but the insight is much
older.

> There is no loss in the verb iyami "I do", for
> instance, or in the suffix -(y)a "and", which is -a after consonants,
> but -ya after vowels(!).

Iyami has glide insertion after a high vowel; it occurs even after e in
ne(y)anzi, udne(y)as. Melchert shouts in boldface (p. 167) that "Hitt.
postvocalic -ya *cannot* represent PIE *-yo"; he posits *-H2o, assuming
special rules for enclitics.

> I don't think the occasional contraction of
> -aya- to -a- (or -iya- to -i-, -uwi- > -u-, -uwa- > -u-) had reached
> the status of a soundlaw in Hittite.  Of course, sporadic glide
> deletion of the very same kind can account for the loss of -y- in the
> oblique forms of the a:-stems outside of Indo-Iranian (*-oya:- >
> *-oa:- > *-a:-), especially in such a common paradigmatic ending as
> the feminine oblique (cf. the colloquial rendering of the past tense
> as -aa in both Polish -al/a and Catalan -ava).
>
> >I do not know why the suffix
> >vowel is -a- and not a more direct reflex of *-e-. I expect *-oy-es,
> >*-ow-es in the nom.pl., but I do not seem to find that in languages that
> >could show the difference, such as Ved. -ayas, -avas; Gk. -ees ; Goth.
> >-eis, -jus; Lith. -ys, OCS -Ije; Toch.B -'i is now generally derived
> from
> >*-ewes too. Thus *-ey-/*-ew- appears to have been generalized, and so I
> >would not like to explain -ay-/-aw- by analogy. I would rather think of
> >*-ew-s going to *-o:s via a stage *-ows which may have been normalized
> to
> >*-owos > -awas (all Hitt. gen.sg. forms end in -as) and subsequently
> used
> >as a model for -ayas.
>
> In the Nom.pl. I expect the same as in the Voc.sg., i.e. -ei-es and
> -ou-es ~ -eu-es.  The -o- in the u-stems is not a "real" *o (Skt. Dat.
> -ave: not *-a:ve:, Nom.pl. *-avas, not *-a:vas) but an Umlaut of *e
> (as in *gWou- "cow", Skt. oblique gavV, not *ga:vV, not *javV), a
> phonetic development which appears to be restricted to u-stems only.

I have the same problem in the vocative then. Ablaut rules working
elsewhere would assign *-ey to a PD pattern, just as voc. *p&'2ter goes
with *p&2té:r, *p&2trós. I would therefore expect voc. either *-i or *-oy,
but I find *-ey. And in the nom.pl. I expect *-oy-es, *-ow-es, but find
*-ey-es, *-ew-es. I would assume that the expected -oy-/-ow- has just been
replaced by -ey-/-ew- by simple levelling in pre-PIE times already. The
expect -oy- is retained in the amphikinetic type alternating *-o:(y),
*-oy-m., gen. *-y-os, nom.pl. *-oy-es, but that is another matter.

Are you seriously assuming that the prestage of Indo-Iranian in which
Brugmann's Law operated could distinguish an -o- that was an
umlaut-product of *-e- from an original -o- and only lengthened the
latter? What happened to that opposition in all the other languages?

> The Hittite i-stems with oblique forms in -aya- (-a-), mainly
> adjectives, but some nouns also show the pattern, are in my opinion a
> compromise between i-stems with oblique -éis etc. (> Hitt. -iyas) and
> a:-stems with obl. -oya:s etc. (> Hitt. -ayas), with additional
> support from the ay-stems (*oi-stems) with Hitt. nom. in -ais, obl.
> -iyas or -ayas.
>
> >[...]
> >
> >> Your original statement was:
> >>
> >> >The Greek aorist épion is of the type élipon, i.e. based on the
> >> >zero-grade with thematic inflection, in this case *pH3i-e/o-. The
> >> >structural basis of this stem type was no doubt the old 3sg middle of
> >> the
> >> >root-aorist, utilized as the stem of a thematic inflection which was
> >> >subsequently made active, i.e. 3sg mid. "*pH3i-é" -> 3sg act.
> >> >"*pH3i-é-t". However, with this specific verb, it is probably better
> to
> >> >depart from the 3pl root aor. *pH3i-ént assuming this structure to
> have
> >> >been transferred to the *wid-é-t/*wid-ó-nt model.
> >>
> >> I interpreted that as saying that the thematic vowel in épion is not
> >> original, but analogical (either after 3sg. middle *-e, or 3pl.
> >> *-e(nt)). That nicely solved a problem of mine, as I would have
> >> expected a thematic vowel to have been deleted if it was original (and
> >> if there was no early laryngeal metathesis in these cases, i.e. epion
> >> < *(h1e)-pih3-o-m [likewise *siH-e-, diH-e-, skiH-e-], in which case
> >> the problem also disappears).
> >
> >There is not a whole lot of evidence I could produce, but at least Hitt.
> >ishiyanzi 'they bind' indicates that there was no such metathesis. What
> IS
> >that rule based on? Why doesn't everybody posit *sH2i-e-, *dHi-e-,
> >*skHi-e-? Laryngeal metathesis does occur, but all examples I know are
> >anteconsonantal (or plainly analogical). I cannot see it matters whether
> >the following vowel is the thematic vowel or not.
>
> I just mentioned the metathesis as a possibility, without expressing
> an opinion.  Whether the following vowel is thematic or not may matter
> if there was a consonantic something about the thematic vowel (I don't
> remember the details now [was it /?e/, with a kind of glottal stop?],
> but I think it was _you_ who suggested that somewhere).

I suggested an analysis of the thematic vowel as something like /e?/ in
order to make the suffix consonant-final, because all other suffixes are.
Still, since there is only one suffix vowel in Indo-European, viz. /e/ and
what comes from it, the special behaviour of the "thematic vowel" could
just as well be ascribed to the special *position* of a suffix-final
vowel, in which case an extra consonant would be unnecessary. The rule is
then that, when a suffix does end in a vowel, that vowel is always the
thematic vowel.
My investigation of the long-diphthong roots showed the very same
behaviour of the root alternants before the thematic vowel as before any
other vowel. In Skt. dhá:y-as- 'Labung' the following vowel is not the
thematic vowel, nor is it in pa:y-ú- 'shepherd', while I'd say it is in
sphá:y-a-te 'thrives', and I don't know for certain in the caus.
pa:y-áya-ti 'makes drink'. I now find it notable that there are hardly any
examples of the *bhoros/*bhoraH2 types, so I do not really know what would
have come out of *peH3y- if it were to form that type; still, I do have
Slavic krajI 'edge, region, land' looking very much like *kroH1y-o-s
'distinction'.

Jens