On Mon, 26 May 2003, fortuna11111 wrote:
> Okay Jens, back to my questions, since you wanted to answer :-)
>
> > > They are listed as separate phonemes.
> >
> > Yes; what I mean is that /d/ and /t/ were even opposed to each
> other in
> > word-final position in PIE, or at least in that prestage of it in
> which
> > the "thematic vowel" was split up into /e/ and /o/ depending on the
> > phonetic character of the following segment (being /o/ before
> [+voice],
> > /e/ elsewhere).
>
> Question: who says this and where? Or if it is your theory, what is
> it based on. I just have a problem with mere statements, I usually
> want to know where they come from. Nothing new to science, or to
> journalism, for that matter.
The short answer: I say this. I have presented it in a chapter of my book
Studien zur Morphophonemik der indogermanischen Grundsprache, Innsbruck
1989. A shorter statement is contained in my contribution to the VIII.
Fachtagung (Leiden): Beekes, Lubotsky, Weitenberg (eds.), Rekonstruktion
und Relative Chronologie, Akten der VIII. Fachtagung der Idg.Gs., publ.
Innsbruck 1992.
That was who and where. I have repeated it many times, but
these references will be the easiest ones to follow.
Now for why: The "thematic vowel" is a special subfield of IE ablaut. The
stem-final vowel of thematic verbs and of pronouns alternate in a way all
its own, being now /e/, now /o/, never zero like the other IE vowels. Its
alternation is independent of the accent: we have *-om, *-es, *-et, *-ome,
*-ete, *-ont no matter where the accent is (and no matter how the ablaut
grade of the rest of the word is), cf. e.g. the types *bher-e/o- with root
accent and the type in *-sk^e/o- which is accented on the suffix. It is
common knowledge that the decisive element in the selection of the two
phenotypes -e- and -o- is the phonetic character of the following segment;
that is not my discovery, Saussure and Hirt agreed on that (and little
else). But it seems to have been too much for the entire field to pin down
the decisive factor. Since we have *-om in the 1sg and the acc.sg., it is
clear that /m/ takes -o-; but 2sg has -es, and the nom.sg. has -os, so it
ceases to look promising. Still, all the rest looks fine, even finer than
anybody had apparently realized:
If we sift the material and discard analogical forms as irrelevant,
we may be left with a core of conclusive facts. The o-stem nouns have
apparently given up the alternation e/o, having almost only -o-, except
for the word-final vocative in *-e and the a-coloured fem./coll. in *-a-h2
(which apparently cannot be based on *-o-h2, but must reflect earlier
*-e-h2 as shown by Beekes in the article with the funny title H2O in Die
Sprache 18 from 1972). But the pronoun has retained much of the old
alternation and in this agrees very precisely with the verb. What we find
is this:
We have -e- in
*-e (vocative domine, imperative lege)
*-es(-) (gen. *tes-yo 'des', 2sg *bhere-s)
*-et(-) (3sg *bheret, mid. *-eto, 2pl *-ete)
*-eh1 (instr. *te-h1 in Goth /the:/, denominative stative *sene-h1-ye/o-)
*-eh2 > *-ah2 coll./fem. *new-ah2; also 1sg.mid. *bher-a-h2-i.
We have -o- in
*-o-e > *-o:(w) (du. *to:[w] 'those two')
*-o-m (acc. *to-m, 1sg *bhero-m, 1pl *bhero-me, 1pl.mid. *bhero-medh&2,
mid.ptc. *bhero-mh1no-)
*-o-nt (3pl *bhero-nt, act.ptc. *bhero-nt-s)
*-o-w- (1du *bhero-we, 1du.mid. *bhero-wedh&2)
*-o-y(-) (nom.pl. *to-y 'they', ntr.du. *to-yh1, opt. *bhero-yh1-t)
*-o-r (impersonal *bhero-r in OIr. -berar; local adv. *to-r in Goth.
/thar/ 'there')
*-o-d (ntr. *to-d 'that');
less certain is the PIE status of:
*-o-dh- (local adv. like Gk. pothi 'where, hothi 'there where')
*-o-bh- (local adv.: Hitt. kuwapi 'where')
In these two lists it is very easy to see that -e- is not followed by
anything voiced, while -o- constantly is. I guess this is so plain that it
cannot be fortuitous, but must be the active factor. Even so, there is a
little worm in the apple, viz. the nom.sg. in *-os. I do not take the easy
way out and shout analogy, for it is also the form of pronouns (rel. *yo-s
in Gk. ho-s, Phryg. io-s, Celtiberian io-s). Instead I take a closer look
at the two s-morphemes. It is well known that the sigmatic nominative have
a set of strange allomorphs, while the same is not seen in the 2sg of the
verb. Specifically, the lengthening seen in the animate nom.sg. (as Skt.
dyau-s) has no counterpart in the 2sg (2sg ipf. a-s'rno-s). I agree with
Szemerenyi in ascribing the lengthening effect to the nominative marker,
but I find no evidence of the same effect in the 2sg marker, although both
surface af /-s/. I can therefore assume that the two s's were once
different from each other. And since the nom. marker makes the thematic
vowel take the form -o- I assume that the nominate marker was a voiced *-z
at the time of the change in the thematic vowel, while that of the 2sg was
not voiced since it takes -e-. Simplicissimus just asks: if they were
different anyway, why not use that to our advantage? Phonetically I see no
reason to assume that the voiced pronunciation of the nominative marker
was still retained in PIE, so I suggest that the merger of *-s and *-z
into *-s should be placed in some prestage of the protolanguage we reach
by comparative reconstruction - but within the time span covered by some
of the operations of internal reconstruction we can make on the basis of
the already-reconstructed PIE.
This e/o rule which applies only to the thematic vowel (which is
descriptively simply a "vowel in stem-final position", since all other
stems end in original consonants) has been accorded little attention,
although divergent formulations are not wanting, every handbook contains
one that just does not fit the facts. I refrain from pronouncing my
opinion of the individual alternatives: the writers are my friends, and
I'd like to keep it that way.
On this list, my e/o rule has been accorded very wide acceptance. We
disagree a bit over the phonetic rationale; some say it's a quantum of
lengthening, and that a somewhat longer e became o; I am now more in
favour of a tonal interpretation, thereby approaching the old stand taken
by Hirt. It *is* at least an increase in sonority induced by the following
voiced segment by whatever exact avenue.
That the e/o rule works only for the thematic vowel is a problem of its
own. I'll leave it be for the moment.
>
> The two phonemes, however, *are* neutralized in
> > Indo-Iranian and Italic.
>
> Examples? In what way are they neutralized?
Word-final *-t has become -d in Italic (Osc. kum-bened 'convenit', Old
Latin esed 'esset'), and so had *-d. In Avestan both yield the special
dental (transcribed by underlined t, according to Gauthiot representing an
unexploded dental stop), in Old Persian they are both lost, and in
Sanskrit they both surface as /d/ before a voiced initial in the following
word, and as /t/ elsewhere.
> It is often forgotten, even in historical
> > grammars, that Sanskrit is incapable of showing whether a final
> stop was
> > voiced or voiceless, since the opposition is neutralized in *all*
> sandhi
> > positions.
>
> That should be so, assuming what you said above about voice/voiceless
> is true or at least a point of scientific agreement.
I think it's true, but it's not agreed upon. I object when ambiguous
evidence is being construed to disprove it. And that was what was going to
happen in case the ablative ending got accepted as "*-od".
[...]
> If *-d and *-t are opposed to each other
> > in PIE, there is a phonemic contrast even in that position.
>
> Okay, to make it clear, what do you mean by opposed? Do you mean
> exclusive? And *in what position* exclusive?
That both exist and are not confused.
>
> A form like
> > *kWod > PGmc. *hwat (Germ. was) would appear to reflect the old
> voicing,
> > if in a funny way.
>
> How does it reflect it? How can you draw a conclusion from just one
> example reflecting something? It could be so coincidentially. Don't
> get me wrong, but in the area of IE linguistics it is full of people
> who are simply being awfully creative. One has to be careful.
You do right in asking directly. Much nonsense has been handed down
because teachers did not address questions, and the next questions were
not even asked. Am I overly naive in seeing a difference in Germanic
between Goth. thata with a particle added to *tod as opposed to the reflex
of *-ti in the prs. <bairith>? Or between Skt. id-am 'it' as opposed to
prs. bharat-i, ipv. bharat-u? I take this to indicate that *-d and *-t had
not merged. I could be wrong about that, but not so easily about the point
I really need to be right, viz. that there was a contrast between *-d and
*-t at some time in the prehistory of PIE.
[... On the tematic ablative]
> > It apparently also matches the gen.sg. of o-stems in Balto-Slavic,
> > Lith. vil~ko, OCS vlUka.
>
> I am not sure about Ablative in OCS. I will look it up. I suspect
> it is just another borrowed genitive ending, as in Sanskrit.
You are potentially up against quite an army here. The BSl. gen. also
covers the syntactic functions of the ablative. If it is an old genitive,
there is *no* basis for it that has even been suggested as far as I am
aware.
>
> Recently the Celtiberian form has been found to
> > be -uz (perhaps with a fricative d at the end).
>
> Who and where?
By Francisco Villar. The quick reference is an Innsbrucker Vortrag (the
thin white papers). As always, it's in Meier-Bruegger's bibliography:
F.V.: A new interpretation of Celtiberian grammar. Innsbruck 1995.
>
> It may aso be identical
> > with the Greek adverb type seen in kalo:~s 'beautifully' (except
> for the
> > -s which may have been taken over from other stem-classes which
> used the
> > genitive form also as ablative).
>
> You can have an -os in all possible other cases. As with the other
> stems in Sanskrit, which have an ablative coinciding with genitive
> (which was -os/-es in PIE). You take a single word in Greek and draw
> theories from it? I just don't get it, sorry.
The ablative is being used as an adverb in many cases. Common opinion is
divided on the question whether the Gk. adverbs in -o:s reflect old
instrumental or old ablatives. Since the instr. has acute tone in
Lithuanian, the circumflex of the endstressed adverbs can be derived
directly from the form of the ablative. In both cases, the IE form in
*-oH1 or *-oat (vel sim.) need the addition of an *-s to produce the Greek
form. If it matters where the *-s comes from I find it easier to assume
that it comes from another ablative form. And all other stem classes had
ablatives ending in *-s.
>
> And another important point is, I am usually very busy, while the
> volume of the list is challenging. So I do not always have time to
> go in depth with questions and comments, etc. In such cases I just
> say: le'me read it somewhere to save us both the time. If I don't
> agree in the end, I will tell ya. That was the portion of
> Hochenglisch for today :-)
We're in the same boat here. Thank you for your time.
Jens