IE vowels: The sequel.

From: Jens Elmegaard Rasmussen
Message: 33298
Date: 2004-06-26

--- In cybalist@yahoogroups.com, enlil@... wrote:

> Jens:
> > However, in very archaic remains, unaccented thematic vowels are
> > reduced to /i/,

> A convenient assumption that isn't a posteriori in any way
> whatsoever.

Oh, it is indeed based on very clear evidence which is even quite
well known. I'll come back to that.

> Instances of thematic *-i- are as commonly unaccented as *-o-.
Where
> we find oscillation between the two, accent differences are simply
> not attested without lurking variables.

When thematic noun stems appear as i-stems in the second part of
compounds the word is a bahuvrihi and so accented on the first part
(Vedic gandhá- => dhu:má-gandhi-). The interrogative pronoun stem
forms Ved. kás'-cit with unreduced thematic vowel under the accent
and -i- in enclitic position, thereby matching the behaviour of the
indefinite which is enclitic and has -i- in Gk. tis (unaccented) and
Lat. si:-quis. The pronominal forms *im, *id are only known as
enclitics. The passive participle in *-tó- is opposed to
substantival forms in *-tu- and *-ti-; *-tu- is used in simplicia
with root accent, *-ti- in composition, both are unaccented. There
are other indications, which I can tell you about if you want me to.
These are the clear cases of surviving archaisms where we have both
forms in opposition to each other and can still show how they were
distributed according to a very clear accentual principle. It even
makes typological sense: the vowel get smaller the more reduced they
are.

End-stressed i-stems are not very common. Greek has none at all.
Vedic has a number, but some are plainly secondary: rayís, ra:yás
from *reH1-i-s, *reH1-y-ós must have had accent on the full-grade
root somewhere, and that can only be in the strong cases, so this
was a root-accented i-stem, so *i is unaccented. Ved. kaví- 'sage''
corresponds to Avest. kauua:, kauuae:m from IIr. **kava:, kavayam
with a suffix *-éy- in the strong cases, so Vedic -i- is the reduced
product of the weak cases.

> For one thing, it's established
> that the shift of accent is partly dependent on the morphological
> differences of noun and adjective. Yes, we see that adjectives have
> accented thematic (regardless of *o or *i, btw) but we also find
them
> unaccented in nouns too. Which one is the deciding factor? Well,
while
> the accent of nominative adjectives are simply fixed to the last
> syllable, nouns enjoy a wide array of accent patterns.

Yes, sort of, it's the way I have just shown.

> As we all know,
> the acrostatic paradigm is the most regular of all, with the others
> showing the older state of affairs (accent alternation).

The acrostatic paradigms also reflect accent alternation, namely in
the vowel gradation of the root.

> We also see
> the same acrostatic pattern of thematic nouns in thematic _verbs_
like
> *bHer-e-ti.

This is the stem of a subjunctive with a radical accent which may be
troubling to all. I do not know why the subjunctive type of Vedic
áyati, ásati, hánati, dehat, dohat, yunájat, kr.n.ávat, kárati,
gámat, bhédati have accent on the root and an unaccented thematic
vowel. The effect of the thematic vowel of the subjunctives of
reduplicated verbs is plainly that of an earlier accented vowel, cf.
dádhati, intens. jánghanat, cékitat with the same accent and ablaut
behaviour as a weak form like 3pl.act. dádhati, várvr.tati.

The structure *CéC-e-ti did arise with subjunctives from acrostatic
verbs, cf. stávat (prs. stem /sta:v-/), váks.ati (s-aor. /va:ks.-/),
where original accentuation of the added thematic vowel would indeed
produce the result found, but what is the structure doing in the
subjunctive of root verbs with normal ablaut e/zero? It could be a
simple act of analogy, but it does not look satisfactory. I think we
have to humbly await new discoveries or better ideas concerning this
point. We have evidence enough to believe that *bher- formed
lengthened-grade (Naerten) verbal forms, as the Tocharian ipf. of
pär- 'carry' which is based on *bhe:r-; perhaps *bhér-e-t(i) is
based on the acrostatic structure and has set the model for the type
as such.

> It's clear then that the acrostatic pattern covers thematic stems
of all
> word types except adjectives and that the adjective is the one
that must
> hold any trace of the original accent in the thematic
noun/adjectives
> pairs.

Yes, the pure thematic forms were adjectival in origin, and they
generally show accent on the suffix. That, however, has nothing to
do with suffixal formations containing suffixes of the structure *-
Ce/o- which have many other functions and are almost invariably
accented on the thematic vowel. There is no way one can derive the
accented thematic vowel of *pr.k^-sk^é-ti 'asks' from an genitive
morpheme, nor the accented *-H1-yé/ó- of the passive present, as
Ved. su:yáte 'is being pressed' (su-). It could be attempted for
formations in *-ró-, *-nó-, *-tó- which generally *are* adjectives,
and little would have to be altered if it is correct, except that
the pronominal genitive ii *-esyo looks unfavourable to it in my
eyes.

> As I've said, the fact that the them.nom. adjective looks just
> like a genitive is not to be ignored. Anatolian confuses the two,
it would
> seem, as do many other languages.

There are very few thematic adjectives in Hittite, so they did not
utilize the identity for much. You are shooting with blanks again.

> Even in Mandarin, adjectives are formed
> in the exact same way as genitives without speakers being in any
way
> confused (eg: xiao-de haizi "the small child", Gelan-de shu "Glen's
> book").

I do not doubt that this is correct.

Hence, the genitival origin of adjectives is painfully clear.

No, this is painfully incorrect.

> The eLIE confusion between genitive *-ás and an already existent
thematic
> nominative *-a-s is a natural mistake that serves to also explain
the
> resulting adjective-noun case agreement as well as the
supposed 'accent
> on thematic vowel'.

For the sake of the argument, of what was *H2ug-ró-s 'strong' (Ved.
urgá-) the genitive? Of what was the ptc. *bhr.-tó-s 'carried' the
genitive? Of what was the ptc. of the causative *sod-i-tó-s 'made to
sit down' the genitive?

> But there's more. In fact, the occurence of *-o-m in
> the neuter adjective, with a clear relationship to inanimate
collectives
> in *-om, which in turn are based on the plural animate genitive,
> emphatically terminates any counterarguement.

Only in the sense that you are changing the subject and whoevber
cares to read your post loses the thread; *-om is not collective.
That the message is emphatically worded gives it no special status
of a positive nature.

> All the evidence, including
> a universal association between adjectives and genitives in
countless
> languages from Etruscan to Mandarin, emphatically shows that IE
adjectives
> were previously genitive constructs in IE, thereby explaining the
special
> adjectival accent in the first place.

No, it means it could be so. Many other languages have it
differently, what if *they* are parallels? Then it is not so. Blanks
again.

> So ironically this just plays into my hands and supports all that
I'm
> saying. Once there was confusion between genitive-derived
adjectives
> and thematic nouns, a new morphologically-driven shift of accent
> occured between finally-accented adjectives and initially-accented
nouns.

There was indeed an interplay of oxytone adjectives and barytone
substantives. It quite probably had indeed come about by the
formation of adjectives which had added an accented thematic vowel;
the system tómos : tomós may well be a reflexion of this. It is
connected with very grave difficulties however to derive the
thematic vowel itself by simple resegmentation of genitives. It is
what we have all tried to do, but perhaps I am the only one who has
really been there and done that and lived to tell about the
frustrations. I can do that in greater detail if you care.

> This was brought about by plain ol' common sense analogy by the
speakers
> of IE in general.

As I said, that's what I once thought too. I got wiser.

>
> You claim that *i is the accented variant of *o but the evidence
doesn't
> show this at all. Unsubstantiated. The data simply shows that *i
is a
> variant of *o, either accented or unaccented in both cases. The
nature
> of the adjectival accent can only be understood by understanding
its
> relationship to genitives.

Typo? I claim that *-i- is, in a series of archaic remains of a
morphological variation later given up, the *unaccented* counterpart
of the thematic vowel. The facts are all in Wackernagel's and even
Brugmann's handbooks; Benveniste added many observations saying the
same thing.

> >> To have the plural *-es be properly etymologized and connected
to
> >> other Steppe languages, we need a contrast of at least *e and *a
> >> in all stages of pre-IE.
> >
> > No, an opposition between vowel and zero will do.
>
> You must be joking.

No, I consider the -e- of the nom.pl. in *-es an anaptyctic vowel.

> It is in fact the zero-grade that is testimony
> to a previous alternation of accented *e with a reduced unaccented
*a
> in MIE.

Indeed, in so far as this means that *o was an intermediate stage between
*e and zero. We see this when unaccented *e is lengthened and then
survives as */o:/.

> That unaccented vowel drops causing the odd pattern. Again,
> another language universal where unaccented vowels tend to drop
off.
> Most people working on pre-IE or Nostratic understand the need for
> Syncope already. I guess you can't be helped.

I did not need this help. We all agree that unaccented short vowels are
lost in most positions. I do not think they were lost in stem-final
position where we find the thematic vowel instead, but I have yet to
consider several options as to the cause of this.


> > Indeed, it is a prestage of PIE. It is beyond the method to tell
us
> > how much older that prestage is, so you may call it what you
like.
>
> Actually, it's not beyond our methods at all. I would hope that any
> respected comparative linguist pays attention to chronological
order.


My account is one of relative chronological through and through, if only
in points where there is evidence to decide on that. I have as many stages
as I have changes (plus one).

> The names are simply a mnemonic to remember this order for me and
for
> any others who wish to understand what I'm saying. By nailing
myself
> to a strict chronological order, it allows others to cross-examine
> the current form of my theories and identify any paradoxes which
will
> then help me evolve better versions in the future.


In this we work alike. However your odd names of stages have had the
effect of a shield behind which you have been hiding. Nobody else (I would
believe) knows what your labels mean, at least I do not know it, and I do
not really care, for these are stages that should be adjusted as you go
along. And you are not at the end of your calculations yet. We can operate
with "before" and "after" of a given change relative to other changes,
"feeding order" and "bleeding order" of generative linguistics which is
what this is in essence.

> Since it is above you to pay attention to this important detail,
you
> don't allow anyone to question you and you fall back on your mantra
> of "This is beyond our understanding".

Many things are, given the lacunae in the evidence. But based on what we
have, I do operate with ordered rules. The rules you have been discussing
on this list over the past few years were, to the extent I have found them
sensible, originallly published by me in a long article in 1978 that
presented them in chronolocal order. The article was recently reprinted
with a few slight updates.

Like a psychological
resistance
> against admitting to being incorrect as if being incorrect were
somehow
> a reflection of your own worth, which it of course isn't, but I
guess
> people with PhD's like to think that way. My god, if it were, with
all
> the mistakes I've accomplished over the years, I'd be completely
> worthless, hehe. You sir, are an intriguing case study. []


Well, thank you. But as before you do not at all seem to care to check who
you are really faced with: neither do I. I do not think that is relevant,
but it becomes annoying when you choose to delve so much in it and just
blasts out all of your psychological projections. Let's drop that subject.

Jens