Re: [tied] IE vowels: The sequel.

From: enlil@...
Message: 33312
Date: 2004-06-29

Jens:
> But is that more than a statistic thing? Does it really matter that
> much?

In my theory, that seemingly trivial matter matters very much. By
recognizing that the same vowel that underlies the thematic alternations
in *e/*o (namely, my unaccented *a in stages previous to Schwa Merger)
becomes *i in certain conditions is very important. The pesky *-i-
naturally doesn't occur in all accented positions or even majoritarily
in accented positions, but rather tends to be _unaccented_. When
accentuation of *i does occur, it can be explained by the existent
morphology in one sentence. The *i-reduplicated verbs, for example,
show the Caland variation of former *a before accent, but yet there's a
strong motivation for initial accentuation here because of the Acrostatic
Regularization that affected all thematic stems at the time.

There's nothing statistical here, nor any logical fact at all, to make us
feel that the _presence_ of accent has anything to do with this. It's the
absence of it that is one motivating factor. And yes, it's a statistical
thing because it's important to base our views on logical facts as much
as we can. I'm sure you agree. Statistics is a wonderful tool.

However, while the lack of accent explains thematic vowels in general,
it doesn't explain their variation. As I've observed, *i is normally
pretonic overall while *o is seen posttonically (and thus more common).
Of course, we can do nothing about exceptions to the rule (they will
always exist) but as long as they are the minority situation, we're
doing good.

My "pretonic" explanation of *i-alternations of the thematic vowel even
makes perfect sense phonetically. In anticipation of the accent, a
vowel will become tense and rise as we find here. When after accent, the
vowel will tend to "loosen up" and drop to [&] as I suggest. Phonemes
after the accent will tend to be less audibly clear afterall.


> Sorry, I'll leave that matter be in future when addressing you if
> you prefer it that way.

Well, you may call me crazy but I see the *o/*i variations clear as
day now (thanks), but *o/*u variations? Do we find **xargu-? Can it be
said to mean the same as *xargi- or *xargo-, let's say? It doesn't look
like a typical pattern like *o/*i is. And we don't even see
*u-reduplicated verbs.


>> Or... everything about *bHéreti is as we expect it to be.
>
> Sure, if we make enough correct predictions it is. I can explain the
> form as a regular Narten-present subjunctive, and I find it hard to
> see what else it could possibly be.

Why can't thematic verbs be what they are: verbs? It's the athematic
verbs, not thematic ones, which appear irregular and are even less
common. Why can't they just be what they are? Why does the thematic
vowel have to be explained here? You are trying to explain something
that isn't begging an answer. It's like asking why 'parking' starts
with 'p' instead of a clic sound. Why ask the question and why do
we need an answer... even then, how _can_ we answer that?

It looks to me that you take the term 'thematic vowel' to be a real
entity in itself when, as I see it, the term is nothing more than a
convenient name for a grammaticist to use when describing the kooky
morphological processes of our beloved protolanguage. If I'm right,
the thematic vowel is nothing but an illusion (etymologically speaking,
that is) being created out of very slight and practically unnoticeable
changes that fortuitously happened to coalesce into a pattern we may
now call 'thematicism'.

You believe that MIE perfect 1pp *wait:a-he for later *woidxe is a
pointless addition of thematic vowel where there isn't one but it does
iron out why it is that the aorist or perfect should have no thematic
vowel on a whole while the durative does. There is no obvious logical
reason for this. The thematic vowel, despite Miguel's assertions, doesn't
show any relevant semantic value and is far too widespread to have
possibly had such a value. So, it's most reasonable to presume that it
is something automatic and trivial. If so, the differences should be
explained and if the solution should be a trivial one, what is it?

Well, my trivial solution is that the fundamental reason for the lack
of thematic vowel in certain instances is simply the quality of vowel
(*a disappears, *e doesn't). The *a-vocalism happens to go securely
with the perfect and aorist conjugations in MIE while *e-vocalism
(which we later call 'thematic') shows up in the durative. The whole
reason for the _presence_ of this thematic vowel is just as trivial:
phonotactics which barred any syllables that were more complex than
CVC.

We can thus only explain *woidxe as < *wait:a-he and that not only
conforms to QAR and the phonotactics observed in my other examples but
also explains the strange lack of thematic vowel in these forms.
Nothing like killing ten birds with one stone, which is exactly what
a good theory should do.


> But I am severely embarrassed by the fact that non-Narten presents
> and aorists form their subjunctives in the very same way.

I'm now drowning in terms and I'm not sure what you're getting at.
Maybe this nagging headache today (which has nothing to do with you)
is making me feel blurry. The subjunctive of present *bHereti &
*esti is respectively *bHere:t and *eset, no? Are you saying that
you're embarrassed that Narten's add *-e- too? This may be a problem
on your end. In my mind, there's nothing odd at all. What am I missing?


> So I read, but of what? And if what is *sod-i-tó-s 'made sit down'
> the genitive? And *néw-o-s? And *bhér-o-m&1no-s 'being carried' and
> *dhugh-m.H1nó-s 'being utilized'?

This is what I'm saying. I've given the Mandarin example that shows
that this is a real possibility that can't be discounted.


> Except that we do not really see the language act this way. It may
> be pure make-believe.

As I've already outlined, it _isn't_ what we see the language (that is,
the latest stage of IE) behave like! I'm describing an older stage with
a necessarily different grammar. Sanskrit grammar is different from
IndoIranian grammar and French grammar is different from Latin grammar.
Why insist that all stages of preIE have the same grammar as IE?!

I've showed how there are hints in IE itself at the former grammar I'm
explaining for early Late IE such as the *s/*m gender opposition that
mirrors the thematic nominal opposition of *-os/*-om. This and so many
other things show how the adjectival declension is based on the
nominal declension... suggesting that adjectives were once undeclined
because we are left with nothing else.


> You are not honestly trying to defend the position that *all*
> languages are like Mandarin, are you?

Did I say I was? I'm saying *most* languages naturally associate
genitive constructs of some kind with adjectives. It's a universal
semantic association that we will find in many languages, of which
Mandarin is perfect example. Since IE is, as far as I'm concerned,
supposed to reflect a natural language that once existed, it should
conform to universals too.


> There is nothing a priori to indicate that it would be parallel to
> precisely that language,

No, there isn't. I only use Mandarin when I can because it is quite
different from English and it's a language that I'm familiar with.
I could just as well use an Anatolian language to show the exact
same pattern or Etruscan. Again, the genitive<=>adjective connection
is a universal semantic link. If your brain can process information,
it should be able to see how adjectives can be formed on genitives.

Furthermore, it seems that you expect all languages like IE to
function on biased structures existent in your language. Not every
language has a clear category called 'adjective' and while IE itself
seems to have had that category, I'm open-minded enough to see that
this is not necessarily the case for Mid IE or earlier.


> When I "was there and did that" I found that the genitive of the
> thematic class is not identical with the nominative: The nom.sg. in
> *-o-s cannot have the same sibilant as the genitive in *-es-yo which
> has thematic vowel /e/,

Right. I understood this. We've been over that. You know that I took
a different path from that point, theorizing that your **-z is merely
an already-existent phonetic allophone of *s. The differences you
see can and have been worked out using this line of thinking. Now,
this path I now go down is somewhere you haven't been. You need to
be there (by understanding my angle), in order to argue effectively
your case. With a phonetic [z] in final positions, genitives and
nominatives can indeed be interchangeable if analogy makes it so.

It would seem that since your arguement against my Nominative Misanalysis
depends on my accepting a phonemic *z, you must convincingly argue
(beyond just vowel alternation) why we should accept this. You don't
seem to understand how well an allophone can explain what you take for
granted as a new phoneme. Have you really "been _there_ done _that_"
or did you just go straight for **z without considering allophony?


>> Alright, then it seems the matter between us is simply the idea of
>> whether *i is the reflex of thematic *o pretonically or not. I've
>> been claiming it is so far.
>
> Do you really invest that much in statistics?

Yes, statistics is logical but it isn't the only logical tool. Take
what you will from that statement. It seems you want to assume the
opposite of what we observe here without any other facts being
considered. Here, I happen to use statistics as a basis, but at least
I'm basing it on something. What exactly is your basis here?


On plural *-es, Jens:
> My analysis at least gets the accent right; presence of a vowel
> would make the nom.pl. a weak case which it is not.

I disagree. The presence of vowel doesn't make it a weak case for the
very fact that we find a vowel in the plural! What we observe is in
this case not what you conclude and you change the facts to suit a
whim. We not only find 'weak' *-es in the nominative plural but we
also find another 'weak' *-ax in the feminine nominative. If you must
continue to assert the above, so be it, but it's not what we observe.

So your assumption is baseless.

We may observe no such correlation between weak case and presence of
initial suffix vowel. Ironically, QAR shows that it is an earlier
presence of a lost _final_ vowel that determines placement of accent
and 'presence of (accented) vowel' as we find it in weak cases. Thus we
have two counterexamples against you:

*-es < MIE *-es [pl]
*-ex < MIE *-he [collective inanimate]

As opposed to accented weak case markers:

*-ós < MIE *-ása [gen]
*-ód < MIE *-áta [abl/part]

These are almost all attested with their respective vowels where we
should find them via QAR in Tyrrhenian languages like Etruscan, btw:

Tyr MIE
*-er (Etr /-ar/) = *-es (*-es)
*-he (Etr /-(cH)va/) = *-he (*-ex/*-ix [fem]; *-x [inanim])
*-ase (Etr /-asa/) = *-ása (*-os [gen])

Again, what's your basis for this? It's clearly not IE as we know it.
There is no correlation you speak of.


> Even so, I would of course have liked the emergence of the vowel in
> *-es to be supported by parallel cases. That wish however has not
> been granted.

We are both plagued by this, despite our differences. It appears to be
a matter that can't be helped in this sole situation unless others have
a clever solution.


> Of course rounding can be added, many languages have done so.

Not when in a lesser position of all things while the accented position
_isn't_ [+round]. It defies the pattern. It works against the grain.
We'd also expect _stressed_ or _accented_ vowels to be especially
lengthened or given any other quality over _unstressed_ or _unaccented_
vowels. This all comes right back to simple phonetics where the latter
vowels will naturally be less audible and less likely to uphold
contrasts or gain new qualities over the former.


> You also have it emerge from an additional dose of quantity, not by
> influence from anything rounded.

Yes. You are speaking of *[&] > *e versus *[&.] > *o before voiced
segments. The important distinction here however is that lengthening
before voiced segments is a well attested phenomenon, even in English.
Rounding before phonemes that aren't necessarily rounded makes no
sense to me, nor do I recall you coughing up a list of examples of this
showing that this is not only attested but as common as the phenomonon
I support.


> I hardly need to say that this was not my first guess, but also took
> a while to get round to.

Alright, how did you 'get round to' it? Why is lengthening, as attested
as it is, inferior to your solution? We've been arguing this for many
many moons but I still don't know what the answer is.


> I see the vowel is there, to me that's as good a reason as any to
> think it has not been lost.

Naturally :) But the cause is the question. There are cases of vowels
disappearing and cases of vowels not disappearing. Looks like _two_
vowels at work, to me. One that disappears (*a) and one that doesn't
(*e > eLIE *a).


> No, I think it is up to you to communicate. It's about time.

:) If I didn't communicate, I wouldn't have sent so many posts to this
Forum and I wouldn't get into as much trouble as I do.


= gLeN