Re: [tied] IE genitive

On Fri, 9 May 2003, Glen Gordon wrote:

>
> Jens:
> >If the underlying base denoted something to which the wolf belongs, then
> >of course the wolf be "of (that thing)", and the genitive expressing this
> >would be open to reanalysis as an adjective characterizing "the one of
> >(that thing)", and in substantivized use it could simply come to mean
> >that. By having zero-grade the word wolf is revealed as older than the
> >ablaut proper, while the substantivization, by accenting the zero-grade,
> >is revealed to be younger than the ablaut.
>
> Disagree. I assume that you're using "ablaut" to mean "loss of unstressed
> vowels", which shouldn't be confused. The former is a long-standing
> morphological process, the latter a single and abrupt event marking a
> severe change in accent, pronunciation and syllabics which provoked
> morphological change.

I'm not confusing anybody - see how well you understood it.

> The accentuation shows that the _accent_ is unoriginal, not that
> the entire form is "younger than the ablaut". The zerograding does not
> reveal that it is "older than the ablaut" -- It merely shows us that the
> word
> is from a time when quantitative ablaut and the position of accent once
> correlated, which could be long after the loss of unstressed vowels.
>
> To summarize:
> 1 - loss of unstressed vowels
> 2 - quantitative ablaut and accent correlate
> 3 - accent changes occur (acrostatic reg. among others)
> 4 - quantitative ablaut and accent no longer correlate
>
> So, *wlkWos shows us that the accent change occured at stage 3,
> whereas the form as a whole shows us that it dates to stage 2 or
> earlier.

THAT'S WHAT I SAID. Sometimes you must accept that we agree, however
much it hurts.

> This is how we should properly proceed with the etymology of this word.
> Given the grading, the accent is expected to be final. So we begin by
> postulating that this was once the case, but by doing so, it now appears
> strongly that this word was originally an adjective -- an intriguing but
> efficient conclusion.
>
> Due to the zero-grading, the adjective itself has been formed while ablaut
> was in full operation,

No, before - so that it was present in the language when unaccented vowel
loss occurred.

> not necessarily during the abrupt event of vowel
> loss, because ablaut had obviously become a morphological process
> occuring long after the vowel loss. We don't know whether *wlkWos
> predates vowel loss by simply looking at the word.

If it postdated any productive period of vowel loss there would be no
vowel loss in it. That can indeed be seen by looking at the word.

>
> We musn't assume too boldly and by doing this, Jens, you've skipped
> a logical step. At any rate, I think we both agree about the adjectival
> origin of this word.

Not necessarily. I can only see it was once endstressed, i can't see for
what reason it was that. It may be by a switch from one part of speech to
another, and substantive-to-adjective is a fair guess, but
collective-to-singular could be another, and perhaps others. The grammar
of the language stage we're talking about is not fully known.

>
> >I'd say, great enough to be taken seriously. Still, as I have made
> >explicit, there are other reasons to have doubts. The vowel difference
> >between the pronominal genitive *-e-s-yo and the nom. in *-o-s points
> >to an older difference between the two sibilants involved,
>
> This is another leap of logic. We observe that the thematic vowel
> alternates based on voicing. Therefore, the thematic vowel was once
> a single vowel. We observe that the likeliest solution is to propose that
> this single vowel has been lengthened (seen readily in many languages),
> not rounded or derounded (which is NOT seen in many languages), by
> the following voiced segment. Thus, we must conclude that the
> unexpected *o before the nominative has been secondarily lengthened.

You can't know that. We are not dealing with many languages, but with a
single one - this one.

> However this is acceptable because the nominative lengthens vowels
> elsewhere in athematic declensions. But then, of course, we must
> wonder what has caused the lengthening.
>
> Well, it is assuredly unacceptable to propose that the most commonly
> used suffix in IE employs the most uncommon phoneme in IE.

You don't know how frequent the voiced sibilant was before it coalesced
with one or more other phonemes into the single PIE /s/. What if it was
lost in some environments - how frequent is zero? What is statistics doing
here? This is just silly. Even if the statistics would turn out to be
right, they would not be decisive. A funny case is the Eskimo ergative
which ends in *-m; that is the only case ending of the language (others
are based on it), and apart from the ergative, a word just cannot end in
*-m in Eskimo.

> That's
> why the theory of a nominative in **-z(zzz) is ridiculously assumptive,
> opposing linguistic common sense, and therefore not worthy of further
> consideration.

Then don't. You aren't showing much.

> The theory is an example of a simplicity of solution that
> becomes too absurd to be truely logical and efficient.

There is nothing absurd about tonal lowering in a tone language,
especially if tonal lowering is already known to operate this way. What
appears absurd would rather be a belligerent attitude to ignore it at all
costs.

>
> The real and less problematic solution is "clipping".

You don't know that. You only clip off what you have glued on yourself.

> Since it's logical to
> presume that the origin of the nominative *-s relates to the commonly
> used demonstrative *so-,

Nor do you know that.

> unproblematic in terms of phonetics and
> function,

An *s- alternating with *t- and followed by a thematic vowel to form a
demonstrative stem is far from unproblematic as identical with a single
final nominative-marking *-s showing the effect of voicing.

> we may observe that the terminating vowel of a nominative
> in original *-s& must have been lost.

This is at variance with the status of the nom.sg. as a strong case. Or do
we have to add fanciful vowels to *all* reconstructions? Did no
inflectional endings end in consonants?

> However, because of lack of
> accent, the loss must have occured after the major event of losing
> unstressed vowels.

That statement directly says the opposite. Supposing (for the sake of the
argument) that the vowel was once there and was unstressed, and that
unstressed vowels were lost, then it looks fine that it is not present
anymore. There are other things that are far from fine, but not on this
particular spot.

> (I've already justified my penultimate accent rule
> time and time again, therefore I can employ it here.)

You have mentioned the terrible rule and appealed to it over and over
again, but never justified it.

>
> We all know that loss of a vowel can cause compensatory lengthening,
> one of the most simple and everyday linguistic processes we can use.

Yes, - *can* cause.

> Thus, we can now trivially unify thematic *-o- and athematic nominative
> lengthening together as a single process of lengthening by the
> nominative suffix, as well as explain the lengthening of the nominative
> in the first place via compensatory lengthening.

We could if we had rules of that kind, but we don't.

>
> Jens, will now resist the above logic as usual so I will restress that Jens
> uses the most _unusual_ linguistic processes to explain IE: the most
> common case suffix with a most uncommon phoneme imaginable, a
> strange process of vowel rounding by voicing (??), and the feably
> justified abuse of a distinct phoneme **z, which is otherwise only
> an allophone of *s in the IE we all know of.

Yes, I resist the above "logic" because it is unsupported by the evidence
of the language we are looking at (PIE). Tone-governed change of (the
prestage of) /e/ to (the prestage of) /o/ is a fact of that language as
seen by Hirt and Gu"ntert. Anyone who cares should read the short article
on Tone Languages in the Encyclopedia of Language and Linguistics. Then
they will see how easy this really is.

> (Please don't forget our
> discussion of the origin of vowel length where Jens again reconstructed
> an equally unsual three-way length contrast for pre-IE which is too
> rare to be considered.)

There *are* three degrees of length in PIE; that is not jamais-vu. Adding
a semi-long /&./ as input to /o/ would make it that.

>
> In comparison, I use compensatory lengthening, regular IE phonemes,
> and lengthening by voice, all of which are trivial things that require no
> further assumptions. Jens says I "misapply" Occam's Razor. Last time
> I checked, Occam's Razor was about efficiency of solution and avoiding
> the multiplication of hypotheses. I've demonstrated more than
> adequately that my solutions conform to this principle.

The opposite has been demonstrated ad nauseam: Accented and unaccented
thematic vowels are not both permitted, yet are both found; verbal and
nominal thematic vowels are not given the the same origin, yet alternate
the same. The constant position of the thematic vowel in stem-final
position is accorded no role in the account. In contrast to this, it it
extremely simple to just say that the rules governing the thematic vowel
alternation operate on stem-final vowels, while the other rules operate on
the other vowels.

>
> That rare processes are being used by Jens shows us all that he is
> multiplying hypotheses exponentially, forcing us to swallow _rare_
> processes that automatically require assumptions that IE must have
> used these rare processes. It's easier to assume that IE was just a
> normal language.

IE was a normal language except where it causes problems; problems are
what we talk about. I haven't found occasion to say that IE was oral.

Jens