I don't think I replied to this message yet. Forgive me if I have.
Jens:
> (This damn thing went off while I was cleaning it. I resume my reply
> now. - JER.)
Hehe. I'm sure I've heard that line before somewhere. Was it "Three's
Company" or "Are you Being Served?", I wonder <:)
> But I haven't based anything on the assumption of a single vowel for
> (an undefined prestage of) PIE. If what else I say is wrong that must
> have other reasons.
Then what's with the obsession with monovocalism?
On the formation of thematic vowel:
> Well, the truth just is the oppposite of that. I do not know what may
> have led you to make this peculiar choice, but it is quite plain that
> the thematic vowel which may even be the *only* vowel of a word was
> originally accented.
Nominal declension doesn't show that the thematic vowel was accented.
It shows the opposite. If one can accept Acrostatic Regularization,
we must of course accept that the nominal thematic stem may no longer
retain the original accentuation. However, in many many cases, we see
a full vowel in the core of the stem. Sadly, we also see that
the thematic vowel is full. What we learn from this is that these
thematic nouns are a recent product because they fail to conform to
the ancient quantitative ablaut where zeroing is expected in one of
these syllables, not full-grade in both. Sometimes it's even weirder
and we have *wlkWos completely contradicting quantitative ablaut
outright with accent on the zeroed syllable and an unaccented but guna
thematic vowel. Thankfully, we can understand in this case that *wlkWos
is derived from an adjective where the vocalism would then be in line
with the accent.
Now if we have a stem like CVC-o- with fullgrade in both syllables, we
can either put accent on the initial syllable in the preform, or on
the thematic vowel. If we put it on the thematic vowel, it admittedly
conforms nicely with the observed adjectival accentuation. The only
problem is that adjectival accentuation is just as clearly regularized
as nominal thematic stems and its accentuation must be under suspicion
too.
The other option is that the accent is on the first syllable, the core
of the stem, and this has the added attraction that it acknowledges that
just by removing the unaccented thematic, we have its athematic
equivalent. Pretty simple stuff. If the thematic were accented, we'd
also have to explain the alternation of accent between the athematic and
thematic variants. This is unneeded and not even what we see anyways.
From *dye:us and *deiw-o-, are we going to claim that there was once
**deiw-ó-? This still doesn't explain *dyeu-! Why make this complicated?
Instead, it's more clear when we accept that *deiw-o- is fixed with
initial accent because of Acrostatic Regularization and *dyeu- shows
the older accent with a former but now desyllabified first syllable *dy-
and accented second syllable shown by the segment *-eu-, indicating
earlier MIE *dayéu- underlying both forms. Acrostatic Regularization
would cause a thematicized noun *diéwa-s to become *déiwa-s while
ignoring *die:u-s.
Furthermore, it's sensible that the etymological core of such stems
should contain accent, because it is afterall the semantic heart of the
word. So nominal *bHor-o-s should as a result be attributed to earlier
*bár-a-, not *bar-á- (more appropriate for an adjective).
The alternation of *e and *o by the thematic vowel is quite radical,
most efficiently explainable by lengthening before voiced segments as
found elsewhere in languages. The length differences then seed the
further phonemic differences as we see them. It's hard to believe that
this alternation could be brought about so regularly if the accent were
present on that syllable in some cases and not in others. It also seems
just as unlikely that these subtle length contrasts between the
protoforms of *e and *o could be possible in accented syllables that
already are well known to show length contrasts of *e/*e: and *o/*o:.
Either we may deduce that the thematic vowel is accented and then wrestle
with exceedingly rare triple-length contrasts as you do, Jens, or we may
come to our senses by siding with the inevitable, that the syllable of
the thematic vowel was unaccented and thus any length contrasts therein
went towards producing later *e and *o.
> Further down you're gloating about my acceptance of accent on
> the vowel marking the subjunctive (although that is not the best of the
> examples, far from it); now that is the thematic vowel.
Actually no. It's not. It's ultimately a suffix *-he-. Thus, what is
transcribed as *bHere:t may in fact be *bHer-e-he-t. By laryngeal loss
and concatenation, that still yields *e: here.
> For some reason you do not accept the stem-elaborating vowel of the
> subjunctive of reduplicated formations as a representative of the
> thematic vowel; perhaps because it means something and *is*
> underlyingly accented?
No, because you misunderstand this suffix and it actually represents
*-he- with or without the _real_ thematic vowel that precedes it.
> By you I meant you, sir. I cannot be content with departing from
> what you know.
Your loss.
> I was not referring to anything in Akkadian. But do you have to scrape
> for a specific Akkadian word to exemplify a varied vocalism? That is
> remarkable in itself.
It's remarkable when any language has no irregularities. So when Grande
says what he says, I don't feel obliged to listen when it violates
the tendency of natural languages. Why should Semitic be regular?
Akkadian isn't perfectly regular. Is any Semitic language truly so
perfect that there is _no_ variation whatsoever? You're saying without
shame that a theoretical construct like Proto-Semitic truly has no
deviation from expected morphological vowel alternations? Or are you
just too stubborn to admit that we probably don't know enough about
it to identify the irregularities that surely exist within it? That
sounds astonishing. Forgive me if I can't come to grips with that
unlikelihood.
> I do not have too much literature on Semitic at hand, but I read
> in B.M.Grande, Vvedenie v sravnitel'noe izuc^enie semitskix jazykov
> (Moskva 1972): "Thus vowels fulfil the function of grammatical elements
> in the words, their role in a word being auxiliary. The character and
> consistency of vowel sounds in the word are one [sic] of the basic
> morphological means in the Semitic languages" (103).
I know the role vowels play, but I doubt the degree of 'consistency'.
There must be exceptions, even if they aren't found yet. Even if it's
99% consistent, that is more believable than being 100% consistent.
> And later: "2.3.2. The general formula given under 2.3.1. and
> characterizing the Semitic root is actually completely valid only
> for Arabic and the Southern Peripheral Semitic languages. It is valid
> there for all verbal as well as for all nominal roots, ..." (64).
So by this quote, are you intending to stand behind it and assert to
us that Arabic or Southern Peripheral Semitic languages contain no
irregularities in vocalism, whether they be recent or historic? All
vowels in those languages serve SOLELY morphological purposes? I think
that's a very dangerous position to uphold, this is.
> I have no serious insight by which I could evaluate the correctness of
> these statements independently. Are they *very* wrong?
No. I'd say they are 'slightly' wrong. A matter of mistaking an almost
regular situation for an absolute state of affairs.
> Against what illusory rule is Akkadian illik a counterexample?
Against the rules by which Akkadian normally operates. It is an
'irregular' verb as all can see in any textbook, meaning that it isn't
the expected form based on most other verbs. Since Akkadian has this and
other irregular verbs, we can state comfortably that this language isn't
perfect and does indeed have some variation in vocalism in the verb stem,
beyond just the typical morphophonological alternations we normally see.
Examples like these would have us accept that there must have also been
such inescapable surprises in IE morphology. We know of these already
but it's a matter of simply surrendering ourselves to the imperfect
humanity of the proto-language we're reconstructing. We must allow
ourselves to accept what is attested that defy the overall pattern:
*o-grade presents, Narten verbs, stems with original *a-vocalism, etc.
> You are changing the subject again. I never claimed I did, you did.
I merely am offering a different viewpoint. If it should come across
that I "know" pre-IE, perhaps it's only because I find it often
pointless to state "I feel" or "I think" when what we are discussing
is evidently theoretical in nature anyway. It's implicitly understood
that we think and feel the way we do and that we are all grappling
in the dark.
>> Again, no language I know of "weakens" /e/ to /o/.
>
> Yacketi-yak:
Translation: "I can't bring myself to accept that gLeN has a point here.
Instead I will reassert silly things like 'It's a commonly used short way
of saying the unaccented form of what would otherwise have ended up as IE
/e/ is weakened to what in the end surfaces as IE /o/' in order to
neglect my obligation to logically defend my point of view."
Evidently if you can't explain the in-betweens of your theory, they are
underdevelopped and lacking. It's understandable then, that I reject
your views on the exact development of *e/*o alternation and stick
with the 'lengthening' idea.
You say "I thereby avoid specifying things I do not know." Then why
are you talking at all on this forum since you don't know pre-IE
anymore than I 'know' it? It's because you have a view to defend.
Defend it, man, or drop your case!
> What intuition, if that is what it is, is in a position to inform you
> that the present aspect and the perfect stem should be parallel?
If aorists and perfects are normally athematic and duratives are
thematic, there is an unintuitive rift between the two groups. It
just so happens that QAR and the other rules I've devised beforehand
lead me straight to differing _uniform_ vocalisms for durative on the one
hand in *e and aorists and perfects in *a on the other. For example,
we seem to have an MIE perfect *ba-bára-he 'I have carried' with uniform
use of *a throughout the reduplicated stem. This yields eLIE
*bHebHár-xa because the first *a is preserved by Paradigmatic Resistance,
the second *a is accented and is preserved, and finally the *e in the
suffix is reduced to *a [&]. All is normal. From here, it's pretty
straightforward and we get *bHebHor-xe after Schwa Merger and Vowel Shift.
BTW, Paradigmatic Resistance also explains how other instances of MIE
*a become strengthened *e, such as MIE *pát:-sa and pat:-ása, vocalically
uniform as the paradigm once was, becoming eLIE *pa:ds/*pedás >
*po:ds/*pedós. All these rules are supporting each other in unison,
indicating that my solution is too damned clever to be off-base.
These rules go further to predict the uniform *e-vocalism of reduplicated
duratives like *bHibHereti. I am led to MIE *be-bere-ta, yielding
after Syncope *bHabHérat in the non-indicative and a new indicative
*bHabHérati fashioned from locative demonstrative *-i. Now, here is
where unaccented *a becomes *i pretonically, yielding *bHibHérati.
Acrostatic Regularization shoves the accent back and Schwa Merger yields
*bHíbHereti as a result. Notice that the first unaccented *e here in MIE
normally survives syncope to become *a. Therefore there is no Paradigmatic
Resistance that is necessary in the perfect showing *a-vocalism. So this
explains why the reduplicated *e in the reduplicated durative becomes
eLIE *a (later *i) while reduplicated *a in the perfect becomes
strengthened *e.
Again, we only arrive at such a uniform vocalism in MIE if we accept
certain things that you remain stubborn about, such as the conclusion
that *i is the reflex of 'thematic vowel' when preceding the early Late
IE accent.
So as a result, this all indicates with a fair degree of certainty
that the durative had *e-vocalism throughout the stem and the perfect
had *a-vocalism throughout its stem and when reduplication occured,
the reduplication also reflected the proper aspectual vocalism. From
there, it stands to reason that the aorist probably distinguished itself
from the other two aspects by containing a different vocalism from
the other two while using durative or durative-like endings. In this
way, it remained unique and prevented it from disappearing right up to
IE times.
> Why is the present aspect patently parallel with the aorist in the
> selection of personal endings and so unparallel with the perfect?
In a nutshell, two reasons:
1) It nicely makes irregular things regular and uniform
2) My rules that work for other phenomena such as QAR,
eLIE phonotactics and Paradigmatic Resistance are
predicting the uniform vocalism as a describe for MIE.
>> Do I use "hundreds of names"? No. An exaggeration.
>
> But you should. There should be a new name each time there has been a
> change. So this is poor design.
You're now trying to find fault in silly things.
> But we can't have each opponent calling each generation's language by
> names that are not shared by the group of discussants.
There's no fear of that. I haven't seen anyone use any names at all
because they haven't bothered to figure anything out yet about pre-IE.
I still keep on seeing people, learned and amateur alike, expressing
their very vague observations of pre-IE but nothing has really been
seriously detailed by anyone from what I've seen. People seem more
concerned with learning what everybody else knows instead of getting
their hands dirty and doing a little discovery.
> This is destructive to the debate.
It's destructive to the debate if you don't think clearly about what
you're saying. I try to be as precise as possible about my views of
pre-IE so that I know exactly where I went wrong. If I never stand
firmly on _one_ specific point of view, I'll never be able to stand
still long enough to know what's right or wrong. It's this hyperactivity
that probably prevents most other people from developping detailed
thoughts on the exact processes of pre-IE.
>> Do you mean here that *woid-, being unreduplicated, is not "normative"
>> compared to other perfects?
>
> Yes.
Well, the lack of reduplication isn't based on the form of the word,
so what then is it supposed to be, in your view?
> No, it does matter, for the full picture of its forms shows very
> definitely that it has been reduplicated.
I can just as well reconstruct *wa-wait:a-he for the 1ps perfect and
would still yield *woid-xe but only if it can be explained why the
reduplication was lost here. I don't see any motivating factor.
>> Here we go again. It's "not of phonotactic origin", but what is the
>> _evidence_ in one short paragraph? These are still idle assertions.
>
> There is much more solid evidence showing exactly what you refuse to
> even observe than could possibly be contained in one short paragraph.
My philosophy is that if someone can't explain something concisely,
they don't know enough about what they're talking about, or
they just don't have any evidence. I know it's a strong view of yours
but I wish I could see your reason for it.
Concerning eLIE's intolerance of -CCC as I formulate it:
> There is no rule complying with this brand of common sense as you
> call it. Whoever issues such a statement must have knowledge of a
> different treatment of IE clustering across the boundary between
> root and flexive in primary word-formations in dependency of the
> number of consonants that are brought together.
Yes, exactly. The phonotactics of eLIE are not the same but in order
to arrive at this understanding, we need to understand the rules
that have been logically devised previous to it and their reasons. I've
described where *bHe:rst comes from, whose explanation is based
on Clipping (itself dependent on Szemerenyi Lengthening and Syncope)
to explain the vocalic length caused by aorist *-s- (from a nominal
ending *-es) and 3ps *-t (from *-ta, a demonstrative).
Other examples can also be explained by these and other rules that
have substantiated themselves independently.
> Whether a root ends in one, two or even three consonants, it is treated
> the same;
Yes, but just in the latest form of IE as it has come to be. We need to
discuss the individual cases of -CCC and all the rules I have listed
plus their proofs to understand this. Unfortunately, this particular
topic is a very complicated one. It's not like accentuation which is
certainly easier in comparison.
> Your terminology is not everybody's. Not that I care much, but for
> heaven's sake, let me be patient, so what is QAR?
I've discussed this many times before: The Quasi-penultimate Accentuation
Rule. It was formerly the Penultimate Accentuation Rule (PAR) but I
realized that phonotactics and some unexpected problems in accentuation
which arise with items like *woid-xe showed that penultimate accentuation,
while good, didn't explain the IE accent enough.
Instead, it turned out that a quasi-penultimate accent, one which allowed
the accent to either be on the second- or third-to-last syllable of a
word described the IE situation much more completely. The interesting
pattern allowed me to push the age of the animate nominative further
back to IndoTyrrhenian by allowing *-sa in MIE. Ironically, it also shows
that while the accent has two different positions for the word as whole,
the stem itself could ONLY have penultimate accent, similar to my previous
conclusion to begin with. So while the accentuation I describe sounds
like it's unpredictable with two possible accentual positions in the
word, it is in fact completely predictable.
So *kawána-sa (> *kwo:n) defies former PAR but is regular under QAR
and shows a fixed _penultimate_ accent on the stem *kawana-, the only
syllable that the accent can be found on by this rule. When a disyllabic
suffix steals the accent, as with the genitive, the accent has absolutely
no choice but to migrate to the penultimate syllable on the suffix. Thus
*kawan-ása (> *kunós).
Frankly, a robot couldn't have come up with a better rule :)
> The subjunctive of the perfect of *weid- turns up with irregular but
> identical forms in Indic, Greek and Celtic. That cannot be an
> innovation.
It was a feature that certainly existed in IE itself. I meant an
innovation in a late stage of _pre-IE_.
> Then that would reduce the number of individual a's.
Yes, Vowel Shift reduced the number of individual *a's but they
were not abolished completely, nor would they have been.
> So you ascribe the vocalism of the hi-conjugation to the lexical
> vocalism of the specific roots concerned? Could you explain and
> justify that? Is it credible that roots with a specific vocalism
> formed their finite forms with *inflections* of their own?
The *o-grade is not an inflection in all cases. That would be like
saying that because we have English "sing" and inflected "sang",
we must think of "can" as the past tense. Why are YOU assuming such
a funny thing?
> The normal vocalism observed in the root aorist is *e. This even
> comprises the aorist stems found in Hittite.
This would do just as well. If that's the case, we'd have to reconstruct
MIE *CeC-a- as the antecedent form of these stems, still with different
vocalism from duratives in *CeC-e-. The thematic *-a- in the aorist would
still be lost by way of Syncope.
>> Not that "plainly". Evidently your idea is dependent on the antiquity
>> of the aorist subjunctive which I don't accept. The subjunctive was
>> originally unspecific to aspect, I figure.
>
> There are retained present, aorist, and perfect subjunctives that agree
> with each other in all corners of the IE territory where there is a
> subjunctive at all.
Yes, I won't bother questioning that, but I question the _antiquity_ in
pre-IE of these forms as I was trying to explain. I've said that the
subjunctive is not a thematic vowel but derived from a special affix
*-he-. It would derive from a bare verb stem, plus or minus the thematic
vowel, and a postfixed ending. If we think about what the function of
subjunctives really are and how they are used, we notice that they are
often secondary 'imaginary-time' verbs in subordinance to the main
'real-time' verb. In other words, it suggests a relative clause.
It wouldn't be too farfetched to expect that *bHerehet (*bHere:t) is
nothing more than the reflex of eLIE *bera-he, originally unmarked for
person and perhaps with a more liberal usage as simply a verb for a
relative clause, affixed with postfixed *e to indicate 'then' or 'thus'.
This account would also explain why the marker *-he- follows a thematic
verb, which is kind of irregular in IE morphology -- It's only based on
the unmarked non-person-specific form of the verb. The verb is also
necessarily in the aorist here, to indicate a timeless or, in this case,
imaginary ('non-time') action.
Unfortunately, this idea works against the notion that subjunctives had
always marked aspect in the past. It would have us believe as a result
that aspect-marking came later to subjunctives.
= gLeN