Me:
>> Now we go over this new theory with a fine-toothed comb
>> to make sure it's not deficient.
Jens:
> That won't be enough, it needs something good in its favour too.
The "good" in its favour is we keep a sensible etymology of
nominative *-s from *so (an obvious one at that) and we don't
need to reconstruct *z's where there are clearly none in IE
itself. You feel the need to do this and that is your weakness
from what I can assess.
> The dreamt-up *-s (< *-z) gives a sensible morphology to items
> which otherwise have none. It also offers a motivation for the vowel
> timbre of the incomprehensible final *-o of both forms.
No, it doesn't. Rather, an allophonic [z] does. I can understand
why this could be confusing but it all can be explained with
allophony alone.
You see, the relationship of *so and *tod to animate nominative *-s
and inanimate nominoaccusative *-d in pronominal stems shows
that *t must have been voiced to *d unless analogy with alternative
forms with medial *t operated against the voicing. So in other words,
the relationship shows that voicing of all final phonemes occured
some time in Late IE. (This finally provides the basis for relating
Uralic *-ta with IE *-od and has external implications to IE studies.)
An interesting side-conclusion is that [b] could potentially surface
in cases of *-p.
This voicing is also proven by the resulting quality of the thematic
vowel, as you keep on stressing. This is because nominative *-s,
just like inanimate *-d, was ALWAYS in final position and never with
an alternative suffix where the phoneme occured medially (unlike
3ps *-t, alternatively indicative *-ti). Voicing of all final
phonemes is as good as proven.
Our struggle involves the question of allophony versus phonemicization.
but since phonemicization is evidently more complex to substantiate,
you are making unneeded work for yourself.
Your Flaw #1:
Failure to accept the absence of **z in IE
------------------------------------------
There's simply no point in theorizing an evidently temporary existance
of *z when no such phoneme exists for IE itself. The obvious reason
why instances of *-d survive and instances of **-z don't is simply
because of the obvious fact that **z wasn't phonemic while *d is a
necessary and distinct phoneme for Pre-IE just as in IE. The stem
*nisdo- shows that [z] did in fact occur and that [z] is a proven
example of one of the possible realizations of a single phoneme *s
based on actual evidence.
So where *-t changed to an already phonemically distinct *-d without
analogical reversal, the change was totally and utterly complete and
irreversible.
Whereas, *z was never a distinct phoneme, so when *-s changed to
[z], it only changed its surface phonetics but did not change into
**z because it was still just *s with voicing. Similar, we find in
English [p] in "spade" and [pH] in "pull" even though they are
still understood as the same phoneme /p/. There's still potential
for both of these instances of /p/ to change into something else
like /f/ together in the future, despite phonetic differences.
So instances of **z never surfaced, exactly as we find, since [z]
was always just a variant of *s as in *nisdo-. This is sufficient
and inarguable. It even explains why thematic *& surfaced as *o
before nominative *-s. Obviously so if *-s was pronounced [-z] at
the time! The plural in *-es does not contain underlying *& (and
we never ever ever see alternation of *e/*o here, thereby proving
that point) and was unaffected when pronounced as [-Ez]. Likewise,
so was accented genitive *-as [-az] > *-os.
This is your Flaw #1 that can't explain the fact that there is a
_lack_ of **z in IE and also a _lack_ of need for hypothesizing it.
Instead of accepting facts, you do what you accuse me of: Twisting
IE to suit your assumptions. Thus, claiming another empty and
circular conjecture: That *z was somehow lost or remerged with *s.
We don't need to theorize this plug-hole theory. Unnecessary
conjecture!
Since your whole thesis is based on this empty conjecture it's
fair to reject it from this point on unless something in it is
modified. My theory is indeed such a modification.
Conclusion: [z] is a necessary allophone of *s,
not a seperate phoneme
Flaw #2:
Confusion between [z] and [S], two facets of IE *s
--------------------------------------------------
In order to explain how a set {*r, *l, *n, *y} could in
unity destroy *-s and only in the nominative, we need, as I
stated earlier, to accept that they all changed [s] to [S],
NOT [s] to [z]. (Note how your Norse example ignores a very
big monkeywrench in this example, namely the presence of *y
and the absence of *w in this set of influential phonemes.)
Voicing is immediately disqualified from this scenario since
we don't find *w in this mix, yet we know that *w had to
have been voiced. Therefore, to continue thinking voicing
has something to do with this is to perpetuate the intolerable
paradox.
Rather, the only way that the above phonemes could have
done what they did is by retracting an apicodental *s to an
alveolar position. This requires that *r, *l, and *n are
alveolar as well while *y's natural palatal quality can also
change [s] to, at the very least, [S^]. As can be seen, both
result in a "sh"-sound.
We can see in many languages how shibilants are particularly
prone to becoming "h" or zeroing altogether. This is because
of the static, unclearness that naturally accompanies such a
phoneme which opposes the relative crispness of /s/.
So [S] and [z] together are necessary to explain the phonemonon
that you cannot properly explain with a single and, as shown
above, non-existent **z.
Conclusion: Since [S] and [z] together address the facts as
we see them, necessitating only different
incarnations of a single phoneme *s, your theory
is comparatively deficient, even when the paradoxes
bred in your conclusions are ignored.
> The fact that both examples have an /s/ preceding the place
> where an *-s seems lacking is taken by you to count for nothing
It _is_ nothing. Nothing in *so shows that *-s ~ *-z must have
been there. It's your dillusion. The stem shows in the IE that
we reconstruct that it is entirely unmarked for case ending...
So the story is clear from the get-go and there was never any
nominative *-s here. If we saw accusative **so-m then perhaps
one would have a leg to stand on, but in reality your view is
baseless and shamefully reinvents IE morphology.
Similarly, *-syo can be functionally explained without any need
for a case ending after *-yo, since it can be seen as an ancient
_locative_ and arguements have been made in favour of a more
ancient and unmarked locative by others. In fact, I recall Mr.
Burrow mentioning some interesting and valid points about that
very subject. I've shown how insisting on a nominative *-s is a
paradox and subsequently a mad pasttime. So be it. It's your
path.
> What is so terrible in explaning a strange nominative
> form as an old nominative?
Because it presumes that *so was once marked with a case ending
when there is no indication of it at all. Ergo, it is unnecessary
assumption by Occam's Razor. It's immediately illogical conjecture.
> And what is terrible in explaining an inflectoinless cliticized
> form of the realtive pronoun as an old nominative which must have
> made good sense in many sentences where it was used?
Because there's nothing a priori that shows that it was a nominative
and since a locative in this instance makes just as much sense if
not more and since we know endingless locatives to have existed,
lacking any need for imaginary **-z, the nominative is again
unnecessary conjecture.
> The lengthening effect of the s-aorist marker which is
> never final also appears to demand a separate phoneme,
No. It necessitates a seperate phonetic realization of an already
reconstructed *s. Don't jump to conclusions. The commonality of
aorist and nominative *s, as I already stated, involves their
assyllabicity. In other words, Szemerenyi discovered what is
plainly compensatory lengthening of suffixes that unexpectedly
lost their syllabicity. So, in other words, common sense tells us
that the lengthening of the nominative occured long before the
phonetic voicing of *-s or the alveolarization to [S] after
certain phonemes in final position. The lengthening is caused
by desyllabification, syllable-timing and compensation.
Examples in other languages show that /s/ can even be more prone
to desyllabification in a syllable. Just look at Japanese again.
Romaji notation shows /arimasu/ but it surfaces as [aRimas:],
not expected *[aRimasU]. This happens specifically with /s/.
> since other s's cannot be assumed to have the same effect.
Obviously since no other instances of *s involve desyllabification
of the morpheme except the nominative and the aorist.
> Also the nominative marker I have inherent in the nom.pl.
> marking *-z-D which ends up being *-es without accent shift and
> with all the effects of earlier lengthening in the stem will
> seem to demand a phoneme doing something that cannot really be
> ascribed to just any sibilant.
Your theories are too demanding. The plural *-es is simply what
it is and had survived Syncope by the Suffix Resistance rule which
forbade desyllabification (expect in the rare circumstances above).
> But if "other sibilants" were other things at the relevant time,
> [...] Still that is grossly overstating the limited knowledge we
> have.
As I said, it has nothing to do with phonetics here. It involves
morphophonology and yes, you are grossly overstating.
>> Rather *so never had an *s and doesn't need to because
>> the morphological evidence shows that *so was uninflected
>> for case... This is what we find.
>
> This is circular.
It's circular to accept what we observe? Oh dearie. I'm
frightened of you now <:(
> We dont se a stem *so- in *any* other forms; we see *se- in *sa-H2,
> and that's it! This is no serious basis for far-reaching
> generalizations.
Exactly, so we accept what we see: *so is undeclined for case.
On "shortcomings" involving my view:
> How can *-yo be an endingless locative? Where do we find such a
> form? And where does this language show that it expresses
> possessive relations as if they were local? If a nominative *-s
> is added there is no problem with the -o-. And what is a
> nominative *-s doing in the neuter? Does that not "exascerbate[]
> the semantics" just as much?
The *-yo here evidently lacks a case ending, we see. The lack of
case ending is intolerable according to IE morphology that we
are aware of... but that's what we see. Either we accept that it
is endingless or we surmise why an expected ending disappeared.
The latter path is non-optimal, the very one you take.
Well, for god's sake, Jens. ANY case ending or phoneme could have
been there since many phonemes have disappeared from laryngeals
to potentially the *m in the thematic 1ps in *-o:.
You immediately jump to a bias conclusion but the discussion
between Richard and I emphasizes how the application of a
nominative is irrational and opposes the facts that we observe
(there is no **-syom, for example) and is dysfunctional
morphologically.
Since locatives can also be endingless as we observe, and since
it fits functionally, and since *o need not be explained
because we can't ASSUME as you do that it derives from a variable
thematic *& (> *e/*o), accepting a locative here is optimal.
Either way, we must assume something but in this view, we need
only assume that *ya (> **yo) was the original locative of a
newly replaced form and other instances of endingless locatives
in IE support that conclusion. In contrast, your view forces
us to accept the absurd and unobservable: That *-syo < **-syoz
when **-z is a non-obvious leap of faith beyond the facts that
we observe.
Nothing of what you say forces us in any way to conclude an
unobserved **z.
> Not if you read the suggestion correctly and tag *yos on to a
> *genitive* which would also respect the function,
Alright, but again, **-z is unnecessary and beyond what we in
fact observe. Therefore, your view remains weak.
> Don't you know that stems are inflected in IE?
I never said otherwise. However, there is strong case to view
the locative as originally uninflected and using a seperate
postposed particle like *bHi, *dHi or for that matter, the
ancient endingless locative *i from which the locative
ironically derives and which also surfaces in the indicative as
well.
Since an endingless locative *i is suggested for *is in the
above examples, an ancient endingless locative *yo for *yos is
also even more likely.
> Don't you know that the thematic vowel is *-e when word-final?
Yes it is, hence the vocative. But don't YOU know that the
ending *-syo does not automatically mean that *o < *&. It
could just as well be *o < *a since we all should know that
the genitive ending here _never_ alternates with **-sye.
> Don't you know that the thematic vowel is /e/ in the genitive
> of pronouns, but /o/ in the nominative *-os? Is this nothing?
You know I do since I accept the thematic *& underlying some
instances of *e/*o alternation. However, other stems would have
identical nominative and genitive forms without *-yo. Analogy
united all thematic stems together into a single paradigm, as
I said before.
> And is one suggestion of dissimilation applying to both cases much
> ado?
Yes. There's nothing you've said that shows that **-z must be
accepted on any level and I've provided an alternative suggestion
that makes this unnecessary assumption. Further, your Norse
example falls short since it doesn't take into account the
problem with *y.
= gLeN