--- In
cybalist@yahoogroups.com, enlil@... wrote:
> 1. Syncope on some level MUST exist in Pre-IE
> ---------------------------------------------
> If anything can be agreed on about Pre-IE, it's that it
> had a stage where syncope of some extent had taken place. That
> can be shown simply by alternations of the *pAlh-/plAh-/plh-
> type (*A = any vowel).
This premise is not true. It is identical with what many used to
believe, myself included, until some who cared enough took the
trouble to check it. And what they found is not compatible with this
theory, whích is actually as old as the knowledge about ablaut. The
fundamental change was achieved by Anttila who made a thorough
investigation and got susprising results which cannot be brushed
away by an invocation appealing to a principle that cannot hold. In
his 1969 Los Angeles monograph Proto-Indo-European Schwebeablaut
Raimo Anttila presented a full survey of roots that apparently
appear both as Benveniste's Thème I and as Thème II. The findings
are momentous, for they show that alternants like *perk^- and *prek^-
are not on a par, for only one is old, while the other one is a
backformation based on the zero-grade which was opaque. On the basis
of zero-grade *pr.k^- speakers could not see or hear where the full-
grade vowel had been lost from, so they sometimes made wrong guesses
and happened to insert the full-grade vowel in its wrong place, thus
getting phony full-grades like Lith. pers^-ù 'I woo', while Latin
precor and all other full-grade forms of that root present the
structure *prek^-. That can only mean that the position of the full-
grade vowel is a lexical given, meaning that the appearance of a
given root as a Benvenistian Thème I or Thème II is idiosyncratic
and must be accepted as a property of the root itself. Thus it is
wrong to derive both alternants from a common prestage of the shape
*perek^- of which the alternants would the present two accent-
conditioned variants. While this of course makes excellent sense on
the phonological plane, it is absolute nonse in morphology. The
revealing point is that the categories showing Thème I are the very
same as those presenting Thème II. Thus, also with 'fill', the old
form was plainly *pleH1-, and the mirror-image variant *pelH1- is
nothing but a post-zero-grade analogical backformation. Anttila's
book is full of such examples that illustrate very clearly how a
given rood adheres to a specific shape for its full-grade variant,
thus with almost complete consistency *g^enH1- 'to beget', but
*g^neH3- where one cannot sensibly ascribe the choice of Th.I
in 'beget' vs. Th.II in 'know' to the differenc in the index number
hanging on the laryngeal. A most spectacular advance achieved by
Anntila's investigation was that it could be seen how often an
alleged alternation *TeRH-/*TReH- (Gk. TeRe/TERa/Tero- [-> ToRe-]
vs. TRe:/TRa:/TRo:-) could in practically all cases be instead
nothing but the basic alternation between e and zero, because TRe:-
/TRa:-/TRo: were also the normal zero-grade forms of roots of the
structures TeRH- and TReH- alike. That did away with by far the
largest amount of potential Schwebeablaut examples. So much so in
fact that what little remained could easily be ascribed to analogy
(which of course had been a possibility all along.)
>
> By adding vowels within that root and others, we can regularize
> its shape. We learn from this exercise that accent is the cause
> of this Syncope event:
>
> pre-Syncope
> *pA'lAh- > *pAlh- (stress on 1st syllable)
> *pAlA'h- > *plAh- (stress on 2nd syllable)
> *pAlAh- > *plh- (no stress)
We do not learn that, for only *plAh- is old, while *plh- is its
reduction caused by syncope, and *pAlh- is a backformed pseudo-
fullgrade made on the basis of the opaque zero-grade which gave
speakers no hint at where to put in the full-grade vowel if they had
forgotten or didn't care.
> It suggests an earlier stress accent, rather than the tonal
> one that must be reconstructed for Reconstructed IE proper in
> order to best account for the vocalic wear and tear.
It does indeed point to an accent of some expiratory prominence, but
it does not in any way exclude that the accent had other properties
also, such as tonal height. Nor does the tonal accent invoked to
account for some other ablaut-like alternations exclude that these
tonal properties were accompanied by differences in expiratory
prominence. It has never been proved that the character of the IE
accent changed.
> At least
> some syncope is needed for any good theory on pre-IE.
Certainly.
> It
> must also be assumed until shown otherwise that Syncope was
> a single event by way of logical simplicity.
That is not a valid inference. In its simplest form it would demand
that the vowel loss occurred without an intermediate stage, i.e.
with full loss from one day to the next with no weakening on the
way. I even believe we know this is wrong.
>
> So basically: You must accept Syncope in pre-IE!
Who doesn't?
> 2. ALL initial consonant clusters are caused by Syncope
> -------------------------------------------------------
> The issue between us (Jens and I) apparently involves our
> different views on what is logically the simplest solution
> concerning the _extent_ of Syncope. A clash of "simplicities",
> as it were.
Guilty as charged.
> Jens seems to be saying that it is simpler to reconstruct
> these initial consonant clusters to the farthest reaches of
> pre-IE in our ignorance. I say that it is simpler to
> reconstruct this previous stage with a same clear restriction
> on syllable shape to the form CV(C) for ALL pre-IE syllables.
> Afterall, every human language has restrictions on syllable
> shape. How can we possibly ignore this?
Hey, I have not advocated unrestricted initial clustering in IE. Is
this at all serious, or is it still personal?
> Luckily for me, everything about IE shows that there is
> indeed this simple, underlying CV(C) shape. We may find
> complex medial consonant clusters as in *we:gHst but of
> course, this is always the product of multiple affixes
> (*we:gH-s-t). Perhaps *kerd- comes the closest to counterproof
> but because of Syncope, it could derive from *kerAd-, not
> *kerd-, if all we know. An indivisible root with a pattern
> like CVCCCV(C)- for example could serve to break our simple
> CVC notion but I'm not aware of any such root in PIE. I'm
> not aware of any root that undermines this CVC solution.
Was that "everything about IE", and did it show the all-pervasive
status of "CV(C)"? I didn't see it showed anything. For all
morphological operations in IE *g^enH1- is a root, and so is *g^neH3-
. The former certainly had a root-final cluster, and the second may
or may not have had an initial cluster. We cannot tell on this
basis - but I claim we can on the basis of the o-infix observations.
>
> Now, to reconstruct any initial consonant clusters in this
> pre-Syncope stage as Jens would have us do forces us to
> either accept that (C)CCVC is allowed only for initial
> syllables, oddly without any proof of this shape in
> subsequent syllables, or it forces us to accept that the same
> shape was allowed for other syllables with some hypothetical
> phonetic erosion perhaps. We certainly have more questions
> but what have we answered by this? This is a deceptive excuse
> of an "answer" to pre-IE Syncope. By simply pushing back
> initial consonant clusters in PIE (eg *stex-) to an earlier
> stage, we explain absolutely nothing at all. An unnecessary
> multiplication of hypotheses quickly ensues. Nice magic act.
In fact IE morpheme structure does teach us that morphological
elements following the root are markedly simpler than the radical
elements. The suffixes never have any other vowels than short //e//
(or no vowel) underlyingly, while the desinences have more,
viz. //e//, //a//, //i//, //o//, //u//, but, in contradistinction to
the roots, desinences never have long vowels underlyingly. There are
well-known languages doing something of the same. German has no
vowel oppositions after the root segment, and Latin has very little
in the middle, and less in the final syllable than in the first. The
Latin structure, while demonstrably an innovation, offers a quite
neat parallel to IE itself. The reason for this uneven structuring
seems clear to me: The suffixes are very reduced forms of stem
elements that followed the root, while the desinences are reduced
forms of more independent morphemes that were added after the entire
stem.
The premises invoked are not even quite true. The structure of the
middle-voice participle *-mH1no- is not much simpler than some of
the roots you won't accept; the suffix may be analyzable into
smaller units, but what if we did not know, or we are wrong about
it? There is also an extended form of the men-suffix, i.e. -smen-,
which even has a variant *-sment-, which is not so simple either.
But basically your simple dream has come true in the suffixal parts
of IE word stems of some length.
> So with *kwon-/*kun-, it may either derive from *kwA'n-/*kwAn-
> (a la Jens) or *kAwA'n-/*kAwAn- (a la gLeN). However, the latter
> is logically superior because by conforming to a strict CV(C)
> structure for all syllables, we avoid any further phonotactic
> issues that the former solution overloads us with while offering
> an ordered language structure right down to the last coda.
The "ordered language structure" should be the result of an unbiased
inspection of the facts, not just a dream about a maximum simple
language structure, as it is here.
Using this as license, Icould insist that an old lost language of
the South Pole had only open syllables. If someone ask why, I can
refer to your authority and shout, Shut up, Dummy, open syllables
are simpler! I fail to see what kind of progress this creates.
>
>
> 3. Stop to breathe
> ------------------
> By this point, we should see that the solution I side with is
> the most intuitive, doesn't raise added problems, and can solve a
> wide array of otherwise arcane alternations such as *dyeu-/*deiwo-.
> We now see that all initial consonant clusters must be the
> product of Syncope until it is shown otherwise that a CV(C)
> shape is insufficient to explain early IE forms.
It is intuitive if one cares for only a self-chosen subset of the
facts; the theory does not command the rest of the facts. The whole
set-up had the purpose of making it predictable when a given root
(or pseudo-root, base, thème, whatever), surfaces as a Thème I and
when as a Thème II. It does not do that. In fact it *never* does
that, for the alternating Thèmes I/II belong to the same categories
and so should be subject to the same accentual effects, which means
that accent doublets is something they just are not.
Your theory demands that all clusters be derived from older
sequences with vowels beween them, not because you find that or
indirect reflexes of it, but solely on the basis of a concept of
simplicity for its own sake. The part of the theory saying that
clusters following the root vowel have lost a vowel is disproved by
the way the inflectional accent moves, since it always leaves the
complete frame of the root, no matter how long it is after the root
vowel. That part of the theory must be wrong.
But the part of your theory that reckons with possible or even
mandatory underlying vowels in clusters preceding the root can in
principle be right or wrong - to most scholars. To me it can only be
wrong, for whoever takes the observations attached to the o-infix
seriously is forced to exclude the existence of underlying vowels in
the roots earlier than the position to which the o-element moves in
the structures where it ends up being infixed as a consequence of
metathesis.
It is cetainly simpler to allow for the presence of less than the
theoretical maximum of simplicity in structures we cannot inspect,
than it is to insist that everything unknown always lived up to the
very theoretical maximum in terms of simplicity. Also simplicity may
be expected to have extremes that are less common than partial
simplicity. Then what do we do? We keep the unknown open, for it is
not even simple to guess at maximum simplicity. I'm sure Ockey would
have understood this, if he was so smart as I have seen him
portrayed.
> Of course beyond the above lies my Penultimate Accent Rule (which
> requires accepting that there was syncope at the END of a root
> too (no big revelation either). But the gang needs to accept the
> above points first before that rule can be fairly examined.
So much for its prospects. Need I add that it has been tried many
times and has failed every time?
Jens