Jens:
> The original form of the causative developed into the
> structure *luk-éye-.
Ah. Why would it? Did the O-fix jump out this time?
Maybe it couldn't get along with the "k". No, honestly,
why would it?
> The fool then began forming new causatives with over /o/
> in them all,
Erh, is this "fool" you mention here meant to be a
long-deceased Indo-European speaking hunter, or a
particular Indoeuropeanologist you've read?
> But if you have a predictable accent on a specific
> syllable counted from the end, should the accent not
> move to the following vowel if a syllabic ending is
> added?
Yes. Actually I admit that it sounds like a ramble and
I was paying attention to more than one thing while I
was typing that. However, strangely it does make some
sense. If we form a hypothetical compound taking, say,
*kawana (which we know to have *-a at the end because
this does regularize the accent to the penultimate in
its paradigm) and we take, say, *werta, we might end
up with a senseless compound like *kawana-werta. Don't
worry about what it means. We're merely exploring
accentuation in compounds in this example.
So, you're trying to goad me into reconstructing
*wert with CVCC structure in MIE when I'm telling you
that CV(C) works just fine. So let's see what happens
here. We then would have *kawana-wert and guess
where the accent would be -- perhaps on the last *-a of
*kawana. Unfortunately for you, that kind of accent
pattern doesn't happen, so we can forget that. The
other possibility is to have the accent on *wa but
this still places accent on the FIRST element which
is not what we find in IE unless accent has been
altered by later rules like Acrostatic Regularization.
We then are stuck with *-a in *werta unless we want to
complicate our lives or destroy the regular accent
that we've achieved with CV(C) syllabics. Both actions
are anti-Occam.
Further, since we initially have a choice of *wert-
or *werta- underlying our *wert-, and since
compounding seems to favour a CV(C) structure with
*werta-, we have *wert- < *werta-. So if this is
true for this example, it must be true for all
relevant examples until evidence shows otherwise.
Again note that I'm not denying the other
possibilities but we need to prioritize what is
likeliest otherwise we won't be able to move on.
To boot, even suspected Semitic loans would seem to
indicate that a final vowel may have been present
because they exist in the donor language. It seems
strange that IE speakers forgot to pronounce the
final vowel all the time. (eg: *s^idc^u > *sweks)
Strangely then, yes, the lack of *-a WOULD destroy
the regular pattern we've attained with QAR and
would ignore a lot of other facts for no reason.
Didn't see THAT one coming, didja :)
> That is explained already: The accent moves to the
> following vowel if a flexive containing a vowel is
> added to the stem, but not if an added desinence has
> no vowels in it.
Wrong. The nominative ending originally contained a
vowel and was *-sa since it derives from a demonstrative,
remember? However it does not steal accent. Neither does
1ps perfect *-ha (> *-xe).
The actual rule is: The accent remains on the penultimate
syllable of the stem unless a _polysyllabic_ suffix is
appended to it, like partitive *-ata, for example. The
result is a regular accent in a complete word that falls
either on the penultimate or on the antepenultimate.
You might have confused this rule with the resulting
conclusion of the rule: that the instance of accent on
the final syllable suggests a terminating vowel that
has been lost. (eg: genitive *-os < *-asa)
>> > The IE word consists of root + suffix + desinence,
>> > sometimes with multiple suffixes, and the ablaut worked
>> > on the lot.
>>
>> But it wasn't always this way so it's absurd to impose
>> these same rules on Mid IE or your version of Pre-IE.
>
> It definitely was that way already when the Schwundablaut
> (syncope) worked. If you do not accept the morphology on
> which we observe that the ablaut works, you are limiting
> your scope arbitrarily to the small parts of the language
> that fit your preconceived ideas.
Yes, but I wasn't denying IE morphology. I was simply saying
that the underlying morphology is different than the last
layer IEists reconstruct. The analysis of IE morphology
too can be looked at different through preceding layers
of it. For example, we speak of "thematic vowels" in verbs
which are, as far as I see, unanalysable in MIE since they
are part of the root that is being conjugated or the suffix.
We can analyse a form *bHeronti within IE itself as
*bHer-o-nt-i but its presumed MIE ancestor *barena can only
be analysed as *bar-ena, where the *e that would become the
thematic vowel *o was once part of the suffix. Within that
layer of MIE, a grammaticist could not speak of a "thematic
vowel" in these paradigms. That's all I was trying to get
across.
> I'm afraid this is where we go separate ways. If you are
> not prepared to allow for the existence of suffixed
> formations in the corpus of wordforms that have been hit
> by the purely phonetic process of vowel loss caused by
> the accent (syncope), then there is no communication.
You're putting words in mouth and confusing what can't
be allowed in a logical discussion of Syncope with
what doesn't exist. I didn't say that these forms don't
exist but we can't allow them in a discussion on the
proof of Syncope because they introduce too many unknowns.
It can be summed up in Statistics: lurking variables. So
let's stop introducing them, shall we?
> The causative is included into the full picture of IE
> morphophonemics if given its proper input form.
Yet I still fail to see the motivation for applying your
O-fix process here. Certainly, you may apply it but we
can apply it to everything, if we fly caution to the wind.
I respect your attempt at regularizing the causative but
I don't see there being a sweeping general rule here.
It's not on the same level as QAR, a rule that applies
to most paradigms, verb or noun, with alternating accent,
or to our mutually accepted Syncope which applies even
more generally.
This O-fix rule in comparison just doesn't apply as
universally in the least, so don't you think it is wiser
to apply more general rules FIRST before lumping the
causative into this? If you do, you can better ascertain
what should be included as being due to this O-fix
phenomenon and what should not be included because of
these other more general and conflicting rules.
> Okay, I went too far. But the introduction of an
> "initial" vowel cannot have much bearing on the placing
> of the accent which is oriented relative to the end of
> the word, can it?
Well, it would affect accent in compounds if it were
the second element of the word.
> Is it not a strange thing that this putative initial vowel is never
> accented?
Not really, because MIE *esam "I am" shows that there
are indeed initial vowels that are accented, with automatic
preceding glottal stop of course.
> And is it not strange that the language has no roots of the
> structure VC- if it has CVC- and CCVC- and VCCVC-.
Hunh? Oh, I see what you mean. You probably would like a
counterexample such as *en "in", though really because of
the automatic glottal stop, we're speaking of CVC.
Technically, a form like *asteh- would be
CVC-CVC- if you count the glottal stop in [?as'tEh-].
Lacking alternation as with *es-, there would be little
need of the *?-. Of course, this is all assuming that it
did being with a vowel. It could equally have been *sateh-,
but since *a-Epenthesis doesn't apply here and since the
causative can be explained as simple *o-grade, things
are just fine.
> What *is* your basis for the assumption of an initial
> vowel before clusters?
Occam's Razor. Since there is nothing conclusively
pointing to consonant clustering in pre-Syncope MIE but
everything supporting a simple CVC structure throughout,
we needn't fret on this supposed clustering that we
don't find. Or rather, unnecessary complexity bites!
The rules on syllabics are automatic, so if one were
to ask a speaker of this language about it, they wouldn't
have a clue. It would be as second-nature to them as me
tapping the "t"'s in "little".
= gLeN