[tied] Re: Syncope

From: elmeras2000
Message: 31575
Date: 2004-03-27

--- In cybalist@yahoogroups.com, enlil@... wrote:

> The causative has o-grade [...] by *-(e-)ye-, which derives
> simply from the relative stem from which we get *yo-. It
> was tacked on to the verb to mark an indefinite object
> (ie "someone" or "something"), thereby transitivizing
> intransitive verbs. So with *mon-e-ye-, the semantics are
> effectively "to think somebody", or rather "to cause someone
> to think"... kinda like what I'm doing :)

I thought we were past this, but I was apparently wrong: The *-ye/o-
part of denominatives forms the present-aspect only. In old
participles of denominative verbs from s-stems like Latin modestus
the suffix *-ye/o- is not present, which is because its function was
not to form denominatives, but to mark the durative aspect of stems
like (suffixless) denominatives. This is also seen in the Gk. type
teleío:, aor. telésai. If the suffix *-ye/o- of derived (secondary)
durative stems were the relative pronoun it would surely be as much
needed outside the present system as in it. Also, the prs.-forming *-
ye/o- does not transitivize; it is actually the most common present
form of passives. You still have not done anything to account for
the -o- of the causative which was the whole object of the
discussion. Your alternative, topping the list of your priorities,
is only to change the subject. You are making me think, but not as
you think.

> The uses of the relative stem were productive elsewhere,
> such as in denominal verbs in *-ye- and in duratives,
> and in thematic stems to disambiguate the nominative from
> the genitive (ie: *-s-yo). Its use in the causitive to mark
> an indefinite object is particularly "synthetic" and thus
> emphasizing its youth. Certainly later than Syncope,
> otherwise we'd expect it to follow quant.ablaut better,
> wouldn't we.

There is no such IE morphology: the causative is not restricted to
sentences where the object is indefinite. It is even regularly used
reflexively (Russ. loz^ít'sja 'lie down') where the object must be
very high-ranking on the scale of topicality. And *if* this were a
sign of lateness (even "certainly" so) we would expect this feature
of the grammar to be reflected somewhere. I don't think we see that.

I accept the identification of the last part of the gen.sg. ending *-
o-syo with the relative pronoun, only I think it has been added to a
genitive form, not a nominative. The old underlying syntax would be
just as in Albanian and the history of Persian.

> Again, I see that there is nothing "ancient" about the
> causative.

Well, you can't *see* that there isn't. Funnily, I can see that
there is, see below and be reminded of earlier messages.

> As we both see, it violates normal quantitative
> ablaut, one of the ancient processes in IE that we can
> be sure of.

Not if derived from something which, through the changes of normal
quantitative ablaut, turns out this way. That is what I have shown.
You have even accepted its possibility, questioning only its maximum
simplicity. You may make things simpler, but you keep silent about
half the language. This is a matter of mileage. I'm sure Occam would
agree if he was so smart as you say.

> It doesn't follow any other ancient pattern
> either, unless we use O-fixing without restraint but we
> could almost apply that to any instance of an o-grade
> and it wouldn't tell us much. Maybe *wlkWo- is really
> *wlkWOe- :)

I couldn't disprove it if you were to insist upon it. I do not know
the rationale behind the naming of the wolf in IE, do you? And I am
sure the last word has not been spoken about the infixal o, so who
is anybody to say it was not present where we see an /o/ we don't
understand fully? I can only say that it solves nothing to bring it
in here as far as I can see, while it solves just about everything
in the categories where it imposes itself.

> So it's overdramatic to say that I'm just calling it
> "young and reckless" out of the top of my head. Given
> the idea above, the causative can indeed be young
> because its derivation is indeed transparent.

But its choice of -o- for its vocalism has not been made tranparent
by you - not for want of trying. And the "verb" to which *-ye- is
supposedly tacked on, i.e. *moné-, is not something we know, so what
is so transparent here? Where are all the other young forms of the
language that do the same?

> Of course,
> it's not "reckless" because it follows the most recent
> rules of IE -- a lack of regard for quantitative ablaut.

That is simply not true. Qualitative ablaut has indeed worked here.
And if it hasn't, and what we find is not reckless but in keeping
with morphological habits of younger periods I would like to see
other transparent younger formations that productively behave in the
same way. You asked for the question, now you got it: where is that
transparent basis of yours?

> > The very fact that it has a Narten counterpart should
> > also be ranked as a sure sign of its extreme antiquity.
>
> Not the sure sign I was looking for. Why is it "sure"? And
> define "extreme" in "extreme antiquity".

There is hardly anything in IE morphology as brittle as the Narten
ablaut variants. They survive only in isolated lexicalizations,
being constantly replaced by new forms that go by the normal ablaut
é/zero. Yet, their remains do form the contours of a system clear
enough to permit a very conservative and careful philologist like
Johanna Narten to lay it out in a masterpiece of a short and lucid
article in 1968 now broadly respected as a milestone in the history
of the field. There is no way such forms can be younger than the
basic quantitative ablaut rule (short vowel vs. zero depending on
accent). Consequently, if a morphological category exhibits Narten
variants that category is older than the ablaut proper. For the
causative Klingenschmitt has shown that it does.

> It just sounds
> like you're mixing the date of the use of morphemes together
> in a particular word with the date of the morphemes
> themselves. Those dates aren't the same just as the dates
> of the first instance of *wertmn versus the first instance
> of *wert- or of *-mn aren't the same.

This is about the actual use of the said morphemes in very specific
lexicalized coinings. They are manifestly older than the ablaut we
are talking about. You *need* all forms of the causative to be
younger than that ablaut. That is just not possible.

>
> > In a few very specific cases the element appears prefixed,
> > so there it is a prefix.
>
> Wait, a minute. What did I miss? Prefixing of *o- occurs
> in causatives? What causatives?

Time's up, so I'll reply now. Yes, prefixed o- in some causative
also, indeed that's how this thread began. I'll sketch it briefly
(from memory), though it may take more than a single sentence:

In the Watkins Festschrift Brent Vine recently analyzed the Greek
verbal-noun type /ago:gé:/ 'act of
leading', /edo:dé:/ 'eating', /odo:dé:/ 'smell' which is very odd
and not parallelled anywhere except partly in the Greek perfect
where 'smell' forms ódo:da and 'eat' forms ede:d- (ptc. ede:dó:s).
Vine's idea is that the funny perfect forms had been transformed
from the expected forms which would have been *o:d- (from both
strong *H3o-H3ód- and weak *H3o-H3d-), *e:d- (from *H1e-H1d- only,
since the strong form *H1e-H1ód- would have merged with 'smell'). In
their transformation the forms were given overt reduplications /ed-/
and /od-/, this giving the stems /edo:d-/, /odo:d-/. Vine's main
objective is the Mycenaean noun <o-ka> found to mean 'leadership'
or 'command' and thus synonymous with later Greek /ago:gé:/,
Doric /ago:gá:/. Vine now assumes that the Proto-Greek forms of the
nouns were *o:da: (1. 'eating', 2. 'smell'), o:ga: ('leadership',
according to Vine attested directly in Myc. o-ka which he
reads /o:ga:/); he further assumes the forms were changed to ed-
o:da:, od-o:da:, ag-o:ga: on the pattern of the perfects since both
types had IE *-o- vocalism (of some sort). That is a fine and well-
argued philology-based article that makes excellent sense
linguistically. -- Now, to me that prompts the questions how PGk.
*o:ga: was formed. Since the o-vocalism of the "toga type", which
this is, is my infixal -o- of prefixal origin, I could not well have
the expected -o- infixed, for that would lead to *H2og^áH2 >
Gk. "ogá:", not /o:ga:/. But with non-metathesized, i.e. prefixed o-
, it works fine: before the ablaut *O-H2eg^-é-H2, after ablaut (and
vocalization of the /O/, plus laryngeal coloration) IE *o-H2g^-á-H2,
which would in fact yield Greek /ago:gá:/, Attic /ago:gé:/. Likewise
for 'eat' and 'smell', IE *o-H1d-á-H2 > *o:dá: => /ed-o:dé:/, IE *o-
H3d-á-H2 > *o:dá: => /od-o:dé:/. Now for the real problem: by what
rule did the prefixed *O- abstain from getting metathesized and end
up an infix? Well, all three examples are roots of the structure
*HeT- (laryngeal + vowel + stop), so that should now be expected to
do just that. And indeed it is found to: The causative-iterative
formation of *H1ed- underlying Armenian utem 'I eat', already
generally derived from *o:d-eye-ti, can now be understood: Before
the ablaut *O-H1ed-éye-ti > after *o-H1d-éye-ti, exactly like Vine's
*/o:da:/. From the root of Latin opus, Skt. ápas- 'work', IE *H3ep-
(if not *H3epH-, which would not change anything as far as can be
seen), we get pre-ablaut *O-H3ep(H)-éye- > post-ablaut IE *o-H3p-éye-
ti, whence Germanic *o:bia-, which is German <üben> (OSax. o:bian).
The accent is the same in the Germanic verb and in Greek /ago:gé:/,
and the ablaut conforms to it in both cases. There is no way one can
get these forms by simply changing the vowel of *H1ed- or *H3ep(H)-
into (stable) /-o-/, for the vowel is long here. Nor does it have
much chance to be a sign of Narten ablaut, for the vowel is not
accented where we can check it. Even regarded in isolation these
forms strictly demand the positing of a prefixed, originally non-
syllabic, element which later changes into a vowel /o/. This looks
like a confirmation of what I have been saying.


> > In the categories that have this vocalism we find laryngeal
> > deletion which is absurd with a vowel and does not occur any
> > real vowel, not even -o- from other sources.
>
> Following the interpretation that laryngeals are "Saussured"
> because of anti-structure zeroed syllables during Syncope,
> there could certainly be some laryngeals that "appear" to not
> follow the rule, particularly if they postdate Syncope and
> Saussure's Rule. Why then must we pursue this further? What's
> so special about these not-so-missing laryngeals?

That they all belong to other morphological categories, and all ones
that also follow the oldest rules of IE ablaut we have. So these
formations are just as old as the others and indeed go back all the
way to the period when the (zero-grade triggering) ablaut worked. It
is only o-formations for which the infix analysis was set up and
independently motivated that delete laryngeals in the two sets of
environment #_Ro and oR_C. There is no deletion if the -o- is that
of reduplicated verbal formations, the acrostatic type ó/é, or the
thematic vowel e/o, nor indeed if the -o- is just a lexical given
root vowel. There has been a frantic chase going on for 25 years
trying to find schwa-deletion in other o-types as well, but the
harvest has been pitiful. I guess I know why.


> > We also find a variant form without the -o-, restricted to
> > specific root structures of a make that invites the
> > interpretation that the element had become an infix here
> > too but was lost before it was syllabified.
>
> I'll pass on the invitation for now. The above makes it
> sound like your rule isn't as ironclad as we would hope it
> to be.

Thanks for your kind hopes, but I prefer to stay objective. If two
sets of o-infix formations *alternate* the same, having both -o- and
zero in dependency of the root structure, they are either
functionally connected or composed of the same material to which the
observed rules then apply. In both cases the rules are impeccable.
What is wrong with alternation all of a suddden? The laryngeals were
found by analysis of alternations, and the same goes for this
element.


> Simply, the causative
> is an action with stative vocalism and a relative clause
> appended to it.

Simply? Why would the root segment of a causative have a stative
vocalism? And why would that vocalism alternate in a way completely
different from that of the perfect? Why is the vocalism of the
perfect combined with reduplication and this one never?


> So I'm thinking that our *mon-eye-ti comes from a purely
> mLIE construct *mané-y&-ti. Similarly mLIE *hWna:mn-y&-ti >
> *hWno:mn-ye-ti "to name something" or *wagH&-y&-ti "to
> wagon something" > *wogHe-ye-ti (well, only if wagons
> existed c.4500 BCE!)

The denominative has accent on the *-yé-, the causative is accented
*-éye-, so they are not the same, certainly not if they are of the
same vintage. But I have of course said that many times already.


Jens