On Tue, 13 Jul 2004 12:57:13 +0200 (MET DST), Jens Elmegaard
Rasmussen <
jer@...> wrote:
>On Tue, 13 Jul 2004, Miguel Carrasquer wrote:
>
>> The reason is that I like to think of these suffixes as
>> erstwhile independent words. I can see how desinences may
>> have no underlying vowel, but if suffixes were once separate
>> words, I'd expect them to always have an underlying vowel.
>>
>> This is of course just an impression of mine, subject to
>> revision, and it's hard to offer solid proof of why suffixes
>> must once have been independent words. The best argument is
>> perhaps that it's the suffix, after all, which carries the
>> case endings. This is perfectly understandable if deriving
>> from Gruppenflexion of a noun phrase consisting of two
>> independent entities: the root and the suffix.
>
>I see it a little differently, but not much. I have noticed that the
>vocalism of suffixes is completely monotonous, being only /e/ or
>what comes from it. It is even worse if the /e/ is secondary itself. IN
>the desinences however I notice other vocalisms also: gen. *-os, even
>instr.pl. *-bhis, loc.pl. *-su, perhaps allative *-a(i), imperative *-dhi
>which may all take the accent and reduce the rest of the wprd to
>zero-grade. I therefore find it difficult to regard the desinences as more
>reduced than the suffixes.
I didn't say the desinences were more reduced in general
than the suffixes. I said some desinences are vowelless
(and therefore unable to carry the accent). There's no
reason why other desinences, especially stressed vowelled
ones, should suffer more reduction than roots or suffixes.
As to the monotony of suffix vocalism, I have a different
opinion. The vocalism of suffixes is as monotonous as that
of roots (in other words, pretty monotonous), but it is
still possible to discern traces of an earlier
multivocalism. As elsewhere, **a, **i and **u have merged
when standing between consonants [apart from their effect on
neighbouring consonants], but **i and **u have separate
reflexes when at the beginning or at the end of the suffix
(or root, or desinence). The status of i-stem *i and u-stem
*u as suffixes on a par with, say, *-men or *-ter is perhaps
debatable. I certainly think there are recognisable traces
of a trio of suffixes **-an, **-in and **-un, reflected (in
neuters) as -r/-n- (*wodr, *wednos < **wá:d-an,
**wa:d-án-a:s), -i (-ir) /-n- ~ -y- (*h2ósth2-i (Arm. oskr),
*h2ésth2nos/*h2ésth2yos < **h2á:sth2-in,
**h2a:sth2-ín(Y)-a:s) and -u(r)/-w- (*pók^u(r) ~ *pék^u(r),
*pék^wos ~ *pk^wós < **pá(:)k^-un, **pa(:)k^-ún(W)-a:s).
For the same reason (labialization of *t > *sW (> *s) after
stressed *ú(:)), I reconstruct the suffix in the word
*méh1-not-, *m(e)h1-nés- "month, moon" as **-nut- (this is I
believe confirmed by Kartvelian *mttute "moon" if from
*m(V)t-nút(-)e).
There appear to be no suffixes with inherent long vowels
(**a:, **u: > *o, **i: > *e:), as some roots have, but the
suffix vowel is liable to be lengthened by phonetical
("svarita" lengthening as in *h2ák-mon-) and morphological
(collective forms such as **wad-á:n > *udór-) lengthening
processes. But then the length of (nominal) roots may
perhaps also be secondary, as suggested by the fact that
/e/-grade is often found where a corresponding verbal form
exists.
> I would suppose the desinences were once
>independent words (perhaps very long ago, seeing that some of them do not
>contain vowels I can detect)
This is obviously so in the case of e.g. the active personal
endings, which are clearly based on agglutinated personal
pronouns. In general, however, the desinences were not
separate words anymore at the stage I was describing, the
stage at which root and suffix formed a noun phrase with
Gruppenflexion (on the suffix, obviously).
>I am quite willing to accept that the suffix *-t(e)r- when forming
>oxytone agent nouns did so due to its high degree of animacy. That would
>allow its identification with the action noun suffix *-t(e)r of *-tr,
>*-ten- (Lat. iter). That would mean the suffix was originally vowelless,
>so this changes by the minute now.
I forgot to mention that there is one aspect of PIE
morphology that has been bothering me, and that perhaps can
be illuminated if there *were* (despite my unwillingness to
believe in them) vowelled and unvowelled suffixes: the
feminine suffix *-ih2 is generally pasted onto a stem (root
+ suffix) containing the suffix in its weakest form, but in
some cases (most notably with the suffix *-wen-/*-wer-) it
is not (e.g. *pih1-wer-ih2). Any other cases?
=======================
Miguel Carrasquer Vidal
mcv@...