From: elmeras2000
Message: 32847
Date: 2004-05-21
--- In cybalist@yahoogroups.com, "Rob" <magwich78@...> wrote:
Continuation: Your cue was:
> It seems that the 'house' root is tricky indeed. How do you
explain
> the contrast between e- and o-vocalism?
I take acrostatic o to represent the timbre assumed by lengthening
of already-long /e:/. That would explain that we find o-vocalism
where the verbs and some neuters have /é:/, and that the weak
alternant is /é/ (with accent retraction as I assume additionally).
I have added the rule that a long vowel in nominatives (and
collectives) was shortened if the stem ended in two consonants (or
perhaps before a cluster of three word-final consonants). That
explains the short vowel of nominatives in *-ont-s (Ved. -an), *-onk-
s (Ved. -aN) and a lexeme like *nókWt-s. I was anticipated in this
by close to a century by Møller and Streitberg who said the same
(buried in a lot of nonsense) in 1880 and 1894 respectively.
That rule comes in handy to explain the nom.pl. in unaccented *-es.
I depart from two sibilants, one marking the nominative, the other
the plural, i.e. nom.pl. *-s-s like acc.pl. *-m-s. If the nominative
sibilant was voiced we may write the original nom.pl. *-z-s. Now, in
a form like *H2nér-z-s 'men', the nom. /z/ should lengthen, and the
clustering should shorten, so we get back to the point of departure.
I then assume that, after ablaut is over, the final cluster *-C-z-s
is broken up by an epenthetic -e-, giving *-C-e-zs, whence IE *-C-
es, e.g. *H2nér-es, Gk. anéres, Ved. náras. With suffixed nouns like
*H2ék^-men- 'stone', we get *H2ék^-men-z-s, for which now both
processes have to be assumed: first reduction of unaccented -e- to
*H2ék^-mon-z-s, then nominative lengthening caused by /z/ giving
*H2ék^mo:nzs, meaning that at the time when unaccented short vowels
were lost this word had a long vowel in its suffix and so retained
the suffix vowel; next, shortening caused *H2ék^monzs, and
epenthesis made it into *H2ék^monezs, whence PIE *H2ák^mones, Gk.
ákmones.
The same shortening rule will also process the nom.pl. of 'foot'
and 'night': 1. *pé:d-z-s > *pé::dzs > *pó:(:)dzs > *pódzs > *pódezs
> IE *pódes, Gk. pódes, Ved. pá:das, Arm. otk'. 2. *né:kWt-z-s >
*né::kWtzs > *nó:(:)kWtzs > *nókWtzs > *nókWtezs > IE *nókWtes, Gk.
núktes, Lith. na~ktes, ON nættr. I have bracketed the second length
mark of teh stage /o:(:)/ since I do not know how it was realized.
That brings me back to the core of the question:
The passage of lengthened /e::/ to an o-timbred vowel can be
regarded as regular if unaccented /e/, on its way to zero, passes
through a stage with o-timbre. I believe it does; better, perhaps,
better, the prestage of /e/ is reduced to the prestage of /o/. That
can be used here if a trimoric vowel had two syllabic peaks (as is
generally assumed for long circumflex vowels anyway); then *pé::dz
is the same as *péedz (disyllable), and the unaccented half would
now change to /o/, whereupon the resulting *péodz would be expected
to come out as *pó:ds by mere contraction. In the nom.pl., we would
get *péedzs > *péodzs > *pó:dzs > *pódzs > *pódezs > PIE *pódes.
That in fact is what we find.
This should not be overlooked when dealing with the acc.sg. I see no
way of accounting for forms like *H2ák^-mon-m. and *pód-m. as
phonetically regular, but I can account for them very easily by
analogy. End-stressed types such as *p&2-té:r, *H2né:r had acc.sg.
in *-tér-m., *H2nér-m. (Gk. anéra, Ved. náram), i.e. with the same
vowel as the nominative, only short; or, with the same vowel as the
nom.pl. *p&2téres, *H2néres. Thus, to nom.sg. *pó:d-s, nom.pl. *pód.-
es, thee expected *analogical* acc.sg. would now be *pód-m., which
is what we have (Gk. póda, Ved. pá:dam, Arm. otn). The phonetically
regular forms with long /e:/ in root nouns do not seem to exist at
all, while the regular forms with zero vocalism in suffixed
formations do seem to have left a few traces.
The word *do(:)m-/*dem- is not easy. One would like to regard it as
a root noun of *demH2- 'build' (Gk. démo., dédme:mai, neó-dme:tos,
Doric -a:-). That provokes the question, where did laryngeal go in
the gen. *dém-s? LIV suggests it was lost in the nom. *dó:m(s) which
had o-timbre, but that is not the way I see the socalled "Saussure
effect" which according to my own results demands an o-vocalism of a
very special background which is not offered by the acrostatic
nominatives (I find the many attempts to salvage it without the
restriction I have found pitiful and deliberately biased). I would
suggest that the laryngeal in *démH2-s + poti-s 'master of the
house' was lost in the compound-like juxtaposition. It is a well-
known rule that there are no schwas in compounds, so perhaps they
weer also ironed out in justapositions above a certain degree of
closeness. That would account for all without compromising knowledge
we already have.
> If all of this is the case, then all that one must further assume
is
> that (most of) the substratum languages had stress accent.
I don't see the real relevance. The accent retraction in acrostatic
paradigms was followed by further reductions in the part where the
accent came from, and that is all pan-IE. To be of relevance, it
would have to be a prestage of PIE itself that was adopted by a
foreign population and got disfigured before the time of the
protolanguage. Now, initial accent is not strange at all, certainly
no more stange than many of the sound changes we ascribe to the
individual languages without accusing foreigners of having
introduced them.
[...]
> That having been said, did PIE *kWo-/kwo- > L ca-?
No, Schrijver's idea is that *ke- (unpalatal k + e) did, and also
that *mo- and *wo- gave ma-, va- (there is also ve- however, as in
vester). It is hard to find a good example of kw- in Latin (and
almost anywhere): ca:seus and vapor do not seem to have been
processed the same way, so what *were* the rules like?
> There are many
> Latin words (verbal and nominal) that have a-vocalism where e-
> vocalism is expected based on the reconstructed PIE etymologies.
Schrijver has addressed many of those, adding Celtic as a source of
many unexpected a's. The rules look very much alike.
[...]
> Interesting. Is it also possible that the L -is genitive resulted
> from a contamination with i-stems? There was some confusion
between
> root nouns and i-stems, with nom./acc. pl. -e:s becoming shared
> between the two (the nom. pl. from the i-stems, acc. pl. from root
> nouns). Also cf. root noun dat./abl. pl. -ibus for expected -bus
> (e.g. regibus instead of regbus).
The genitive of i-stems was *-eis which was in fact generalized in
Oscan and Umbrian (Osc. -eis of o-stems, i-stems and consonant
stems). I find it quite possible that the reflex of *-eis which
would be *-e:s with a narrow e-vowel, was replaced by *-es of the
consonant stems (ablative?) before it reached the stage *-i:s. That
would bring the Italic languages a bit closer together.
Jens