On Fri, 21 May 2004 05:29:48 -0700 (PDT),
enlil@... wrote:
>Miguel:
>> Oh come on. It's essentially Jens' theory. Rob, like Jens
>> and myself, has the common sense to recognize that there was
>> only _one_ suffix *-mVn- (or *-tVr-), which developed in two
>> different ways according to the accent. None of that
>> nonsense about two different suffixes.
>
>After thinking about it some more, I developped a more optimal
>solution. Both *-me:n and *-mo:n develop from the same suffix
>*-mén-s/*-m&n-s depending on accent, yes. Also then, *-te:r and
>*-to:r simply from *-tér-s/*-t&r-s.
Does this mean that you now accept the fact that in compound
nouns, PIE could stress either the root (proterodynamic) or
the suffix (hysterodynamic)?
>> In the accusative, that is not possible. We would expect
>> the unstressed variants to have *-mn-m. (*-tr-m.) instead of
>> attested *-mon-m. (*-tor-m.). Jens explains this as analogy
>> after the nominative (with a handful of "original"
>> accusatives preserved, such as Old Persian pathim (<
>> *pónth2m.)).
>
>You're mad. Nothing is motivated by this complexity. Rather,
>*-m&n-m transparently becomes *-mon-m since *& lengthens before
>the following *n as already theorized by Jens himself (with a
>few modifications of my own, of course)
So why does it only lengthen in the proterodynamic strong
forms? Why is unstressed *& after *m and before *n simply
deleted in the hysterodynamic weak cases? What about all
the other suffixes, such as *-ter-/*-tor-/*-tr-? What role
does "lengthening before *n" play there?
The forms that need to be explained are:
PD HD AD ST
nom. *'-to:r *-té:r *'-to:r *'-to:r
acc. *'-tor-m *-tér-m *'-tr-m. *'-tr-m.
gen. *-tér-os *-tr-és *-tr-ós *'-tr-(o)s
Now, I understand Jens' theory about this. I don't fully
agree with it (otherwise I wouldn't have bothered to come up
with my own modifications), but I understand it:
PD/AD *'-ter-z > *'-tor-z > *'-to:r-z > *'-to:r
*'-ter-m > *'-tor-m > *'-tr-m (PD: analog. *-tor-m.)
*-tér-os > *-tér-s (analog. PD *-tér-os, AD *-tr-ós?)
HD *-tér-z > *-té:r-z > *-té:r
*-tér-m (unchanged)
*-ter-ós > *-tor-ós > *-tr-ós
ST *'-ter-z > *'-tor-z > *'-to:r-z > *'-to:r
*'-ter-m > *'-tor-m > *'-tr-m.
*-tér-os ~> *'-tor-os > *-tr.-s
(i.e. *é stays, *e > *o > 0, except when lengthened before
*-Cz, when we have *e > *o > *o:, initial accent rule for
static roots).
My theory is (svarita lengthening marked as â):
PD *'-târ-z > *'-tor-z > *'-to:r-z > *'-to:r
*'-târ-m > *'-tor-m
*-tár-âs > *-tér-os
AD *'-tar-z > *'-t&r-z > *'-to:r-z > *'-to:r
*'-tar-m > *'-t&r-m > *'-tr-m.
*-tár-âs ~> *-t&r-á:s > *-tr-ós
HD *-tár-z > *-tér-z > *-té:r-z > *-té:r
*-tár-m > *-tér-m.
*-tar-ás > *-t&r-és > *-tr-és
ST *'-tar-z > *'-t&r-z > *'-to:r-z > *'-to:r
*'-tar-m > *'-t&r-m > *'-tr-m.
*-tár-âs ~> *'-t&r-os > *-tr-os
(i.e. *á > *é, *á: and *â > *ó/*o, *a > *& > 0, except when
lengthened before *-z, when we have *a > *& > *&: > *o,
initial accent rule for static roots).
So what is your theory exactly (preferrably in a form that
is machine analyzable by the Sound Change applier)?
Mine is:
inputs:
* PD
các-tar-z
các-tar-m
cac-tár-as
* AD
cácc-tar-z
cácc-tar-m
cacc-tár-as
* HD
cac-tár-z
cac-tár-m
cac-tar-ás
* ST(1)
cá:c-tar-z
cá:c-tar-m
ca:c-tár-as
* ST(2)
ca:c-tár-z
ca:c-tár-m
ca:c-tar-ás
rules:
S=áíú
V=aiu
C=trcszm
R=lmnry
* svarita rule
a/â/SC-(C)_C
* AD rule
á/&/CC-C_(C)-(C)â
â/á:/CC-C&(C)-(C)_
* initial accent rule
á/&/a:C(C)-C_
á/&/a:C(C)-CVC(C)-_
a:/é/_
* reduction
a/&/_
* vowel quality
á:/ó/_
â/o/_
á/é/_
* [rules for *i(:) *u(:) omitted]
* Szemerényi lengthening
&/o:/_C-z
é/é:/_C-z
ó/ó:/_C-z
o/o:/_C-z
* zero grade
&//_
* loss of -z after R
z//R-_
* get rid of -
-//_
outputs:
* PD
cécto:r
céctorm
cctéros
* AD
céccto:r
cécctrm
ccctrós
* HD
ccté:r
cctérm
cctrés
* ST/PD
cócto:r
cóctrm
céctros
* ST/HD
cécto:r
céctrm
céctrs
=======================
Miguel Carrasquer Vidal
mcv@...