Re: [tied] IE genitive

On Fri, 09 May 2003 21:04:54 +0200 (MET DST), Jens Elmegaard Rasmussen
<jer@...> wrote:

>> Finally there was a second zero grade
>> rule that further reduced *e, but not *o, in posttonic position.
>
>Almost so: The vowel from which the accent was removed by the initial
>accent rule was itself lost. Long posttonic vowels were retained.
>Deaccented /e/ (in reduplicated wordforms) was not lengthened, and so was
>lost. The reduction product of unaccented *e, which was (the prestage of)
>*o, was itself lost if not saved by lengthening. Some new unaccented o's
>were created by analogy (acc.sg.).

This is where we differ. I don't think *o is the reduction product of
anything: it's an original long vowel. The only case where we have *o
from *& is the thematic vowel (before a voiced segment), and in _some_
nominative singulars, where *-&C-z was lengthened to *-&:Cz >
*-o:C(s). The *-o- in the acc.sg. is not analogical.

>> I think the thematic vowel survived the first zero grade simply by the
>> fact that it was initially always stressed. For nouns such as
>> *wl.kWós or verbs such as *tudéti this simply follows from the
>> attested facts. For verbs such as *bhéreti or nouns (adjectives) such
>> as *néwos this requires original long grade (vr.ddhi) of the root
>> (**bha:r-á-t(i) > *bharát(i) > *bhéret(i); **na:w-á-z > **nawá:z >
>> *néwos), which is in itself not a problem (Sanskrit .

[I meant to say: "(In Sanskrit vrddhi with thematic vowel was still a
living process)"]

>Other saved vowels do not alternate like the thematic vowel, so this
>doesn't help.

The thematic vowel rule came *after* this initial "reduction" stage.

>> In the case of *wl.kWós or *tudéti we also have no problem with the
>> "second" zero-grade rule, but such a problem does arise with e.g.
>> post-initial-accent *bhéreti, which should have given +bhérti.
>>
>> My solution lies in the re-evaluation of the vowel qualities/
>> quantities in combination with a slightly different working of the
>> zero-grade rules (where zero-grade rule I becomes a reduction-rule,
>> creating schwa's, and the real zero-grade (schwa > 0) is the work of
>> zero-grade rule II).
>>
>> Imagine a pre-PIE vowel system consisting of:
>>
>> /a/ /i/ /u/
>> /a:/ /i:/ /u:/
>>
>> both stressed and unstressed.
>>
>> The vowel reduction rule reshaped this to:
>>
>> stressed svarita unstressed
>> a > &' o: &
>> i > (y)&' (y)e: (y)&
>> u > (w)&' (w)o: (w)&
>> a: > ó: a
>> i: > (y)é: i > (y)&
>> u: > (w)ó: u > (w)&
>
>Why would stressed vowels be reduced?

It happened in e.g. Tocharian.

>It seems to pull the carpet from
>under the initial insight that Schwundablaut is accent-governed.

Read on...

>> Disregarding the labializing and palatalizing effects of former *i(:)
>> and *u(:), this amounts to a system where we have:
>>
>> stressed unstressed
>> &' &
>> a
>> ó: o(:)
>> é: e(:)
>>
>> The initial accent rule turns all cases of /a/ (from unstressed /a:/)
>> into stressed /á/, and reduces any following stressed /&'/ (from
>> stressed /á/, /í/, /ú/) to unstressed /&/.
>>
>> We now have:
>> stressed unstressed
>> &' &
>> á
>> é: e(:)
>> ó: o(:)
>>
>> Stressed /&'/ and /á/ merge as /é/, and the zero grade rule then turns
>> all unstressed /&/'s into zero, giving:
>>
>> stressed unstressed
>> é
>> é: e
>> ó o
>>
>>
>> This accounts for most noun patterns:
>> root: **sám-z, G. **sam-ás -> *séms, *smés
>> root static: **pá:d-z, G. **pa:d-ás -> *pó:ds, *péds
>> PD: **h2ák-ma:n-z, G. **h2ak-mán-a:s -> *h2ák^mo:n, *&2k^ménos
>> HD: *pa-h2tár-z, G. *pa-h2tar-ás -> *p&2té:r, *p&2trés
>> PD static: **wá:d-an, G. *wa:d-án-a:s -> *wódr, *wédnos
>> collective: **wad-á:n-h2, G. *wad-a:n-ás -> *udó:r, *udéns
>>
>> For instance, the G. **pa:d-ás becomes **pad&'s (reduction rule), then
>> **pád&s (initial accent rule) then *péds (zero-grade rule).
>> Likewise **wa:d-án-a:s > **wad&'nos > **wád&nos > *wédnos, or
>> **wad-a:n-ás > **w&dan&'s > **w&dán&s > *udéns.
>
>This looks like all the rules I made twenty-five years ago, with only
>slight deviations. You just use a different notation for not-yet-surfaced
>full vowels.

Of course most of it is based on your rules. I can't help it if you
were mainly correct. But my amendments do amount to more than simply
a different notation. I have completely different views on the origin
of *o [as outlined above], and the version of the "zero-grade rule"
(reduction rule) I present here has the advantage of (a) being very
different from the second [now the *real*] zero-grade rule (which
makes the existence of the two rules as separate entities more
understandable); (b) helping to understand the working of the initial
accent rule [there is no need anymore to justify why the accent was
not drawn back by an /i/ or a /u/ or one of the syllabic resonants, as
there were't any yet]; and (c) it makes the possibilities for survival
of the thematic vowel a lot greater: now all they had to do to escape
deletion was to not be schwas, which is exactly what the thematic
vowel rule [which we need in any case] can easily be made to do (& > a

> e before voiceless, & > a: > o before voiced).

>> To explain the thematic forms, all we need is to suppose that the
>> thematic vowel /&/ was affected not only by a following voiced segment
>> (giving /&:/ > /o/), but also by a following voiceless segment (giving
>> strengthened **/a/ > /e/, the same sound that resulted from unstressed
>> **/a:/). In other words, the "thematic vowel rule" is a rule that
>> turns morpheme-final schwa's into "clear" vowels.
>
>It processes stem-final (not word-final) vowels. There are no other
>stem-final vowels than the thematic vowels.

Stem-final is fine. I was just taking a broader view based on the
o-stem vocative *-e. Some verbal endings [*-h2a, *-th2a, *-e; *-mé,
*-té] also have word final vowels (which are not dropped).

>> We then have:
>>
>> reduct. them-vow. init-acc. zero
>> **tawd-á-t t&wd&'t t&wdát t&wdét tudét
>> **tawd-a-mán t&wd&m&'n t&wda:m&'n t&wdóm&n tudómen (!)
>
>The tudati type is not that old. This is otiose.

The avidat type then (thematic root aorist). Same thing, formally.

>> **bha:r-á-t bhar&'t bharát bhéret bhéret
>> **bha:r-a-mán bhar&m&'n bhara:m&'n bhérom&n bhéromen (!)
>>
>> Note that the ending -men (-mes) is also unaffected by zero-grade, but
>> that might be analogical after normal stressed -mén (-més).
>
>It's rather a matter of what that ending really contained. The same
>surprise is seen in the other 1./2. du./pl. endings, act. and mid.

In any case (except perhaps for the verbal endings I mentioned above)
it's a different story than the story of the thematic vowel.

=======================
Miguel Carrasquer Vidal
mcv@...