Skt V>VV (was: Re: *pYerkW+)

From: stlatos
Message: 49449
Date: 2007-07-31

--- In cybalist@yahoogroups.com, "stlatos" <stlatos@...> wrote:
>
> --- In cybalist@yahoogroups.com, "gprosti" <gprosti@> wrote:

> > Are you saying that *C-ry/rw/ly/my/etc. give *C-yy/C-ww
> > (sporadically or otherwise)? If so, what is your evidence for this?
>
> Yes. The first indication came from studying the comparative forms
> of adjectives, mainly in Sanskrit.

In post-Vedic Sanskrit there was a rule that lengthened a vowel that
was in a syllable by itself between two syllables. This usually
affected a V from C+syl, as:

*pYel-xY-n.of > pári:n,as- 'plenty'

*pr,zd-n-ku+s > pr'da:ku-s 'leopard'

No lengthening occured if the vowel was in a syllable with another
sound, as:

ks,atríya- 'sovereign' (ks,at-rí-ya)

However, in PIE a dental was put in a different syllable before y
(if possible). When y>iy after a heavy syl. in many IE languages the
syllabification remained, so:

*tr,t-yo+s > *tr,t-i-yo+s > tr,tí:ya- 'third' 1

*gWhr,n.-yo+s
*gWhr,n.-i+went+ > ghr'n.i:vant- 'glowing' 2

No lengthening occured if the accent followed the vowel that was in
a syllable by itself, as:

*xYis-x-ro+ > is.irá- 'vigorous'

*pYlet-x-mo:n. -(o)n.+ > prathimán- 'wideness'

But the opposite accent-scheme applies to between P and r (including
v_r):

*stew-x-ro+ > sthávira- 'sturdy'

*gWmY,bh-x-ro+ > gabhi:rá- 'deep' 3

Before a plain m in the previous syl., x,>u:

*domxo+s *dem-x+ 'house'

*dem-x+ >> dámu:nas- 'dear to the house(hold)' 4

Indic had x(),>0 between dental obs. between syl. (with some analogy
in paradigms). This wasn't part of a PIE loss in compounds. The
older syllabification is still seen in:

*per i dxW,-to- > pári:tta- 'given away'

by i>i: between syl. which wouldn't occur if *per id-to- originally in
PIE.

The timing was post-Vedic because the optional change of unaccented
i>0 before y (if possible) had already happened. Optional comparative
forms like:

tèw-i-yós+
táw-i-yas+
táw-i-yas+ or táw-yas+
táw-i:-yas+ or táw-yas+
táv-i:-yas+ or táv-yas+ 'very strong' 5

Analogy between closely related words can obscure the above rules:

Regular:

*xYer-xY-tor+ > Skt aritár- 'rower'

*pew-x-tor+ > Skt pavitár- 'purifier'

So no i: in their derivatives:

*xYer-xY-trom xYr,xY-tr(e)+ > Skt arítra- 'oar'; Lith irklas

*pew-x-tro+ > Skt pavítra- 'filter, etc.'

*kWen.-x-tro+ > Skt khanítra- 'spade' 6

The other option (that the accent shifted to i _after_ this rule)
doesn't work out because of specific ordering needed for other rules.

1
The form *tr,tyo+s not *trito+s because these weren't originally
formed by *+to+ at all:

Metathesis of tkW>kWt after PIE is seen in *pn,tkWus
> Hittite pankus but Latin *kWn,kWtus > cu:nctus.
Even in PIE there was some met.: adj. *p(e)ntkW+o+s >
*p(e)nkWtos (then analogy treating ordinals as pp,
seen clearly in *tr,yos + tos > *tr,tyos since a truly
old form in *+tos would be *tritos).

2
*gWhr,n.o+s *gWhr,n.(e)+ 'heat, sunlight'

*gWhr,n.o+y+s > *gWhr,n.-yo+s *gWhr,n.-i+
'(in the) heat, sunlight'

*gWhr,n.-yo+s > *gWhr,n.-i-yo+s > Celtic *gWarn.-i-yo+s > *gWriyan.o+s
> OIr grían 'sun'
*gWhr,n.-i+ > Skt ghr'n.i-s 'heat, sunlight'

3
It applies only to between P and r:
gabhis.ák 'far down'

4
Before a plain P in the previous syl., x,>u and
before a plain P, R,x>uR in Latin (domo+ / domu+).
Before a plain P, any x(),>u in Slavic and Baltic.

5
A PIE change of
C-ry/rw/ly/my/xy/etc. > C-yy > C-i-y

Comparatives show C-nyos > C-i-yos (among other changes).

The retention of the old syllable boundary between whatever C and the
n>y>i puts the new V in its own syllable. Therefore, in Sanskrit I
saw i/u > +long between syllables unless followed by a syl.+tone.

This PIE rule also works for Cnw, etc., or else there would be no
change causing n/0 in 'river' or 'lynx' as:

daxnus . daxnwos . luugYsnus . luugYsnwos
daxnus . daxwwos . luugYsnus . luugYswwos
daxnus . daxuwos . luugYsnus . luugYsuwos

6
*xYer-xY-tor+ > Skt aritár- shows no i>i: since the accent follows i;
but arítra- should be *arí:tra- which indicates analogy. But that
doesn't happen in PIE; Skt. has x(),>0 between dental obs. between
syl. seen in