Re: Re[4]: [tied] Re: *a/*a: ablaut

From: Miguel Carrasquer Vidal
Message: 53576
Date: 2008-02-17

On Sun, 17 Feb 2008 22:12:02 +0100, "fournet.arnaud"
<fournet.arnaud@...> wrote:

>Thank you for taking the time to
>explain your system.
>It helps understand
>the way you think
>even though I may disagree.
>Arnaud
>
>I understood :
>a i u > e
>a: u: > o
>i: > e:
>===============
>
>That's not the system I propose. My proposal involves 3x2
>vowels in Pre-PIE (*a(:), *i(:) and *u(:)).
>
>As to PIE *o (when not derived from *h3e), it shows the
>following characteristics in Indo-European:
>
>1) it is reflected as /a:/ in open syllables in
>Indo-Iranian;
>2) it is not reduced to <ä> in Tocharian (*o > e, while
>*e/*i/*u > ä)
>3) it is not coloured by a laryngeal (*h2o > o, *oh2 > o:),
>a feature that it shares with *e:.
>
>Miguel
>====================
>My spontaneous linguistic objection
>is that your system makes long vowels
>[a:] [o:] > [o] as frequent
>as short ones.

No, short vowels are more frequent, especially in the verbal
system. Where it is hard to find a short vowel is in PIE
root nouns, which show almost exclusively long vocalism (*o
[as in *pod- "foot", G. *péd(e)s, or *nókWts, *n(e)kWtós] or
*e: [as in *k^é:rd, G. *k^r.dés "heart"]). The only example
of a short-vowelled root noun may be *sem- (G. *smés) "one",
but even in that one there are variants with long vocalism
(both *som- and *se:m-).

>I conclude that *o was originally a long vowel **a: (or
>**u:). The loss of length (except partially in Indo-Iranian)
>is due to the fact that there was no short /o/ in the
>system. Something similar happened much later in
>Indo-Iranian with the reflexes of PIE *Vi and *Vu, which in
>Pali are long /e:/, /o:/ in open syllables, short /e/, /o/
>in closed syllables, while they are short /e/ and /o/ in
>most other Prakrits.
>
>-is due- ?
>It sounds a bit adhoc.
>I'm really sceptical that a long vowel could shorten
>in a system where the general direction is
>more and more long vowels arise
>because of the fusion with decaying laryngeals.

The laryngeals didn't "decay" until much later.

>The reflex of long **i: is PIE *e:, which remained distinct
>from PIE short *e. It appears as "fundamental *e:" in a
>number of words like for instance *yé:kWr. "liver" or the
>"Narten" verbal roots such as *ste:u- "to praise".
>
>In general, non-laryngeal length in PIE appears in:
>
>1) "fundamental e:" (in my view, from pre-PIE *i:;
>fundamental **a: and **u: appear as *o).
>
>How many words like *ye:kwr exist ?

A number: *yé:kWr.(t), *yikWnós (*yékWnos) (*lé:pr.(t),
*lipnós) "liver" [Skt. yá:kr.t, yáknas], *pé:r(r), *pr.nés
"house" [Hitt. pi:r, parnas], *k^é:rd, *k^r.dés "heart"
[Hitt. kir, [kardiyas]; Arm sirt], *d(h)é:g^ho:m,
*d(h)g^hmós, coll. *d(h)g^hó:m "earth" [Hitt. te:kan,
tagnas], perhaps *g^hwé:r "wild animal" [Grk. thé:r, Lith
z^ve:rís], etc.

>How can we be sure that this is not a innovation ?

Because they are irregularities. Irregularities are
generally archaisms, not innovations.

>How do you account for latin jecur with short e ?
>I see no reason why it should have shortened.

Latin shows a marked preference for the oblique stem in
cases like this (iecur, cor, ferus, etc.).

>2) Lengthening of PIE *e and *o in the sequence *VCF#
>(vowel-consonant-fricative-morpheme boundary). This worked
>after the merger of *a/*i/*u to *e, but before the complete
>loss of unstressed *& (shwa) [*& lengthens to *o]. We find
>it mostly in the Nsg. of non-neuters (*hák^mo:n, *pó:ds,
>*p&2té:rs, etc.), in the NA neuter (*udó:r(&2), *nebhós&2,
[should be *p&2té:r and *nebhó:s&2]
>etc.) and in the s-aorist *bhé:rs-t.
>
>3) Vr.ddhi of short *e, *o in thematic derivatives (e.g.
>*o:ujóm "egg", *me:msóm [this could also fall under (2)]).
>
>I disagree with both reconstructions.

Which ones?

>Do you have other words ?

With vr.ddhi? Yes. The process was still productive in
Sanskrit.


=======================
Miguel Carrasquer Vidal
miguelc@...