Re: [tied] Morphology (10/20)

--- In cybalist@..., Miguel Carrasquer <mcv@...> wrote:

> On Sat, 31 Aug 2002 00:36:33 +0000, "Glen Gordon"

<glengordon01@...>

> wrote:
>
> >While you feel the need to reconstruct a 9x9 "grid" system of

personal endings,

> >what language reflects this grid at all?? We can see this is

immediately

> >misguided.
>
> Of course no (IE) language reflects such a 9x9 grid, but that's the

whole point.

> If such a grid were actually attested, the origin of the middle

would have been

> obvious. To use an example recently discussed on this list, if we

take the

> modern Romance languages, and we see plural forms such as *muros in

the West,

> *muri in the East, is it really misguided to reconstruct a "2-

dimensional grid"

> *muri, *muros for the ancestor of Romance? Of course not: that's

what it had.

> Likewise, if in the middle 3pl. we find *-ntom, *-ntor(i), *-nto

(i), *-ntodh(i),

> the possibility that all these forms represent the debris of an old

paradigm

> where each form had its proper function is worth investigating.
>
>
> =======================
> Miguel Carrasquer Vidal
> mcv@...

If we accept this argument, why doesn't it apply it to the
centum/satem reflexes of PIE velars (k/s etc)? In the 25 odd years
I've been away from linguistics, I learn that in what used to be
causeless vacilation in the alternations of the ablaut vowel, its
various alternations have now been back-assigned to forms of eg. the
original cases and numbers of PIE. Why not do a similar "back-
assignment" of the reflexes of velars, assuming on original
alternation k/c^ etc, with later generalization, just as -s and -V
were generalized in West and East Romance?

Torsten