On Tue, 28 Jan 2003 12:07:35 +0100, Piotr Gasiorowski
>While many of your examples support the theory, some look questionable
>to me. *h1dont-, for example, clearly had the oblique stem *h1dn.t-,
>unambiguously attested in groups as far apart as Indo-Iranian and Germanic
>(or Celtic), while the alleged alternation *h1don- ~ *h1dent- does not seem
>to me to be supported by any real data.
I believe it is supported by Latin dens. Within a static paradigm
like *h1dónts, G. *h1dénts, Latin had a strong tendency to confuse the
genitive in *-s with the nominative, and to create the paradigm anew
based on the oblique stem: e.g. hiems, dens, pe:s, all nominatives
from original G. *ghyéms, *h1dénts, *péds.
Of course the oblique stem *h1dn.t- is a reality, and has to be
explained. My reconstruction *h1dónts, *h1dóntm., *h1dénts
presupposes an animatized collective. Other possible patterns are
hysterodynamic *h1dénts, *h1déntm., *h1dn.tés, or the normal present
participle construction *h1édonts, *h1édn.tm., *h1dénts.
I don't have a good answer at the moment as to how these different
forms interacted to result in the attested Indo-European forms.
>Do you regard oblique allomorphs like *dreu- ~ *dru- (from *doru),
>*h2jeu- ~ h2ju- (from *h2oju), etc., as analogical? Analogical to what?
The word for "tree" had at least two different shapes: **dá:r-un (n.),
G. **da:r-ún-âs, and a variant with short root vowel **dár-ûn (n.), G.
**dar-ún-âs. These regularly develop into:
*dér-wor, *dr-éw-os (without labialization of /n/: *dr-ún-os).
We have: Skt. da:ru, G. drún.as or dró:s, Loc. da:runi, Ins. drúna:
Grk. doru, douros < *dorwos, Att. G. dóratos < *dorn.tos
Slav. *derwo-, Gmc. triu (*drew-), etc.
In my previous message I was only discussing the paradigms resulting
from original pre-PIE **a: (> PIE *o, "zero-grade" *e), and I
consciously left out those forms that I believe go back to pre-PIE
**u: (> *o) and **i: (> *e:), both with zero-grade zero. A good
example with **u: is for instance:
**pú:nt-ax-z > *póntoh2s
**pú:nt-ax-m > *pónth2m.
**pu:nt-áx-âs > *pn.th2ós (amphidynamic)
Another example may well be **xú:y-u-, obl. **xu:y-ú (> *h2óy-u-,
Miguel Carrasquer Vidal