what's in a root? (was Re: IE and Semitic roots).

[Jacob Baltuch]

I have a few questions regarding Sanskrit roots and I'll be grateful for
any enlightment:

a/ Is there a specific name for the unmodified grade of a root
   (by opposition to guNa or vRddhi grade) or do we have to continue
   using "zero grade"? I don't care if it is a Sanskrit neologism or
   a traditional term. It is just esthetically unsatisfying to go
   "zero grade, guNa, vRddhi"...

b/ What was the way grammarians viewed their roots? Were they taken to be
   a convention for the sake of the convenience of description (e.g. is
   the statement ever explicitly made that a grammarian is free to posit
   roots any way that simplifies his description) or are they deemed
   to have some sort of objective existence in the "competence" of
   the speakers and thus debate about the "true" form of the roots
   (beyond debate around the simplicity of a description) is legitimate?

   One related question is the use of bare roots in composition
   (quite often with some accomodation)

     kaala-jJa, sarva-dRz, vighna-kRt, sarva-jit, satya-jit,    maargha-stha,
     jala-ja,   jala-ruh,  jala-ruha,  agree-ga,  hRdayaM-gama, graama-gamin,
     kSiira-pa, ahi-hata (but here -hata could be ppp?), etc.

   Is that a usage which arose precisely under the influence of the gram-
   matical analysis or is that a completely independant and native mech-
   anism which only by coincidence uses forms which are often similar to
   the abstract roots posited by the grammarians? Are there any other uses
   in the language (as opposed to the grammatical meta-language) for
   (forms which resemble) bare roots?


Also Vidhyanath Rao wrote:

>These is where I get lost in laryngeal theory. Treatments of
>laryngeals do not get into the nitty-gritty. What we need is a
>history of Sanskrit starting with the `new look PIE'. [Does the new
>Mayhoffer have anything on this?]

Maybe what I'd like to hear of is a reconstruction/history of Indo-Aryan,
Iranian and Indo-Iranian done exclusively from Indian and Iranian hard data
just to see what that gives. Since Indo-Iranian is basically the
only universally accepted wide grouping in the IE family shouldn't one
have taken advantage of this? Has anyone? Also people should try to put
some order into that messy poorly understood business of what happened from
PIE to Proto-Indo-Iranian which is of course connected with the question
of the relationships betweent the ten or so branches of IE which nobody
seems to have any clear idea of. While not universally accepted I've often
heard the statement that Greek and Indo-Iranian (with maybe Armenian) might
form a natural (that is a genetic) grouping within IE and maybe one could
try to proceed to a reconstruction of "Indo-Greek", or "Indo-Armeno-Greek"
as the case may be, and see what happens. In any case it seems to me that
it is more prudent to work from data to conjecture than the other way round,
but maybe I just don't get it.

I guess what I'm trying to say is that I don't see why a history of Indo-
Iranian can't be written which will last longer than the currently accepted
version of PIE. But then, it might just be me.

>Roots ending in e/ai are really roots ending in aa, which show a
>glide `y' without/with a shortening of the preceding aa in the
>present stem. I don't know about o/au. Glide v is not that common
>in Indo-Aryan between a's. [Are there really roots ending in au?]

Well, all the forms of dhaav ('to run' but also 'to wash') I know or
have access to right now, seem to be compatible with a root dhau given
internal sandhi rules. On the other hand the ppp dhauta (when in the
sense of 'wash') does _not_ seem to be derivable from *dhaav-ta which
seems to give (the incorrect) *dhaa-ta...

Can someone explain why positing dhaav is simpler than positing dhau?
Is dhau ever given as an alternate form for that root anywhere?