Indo-Aryan Invasion (focussed discussion)

[George Thompson]

For the past week I have taken every free moment I've had to re-examine
Hock's 1996 article [in IDEOLOGY AND STATUS OF SANSKRIT: CONTRIBUTIONS TO
THE HISTORY OF THE SANSKRIT LANGUAGE. E.J.BRILL], in light of some of the
earlier literature available to me. I would like to confine my comments to
the problem of retroflexion, and in particular to retroflexion in the RV.
This is not to say that other issues [e.g., syntactic and lexical evidence]
raised in Hock's article are unimportant. It is just to take one issue at a
time.

I should say at the outset that for a long time I have been sympathetic to
Madhav Deshpande's views re RV retroflexion: that it may well be that the
retroflexion which we see in the only recension of the RV that we have was
introduced into the text by "Dravidianized oral transmission of the text"
[see most recently Deshpande's article, "Vedic Aryans, non-Vedic Aryans,
and non-Aryans: Judging the Linguistic evidence of the Veda", in the same
volume as Hock's article, cited above]. At the AAR panel in San Francisco
which Edwin Bryant referred to in his first post on this thread, I briefly
mentioned my view of RV retroflexion,in a comment after my paper, though
without mentioning Deshpande by name. [I hope that he does not mind my
raising this issue on the list!]

In no way do I consider the question settled, and I have brought it to the
list in order to see if the question could be illuminated [which I would
hope], if not resolved [which I do not expect].

I think that there is general agreement that retroflexion in the RV is
rather incipient, that there is much less of it in the RV than in classical
Sanskrit. However, it would be nice to see this general impression
confirmed by something specific. Perhaps the statisticians on the list
could easily produce a retroflex-to-total letter count in every book of the
RV and compare that with the same in the MBH [just kidding].

It has also occurred to me, after soliciting minimal pairs in Norwegian
from Lars Martin Fosse, that there actually are not very many minimal pairs
that one can produce from the RV itself. For example, the pair frequently
cited by Hock:

        pAta 'flight'   vs.     pATa 'portion'

is not attested in the RV [in fact, neither word is attested there].

I'd be grateful if others on the list would call some RV pairs to my
attention. There aren't many [e.g., the pair kuTas, apparently a proper
name at 1.46.4, vs kutas 'whence'].

Another puzzling detail, often mentioned in the literature I think, is that
there is only one word-initial retroflex consonant in the RV: SaT [and its
derivatives]. Historically, /S/ appears to be an allophone of /s/ by the
ruki rule and in other contexts.

It is not clear to me to what extent retroflexion in the RV is a phonemic
process, as opposed to a purely allophonic one triggered by mechanical
application of recitation techniques, such as those referred to by
Deshpande. Perhaps the learned of the list can enlighten me about this
problem, which I acknowledge has been treated already not only by those
already cited but also by previous scholars such as Emeneau and Kuiper [I
have also consulted Vedic Variants, vol 2, and Wackernagel, vol 1].

Perhaps a re-examination of this problem would be of use to others as well
as myself. I have no thesis. I am just trying to figure things out. If
these matters have been dealt with in literature that I am unfamiliar with,
I would be grateful for relevant references, and will no longer trouble the
list about them.

Best wishes,

George Thompson