Retroflexion in IA

[George Thompson]

My apologies for not responding earlier. During his last series of storms,
very violent things they were, Indra struck my modem with his vajra, and
utterly destroyed it [I don't know what sins I may have committed against
him, but be assured that I have been performing all of the relevant
prAazcittas]. I am only now starting to get myself back on track [I've also
switched from Mac to Windows 95: not as easy as I had hoped].

Perhaps a Dravidianist will step in and help me, but in the meantime here
is a response to Vidhyanath Rao's observations:

>How did the bilingual children cause the splitting of `n'? In
>Dravidian, dental and alveolar n's are allophones of the same phoneme,
>contrasted with the retroflex. To judge by the situation in South
>Dravidian languages, dental n occurs word initially and before dental
>stops, alveolar n elsewhere. It does not matter whether you posit
>bilingual children or pidgins or creoles. Why would they split what is
>one phoneme in >both< languages involved, while preserving the dental
>where it could not occur in Dravidian?
>
I don't see the problem here. Retroflexes, alveolars, and dentals, at least
phonetically, are nothing but a series of gradients formed by contact of
the tongue to various points along the roof of the mouth, ending at the
teeth [or, if you will, the root of the teeth]. In proto-Indo-Iranian what
we call dentals could have been formed by contact of the tongue with a
relatively wide range of points, well into the range that we would call
alveolar. Articulation at the extremes of this spectrum may have seemed
unusual pronunciation, but since there was no phonemic opposition, such
variation would not have caused any significant disturbance to the system.

But when even rudimentary contact first occurred between IA speakers and
Dravidian speakers, IA speakers must have quickly noticed that Dravidian
speakers observed a phonemic distinction between dentals-alveolars on the
one hand and retroflexes on the other. Such recognition, in my view, would
have been enough to trigger a similar though not necessarily an identical
segmentation in their own language [that is, in their own phonemic system].

Bilingual speakers work with two phonemic systems. Prolonged contact
between two phonemic systems will inevitably lead to some amount of
convergence, I think. But the two systms will remain distinct. So the
different sorts of permitted segmentations that one sees in IA and in
Dravidian would not have been a problem. Bilinguals switch back and forth
like this all the time. This is not a problem. One expects the two systems
to remain distinct.

>And finally, a practical question: What did the children speak to their
>parents? IA to one and Dravidian to another? In such cases, as long as
>given person speaks the same language to the child at all times, modern
>research indicates that the child will pick up both languages well,
>speaking exactly like their model.

I think that they probably spoke both to both parents [both of whom became
increasingly bilingual themselves, of course], depending on context. I
assume that extensive bilingualism gradually increased as IA speakers
migrated deeper into the subcontinent. As far as I can tell, there is
little disagreement that eventually there was extensive convergenece
between Dravidian and IA [which is why we can now talk about a South Asian
linguistic area]. The real dispute, it seems to me, revolves around *when*
this convergence is assumed to have taken place, not *whether* it took
place. Perhaps V. Rao and I actually agree in this sense: significant
convergence may have been relatively late.

As the list well knows, Michael Witzel has been working on identifying with
greater precision the many stages of development of Vedic language as well
as Vedic culture. He informs me that he has measured increasing
retroflexion as we move from the family books of the RV to the later books.
It is reasonable to assume that future research will disclose other
evidence, syntactic and lexical, of the sort of convergence that we are
talking about. In any case, I think that the issue of *when* retroflexion
became a phonemic process in IA is still an open one. Early Vedic does not
seem to present a lot of evidence for convergence with Dravidian. As far as
I can tell, Deshpande's thesis remains viable, though certainly I would
agree with V Rao that it is open to debate.

>This is likely to be the explanation for the AA passage mentioned
>elsewhere: If the pronunciation mo-Su-NaH was limited to rcs, while in
>ordinary speech people said mo-su-naH, then the point would have been to
>preserve the original pronunciation instead of using the daily
>pronunciation.
>
Agreed. But the saMhitA text represents "the original pronunciation" only
within the context of surviving Vedic traditions of recitation [and who
knows what the recensions that we have lost read?]. Diachronically, the
reading mo-Su-NaH is clearly *later* than the "daily pronunciation"
mo-su-naH. It seems to me that those who objected to the retroflexed
version were objecting to a mechanical recitation tradition that differed
significantly from standard daily usage [i.e., . According to Deshpande the
Ur-RV had 'mo-su-naH', rather than what has been handed down in zAkalya's
recension. This still makes sense to me. Why assume that the zAkalya
recension represents the Ur-RV?

V. Rao continues:

>But many of these features don't appear till much later than RV. On the
>other hand, retroflexion goes back much further. In fact, in RV there
>is less difference between internal sandhi and external sandhi in this
>matter. How do you explain this? [This is also one of Hock's objections.]

Again, I think we may be talking about a recitation tradition, not a
linguistic one.

Enough for now. More later, if there is interest.

George Thompson

p.s. Would Indologists really prefer to see this sort of discussion removed
to an Indo-Iranian list?