Retroflex sounds

[Bh.Krishnamurti]

At 02:51 21/06/98 +0100, you wrote:
   In any case the examples adduced so
>far also seem to contrast with the Indian example in that they are
>not a system involving a whole series of phonemes of the language.
>I would imagine such isolated features would, for the sake of economy,
>tend to get simplified out of the system (even if it is the case that
>they are phonemic). Which brings me to the following: isn't a unique
>feature of the Indian linguistic area the stability of retroflexes?
>If proto-Dravidian had them, that's something like 5000 years of retro-
>flexion. This doesn't seem to be the case in the other examples where
>the retroflexes seem to be (relatively) ephemeral intermediate steps
>in sequences of phonetic changes. Is that really the case?

Dear Indologists;

This is the most crucial argument. Let us make two assumptions:

Assumption 1: A non-inherited phonological feature can develop internally
within a language,  not induced by contact with other languages.

Assumption 2: A non-inherited phonological feature can develop internally
within a language induced by contact with other languages, which genetically
possess such a feature.

By A1, several languages may have developed retroflexion as isolated
instances,like Sweedish, Norwegian, certain Italian dialects, etc., first
allophonically and later perhaps phonemically. They only change the
distribution of certain phonemes marginally but not the whole phonological
system; mid-Western American English retro r in girl, world, word, is also
part of this phenomenon. The case of Australian languages is different,
since they have retroflexes as part of their genetic make-up.

By A2, through prolonged contact a language naturalizes a foreign
phonological feature by integrating it its system. Sanskrit retroflexion is
one such phenomenon. Look at the chronological profile.

Stage 1: Indo-Iranian and Slavonic share a phonetic rule (later called
ruki-rule in Sanskrit), i.e PIE dental s becomes retracted to alveopalatal
postion as s' or s" (like English sh)when preceded by high  vowels, r, .r
and k, all of which carried the feature [+hi]. This is a reconstructable
allophoneic change at an undivided stage.

Stage 2: Both Indic and Iranian perhaps had the same pronunciation, but when
Indic seperated,  Iranian retained s" (s with wedge=sh) but Old Ind speakers
phonetically changed this sound into a retroflex. At this point the
different phonetic representations betwen Iranian and Indic need to be
explained. This could be the impact of initial contact with Dravidian.

Stage 3: Maybe simultaneously with Stage2 was the developemnt of.t, .th, .d
and .dh from internal sandhi of *s"t, *s"th, *z"d, *z"dh; the dental stops
first became retroflexes phonetically after s" which then changed s" to a
retroflex. The articulation of phonemic s" (<*s *s') in Pre-OInd could have
been retroflexed by then. OI v.ri.s.ti, ni:.da, u:.dha are thus accounted
for. These do not account for intervocalic .t and .th which  deveoped later
during the Middle Indic period.

Stage 4: A precedig r or .r changed a dental to retroflex; thus we get
na.ta- from n.rt-, vika.ta- from vik.rta-; we notice this happening already
by the time of As'okan inscriptions. H.Hock says that the retroflex
representation was more in Northwest and East than in West and Central. This
distribution broadly corresponds to the concentration of Dravidian (and
Munda in the East)speaking peoples of that time.

There could be a few  missing links in the above scenario.I am not a
Sanskritist or a IE-ist. Such a transformation of Skt phonological system is
possible only under a contact situtaion. And retroflexion is one  of the so
many other features of Middle Indic both in phonology, morphology and syntax
that look to a Dravidian source. A phonlogical rule shared by both IA and
Drav is the emergence of the favourite syllable type (C)V:C or (C)VCC to the
elimination of *(C)V:CC.

H.Hock's argument that both IA and Drav developed retroflexes on parallel
lines is not defensible, because Dravidian had both *_t and *.t basically in
a large no of morphmes. It is true that in sandhi also _t and .t develop
from l+t, and .l+t. But Middle Indic retroflexes are a result of sandhi
alone and became phenemic later to fill a gap in the phonological system.
The Drav sandhi may have triggered these but such a mysterious convergence
without contact and large scale bilingualism does not make sense.

Aryan Migration from India westward. One can go from stage 1  to stage 2 and
not the other way round. To be explicit, we cannot  derive niz"d- (Indo-Ir.)
from ni:.da (old Indic).  It is impossible to a reverse a merger of that
kind by a subsequent sound change. If Aryans went westward, how do we get
retentions of an older stage in Avestan, and still older ones
reconstructable to Indo-Ir-Slavonic. Professor Witzel, I remember has
mentiond this point,  but I did not follow the thread later.

Regards, Bh.K.


Bh.Krishnamurti
H.No. 12-13-1233, "Bhaarati"
Street 9, Tarnaka
Hyderabad 500017
Telephone:40-7019665