I know some of the Pratisakhyas first-hand,
and my use of the word "naive" implies no offence. The Pratisakhya
authors were "naive" (as observers), since their phonetic observations were
clearly pre-scientific and mostly introspective. That said, they were
_first-class_ naive observers. Some of their ideas (such as the rigorous
distinction between the place of articulation and the active articulator, and
the establishment of consonant "series") were brilliant indeed.
They attained a level of descriptive precision that was unimaginable in
Europe until the second half of the nineteenth century. Nevertheless, the
Pratisakhya authors had to work without sound spectrograms, X-ray equipment,
palatograms and other things that make a phonetician's life easier nowadays
by enabling objective observation; also the Sanskrit anatomical
terminology, impressive as it was, was
not always sufficiently precise to render verbal descriptions unambiguous.
Modern phoneticians make great use of diagrams (showing the configuration of the
vocal tract), which often tell you more that several paragraphs of text; the
Pratisakhyas have none. The question remains, of course, to what extent the
prescribed pronunciation reflects the actual usage of the Vedic
dialects.
As for the "r question", the Pratisakhyas
are not entirely clear about minute distinctions such as "trill, flap or
approximant?", but they suggest some degree of retroflection or at least a
markedly postalveolar articulation for syllabic "r", which is said to be
pronounced with the tip of the tongue approaching the upper back gums.
Descriptions of non-syllabic /l/ and /r/ also show that while the former was
dental, pronounced "at the root of the teeth (dantamu:les.u)", the latter was
retracted to "behind the roots of the teeth (pratyadantamu:lebHyaH)", i.e.
alveolar (not excluding postalveolar, in my opinion).
If one were to make sense of the
Pratisakhya descriptions _together with_ phonological processes such as
retroflex assimilation (with regard to which /r/ unambiguously patterns with
the cerebral series), the following is a likely scenario:
The inherited *r phoneme was an alveolar or
postalveolar trill or tap with a tendency to be reduced to a
fricative/approximant in syllable-final positions. It patterned phonologically
with the dental series until the phonologisation in pre-Old Indo-Aryan of the
"cerebral" series, with which it shared its apical articulation and retracted
position (with respect to dentals, as in Swedish) -- sufficiently distinctive to
turn a following *s or *n into /s./ or /n./ by assimilation.
Perhaps full retroflexion in the cerebral
series (with the tip of the tongue curled back: "jihva:grena pratives.tya
mu:rdhani...") developed still later thanks to substrate influence (through the
identification of the retracted articulation with Dravidian-style retroflexion),
and as the subapical articulation of "cerebral" obstruents and nasals became the
rule, /r/, or at least some of its allophones, followed suit. Retroflexion would
have been functionally desirable in the partly vocalised syllable-final
allophone and in syllabic "r", since its acoustic effect was to emphasise the
rhotic character of the sound (also American /r/ tends to be clearly retroflex
or "bunched" in those positions but merely postalveolar prevocalically).
Cf. Prakrit developments like *bHrta- > bHat.a 'mercenary', where retroflex
assimilation extends to oral stops.
Similar developments are not unknown to
English. For example, in the traditional non-rhotic dialect of Dentdale (in the
western part of the Yorkshire Dales), <short, cord> and <shot, cod>
contrast by having a different final consonant -- a semi-retroflex (apical)
postalveolar [t.] in <short> and a (laminal) dental [t] in <shot>.
In the north of England fully retroflex /r/, such as that of West-Country or
Irish English, would be unusual. If anything, there are, or used to be, uvular
or velar rhotics in Northumberland (the so-called Northumbrian "burr" or
"wharl"). In the north, the phoneme /r/ is generally realised as an
alveolar approximant/fricative or tap in prevocalic positions and a
postalveolar semivowel syllable-finally in those few Northern dialects that
remain rhotic (it manifests itself by r-colouring the preceding vowel).
Evidently, "Proto-Dentdale" postvocalic /r/ had become retracted enough to
affect the place of articulation of adjacent consonants before it
disappeared.
Piotr
----- Original Message -----
Sent: Monday, June 25, 2001 4:01 AM
Subject: Re: [tied] Sanskrit /r/
--- In cybalist@......, "Piotr
Gasiorowski" <gpiotr@......> wrote:
> If
the flapped or tapped /r/ was retracted but not subapical,
> and if the
retroflex variant was only slightly so, how would a
> naive observer have
known the difference?
I am curious why you consider the authors of the
praati"sakhyas
as "naive observers"? Is this based on first hand knowledge of
the
praati"sakhyas or an assumption? [This is a genuine question.]
If the
former, can you explain your evaluation further?
The authors of the
praati"sakhyas seem to have gone to great
trouble to ensure that their
students reproduce the vedas
"correctly", giving minute details as to what
they considered
to be correct and incorrect pronunciations. This is
hardly
what I would associate with "naive observers".