Nasals/ Editing problems
witzel at HUSC3.HARVARD.EDU
witzel at HUSC3.HARVARD.EDU
Sun Feb 11 18:51:52 UTC 1996
On Wed, 31 Jan 1996, Birgit Kellner asked:
>>> I would like to have as many opinions as possible on
Sandhi-conventions in critical editions of Sanskrit texts, especially on
what to do with nasals ...<<<
There have been only two answers -- a surprising fact on this list. How
is it possible that we mindlessly just follow the features of the printed
editions of the ``Benares/Poona tradition`` of the last 150 years or so,
features which have also been copied in most western editions?
This is not the solution, it is the problem. B. Kellner`s case is a
fairly standard one in better MSS *outside* the Poona/Benares tradition:
>>> the manuscript I am currently using, for example, generally ...
assimilates nasals in pausa, when the following word starts with ca, ta
or va (i.e. ki~n ca, kin tu, kim vaa). It also sometimes writes ma in
pausa, when the following word begins with bha, ba or na, but never with
any other consonants. <<<<
This is one of the ways to write such sandhis (not pause, incidentally;
or do I misunderstand her here?) Poona/Benares editions of course would
write anusvaara (M) everywhere. Part of the indological tradition (in
India and elsewhere) even writes Anusvaara (M) in cases such as
``daNDa`` which you can find, especially in lithographs and Nirnayasagara
prints as ``daMDa`` -- thus even inside words, and compounds, in internal
sandhi... The easy-going practise to write M in *external* sandhi
(graamaM gaccha!) instead of homorganic nasal (as good Vedic MSS and
many local traditions do) is as old as the oldest Skt. MSS from
Hsinkiang... (and the sometimes resulting Pali style pronunciation of
Anusvaara as |ng| is scolded in the medieval Siksa literature).
((( And I do not even touch the distinction between anusvaara (M) and
anunaasika (&), due to Panini`s not so clear distinction of the two. Thus
you can see haMsa or ha&sa, but again, Poona/Benares usually prefer the
easy-going haMsa. The case is altogether more complicated in Vedic
recitation (see anunasika, IIJ 25) and Vedic MSS-- and different in every
school (zaakhaa)-- but as usual, obscured by the current editions...)))
>>>>B.Kellner: >Is one supposed to replace ma at the end of a word, just
as the scribe seems to have done, by the homorganic nasal, ... is one
supposed to carry out this replacement only within compounds?<<<
See above, and below. The question is which ``norm`` to follow.
Prof. v. Simson wrote:
>>>>But since the scribes are not consistent in their use of sandhi, I
think it is best to standardize the orthography and to follow Panini's
rules as far as sandhi is concerned. This makes it also easier for the
user of your edition. You may describe the actual practice of the scribes
in your introduction or you can give the writings of the manuscripts in
the critical apparatus. <<<
This is reasonable practice. However, I think the this practice *is* the
problem. At any rate, the case is more complicated. Once you start
comparing MSS from various areas of medieval India you notice clearly
defined local styles: the Kashmirians have one ``orthography`` of
Sanskrit, the Newars of the Kathmandu valley another, the Gujaratis,
Oriyas, Tamils, Nambudiris still another, and so on... Apart from
occasionl remarks (e.g. : this is Dravidian ``ra`` for vowel ``r``) the
problem has hardly been noticed. (I think I have referred to it here and
there in articles on the Paippalada Atharvaveda; or see Lubotsky in IIJ
25 for Maitrayani Samhita /Gujarati practise which is surprisingly
different from what we learn in school; cf. also Prof. Rao`s example of
pronunciation of vowel R in sandhi ).
I say ``problem`` because we have to KNOW the local NORM in order to
understand what kind of writing mistakes might occur (of course you also
have to know the local paleography; --- for examples of all of this, see
articles in: Journal of the Ganganath Jha Research Institute, Vol. 29
(1973) 463-488; Vol. 32 (1976), 137-168 (both with innumerable
misprints, for correct copies ask me!); MSS 44 (1985), 259-287, IIJ 25
(1983) p. 19, IIJ 29 (1986), 249-259, and: On Kashmiri MSS and
pronunciation, in: Studies on the Nilamata-Purana, ed. by Y. Ikari,
Kyoto 1995).
If we paint over the local style by applying Paninean norms we will never
know, especially if the ``orthographic peculiarities`` are merely
summarized in the introduction. I recognize that it may be too much to
put every anusvaara (M ) found in your MSS (instead of correct n,m,N etc.
) into the variant readings. A middle path has to be found here.
But it is an altogether different thing if we as editors while
normalizing hide the fact that, say, the Nepal scribes traditionally
write cch- even at the beginning of words, that the Kashmiris hardly
employ Visarga (only in pausa!) but a lot of Upadhmaniya and Jihvamuliya,
and z, S, s before the same consonants, even in classical MSS ... etc..
If you do not draw the attention of the reader to these facts in your
edition by actually printing --devas sam-- and the like, people will
never know -- because more than 100 years experience teaches that
virtually no one reads introductions (note the case of Geldner`s Avesta
or S.P.Pandit`s Atharvaveda ...) and even *if* people do so, they quickly
forget (``another orthographic peculiarity``: we want to read proper
Sanskrit...). Well, Skt. was not, at any place, so proper Paninean
throughout history as the grammarians think and want us to believe...
In short:
*** preserve the local norm (once you know it)***
This helps to understand many mistakes made by (later) scribes and thus
helps you to reconstruct the text (archetype or not, see last summer`s
unfinished discussion ).
I therefore agree with Prof. Vidhyanath K. Rao:
>>> When standardzing sandhi conventions, it may be worthwhile to record,
perhaps even follow the main manuscripts, when one of the several options
allowed by Panini/vartikakara are generally preferred.
It seems that "deferring to common practice", with the attendant
elimination of indications about dialectical variations would lead to
wrong impression by those who consider the printed editions to be the
only "scientific" evidence for historical purposes. <<<
It is indicative of our editorial skills and practice that a
Sanskrit-loving prof. of Mathematics has to tell us what to do! This HAS
been the problem in Indologyy, starting with our very conscientious and
productive predecessors in the 19th century (Keith`s AA, quoted by Prof.
Rao, included. Keith, just like the others, also standardized).
Problems arise, of course, when we get MSS from various parts of India
(not to speak of recitation, see again Prof. Rao`s post) . Which
``norm`` to use if we do not know the locality of the author or if there
is neither a single author nor a single location of the authors/
redactors ( Epics, Puranas... though even here, much more could be found
out if we would use some of the above...)
But even then, we may use the fact of MSS stemming from various areas to
our advantage. Take a look at Kuiper`s Gopalakelicandrika which we both
discussed before publication. It seems that we have a Gujarati author but
the MSS also show Tamil, Nandinagari and other peculiarities... (A
similar case is that of Schokker`s Padataditaka: from early Kashmir but
with Malayalam MSS and both peculiarities...) This helps to understand
the history of transmission -- and will help the editor to make
intelligent choices in editing...
In my opinion, editors too easily fall back on what I call the
Benares/Poona tradition of the last few centuries...
(which has ``invaded`` Gujarat, Orissa, Kashmir etc. only during the
last century!). They make it THE NORM for editing Skt. texts, --- down to
details such as use of avagragha and double avagraha which I have not
seen in any MS so far (anybody? -- Note also the ``traditional``
N.American transcription of ai, au!) We should be aware of the fact
that, down from our oldest Skt. MSS (roughly Kushana time, from
``Chinese Turkestan``/Hsinkiang/Xinjiang), there have been many ``norms``
and that the one we follow now is just *one* of them, and artficial in
places (just think of the word division in printed Nagari :: word
division in Roman transliteration :: lectio continua in the MSS; and this
even obscures certain facts of sandhi!).
Another problem which we face, e.g. with the Kashmirian version of the
Paippalada Atharvaveda, has been alluded to by Prof. v. Simson:
>>>> Franz Bernhard's edition of the Udaanavarga, Vol.I, Goettingen
1965, where the orthographical variants are kept apart from the others
(and more important ones) as a category of its own. <<<
There are, as I mentioned above, the innumerable cases of M :: n,m,N
which seem to be useless in a list of variants. However, even these can
become important in a tradition which is limited to a few MSS, say in the
case of the Paippalada Samhita. Here, it would be worthwhile to record
them as the can provide clues about the stemma and the various local
lines of transmission.
Secondly, some texts (as in the Atharva tradition, especially the Kash.
Paipp.) bristle with local mistakes based on pronunciation which render
the text unitelligeable in places. They may very well be recorded in a
special appendix. However, the problem even then is how to distinguish
between ``useless`` orthography (or recitation) and ``important``
variants. In short: record both, in one way or another: our successors
will be grateful to you. In the words of a venerable octogenarian Indian
colleague. B.R. Sharma, who has, single-handedly, edited most of the
Samavedic texts (a new SV edition with comm. to come out in HOS):
``I give all the variants (since I have the MSS), and you have all the
time and can figure out where to correct me.``
If only more scholars would follow this dictum!
M.Witzel, Sanskrit
Harvard U.
witzel at husc3.harvard.edu
More information about the INDOLOGY
mailing list