Indian musical theory

[l.m.fosse at easteur-orient.uio.no]

>Lars--
>
>I have been away for some time and just found your message.  You might have 
>your student contact Prof. Stephen Slawek, Music Dept., MBE 3.210, University
>of Texas, Austin, Texas 78712, USA.
>
>Richard
> 

Thank you! I have already got another address, which I gave to my student,
but I'll let him have this one too.

Since you asked to keep you informed about the progress of my work, let me
use this opportunity to tell you that everything is going according to plan
so far. I have begun to write the first couple of chapters of my thesis
(which will partly deal with the history of statistics in Sanskrit
philology, partly discuss methodological problems). The next year will be
crucial, because I shall then have to do my pilot study. If the pilot study
breaks down due to practical of theoretical problems, I'm in deep trouble.
I not, the future still looks reasonably bright. I plan to participate at
the IX. Sanskrit World Congress in Melbourne, perhaps I will see you there!

Best regards,

Lars Martin 

P. S.: In case you are interested, I give you the abstract of the paper I
would like to present in Melbourne (if they accept it!)

               IS THERE A FUTURE FOR STATISTICS IN INDOLOGY?
                            Lars Martin Fosse

Statistical methods have been used in philological studies  since the
1880's . Within the field of modern languages  in particular, a great
number of studies based on such techniques have been published during the
last hundred years.  Statistics have been very successful in clarifying
authorship problems. Although these techniques  can never prove an
hypothesis, they give a mathematically defined estimate of  how probable
(or improbable) a given hypothesis is.  Multivariate techniques enable the
scholar to evaluate a large number of criteria and to draw conclusions with
a high degree of probability even on the basis of relatively small samples,
as is shown in O'Donnell's study of The O'Ruddy.  
        Despite the obvious usefulness of statistics in studying and
clarifying a number of textual, linguistic or philological problems, the
method has never really gained acceptance in indological circles.  In the
period  from 1880 till 1905, a number of studies were published  by  among
others H. Edgren, C. R. Lanman and E. V. Arnold where descriptive
statistics were used, the most well known work being Arnold's Vedic  metre
in its historical development.  The next major attempt at using statistics
in an indological context was made by Walter Wust, who in 1928 published
his Stilgeschichte und Chronologie des Rigveda.  He gives no mathematical
evaluation of the  numbers  he has compiled , and the conclusions are not
supported  by reference to statistical theory. The philological part of the
study was criticized  by Edgerton, but an informed statistical discussion
had  to wait until A. S. C. Ross published his  article "Philological
Probability  Problems"  in the Journal of the  Royal Statistical Society in
1950. In the Discussion on the Paper,  P. A. P. Moran, B. Babington-Smith
and J. Gillis offer solutions to the numerical material provided by Wust
based on proper statistical principles. Wust's intuitive conclusions were 
by and large supported by  the statisticians.  It is, however, debatable
whether Wust actually managed  to say something about the relative age of
the Rigvedic books, or  whether he  was  merely able to group them
according  to certain numerical principles. Pavel Poucha's attempt in the
'forties to deduce the relative age of the Rigvedic hymns on a hymn-by-hymn
basis has  already been rejected by Gonda on philological grounds. It
should be added that the numerical principles used by Poucha can at best be
described as naive. But his basic perception that the Rigvedic books all
contain both old and young material is most probably correct, and  since
some books  probably contain more ancient material than others,  this may
create a false impression of  books having been edited at different times.
        In the fifties and sixties a small number of works  appeared where
certain statistical figures were presented. Most notable are the three
articles by R. Morton Smith on the stories of Amba, Sakuntala and Nala in
the Mahabharata.  Significance testing was not applied.  It was only with
T. R. Trautmann's study of the Arthashastra  that more sophisticated
methods were introduced.  However, the statistical part of Trautmann's work
was unfavourably reviewed  by Sternbach, who objected  to the limited 
amount of discriminators used  by  Trautmann. Trautmann's study is
certainly not devoid of merit, although a more comprehensive choice of
discriminators would  have  given it greater significance.  As it is,
Trautmann's work needs confirmation. Since Trautmann, no major statistical
studies have been published within the field of Indology apart from M. R.
Yardi's work on the Mahabharata and the Ramayana.
        The general impression conferred by earlier statistical studies 
within the field of Indology is one of numerical naivity. The techniques
used are quite rudimentary, and there has been no proper discussion of what
the numbers actually mean. Establishing a set of numbers (with or without
significance testing or the use of other techniques) is only part of the
job. Interpreting the numbers is a different matter.  This calls for a
milieu of scholars well versed in statistical theory as well as Indology. 
It also calls for the establishment of large  electronic corpora of
Sanskrit texts available to all. Such corpora already exist within  other
philological fields, e.g. modern languages, biblical studies, ancient Greek
etc.  They serve as the basis for linguistic, literary and  stylistic
studies.  Furthermore, there is a need for a uniform format for the
registration of Sanskrit texts.  It is particularly important that
compounds be  dissolved. A  common system of markup (tagging)  for
grammatical and stylistic features  would be a great advantage.  This would
facilitate the exchange of files enabling scholars with different
objectives to  profit from the work of others.  It would  then be possible
to conduct large scale statistical studies of Sanskrit texts and to develop
statistical techniques adapted to the special requirements of Indology. 


Lars Martin Fosse
Department of East European
and Oriental Studies
P. O. Box 1030, Blindern
N-0315 OSLO Norway

Tel: +47 22 85 68 48
Fax: +47 22 85 41 40

E-mail: l.m.fosse at easteur-orient.uio.no