The best transliteration scheme for Sanskrit etc.

[Peter_Scharf at brown.edu]

At 4:00 pm 18/12/94 Dominik wrote regarding transliteration schemes:

>There is, of course, no such thing, any more than there is a best
>grammar of Sanskrit, or a best introductory reader, etc. (except
>that Macdonnell *is* the best grammar, and Lanman *is* the best reader :-)

>There are many, many rational and useful input schemes for Sanskrit,
>and there is really no ground for judging one better than another.
>As long as a system is unambiguous, it is good.

Lack of ambiguity is of course exacltly the minimum criterion of a good
transliteration scheme.  Other criteria are ease of use on a variety of
systems and transferability across the network.  These qualities were also
met by Michio Yano's and mine.  Mine in addition has the virtue of
requiring a single key stroke for each Sanskrit sound.  This makes it
faster to type with, leaves fewer characters to store and leaves nothing to
context when it comes to transliterating to another Font.

>If you want to promote one scheme over another (I don't) then you should
>at the outset make clear whether you are using a 7-bit or an 8-bit
>or a 16-bit character set, and why you have made this choice.  Then you
>have to make clear whether your scheme uses di-graphs or tri-graphs
>to represent single Indic characters.  Finally, you must indicate whether
>your chosen scheme permits or requires code-switching, and what the
>code-switch character is.  I.e., is "a" in English language represented
>by the same code-point (character number) as "a" in Sanskrit, say.

All of this is irrelevant since we are not speaking of anything but what is
common to all these machines and to the usual roman upper and lower case
character set.

>when mixing English and Sanskrit you must have a marker to say
>"here endeth the English and here beginneth the Sanskrit" and vice versa.
>If none of the characters do double duty (like IISCI) then you don't need
>to switch codes in this way.

We are speaking of the roman script used exclusively for Sanskrit, not for
mixing languages.  This is the use required when making a large file of
Sanskrit Text.  When making data files of Sanskirt text it is unnecessary
to use a scheme which makes distinctions between English and Sanskrit at
the expense of the factors I mentioned above under criteria of a good
transliteration scheme.

My remarks were adressed to promoting adequate and useful input of Sanskrit
texts in the wake of recent input that left room for improvement.  Let us
hope that others discern the reasonableness of my suggestions.

>If you intend to distribute texts in bulk, then please make your coding
>scheme publicly documented, as have the Kyoto group, the Vienna group (CSX)
>and Peter Schreiner, etc.

Michio Yano has been thoughful to publish his.  I would like to see the CSX
and Peter Schreiner's.

Peter M. Scharf
Department of Classics
Brown University
P.O. Box 1856
Providence, RI 02912
U.S.A.