compound analysis in e-texts

[Lars Martin Fosse]

Birgit Kellner wrote:

>Lars Martin Fosse wrote: 
>>  May I, by the way,
>> suggest that all typing of Sanskrit texts should be done with compound
>> analysis the way it is done in the TZ-format? Proper compound analysis makes
>> the texts fare more valuable for electronic analysis than entering the texts
>> in the way they appear in a Sanskrit book or manuscript. It is definitely
>> "value added" text entry!
>
>I strongly disagree, or, at least, I would like to express some reservations
>as to compound-analysis or Sandhi-separation (but maybe Lars did not refer
>to the latter, anyway).

Yes, I did, although not explicitly.

 Most of the time, I type texts which I do not yet
>fully understand, or I type texts, because I'm paid for it, and I do not
>aspire to fully understand them. There are critical cases of compound
>analysis, i.e. ambiguous cases (sometimes it's difficult to even tell
>whether a certain cluster is a compound or not), and analysis carried out by
>somebody who actually did not work closely with the text is more detrimental
>than no analysis at all. 

This is not a good argument against compound analysis as such. I have worked
with people who typed text for me, some of them analysing compounds very
competently, others not. When I got non-analysed text, I sent it to another
person who was able to analyse the compounds. Using the TZ-format, I never
have any problem recreating a sandhi-text by means of macros, thus getting
the best of two worlds. The compound-analysed text is essential for a number
of analytical tasks performed by computer, e.g. language statistics, word
collocation studies etc. 

>On a more mundane level - I am using texts where compounds are analyzed
>(i.e. compound-elements are separated with dashes), and quite frequently, I
>paste quotations from those texts into foot-notes of other files. It's quite
>a hassle to get rid of all the dashes, and I would much more prefer a file
>which didn't have analyzed compounds. 

Read the analysed text into a text editor with a good macro program and
change the text so that it looks the way you want it before you start
copying quotations.

>I know, everybody else seems to be perfectly happy with compound-analysis,
>so either I am suffering from a chronical misunderstanding, or my knowledge
>of Sanskrit is even worse than I thought ...

I suspect that our knowledge of Sanskrit is probably always worse than we
think. It is not an easy language. My impression is that for fresh Sanskrit
learners, working with a compound analysed text is quite a relief. I have
utilized some of my analysed texts in class, producing a double set of
texts, one without compound analysis (but romanized), another with compound
analysis (romanized) and of course the same text in devanagari. This saved
time at a stage of study where the students in my opinion should not spend a
lot of time trying to reinvent the wheel. The students were happy and did
quite well at their examinations, where they had to face the texts as they
really are. 

Best regards,

Lars Martin



> From 101621.104 at CompuServe.COM 26 96 Aug EDT 02:54:04
Date: 26 Aug 96 02:54:04 EDT
From: Anthony P Stone <101621.104 at CompuServe.COM>
Subject: Re: typing Sanskrit

On 24 Aug 1996,  Jakub Cejka wrote:

> a, aa This has, IMHO, a great drawback: Using sequence aa for long vowel
makes it impossible to type also Prakrit in the same way. So it's for  example
unusable for inputting "Sanskrit" dramas.

But  with aa as the long vowel, two short vowels could be typed a.a in BO.

Tony Stone