compound analysis in e-texts

[Madhav Deshpande]

	Since John Gardner has brought up the question of Vedic Padapathas
in connection with compound analysis in e-texts, it may be well worth to
keep in mind some issues relating to the Padapathas.  As the evidence in
Yaska's Nirukta and Patanjali's Mahabhasya shows, the Pada divisions
proposed by specific Padakaaras were not universally accepted by others. 
The Padapathas for different Samhitaas occasionally differ from each
other.  Often the Padapathas leave expressions unanalyzed and the
Praatishaakhyas explain this by saying that the division was not shown
because of sandeha 'confusion', anyaayasamaasa 'irregular compounding'
etc.  That means even the authors of the Padapatha did not pretend to be
able to a 100 % job of analysing.
	E-texts with compound analysis are indeed some kind of Padapathas
for those texts.  Up to a certain extent they are useful, but one must
never confuse them with original texts, which do not have even word breaks
in the manuscripts or recitations.
	Using the e-texts with compound analysis or any other kind of
analysis is indeed a mixed blessing.  I found this while using the Sri
Lankan Pali canonical texts in e-text form.  There are numerous errors of
analysis and typing.  The same happened even with the epic texts of
Tokunaga.  Using these e-texts for searching expressions etc., therefore,
gives us perhaps a majority of the occurrences, but one can never be sure
that we have them all.  
	Hence, it would be useful to have a given text available in two
forms, the Samhitaa form and the Pada form, even in the age of e-texts.
	Madhav Deshpande

 On Tue, 27 Aug 1996, JR Gardner wrote:

> I'm just curious, considering the RV padapaatha is, in effect, a 
> "commentary"--is this what the rendering of texts should be?  Or whould 
> the analyzing per akSara not be left to the citation of specific passages 
> in analytical studies.
> 
> Viz. e-texts, preparing them is mammoth enough considering limitations, 
> still, of non-roman-script computing.  OCR work would still require 
> line-by-line analysis and search engines now--for a careful scholar--can 
> pick out units of words sought without akSara analysis.
> 
> John Robert Gardner
> University of Iowa
> 
> 
>