This repository contains a library of Emacs Elisp functions that perform various computations of Sanskrit Grammar. The system was begun in about 2002, with some revisions in 2006, and currently, in 2015.

The principles of grammar were adapted from reprints of two texts:

• Robert Antoine, A Sanskrit manual for high schools, 1914
• M.R. Kale, A Higher Sanskrit Grammar, 1894.

The general approach was to develop functions which take as inputs material drawn from a Sanskrit dictionary and which generate inflected forms (declensions of nominals and conjugations of verbs). The Sanskrit dictionary used was that developed in the 1990s by Thomas Malten of Cologne University.

The Sanskrit texts mentioned above present Sanskrit grammar in what I think of as a model-based form. The lexicon provides information to identify the model to be applied, and the Sanskrit grammars present the algorithms that pertain to a given model. This is likely conceptually different from the approach of Panini's grammar, although, as I understand it, this approach is based upon an interpretation of Panini.

The choice of Elisp was made because of its free availability, ease of installation, and computational completeness.

At this current time in 2015, it seems desireable to revisit this earlier work. There have been substantial improvements in Malten's digitization of lexicons. I would like to adapt this Elisp work using the Python programming language. In this adaptation, I would hope to take advantage of the dictionary improvements and also to improve and extend the algorithms. (One notable extension would be for various verb forms: aorist, desiderative, intensives.) Such improvements would help improve the grammatical accuracy of the computed forms.

In the present form of this repository, the most accessible parts reside in the grammar/prod/outputs directory. This directory contains a large number of computed forms in the various text files. All of the Sanskrit words in these files are spelled using the SLP1 transliteration system for Sanskrit. The file readme-batch.org explains how these files were constructed. These constructed forms are generally believed to be accurate, but it is likely that many undiscovered errors are present.

The readme-interactive.org file provides several examples of using some of the Elisp functions interactively, to generate various declensions and conjugations.

For someone interested in examining the code base itself, the files in the static-code-analysis directory would be useful; a brief explanation appears in the readme file therein.

The pysanskrit directory contains a rewriting of the elispsanskrit functions in the Python programming language. There is a comprehensive set of tests to confirm the code equivalence. See the readme.md file of pysanskrit directory for details.

Compare the inflected forms of elispsanskrit/grammar/prod/outputs to reference material from Gerard Huet.