Sanskrit Computational Linguistics

Sanskrit is the primary culture-bearing language of India, with a continuous production of literature in all ?elds of human endeavor over the course of four millennia. Precededbyastrongoraltraditionofknowledgetransmission,records of written Sanskrit remain in the form of inscriptions dating back to the ?rst centuryB. C. E. Extantmanuscriptsin Sanskritnumber over30million,one h- dred times those in Greek and Latin combined, constituting the largest cultural heritage that any civilization has produced prior to the invention of the printing press. Sanskrit works include extensive epics; subtle and intricate philosophical, mathematical, medical, legal, and scienti?c treatises; and imaginative and rich literary,poetic,anddramatictexts. WhiletheSanskritlanguageisofpreeminent importance to the intellectual and cultural heritage of India, the importance of the intellectual and cultural heritage of India to the rest of the world during the pastfewmillennia andinthe presenteracanhardlybe overestimated. The int- lectualandculturalheritageofIndia hasbeen amajorfactor inthedevelopment of the world's religions, languages, literature, arts, sciences, and history. Sanskritdocumentsaremovingintothedigitalmedium. Recentdecadeshave witnessed the growth of machine-readable Sanskrit texts in archives such as 1 the Thesaurus Indogermanischer Text- und Sprachmaterialien (TITUS), Ky- 2 3 oto University, Indology, the Gottingen ¨ Register of Electronic Texts in In- 4 dian Languages. The last few years have witnessed a burgeoning of digital images of Sanskrit manuscripts and books hosted on-line.

This volume presents the proceedings of the Third International Sanskrit C- putational Linguistics Symposium hosted by the University of Hyderabad, Hyderabad,IndiaduringJanuary15–17,2009.TheseriesofsymposiaonSanskrit Computational Linguistics began in 2007. The ?rst symposium was hosted by INRIA atRocquencourt,Francein October 2007asa partofthe jointcollabo- tion between INRIA and the University of Hyderabad. This joint collaboration expanded both geographically as well as academically covering more facets of Sanskrit Computaional Linguistics, when the second symposium was hosted by Brown University, USA in May 2008. We received 16 submissions, which were reviewed by the members of the Program Committee. After discussion, nine of them were selected for presen- tion. These nine papers fall under four broad categories: four papers deal with the structure of Pan ¯ ini's Astad ¯ hyay ¯ ¯ ?. Two of them deal with parsing issues, . .. two with various aspects of machine translation, and the last one with the Web concordance of an important Sanskrit text. Ifwelookretrospectivelyoverthelasttwoyears,thethreesymposiainsucc- sion have seen not only continuity of some of the themes, but also steady growth of the community. As is evident, researchers from diverse disciplines such as l- guistics, computer science, philology, and vy¯ akarana are collaborating with the . scholars from other disciplines, witnessing the growth of Sanskrit computational linguistics as an emergent discipline. We are grateful to S.D. Joshi, Jan Houben, and K.V.R. Krishnamacharyulu for accepting our invitation to deliver the invited speeches.

This volume constitutes the refereed proceedings of the 4th International Symposium on Sanskrit Computational Linguistics, held in New Delhi, India, in December 2010. The 18 revised full papers presented were carefully reviewed and selected from numerous submissions. The papers can be categorized under following broad areas such as phonology and speech technology; morphology and shallow parsing; syntax, semantics and parsing; lexical resources, annotation and search; machine translation and ambiguity resolution.

About the Book India has a rich grammatical tradition, still extant in the form of PÀõini’s grammar as well as the theories of verbal cognition. These two together provide a formal theory of language communication. The formal nature of the theory makes it directly relevant to the new technology called Natural Language Processing. This book, first presents the key concepts from the Indian Grammatical Tradition (IGT) that are necessary for understanding the information flow in a language string and its dynamics. A fresh look at these concepts from the perspective of Natural Language Processing is provided. This is then followed by a concrete application of building a parser for Sanskrit using the framework of Indian Grammatical Tradition. This book not only documents the salient pieces of work carried out over the last quarter century under Computational Paninian Grammar, but provides the first comprehensive exposition of the ideas involved. It fills a gap for students of Computational Linguistics/Natural Language Processing who are working on Indian languages using PÀõinian Grammatical Framework for developing their computational models and do not have direct access to the texts in Sanskrit. Similarly for the Sanskrit scholars and the students it provides an example of concrete application of the Indian theories to solve a contemporary problem. About the Author Amba Kulkarni is a computational linguist. Since 1991 she has been engaged in showing the relevance of Indian Grammatical Tradition to the field of computational linguistics. She has contributed towards the building of Anusaarakas (language accessors) among English and Indian languages. She is the founder head of the Department of Sanskrit Studies, University of Hyderabad established in 2006. Since then her focus of research is on use of Indian grammatical theories for computational processing of Sanskrit texts. Under her leadership, a consortium of institutes developed several computational tools for Sanskrit and also a prototype of Sanskrit–Hindi Machine Translation system. In 2015, she was awarded a “Vishishta Sanskrit Sevavrati Sammana” by the Rashtriya Sanskrit Sansthan, New Delhi for her contribution to the studies and research on Sanskrit-based knowledge system. She was a fellow at the Indian Institute of Advanced Study, Shimla during 2015-17.

Computer Processing of Sanskrit Nominal Inflections: Methods and Implementation is the result of Research and Development (R&D) at the Master of Philosophy (MPhil) level at Jawaharlal Nehru University, New Delhi, India. The title of the dissertation was “Machine Recognition and Morphological Analysis of Subanta-Padas.” The work, which is based on the reverse engineering implementation of Panini’s Sanskrit Grammar, brings together new and original studies in the area of computational linguistics, language technology and natural language processing with reference to parsing Sanskrit nominal inflections. On the surface level, Panini has defined rules in a forward looking generative fashion which makes reverse analysis necessary for parsing. Since parsing inflections is the first basic step towards complete analysis, the present work has relevance for any larger system that may evolve in future.

Sanskrit Informatics is intended as a study guide for Sanskrit Students attending methodology courses on Informatics. It can create awareness about the available digital resources on Sanskrit and Indology, and introduce the basics of ICT skills for effectively accessing, processing and using such resources