A Paninian perspective to information dynamics in language mapping structures between English and Hindi

Abstract

In the era of globalization, vast amount of information is available on World Wide Web. In order to access this information, the speakers of different languages have to rely on translation, hence the need of Machine Translation (MT) arises. newline newlinePaninian Grammar (PG) universally admired for its insightful analysis of Sanskrit offers a perspective to language analysis. The main focus of PG is to explain how, where and how much information a language encodes which helps in capturing information dynamics in MT. In order to use Paninian theory to analyse other languages, we model these languages in terms of Paninian primitives such as sup, tin, pada, samasta-pada, etc. In our research, we have used these concepts to analyse English from Paninian perspective and then map English structures into Hindi for handling following major linguistic tasks: newline newline1. Identifying the Meaningful Linguistic Units That Can Participate in a Sentence: This step theoretically accounts English sentence in terms of Paninian primitives such as sup, tiand#729;n, pada and samasta-pada. newline2. Mapping Dependency Parsers Output into a Uniform Schema: Bringing parsers output into one uniform notation provides the system an ability to plug-in any of these parsers without modifying other sub-sequent modules thereby saving large amount of manual work, cost and time. newline3. Framing Transfer Grammar Rules: Word ordering plays an important role in the translation process between languages. Using the concept of pada and samasta-pada the source language sentence is manipulated into target language sentence for natural translation. newline newlineThe main focus of this study is on MT between two diverse languages English and Hindi, But the research carried out in this thesis is generic enough to be applied to any English-Indian Language pair. Because most of the Indian languages have many common features and the transparency of the tools built during this research allows to cover diverse cases easily.