Phrase markers in Indian languages

Abstract

Analysis of a natural language input typically consists of the stages of syntactic and semantic analysis. The standard technique is to use a context-free grammar (CFG) and an augmented transition network (ATN), and build a parse tree before processing further. Such a parse tree consists of nou-phrases (NP) and verb-phrases (VP), which are centered around a noun or a verb accordingly. In this paper, we suggest an alternative approach to the concept of phrases which appears to be more meaningful in the context of Indian languages. Instead of the conventional Np· VP grammars, we adopt verb boundaries as phrase markers and build phrase·level representations to convey the meaning of a sentence. The validity of the approach is established through studies on the application-oriented experiments carried out for Indian languages.