Phrase and Idiom Identification in Assamese

Identification of phrases and idioms is an indispensable part of computational linguistics work. In case of Assamese, this is a challenging topic mainly because of the cases and affixes used in the language. Though, this language is an Eastern Indo-Aryan language spoken by around 30 million people, this topic has not been studied much, as very little computational linguistics work has been done for this language. Assamese language is a relatively free word order language. Context Free Grammar (CFG) can be applied in phrase level by taking extra care in defining the production rules. In this paper, we explain about a method which can be considered as modified context free grammar. Different production rules for phrases can be defined using this modified context free grammar. In this method, the right hand side of the production rules is treated as a free string. So that free word order phenomenon can be dealt with. Different idioms are also analyzed in terms of their syntax and use, to find out the similarities among them to build a dictionary of idioms. Difficulties in parsing phrases and idioms are also discussed and some of the techniques are also provided to overcome those difficulties.