Morphological analysis of inflecting compound words in Bangla

Abstract

The addition of inflectional suffixes in Bangla com-pound words is fairly complex. A compound is a word that is formed by two or more different words acting as a single entity. One of the key distinguishing features of compounds is the absence of inflectional morphology between the constituents of a compound. In Bangla, however, the constituents may retain inflectional suffixes on either or both the constituents and the resultant compound may then be inflected further as a whole word. Such inflection creates ambiguities as context-free word grammar is unable to recognize whether the inflectional suffix is an inflectional property of the last constituent root-word or of the compound as a whole. We use a feature unification based morphological parser, which can successfully and efficiently parse compound words that retain such inflectional morphology and at the same time resolve such ambiguities.