Urdu Conjunct Predicates (N+V) Inventory from Urdu Universal Dependency Corpus

Abstract

This research study aims to develop a semantic inventory of Urdu nouns which may serve as a useful resource for developing natural language processing tools. It is an effort towards improving the severely under-resourced status of Urdu. Conjunct predicate is a type of complex predicate where a noun is followed by a light verb and both work as a single syntactic constituent. Conjunct predicate N+V collocation is extracted from universal dependency annotated Urdu corpus i.e., URDU_UD_UTB (Bhat et al., 2017). Resultant data provided adequate information to categorize the pattern of nouns compatible with light verbs in their all-possible morphological forms. This research yields a sizeable repository of Urdu conjunct predicate along with figuring out a range of case markers licensed by N+V collocation as a constituent which does further implication on the volitionality. Resultant mined data can be used in some future research work to train the data in some cross-linguistic computational programs.