Mapping of Semantic Class of Nouns in Urdu Conjunct Predicates to the English WordNet

Description

The study aims to develop an inventory of Urdu conjunct predicates which helps to draw semantic classes of Urdu nouns, and to investigate compatibility patterns between Urdu nouns and light verbs such as argument structure, case marking, and semantic role of the subject argument influenced by Urdu host nouns in the clause, and to develop a tool to identify true Urdu conjunct predicate. The complex predicate and its subtypes are considered as single constituents where a light verb is hosted by a verb (V1+V2), a noun (N+V), or an adjective (Adj+V) and known as compound predicate and conjunct predicate respectively. Complex predicates are a quite prolific and productive phenomenon in the languages of Indo-Aryan origin including the Urdu language. Urdu Conjunct predicate (N+V), a type of complex predicate, has not been fully explored to date for its semantic and syntactic information. The lack of adequate lexical resources related to Urdu conjunct predicates has been a challenge for Natural Language Processing applications. To develop an Urdu N+V inventory, two Urdu Corpora i.e., Universal Dependency Urdu Treebank Corpus and Urdu WordNet Wordlist 0.1 are used. To draw semantic classes of Urdu nouns, semantic senses of Urdu nouns are mapped onto twenty-five semantic noun primes in English WordNet. To elaborate on combinatory restrictions between Urdu nouns and light verbs, a theory of correlation between semantic and syntactic context has been employed. A tool is developed to identify true conjunct predicates so that these can be inserted into Urdu electronic resources such as Urdu WordNet. Achievement of these research objectives may contribute to the digitally enabled status of the Urdu language which is less resourced to date.

Period	7 Nov 2023
Examinee	Farhat ABDULLAH
Examination held at	Air University, Islamabad Nanyang Technological University
Degree of Recognition	International