Mapping the Linguistic Landscape: Progress Towards a Bengali FrameNet

FrameNet, a lexical database based on Frame Semantics, has been extensively developed for English but lacks comprehensive versions in other languages, such as Bengali, limiting its application in Natural Language Processing(NLP). This research addresses this gap by adapting Frame Semantics to Bengali, resulting in the development of a Bengali FrameNet. The process involved identifying semantic frames, frame elements, and lexical units specific to Bengali, a language characterized by rich morphology and significant syntactic differences from English. Manual annotation of Bengali texts was carried out using frames from the English FrameNet, with the expertise of linguistic annotators. Key stages of this work included corpus collection and preprocessing, as well as the identification of frames, frame elements, and lexical units, leading to the successful mapping of 31 frames and 120 frame elements. These annotations are vital for enhancing Bengali NLP tasks such as semantic parsing, information extraction, and machine translation.