Finite-state Scriptural Translation

We use robust and fast Finite-State Machines (FSMs) to solve scriptural translation problems. We describe a phonetico-morphotactic pivot UIT (universal intermediate transcription), based on the common phonetic repository of Indo-Pak languages. It is also extendable to other language groups. We describe a finitestate scriptural translation model based on finite-state transducers and UIT. We report its performance on Hindi, Urdu, Punjabi and Seraiki corpora. For evaluation, we design two classification scales based on the word and sentence accuracies for translation system classifications. We also show that subjective evaluations are vital for real life usage of a translation system in addition to objective evaluations.

• Type: Journal
• Year: 2010
• Where: COLING
• Authors: M. G. Abbas Malik, Christian Boitet, Pushpak Bhattacharyya
• Keywords: COLING 2010, Computational Linguistics, Phonetico-morphotactic Pivot Uit, Scriptural Translation, Scriptural Translation Problems