Code-Mixed to Monolingual Translation Framework

The use of multilingualism among the new generation is widespread in the form of code-mixed data on social media, and therefore a robust translation system is required for catering to the novice and monolingual users. In this work, we present a translation framework that uses a translation-transliteration strategy for translating code-mixed data into their equivalent monolingual instances. One of the goals of this work is to translate a code-mixed source (written in Roman script) to a Bengali target (written in Devanagari script), where the source may contain English, along with transliterated Bengali. Finally, to convert the output to a more readable form, it is reordered using a target language model. The decisive advantage of the proposed framework is that it does not require a code-mixed to monolingual parallel corpus for training and decoding. On testing the framework, it achieved BLEU and TER scores of 16.47 and 55.45, respectively. Since the proposed framework deals with various sub-modules, we dive deeper into the importance of each of them, analyze the errors and finally, discuss some improvement strategies.