This work focuses on building language models (LMs) for code-switched text. We propose two techniques that significantly improve these LMs: 1) A novel recurrent neural network unit with dual components that focus on each language in the code-switched text separately 2) Pretraining the LM using synthetic text from a generative model estimated using the training data. We demonstrate the effectiveness of our proposed techniques by reporting perplexities on a Mandarin-English task and derive significant reductions in perplexity.
@article{arxiv.1809.01962,
title = {Code-switched Language Models Using Dual RNNs and Same-Source Pretraining},
author = {Saurabh Garg and Tanmay Parekh and Preethi Jyothi},
journal= {arXiv preprint arXiv:1809.01962},
year = {2018}
}