The George Washington University System for the Code-Switching Workshop Shared Task 2016

We describe our work in the EMNLP 2016 second code-switching shared task; a generic language independent framework for linguistic code switch point detection (LCSPD). The system uses characters level 5-grams and word level unigram language models to train a conditional random fields (CRF) model for classifying input words into various languages. We participated in the Modern Standard Arabic (MSA)-dialectal Arabic (DA) and Spanish-English tracks, obtaining a weighted average F-scores of 0.83 and 0.91 on MSA-DA and EN-SP respectively.