corpus: data used for training LM: language model built from training data demo: demo used for skunkworksdata used for training working: home of translation engine translate: scripts to generate test data and translations