Travel Conversation Exercise Corpus

The Travel Conversation Exercise (TCE) corpus contains 888 parallel sentences and their zero pronoun label (true or false) which are written in Japanese and English.The sentences in this dataset are sentences that are often used to practice English conversations, when travelling abroad, for L2 English learners in Japan.Some of the conversation themes include (but are not limited to) conversations at the airport, hotel reservations, sightseeing, and emergencies.Several example usages include using the dataset as an evaluation set to compare machine translation results, compare machine translation models' ability in translating zero pronoun sentences and those which are not, and build a zero pronoun binary classification model to automatically label new Japanese-English parallel sentences.