A number of approaches have been taken to improve lexical consistency in Statistical Machine Translation. However, little has been written on the subject of where and when to encourage consistency. I present an analysis of human authored translations, focussing on words belonging to different parts-of-speech across a number of different genres.
展开▼
