摘要：This paper conducts an empirical investigation on the errors in English-Chinese translation memories in computer-aided translation (CAT) and shows that the intention of error occurrence and the statistical difference in three common types: fixed expressions, omissions and symbols. It reveals that the translation errors in sentence pairs of translation memories reach as high as over 14%, and among all the errors in target texts punctuation errors account for nearly 46%. These findings are of great significance in improving the target text quality as well as in lowering the cost in CAT.
