POST-EDITING PROPORTION OF GOOGLE TRANSLATE IN INFORMATIVE AND EXPRESSIVE TEXTS

The massive development of Google Translate (GT) is remarkable and there are many people all over the world use it. Yet, the texts that have been translated by GT still need post-editing by the human translator. This research aims to find the proportion that translator needed in post-editing when using GT in translating of informative and expressive text from English to Indonesia and what they need to pay attention in translating informative and expressive text using GT. The data in this research were words, phrases, and sentences that were analyzed using descriptive-comparative with error analysis at three different levels: accuracy and acceptability. The result shows that the proportion of post-editing for informative text is 5% for accuracy and 22% for acceptability, while in the expressive text, 33% for accuracy and 47% for acceptability. Humans need more effort in post-editing the expressive text because the structure of the sentences in the expressive text is more complex and longer. Furthermore, there are many CSI (Cultural Specific Items) found in the expressive text that makes the result unacceptable for TT readers. Another problem in using GT that the result of GT translation is literal while the expressive text has many of figurative meaning. It means that GT is acceptable for semantic translation. Thus, this research suggests GT can be used directly in translating informative text because informative, as long as the users conduct post-editing and the users should not use GT in translating expressive text directly except for decoding the semantic, pragmatic, and contextual meaning to find the suitable translation.