The Analysis of English-Indonesian Translation of Poetry on Chat Generative Pre-Training Transformer (GPT)

Abstract

The purpose of this study is to describe the quality of translation (accuracy, acceptability, readability). This research uses descriptive qualitative method. The data were collected using purposive sampling method. To collect data, researchers used documentation techniques. This research instrument was validated and the quality of translation was assessed by rater. The first finding in this study is that in the first poem the accuracy level reached 75%, less accurate 25% and inaccurate 0%. The second poem had an accuracy rate of 95%, less accurate 5% and inaccurate 0%. The third poem has an accuracy rate of 88%, less accurate 12 and inaccurate %. The second finding, in the first poem, the acceptable level was only 8%, less acceptable 88% and unacceptable 4%. The second poem, the acceptable level reached 60%, less acceptable 25% and unacceptable 15%. The third poem has an acceptable level of 58%, less acceptable 38% and unacceptable 4%. The third finding is that the first poem has a low readable level of 13%, less readable 83% and unreadable 4%. The second poem has a readable level reaching 60%, less readable 20% and unreadable 20%. The third poem has a readable level with a percentage of 63%, less readable 37% and unreadable 0%