Error Analysis of AI Translation in Korean-English Legal Text and Post-editing Guidelines for Articles: Focusing on Pattern in Article Errors

This paper presents the new translation guidelines of the use of article when post-editors revise the sentences translated by AI. So it's the key point of this work to analyze what kinds of errors can happen in using articles between the sentences authored in Korean by humans and the translated sentences reproduced by AI. This framework is based upon three theories suggested by Christoper(1939), Jespersen(1949), Allan(1980) to analyze the articles in the sentences produced by humans, AI, and post editor with the comparison of the Post-editing guidelines by Ke Hu and Patrick Cadwell(2016). This research on AI translation post-editing is to establish several types of guidelines as a result of pattern in errors. That's why post editors judge to have article errors including grammatical errors to be fixed exactly. These comparisons focus only on the drafting guidelines on how to appropriately use the articles in post editing. Based on these guidelines, post editors can carry out post-editing assignments more easily. In this point, this paper is to put forward whether the guidelines is applicable for a wide range of users who focus on legal texts. However, there is a limitation because much more data should be needed.