earticle

한국어

In the recent machine translation (MT) market, a massive amount of machine translation-related projects are carried out every year, but there are no specific guidelines in Korea that can be used as a common standard by language service providers (LSPs), post-editors, and translators other than the client-specific localization style guides, such as Amazon. Based on the current situation, this study intends to explore the research on the MT post-editing guide conducted by the KIGO Style Guide Association as well as actual MT post-editing (MTPE) projects in the localization industry, which might help LSPs and post-editors coexist in the MT market. Despite the rapid development of neural machine translation (NMT), it is still too early to publish machine translations without human quality assurance (QA) work, meaning that the role of human post-editors is still an inevitable requirement. In other words, this study was initiated with the purpose of serving translators who need to quickly adapt to emerging technologies and expand their careers in the MT market and for establishing a standard for Korean LSPs that can be used in the negotiations between LSPs and their customers. The initial study of the MTPE Style Guide ver. 2018 contains basic MT-related concepts and definitions of MT error types, but it was difficult for users of this guide to understand each error type, as there were no detailed examples provided. Therefore, a follow-up study conducted in 2019-2020 added detailed examples for each error type with relevant description. It is expected that such a study on the MTPE style guide would serve many people in the localization industry, and that additional research would allow for the creation of common guidelines for various localization subject matters related to LSPs and academia in the near future.