Error Analysis In Machine Translation Of Instagram Caption @INFODENPASAR Account

Instagram, one of the most popular social media platforms, is widely used by Indonesian people to connect and communicate with each other worldwide. The impact of Instagram influenced many economic sectors, such as marketing, creative industry, and personal branding. Therefore, social media is utilized through content production to gain additional income and extend engagement. One of the most commonly known Indonesian Instagram accounts @infodenpasar, which focuses on Balinese entertainment news. Most of the posts are wrapped with general Balinese news supported by entertaining captions. However, due to systematic error, the automatic translation feature cannot correctly render the text into the target language of certain users. Therefore, this research aims to identify, categorize, and evaluate errors of automatic machine translation on Instagram caption posts with a case study of @infodenpasar media platform. Qualitative method with descriptive analytic approach was used to analyze the data. The data were obtained from 74 selected posts with short or long captions. Koponen’s theory of error category was applied in this study to describe the translation quality of Instagram captions. The result showed significant errors of the machine translation in identifying local language as well as the contextual and grammatical meaning of the language. Several identified translation errors linked with 5 basic translation concepts were found, namely Omitted Concept (28%), Added Concept (19%), Untranslated Concept (20%), Mistranslated Concept (18%), and Substituted Concept (15%).