Human-Computer Interaction: Method of Translation Facebook Auto Translation in Translating Arabic–Indonesian on Al Jazeera Channel Posts

The main problem of translation lies in the method used by the translator. Newmark suggests in theory eight translation methods commonly used by "human" translators in translating texts. However, whether this theory also applies to machine translators is a question that this study attempts to answer. More specifically, this study aims to analyze the translation method used by Facebook Auto Translation (FAT) in translating Arabic-Indonesian based on Newmark's V diagram theory. This research is a combination of quantitative and qualitative research. Data mapping was carried out quantitatively, then data analysis was carried out qualitatively, in which the collection, analysis and presentation of data was carried out descriptively.The results showed that by the 58 posts and their translated results analyzed, the researchers found that FAT only applied two of the eight methods proposed by Newmark. The two methods are word-for-word translation and literal translation. A total of 6 translation results (10.34%) are classified as word-for-word translations and 52 translation results (89.66%) are classified as literal translations. The translation method applied by FAT is dominated by literal translation because basically Arabic grammar and Indonesian grammar are different. Arabic has a predicate + subject sentence pattern, although there are also some that have a subject + predicate pattern. Besides that, Arabic also has additional letters or words to verbs which do not have to be translated into Indonesian but must still be in the text because their role is very important in determining meaning. The difference in grammatical structure causes the order of translation to also change so that FAT applies more literal translation methods.