Suitability and Accuracy of Memsource in Translating Articles Dealing with Indonesian Cultural Words

Abstract

Official news relating to the state of Germany is published by the government of the Federal Republic of Germany in German, as in the news containing the German state's political issues posted on the Deutscher Bundestag's official website. Because the readers come from various countries including Indonesia and not all readers are proficient in understanding German texts, the use of Google Translate is an alternative solution to overcome these limitations. This study aimed to test the reliability of the GT translation machine based on Memsource's level of suitability. This study also analyzes and describes various forms of inaccuracies that appear in translating German-language political articles into Indonesian, so that GT users will pay more attention to matters related to the development of their usage performance. The data are eight articles on German political issues containing 338 sentences obtained from the official website of the state news website of Germany, namely the Deutscher Bundestag (https://www.bundestag.de). The results showed that the quality of GT translation is generally in the 2nd level of suitability category (85-95) which means that the translation results are almost the same as human translation but still need some editing from experts. It found that there are 237 lexico-semantic errors, and it dominates the types of errors found in the translation results. This implies that GT generally needs to improve in terms of determining the appropriate lexical properties to translate text from the Source Language into the Target Language, such as in choosing one word that best fits the meaning of the text from several available word translation options. The most common morpho-syntactic errors (75 errors) are errors in translating single morphemes that are translated with plural morphemes or vice versa.