Booth Id:
SOFT048
Category:
Year:
2016
Finalist Names:
Paxton, Jesse
Abstract:
This study is the first to examine the accuracy of Google Translate with American vocabulary from actual high school syllabi. Five sentences, chosen for their potential to pose complications for international students, were translated into both Spanish and Korean from English using Google Translate. Four native Spanish speakers and four native Korean speakers, all who were fluent in English as well, were each asked to translate all five English phrases into Spanish or Korean, respectively. As expected, the human translations had slight variations, providing multiple possible matches to serve as references for comparison for assessing the accuracy of Google Translate. Each Google translation was analyzed by using a numerical metric, measuring deviations in length, word choice, and word order, in comparison to the human translations. The deviations were also categorized as lexical, syntactic, or discursive. The accuracy of Google Translate was judged to have higher quality the closer it was to the human translations. This study found that, while Google Translate might be relatively accurate in translating English to Spanish, it is far less accurate translating English to Korean. These results support the hypothesis that high school English phrases found in syllabi pose a greater chance for misunderstanding when translated into Korean using Google Translate. This study has implications, not only for machine translation, but also international students and high school teachers. A more in-depth examination might be warranted, focusing on other East Asian languages as well.