Automatic Captioning Translation: Human Names

Abstract

YouTube is a popular website on which audiovisual content can be consumed and shared online. YouTube provides its users with an automatic captioning service which creates captions for their videos along with an auto-translate service which translates the captions into any preferred language out of 109 languages including Turkish. These services tend to produce inadequate output. In this study, proper nouns and human names have been examined within the scope of automatic captioning translation from English to Turkish. With this aim, data excerpts from a YouTube video were selected on an exemplary basis to be analyzed in terms of speech recognition and machine translation perspectives via the concept of translation adequacy and translation strategies. The analysis demonstrated several striking findings with regard to YouTube’s automatic captioning translation and their implications were evaluated via Relevance Theory within the scope of Audiovisual Translation.