Comparing the impact of subtitles on learning: automatically generated vs. corrected subtitles

Subtitles have been proven to benefit language acquisition, improve comprehension, listening skills and word recognition in an educational context. According to Cognitive Load Theory (CLT), information from different modalities that are presented simultaneously may cause cognitive overload due to the redundancy effect. However, research has shown an increase in learning when information is presented concurrently both visually and auditorily. The result can be explained by Dual Coding Theory: when redundant information is complimenting each other during processing, no cognitive overload results. Adding subtitles is essential for equal access as many online video lectures have become a major channel for learning in education. However, conventional subtitling is both expensive and time consuming, with the result that automated subtitling is a potentially powerful solution if the issue of accuracy and readability can be resolved. The focus of this study is to investigate the impact of automated and corrected subtitles on learning (tested in the context of micro-economic principles). A video was shown to participants, who were randomly assigned to 3 test groups: English lecturer with no subtitles, English lecturer with automated subtitles, and English lecturer with corrected subtitles. Participants were asked to complete a pre-test, an effort test and a post-test. Biographical information was collected for data analysis and interpretation. The objectives of this study are 1) to determine whether adding automated English subtitles and corrected English subtitles to English video improves learning; 2) to determine whether these subtitles impact differently; and 3) to compare the amount of cognitive load induced by automated and corrected subtitles on these groups.