You have no items in your shopping cart.
ABSTRACT
This work carried out a lexico-syntactic analysis of errors in auto-generated subtitles. The study specifically identified the lexico-syntactic features present in the auto-generated subtitles, examined how errors influenced meaning and readability, and analysed the types of sentence structures used, along with their communicative effects on viewers. The data for this study were drawn from three categories of YouTube videos: news, education, and entertainment; each with distinct linguistic characteristics. The analysis was carried out using Systemic Functional Linguistics (SFL) and Error Analysis (EA) Theory, providing insights into how language is structured within auto-generated subtitles and how deviations impact comprehension. Findings from the analysis revealed that lexical choices, syntactic structures, and phonological influences played a significant role in shaping the accuracy of these subtitles. Some mistakes are small, while others change the meaning of sentences completely. The study highlights that, despite improvements in speech-to-text technology, many errors still occur because the system does not fully understand context. This research adds to the discussion on howNatural Language Processing (NLP) systems can be improved in order to make auto-generated subtitles more accurate and useful.
