University of Southampton
Subtitling strategies and quality assessment: human vs artificial intelligence
Abstract
dc:description.abstractThis research investigates the quality and translation strategies of interlingual subtitling, comparing the outputs of professional human translators (HT), generative artificial intelligence (ChatGPT-4o), and neural machine translation (DeepL). The study evaluates English-to-French and English-to-Arabic subtitles across three distinct audiovisual genres: a courtroom drama, a nature documentary, and a children’s animated series. Adopting a mixed-methods approach, subtitle quality is evaluated using Pedersen’s (2017) FAR model, while translation choices are analysed using a novel theoretical framework, the FARD model (Foreignization, Adaptation, Reduction, and Disambiguation). Findings reveal that HT consistently achieves the highest quality across all genres by actively employing various subtitle translation strategies to navigate linguistic and technical constraints. Among the automated systems, ChatGPT outperforms DeepL, demonstrating superior contextual awareness, whereas DeepL frequently generates serious semantic and idiomatic errors. Cross-linguistically, automated systems manage Arabic's spatial constraints more effectively than French due to the language's structural economy, though they struggle with Arabic's morphological complexities and translation of idiomatic expressions. The research concludes that while generative AI shows promise as an assistive tool, its inability to independently manage spatial constraints and contextual nuances necessitates extensive human post-editing, reaffirming the indispensable role of human expertise in audiovisual translation.<br/><br/>
Degree
thesis:*- Name dc:type.qualificationname
- Ph.D.
- Level dc:type.qualificationlevel
- doctoral
- Grantor dc:publisher.institution
- University of Southampton
- Year dc:date.issued
- 2026
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Mechairia, Yasmine
- Advisors dc:contributor.advisor
-
- Mccall, Ian D
- Minney, James
- Borthwick, Kate
- ORCID person:identifier.orcid
Rights
- Language dc:language
- English, eng
Identifiers
dc:identifier.*- DOI dc:identifier.doi
- https://doi.org/10.5258/SOTON/01/T044