Abstract
This doctoral study explores interlingual subtitle voicing (ISV) as an innovative technique
for the creation of interlingual subtitles drawing from traditional translation modes of
subtitling, interpreting and respeaking and primarily relying on speech recognition (SR) as
opposed to the traditional typing. This ISV approach enables subtitlers to voice rather than
type subtitles into a speech recognition tool as an option depending on subtitlers’
preferences and/or work conditions. The rationale for this study was investigating whether
SR can boost efficiency, defined as a confluence of speed and accuracy, and also enhance
the subtitler experience by introducing mobility (ergonomical element, physical health)
and variety (mental wellbeing) into the workflow. For this purpose a hybrid workflow was
created as a part of the ISV training and testing process that relies on already existing skills
utilised in underlying translation modalities, while taking advantage of technologies
emerging in both audiovisual (AVT) translation industry and academic research. Current
research related to SR usually focuses on respeaking in live subtitling for the purposes of
accessibility, primarily in major languages. Research about hybrid workflows is only now
emerging with the fast development of SR, MT and AI, but rarely doesit focus on subtitlers
and their experience and is usually more interested in the process and/or the final product.
ISV study tries to fill this gap by exploring introduction of SR into non-live environments
while enabling users to evaluate and shape future workflows and tools that can include
said additional technologies, which is also a rarity in both the AVT academia and industry.
Such technologies are often not as available or are underdeveloped for low-resourced
languages as opposed to major languages. For that reason and for the purposes of this
doctoral study, a specialised tool was devised specifically for the exploration of various
hybrid workflows in Croatian that improves on the underlying technologies and includes
a possibility of adding other low-resourced languages in the future, both for training and
research purposes. This tool, developed and explored as a part of this doctoral study, is
integrated with specialised software created primarily for the translation of prerecorded
subtitles for non-live broadcasts, streaming services and similar content (e.g. films,
documentaries and TV shows). The main objectives of the research are as follows: (1)
create a theoretical framework for the methodological portion of the study by exploring
the existing literature from traditional translation modes as well as emerging hybrid modes,
especially in AVT; (2) to present the technique; (3) to test whether and how SR technology
and specialised software could improve efficiency (referring to speed and translation
quality, i.e., accuracy in this research cycle) in the translation of subtitles from English into
Croatian; and (4) to explore the process workflow and gather data on subtitlers’ experience
through feedback on both the physical and virtual elements of the ISV technique. This
subtitler-centric research that relies on theoretical notions of action research (AR) and
translator experience (TX) explores various complementary aspects – related to both the
audiovisual translation industry and academia – of a multimodal and multidisciplinary
environment to create a customisable approach and personalised tool that strive to enhance
subtitlers’ workflows and work environments. This study and the technique are a
foundation based on existing translation modalities and drawing from existing academic
literature, while complementing it with knowledge from the most current technological
developments in the AVT industry. From this foundation, recommendations for workflow
variations and professional practice are offered alongside suggestions about further
technological integration and implications for researching and teaching these new hybrid
workflows