This article is part of a Special Issue on The Social Value of European Research on Media Accessibility.
Real-time subtitlers, also called live captioners, produce transcripts of what speakers say in many contexts: cultural events, workplaces, parliamentary assemblies, broadcasts, educational, other. Traditionally subtitlers in these events work in the same language, though often translation is also required when the event has an international participation. Live events subtitling demands require a minimum delay and maximum accuracy and without losing content. However, despite clear EU legislation, the provision and quality of such services are still uneven and sometimes insufficient. The European co-funded project Live Text Access (LTA) aims to bridge this gap by providing harmonised training to professionals in this field.
LTA gathers educational and non-educational partners (trainers, employers, service providers, end users, and certifiers), whose aim is to provide a stable reference for the training of current and future professionals and trainers, through certified open-source learning materials, also accessible to blind trainees.
Two very different professional profiles are developed. The first is that of the respeaker. This is the person who generates subtitles by listening to what is being said and saying it again to a speech to text software. This is in fact the most common – and cheaper – way to produce subtitles these days. The quality may not be perfect, and the reasons for this lack of accuracy may go from the content to be subtitled, the speech to text software used, or the speed of the dialogue or narrative to be subtitled. For example a Woody Allen movie has a dialogue complexity different from a nature documentary. Neither of these two examples are live, still subtitling workflows are changing rapidly.
The second profile is velotypists. Subtitles are produced using a special keyboard and also by pressing different keys at the same time. The result is a text written by syllables, not by letters as we usually do with a QWERTY keyboard.
How the new professional profiles are defined is the start of the LTA project where an online survey was drafted to understand the skills of real-time intralingual subtitlers. The aim was to identify the competence areas and skills that will compose the curriculum for the training of real-time intralingual subtitlers.
The survey departed from a description of skills cards following the European Certification and Qualification Agency (ECQA) guidelines. A complete report is available here. Two different certification avenues are proposed in the LTA project. The first is the university certification measured in ECTS. The second is the professional certification, which can be achieved by doing a test at ECQA. This option opens the door to people with no academic background, still with the right skills to perform this profession.
The project adopted a bottom up approach to gather information towards defining the professional competences. Respondents matched the intended audience of the project: out of 121 respondents, 57 are real-time intralingual subtitlers (mostly freelancers), 29 end users, 17 service providers, 13 trainers, 13 people interested in the profession, and 20 chose Other. Demographic data show that most professionals and service providers have been working in this field for 10 to 19 years, whereas trainers have only been training for 0 to 9 years; only 2 for over 20 years. These results confirm that training lags.
The questionnaire included 48 skills grouped into six competence areas: Knowledge on accessibility, Linguistic Competence, Entrepreneurship and Service Competence, IT Competence, Respeaking Competence and Velotyping Competence. Respondents scored all areas as necessary for the profession (2.1 to 2.7 points out of 3). Results yielded minor changes in the skills cards which included regrouping, specifying, and adding skills such as live editing and easy-to-understand subtitling.
Knowledge about accessibility
Responses show the need to establish quality standards and to clearly describe the role and tasks of professionals. The skills with the highest scores were related to the output quality, to be adjusted to working-settings and classified according to audiovisual media and events.
The skills concerning accuracy and readability obtained the highest scores. Answers reveal two approaches to accuracy: verbatim (word for word transcription) and sensatim (meaning for meaning). Respondents stress the need to train both types of subtitling skills while agreeing on the importance to acquire and develop thematic knowledge.
Entrepreneurship and Service competence
Participants relate the need to acquire these skills to the type of employment. The questionnaire showed that 67% of professionals are freelancers: out of 57 professionals, 34 are freelancers, 14 permanent-employed, and 4 both. Answers provided interpersonal skills necessary for the interaction with users such as a code of conduct, avoiding patronising, and remaining objective and open to dialogue.
The highest scored skills concerned setting up and using the hardware and software for each working setting, followed by the need of identifying potential risks and the ability to solve problems. Professionals described these skills as more relevant and pointed out the difficulty of coping with new developments.
The skill “Communicate with good pronunciation” received the highest score followed by the ability to activate exit strategies in challenging situations (confused or intricate grammar, quick speech-rate, background noise…). Answers triggered two changes in the skills cards: a differentiation between verbatim and sensatim subtitles, and a MARS (Most Accurate and Rapid Speech-to-Text rate [Eugeni, Oncins & Bernabé, forthcoming]) above 120 English words per minute.
The highest scored skills concerned remembering full sentences while lagging and the ability to identify and correct own typing mistakes where necessary. Respondents categorised spelling accuracy and typing speed as subordinated to the overall goals of minimising delay and applying higher-level interpreting strategies. Answers showed the need for clear and grounded parameters to measure accuracy and for defining appropriateness of the technique for specific contexts.
Though conclusions are difficult to derive from a questionnaire which is meant as the starting point of the project, data trigger two interesting reflections: the high number of respondents show the widespread interest for a discipline which is, didactically, still in its infancy; moreover, the agreement on the importance of almost all skills proposed highlights the necessity for a curriculum profiling and training professionals capable of serving multiple needs in the highest number of working settings.
By Rocío Bernabé* from SDI München – University of Applied Languages, Germany
& Carlo Eugeni from SSML – Pisa, Italy
*Corresponding author: Rocio.Bernabe (at) sdi-muenchen.de
Go back to the Special Issue: The Social Value of European Research on Media Accessibility!