TADS: Translation of Audio Description Scripts

Description of the project

Audio description (AD) is a technique to verbalise visual content for blind and visually impaired people. Research on this description technique, in particular applied to films, is carried out in the field of Applied Linguistics within a branch of Translation Studies that deals with intersemiotic forms of translation.

Despite the signing of Directive (EU) 2019/882 of the European Parliament and of the Council of 17 April 2019 on accessibility requirements for products and services, the current supply of audio described films is rather limited. From the scientific side, therefore, the question of optimising the work processes for the production of audio described films has been raised for several years. This includes the translatability of AD scripts for dubbed films and TV series, a topic which is highly relevant for multilingual broadcasters and their suppliers.

Initial research indicates that AD scripts possess specific text type features. We assume that for this reason, AD script translations cannot be used without the following localisation tasks: consider country-specific AD guidelines, fit the AD units into the dialogue gaps of the dubbed film, and maintain coherence between AD units and the dubbed soundtrack.

Based on current findings in AD research, the TADS project pursues two closely related goals:

1. Contrastive research on the text type AD script by compiling, annotating, and evaluating parallel and monolingual corpora containing French and German AD scripts and their corresponding films in original and dubbed versions, if available.

2. Research on the suitability of translated AD scripts for the production of audio described films by means of a study to survey and compare the effort and duration involved in the production of AD scripts: production ex novo versus human translation, or NMT (neural machine translation). We also seek to analyse the quality of the results. The acceptability of the audio described films based on the AD scripts obtained is investigated through a user study.

Our decision to compile a German-French/French German AD script corpus is justified by the fact that to date, little research has been done on ADs in these two languages. Our cooperation partners, the television broadcasters ARTE and SRF, will provide us with AD scripts of dubbed films and series in German and French and for some of them even with their video. In addition, corpus-based and corpus driven research opens up the perspective to identify linguistic features which may help to explain and to face challenges linked to AD script production for dubbed films.

The outcomes of the TADS project are several annotated datasets (some of them multimodal) shared with the scientific community, as well as the developed guidelines for the localisation of translated AD scripts and for AD-specific post-editing.

 

Main team members (alphabetical order):

German Team: Nathalie Mälzer, Saskia Schulz, Maria Wünsche

French Team: Eva Schaeffer-Lacroix

in cooperation with: the broadcasters ARTE, SRF, the association of blind and visually impaired people DBSV, the film describers Anke Nicolai and Laure Morisset 

Consultants: Sylvia Jaki, Ulrich Heid, Serge Heiden and Thomas Schmidt

MA-students supporting in the project: Lena Schneider and Linda Jagonak. 

Call for Papers. Special Issue for the Journal of Audiovisual Translation (JAT)

Beyond Objectivity in Audio Description: New Practices and Perspectives

We gather contributions that identify, discuss, and go beyond the limits of objectivity in audio description of different cultural products such as film, TV programs, theatre or dance performances, opera, or museum exhibitions. We welcome contributions by both researchers and practitioners.

Deadline for Abstracts: March 15th 2022.

Guest Editors are. Nathalie MÄlzer, Eva Schaeffer-Lacroix and Maria Wünsche

Please find more information here.

Conference: The limits of objective audio description

The conference will takte place in Hildesheim from October 5-8th, 2022. 

Organisators: 

Nathalie Mälzer (University of Hildesheim) and Eva Schaeffer-Lacroix (Sorbonne University)

Participants:

Nina Reviers Belgium, Anna Jankowska Belgium, Gert Vercauteren Belgium, Maria Zimina-Poirot France, Kirsten Berland France, Ekaterina Lapshinova-Koltunski Germany, Saskia Josephine Schulz Germany, Sylvia Jaki Germany, Maria Wünsche Germany,Noura Gzara Germany, Ismini Karantzi Greece, Elena di Giovanni Italy, Anna Matamala Spain, Pilar Orero Spain, Yuchen Liu Spain, Susanne Jekat Switzerland

Die Audiodeskription (die Beschreibung visueller Inhalte für Menschen mit Sehbehinderung) ist ein junger Forschungsgegenstand. Ihre wissenschaftliche Untersuchung ist angesichts des steigenden Bedarfs an Audiodeskriptionen (AD) von hoher gesellschaftlicher Relevanz, da mit der Erforschung dieser Textsorte der Abbau von Barrieren im Bereich der ästhetischen Kommunikation unterstützt wird. Im Fokus der Tagung steht die Diskussion der Objektivität von AD sowie die Frage nach länderspezifischen, sprich: sprachlichen und kulturellen Unterschieden bei ihrer Formulierung. Das langfristige Ziel ist eine Optimierung der Arbeitsprozesse durch den Einsatz maschineller Übersetzung. Dies erfordert jedoch die Untersuchung der spezifischen Erfordernisse für eine angemessene Lokalisierung und ein textspezifisches Postediting maschinell übersetzter AD, um kohärente Hörfassungen von Filmen oder Theaterstücken zu gewährleisten. Die Tagung findet im Rahmen des deutsch-französischen Forschungsprojekts TADS (Translation auf Audio Description Scripts) statt, das von arte und dem Schweizer Fernsehen unterstützt wird. Die Konferenzbeiträge werden in einer Sondernummer der internationalen peer-reviewed und open access-Zeitschrift Journal of Audiovisual Translation JAT erscheinen (siehe oben).

 

Audio description (the description of visual content for people with visual impairments) is a young research subject. Its scientific investigation is of high social relevance in view of the increasing need for audio description (AD), as research into this type of text supports the removal of barriers in the field of aesthetic communication. The conference will focus on the discussion of the objectivity of AD as well as the question of country-specific, i.e. linguistic and cultural differences in its formulation. The long-term goal is to optimise work processes through the use of machine translation. However, this requires the investigation of the specific requirements for appropriate localisation and text-specific posting of machine-translated AD in order to ensure coherent audio versions of films or plays. The conference takes place within the framework of the Franco-German research project TADS (Translation on Audio Description Scripts), which is supported by arte and Swiss Television. The conference papers will be published in a special issue of the international peer-reviewed and open access journal Journal of Audiovisual Translation JAT (see below).

 

funded by Procope Plus – DAAD.

 

Lectures

Schaeffer-Lacroix, Eva. "Language learning from an audio description corpus". TaLC13 2018, Faculty of Education, Cambridge (England), July 18-21st, 2018.

Berland, Kirsten / Mälzer, Nathalie / Schaeffer-Lacroix, Eva / Schulz, Saskia. “Purpose and features of the TVseriesAD corpus”, 3rd Swiss Conference on Barrier-free Communication, ZHAW (Virtual), July 1st, 2020

Schaeffer-Lacroix, Eva. "Annotation von Audiodeskriptionsskripten am Beispiel des Korpus zu Neues aus Büttenwarder" (videoconference). Forschungskolloquium Fachbereich 3 (Sprach- und Informationswissenschaften). University of Hildesheim, November 24th 2020.

Mälzer, Nathalie / Schaeffer-Lacroix, Eva. “The French-German TADS Project: A Multi Method Approach to the Translation of AD Scripts”, ARSAD Barcelona 2021, January 26th, 2021

Schaeffer-Lacroix, Eva / Saskia Josephine Schulz. "Creating and translating audio descriptions with the help of Frazier". Colloque interdisciplinaire "Vers une robotique du traduire ?". Université de Strasbourg, 30 September - 1 October 2021. https://robotrad2020.sciencesconf.org/

Schaeffer-Lacroix, Eva / Céline Largier Vié. "Deutsche Adverbien und Adverbiale aus der Sicht des TreeTaggers". Colloque Deutsche Adverbien und Adverbiale aus deskriptiver, theoretischer und vergleichender Sicht. Maison des Sciences de l'Homme d'Aquitaine, Université Bordeaux Montaigne, 29-30 April 2021. 

Mälzer, Nathalie. "Kultur-, sprachenpaar- und AD-spezifische Einschränkungen bei der Übersetzung von synchronisierten Spielfilmen", Bild Ton Sprachtransfer, Innsbruck 28.10.2021

Schaeffer-Lacroix, Eva. "Korpuslinguistische Herangehensweisen an die Audiodeskription", Bild Ton Sprachtransfer, Innsbruck 28.10.2021

Papers

1. Schaeffer-Lacroix, E., Mälzer, N., Berland, K., & Schulz, S. (to appear 2021). Purpose and features of the TVseriesAD corpus. BfC 2020 Conference Proceedings. 3rd Swiss Conference on Barrier-free Communication., Winterthur (Switzerland). https://doi.org/10.21256/zhaw-3001

2. Schaeffer-Lacroix, E., Magnus, G., & Grass, T. (2020). Usages variés de l’outil d’exploration de corpus TXM dans le contexte de l’allemand. In B. Terrisse, R. Lambertz-Pollan, M. Roudaut, J. Dahm, K. Forbrig, C. Fradin, & G. Zielinski (Eds.), Machines / Maschinen. Les machines dans l’espace germanique: De l’automate de Kempelen à Kraftwerk. (Presses Universitaires de Rennes, pp. 313-328.).

3. Schaeffer-Lacroix, E. (2020). "Integrating corpus-based audio description tasks into an intermediate-level German course". International Journal of Applied Linguistics (Wiley), p. 1-20. DOI: http://dx.doi.org/10.1111/ijal.12294

4. Mälzer, N./ Wünsche, M. (2018). Audioeinführungen. In: Christiane Maaß/ Isabel Rink, Hrsg. (2018), Handbuch Barrierefreie Kommunikation. Berlin: Frank & Timme, 471-486.

5. Mälzer, N., Hrsg. (2016). Barrierefreie Kommunikation – Perspektiven aus Theorie und Praxis. Berlin: Frank & Timme.

6. Schaeffer-Lacroix, E. (2015). Analyse de trois systèmes de gestion de corpus pour l'enseignement-apprentissage des langues étrangères. Alsic 18(1). https://alsic.revues.org/2852

7. Mälzer-Semlinger, N. (2012). Narration or Description – What Should Audio Description ‘Look’ Like? In: Elisa Perego (Hrsg.), Emerging Topics in Translating and Interpreting. Triest: EUT, 29-36.

8. Schaeffer-Lacroix, E., & Berland, K. (submitted). Dealing with variation in audio descriptions scripts. JAT (Journal of Audiovisual Translation), 10.

Working Group Meeting

On November 19th 2020, the first meeting of the TADS-group took place. Due to the current situation the meeting was held on Zoom.

The following members attended the meeting:

INSPE Paris: Eva Schaeffer-Lacroix, Pierre Lignee

University of Hildesheim: Nathalie Mälzer, Maria Wünsche, Sylvia Jaki, Saskia Schulz

The second meeting took place  in Paris, in November 24th-27th, 2021.

It was attended by: Eva Schaeffer-Lacroix, Nathalie Mälzer, Saskia Schulz, Maria Wünsche, Antonia Lins, Thomas Schmidt. 

 

funded by Ministerium für Wissenschaft und Kultur (MWK) Niedersachsen

Cooperation partners

Our cooperations partners are ARTE, ZDF, SRF who provide the team with audio descriptions scripts and audiovisual content. We would also like to thank Beta Film, Sommerhaus Filmproduktion for their support.

VIDEO TO VOICE supports the project by granting us free access to their platform Frazier.

The company develops tools that make audiovisual content accessible to all, regardless of disability. The Berlin-based firm’s software incorporates machine translation technology, allowing audio descriptions to be produced in over 40 languages. 

We would like to thank "Video to Voice" for supporting our project.

Find more information on: https://www.videotovoice.com