Abstract

Keywords:

Abstract

Keywords:

The concept of accessibility has emerged as a cornerstone in the modern drive for inclusivity, transforming barriers into bridges and fostering a more equitable society. Whether in audiovisual media, museums, or even the fashion industry, the goal is to ensure that people with disabilities can fully engage with and enjoy diverse cultural, educational, and entertainment experiences. Media accessibility services such as Subtitling for the Deaf and Hard-of-Hearing (SDH), Live Subtitling (LS), and Audio Description (SD) have proven essential for enabling diverse audiences to enjoy movies, television programs, live events, and more. Similarly, museums have embraced accessibility to ensure exhibits can be experienced by all visitors, incorporating features like tactile displays or musical ekphrases for blind patrons, and live or pre-recorded subtitling services for those with hearing loss. Even sectors like the fashion industry have begun to prioritise accessibility, definitively proving that it is not an obstacle to creativity, contrary to longstanding misconceptions.

This widespread adoption of accessibility has been driven by both technological advancements and changing societal attitudes across European countries. Natural Language Processing, Artificial Intelligence, and Augmented Reality have contributed to making accessibility tools more sophisticated and user-friendly. At the same time, legal frameworks such as the UN Convention on the Rights of Persons with Disabilities and the upcoming European Accessibility Act have reinforced the need for inclusivity in both public and private sectors, thus helping raising awareness in the field.

Despite these advances, challenges remain. Accessibility is often unevenly distributed across regions and industries, with some sectors and countries outpacing others. In Audiovisual Translation Studies, the field has evolved from the exploratory works on media accessibility in the early 2000s to the consolidated standards of the 2010s, with current trends increasingly focusing on the detailed, specific needs of specific users of specific services within specific fields.

As this volume demonstrates, there is no one-size-fits-all solution for accessibility, due to the diverse cultural, linguistic, and technological contexts it must address. By gathering voices from such various landscapes, this volume provides a multifaceted exploration of both the universal principles and the local specificities of accessibility services applied to media, arts, and everyday life. It is divided into three thematic sections, featuring 11 double-blind peer-reviewed articles, stemming from contributions to the International Conference on Accessible Intersemiotic Translation (ICAIT), held in Parma on 31 March and 1 April 2022.

The volume starts with Verónica Arnáiz-Uzquiza and Paula Igareda’s "Old Accessibility for a New Television: Current Accessibility Solutions on VOD, TVOD, and Streaming Platforms", which serves as Introductory Reflections on the strides and shortcomings of media accessibility in the streaming era, focusing on European platforms. Their work highlights the gap between legislative aspirations and actual practices, urging improvements in accessible audiovisual services.

Section 1 - Pre-recorded Subtitling starts with María Labarta Postigo investigating idiomatic translation in SDH with "Improving Subtitles: The Relevance of Translating Idioms for d/Deaf and Hard-of-Hearing People." Through a cognitive and contrastive analysis of English idioms translated into European and Latin American Spanish, the paper reveals how inadequate translation can diminish the viewing experience for deaf audiences. Then, Francesca Illiano shifts focus to Italian SDH in "L’italiano parlato nei sottotitoli per sordi e ipoudenti su RaiPlay: uno studio sociolinguistico per l’accessibilità del prodotto audiovisivo." Her sociolinguistic study assesses the accuracy of spoken Italian subtitling on RaiPlay, considering colloquialisms, dialects, and other linguistic features crucial for retaining the authenticity of the original dialogue. Finally, Jurgita Kerevičienė explores accessibility in Lithuania in "Subtitling for the Deaf and Hard of Hearing: An Atypical Audience for Screen Translation in Lithuania?" She contextualises media accessibility within the country’s post-Soviet evolution, examining societal perceptions of disability and legislative advancements.

Section 2 - Live Subtitling begins with Silvia Martínez Martínez and Vicente Bru García delving into museum contexts in "'En vivo y subtitulado': comparativa de las estrategias de traducción en el subtitulado para sordos de productos audiovisuales museísticos en directo y pregrabados." Their study compares subtitling strategies for live versus pre-recorded museum content, emphasising linguistic simplifications for audiences with limited literacy skills. Then, Alice Pagano bridges the gap between live and pre-recorded translation in "Testing Quality of Different Live Subtitling Methods: A Spanish to Italian Case Study." She evaluates five methods of interlingual LS, from human-mediated approaches to automated systems, analysing their efficacy in live broadcasting scenarios. Finally, Carlo Eugeni and Silvia Velardi investigate technological advancements in SDH and LS with "Il contributo dell’accessibilità per sordi alla resocontazione." Their work traces the evolution of Automatic Speech Recognition technology in Italy, highlighting its dual benefits for accessibility and institutional transparency.

Section 3 - Audio Description begins with María J. Valero Gisbert venturing into novel territory with "Reflexiones sobre la audiodescripción aplicada al cómic." She examines the challenges of adapting graphic novels for blind readers, proposing solutions to integrate visual and textual elements into accessible formats. Then, María Fernández De Casadevante Mayordomo presents cutting-edge applications in "La audiodescripción como modalidad de traducción intersemiótica: novedades y retos." From museum navigation tools to accessible fashion, her paper showcases the expansive potential of intersemiotic translation. Małgorzata Korycińska-Wegner explores innovative approaches in "Alternative Audiodeskription: Die Anwendung der musikalischen Ekphrasis im Hörfilm." Her study of musical ekphrasis suggests that AD can transcend verbal cues, offering aesthetic interpretations of cinematic elements. Finally, Laura Martinkutė closes the volume with "Audio Description of Theatre and Cinema Production in Lithuania: Experiences and Needs of Users." She provides a user-focused study on the challenges and successes of AD in Lithuanian cultural contexts, stressing the role of language-specific adaptations.

Overall, the insights offered here inspired all of us—authors, reviewers, and editors. We are confident they will also serve as a valuable resource for students, scholars, practitioners, and policymakers. They point the way toward a more inclusive future, where customised access to culture and information is a common, everyday right, not a privilege. Enjoy the issue!

Carlo Eugeni & María J. Valero Gisbert

Abstract

The extraordinary technological evolution of the last decade has consolidated the production and distribution of tailor-made and easily accessible quality audiovisual materials that users can manage as they wish. As a result of this technological and sociological evolution, streaming, video and TV on demand (VOD and TVOD) are booming. Media accessibility has been promoted by legislative frameworks in many countries and has been trying to meet the needs of these new audiovisual distribution models, but it has struggled to keep pace. The availability of accessible solutions on many of these new platforms is not new, and both the study of alternatives and the quality of the services provided by the various platforms have grown in recent years. However, accessibility is still far from the 100 per cent mark set in recent norms, and, in some cases, figures have shown only modest growth in the last years. This paper presents the results of a study on the accessibility solutions present on streaming, VOD and TVOD platforms in several European countries and their evolution over the last decade.

Keywords: video on-demand VOD, television on-demand TVOD, streaming, accessibility, evolution

1. Introduction

Although audiovisual content has existed for over a hundred years, in the 21st century it has grown dramatically. Technology, aesthetics and user demands, to name but a few factors, have had an impact on the products currently aired, streamed and released. Screens are no longer flat and static scenarios, but an open window to an increasing number of possibilities, from computers to smartphones and other devices. The rapid evolution of technologies ensures the production of hours and hours of easy-to-access audiovisual materials which are no longer short-lived instant experiences that viewers can enjoy at a specific date and time but on-demand experiences that users can manage at will. Technology has radically altered not only the viewing experience of users but also the quality of their experience.

One of the most striking examples of such a changing and increasing evolution in technology and viewing habits is the rise of streaming, video and television on-demand (VOD and TVOD). They have been present now for over two decades in some countries and provide the most representative example of “controlled viewing”: contents, language combinations, layout aspects and viewing times, among other elements, can be customised. However, such a growing selection does not include a particularly wide range of accessibility solutions.

With the help of technology and the impetus provided by legislation and social awareness, accessibility practices have also adapted in order to meet the needs of the new models. Nevertheless, it has not evolved as fast, which has resulted in part of the currently released audiovisual products lacking adequate accessibility solutions, if any.

The availability of accessibility in many of these platforms is not new in the field, and the study of alternatives and the quality of the services provided has grown in recent years (Arias-Badia 2020a and 2020b; Arnáiz-Uzquiza and Igareda 2013; Arrufat-Pérez-de-Zafra et al. 2021; Bruti 2021; Chapdelaine and Gagnon 2009; Ellis 2015; Ellis and Kent 2015; Ellis et al. 2016; Ofcom 2018; Villena et al. 2012; Samčović 2022), but data differ significantly depending on the accessibility solution provided, the country and/or the platform, with quotas not yet close to 100 per cent and figures moving timidly in the last decade in some cases. The number of TVOD platforms has increased considerably, and different accessibility services have been incorporated, at least partially.

Below we present the results of a study on the accessibility solutions provided back in 2013 and in 2023 by the main TV channels in Spain, France, Germany, Portugal and Italy and of three of the most popular TVOD platforms – Netflix, Amazon Prime and HBO, following, in this case, the line of Blanca Arias-Badia’s study (Arias-Badia 2020a). Both quantitative and qualitative analyses were conducted to obtain information on, first, the accessibility provided to their contents – audios, videos, texts, photos and any other source of information – and then on the amount and quality of the accessibility services on offer.  

2. (Media) accessibility: then and now

In order to outline the evolution of the accessibility solutions provided in the audiovisual context to date, it is necessary to determine, albeit briefly, what accessibility is and / or what needs have arisen and been addressed. There are many different kinds of barriers that can arise when it comes to getting access to audiovisual contents, from the most visible ones – physical – to less visible ones, such as sensory or cognitive barriers. For this reason, and as stated in the United Nations Convention on the Rights of Persons with Disabilities (UN 2006), accessibility must (our emphasis): 

enable persons with disabilities to live independently and participate fully in all aspects of life, States Parties shall take appropriate measures to ensure to persons with disabilities access, on an equal basis with others, to the physical environment, to transportation, to information and communications, including information and communications technologies and systems, and to other facilities and services open or provided to the public, both in urban and in rural areas. These measures, which shall include the identification and elimination of obstacles and barriers to accessibility, shall apply to, inter alia:

a) Buildings, roads, transportation and other indoor and outdoor facilities, including schools, housing, medical facilities and workplaces;

b) Information, communications and other services, including electronic services and emergency services.

In accordance with this framework established by the UN, accessibility to audiovisual content is justified by its link to the right to freedom of expression and access to information on equal opportunities and without discrimination (Storch de Gracia y Asensio 2007). As adopted by the UN Convention, this property/quality must overcome all sorts of barriers and has a significant impact on all areas of life (Greco and Jankowska 2020). Moreover, although it is not restricted to the audiovisual framework, it is on this aspect that we focus our study.

From the point of view of media accessibility, perceived as “a set of theories, practices, services, technologies and instruments providing access to audiovisual media content for people that cannot, or cannot properly, access that content in its original form” (Greco 2016: 11), there are various barriers to be overcome, the most notable being those of a linguistic, cognitive or sensory nature. While the first has traditionally been addressed by translation studies and audiovisual translation, somewhat ignoring its basic function as an accessibility solution, the latter two have only been more thoroughly analysed in recent times. Thus, as in many other fields, the need to guarantee access to information pushed professional practice ahead of the academic, regulatory and legislative contexts, leading to the gradual development of different types of audiovisual translation and media accessibility. In a regulatory framework still to be defined in many cases, the most popular – somehow “traditional” – modalities were originally scarce and limited their presence, in most cases, to subtitling for the deaf and hard-of-hearing (SDH), sign language interpreting and audio description (AD) (Ellis 2015), with an uneven presence, but generally extremely restricted. Subsequently, with the adoption of accessibility guidelines in different countries (ANACOM 2011; BOE 2010, 2022; GU 2016) and the proliferation of academic studies, the focus aimed at shifting from quantity – established in a normative/legislative way – to quality, with diverse initiatives from different fields (Richart-Marset and Calamita 2020; Romero-Fresco 2015).

From a practical point of view, the first technically available accessibility solutions were SDH and sign language. However, the ad hoc production of audio descriptions or audio subtitling solutions to overcome communication barriers in audiovisual media or many other varied settings, from performing acts to museums, or the most traditional cinema screens, is well known and is still common practice. In recent years, we have seen how the scenario has expanded to guarantee full access to the media through the application of these same modalities to other contexts, together with the irruption and / or rise of other lesser-known and/or less widely used modalities, as is the case of Easy-to-Read (E2R), Plain Language (PL) and Easy Language (EL) – adapted to audiovisual texts or not – easy-to-understand accessibility services, accessible filmmaking or typhlological solutions, to name but a few (Greco and Jankowska 2020; Richart-Marset and Calamita 2020; Pena-Díaz 2023).

The progressive incorporation of audio subtitling, respeaking, easy-to-read and plain language, among others, to the bulk of modalities and services that have been emerging to overcome cognitive and sensory barriers has enriched the accessibility toolbox currently available to provide media accessibility solutions, but it does not yet guarantee their implementation in the audiovisual scenario.

The European Accessibility Act (EAA), also known as the Directive (EU) 2019/882, was established to address accessibility requirements for products and services. It stems from the EU and member states' commitment to accessibility after ratifying the UN CRPD (Committee on the Rights of Persons with Disabilities). According to Recital 23 of the Audiovisual Media Services Directive (AVMSD) 2018, achieving accessibility in audiovisual media services involves various measures such as sign language, subtitles, spoken subtitles, and audio description. Member states are required to ensure that media service providers report regularly to the national regulatory authority on their progress in making their services accessible to people with disabilities. However, the AVMSD does not specify the percentage of content that should be made accessible. Member states had until 28 June 2022, to transpose the EAA into their national legislation, and the measures will take effect from 28 June 2025. Some countries have gone beyond the literal transposition of the Directive and implemented specific time slots or quotas for accessible content to meet the accessibility requirement of Article 7(1) of the AVMSD.

The countries analysed in this article have specific regulations regarding accessibility for audiovisual media services. In Germany, for example, the Interstate Treaty on Broadcasting and Telemedia, along with the Act on the Broadcasting Corporation under Federal Law (DWG 2004), govern accessibility rules. Broadcasters and telemedia providers are required to offer accessible options within their technical and financial capabilities.

In France, the accessibility regulations are outlined in Law No. 86-1067 of 30 September 1986 on the freedom of communication (Loi Léotard). Agreements between audiovisual media service providers and Arcom determine the accessibility proportions of programs, especially during peak viewing times, for the deaf or hard of hearing. Broadcasters with a significant audience must make their programs accessible, while video-on-demand services have specific requirements.

The regulatory authority in Italy is AGCOM (Autorità per le garanzie nelle comunicazioni), responsible for ensuring compliance with accessibility measures.

Portugal has a range of legislative acts governing broadcasting, including the Television and On-demand Services Law and the Electronic Communications Act. The ERC (Entidade Reguladora para a Comunicação Social) is the media regulator, while ICP-ANACOM is responsible for electronic and postal communications. Accessibility measures in Portugal include subtitling, sign language interpretation, audio description, and user-friendly navigation menus,and non-compliance with accessibility obligations can result in fines (Cabrera Blázquez et al. 2023).

Finally, in Spain, accessibility requirements were only applicable to public linear television channels until the 2010 law (Ley 7/2010), and providers of pay channels and video-on-demand were not subject to these obligations, so it was up to the operators whether or not to include accessibility services (CNMC 2022). The General Law on Audiovisual Communication (LGCA - Ley 13/2022), however, regulates the right of people with visual or hearing disabilities to universal accessibility to the audiovisual communication service, and it substantially expands both the accessibility obligations that had been required and the set of audiovisual providers with the obligation to comply with them (CNMC 2022).

3. Media accessibility in practice

While the regulatory and legislative framework for accessibility to audiovisual content has evolved from an incipient need a decade ago to a present reality at an international level, its implementation has witnessed an uneven pace, often conditioned by the technical / technological evolution and by the changing audiovisual consumption habits of the target audience (Arana, Mimemza and Narbaiza 2020; Bolognesi 2023; Navarro and Vázquez 2020; Telefilm Canada 2021).

Boosted by the evolution of technology, the proliferation of media formats has not only exponentially multiplied the number of audiovisual products available, but it has also been the door to a more widespread presence of accessibility, progressively promoting greater visibility of less widespread practices.

First television and then video some years later have traditionally been the main formats providing accessible audiovisual contents to audiences. Their role, as a tool for social inclusion and equal access to information, in the case of the former mainly, has made them a key element in terms of accessibility (García-Prieto and Aguaded 2021). From the first steps in the 1970s until today, the progressive incorporation of accessibility services in the portfolio that television channels have been offering has continuously increased, and numerous studies have addressed the situation in recent years. The proliferation of studies on general media accessibility (García-Prieto 2018; Rovira-Esteva and Tor-Carroggio 2019; Pena-Díaz 2023) and, more specifically, on the presence of subtitling services, audio description, sign language interpreting (Gil Sabroso and Utray 2015; Bosch-Baliarda et al. 2020:38), audio subtitling (Beseghi 2023) and, more recently, the incorporation of easy-to-read (Bernabé and Orero, 2020) and respeaking (Romero-Fresco 2011) have led to a changing scenario over the last decade.

In the case of video, on the other hand, although its birth, and more specifically the irruption of DVD technology, represented an extraordinary milestone in terms of accessibility with the possibility of incorporating additional content, as was the case of SDH and AD services, the last few decades have witnessed the progressive extinction of traditional physical formats. The different digital supports that emerged in the first decade of the 21st century – DVD, Blu-Ray, HD-Blu-Ray – have now massively been replaced by "non-tangible" products due to the gradual technological virtualisation of new management, dissemination and consumption platforms. Because of this radical transition, we will focus primarily on television in the text below. Its evolution, although also subject to important changes over the last decade, continues to preserve its more traditional essence even today. But the audiovisual scenario has seen the rise, in the last decade, of emerging services that now accompany – or even replace – the leading role of television. In the last decade, and especially boosted by the COVID pandemic, over-the-top (OTT) services have progressively gained ground in the world of user preferences, with over 1.8 thousand million subscribers (Forbes 2024), which means more or less 2.3 platforms per user. According to some studies (SigmaDos30’ 2022; Variety 2022), Netflix (72 per cent), Prime Video (67.8 per cent) and HBOmax (31.3 per cent) are the ones with most users, which is one of the main reasons why we chose them to carry out our research.

But although born at the end of the 2000s, only a limited number of studies have been carried out to date on the accessibility of these not-so-new audiovisual content platforms (Arias-Badia 2020a; Ellis 2015; Herrera-Crespo 2020; Scope 2023), focusing, for the most, in the situation of English-speaking countries. Back in 2015, Katie Ellis (2015) identified the lack of accessibility of some platforms at the time when they were already offering their services in many countries. She mainly highlighted the lack of AD and the scarcity of titles with SDH primarily in countries such as Australia and the US.

Years later, in a study conducted by Cecilia Herrero-Crespo (2020) on the accessibility of the most popular video-on-demand platforms in Spain (Netflix, HBO, Movistar+ and Prime Video), the author concluded that none of them fully complied with the international standard and norms in force on accessibility, much in line with the results presented by María Asunción Arrufat-Pérez-de-Zafra, Liliana Herrera-Nieves and María Agustina Olivencia-Carrión (2021). These authors conclude that in order to reduce the digital gap that occurs on platforms such as Netflix, it is essential to increase the existence and quality of resources that facilitate accessibility. There is also a need for consistency between the needs of subscribers and what complies with the regulations of each country, and an increase in SDH content in more languages, titles with AD, sign-language interpreted and easy-to-read content, as well as improved interaction with the interface on different devices.

Despite the limited examples of research initiatives focusing on the accessibility of streaming platforms at an international level, there is a comprehensive study conducted by Arias-Badia (2020a), who carried out a quantitative study showing the scenario of accessibility services in Spanish and Spanish Sign Language (LSE) of seven platforms accessible from Spain out of a total of 20,661 titles. In November 2020, only AppleTV+ and Prime Video offered specific search filters; Netflix offered them for AD and subtitling. In the other cases (Filmin, Movistar+, HBO and Disney+), the results are based on a detailed analysis of a random sample of titles on each platform. Less than three per cent of the content available on the platforms were then audio described in Spanish, and around 80 per cent were subtitled (but only 7.85 per cent is SDH-specific). LSE interpreted content was less than two per cent. According to this study, in 2020 Filmin, HBO and Disney+ had no accessibility services. Prime Video had search filters, but they were not easily accessible. Netflix also provided search filters for AD and subtitling but did not give specific SDH data, although it was known that there were products that did (mostly local productions). In any case, depending on the device used (SmartTV, mobile phone, tablet), the results were different. Movistar+ made a strong commitment to accessibility with its 5S app, but it failed when it mixed AD and SDH in its search filter, although LSE was with an additional subscription. Arias-Badia (2022) herself concludes in a subsequent article that, as a general rule, platforms have failed to integrate services that enable people with hearing or visual loss to access content and that accessibility is a pending task for video-on-demand platforms.

4. Materials and methods

In order to compare the evolution of accessibility services on television and streaming platforms from 2013 to 2023, our study aimed at focusing on five European countries – Spain, France, Germany, Italy and Portugal – and the most relevant streaming platforms, as previously mentioned: Netflix, Amazon Prime and HBO. Data collection involved, on the one hand, a quantitative data collection process to determine the number of channels present in each case, and, on the other, a qualitative study to determine the type of accessibility solutions offered by each one in each case. Once the channels had been identified, a second qualitative analysis was carried out to identify the main characteristics of the solutions available. The process was then replicated in the case of streaming platforms in order to collect a representative sample in all cases.

5. TV accessibility results of our 2013 study

With the aim of outlining the situation of TV broadcasts and on-demand services in terms of accessibility, back in 2013 we carried out a study (Arnáiz-Uzquiza and Igareda 2013) focusing on the level of accessibility of broadcasters’ websites and examined the level and volume of accessibility and services offered in some European countries: Spain, France, Germany, Italy and Portugal. We found that there was a general lack of international icons in subtitles for the deaf-and-hard-of-hearing, sign language interpreting, audio description and audio subtitles services on TV and Internet broadcasts in most countries.

At the moment of the first part of our study, two Internet release formats were already available through broadcasters’ websites: live and on-demand television. With regards to the former, despite having a longer tradition based on more traditional analogue and digital terrestrial television (DTT), and so presumably, and in reality, more widespread and easy to access, live content broadcasts were available in all the countries. Nevertheless, viewing was only possible under regional restrictions for most, preventing access from foreign locations. However, and even if regionally-limited, accessibility services – mainly SDH – were guaranteed in live broadcasts in most countries, with some rare examples of SLI, mostly in Portugal, but rarely providing AD services.

With regards to the latter, although the first examples, and later explosion, of on-demand television date back to the turn of the present century (Castro and Cascajosa 2020), and despite the fact that by 2013 this was a relatively widespread form of parallel broadcasting among the different television channels, as we will see, programming was still more restricted and scarcer at the time when compared to the most traditional TV formats. Although on-demand television is, or should be, from a practical point of view, a variant of a similar nature to traditional television in which “only” the form of consumption changes, there were major differences a decade ago, as we will see.

Fig. 1: Accessibility of TV in 2013.

At this stage, in the case of Spain there were nine TV Stations with 28 channels, where only 20 out of the channels (from three stations) were accessible on their ordinary TV broadcasts (See Figure 1). No information on the access services provided by TV stations was announced, and only some TV stations included on their Teletext pages brief information announcing, almost exclusively, the SDH service offered in their programmes, which were mostly restricted to national productions. As was ascertained during the manual analysis, in other countries, such as France, Germany and Italy, slightly more information was provided on the access services of the TV stations. When it came to on-demand TV, most stations, and 17 out of the 28 channels, provided these services through their websites, but none of them included accessibility services – be it sign language interpreting, audio description or subtitles for the deaf and hard-of-hearing – to on-demand contents, although they did for ordinary (live and/or prerecorded) TV. After the manual analysis, a detailed analysis of all the channels was carried out, looking closely at the programming on offer, from films to TV series, shows and news programmes. We then found that not a single channel provided accessibility services and only one example of sign language interpreting was identified for parliamentary sessions and for a pre-recorded specialised programme (30 minutes a week) aimed at hearing-impaired audiences in the State TV channel La2.

In France, the State France TV, with six channels, included open information on accessibility to their website for the hearing-impaired, visually impaired and motor impaired. Nevertheless, only three out of the six TV channels, France 2, 3 and 5, provided subtitles for the deaf and hard-of-hearing – the most common accessibility solution traditionally offered – for their on-demand broadcasts.

Germany, with ten TV stations and 21 channels, only had ten channels which offered on-demand and streaming TV, and in only three of them – Das Erste, MDR and ZDF –were contents accessible, with just Das Erste providing audio description services the only example in the countries studied to do so.

In the case of Italy, the macro-station Rai hosted 15 TV channels a decade ago, with only three of them – Rai 1, Rai 2 and Rai 3 – including accessible on-demand contents, mostly limited, like in most countries, to programmes originally broadcasted by Rai TV in their traditional formats, with accessibility restricted to SDH services.

Portugal, the only country from a subtitling tradition included in our study, showed significantly different results. The state station RTP, with nine channels, only had some limited examples of accessibility solutions in two of its channels. While whit may not seem much, it should nevertheless be noted that RTP Play live provided sign language interpreting in their broadcast, a rare example in our study, in which SLI was seldom included in the accessibility portfolio. However, when it came to on-demand TV, no accessible programmes were found back in 2013.

The study of the state-of-art of accessibility to TV broadcasts back in 2013 showed there was a poor panorama in relation to on-demand TV since just nine channels (three in Italy, three in Germany and three in France) provided some kind of accessibility service. In Spain and Portugal, however, the only examples identified of accessible contents were justified by the nature of the programmes – aimed at hearing-impaired audiences, or live broadcasts with a long tradition including sign language interpreting. Nevertheless, the most surprising fact, in terms of accessibility, may be that most on-demand products offered by the different stations through their websites across the different countries, which had been aired through ordinary release formats and were originally made accessible even on live broadcasts, seemed to be later distributed on-demand without the original access services.

But with the change in technologies and audiovisual consumption habits, to some extent conditioned by the new regulatory frameworks, one would have expected the accessibility scenario to have changed over a decade.

6. TV accessibility now

As outlined in previous sections, over ten years later, and subject to a different regulatory framework in many cases, one would expect a different and more positive TV scenario in terms of accessibility at all levels. On the one hand, the continued moves of the audiovisual market with the increase in groups and TV channels would suggest a significant increase in the presence and offer of television products. On the other hand, the greater portfolio of accessibility solutions on the market, linked to the regulatory and legislative impositions in terms of accessibility, seemed to build the ideal context for the proliferation of accessible products.

An analysis of the results obtained in 2023 reveals how, in the case of Spain, although the number of groups and channels available has not changed significantly over the last decade, of the nine groups that are currently present in the Spanish audiovisual market, five of them, that is 21 channels, offer live television via their websites, both through live and on-demand services. Unlike a decade ago, when channel websites rarely included information on the accessibility solutions available in their schedules, today we find that in most cases channels do announce the accessibility mode available, although this information is not always easy to access for users, who often have to enter the secondary menu of the programmes to get this information. While it is true that the volume of accessible production on television has grown progressively over the last decade (CNMC 2020) – especially in the case of SDH – it should be noted that, also here, the 100 per cent quotas have not yet been reached in their production, mostly focused on films (42 per cent), news and sports (17 per cent) and entertainment programmes (26 per cent – 37 per cent). However, it should also be noted that these figures only apply to live broadcasts delivered via traditional formats, DTT, because today, as was the case in 2013, Internet broadcasting of on-demand programmes that were originally delivered through live emissions is not accompanied by the accessibility solution originally available.

In the case of France, the television offer has also increased significantly over the last decade, and there are now 28 television channels, which belong to eight media groups, responsible for broadcasts. As was the case in 2013, access to the content of foreign television channels continues to be restricted for geographical reasons, making access to live broadcasts very difficult, and limiting – if not impeding – access to on-demand broadcasts. Despite this major obstacle, the study of the audiovisual offer currently available highlights the presence on the state channels of news programmes that were originally part of live broadcasts and are later made available on-demand, and that include SDH solutions, a practice which is not common in other cases, such as Spain for example.

In Germany, the television offer is currently made up of 27 television channels belonging to 12 media groups. Although it is not possible to access the live production of any of their broadcasts, 11 of the 27 channels have on-demand content accessible mainly through subtitling for the deaf and hard-of-hearing and audio description. Also, in addition to the exceptions identified in the 2013 study, where audio subtitles were included, there is now Easy Language in the 2023 offer, which is included by the ProSiebenSat Media group in two of its channels. This is an exception to the standardised offer of the different European channels included in this study.

Finally, of the 22 television channels available in Portugal, owned by four major media groups, only two offer some content accessible via subtitling for the deaf and hard-of-hearing, audio description and/or sign language – in radically different proportions – through their on-demand broadcasts. As in the 2013 study, in those channels where it was possible to access live broadcasts – only six out of the 22 – it was found that none of them incorporated any accessibility service at all, which paints a poor picture in terms of accessibility.

Fig. 2: Accessibility of TV in 2023.

If we compare the 2013 and 2023 data, we can see that the audiovisual offer has changed considerably, with a significant increase in terms of TV channels over the last decade. Accessibility, however, has not improved to the same extent. Although the overall situation has improved over the course of this decade, the presence of accessibility solutions is still limited and uneven in the television market, and so it still needs to be improved.

Fig. 3: 2013-2023 Accessibility of Television.

However, as previously pointed out, the irruption of streaming platforms has made TV broadcasters more open to this new form of consumption and distribution, and they are attempting to narrow the marked technological divide between the two. But what is happening in terms of accessibility in the streaming context?

7. Accessibility in streaming platforms

In contrast to the previous sections, where we were able to make a comparison of the analyses carried out in 2013 and 2023, in the case of streaming platforms a comparison is not possible because, while already present back in 2013, they did not have the leading role they have now, and the minor role they did have was not considered in the first part of the study. Nevertheless, based on the study conducted in 2020 by Arias-Badia, who presented the situation regarding the accessibility of the main streaming platforms in Spain, we sought to identify a change compared to 2023, possibly conditioned by a number of causes, such as the COVID pandemic. 

If the user wants to know the titles provided with some type of accessibility, we found that only Netflix gives us the option to filter that search from its web, but only with audio description. The other two platforms studied have no filters at all, and the user has to play the title and then discover the options of accessibility, but inside the menu of the product, which is not particularly accessible. Only Prime Video, and very recently Netflix, put the options that users have in the description of the product so they do not need to play the content and have the information beforehand (i. e. if there are several audio and subtitle options, audio description or not, spatial audio and quality of image). Netflix has even recently rolled out customisable subtitles for TV.

Our study analysed 42 audiovisual products – 14 in each platform – of different genres (see Figure 5), totaling 2409 minutes, mainly from the last ten years. We replicated the quantitative analysis adopted for televisions, and much in line with Arias-Badia’s study (2020). As a result, we found that Prime Video offered some kind of accessibility in all the samples analysed – mainly AD and SDH – but surprisingly only 30 per cent in Spanish (compared to 15 per cent of Italian and 69 per cent in English). Netflix presented a similar situation with worse results regarding Spanish AD (seven per cent) or SDH, while HBO presented a poor accessibility offer (15 per cent) and everything we found was only in English.

Fig. 4:  Accessibility of Streaming Platforms.

Most audiovisual platforms have a portfolio of international titles whose language and accessibility solutions are common in most cases. As we have identified as a result of our analysis, the presence of AD and SDH in English is common in most titles. It is true that, when platforms offer SDH, they usually do so in the source language of the audiovisual product, i. e. intralingual subtitling. And, in that context of the dominance of audiovisual content produced in English-speaking countries (Arias-Badia 2020b), this is a barrier to access for users who do not have the corresponding linguistic competence.

However, the situation changes in the case of titles produced out of the English-speaking countries. In this case, the main accessibility solutions provided, which are SDH and AD as SLI is only rarely present, are mainly offered in the language of the country concerned (Spain, France, Italy and Portugal, in our study).

While in Netflix it is possible to find these versions in the language of each country and sometimes accompanied by their English versions, this is not the case for Amazon and HBO, where the provision of accessibility solutions for national products is present, but only the national language is provided in the case of Amazon, but is rarely included by HBO, much in line to the situation described by Arias-Badia in her study.

In relation to the accessibility provided per audiovisual genre, we have to say that, in the case of the platforms analysed, there are some products such as variety programmes, news or game shows that are not present as they are in TV.

Fig. 5: Accessibility provided per audiovisual genre and platform.

Although not presented in the current study, we also analysed different practices in subtitling for the deaf and hard of hearing because we wanted to compare the way platforms and TV make audiovisual content accessible and to investigate whether these platforms follow the standardised style guides provided. Further research will focus on these aspects.

8. Conclusions

The continuous political, social, cultural and, above all, technological changes in recent years have made accessibility evolve considerably, making a discipline whose practice and study began decades ago be considered by many a young and incipient field of studies due to the radical changes it continuously undergoes.

Throughout these lines we have reviewed how the national and international legal contexts have been expanding and adapting to accommodate and guarantee media accessibility. Beyond its consideration as a tool to guarantee access to information in the different countries, legal frameworks have been adapted not only to enable access to services but also to ensure these are accessible to the majority of the population – considering diverse needs – through a growing portfolio of services, adapting to new platforms and increasingly concerned with quality. However, the pace of these policy changes often lags well behind commercial practices, and the rapid pace of technological changes and new market needs often precedes the regulation of practices.

While the audiovisual market has always been subject to a continuous technological evolution, and the changes have also led to continuous modifications not only in the products and services but also in the habits and tastes of the target audiences, the speed of the technological changes over the last decade has completely transformed the scenario. Traditional physical media have progressively disappeared, being replaced by their digital versions, and traditional broadcast formats, as in the case of television, have diversified their release formats to adapt to the changes, and trying to guarantee, if possible, the continuity of the accessibility services that were traditionally offered in each case.

As we have seen throughout this study, the first steps in the diversification of the TV offer, back in 2013, revealed an uneven panorama in the European context. Traditional broadcasts – DTT – still far from being 100 per cent accessible, focused their efforts on SDH as the main accessibility service, with a limited presence of other modalities that were also widespread, such as AD or SLI. However, television broadcasts through the Internet also revealed a significantly different panorama, with a much more limited – or non-existent – offer of accessible products: of the 41 television channels available in the five countries under study – Spain, France, Germany, Italy and Portugal – only nine, in three countries, offered accessible on-demand contents, at least part of them, back in 2013, barely 20 per cent of the total production on offer. It is particularly significant that, in many cases, part of the productions, coming from live broadcasts of the same TV channels and being originally distributed with accessibility solutions in their live broadcasts, did not incorporate accessibility services into their on-demand offer.

More than a decade after the first part of the study was conducted, in an, in principle, more favorable general context, one could assume a massive presence of media accessibility. However, although the most notable difference lies mainly in a greater offer and variety of audiovisual products and a greater diversification of the forms of distribution of these products, media accessibility is still far from being massively present. Now, as then, DTT still coexists with Internet broadcasts, both for live television and on-demand services. Nevertheless, it is worth noting that the latter has gained weight among target audiences, by providing a new way of accessing television products and building a bridge to the new forms of audiovisual consumption enabled by streaming platforms.

In a different television landscape, with a greater number of channels – over 60 ­– and more sensitive regulatory frameworks to ensure not only the presence but also the quality of accessibility services, the role of accessibility solutions available on television in Europe is still uneven. The massive presence – but still not 100 per cent guaranteed – of SDH has not yet been transferred to accessibility modalities, such as AD, SLI or easy-to-read language, for users with other sensory or cognitive profiles. While these are the main and/or most widely used modalities, the presence of others, such as audio subtitling, is exceptionally rare in most countries.

Although the international legislative frameworks have been forcing media accessibility quotas to reach all audiovisual production, we have seen that, even if from a somewhat more privileged situation, the focus still remains on guaranteeing the presence of all accessibility solutions in audiovisual products in all broadcast formats, whether via television or streaming platforms. In this context, and although it has always been an outstanding field of study for academia, at present quality is a secondary objective in most cases. With the progressive rise of accessibility quotas, the focus will hopefully shift to quality.

Abstract

This paper addresses the issue of the accurate translation of idioms in subtitles, towards a greater understanding of how access to multimodal texts such as film and TV series for people with hearing disabilities can be enhanced.

Captions and subtitles help the d/Deaf and Hard-of-Hearing to enjoy films, TV series and other multimodal texts with audio. They convey dialogue and provide other information, such as plot-related sounds and the kind of music being played. Without such details, watching most TV series and movies would be an unfulfilling experience for people with hearing disabilities. Yet it is clear from almost any internet forum on the subject that the quality of captions and subtitles continues to be an issue.

This paper analyzes the translation of figurative language TV series, focusing on the rendering of idioms from their original (English) version into Spanish subtitles. For the purposes of the analysis, a corpus of TV series aired on the Netflix and Amazon Prime streaming platforms will be used; this includes series using both European and Latin-American Spanish. These data are described from a cognitive and contrastive perspective, following the methodology developed by Labarta Postigo (2020, 2021). The aim of the study is to shed light on the translations themselves, and to explore differences between the original dialogues and the translated subtitles from a metaphorical and idiomatic perspective.

Results of the analysis show that the figurative meaning of English idioms is considerably reduced in the translations into both varieties of Spanish. This entails a considerable loss of meaning, especially for people with hearing disabilities, who do not have the additional information conveyed through actors’ voices, intonation and accent, a significant deficit in that idiomatic utterances often reflect important features of the character in question, such as sociolect, style and sense of humor.

Keywords: multilingual translation, subtitling, idioms, internet-TV series, people with hearing disabilities

1. Introduction

Hearing loss can happen to anyone. According to the World Health Organization[1] more than 1.5 billion[2] people worldwide are currently affected by some sort of loss of hearing; this figure is on the rise and could exceed 2.5 billion by 2050. In addition to genetic predispositions and pathologies, the excessive loud noise we are often exposed to in our daily lives is also a cause of hearing loss; another is aging, since the chance of developing hearing loss increases as we get older.

Taking these factors into account, it is the case that an increasing number of people rely on subtitles when watching films and TV series. People with hearing loss may find it difficult to understand the dialogues of a movie or TV series, and subtitles are a great help here. In the present paper we consider Spanish-language subtitles in original English TV series. The potential viewers of these subtitles are, on the one hand, those who do not understand the original language of the series (or don’t understand it well) and who, despite listening to the dialogue in English, need to read a Spanish translation as they watch; on the other hand, those who cannot listen to the audio, in either English or Spanish, because of hearing problems. In Spain alone there are 1,230,000[3] people with different types and degrees of hearing loss or deafness.

If we look at the audio and subtitle options offered by TV and streaming services, we increasingly find audio descriptions in various languages, which allow access for blind and visually impaired people. The number of series offered in two distinct versions of Spanish (European and Latin-American) has also increased in recent years. Depending on the provider, these two versions are labeled differently. For example:

• Netflix: “European Spanish” and “Spanish”
• Amazon Prime Video: “Spanish (Spain)” and “Spanish (Latin-America)”

However, most series don’t offer a special version of translated subtitles for the d/Deaf or Hard-of-Hearing (HoH).[4] This means that both those who are d/Deaf or HoH and those who can hear the full soundtrack perfectly share the same subtitle options. Whereas accurate translation into Spanish is crucial for any subtitle users, it is especially important for those who cannot hear intonation, the voices of characters, or the soundtrack. Idiomatic expressions in subtitles, as in the dialogue itself, often provide essential insights for the audience, providing important information at the level of implicit meaning, such as irony, humor, or the personal characteristics or nature of a particular character. This paper considers the importance of accuracy in the translation of idioms in subtitles, towards enhancing access to TV series for people with hearing disabilities.

The study takes a qualitative and contrastive approach to figurative language in the form of idioms in subtitles, exploring the metaphorical dimension of idioms in English series and their translation as Spanish subtitles. A further goal of the analysis will be to look at the two varieties of Spanish in question, for which we will follow the terminology used by streaming companies to refer to these varieties: European Spanish (ESp) and Latin-American Spanish (LASp).

The paper is divided into six sections. In what follows an overview is given of idioms and their importance in translated subtitles. Section 3 then discusses different potential addressees with hearing disabilities. In section 4 the theoretical framework for the study, plus the corpus, are described. The following section presents the analysis and the results, illustrated with examples for each translation strategy. Finally, section 6 offers conclusions and some suggestions for further research.

2. The importance of idioms and their translation

According to Maria Labarta Postigo, idioms are considered by providers like Netflix to be one of the greatest challenges for translators (Labarta Postigo 2021: 3). Idioms, as with other figures of speech, transcend the literal, denotative meanings of words and phrases to provide readers with new cultural and imaginative insights. An accurate translation of these figures of speech will greatly improve the quality of subtitles and contribute to a better understanding and reception of the film material by the audience.

In order to establish the limits of the current research aims, I will provide a definition of idioms and an illustrative example. As noted in previous studies (Labarta Postigo 2020, 2021) there are many different definitions of these types of phraseological units. I will follow here Anglo-American thinking, drawing thus on a definition provided by Chitra Fernando and Roger Flavell (1981: 48): “A pure idiom is a non-literal set expression whose meaning is not a compositional function of its syntactic constituents, but which always has a homonymous literal counterpart”. In other words, idioms are set expressions whose overall meanings are not deducible from the composition or sum of their constituent parts. Most idioms are metaphorical and have a figurative meaning. In addition, they always have a homonymous literal counterpart, as the following example illustrates:

Table 1: Literal and figurative meaning of an idiomatic expression

As we can see in the above table, the metaphorical expression may add information, as well as touch on emotion and expressiveness, rather than merely convey the explicit content.

In Labarta Postigo (2020) the importance of idioms in daily communication and in film dialogues was explored. It was stressed that idioms, like other phraseological units, constitute an essential tool for characterization. The fictional dialogues in films and series also use idioms to reflect the reality of verbal communication (the conversational one). They can thus be seen as a resource to enhance expressivity in dialogue, as well as being a linguostylistic device to better convey the specific traits of characters in both audiovisual and literary texts. In addition, they may have an entertainment value and a role in conveying humor. Idioms can provide essential insights through text and convey important information at the level of implicit meaning, e.g. irony, humor, or elements of a character’s specific nature or make-up. It is very often the case that idioms are used to create a special impact in the audience in some way, and thus to ensure their attention.

Considering all these features, idioms have the potential to be far more expressively powerful than the explicit expression of meaning. As we have seen in the example above, the idiom “he/she knocked it out of the park” has the capacity to say far more, that is, to be more expressive and powerful, than simply “he/she was great”.

3.The d/Deaf and Hard-of-Hearing (HoH) as addressees of subtitles in streamed series

Since the 70s special SHD (Subtitles for the HoH and the d/Deaf) have been developed for TV and other media. These subtitles were created first in the US and incorporate the advantages of open subtitles (for addressees who do not speak the language of the series) and close captions (CC, which provide text for all audible information, such as significant noises, sound effects and music, and are aimed at d/Deaf and HoH addressees). So, SHD offer the information typical of CC, as well as all other audio-related information for an audience who doesn’t speak the language of the original audio.

Despite SHD having already been developed, and being offered widely in video games, most of the series and movies provided by the major streaming companies do not offer special versions of subtitles for people with hearing disabilities. There are economic motivations for this. On the one hand, people with hearing disabilities are in the minority, and on the other, there is a wide spectrum of hearing impairments, from HoH to Deaf. According to Josélia Neves (2005: 309-310):

While it is still commercially unviable to offer a number of different SDH subtitling solutions for the same product, SHD will have to provide subtitles that will be useful to a wide variety of addressees, including hearers, as is the case of people who use SHD for language learning or because of environmental noise, for instance. It is a fact that, by providing subtitles for all, not everybody is getting their due.

In another study, Neves (2007: 98) states that although open subtitles in foreign audiovisual products are clearly insufficient to meet the needs of the d/Deaf, they do provide the basic information for these viewers to be able to follow the content. Therefore, so-called “open subtitles”, that is, general multilingual subtitles, are normally the only way to access film material in a foreign language, and thus are used by hearing, d/Deaf and HoH viewers.

To corroborate this, I surveyed 14 d/Deaf and HoH people who communicate in different kinds of language, such as sign language as their first language, lip readers, users of oral language, etc., about the kind of subtitles they use when watching TV series. Their responses indicated that they usually access a foreign film or series using the subtitles offered in their native language, Spanish, thus using the same subtitles as people without hearing problems. A majority (86 per cent) reported using these Spanish open subtitles, whereas 14 per cent used either English CC or Spanish open subtitles depending on the series.

Everything points to the fact that open subtitles (hereafter subtitles) will continue to be the main form of access for the greatest number of addressees, including the d/Deaf, HoH, and listening ones.

In order to obtain further information about addressees’ preferences with regard to figurative language in the translated subtitles, I conducted a pilot survey with the same 14 d/Deaf or HoH respondents[5]. They were asked to complete a 10-item questionnaire about the kinds of subtitles they preferred. Each question had two options for Spanish subtitles translated from English: one contained an idiomatic expression with a figurative meaning; the other was a non-idiomatic expression and the translation was of the explicit meaning of the original English. The goal of this survey was to reveal the opinions of these viewers of series with a hearing impairment and who use the subtitles designed for all audiences. More than half of the respondents expressed a definite preference for subtitles with idiomatic expressions. Eight of 14 chose idiomatic expressions in between 70 per cent and 100 per cent of cases, while for the other six respondents, this preference fell to 40 per cent or less. Furthermore, there were some additional comments about accurate translations. For example, one respondent observed that the translations in the Spanish subtitles should be as close as possible to the original dialogue in English and not simplified in any way.

Of course we cannot draw firm conclusions from such a small survey, but it does give us an indication of the preferences of this specific group of subtitle users/readers, the majority of whom seen to prefere metaphorical translations of the original English idioms. However, more extensive and detailed research would be needed to draw solid conclusions here.

4. Methodology and corpus

The theoretical framework of the current research is based on George Lakoff´s Conceptual Metaphor Theory (Lakoff 1993; Lakoff and  Johnson 1980); it also embraces Raymond W. Gibbs’ work on metaphor and culture (Gibbs 1996, 2001; Gibbs, Bogdanovich et al. 1997) and on idioms and idiomatic/literal meaning (Gibbs, Nayak and Cutting 1989; Gibbs and O’Brien 1990).

Labarta Postigo (2020) developed a methodology to analyze translation strategies used in producing English subtitles for original movies in Spanish and German, drawing on approaches to various translation types and goals (Baker 1992; Conca and Guia, 2014; Corpas, 2000; Díaz Cintas and Anderman, 2009; Díaz Cintas and Remael, 2014; Gottlieb 1992, 1994; Newmark 1988; Oltra, 2015; Trim 2007. In Labarta Postigo (2021), I applied the methodology focusing on the four most frequent categories or strategies, in an analysis of TV series: Total Equivalence (TE), Partial Equivalence (PE), Omission (O), and Explicit meaning (EM). Drawing on this previous work, the current analysis will include two parts: a presentation of the strategies based on a qualitative analysis, followed by a comparative analysis of the translations from English into ESp and LASp.

In the qualitative analysis examples will be given of the following strategies, which use figurative meaning in the translation in the first place:

• Total equivalence (TE): This is, of course, the ideal translation, in which an exact metaphorical equivalent in the TL is found.
• Partial equivalence (PE): Somewhere between total equivalence and an absence of equivalence. The strategy involves translating the idiom of the source language (SL) into a similar, but not identical, idiomatic expression in the target language (TL). The two idioms have identical target domains, but do not fully correspond in their metaphorical structure, image, or source domains.

Following this are examples of strategies in which there is no idiomatic expression in the translation:

• Omission (O): to omit an idiom from the SL in translating it into the TL.
• Explicit meaning (EM): to leave out the idiom from the SL and to translate it using a word or non-idiomatic expression which coincides with the idiom’s figurative meaning.

This methodology will be used to analyze the current corpus as a means of assessing the similarities and differences between the original dialogues and the translated subtitles, as well as between the two Spanish varieties.

The corpus used in this analysis comprises a selection of idiomatic or metaphorical expressions that appear frequently in TV series. Previous studies (Labarta Postigo 2021, 2022) on the translation of idioms from English to Spanish and other languages in film material highlighted the need to develop a larger corpus, towards conducting a more exhaustive analysis. In the current study the reference corpus has been greatly expanded, using data from The TV Corpus[6], one of the largest available corpora of informal English, at 325 million words. To this I have added my own corpus, the result of several previous studies on subtitled TV series from 2000 to date. The contrastive analysis is based in 300 idiomatic expressions from following series, which provide subtitles in both Spanish varieties: Homecoming (2018-2020), The Diplomat (2023), The Man in the High Castle (2015-2019) and The Night Agent (2023). Further details about the series are listed in section 7.

5. Analysis

The qualitative analysis of the corpus focuses on the use of metaphorical or idiomatical expressions in the translated subtitles. The first classification establishes whether or not a translation includes an idiomatic expression. For each of these two options we will present two possible strategies, as explained in section 4, above. To illustrate the results, some examples of each strategy will be presented in the sections 5.1 and 5.2.

The second step will be a contrastive analysis of the translations English-ESp and English-LASp. The results of the quantitative analysis will be presented in section 5.3.

5.1. Translations with figurative meaning

5.1.1. Total Equivalence (TE)

The TE strategy consists of translating the original idiom using a fully equivalent one in the TL, as in the following example:

Table 2: Example “to pull some strings” (series The Man in the High Castle).

In this example the idiom in both languages is the same. According to the Cambridge Dictionary, the meaning of the idiomatic expression “to pull the strings” is: “to secretly use the influence you have over important people in order to get something or to help someone”. An identical idiom exists in Spanish: “tirar de los hilos”. This is an idiomatic expression, and indeed can be found in several languages; as such it belongs to the category of widespread idioms, that is, idioms that exist in various languages with the same or similar structure and figurative meaning (Piirainen, 2006: 158-159).

Other examples from a total of 25 widespread idioms translated as TE in the analysed corpus are as follows:

Table 3: Examples of the translations of widespread idioms

We also find idiomatic expressions in the analized corpus which use strong language and are translated as TE into Spanish:     

Table 4: Examples of the translation strategy TE with strong language

A total of six TE translations contain strong language. Although the use of strong language is not the focus of our study, we found it interesting to note that sometimes the strategy TE also includes the translation of those expressions.

5.1.2 Partial Equivalence (PE)

This strategy consists of translating the original English idiom into a different idiomatic expression in the target language (TL). That is, the form and/or structure in the TL varies but the figurative or metaphorical meaning is preserved, as in the following case:

Table 5: Example “to be nuts” (series Homecoming)

In the example of table 5, both idiomatic expressions serve to ask if someone is crazy. Yet the metaphors in English and Spanish are completely different. A literal translation of the Spanish one would be “is your pot/kettle gone?”, which makes no sense in English.

We find several examples of this strategy in each of the analyzed series. As in the previous tables, table 6 shows the expressions in the SL English on the left, and the subtitles in the TL Spanish on the right:

Table 6: Examples of the translation strategy Partial Equivalence (PE)

This strategy is in general much more frequent than the previous one, reaching a total of 60 examples in the translated subtitles. It is not surprising, since finding a total equivalence is only possible in the case of widespread idioms, and many idioms are not. In section 5.3 we will discuss these data in more detail, as well as the results of the contrastive analysis between the two translation pairs.

5.2 Translations without figurative meaning

5.2.1 Omission (O)

This strategy (O) consists of omitting the translation of the original idiom. One of the few examples found in our data is given here:

Table 7: Example “to piss someone off” (series The Diplomat)

The first sentence with an idiomatic expression is wholly absent in the Spanish translation. This strategy is rare in our corpus, since there are just three examples, yielding fewer than for the other types.

5.2.2 Explicit Meaning (EM)

This strategy consists of translating the original idiom by means of a non-idiomatic expression in the target language, with this translation explaining or paraphrasing the figurative meaning of the original. Hence, no idiomatic or metaphorical expression appears in the subtitles translated into the target language; here, the subtitles have no figurative sense at all, as can be seen in the following example:

Table 8: Example “to touch base” (series The Diplomat)

This idiom is a sporting metaphor from baseball. “To touch base” is a colloquial American English expression, which is culturally specific and has no equivalent with a figurative meaning in Spanish. The metaphor relates to a situation in baseball in which both runners and fielders have to literally “touch the base” in order to be safe (runners) or to eliminate the running player (fielders). In terms of figurative usage, the Cambridge Dictionary, the Collins English Dictionary and the Merrian-Webster Dictionary agree that “to touch base” means to establish contact or communicate with someone, to find out how a person is doing or what he/she thinks about something. Some idioms are culturally specific and thus the only possible translation is to paraphrase or explain them, as in the example in table 8, above: The translated subtitle conveys the figurative meaning of the metaphor only.

However, the presence of untranslatable, unique or culturally-bound idioms is not the only reason to use this strategy. Explaining the meaning is often a choice made by the translator, as we will see in some examples in section 5.3, below.

The following table shows other examples of the EM strategy in the four analyzed series. The original English CC are on the left side. We find on the right side the Spanish subtitles with the corresponding English translation by the author in brackets:

Table 9: Examples of the translation strategy Explicit Meaning (EM)

This strategy is the most frequent one with a total of 114 examples in the translated subtitles. However, the frequency is considerably different in the subtitles translated into each of the two varieties of Spanish. Section 5.3 will further explain these data and provide a detailed explanation of the results of the contrastive analysis.

5.3 Contrastive Analysis of the two translation pairs: English to European Spanish (ESp) vs. English to Latin-American Spanish (LASp)

300 idiomatic expressions were selected from drama series in which subtitles in both Spanish varieties were provided. These included four series that were on air at the time this study was carried out (two from Amazon Prime and two from Netflix). Of the 300 expressions selected, one hundred were the original English idioms, plus the corresponding translations into ESp and LASp. A quantitative study was then carried out to compare the strategies used for the translation of idiomatic expressions in both the Spanish varieties. The following graph shows the results of that analysis:

Figure 1. Results of the contrastive analysis

One of the most surprising findings was the marked difference between the two varieties in terms of idiomaticity. The most frequently used strategy in LASp is explicit meaning (EM): 72 per cent of the subtitles translated into this variety involved no metaphorical or idiomatic expressions; a further 2 per cent were omissions (O); this leaves just 26 per cent of translations with figurative content, little more than a quarter of the total number of idiomatic expressions in the source language.

In the ESp subtitles the translations with figurative meaning represent a total of 59 per cent (41 per cent partial equivalences and 15 per cent total equivalences), more than double what was observed in LASp. By contrast, in the ESp subtitles EM was used in only 40 per cent of cases.

Let us illustrate these differences with some examples:

Figure 2: Example from the series The Diplomat, Netflix, Season 1, Episode 8

The following table shows the full text of the English subtitles containing the idiom in figure 2, as well as the ESp and LASp translations:

Table 10: Example from the series The Diplomat, Netflix, S1, E8

As we see above, “Hacer la pelota a alguien” is an idiomatic expression with a similar meaning to the English one, while “adular a alguien” means to flatter someone. Despite not using strong language, the ESp translation is metaphorical and has a figurative meaning. Meanwhile, the LASp subtitle uses the idea of flattering. Furthermore, the verb “adular” is from a rather formal register in Spanish, and this does not accurately translate the colloquial tone of the English idiom.

Table 11: Example of the Series The Night Agent, Netflix, T1, E9

The idiomatic expression in the English CC is a variation of the idiom “to take something with a grain of salt”, that is, to view something with skepticism. By exaggerating the normal quantity of salt in the idiom to “a giant bucket” the character speaking here expresses her extreme mistrust of her interlocutor in a very sarcastic way.

In the ESp subtitles, the expression "with feet together" alludes to a very old idiom, which according to the Dictionary Real Academia Española (RAE) means "without any doubt". This idiom is usually used when one has blind faith in something or someone. Aligned with the verb “forgive”, the phrase conveys a humorous note, which does in part translate the figurative meaning of the English CC. Thus, whereas this translation lacks the overall force of the sarcasm in the original, the irony is maintained. Less expressive is the explicit translation of the LASp.

Table 12: Example of the Series The Man in the High Castle, Amazon Prime, T3, E1

In the above example the translation into ESp is a TE. The idiom and the bad language are translated into an identical Spanish expression. However, the LASp subtitles conveys the meaning of the phrase with a euphemism and avoids translating the word “shit”. The result is an interpretation of the original English, toning down the language, and the translation thus lacks expressive force: “debes saber cosas de todos allí” (you must know things about everyone there).

Figure 3. Example from the series Homecoming, Amazon Prime, Season 1, Episode 5

The following table shows the full text of the English subtitles containing the idiom in figure 3, as well as the corresponding translations into ESp and LASp:

Table 13: Example of the series Homecoming, Amazon Prime, T1

In the scene shown in figure 3 the main character of the series is being threatened with being fired by her boss. He is very angry and uses an idiomatic expression to make this threat: “you will be out on your ass”. The translation into ESp “te pondré de patitas en la calle” is an idiomatic one, which literally means “I will put you out on your little legs on the street”. A similar idiom in English would be “I will kick you to the curb” or more simply “you’ll be out on the street”. Despite not containing the forceful word “ass” of the original, the ESp subtitles reflect a touch of expressivity through the humor of the diminutive “patitas” (little legs). The LASp option “te despediré de inmediato” simply means you will be fired immediately.

6. Conclusion

The results of the analysis show a general trend towards a reduction of idiomatic and metaphorical expressions in the translated Spanish subtitles, with significant differences between the two varieties analyzed.

Turning to the overall results, only 42.5 per cent of the 200 translations into the two varieties are idiomatic; that is, far less than half of the Spanish subtitles preserve the figurative meaning of the English CC. In terms of translation strategies, the most frequent one with idiomatic content is PE, at an average of 30 per cent, followed by TE, at 12.5 per cent. The number of omissions is very low (1.5 per cent), whereas the number of explicit meanings is the highest of all (112 of the 200 translations). On the one hand, the fact that almost all the idioms have been translated in some way, and only a few have been omitted, is a positive finding with regard to accuracy and quality. On the other hand, 56 per cent of the idioms are translated through expressions with no metaphorical, idiomatic or figurative meaning at all; thus, a possible loss of part of the meaning affects more than half of the translations in our corpus. It seems clear, then, that a greater emphasis on improving the translation of idioms would enhance the overall quality of subtitles in these data.

Reductions in the figurative meaning of the original subtitles differs in the two Spanish varieties under investigation. Indeed, figure 1 illustrates how striking these differences are. The contrastive analysis shows that in the translations into LASp, the explicit meaning strategy is used in 72 per cent of cases, that is, an increase of 32 per cent on those translated into ESp. The LASp subtitles, then, offer far less idiomaticity and fewer metaphorical expressions than the ESp ones.

As noted in section 2, idioms are used frequently in dialogues, adding expressive force and drawing a character’s personal speaking style into the action. Sometimes the use of idioms, or a certain kind of idiom, is indeed a recognizable part of a character’s identity. These features of conversation may be lost when the EM strategy is used for the translation, and the fact is that more than half of the idioms overall, and almost three-quarters in the LASp data, were indeed translated by omitting the idiomatic/metaphorical meaning itself.

The use of the EM strategy might lead to an easier and faster literal understanding on the part of viewers, but it might also have negative consequences in terms of the loss of other levels of meaning. Furthermore, omitting the idiom itself implies a loss of meaning, including information relating to register, humor, or the features of a specific character in the action, in that an idiom tells us something about a speaker: their sociolect, idiolect and style, among many other factors.

In the survey of a group of d/Deaf and HoH people, more than half of the respondents were in favour of translating most idioms using figurative language. Larger and more detailled surveys are needed to shed further light on this issue. Nevertheless, it is important to underline the significance of the opinions of this sector of the TV audience; although a minority of the overall audience share, their judgements are of disproportionate relevance, given the greater degree to which they rely on subtitles when watching TV.

The approach to translation taken in this study, as well as the analysis of findings, have raised new questions that merit further investigation through a more detailed exploration of the reception of subtitles among the d/Deaf and HoH. A survey of a larger number of addressees here would serve to support the findings presented above. Additionally, more work in this area would help us to better understand the reduction of figurative meaning in translated subtitles in general, and in Latin-American Spanish subtitles in particular.

Abstract

English:

The present study analyses Italian subtitles for the deaf and hard of hearing from a sociolinguistic perspective in some Italian audiovisual productions. The aim is to assess the degree of adherence to the original product with a view to accessibility, not only referred to the service per se, but also to the linguistic quality. More specifically, the focus of the analysis corresponds to spoken Italian following its sociolinguistic continuum. It includes colloquialisms, sub-standard syntactic strategies, semantically marked expressions, and swearwords, up to dialect and regional items. Therefore, the intention is to monitor the presence or absence of these spoken Italian features in such subtitles for some films and television productions broadcast on RaiPlay, given its recent development not only as a streaming platform, but also as a channel with original content (Rai.it, 2022). The study will be conducted through an opportunistic corpus, namely SoLiSS (Sociolinguisitca dei Sottotitoli per Sordi – Sociolinguistics of Subtitles for Deaf and Hard of Hearing), which collects the most significant excerpts for this purpose.

Italian:

Il presente studio si pone la finalità di analizzare la resa delle dimensioni sociolinguistiche individuabili nella produzione audiovisiva in lingua italiana nei sottotitoli per sordi e ipoudenti, in modo da valutarne il grado di aderenza al prodotto originale in un’ottica di accessibilità non solo del servizio, ma anche della qualità linguistica. Nello specifico, il focus principale dell’analisi si identifica con l’italiano parlato e con gli elementi ad esso afferenti, tra cui figurano colloquialismi, strategie sintattiche substandard, espressioni semanticamente marcate e il turpiloquio, fino ad arrivare all’estremità più dialettale del continuum linguistico dell’italiano. Pertanto, si intende monitorare la presenza o l’assenza nei suddetti sottotitoli delle caratteristiche dell’italiano parlato in alcune produzioni per cinema e televisione trasmesse su RaiPlay, dato il suo recente sviluppo in qualità non solo di mera piattaforma streaming, ma anche di vero e proprio canale con prodotti originali (Rai.it, 2022). Lo studio sarà condotto con l’ausilio di un corpus opportunistico denominato SoLiSS (Sociolinguistica dei Sottotitoli per Sordi) che raccoglie gli estratti più significativi a questo scopo.

Keywords: traduzione audiovisiva, sottotitoli per sordi, sociolinguistica italiana, linguistica dei corpora

1. Introduzione

Il presente studio analizza la realizzazione dei sottotitoli per sordi e ipoudenti in produzioni audiovisive in lingua italiana secondo le dimensioni sociolinguistiche, in modo da valutarne il grado di aderenza al prodotto originale. Il focus principale dell’analisi si identifica con l’italiano parlato e con gli elementi ad esso afferenti, tra cui figurano colloquialismi, strategie sintattiche, espressioni semanticamente marcate e fraseologismi, fino all’estremità più dialettale del continuum linguistico dell’italiano. Si intende monitorare la presenza o l’assenza nei suddetti sottotitoli delle caratteristiche dell’italiano parlato in alcune produzioni per cinema e televisione, comprendenti documentari, reality show, film e serie televisive, con l’ausilio di un corpus opportunistico che raccoglie gli estratti più significativi a questo scopo. Nello specifico, si considerano alcuni programmi trasmessi su RaiPlay, dato il suo recente sviluppo in qualità non solo di piattaforma streaming, ma anche di vero e proprio canale con prodotti originali (Rai.it 2022).

1.1 Stato dell’arte e scopi della ricerca

È noto come la natura multimediale della sottotitolazione di un prodotto audiovisivo (AV) implichi l’interazione tra la componente verbale orale, l’immagine e la componente verbale scritta nel rispetto di sincronia, dimensione dello schermo e readability del sottotitolo stesso (Perego 2005; Díaz Cintas e Remael 2021)[1]. L’osservazione di tali vincoli spazio-temporali potrebbe portare all’applicazione di strategie di economia testuale (o di semplificazione linguistica, secondo Berruto 1990) nella sottotitolazione per sordi e ipoudenti, data la necessità di riportare anche la componente acustica paralinguistica, extralinguistica e gli effetti sonori, come richiede la struttura intersemiotica del prodotto AV[2]:

Oltre agli elementi vocali, va convertita in forma scritta anche la componente soprasegmentale del parlato e devono essere resi graficamente i rumori, i suoni non vocali, i contenuti musicali, i silenzi; in definitiva tutti i segnali la cui ricezione passa per il canale uditivo. Questo rende i sottotitoli per non udenti estremamente più ricchi di informazioni sui fenomeni sonori di quelli interlinguistici (Orletti ed Eugeni 2019: 532).

Franca Orletti (2016) descrive attentamente questa criticità, menzionando il problema della variazione:

Il problema della condensazione e riduzione del contenuto informativo dei dialoghi è ancora più sentito nei sottotitoli intralinguistici in quanto nei 36/40 caratteri per riga vanno incluse tutte le didascalie descrittive dei fenomeni sonori[3]. I tratti specifici del parlato quali le ripetizioni, gli intercalari, i segnali discorsivi, ma anche l’adozione di forme substandard in diatopia e diastratia scompaiono. […] si perde la caratterizzazione sociolinguistica dei personaggi e delle relazioni sociali espressa proprio dalla variazione in diatopia, diastratia e diafasia della lingua (228).

La semplificazione del testo talvolta porta a un innalzamento del registro, che complica la comprensione del sottotitolo, spesso percepito come “longer to read” e “too demanding” (Bianchi, Eugeni e Grandioso 2020: 24-5)[4].

Se la sottotitolazione per sordi rappresenta un metodo di traduzione intralinguistica volto all’avvicinamento dello spettatore al patrimonio audiovisivo (Neves 2005; Díaz Cintas e Neves 2015; Baños e Díaz Cintas 2018)[5], la volontà di fornire a una tale audience la stessa esperienza di un pubblico udente dovrebbe trovare dei riscontri nel trattamento diamesico delle varietà linguistiche. Per avvalorare questa ipotesi, occorre adottare una prospettiva di ricerca empirica: si propone per il presente lavoro la creazione di un corpus ad hoc, denominato SoLiSS (Sociolinguistica dei Sottotitoli per Sordi), che al momento include i sottotitoli di battute diatopicamente, diafasicamente e diastraticamente significative di produzioni italiane recenti trasmesse in RAI. Lo scopo è di analizzare il trattamento di elementi come parole gergali, dialettali, regionali, connotate in base a registro e stile o proprie dell’italiano neostandard (Berruto 1987, 2012; Tavoni 2002), substandard o dell’uso medio (Sabatini 1985) nei sottotitoli, per constatarne la corrispondenza ai dialoghi originali.

2. L’adattamento linguistico nei sottotitoli per sordi: i dati RAI

La RAI, che trasmette sottotitoli dal 1986 (De Seriis 2006), prevede come paradigma accessibile la corretta conservazione di “contenuto, intenzioni e registri linguistici e stilistici specifici del singolo prodotto audiovisivo [rispettando pienamente] ogni singolo programma e ancor più le esigenze e le aspettative dei fruitori” (Servizio sottotitoli-RAI, 2016: 1). Tuttavia, la più recente normativa RAI al momento del presente lavoro, pubblicata nel giugno 2021, specifica le caratteristiche del pubblico fruitore dei sottotitoli, non limitandosi all’etichettatura di “deficit uditivo” (ibid.), ma includendo “persone sorde dalla nascita o divenute tali nel corso del tempo e [i] deboli di udito, […] telespettatori anziani, stranieri e persino persone udenti che scelgono, al bisogno, di seguire i programmi televisivi senza audio” (RAI 2021a: 1). Questa varietà di utenza sembra suggerire un adattamento linguistico che possa corrispondere piuttosto fedelmente ai dialoghi in funzione dell’attendibilità del servizio offerto.

Parlando di norme e convenzioni editoriali, nella normativa del 2016 la RAI indica che i sottotitoli per sordi devono contenere al massimo 37 caratteri per riga (cpr), ad altezza doppia e trasmessi a blocchi (Servizio Sottotitoli-RAI 2016). Per ciò che concerne le dimensioni fàtica o regionale, la stessa precisa di “non attuare censure o semplificazioni aprioristiche” (4); tuttavia, si impone di offrire traduzioni di contenuti dialettali o espressioni idiomatiche di supposta difficile comprensione, senza un eccessivo discostamento dal lessico di partenza. La normativa del 2021 prevede la medesima strategia, indicando però le sincronie cinetica e labiale come discrimine essenziale per il mantenimento del lessico originale, anche se marcato (RAI 2021a). Ulteriori specifiche vengono introdotte anche circa il linguaggio gergale od offensivo, per cui si suggerisce di non applicare censure o riformulazioni del tutto arbitrarie, ma di “valutare caso per caso se vi siano le condizioni spazio-temporali per preservare il linguaggio originale” (RAI 2021a: 17). Tale precisazione non compare nel documento precedente, che infatti parrebbe avere un approccio ormai aprioristico.

Dal confronto delle normative RAI si può anche identificare una svolta sulla quantità e sulla modalità in cui la sottotitolazione accessibile è resa disponibile. Il Bilancio di Sostenibilità del Gruppo Rai (RAI 2021b) mostra che “nel 2021 Rai ha sottotitolato 17.540 ore, pari a circa il 95% della programmazione” e “oltre 3.000 ore […] sul web” (187)[6], con un indice di soddisfazione per i servizi offerti agli spettatori con disabilità pari al 7,3 percento (69). In merito al web, il crescente utilizzo della piattaforma RaiPlay negli ultimi anni come servizio streaming e catalogo on demand[7] ha reso necessaria l’inclusione di servizi per l’accessibilità, tra cui i sottotitoli per persone sorde; tuttavia, occorre fare una riflessione. Se si comparano le ore sottotitolate sulla piattaforma web nel 2021 alle ore sottotitolate sul digitale terrestre (RAI 2021b: 187), la differenza risulta rilevante. Inoltre, considerando che, secondo una ricerca condotta nel 2018 dal Censis, 7,2 milioni di persone su 60 milioni di abitanti in Italia sono ipoudenti, si deduce che circa il 12,1% della popolazione residente (Censis 2019: 6-7), equivalente a più di un italiano su dieci, presenta problematiche relative alla totale accessibilità del patrimonio audiovisivo del Paese, senza contare chi ancora non risiede formalmente sul territorio. A questo si aggiungono alcuni disagi dichiarati dagli utenti sordi in merito alla funzionalità dei sottotitoli RAI (Echites 2016)[8].

È pur vero che il servizio di sottotitolazione sul web ha visto una vera e propria regolamentazione nei tempi recenti. Infatti, nella normativa del 2016 (compresi gli aggiornamenti del 2018) non è presente alcun tipo di normativa sulla gestione dei sottotitoli per sordi specificamente per il portale, nonostante questo fosse già in funzione. D’altro canto, nel documento del 2021 è visibile in tutte le sezioni l’aggiunta dell’etichetta WEB per descrivere le modalità di composizione del sottotitolo destinato alla piattaforma. Inoltre, le potenzialità tecnologiche di RaiPlay giustificherebbero la scelta di utilizzarla come veicolo primario per la trasmissione di eventi nazionali culturalmente incisivi adattati a un pubblico sordo. Ad esempio, in occasione del 72° Festival di Sanremo, tenutosi dal 1° al 5 febbraio 2022, la piattaforma ha fornito per la prima volta un servizio di diretta con sottotitoli contrassegnati da colori, alla stregua di quelli trasmessi alla pagina 777 del Televideo, e con interpreti LIS che accompagnavano le esibizioni degli artisti. Questa soluzione sembrerebbe dimostrare la sempre maggior incidenza del portale multimediale sull’offerta di prodotti accessibili per spettatori con ‘disabilità’ sensoriali. Proprio per il generale sviluppo della RAI in termini di accessibilità digitale, lo studio si concentra sull’analisi dei sottotitoli proposti su RaiPlay, dopo aver valutato un’effettiva corrispondenza con il Televideo.

3. Variazione sociolinguistica e sordità

Per comprendere la complessità della resa delle varietà dell’italiano nei sottotitoli per sordi occorre innanzitutto partire dalla considerazione della lingua non come sistema di frasi isolate, ma come azione che si compie in uno specifico contesto sociale e comunicativo, con focus sugli utenti e sulle interazioni tra di essi (Dardano 2005; Baldi e Savoia 2009). Per questo motivo risulta necessaria un’indagine bilaterale secondo i molteplici aspetti della pragmatica (Levinson 1983): da un lato, sull’uso di varianti verbali nel parlato come discrimine per la caratterizzazione del personaggio; dall’altro, sull’influsso di analoga variazione nel patrimonio linguistico di persone sorde, specialmente segnanti e con sordità dalla nascita. Chiara Branchini e Lara Mantovan (2022) sostengono che la LIS si realizzi concretamente più nelle sue varietà sociolinguistiche[9] che come standard a causa di quattro fattori:

1. un riconoscimento ufficiale della lingua dei segni tardivo, avvenuto solo di recente[10] ii) la pressione causata dalla lingua italiana, che tende ad essere considerata più prestigiosa, iii) la scarsità di programmi scolastici bilingui (italiano-LIS) e iv) l’assenza di una forma scritta diffusa e condivisa della LIS (110).

Per fornire degli esempi pratici, le due autrici menzionano l’incidenza delle variazioni diastratica e diatopica nell’ordine dei costituenti all’interno della frase: i parlanti più anziani utilizzano l’ordine SVO, al contrario dei più giovani, che tendono a preferire l’ordine SOV; analogamente, le varietà del nord Italia tendono a prevedere un ordine di tipo SVO, mentre le varietà del sud Italia applicano l’ordine SOV (le Figure 1 e 2 riportano gli screenshot in ordine temporale dei video in Branchini e Mantovan 2022:

a. Gianni comprare casa

‘Gianni ha comprato una casa’

Fig. 1: Screenshot della frase ‘Gianni comprare casa’ segnata in LIS – variante SVO (111).

b. Gianni casa comprare

‘Gianni ha comprato una casa’

Fig. 2: Screenshot della frase ‘Gianni casa comprare’ segnata in LIS – variante SOV (ibid.).

Per quanto riguarda il lessico, sono presenti esempi di varianti diatopiche: il segno uno viene realizzato con l’indice nelle aree meridionali, mentre con il pollice nelle aree settentrionali (112). La variazione geolinguistica è inoltre giustificata dalla produzione di dizionari che negli anni hanno caratterizzato la promozione e la diffusione di segni regionalmente connotati, come è accaduto per il Dizionario Regionale del Linguaggio Mimico Gestuale Marchigiano (92). È stato anche mostrato come le varietà nella LIS siano dovute ai diversi istituti frequentati dalle persone sorde, per cui si riscontrano più varianti per uno stesso segno in una stessa aera geografica (ibid.).

La dimostrazione della variazione sociolinguistica nella LIS sembrerebbe giustificare una resa fedele della variazione nel più ampio spettro dell’italiano parlato anche nei sottotitoli inclusivi, pur rispettando le convenzioni tecniche. Inoltre, un’alternativa verbatim coincide con la volontà degli autori di definire il personaggio nel suo idioletto, scongiurando il rischio di una comprensione solo parziale di dettagli utili alla trama da parte di spettatori sordi o ipoudenti in virtù della sospensione dell’incredulità.

4. Metodologia d’indagine: il corpus SoLiSS

Al fine di analizzare il comportamento dei sottotitoli chiusi RAI per la pubblica utilità è stato ideato e creato un corpus contenente sia gli script originali che i sottotitoli corrispondenti in ottica comparativa. Il corpus, chiamato SoLiSS – Sociolinguistica dei Sottotitoli per Sordi, è rappresentativo dei principi teorici secondo cui la sottotitolazione per sordi si realizza relativamente alla trasformazione diamesica in produzioni televisivo-cinematografiche italiane RAI. Esso è opportunistico in quanto realizzato specificamente per lo studio in questione; tuttavia, è suscettibile di ampliamenti e modifiche, in base alle nuove produzioni potenzialmente idonee allo scopo della ricerca. Inoltre, permette la comparazione con le ricerche sul campo dell’effettiva percezione delle varietà del continuum dell’italiano da parte di sordi e ipoudenti, tenendo conto della gamma di caratteristiche fisiologiche in termini di udibilità[11].

SoLiSS include dieci programmi televisivi italiani su RaiPlay comprendenti fiction, reality preregistrati, film, spettacoli in diretta e documentari, i cui sottotitoli sono stati trascritti manualmente. Nello specifico, sono stati selezionati i seguenti titoli:

• Veloce come il Vento (film; Italia, 2016)
• Vola, Luna Rossa (documentario; Italia, 2021)
• Un Posto al Sole – episodio n.5871 (fiction; Italia, 2022)
• 72° Festival di Sanremo[12] – estratto monologo di Sabrina Ferilli (spettacolo; Italia, 2022)
• 72° Festival di Sanremo – estratto dialogo tra Amadeus e Rettore (spettacolo, Italia, 2022)
• L’Amica Geniale – stagione 3, episodio 2, “La Febbre” (fiction; Italia, 2022)
• Il Collegio[13] – stagione 6, episodio 8 (reality; Italia, 2021)
• Vostro Onore – stagione 1, episodio 1 (fiction; Italia, 2022)
• Màkari – stagione 2, episodio 1 (fiction; Italia, 2022)
• Noi – stagione 1, episodio 1 (fiction; Italia, 2022).

La scelta di analizzare i sottotitoli di queste produzioni italiane recenti deriva dall’intenzione di constatare l’applicazione delle norme nel documento RAI del 2021 sulle variazioni. Unica eccezione è il film del 2016 Veloce come il Vento, che è stato caricato sulla piattaforma negli anni seguenti. Occorre inoltre specificare che si sono riscontrate problematiche nella reperibilità di programmi preregistrati con sottotitoli su RaiPlay, il che ha posto dei limiti nella selezione dei dati; tuttavia, il materiale raccolto risulta sufficiente alla ricerca.

SoLiSS è composto complessivamente da 40.357 tokens e 8.163 types ed è stato taggato manualmente in maniera sperimentale utilizzando la stessa modalità di etichettatura del POS-tagging (Part Of Speech tag), in quanto di facile composizione e di immediato riscontro in fase analitica. Esso è composto da due sub-corpora: il primo, denominato SoLiSS_script, racchiude gli script originali (o porzioni di script) per un totale di 21.597 tokens e 4.221 types; il secondo, denominato SoLiSS_subs, contiene i sottotitoli corrispondenti e ha in totale 18.760 tokens e 3.942 types. Ciascuno è composto da dieci file di testo, uno per programma. In essi, sono stati oggetto di tag gli elementi lessicali più rappresentativi delle varietà dell’italiano:

• #_COL = colloquialismo
• #_REG = regionalismo
• #_DIAL = dialetto
• #_TUR = turpiloquio
• #_PAR = parlato.

Con colloquialismo si intendono sia un lessico informale e non connotato per gruppi sociali, sia elementi gergali; il regionalismo è qui inteso come elemento linguistico geograficamente connotato su una o più regioni (amministrative o linguistiche), senza però specificare il territorio di appartenenza o l’estensione areale, mantenendo una definizione generalizzata (De Blasi 2014); sotto l’etichetta di dialetto si raggruppano koinè dialettale regionale e dialetto rustico arcaico (ibid.); il turpiloquio indica gli elementi osceni, triviali, blasfemi e offensivi; infine, con parlato si indica qualsiasi strategia sintattica marcata appartenente all’italiano neostandard e substandard[14]. A tal proposito non sono state previste procedure di parsing per l’analisi sintagmatica e sintattica, in quanto di eccessiva complessità per il lavoro in questione.

SoLiSS è stato successivamente caricato sul software LancsBox (versione 6.0) per l’analisi dei corpora, sviluppato dall’Università di Lancaster e scaricabile gratuitamente sulla piattaforma dedicata. La scelta di questo strumento è dettata, in primis, dalla possibilità di impostare la lingua italiana tra le opzioni per il riconoscimento linguistico del corpus, scongiurando errori di codifica e decodifica dei caratteri accentati e di identificazione lessicale. Inoltre, LancsBox permette di confrontare due corpora o sub-corpora diversi sulla stessa schermata tramite l’opzione di separazione del riquadro di visualizzazione testuale, in modo da avere due diversi riscontri per una stessa richiesta. Infine, è possibile effettuare una ricerca nella sezione KWIC (Key Words In Context) specifica per i tag, riportando nell’apposito riquadro, denominato POS, il nome dell’etichetta senza aggiungere il trattino basso, poiché il sistema provvede automaticamente al suo riconoscimento.

Fig. 3: Ricerca del tag REG direttamente come POS nella sezione KWIC di LancsBox.

Questa modalità di ricerca evita che il software mostri i tag per esteso attaccati a ciascuna parola, preferendo porre l’accento direttamente sul lessico. Infatti, nella schermata KWIC comparirà come nodo del contesto solo la parola associata al tag ricercato, senza che quest’ultimo disturbi la visualizzazione delle parole da analizzare.

Fig. 4: Schermata KWIC di LancsBox per la ricerca del tag REG
nei sub-corpora SoLiSS_subs e SoLiSS_script

5. Analisi dei dati e discussione

Il corpus SoLiSS così composto e caricato è stato analizzato partendo dal tag REG. Esso ha mostrato un’occorrenza di 81 volte per il sotto-corpus SoLiSS_subs e di 257 volte per il sotto-corpus SoLiSS_script, dimostrando che a livello quantitativo i regionalismi generalmente rispettati nei sottotitoli corrispondono a circa 1/3 di quelli effettivamente detti. Inoltre, in SoLiSS_subs sono risultati sette testi su dieci come contenenti il tag; di contro, SoLiSS_script ne possiede di più, nello specifico nove: il testo mancante in entrambi i sotto-corpora è il documentario Vola, Luna Rossa, che non prevede elementi regionali per il suo carattere di monoreferenzialità. Tuttavia, i restanti due pacchetti di sottotitoli non contenenti regionalismi sono Noi e Un Posto Al Sole, che invece sono presenti nelle battute originali.

Per quanto concerne Noi, la serie utilizza un italiano piuttosto standard, per cui l’assenza di regionalismi nei sottotitoli risulta prevedibile. Nello script originale figurano come tali solo due casi: il pronome personale di forma tonica te in qualità di soggetto, tipico delle parlate di Roma, oltre che del nord Italia e della Toscana (sebbene una sua diffusione sia riscontrabile su tutta la penisola), e il troncamento o apocope della forma alla terza persona plurale del verbo essere al presente indicativo son, particolarmente diffuso nelle parlate regionali del nord Italia (Grassi, Sobrero e Telmon 1997: 96-7):

l’hai detto anche te prima![15] (personaggio con cadenza romanesca);

Claudio, i copioni son sempre quelli (personaggio con cadenza regionale settentrionale).

Risulta comprensibile la scelta di non includere queste due varianti nei sottotitoli, in quanto avrebbero innanzitutto dato un contributo pressoché nullo alla caratterizzazione dei personaggi. Inoltre, si sarebbe corso il rischio di confusione omografica nel caso di un te polisemico in una frase con lieve commutazione di codice tra standard e regionale in un discorso complessivamente standard, considerando la breve permanenza del sottotitolo sullo schermo.

La situazione di Un Posto Al Sole risulta diversa. Nello script sono visibili dieci tag REG, ciascuno particolarmente significativo:

• Uè! Ti stavo inviando un messaggio proprio adesso.
• Uè¸ Nunzio!
• Ma veramente tu stai attaccato proprio…stai con la demenza senile?
• Ieri te l’ho detto, questo fatto dell’aria
• Io non so’ tagliato per ‘ste cose materiali
• Raffae’, ma questo niente niente è proprio il famoso Giancarlo?
• Ma vai a controllare i termosifoni e lascia stare a Silvia!
• […] lui di solito mangia delle schifezze […] della cucina di mammà?

Gli esempi riportati dimostrano che nella fiction ambientata a Napoli sono presenti inflessioni regionali nei dialoghi dei personaggi, caratterizzati proprio da tali scelte linguistiche. La maggior parte dei suddetti interventi vengono pronunciati dai personaggi di Raffaele Giordano e Renato Poggi, due cognati di circa mezza età che tendono a parlare abitualmente un italiano regionale con accento napoletano. Tuttavia, la terza frase sopra citata ha come intero sottotitolo:

Scusa, per caso hai la demenza senile?

che, sebbene da un lato conservi il riferimento chiaramente umoristico a una condizione neurologica, dall’altro rischia un appiattimento per l’eccessiva standardizzazione della frase. Nonostante si giustifichi l’eliminazione della falsa partenza (ma veramente tu stai attaccato proprio…), una possibile mediazione sarebbe stata la sostituzione di per caso con ma e di hai con la voce verbale tieni:

*Scusa, ma tieni la demenza senile?

La frase ottenuta sarebbe stata più regionalmente connotata e caratterizzante i due personaggi. Analogamente, l’interiezione uè, l’apocope so’, l’aferesi ‘sto e mammà negli altri esempi riportati hanno subìto lo stesso processo di cancellazione, nel primo caso, e standardizzazione, nei restanti casi; in questo modo i sottotitoli peccano di caratterizzazione regionale. Per ciò che concerne invece la ripetizione niente niente e l’accusativo preposizionale a Silvia, la scelta di una loro omissione è del tutto comprensibile, in quanto suscettibili di fraintendimenti tipografici e grammaticali.

Procedendo per ordine, una fiction che ha mostrato solo un’occorrenza di elementi regionali nei sottotitoli contro le quattro nello script è Vostro Onore, ambientata a Milano e caratterizzata da una varietà piuttosto standard. Nello specifico, l’elemento contrassegnato come regionale (e udibile come tale data l’inflessione accentuale) è l’uso dell’articolo determinativo la davanti a nomi propri femminili, tipico delle aree settentrionale e centrale (Toscana) in situazioni fàtiche informali:

ho parlato con la Vichi.

I restanti regionalismi sono il troncamento delle forme abbiamo e sono e il già citato pronome tonico te come soggetto, non riportati nei sottotitoli.

Un esempio piuttosto notevole di conservazione dei regionalismi nei sottotitoli è invece dato dal monologo di Sabrina Ferilli al 72° Festival di Sanremo. Sebbene non tutti gli elementi regionalmente connotati siano stati riportati, i sottotitoli, in diretta, hanno rispettato forme quali

• e ancora adesso il petto qui me vibra
• e che sorpresa quando […] vinsero […] i MANESCHINE,

quest’ultimo caso per il rispetto della sincronia labiale e dell’intenzionalità di una storpiatura comica del nome Måneskin.

Per quanto riguarda gli altri programmi, ne Il Collegio REG compare appena sette volte nei sottotitoli in confronto alle 38 dei dialoghi originali. Un caso notevole è la parola impazzuti, correttamente riportata nei sottotitoli quale idioletto di un ragazzino pugliese. Tuttavia, in altri casi i sottotitoli non riportano alcune espressioni caratterizzanti: l’assenza dell’aggettivo fessi e della reiterazione del verbo stare per essere, tipicamente meridionali, influisce negativamente sulla caratterizzazione dei personaggi, invece pienamente colta dal pubblico udente.

La fiction Màkari, ambientata in Sicilia, prevede appena 17 occorrenze di regionalismi nei sottotitoli in confronto alle 39 dei dialoghi originali. Tuttavia, sono riportati elementi significativi: la parola zita per indicare fidanzata, piuttosto diffusa in Sicilia, Puglia, Calabria, Basilicata e Molise, l’aggettivo bona per bella/formosa/avvenente, diffuso principalmente nelle aree meridionali a partire da Roma, e, analogamente, l’uso del congiuntivo imperfetto in subordinata si facesse i fatti suoi e la risalita del pronome atono in vatti a fare la valigia.

Il film Veloce Come il Vento costituisce un caso particolare. Lo script è composto da 74 tag REG, che comprendono numerosi troncamenti delle forme coniugate al presente indicativo di avere ed essere, l’uso dell’articolo determinativo davanti agli aggettivi possessivi (è la tua sorella), l’uso preposizionale dell’avverbio davanti senza essere seguito dalla preposizione a ([ho] il sette davanti l’età) e numerosi te come pronome personale soggetto. Tali espressioni non occorrono nel sub-corpus dei sottotitoli, probabilmente a causa dei già citati possibili fraintendimenti nella forma scritta. Tuttavia, si ripresenta la problematica della caratterizzazione del personaggio che ripetutamente utilizza le suddette forme regionali. Il rischio è di perdere la sincronia tra codice linguistico e codice iconografico (Chaume 2004), intesa come la corrispondenza tra registro utilizzato e rappresentazione visiva (a tratti caricaturale) dei personaggi in quanto emblemi dello strato sociale di appartenenza. Infatti, lo stesso protagonista, Loris, è rappresentato come un tossicodipendente, irascibile, dalla condizione socioeconomica difficile e con un forte legame con il suo territorio, l’Emilia: questi elementi sono identificabili anche dal suo idioletto, caratterizzato da un registro basso utilizzato anche in contesti non strettamente confidenziali e ricco di parole scurrili e dall’accento emiliano marcato. Nonostante il rischio di appiattimento linguistico, nei sottotitoli sono comunque visibili strategie di compensazione tramite il mantenimento di regionalismi quali [fare] fuga per indicare il marinare la scuola (segnato nei sottotitoli con le virgolette alte) e, in maniera particolare, le espressioni volgari vacca boia e due maroni, che da sole costituiscono più della metà delle occorrenze di REG in tutto il film.

Per L’Amica Geniale, serie ambientata nella Napoli degli anni Settanta, il tag è presente per 54 volte nel sub-corpus degli script contro le sole 19 nel sub-corpus dei sottotitoli. Tuttavia, gli elementi conservati sono piuttosto significativi dell’area napoletana: professo’ (sebbene ricorra solo un paio di volte), il verbo dìcere per dire (evidenziato dall’intonazione marcata), l’interiezione uè, la parola fatìca come sinonimo di lavoro e il verbo tenere generalizzato. Occorre però specificare che la fiction è composta da una maggioranza di dialoghi originali in dialetto napoletano tradotti in italiano standard nei sottotitoli aperti[16], i quali in un certo senso favoriscono una risalita del continuum sociolinguistico verso la standardizzazione dei sottotitoli chiusi[17]. Il risultato è, come prevedibile, il parziale mantenimento delle espressioni regionali piuttosto che strettamente dialettali, così da dare un equilibrio linguistico-stilistico generale al prodotto finale.

L’ambito dialettale è stato analizzato più approfonditamente attraverso il tag DIAL. In questo caso, la differenza tra i due sub-corpora è evidente: in SoLiSS_script le occorrenze totali sono 125, mentre in SoLiSS_subs sono appena 37. Tuttavia, il corpus complessivo mostra pochi prodotti in cui il dialetto è effettivamente utilizzato: Il Collegio, L’Amica Geniale¸ Màkari e Veloce Come il Vento per gli script e soltanto Il Collegio e Màkari per i sottotitoli.

Il Collegio evidenzia una situazione di intenzionalità di standardizzazione nei sottotitoli che ha compromesso in maniera particolare la caratterizzazione di uno dei protagonisti: un ragazzino campano particolarmente esuberante e affettuoso. Nei sottotitoli mancano i riferimenti al suo dialetto: un esempio è il verbo rìcere (dire), che ha infatti subìto una cancellazione. Si può a questo punto fare un confronto con L’Amica Geniale, che pure fa riferimento alla Campania (vedasi il mantenimento di dìcere[18]). In realtà, le forme più dialettali sono state tradotte in italiano anche nella fiction; ciononostante, si notano strategie di compensazione conservando alcune espressioni regionali nella stessa scena (ad esempio: Auguri, a mamma). Strategie simili in circostanze simili invece non sono presenti ne Il Collegio, che ha visto una standardizzazione completa nei sottotitoli e un appiattimento del personaggio del suddetto adolescente. Ad esempio, la frase originale

[…] nun sto capenn’ proprj nient’ d’ chell’ ch’ m’ sta ricenn’

diventa

In realtà non ho capito niente.

Nonostante le evidenti difficoltà di trascrizione di elementi dialettali, il risultato finale eccede nella standardizzazione e si discosta totalmente dall’originale sia in termini diatopici, sia diafasici. Al contrario, la parola pota in dialetto bresciano è stata riportata fedelmente, caratterizzando adeguatamente il personaggio di una studentessa bresciana dall’accento marcato.

Màkari è sicuramente la pellicola i cui sottotitoli hanno maggiormente rispecchiato la verosimiglianza del parlato spontaneo in relazione all’uso del dialetto in un contesto siciliano. Già dalla sigla di apertura, parzialmente cantata in dialetto, si evince il rispetto della varietà linguistica nei sottotitoli:

[# CHISTA E' NA STORIA
NU POCU CURIUSA]
[# NU POCU SCUNTRUSA
CHE PARLA D'AMORE].

Parole conservate nei sottotitoli sono [broccoli] arriminati (mescolati; di specifico riferimento), camurriuso (seccante), sardi a beccaficu (per cui occorre mantenere la dicitura originale, data la specificità del riferimento), il lemma picciotto (ragazzo), babbiare (scherzare), l’interiezione ammunì (andiamo), il vezzeggiativo picciriddu (ragazzino), il participio patuto (sofferente), pìccioli (soldi), l’aggettivo meschino (indicante poveretto, anch’esso con senso vezzeggiativo) e il lemma minchia[19], nell’uso comune privo dell’originale denotazione della sfera sessuale, molto presente nei dialoghi come esclamazione anche in forme nominalizzate come minchiata. La conservazione di elementi dialettali così caratterizzanti favorisce la costruzione dei personaggi della serie, in special modo dei protagonisti. D’altro canto, si nota una certa standardizzazione nei sottotitoli delle battute di personaggi secondari, sebbene in taluni casi essi costituiscano di per sé codice iconografico. È il caso di un contadino che, durante l’episodio, si mostra come parlante il siciliano piuttosto che l’italiano standard, ma che ha subìto un processo di italianizzazione nei sottotitoli. Il rischio è quello di cozzare con l’immagine di uomo abituato agli ambienti rurali e apparentemente poco istruito. Il risultato del processo è quanto segue:

Lui a truvò / Era ccà sulo / Nun lu sacciu. / Cu mmìa non ci vole parlare. / Lillo, cuntaci tuttu chillu ca cuntasti a mmìa,

che diventa:

Lui lo trovò / Era da solo / Non lo so. / Con me non ci vuole parlare. / (siciliano) Racconta al dottore / quello che mi hai raccontato.

Come già detto, la difficoltà nella resa grafica di un dialetto impone delle strategie compensative che possano garantire la trasmissione corretta di un messaggio. Qui il sottotitolo riesce comunque a delineare il profilo originale del personaggio mantenendo il ci con funzione avverbiale in una dislocazione, diffuso nell’italiano parlato. Inoltre, si nota la strategia della specificazione tra parentesi tonde tipica della sottotitolazione per sordi in presenza di un dialetto non del tutto comprensibile, che completa il ritratto del personaggio.

Il tag TUR è stato identificato in entrambi i sub-corpora: SoLiSS_script possiede 173 occorrenze, mentre SoLiSS_subs ne possiede 159, per un totale di 331 occorrenze totali nel corpus SoLiSS nel suo complesso. Ne consegue che il turpiloquio è l’elemento dell’italiano parlato che ha lo scarto minore tra i sottotitoli e i dialoghi originali; in questo modo, si rispecchia la volontà di conservare un tratto del parlato solitamente oggetto di censure (Beseghi 2016). Nel sub-corpus degli script otto testi su dieci propongono l’uso del linguaggio scurrile, afferente alle sfere religiosa, sessuale, di abusi, scatologica e di handicap mentale (McEnery in Beseghi 2016); le stesse compaiono in SoLiSS_subs, riscontrabili in sette testi su dieci. Gli assenti sono il dialogo tra Amadeus e Rettore a Sanremo e Vola, Luna Rossa in entrambi i casi; Un Posto Al Sole presenta il turpiloquio solo nel corpus dei dialoghi.

In SoLiSS_subs, su un totale di 159 volte, 76 sono afferenti al lemma cazzo, costituendo circa la metà delle parole volgari utilizzate nei sottotitoli. Da questi risultati si conferma in un certo senso il suo status di turpiloquio più diffuso nella lingua italiana[20]. La maggior parte degli usi del lemma è associabile a una funzione catartica (Pinker 2007):

• Se è sbagliato, ‘sti cazzi (Il Collegio)
• Che cazzo hai nella testa? (Veloce Come il Vento)
• Oh, cazzo! (Vostro Onore)
• Chi cazzo ha detto che un insegnante deve avere gli occhiali? (Noi).

Veloce Come il Vento rappresenta la pellicola con le maggiori occorrenze di turpiloqui (119 occorrenze nei sottotitoli e 124 nei dialoghi originali), in special modo riferite al personaggio di Loris. La scelta di preservare le parole volgari nei sottotitoli ha come ulteriore utilità, oltre alla già citata caratterizzazione del personaggio, anche la volontà di compensare la cancellazione di quegli elementi regionalmente connotati, ma semanticamente ambigui. Notevole è il mantenimento nei sottotitoli, anche se in quantità ridotta rispetto all’utilizzo effettivo, dell’esclamazione blasfema Cristo! detta dallo stesso protagonista. Diverso è il caso de Il Collegio, dove l’espressione Mado’! non è stata riportata nei sottotitoli, forse per rispettare i caratteri per secondo (cps), forse per evitare di sottolineare il tratto blasfemo dell’espressione nei discorsi di giovani adolescenti. Invece, in Vostro Onore il turpiloquio risulta piuttosto presente nelle situazioni concitate e caratterizzate da registro e stile colloquiali, e viene perciò reso anche nei sottotitoli:

• De Mauri è un coglione
• Dimmi cosa sei? Una testa di cazzo.
• E allora è uno stronzo
• Vaffanculo, Matteo.
• Non sai accendere un’auto, però le cazzate le dici bene.

Le prime due frasi vengono dette dal protagonista, un giudice milanese, nei confronti di suo figlio e le ultime tre sono pronunciate da ragazzi liceali. Ultimo elemento significativo è il mantenimento del turpiloquio regionalmente connotato sfancula, di ambito sessuale-abusivo, che Sabrina Ferilli pronuncia nel suo monologo a Sanremo. Esso è stato tenuto probabilmente sia per veicolare il parlato informale e umoristico, sia per i tempi concitati di stesura dei sottotitoli durante una diretta.

Il tag PAR, riferito a forme sintattiche marcate, è rappresentativo delle strategie di condensazione. Esso ricorre nel corpus SoLiSS per 434 volte, di cui 107 appartengono al sub-corpus dei sottotitoli e 327 a quello degli script, di ben tre volte superiore rispetto ai sottotitoli. Tuttavia, le motivazioni principali di tale differenza sembrerebbero derivare dalle classiche convenzioni spazio-temporali. Qui gli elementi più conservati nei sottotitoli sono la modifica dei tempi verbali in presenza di periodi ipotetici:

• Se non me lo danno, significa che non dovevo averlo (Il Collegio)
• Se facciamo i diagrammi a blocchi, li trasformiamo in figure geometriche (L’Amica Geniale)
• Se torna tua figlia? (Un Posto Al Sole)
• Se lo faceva un altro, sai che cosa gli dicevano? (Veloce Come il Vento),

così come la dislocazione, principalmente a sinistra:

• questa conversazione me la porterò a lungo dietro (Il Collegio)
• alcuni segnali di cambiamento […] li ho colti (Il Collegio)
• il programma d’esame me l’aspettavo (Il Collegio)
• il bambino lo cresco io (L’Amica Geniale)
• a te non ti devono pigliare per fesso (Màkari)
• le mie reazioni le decido io (Un Posto Al Sole)
• che vi prende a tutti e due? (Veloce Come il Vento)
• lo ridai, l’esame (Vostro Onore)
• a parlare, devono essere le vittorie (Vola, Luna Rossa),

quest’ultimo caso senza clitico di ripresa, ma con l’oggetto dislocato come focus contrastivo (Rizzi 1997; Benincà 2001), identificabile dall’intonazione del parlante.

Sono inoltre riconoscibili il mantenimento della frase scissa:

• questo collegio non è che mi ha spento (Il Collegio)
• chi è che l’ha scritto? (Il Collegio)
• che cos’è che l’ha fatta agitare? (Il Collegio)
• è che non mi piace essere preso per fesso (Màkari)
• è che di me non ti fidi (Un Posto Al Sole)
• è lui che si è buttato (Vostro Onore),

il che polivalente:

• sai se c’è della legna, che devo fare il barbecue? (Veloce Come il Vento)
• non dare i ciccioli a Pina che le fanno male (Veloce Come il Vento)
• levati quel cazzo di tutore che non c’è niente lì sotto (Veloce Come il Vento)

e il ci attualizzante, talvolta con funzione di clitico:

• con le rime c’ho pure dei problemi (Ferilli, Sanremo)
• penso ai pettorali, non tuoi, ma che c’ha Ibra (Ferilli, Sanremo)
• [fesso] ci sei di natura (Màkari)
• secondo me non ci stanno bene gli interventi delle mamme (Noi)
• io non ci torno in quella scuola di merda (Veloce Come il Vento).

L’impiego della forma in aferesi ‘sto per l’aggettivo dimostrativo questo in funzione deittica è poco ricorrente nei sottotitoli (otto volte contro le 42 nei dialoghi originali) e riscontrabile in otto testi su dieci:

• Eppure ‘sto ragazzo insieme a Fiore (Ferilli, Sanremo)
• Non vi muovete, o ammazzo ‘sto cornuto! (Màkari);

sono inoltre visibili l’uso di aggettivi in funzione avverbiale e bella come rafforzativo di altri aggettivi:

• Va più veloce (Veloce come il Vento)
• Uno che andava forte (Veloce come il Vento)
• Cinque serate belle emozionanti (Ferilli, Sanremo)
• È bella sana (Il Collegio).

Infine, il tag COL ricorre nel corpus SoLiSS per 219 volte: 84 in SoLiSS_subs e 135 in SoLiSS_script. Notevole è la conservazione nei sottotitoli de Il Collegio della forma tronca raga o ragà o anche raga’ stante per ragazzi e ragazze, dove una sua cancellazione influirebbe negativamente sulla percezione generale dell’immagine dell’adolescente medio. Lo stesso accade per vabè o vabbè, di uso colloquiale e informale, aspe’ per aspetta con valore imperativo e il vocativo prof, di utilità anche spaziale. I sottotitoli riportano fedelmente anche i prestiti dall’inglese tipici del gergo giovanile: mood (spirito, attitudine), fly down (càlmati, imperativo), bestie (migliore amico), come anche l’espressione aggettivale più diacronicamente lontana fru fru (lezioso, frivolo). Altri colloquialismi frequenti nel parlato e inseriti nei sottotitoli sono: casino, broda, rimorchiare, beccare, sfigato, figo, [non capirci una] mazza, scemo, filarsela, strafiche, bucarsi e rompere nel senso di infastidire. Tra i vocativi figurano Ciccio! con valore peggiorativo, pa’ per papà e Rebby per Rebecca con valore vezzeggiativo:

• Ciccio! Mi senti? Stai calmo. Ti togli dai coglioni e ti prendi questo bel regalino. Dài! (Veloce come il Vento)
• Pa’, mi senti? (Veloce come il Vento)
• Rebby, te con i capelli lunghi! (Il Collegio).

Questi ultimi, così come l’abbreviazione prof (presente solo ne Il Collegio), sono meno ricorrenti in SoLiSS_subs rispetto a SoLiSS_script, ma ciò è dettato da ragioni convenzionali riguardo i cps.

Per riassumere, le etichette hanno mostrato dei comportamenti diversi nei due sub-corpora. Le frequenze relative (calcolate su diecimila token) delle occorrenze di ciascun tag come mostrato su LancsBox sono illustrate sotto:

Tabella 1: Frequenze relative dei tag e differenze nei due sub-corpora in SoLiSS.

Dalle differenze tra i due valori dei corrispettivi sub-corpora si conferma che il turpiloquio è l’elemento maggiormente rispettato nella sottotitolazione in confronto al parlato originale perché riconoscibili, con una differenza di appena 5,11; seguono i colloquialismi, con 17,74 di differenza; invece, le feature dialettali, i regionalismi e le forme sintattiche marcate sono quelle più riadattate con le forme standard per ragioni di comprensibilità con rispettivamente 38,16, 75,82 e 94,37 di differenza.

Gli esempi sopra mostrati e i dati riportati in tabella dimostrano una generale attenzione a rispettare la variazione nei sottotitoli di recenti produzioni italiane su RaiPlay secondo la normativa RAI del 2021. Nonostante un simile rispetto, ci sono alcuni casi, specialmente riguardanti variazione diatopica e marcatezza sintattica, in cui le strategie adottate hanno piuttosto costituito un ostacolo alla contestualizzazione dei personaggi. Sebbene la tendenza ad attuare strategie di semplificazione sia talvolta necessaria per l’intelligibilità, in taluni casi si corre il rischio di peccare di disattenzione nel comunicare efficacemente la caratterizzazione complessiva di un personaggio nel rispetto dei codici di significato.

6. Conclusioni

Il presente lavoro ha dimostrato un generale rispetto delle varietà sociolinguistiche dell’italiano parlato nei sottotitoli accessibili di produzioni AV RAI. L’indagine ha coinvolto un totale di dieci testi derivanti da altrettanti programmi televisivi recenti che avrebbero potuto costituire esempi concreti del trattamento dell’italiano parlato in fase di trasposizione scritta.

Il corpus ad hoc SoLiSS ha raccolto sia i dialoghi originali, che i sottotitoli corrispondenti al fine di offrire un’indagine comparativa. Questi ultimi sono stati estrapolati da RaiPlay per due ragioni: verificare il trattamento della variazione nei sottotitoli su una piattaforma streaming di crescente utilizzo e usufruire della costante disponibilità dei programmi a prescindere dalla loro messa in onda sulle varie reti. Il corpus è stato taggato in via preliminare seguendo l’esempio del POS-tagging per favorire immediatezza nella ricerca in sé e nella processazione delle informazioni da parte del software LancsBox. I tag #_REG, #_DIAL, #_TUR, #_PAR e #_COL sono stati ideati tentando di rispecchiare gli elementi lessicali e grammaticali più rappresentativi dell’italiano parlato.

Dai dati mostrati è emerso che il turpiloquio è l’elemento maggiormente rispettato nei sottotitoli, seguito dai colloquialismi, mentre le features dialettali, regionali e sintatticamente marcate vengono principalmente standardizzate per rendere i sottotitoli facilmente fruibili dagli spettatori sordi, considerandone la velocità di lettura in presenza di elementi substandard[21]. Come sottolinea Orletti (2016): “Vanno preferite le parole brevi, di alta frequenza, le forme verbali semplici, le frasi semplici, la paratassi all’ipotassi” (225). Tuttavia, la prassi della riduzione linguistica nei sottotitoli “rende omogenea una fascia della popolazione che omogenea non è in termini di padronanza linguistica, appiattendo, anche oltre il necessario, la varietà linguistica” (Orletti ed Eugeni 2019: 534). Pertanto, un compromesso tra indagine scientifica e attività professionale permetterebbe di garantire la qualità dei sottotitoli in maniera più precisa, valutando caso per caso la necessità di adattare i dialoghi senza rischiare di perdere informazioni necessarie alla descrizione di personaggi e scene. Jorge Díaz Cintas e Aline Remael (2021) pongono l’accento proprio sulla funzione della variazione, specificando che occorre preservarla nei sottotitoli se caratteristica di un personaggio o di una situazione (182-3)[22]. Anche Soledad Zárate (2021) sottolinea questo aspetto:

[…] certain features typical of oral discourse, such as repetitions, hesitations, fillers and redundancy, can be partially omitted in the migration to subtitles without affecting the message too negatively. Yet some indications of such features can still be given in the written version, especially if they contribute to characterisation and have diegetic value, that is, narrative meaning (41).

In merito al problema della reperibilità dei prodotti sottotitolati su RaiPlay, lo studio ha mostrato quanto il picco di utilizzo della piattaforma nel 2021 rispetto all’anno precedente sembri essere in dissonanza con l’effettiva quantità di ore sottotitolate che la stessa propone. Sarebbe interessante condurre un'indagine più approfondita sulla tipologia di programmazioni sottotitolate e sull’ammontare dei tempi di fruizione del servizio da parte del pubblico sordo e ipoudente.

Dal punto di vista metodologico, lo studio pecca di una mancata applicazione di sistemi di parsing che spezzino la frase in sintagmi o porzioni di testo a favore di un’analisi computazionale sintattica e semantica. Inoltre, SoLiSS è ben lontano dall’essere un database ufficiale dei sottotitoli per sordi di RaiPlay, in quanto progetto preliminare. Come sviluppo futuro, la creazione di una banca dati ufficiale permetterebbe di monitorare sia la qualità generale della sottotitolazione, in special modo riferita agli elementi del parlato che sottostanno alle maggiori difficoltà diamesiche, sia la quantità di ore sottotitolate sul servizio streaming.

Abstract

In every society, there are individuals with different needs and abilities. This makes every society unique and distinctive. In order to ensure that social groups with different abilities are fully integrated into public life, a wide range of accessibility issues are relevant. Media accessibility for deaf and hard-of-hearing people is one of them. After the Soviet occupation, when the Baltic countries - Lithuania, Latvia, and Estonia - regained their status as independent states, the distorted public perception of people with disabilities, their adaptation possibilities, and a widespread frame of mind for their social life still face the relics of the Soviet occupation regime - a certain stigmatisation of these social groups. Nonetheless, with changes in the legislative framework, modern foreign practices and encouraging examples, this flawed attitude towards people with disabilities is gradually changing. This paper aims to present the needs and expectations of Lithuanian hearing-impaired people and discuss perspectives on media accessibility for this particular audience in Lithuania. Firstly, statistical information about the Lithuanian target group will be provided, then the principles of media adaptation for the Lithuanian deaf and hard of hearing will be discussed and finally, the needs of the target group in terms of media accessibility will be considered. The data presented in this paper includes results from the research project “Inclusive Culture: the Study on Accessibility of Audiovisual Products for the Visually and Hearing Impaired” (KlaRega), which was conducted between 2021 and 2023.

Keywords: Lithuanian audience, audiovisual translation, media accessibility, subtitling for the Ddeaf and hard of hearing

Introduction

Lithuania is one of the Baltic countries that was under the regime of the Soviet Union for over 40 years. “Not only was the expression of Baltic culture strongly suppressed in every respect” (Tulun 2014: 140), but also an inadequate distorted awareness of the harmonious society raised by the Soviet Union had a significant impact on the social attitude towards the abilities and needs of persons with disabilities. The promoting idea about the non-existence of this social group did not stimulate any discussion about the accessibility of cultural products for this marginalised type of consumer. Furthermore, such stigmatisation has given rise to two-sided consequences; namely, a disinterested, unsympathetic and irresponsible social approach to people with disabilities and their needs has been developed; secondly, self-isolation of this group and their decision to go into a social decline have been experienced. In Lithuania, the onset of a novel perspective for people with disabilities began with the implementation of the Law on the Social Integration of the Disabled (1991, 2005) and the United Nations Convention on the Rights of Persons with Disabilities and Optional Protocol (2008), which have highlighted the state policy for the integration of people with disabilities through physical, medical, social, educational and cultural spheres of life by treating those persons with “respect for their inherent dignity” (Article 1:4). Therefore, a barrier-free life for people with disabilities must be ensured in Lithuania ever since. According to the state policy, access to audiovisual content in the areas of cinema/theatre, television and the Internet should also be guaranteed. However, due to a lack of both specialists’ knowledge and competence in producing audiovisual material accessible to deaf and hard-of-hearing viewers, the implementation of the measures has not been put into standard and accepted practice yet. The first attempts to adapt films and performances for the hearing impaired have only been made in Lithuania since 2017.[1]

In this context, the paper aims to provide an overview of Lithuanian practice in relation to the accessibility of audiovisual material for deaf and hard-of-hearing people, both in terms of intralingual and interlingual parameters. Firstly, historical aspects related to the notion and tendencies of accessibility of audiovisual content for the target audience in Lithuania will be presented. Then, principles and strategies associated with subtitling for this audience in the Lithuanian setting as well as respective accessibility options suggested by the target viewers will be discussed.

The paper focuses on the results of the research project “Inclusive culture: the study on accessibility of audiovisual products for the visually and hearing impaired” (KlaRega) (2021-2023) which was conducted by the author and her colleague, Laura Niedzviegienė (Vilnius University, Lithuania). In this case, empirical subtitling research relevant to the issue, dominant features and principles of subtitling audiovisual products for the deaf and hard of hearing (SDH) is introduced. First and foremost, this includes the subtitling needs of hearing-impaired viewers. As this is a heterogeneous target audience offered to enjoy films with standard subtitles in Lithuania, certain preferences for acceptable Lithuanian subtitles to ensure adequate readability for most viewers in this group are reviewed. Finally, the major findings and insights of this research are briefly discussed and suggestions for further research are presented.

1. The hearing impaired as a certain social cluster in Lithuania

1.1. A stigmatising historical perspective on persons with disabilities

For more than 40 years, Lithuania was suppressed in various aspects by the Soviet regime. Not only language policy, but also cultural events such as music, art, literature, theatre and films were affected by the strict Soviet guidelines. Alongside the suppression of culture, the ethnic composition of the Baltic region and the apparent exclusion of the mother tongue from various areas of social life, the issue of disability had the tendency to be ignored as well. As rightly pointed out by Oleg Poloziuk (2005: 9), “there was no sex or disability in the Soviet Union. Under Soviet rule, these things were unrealistic and non-existent; therefore, persons with disabilities in the former Soviet Union remain in many respects an unknown population”. This reality can be illustrated by one historical fact: “During the 1980 Olympic games in Moscow, a Western journalist inquired whether the Soviet Union would participate in the first Paralympic games, scheduled to take place in Great Britain later that year. The reply from a Soviet representative was swift, firm, and puzzling: ‘There are no invalids in the USSR!’” (Fefelov 1986 in Phillips 2009). This official’s denial “of the very existence of citizens with disabilities encapsulated the politics of exclusion and social distancing that characterised disability policy under state socialism. Historically throughout the former Soviet bloc, persons with physical and mental disabilities have been stigmatised, hidden from the public, and thus made seemingly invisible (Dunn and Dunn 1989 in Phillips 2009). Due to these ideological factors, hearing-impaired people in Lithuanian were also pushed to the margins of social life: they lived in isolated spaces (either with relatives or in special boarding schools); cultural life was also inaccessible to them, even inclusion in world sporting competitions such as the Paralympic Games was only possible for them to a limited extent.[2] Furthermore, the language used by the Soviet regime about deaf people was also often not very inclusive. In many cases, Soviet authorities did not employ what is now considered correct or respectful terminology when referring to deaf individuals. Instead, they used outdated or derogatory terms that were stigmatising or insensitive. In the post-Soviet states, for instance, the practice of referring to deaf and hard-of-hearing individuals as deaf-mute, deaf-dumb, or even invalids, disabled was frequently used by hearing people. The attitude towards sign language has also been inappropriate and disrespectful since the Soviet era. Sign language as such was not regarded as an effective means of communication for deaf people; on the contrary, in the social context, it was portrayed as a problem of backwardness (Show 2011: 192).[3] Consequently, since the Soviet times, deaf people have developed the audist notion[4], that is, they have considered themselves as inferior members of the hearing society; therefore, they had no expectations of better access to cultural life. Although Lenin’s propaganda message that “of all the arts the most important for us is the cinema” (first published in Kinonedelia 1925) was widespread in the country, audiovisual material was not accessible to the Lithuanian deaf and hard of hearing either. The first reason for this was related to the predominant film translation mode – the so-called ‘Soviet voice-over’, which was performed “in a straight, formal tone without regard to lip movements, gender, or character emotions and the original audio was still audible in the background” (Shih 2020). Subtitled films were the exception rather than the norm. On the other hand, as already mentioned, the deaf audience like all persons with disability were “invisible”, non-existent and any ideas about film adaptations were regarded as utopia, as fantasy rather than reality. As a result, deaf and hard-of-hearing viewers in Lithuania at that time, were more likely to enjoy the visuals than to engage in the film plot, read the characters’ expressed emotions, experience the mood-setting music, and so on. This distorted perception of media accessibility, both in society and among deaf people themselves, was particularly pervasive until a legal basis (the Law on the Social Integration of the Disabled [1991, 2005] and the United Nations Convention on the Rights of Persons with Disabilities and Optional Protocol [2008]) has emerged to change this situation in Lithuania. Nowadays, according to the state policy, in the country access to all types of media should be equally guaranteed for everyone, regardless of physical and/or mental disability. Therefore, filmmakers, broadcasters and distributors in Lithuania have launched new initiatives to adapt the media for the target viewers, taking into account their physical and sensory abilities.

1.2. Statistical and socio-cultural overview of deaf and hard-of-hearing people in Lithuania

As far as the size of the target audience is concerned, it should be noted that there are around 30,000 hearing-impaired people in Lithuania.[5] According to the statistics of the Lithuanian Association of the Deaf, there were about 8000 deaf people in Lithuania in 2022 and more than 3000 of them were registered as members of the Lithuanian Association of the Deaf.

Comparing the data with other groups of people with disabilities, according to the Ministry of Social Security and Labour (2022), there were about 223,000 people with disabilities in Lithuania at the end of 2022, of whom almost nearly three per cent were hearing impaired.

From a socio-cultural perspective, deaf and hard-of-hearing people in Lithuania, as in other countries, represent a heterogeneous group from a cultural and sociolinguistic point of view, according to the model of identity development of deaf people proposed by Neil Stephen Glickman (1993: 62-63) and the scientific findings of Agnieszka Szarkowska (2013). In general, four large groups can be distinguished. The first group is culturally hearing and includes persons who have higher degrees of residual hearing, perceive deafness as a disability (rather than a cultural difference), typically identify themselves with the hearing culture instead of the Deaf culture (Iriarte 2017: 22) and give preference to oral communication. Their oral language skills vary depending on “when the hearing loss was sustained, that is in childhood or adulthood, the degree and nature of the hearing loss, and the effectiveness of subsequent therapeutic programs” (Ross 2005: 7). The second, culturally marginalised group, includes individuals who are part of deaf and the hearing world but do not feel comfortable in either world (Glickman 1993: 93); for instance, deaf children born into hearing families. The third group is associated with the process of immersion in the deaf world; it thus includes people who do not see their deafness as a disability and view themselves as a “cultural linguistic minority” (Adams and Rohring 2004: 70), using only sign language and never speaking with their voice (Glickman and Carey 1993: 277; Glickman 1993: 99). Finally, the last – bicultural – group, which includes the hearing impaired who recognise deafness as a cultural difference and feel comfortable in both deaf and hearing worlds (Glickman 1993: 100-104). Consequently, “there are those who were born deaf, use sign language as their mother tongue and identify themselves with the Deaf community; those who are hard of hearing, who often have residual hearing; and those who are deafened and who lost hearing at a later stage in their lives and have an oral language as their first language” (Szarkowska 2013: 69).

Despite these explicit groupings, as the data from the research project “KlaRega” reveals, deaf people in Lithuania can still be relatively divided into those born before 1995 (when the Lithuanian government signed a resolution legalising sign language as the official language of the deaf community) and those born after 1995. The former group includes persons who lived under the Soviet regime, where communication in sign language was humiliated, as mentioned above. Therefore, persons from this group are excellent lip-readers and have a sufficiently sophisticated auditory apparatus to pronounce words clearly and combine them into sentences. There are also some who are bilingual and able to lip-read in several languages, mostly Lithuanian and Russian. Deaf people born after 1995 are young people who are less likely to be able to read Lithuanian texts, and only a few can lip-read. Due to the pervasive influence of English in all areas of their lives, small English texts have risen in their favor. However, the preference is given for sign language interpreting in all areas of their lives, including media accessibility (Kerevičienė and Niedzviegienė 2022a).

Considering the heterogeneous nature and different cognitive abilities of the Lithuanian deaf and hard of hearing, their needs for media accessibility also appear to be diverse.

2. Deaf and hard-of-hearing viewers as a specific audience in Lithuania

2.1. Research analysis and methodology

Nowadays media accessibility is increasingly becoming a realistic ambition for producers of audiovisual material. In Lithuania, filmmakers, national broadcasters, audiovisual distributors are targeting socially marginalised audiences and trying to apply foreign practices and adapt audiovisual products for the Lithuanian deaf and blind. However, the adaptation methods tend to vary as media practitioners do not sufficiently understand the specificities, physical abilities, and needs of the target groups. Therefore, this variation does not always result in a quality adaptation of audiovisual (AV) material for the target audiences. Therefore, the research project “Inclusive culture: the study on the accessibility of audiovisual products for the visually and hearing impaired” (KlaRega) (2021-2023) was conducted. One of its main tasks was to identify the needs of the Lithuanian deaf and hard of hearing in terms of media accessibility. The project is included in the national research program “Welfare Society” with the purpose “to carry out integrated scientific studies of the preconditions for a welfare society and its development in Lithuania” as well as making recommendations for the development of audiovisual accessibility for the special social groups of the Lithuanian population.[6]

Methodologically, an online survey and interviews were used to identify the needs of Lithuanian deaf and hard of hearing in terms of already adapted AV products in Lithuania. All persons with hearing impairment and those registered as members of various deaf organisations in Lithuania were invited to participate in this online survey. However, due to objective circumstances related to the threatening situation of COVID-19 and the low activity of participants, only a small number of deaf individuals expressed their willingness to participate in the study. Thus, a total of 58 people from various deaf associations in six major Lithuanian cities (Vilnius, Kaunas, Panevėžys, Šiauliai, Klaipėda, Kėdainiai) took part in the online survey. As the results of the survey revealed, about half of the respondents were middle-aged and older (32 per cent of those over 50 and 21 per cent between 41–50). 16 per cent of participants were 31–40 years old and 14 per cent of respondents were between 18–25 years old. Most passive respondents were in the 26–30 age group (nine per cent). Secondly, the majority (83 per cent) of respondents stated that sign language is their first language and only a small proportion of respondents (16 per cent) indicated that their first language is still Lithuanian. Due to special communication and learning needs, almost all respondents (93 per cent) were educated in centres for deaf and hard-of-hearing pupils either from infancy (33 per cent) or from the first grade (48 per cent), except the small group of the interviewees (seven per cent) who attended national standard schools for hearing students. What concerns education, it was indicated that the majority of deaf and hard-of-hearing respondents (64 per cent) had only completed secondary school, while some respondents managed to achieve high and higher education (12 per cent and 17 per cent respectively). In terms of the degree of hearing loss, the majority of the survey participants were deaf (72 per cent), some were hearing impaired from birth (81 per cent), others were deafened in childhood (eight per cent) or later (nine per cent). Despite their sensory diversity, the majority of respondents lived in hearing families (76 per cent), other deaf and hard-of-hearing interviewees lived either in deaf (17 per cent), hard of hearing (three per cent), or mixed (three per cent) families.

In addition, a comprehensive online questionnaire with 55 questions on the needs and experiences of deaf and hard-of-hearing people in Lithuania concerning the adaptation of AV products was developed for the study. The questionnaire was conducted using a Google survey form. The main part of the survey form was focused on the qualitative and quantitative aspects of AV products adapted in Lithuanian theatres, broadcast on TV channels and distributed on various film platforms. The questionnaire contains questions to recognise respondents' interests and habits concerning AV production (for example, what kind of films and performances they like to see, what their preferences are: watching TV, going to the cinema, using Internet platforms, which subscription to streaming AV production they prefer; how often they visit public places related to the screening of AV productions (cinema, theatre), what their opinion about films with SDH is, what benefits and drawbacks they recognise, what suggestions they would make to improve subtitles and surtitles in order to gain more useful audible information, and so one.). In addition to the questions, in this survey form, short (up to one minute) video fragments of AV products were presented, which were adapted for the target audience according to the proposed methodology. The video excerpts were used to identify the needs, preferences and interests of the target audience and, on this basis, to create guidelines for adapting AV products to the target audience[7]. To promote the survey, the Lithuanian Association of the Deaf prepared a short video in sign language to provide information about the survey and the data to be collected. Besides, the questionnaire was also adapted for deaf respondents; that is, questions and possible suggested answers were interpreted into Lithuanian sign language and recorded. In addition to this questionnaire, a free-form interview-survey method was also used to investigate the needs of members of the target groups. It was designed to assess the quality of accessibility of the already adapted AV products in Lithuania and distinguish the benefits of the modified AV material. Therefore, contact meetings were organised with members of Lithuanian deaf associations. Each meeting-interview was scheduled to last up to one hour. First, a short presentation was given on the practice of accessibility of AV products (films and performances) abroad and in Lithuania; afterwards, various excerpts from films, cartoons and performances with specially designed SDH were demonstrated. After this, interviews were conducted with deaf and hard-of-hearing people. They included a series of pre-determined targeted questions aimed at clarifying the needs of the target groups; namely, scale-type questions (with response options such as “I agree/disagree/partially agree with statement X”) and funnel-type questions (where a broad, usually open-ended question is given at the beginning, and additional questions later narrow down the topic by detailing the information). In order to establish principles for the adaptation of AV material for deaf and hard-of-hearing viewers and to identify their needs for AV production, the following topics were discussed: strategies for rendition of audible information, including off-screen sounds, personages’ manner of speaking, the background music and finally the linguistic character of the subtitled text. Hence, the results of the questionnaire and the insights from the discussion helped to determine general trends in the reception of adapted AV products and establish the needs of the Lithuanian target group (that are briefly presented hereinafter).

2.2. Delivery modes of audiovisual material for Lithuanian hearing-impaired viewers

Deaf and hard-of-hearing people are a special audience as they are not receptive to auditory information due to their physiological characteristics. For this reason, AV material is either interpreted into sign language, subtitled or delivered by a special AV mode - subtitles for deaf and hard-of-hearing viewers (SDH). Due to the aforementioned prevailing Soviet attitude towards the hearing impaired, films and performances in Lithuania (both in cinemas and theatres) are still presented with standard subtitles. In the case of the Lithuanian national broadcaster, AV material is usually displayed with standard subtitles as well; although daily news or other current affairs programmes (political debates, interviews, and so one.) are broadcast with sign language interpretation. Regarding the needs of the target viewers, as the research detected, deaf socio-cultural subgroups of the older generation in Lithuania would prefer edited or standard subtitles, while the hard of hearing opt for verbatim[8] or standard subtitles. The respondents explain that the AV delivery format with standard subtitles is sufficiently suitable for them, as they are already used to it. During the Soviet era, sign language was not recognised as their mother tongue. Therefore, deaf people were taught to lip-read, and better reading skills were developed in educational institutions. For this reason, the Lithuanian target viewers are now able to watch films with standard subtitles by using their lip-reading skills. The situation is different for the younger deaf generation. They would prefer subtitled films and SDH. However, it is expected that sign language interpreters will be used for most audiovisual products. Again, their preference is associated with the fact that the younger deaf viewers in Lithuania have not experienced such social pressure and received information mainly in sign language. Reading long texts therefore seems to require more effort and concentration. On the other hand, viewers in this age group are more technologically literate and have seen foreign AV productions with SDH. Despite these differences, the Lithuanian deaf and hard-of-hearing respondents do not make any complaints about their reading skills and therefore do not question the simplification of subtitled language. The most common reading difficulties encountered by the respondents are related to the inappropriate spotting time of subtitles.

As the research results showed, Lithuanian hearing-impaired people who have developed good reading skills can be considered regular media consumers. Most of the target audience (38 per cent of the respondents) subscribed to streaming AV productions either from various foreign film platforms, or from the Lithuanian Internet platform “Mediateka” (31 per cent of the interviewees). Only some of them consume AV productions from various Lithuanian film platforms (21 per cent of the respondents). As a rule, the majority of the viewers spend more than two hours daily watching TV (almost 80 per cent of the surveyed) or enjoying AV material on the Internet (almost 70 per cent of the interviewees). Only 17 per cent of respondents stated that they have never watched films on the Internet. Regarding viewers’ choice of AV products to watch, the following chart shows the preferences of the Lithuanian deaf and hard of hearing when watching TV (in percentages):

Fig. 1. TV viewing preferences of Lithuanian deaf and hard-of-hearing viewers

Obviously, the news is of the utmost importance to Lithuanian hearing-impaired viewers. On the other hand, the target audience are also engaged in watching films of different genres and enjoy various television programmes.

Therefore, Lithuanian deaf and hard of hearing of different subgroups take an interest in watching AV subtitled productions. They are quite active media users with sufficient reading skills and the vast majority of them (almost 90 per cent) consider it highly relevant to be familiar with all the information they find in the AV material. Both the spoken language - dialogues - and off-screen audible information seem to be significant for the complete understanding of the AV material. Thus, according to the respondents, films and performances with SDH would be the major mode of media accessibility for the Lithuanian target audience.

2.3. Off-screen audible information for Lithuanian deaf and hard-of-hearing viewers

As mentioned above, most of AV products in Lithuania use standard subtitles; therefore, the priority of the hearing impaired is given to SDH, where audible information should be indicated. As the results of the survey show, the majority of the target audience emphasise the importance of information about off-screen sounds in films and performances. In particular, the need to include information about natural and environmental sounds in subtitles is notably highlighted. According to the respondents, the most relevant off-screen audible information includes nature sounds (such as ‘waves crashing’, ‘leaves rustling’), city noises (for instance, ‘traffic noise’, ‘bells ringing’), animal noises (like ‘dogs barking’, ‘horses neighing’), human sounds (as ‘laughter’, ‘applause’), and noises caused by physical actions (for instance, ‘slammed doors’, ‘loudly shutting windows’). The indication of these sounds allows the deaf viewer to understand what is happening in the auditory background of the film, to predict what may relate to the characters’ future actions, to recognise what causes the characters’ reactions and to anticipate what will happen next. The appropriate position of subtitles for rendering such off-screen audible information, according to the respondents, should be in the centre, at the bottom of the screen; they can also be placed at the top or in the corner if important information is displayed below.

In addition to this, the references to music are also worth mentioning. Lithuanian deaf and hard-of-hearing viewers would like to read information about the rhythm of the music, the mood it creates, as well as the changes in the music such as its softness or loudness, pauses, and so one. For clarity, the respondents suggest that various linguistic comparisons such as ‘like a police siren’ and so on, could be used to describe this type of information. Less relevant information for the target audience would relate to the descriptions of the music genre or the musical instruments used. After watching film excerpts with SDH, the respondents prioritised the informative descriptions of sounds as follows:

Fig. 2. Preferences of paralingual information about music in Lithuanian SDH

As for the details that should be indicated when describing the type of music in films and performances, the following aspects were highlighted. Firstly, the majority of the Lithuanian deaf and hard of hearing suggest including the lyrics of a song in the subtitles and adding a musical note (♪ or ♫) at the beginning and end of each subtitle; others recommend adding the name of the song, artists and composers. Secondly, according to the respondents, information about the rhythm of the music and the main musical instruments would also be relevant. In the case of a foreign song, Lithuanian viewers tend to emphasise the translation of lyrics in the subtitles. Conversely, brief information in which only the singer or the name of the song is mentioned is not favourably supported by the respondents. Information on the gender of the singer and the genre of the music in the film was also not necessary for the audience. Thus, summarising the variants suggested by the respondents, it can be concluded that deaf and hard-of-hearing people in Lithuania prefer SDH that contain more extensive information about the off-screen sounds and music played in the film or performance; and, if possible, the translation of the lyrics of foreign songs into Lithuanian should be provided. In addition, the song lyrics should be distinguished from the dialogue subtitles by marking them with musical notes.

2.4. Language rendition in subtitles for the Lithuanian target audience

The plot of the film is inextricably linked to the thoughts and dialogues of the film characters. When the actors’ faces are clearly visible on the screen, deaf people can understand the spoken text by watching their lips. However, if the speakers are in the background or not visible, the content of the dialogue becomes incomprehensible to the deaf viewer. In this case, an additional identification of the speakers becomes significant. Following the practices of various foreign broadcasters, the speakers can be distinguished either by name tags, colours or speaker-dependent placement. The Lithuanian target audience express their preference either to use name tags in capital letters (14 per cent of the interviewees) or colour-coding (26 per cent of respondents). Only a small number of respondents (ten per cent) reject the identification of characters, stating that the content of the speech, not the speakers, included in the standard subtitles is essential in a film. They therefore prefer single colour subtitles. The discussions revealed that this opinion is mainly held by the hard-of-hearing and younger deaf viewers.

Furthermore, various feelings and emotional states experienced by characters are encoded not only in the film content that is uttered but also in the manner in which it is said. Although these audible aspects are conventional to the ordinary viewer, they are inaccessible to deaf persons without specific written textual references. Consequently, Lithuanian hearing-impaired viewers are concerned about this kind of information being added to the subtitles. Again, the indication of the speech features is particularly important when the actors’ faces are not visible. As for the positioning and formal presentation of subtitles describing the issues, informative references should be displayed in square brackets next to the subtitled dialogue.

When subtitling films with multilingual content, the Lithuanian hearing impaired express their suggestions for indicating the foreign language spoken by the characters with an additional label in the subtitles and providing a translated version of the dialogue. According to the respondents, this method would not require their deeper linguistic knowledge; on the contrary, it would reveal information about the content of the dialogue, and at the same time help to understand the idea of the AV material. Therefore, the Lithuanian hard of hearing prefer the translation of a multilingual text together with a foreign language labelling.

Since sign language is the native language for the majority of deaf people and another language, such as Lithuanian, is already the second one that the target viewer has to learn and use daily, it is common in foreign practices to simplify the subtitled text in SDH. Regarding this aspect, the results of the questionnaire and the discussions identified two different groups of opinions among the Lithuanian deaf and hard of hearing: some respondents express their preference for the comprehensible, grammatically correct, edited subtitle text that conveys only the main ideas to read and thus provides the opportunity to grasp the information quickly; consequently, viewers will gain the great opportunity to expand their language skills. Others, however, prefer a less condensed and edited, non-standard subtitled text (that is less censored conversational content) and therefore retain the original, albeit long, subtitled version of the dialogues. However, this variant of the subtitled text would require the audience to have good speed-reading skills. Yet again, the diversity of opinions and suggestions seemingly is based on the experienced insufficient social influence on the deaf, which leads to different cognitive abilities and preferences of subgroups of the target viewers.

3. Conclusions

In view of the above, the results of the project research can conclude that Lithuanian deaf and hard of hearing constitute a heterogeneous group of viewers whose varying needs in terms of media adaptation are related to their age, previous experience, reading skills and technological literacy.

Despite this diversity, relying on the research data, the Lithuanian target audience emphasise the value of specialised subtitles (SDH), which would contain clearly visible and comprehensible subtitles with informative cues about off-screen sounds (highlighted by 88 per cent of the respondents), speaker identification and references to the way the characters speak, their emotional state (pointed out by 45 per cent), the inclusion of information about the background music, the pieces of music played in the film (emphasised by 35 per cent), references to the foreign languages spoken in the multilingual film and the translated content of the dialogue (mentioned by 35 per cent). Some respondents (38 per cent) emphasise the specificity of the characters’ speech, where any censorship of dialogue, including swearing, is rejected. Finally, the majority of deaf and hard-of-hearing people (65 per cent) stress the grammatical correctness of the language in the subtitles. In addition, the results of the survey showed that deaf and hard-of-hearing viewers consider the subtitling practices in Lithuania unacceptable due to the limited diversity of AV production with SDH.

Furthermore, Lithuanian hearing-impaired viewers will no longer be considered atypical if they are no longer stigmatised any longer. In the future, they will communicate and collaborate with AV specialists, researchers, filmmakers and broadcasters to share information about their specific needs and abilities. On the other hand, media accessibility will improve significantly when professionals are more aware of techniques for adapting AV productions to the target audience and implement guidelines on how to make all types of AV material accessible to Lithuanian deaf and hard-of-hearing audiences.

Abstract

English:

This article is an initial approach to accessibility for D/deaf and hard of hearing people in audiovisual products for museums. It is based on the study of the translation strategies used in the subtitling for D/deaf and Hard of Hearing people in a corpus of two different types of videos created for museums: unscripted live videos and pre-recorded videos. To this end, an analysis is made of the differences between these two types of videos, the challenges they pose for the D/deaf community and the needs from this community when accessing specialized or semi-specialized content. Special emphasis is placed on the use and relevance of simplification and linguistic adaptation strategies, usually found in easy-to-read texts, to transmit information in a more orderly and coherent way, as well as to facilitate access to knowledge for those members of the D/deaf community with greater reading and writing difficulties.

Spanish:

Este artículo supone una investigación inicial sobre la accesibilidad para personas con discapacidad auditivas en contextos audiovisuales museísticos a partir del estudio de las estrategias de traducción utilizadas en el subtitulado para personas sordas de un corpus de vídeos creados para museos de dos tipologías diferentes: vídeos creados en directo y vídeos pregrabados. Para ello, se realiza un análisis de las diferencias entre los dos tipos de vídeos, los retos que plantean para la comunidad sorda y las necesidades planteadas por esta comunidad a la hora de acceder a contenido especializado o semiespecializado. Se hace hincapié en el uso y la pertinencia de las estrategias de simplificación y adaptación lingüística, propias de la modalidad traductológica de la lectura fácil, a la hora de transmitir la información de forma más ordenada y coherente, así como para facilitar el acceso al conocimiento para aquellas personas de la comunidad sorda con mayores dificultades lectoescritoras.

Keywords: subtitling for the Ddeaf and hard of hearing, audiovisual products for museums, simplified language, Ddeaf comunity, subtitulación para personas sordas, productos audiovisuales museísticos, lenguaje simplificado, comunidad sorda

1. Introducción

En la actual y globalizada era digital, resulta evidente que los museos ya han pasado de ser un lugar casi sacrosanto donde se exponían artefactos de épocas anteriores a convertirse en un lugar de acceso al conocimiento donde el principal agente es el usuario (Mairesse 2020: 37).

Una de las técnicas utilizadas para ofrecer una visión contemporánea, interactiva y atractiva es el uso de productos audiovisuales en multitud de formatos y contextos (Serrell 2002). Desde visitas interactivas desde la comodidad del hogar con Google Arts and Culture (Reigosa Lombao 2020) al uso de realidad aumentada para ampliar las fronteras del conocimiento (Rhodes 2015), los vídeos ya forman una parte esencial tanto de la estrategia de marketing en museos como de la experiencia museística.

Por otra parte, el acceso equitativo al conocimiento supone cada día más una preocupación emergente en el ámbito de la investigación (Massarani y Merzagora 2014) y numerosos ámbitos relacionados con el acceso al conocimiento se interesan por ofrecer sus servicios en un marco más justo e igualitario, como es el caso de los museos y otros entornos de ocio cultural y educativo. Para poder realizarlo de forma eficiente, tal y como argumenta Viviane Panelli Sarraf (2008: 38), es necesario plantear todos los espacios y materiales desde una óptica accesible para que cualquier persona, independientemente de sus necesidades, pueda acceder al conocimiento.

La presente contribución se sitúa en esta convergencia y profundiza en las estrategias traductológicas utilizadas en el subtitulado para personas sordas (SpS) de vídeos alojados en sitios web de museos y que pertenecen a dos tipologías diferentes: vídeos pregrabados y vídeos grabados en directo. A partir de un estudio de corpus de cincuenta y tres textos audiovisuales (22 vídeos pregrabados y 31 vídeos grabados en directo, todos ellos del año 2020) con SpS, pertenecientes a dos museos de la ciudad de Granada (España), el Museo Arqueológico y Etnológico de Granada (MAEG) y el espacio BioDomo del Parque de las Ciencias, se analiza este fenómeno transformador y comprendemos mejor el papel que estos desempeñan en la difusión de conocimiento. Por último, se observa, a raíz del etiquetado con el software de análisis cualitativo MAXQDA de las estrategias de traducción según la clasificación de Silvia Martínez-Martínez (2015) que se han utilizado, la pertinencia o no del uso de técnicas de simplificación intralingüística en los productos en directo para contrarrestar la velocidad y los elementos propios de la oralidad, así como la presencia de simplificación léxica en los productos pregrabados para la mejora de la comprensión de los términos especializados.

En este estudio se persiguen varios objetivos. En primer lugar, se estudian los desafíos a los que se enfrentan los subtituladores y las subtituladoras de productos fílmicos especializados en museos en términos de accesibilidad para personas con discapacidad auditiva. Además, se valoran la pertinencia de las estrategias de lenguaje simplificado, propias de la modalidad traductológica de la Lectura Fácil (LF), en el SpS en estos contextos a partir de un análisis de corpus. Por último, se analizan las estrategias de traducción más utilizadas en el subtitulado de cada uno de los tipos de productos audiovisuales museísticos.

Para ello, comenzamos con un apartado dedicado al objeto de estudio, los productos audiovisuales museísticos, valorando el porqué de su utilización y su éxito; continuaremos con un apartado dedicado a los usuarios, la comunidad sorda, analizando su acceso al conocimiento especializado; a continuación, procederemos a describir la metodología y finalizaremos con el análisis de los resultados, prestando especial atención a las diferencias encontradas entre los distintos tipos de vídeos o las estrategias simplificadoras.

2. Vídeos en museos: la nueva revolución digital

Para comprender mejor el tipo de texto que analizaremos en este estudio, es necesario entender su uso actual y los retos que estos plantean para su traducción intralingüística en la modalidad del subtitulado para personas sordas.

En una investigación previa, Vicente Bru García y Silvia Martínez-Martínez (2024) describieron cómo la realidad de los museos se había transformado radicalmente en las últimas décadas, evolucionando hacia un lugar donde lo esencial es el trasvase de conocimiento y la interacción con la sociedad actual. En un mundo digital en el que las pantallas se convierten en artículos de necesidad para cualquier actividad humana, desde la medicina hasta la administración, se ha redefinido el concepto de alfabetización, conjugándose con la realidad de las pantallas en la que nos encontramos (Rowsell 2013). Los museos también han adoptado estas herramientas para hacer más atractivas y relevantes sus exposiciones, apoyando los textos con un soporte visual.

Irene Camps-Ortueta, Luis Deltell-Escolar y María Francisca Blasco-López (2021) identifican numerosos tipos de vídeos con diferentes características y diversos usos: vídeos pregrabados, vídeos interactivos o vídeos en directo. Todos ellos aportan interactividad al museo, situando el foco del aprendizaje en el usuario.

Otra de las grandes ventajas que relatan los usuarios del uso de vídeos en los museos es la sensación de autenticidad y conexión con el mundo exterior (Hakulinen et al. 2018). Una visita guiada por el director de un museo o un vídeo explicativo en museos de ciencias ayudan a que se fije el conocimiento y que se atraiga a todo tipo de público. Además, el uso de música y efectos sonoros apoyan esta visión de inmersión que buscan las instituciones para impactar u ofrecer un enfoque didáctico de todos aquellos conocimientos que ofrece cada museo (Carrozzino y Bergamasco 2010).

Centrándonos en los vídeos en sí, si bien es cierto que los museos apuestan cada vez más por una presencia en redes sociales, lo cual ha suscitado un gran interés en la literatura desde la óptica del marketing (Marín-Cepeda 2021), decidimos centrarnos en dos tipos que suponen una tipología muy difundida en museos y que cuentan con ejemplos con SpS: los vídeos pregrabados y los vídeos en directo.

Los vídeos pregrabados son todos aquellos productos audiovisuales que se producen en un entorno controlado, con un guion establecido previamente y que se muestran tanto en las pantallas de los museos como en redes sociales o páginas web institucionales. Estos productos audiovisuales pueden variar en género y en funcionalidad: un comentario de un curador sobre un artefacto, un vídeo interactivo que explique un proceso o incluso un testimonio sobre un evento histórico (Reading 2003).

La principal característica de estos vídeos es precisamente el hecho de que se han filmado con una visión creativa, es decir, utilizando un lenguaje audiovisual y verbal sumamente cuidado para que el público pueda comprender los conocimientos que se quieren transmitir de una manera ordenada, original y atractiva. No es de extrañar que en estos vídeos encontremos componentes musicales y efectos sonoros, al igual que los podríamos encontrar en la banda sonora de cualquier producto fílmico. La música y los efectos sonoros añaden a la inmersión que se persigue con el uso de estos materiales audiovisuales, una de las ventajas antes descritas de estas herramientas museísticas.

En lo referido al contenido verbal, estos vídeos son siempre guionizados, es decir, presentan un nivel de oralidad prefabricada similar al que podemos encontrar en un documental (Ogea Pozo 2012). De esta forma, podemos encontrar un léxico mucho más especializado, explicado de forma más instructiva y ordenada, y unas estructuras sintácticas más elaboradas o cultas para transmitir la sensación de rigor. El destinatario de estos vídeos son aquellos visitantes que quieran profundizar en un artefacto o sección de la exposición debido a su interés o curiosidad; que posea un nivel de especialización medio para poder comprender su contenido, normalmente creado por expertos en la materia; o que tengan el tiempo suficiente para detenerse y ver el vídeo en su totalidad.

Finalmente, a diferencia de los vídeos creados en directo, los vídeos pregrabados suelen ser menos extensos. Los visitantes del museo no cuentan con mucho tiempo para consumir estos productos audiovisuales, que suelen ser complementarios a la visita a los propios objetos u obras pictóricas del museo. Es por esto por lo que la concepción de estos vídeos desde su planificación es de una naturaleza corta para que puedan ser consumidos en cualquier momento de la visita y que, aunque requieran de un esfuerzo cognitivo, sean atractivos y entretenidos.

En lo referente a nuestro enfoque sobre la accesibilidad de estos productos audiovisuales, los vídeos pregrabados resultan esenciales para la comprensión de ciertas exposiciones, por lo que las personas con discapacidad auditiva no podrán acceder a ellos si no se realiza una subtitulación de estos. Por otra parte, como veremos más adelante, el hecho de que el texto sea semiespecializado y la velocidad de locución sea más rápida de lo normal hace que las personas sordas puedan tener problemas a la hora de leer y procesar cognitivamente el texto, incluso si cuentan con la transcripción o con unos SpS que no estén adaptados.

Los vídeos creados en directo, por su parte, son vídeos elaborados a partir de un encuentro entre un experto en la materia con público o con un público simulado, improvisado o casi improvisado, y cuyo atractivo principal radica en su espontaneidad, su naturalidad y su divulgación. A diferencia de los vídeos pregrabados, estos productos audiovisuales se crean sin guion establecido previamente: el conferenciante puede desviarse del tema principal contando anécdotas; encontraremos un uso abundante de marcadores de la oralidad como muletillas o conectores; y es posible que incluso se detecten errores de contenido en el discurso, lo que hace que se falsee la información.

Los tipos de productos audiovisuales museísticos producidos en directo suelen ser dos: la entrevista y la visita con un curador de la institución. Como indica Sonia González Cruz (2016), estos productos están caracterizados por una oralidad espontánea y un componente cercano, lo que hace que la duración sea mayor, dada la escasa planificación o la ausencia de hilo conductor. Su atractivo recae precisamente en la autenticidad que transmiten, la cercanía y la interactividad que emanan. Es por ello también por lo que encontramos este tipo de material audiovisual en los ‘museos virtuales’ (Hereniko 2019), es decir, en las páginas web, redes sociales o cualquier soporte virtual en el que se puedan difundir o acceder a los artefactos de los museos.

En lo referente al componente verbal, si bien encontramos términos especializados como en cualquier conferencia o producto relacionado con un área de conocimiento especializada, los marcadores discursivos orales son los que caracterizan estos textos (Vázquez y Fernández Bernárdez 1995). Las oraciones sin terminar y la abundancia de conectores y de elementos propios de la función fática también configuran estos textos.

A la hora de realizar el trabajo de SpS en estos vídeos, es necesario realizar una documentación exhaustiva para detectar aquellos errores cometidos por el conferenciante debido al carácter espontáneo de la locución (González Cruz 2016), eliminar todos aquellos datos superfluos y tratar de elaborar un discurso más cohesionado y ordenado para que la información pueda comprenderse de una forma más inmediata.

Finalizamos este apartado reflexionando acerca de las diferencias existentes entre las dos tipologías y las estrategias de traducción que suponemos que se utilizarán más dependiendo de esta naturaleza ya mencionada: el subtitulado de los vídeos pregrabados deberá adaptar toda aquella terminología que pueda resultar compleja o extraña para los usuarios con discapacidad auditiva, mientras que los principales retos de los vídeos en directo serán la reformulación hacia un discurso más estructurado y con una expresión más comprensible o pulida.

3. La comunidad sorda: estrategias que permitan su acceso al conocimiento especializado

En este apartado, analizaremos a los sujetos de este estudio, las personas con discapacidad auditiva, desde una perspectiva física, sociológica y cultural. Para ello, comenzaremos describiendo los retos que plantea la definición de esta comunidad tan heterogénea; observaremos las herramientas que se han establecido para su acceso al conocimiento, comparando y ofreciendo las ventajas de cada una de ellas; y por último, comentaremos los desafíos a los que se enfrentan los subtituladores y las subtituladoras a la hora de permitir el acceso al conocimiento especializado por parte de una parte de estos usuarios, así como las estrategias existentes en la modalidad de LF que pueden aplicarse en el SpS para hacer más comprensible cualquier texto especializado o semiespecializado.

Comenzaremos señalando que la comunidad a la que va dirigida los productos que se analizan en este estudio es muy heterogénea. La comunidad sorda, de esta forma, comprende desde usuarios prelocutivos, que perdieron la audición desde muy pequeños o que nacieron sin este sentido; a usuarios postlocutivos, es decir, personas con algún resto auditivo, con implante coclear o incluso usuarios oyentes pero que son familiares de personas sordas (Neumann y Stephens 2011). Esta comunidad posee diversas culturas, adaptadas al espacio y tiempo donde se desarrollan (Leigh et al. 2022) y es muy activa políticamente con el objetivo de poner de relieve sus derechos.

Dada la heterogeneidad de esta comunidad ya descrita, es necesario precisar que cualquier afirmación que hagamos en este artículo será una generalización que puede no aplicarse en algunos casos, especialmente en lo referido al nivel de lectoescritura, acceso al contenido auditivo o al lenguaje especializado. De esta forma, si bien analizaremos a continuación las diferentes herramientas que utiliza la comunidad sorda para acceder al conocimiento, centraremos las conclusiones y las observaciones en los usuarios prelocutivos con la lengua de signos española (LSE) como lengua materna, tal y como se ha hecho en la literatura en español tradicionalmente (Martínez-Martínez 2015; Ferrández Melero 2013).

Observemos ahora las dos técnicas traductológicas más extendidas a la hora de ofrecer una traducción accesible del contenido de la banda sonora de un producto audiovisual.

Quizás el método más difundido para el acceso al conocimiento audiovisual entre las personas sordas sea la interpretación en LSE, especialmente en aquellas comunidades en las que los usuarios tengan como lengua materna una lengua de signos. Esta funciona como una traducción interlingüística que permite a los usuarios acceder a la información en su lengua materna. Sin embargo, tal y como señalan Bru García y Martínez-Martínez (2024), dada la amplia variedad de lenguas de signos en todo el mundo, incluso en lugares donde la lengua oral es la misma, esta solución solo es adecuada a un grupo de usuarios de la misma comunidad espacial en la que se encuentra el museo. Una persona sorda de Colombia, por ejemplo, no podría acceder completamente a la información de un producto audiovisual del Museo del Prado, en Madrid, dado que estaría signado en lengua de signos española.

La otra herramienta, por tanto, que puede considerarse con un componente más internacional y que goza de popularidad entre el colectivo sordo es el subtitulado para personas sordas. Se trata de una técnica que consiste en la traducción intersemiótica e intralingüística de la banda sonora fílmica (diálogos, música y efectos sonoros), así como de aquellos textos en pantalla que se ofrecen en otro idioma (traducción interlingüística) a través del uso de un soporte textual colocado normalmente en la parte inferior de la pantalla (Bru García y Martínez-Martínez 2024).

Esta herramienta, además de ser más accesible para un número mayor de usuarios con discapacidad auditiva, está regulada por normas y estándares como la Norma UNE 153010 (2012) en España o las guías de estilo de subtitulado como la de Netflix para el español de España (2023), así como leyes, como el artículo 8 de la Ley General de la Comunicación Audiovisual de España (2022), que establece que el 90 por ciento de las emisiones en los canales de servicio público estén subtituladas,  que garantizan su uso en plataformas de vídeo bajo demanda (Zárate Campos 2021).

Asimismo, es importante señalar que la defensa de la accesibilidad para los usuarios antes descritos, así como para cualquier persona que padezca diversidad funcional sensorial o intelectual, no es algo novedoso o que se circunscriba al ámbito académico. Es evidente el creciente esfuerzo de las instituciones patrimoniales por brindar igualdad de oportunidades a las personas con discapacidad, no solo en contextos administrativos o médicos, sino también en entornos culturales y de entretenimiento, tal y como reivindica la Convención Internacional sobre los Derechos de las Personas con Discapacidad de la Organización Nacional de las Naciones Unidas de 2006.

En el ámbito de nuestra investigación, los museos, se debe señalar que se han elaborado numerosos manuales y directrices que abogan por la mejora de la calidad del acceso al conocimiento por parte de los usuarios con discapacidad, como el Manual de accesibilidad para museos de Ibermuseos (2019) o Hacia una cultura inclusiva. Museos para todas y todos (2020).

En el caso de las personas con discapacidad auditiva, además de las guías anteriores, debemos citar la Guía de accesibilidad para personas sordas en las industrias culturales, elaborada por la Confederación Estatal de Personas Sordas (CNSE) en 2013. Algunos de los recursos que esta recomienda para mejorar la accesibilidad, además de la instalación de bucles magnéticos para personas con implante coclear, son la habilitación en todos los recursos del museo de las dos herramientas ya descritas: la interpretación a la lengua de signos y las signoguías; y el subtitulado para sordos (Guía de accesibilidad para personas sordas en las industrias culturales, 2013: 13-22, 25).

Sin embargo, es importante destacar que no existe ninguna directriz ni estrategia propia para el SpS en productos museísticos, más allá de las ya indicadas para productos audiovisuales fílmicos. En estos casos, además de los problemas asociados a la accesibilidad fílmica (Martínez-Martínez 2015), encontramos el problema principal del acceso al conocimiento especializado en usuarios sordos con problemas de lectoescritura tales como dificultad para establecer la relación entre significante y significado (Matamala Ripoll y Orero Clavero 2010); comprensión limitada de abstracciones, metáforas y juegos de palabras (Asensi Borrás 2003; García Muñoz 2012), dificultad de acceso léxico y vocabulario reducido (Asensi Borrás 2003) o bien comprensión limitada de las palabras función (Asensi Borrás 2003).

No obstante, un aspecto común de la comunidad sorda es que existe una preferencia por mantener de forma literal en el SpS el contenido verbal del producto multimedia siempre que sea posible. Sin embargo, si bien estos usuarios alegan comprender perfectamente estos subtítulos sin editar (o verbatim), existen estudios que demuestran que la realidad es diferente (Perego 2020) y que la simplificación del contenido verbal puede serles de gran utilidad y facilitar notablemente el acceso a la información (Bredel y Maaß 2016). La situación ideal sería dar respuestas a todas estas necesidades a partir de la creación de diferentes tipos de SpS de acuerdo con las características de los usuarios, es decir, tal como apuntaba Cristina Cambra (2006: 28), según la edad, el nivel educativo, etc. Desafortunadamente, esto no es más que una utopía, ya que llevar este propósito a cabo supondría una inversión económica cuya consecución es difícil para museos y productoras audiovisuales.

Teniendo en cuenta la necesidad de realizar una adaptación para mejorar el acceso al conocimiento por parte de las personas sordas, planteamos en este estudio la posibilidad de utilizar estrategias provenientes de otra modalidad traductológica accesible, la Lectura Fácil, para mejorar el acceso a la información por parte de los usuarios con discapacidad auditiva. Como indica Silvia Toribio Camuñas (2021), esta modalidad, destinada a facilitar la comprensión de los textos por parte de usuarios con diversidad funcional cognitiva, no cuenta aún con unos fundamentos teóricos que puedan responder desde un plano estructurado a sus retos lingüísticos, si bien existen algunas directrices (García Muñoz 2012) o normas como la UNE 153101: EX (2018) que pretenden ofrecer algunos estándares o patrones para mejorar la presentación del texto y, por consiguiente, del acceso a la información.

A pesar de la novedad de este enfoque, cabe señalar que existen investigaciones previas que han intentado aunar las dos modalidades o, al menos, utilizar estrategias simplificadoras en el SpS debido a las razones antes señaladas. En Alemania, este planteamiento ya se puso en práctica de la mano de Laura Marmit (2021). La autora ya señalaba que sería apropiado utilizar el Leichte Sprache (equivalente alemán a Lectura Fácil) para la subtitulación para personas sordas. Esto no sería una utopía, ya que tradicionalmente en el subtitulado para normoyentes ya se utilizan de por sí técnicas de adaptación para la reducción como la simplificación de estructuras complejas (Marmit 2021: 21). Por su parte, Cintia Espinosa Martínez (2022) realizó un estudio piloto basado en vídeos del ámbito biosanitario con SpS simplificado y que se realizó con estudiantes de la asignatura Traducción para Entornos Multimedia (Alemán) del grado en Traducción e Interpretación de la Universidad de Granada.

Laura Marmit (2021) indica que las principales estrategias de LF que pueden ser de utilidad para la creación de SpS son la simplificación de estructuras sintácticas complejas y el uso de un estilo verbal frente al nominal. Esto es especialmente útil en los videos grabados en directo, donde las estructuras sintácticas son algo más complejas debido a la falta de preparación, así como aquellos productos pregrabados que utilicen un vocabulario más elevado o unas estructuras más complejas para transmitir la sensación de rigor.

Sea como fuere, resulta interesante que estas estrategias ya se vieran plasmadas, aunque no vinculándolas directamente con la LF o con el lenguaje simplificado, en la tesis de Martínez-Martínez (2015), a partir del estudio de corpus realizado sobre el SpS de diferentes productos fílmicos. La concreción de estas estrategias se ofrece en el siguiente apartado, donde se observará el sistema que hemos utilizado para evaluar la utilización o no de estas.

4. Metodología

Una vez establecido el marco teórico en el que se encuentra nuestra investigación, es necesario describir los pasos que hemos realizado para la consecución de los objetivos expuestos al comienzo de este trabajo.

Dado que uno de nuestros objetivos es la pertinencia de las estrategias simplificadoras en el SpS, hemos decidido evaluar la aparición (o no) y la frecuencia de uso de estas estrategias en productos audiovisuales museísticos ya confeccionados. Para ello, hemos creado un corpus a partir de todos los vídeos alojados en la plataforma de accesibilidad museística AL-MUSACTRA.

Esta plataforma, fruto del proyecto i+D Acceso universal a museos andaluces a través de la traducción (AL-MUSACTRA) (B-TIC-352-UGR18), ofrece vídeos pregrabados y visitas con curadores, tanto con SpS como con interpretación a la lengua de signos española. El SpS fue realizado por parte de expertos del grupo TRACCE, coordinador de este proyecto i+D, así como alumnos y alumnas y subtituladores y subtituladoras en proceso de formación de la Universidad de Granada, que subtitularon estos vídeos siguiendo su propia intuición y la norma UNE 153010: 2012 para SpS, de vigencia en España.

Se trata de un total de 22 vídeos pregrabados de cinco minutos cada uno, pertenecientes al espacio BioDomo del Parque de las Ciencias (Granada) y 31 vídeos de quince minutos cada uno de una visita realizada por el director del Museo Arqueológico y Etnológico de Granada. Se debe observar que la velocidad del discurso en todos los vídeos es normal (17 caracteres por segundo aproximadamente en los pregrabados y 20 caracteres por segundo en el caso de los grabados en directo), por lo que la simplificación no parece responder a la necesidad de reducir en demasía el contenido para que pueda ser leído por parte del usuario sordo. No obstante, se observará este aspecto en los resultados y conclusiones.

Dado que contábamos con los archivos de subtítulos, decidimos colocarlos en una tabla comparativa de tres columnas: una para el texto origen (la transcripción del audio), otra para el subtitulado para sordos y la última para señalar las estrategias utilizadas.

En lo referente a la herramienta investigadora, si bien existen otros estudios sobre la traducción intralingüística (Eugeni y Gambier 2023) nos hemos basado en la clasificación de estrategias traductológicas creadas por Martínez-Martínez (2015: 231-243). El objetivo principal del sistema de este etiquetado consistía en identificar, analizar y clasificar los elementos acústicos (verbales y no verbales) del texto origen que el traductor había considerado relevantes para el significado multimodal del producto audiovisual y relacionarlos con las técnicas de traducción interlingüística, intralingüística o intersemiótica empleadas para su traslado al subtítulo (Martínez-Martínez y Álvarez de Morales Mercado 2022: 184). Si bien existe un primer nivel en el que se describe la procedencia del sonido, nos decidimos por utilizar únicamente el segundo nivel, que es el que se centra en las técnicas de traducción.

Hemos seleccionado las estrategias intralingüísticas, dado que son las que se corresponden con el tratamiento del componente verbal, y las estrategias intersemióticas, dado el componente inmersivo propio de los vídeos pregrabados, conseguidos a partir de efectos sonoros y música. A continuación, procederemos a explicar en qué consisten cada una de estas técnicas y las ilustraremos con ejemplos extraídos de nuestro corpus de trabajo.

En lo referente a las estrategias intersemióticas, señalamos las tres estrategias más utilizadas según el estudio de Martínez-Martínez (2015): categorización, o asignación de una categoría conceptual de un sonido (música, sonido, lluvia), el agente que lo produce (pájaros, campana, puerta), la acción que lo produce (ríe, grita, toca el piano) o el resultado de dicha acción (risa, grito); atribución o asignación de una cualidad o atributo al sonido que puede ser explícita (música melodiosa) o implícita ([habla] nervioso); y explicación, o explicitación de la información que se proporciona en el subtítulo, como duración (continúa la música) o detalles (suena la 5ª sinfonía de Beethoven).

En cuanto a la traducción intralingüística, señalamos dos grandes bloques: el referente al verbatim y los referentes a la adaptación lingüística. La traducción literal, única estrategia del primer bloque, es la plasmación íntegra de los diálogos de los personajes. Tal como veremos en el siguiente apartado, referente al análisis de los resultados, la frecuencia de uso de esta estrategia suele ser baja debido al carácter especializado del contenido y el orden incoherente de las oraciones de los vídeos grabados en directo.

El segundo bloque lo conforman estrategias reductoras como la simplificación y la reducción. La primera de ellas consiste en el uso de un léxico o sintaxis más sencillos en el subtítulo que los utilizados en el audio original. Esta técnica se divide en dos: simplificación léxica (‘estatua’ en vez de ‘estatuaria’ o ‘proporcionar alimento’ por ‘alimentar’) y simplificación sintáctica (‘Esperamos que vuestra visita sea una aventura y que os despierte el interés por estas zonas’ por ‘Esperamos que vuestra visita sea una aventura y que os interesen estas zonas’).

La reducción, por su parte, consiste en un reajuste del contenido del audio original en el subtítulo. Esta puede consistir en un resumen o condensación (‘lo cual prueba que el foro y sus templos eran muy importantes.’ por ‘lo cual te dice que el foro de Almuñécar, los templos de Almuñécar tenían que ser muy potentes’) o en la eliminación de parte de este (‘En este acuario de agua dulce,’ por ‘A continuación, en este acuario de agua dulce,’).

Para finalizar este bloque, hemos añadido la estrategia de expansión, debido a que nos hemos encontrado un alto número de casos en nuestro corpus. Esta se utiliza para completar las carencias lingüísticas y culturales que puedan tener los usuarios sordos (Pazó Lorenzo 2011: 210) y favorecer de este modo su comprensión. En los vídeos con los que hemos trabajado, hemos observado cómo se han añadido elementos aclaratorios para que se pudiera entender la totalidad del contenido del mensaje. Es el caso de ‘su tamaño supera la escala natural y está esculpida en mármol de calidad’ para el parco comentario del director ‘es más que a escala natural en un buen mármol’.

En el siguiente apartado, presentaremos los resultados de la investigación en gráficos elaborados con el propósito de hacer los datos más accesibles, así como procederemos a realizar las interpretaciones de estos.

5. Resultados y discusión

Con el objetivo de presentar los datos extraídos de forma ordenada y consecutiva, observaremos de forma comparativa los resultados de las estrategias de traducción de los vídeos pregrabados pertenecientes al museo de ciencias BioDomo y las utilizadas en los vídeos creados en directo en el Museo Arqueológico y Etnológico de Granada, así como ejemplos que responden a los retos antes descritos sobre los productos audiovisuales museísticos.

Para comenzar, mostraremos los gráficos que resumen las estrategias de traducción intralingüística que se han utilizado en estos vídeos:

Fig. 1: Frecuencia de uso de las estrategias de traducción intralingüística del BioDomo

Fig. 2: Frecuencia de uso de las estrategias de traducción intralingüística del MAEG

Como podemos observar, la variedad de estrategias utilizadas es evidente. Este hecho confirma que estas difieren atendiendo al tipo de vídeo, tal como se ha explicado en el apartado 2.

En el caso de los vídeos pregrabados, destaca significativamente el empleo de la simplificación léxica con un 40 por ciento, mientras que en los vídeos grabados en directo desciende a un 17 por ciento. Este dato no es de extrañar, puesto que el nivel de especialización de los textos audiovisuales del BioDomo es mayor que la divulgación que se persigue en la visita guiada del director del Museo Arqueológico y Etnológico.

En cuanto a la técnica más utilizada en los vídeos del MAEG, la más destacada es la condensación, con un 29 por ciento frente al 21 por ciento de los del BioDomo. Como se ha mencionado anteriormente, los vídeos grabados en directo son vídeos improvisados que suelen desviarse en su discurso del tema principal contando anécdotas o explicando de forma coloquial aquellos elementos que no se comprenden fácilmente. Es por ello por lo que parece evidente que haya que condensar para que la información pueda plasmarse en el subtítulo de forma clara y sin circunloquios, además de respetar las estrictas limitaciones temporales y espaciales (Díaz-Cintas 2020: 152).

Tampoco debe resultar extraño que la segunda técnica más utilizada en los vídeos grabados en directo sea la de eliminación con un 23 por ciento, utilizada principalmente para omitir muletillas y anécdotas fruto del carácter improvisado de este tipo de vídeos. Sin embargo, en los vídeos pregrabados, esta técnica también resulta relevante, ya que supone un 21 por ciento de todas las estrategias de traducción utilizadas. No obstante, el motivo que subyace tras este uso es diferente: el subtitulador o la subtituladora ha utilizado esta técnica para eliminar información superflua, estética desde el punto de vista funcional, cuya eliminación no afecta en la comprensibilidad del mensaje.

En lo que respecta a la técnica que hemos añadido al sistema de etiquetado para este estudio, expansión, su uso arroja también unos datos significativos en ambos subcorpus. Destaca su frecuencia en los vídeos grabados en directo con un 18 por ciento mientras que, en los vídeos pregrabados, esta cifra asciende tan solo al 9 por ciento. En ambos casos se utiliza para proporcionar una referencia más clara y específica del lugar o elemento al que se hace mención.

Para finalizar, nos encontramos con dos técnicas que presentan una baja frecuencia de uso. La primera de ellas es la traducción literal, que en el BioDomo es de un 5 por ciento y en el MAEG de un 12 por ciento. Es lógico que esta estrategia sea poco usada, ya que como sabemos, el subtitulado es una modalidad de traducción que presenta una serie de restricciones espacio-temporales, lo que significa que la traducción literal no siempre es posible. No obstante, como se ha observado, la velocidad del discurso, así como la velocidad de lectura de los subtítulos permitía la traducción literal, algo que no se ha visto realizado al optar por una simplificación que responde a la tradición de utilizar estrategias simplificadoras en el SpS.

Por último, nos encontramos con la simplificación sintáctica, estrategia que, en un principio, pudiera parecer muy utilizada, pero que la realidad es diferente. La razón principal recae en que otra estrategia, la condensación, permite reducir de forma más eficiente. Es por ello por lo que observamos en los gráficos que los datos de esta estrategia son un 3 por ciento en el BioDomo y apenas un 1 por ciento en el MAEG.

En el siguiente gráfico (Fig. 3), podemos observar un resumen comparativo de las técnicas de traducción intralingüística en ambos subcorpus.

Fig. 3: Comparación de las estrategias de traducción intralingüística utilizadas

Dado que los resultados de la simplificación léxica han resultado muy llamativos, se ha planteado una explicitación de las categorías gramaticales que se han simplificado más y hemos realizado estos gráficos en función de la frecuencia y recurrencia de sus casos, tal como podemos ver en las siguientes figuras.

Fig. 4: Explicitación de la frecuencia de uso de la estrategia simplificación léxica en el BioDomo

Fig. 5: Explicitación de la frecuencia de uso de la estrategia simplificación léxica en el MAEG

Como se puede observar, también existen discrepancias entre los diferentes tipos de productos audiovisuales. En los videos pregrabados del BioDomo, encontramos un mayor número de simplificaciones verbales (‘alimentar’ por ‘proporcionar alimento’) y de adjetivos (‘debajo del agua’ por ‘subacuático’), debido al registro elevado que se utiliza para aumentar el rigor y la ostentación propia de estos productos pregrabados.

En lo que respecta a los vídeos en directo del MAEG, se han simplificado aún más otros elementos lingüísticos como preposiciones (‘por’ en lugar de ‘durante’, más largo), deícticos (‘aquí’ por ‘por allí’, más claro) y de verbos y sustantivos, también por el hecho de usar palabras más en desuso (‘reflejo’ por ‘trasunto’) o menos frecuentes (‘está hecho’ por ‘fabricó’).

El tratamiento de los deícticos resulta llamativo, especialmente en el caso de frases en las que no resulta claro a qué parte del museo se refiere. Es el caso del ejemplo extraído de los vídeos del BioDomo, donde se sustituye ‘enfrente’ por ‘enfrente de las medusas’, más claro para la persona que se encuentra realizando esta visita.

Como resumen, mostramos un gráfico en el que se puede ver de forma más clara las diferencias sobre esta técnica entre ambos subcorpus.

Fig. 6: Comparación de la recurrencia de casos de la estrategia simplificación léxica

Finalmente, señalaremos las estrategias intersemióticas explicadas en el apartado anterior que se han utilizado en los vídeos. No hemos encontrado ejemplos del uso de estas estrategias en el subcorpus de los vídeos grabados en directo, posiblemente porque estos se centran exclusivamente en la información verbal y no se ven complementados por ninguna información musical o sonora preestablecida, como es el caso de los vídeos pregrabados.

Fig. 7: Resultados de la frecuencia de uso de la traducción intersemiótica en el BioDomo

En el caso de los vídeos pregrabados, se han utilizado técnicas de traducción de la música y los efectos sonoros que aparecen al principio y al final de las explicaciones, enmarcando el lugar que se va a describir y permitiendo al visitante que se inmerse en la naturaleza propia de este espacio. Se ha utilizado una atribución (‘música tranquila’) en la mayoría de los casos, para evocar las sensaciones que se crean a partir de estos sonidos, en vez de identificar la fuente del sonido (’sonido de agua’).

6. Conclusiones

En este artículo de investigación se han presentado los resultados de un análisis de corpus de vídeos pregrabados y vídeos grabados en directo de museos en español. Este estudio ha contribuido a mejorar la comprensión de las estrategias de traducción intralingüística utilizadas en el SpS en museos y ha señalado la necesidad de seguir investigando en este campo para mejorar la accesibilidad de los productos multimedia para personas con discapacidad auditiva.

Hemos analizado las diferencias existentes entre dos tipos de productos audiovisuales museísticos, especialmente en lo referente a los retos que se plantean para los subtituladores que realicen SpS. Además de valorar las ventajas que aportan cada uno, como la inmersión, la espontaneidad o el rigor, hemos reflexionado sobre el papel que desempeñan estos productos audiovisuales en la nueva realidad del museo virtual e interactivo, cada día más presente en nuestra sociedad.

El estudio de corpus, por su parte, ha mostrado que se sigue defendiendo la idea de que, para lograr una efectiva accesibilidad en el subtitulado para personas sordas, es necesario considerar que los usuarios generalmente tienen una competencia limitada en su capacidad lectora, lo cual se pone en práctica a partir de la adaptación del contenido lingüístico. A diferencia de lo que ocurre en las películas y las series en streaming en los que ‘cualquier técnica de adaptación está perdiendo relevancia’ (Martínez-Martínez 2022) vemos que esto no ocurre con los vídeos multimedia en museos, debido a la idea de que estos textos semiespecializados deben ser adaptados para que sean fácilmente comprensibles por parte de la comunidad sorda. No obstante, consideramos necesario realizar un estudio de recepción que demuestre la validación de esta afirmación, así como su grado de realidad en cada uno de los diferentes tipos de discapacidad auditiva.

En lo referente a las estrategias en sí, aunque la frecuencia de uso de estas varía, la simplificación léxica es la técnica más utilizada en los vídeos pregrabados, mientras que la condensación es la más empleada en los vídeos grabados en directo. El resto de las estrategias aparecen con una frecuencia menor, aunque también son relevantes. Este hecho no hace más que poner de manifiesto las claras diferencias que existen entre las dos tipologías de vídeos que hemos analizado y que confirman nuestras premisas. Además, creemos que en un futuro también sería interesante realizar estudios similares en otras lenguas para observar si estos resultados son comunes en diferentes idiomas o si el SpS de productos multimedia en museos refleja las idiosincrasias propias de la percepción de la comunidad sorda de cada país. También destacamos la relevancia de futuros estudios sobre parámetros lingüísticos comunes de lenguaje simplificado en cada una de las lenguas de estudio.

Por último, destacamos la complejidad y la diversidad de destrezas que un subtitulador o una subtituladora que realice SpS debe poseer para realizar un subtitulado para sordos en museos, lo que sugiere la necesidad de crear un perfil específico para esta figura en el futuro.

Agradecimientos

Esta investigación se ha llevado a cabo dentro del Proyecto de Investigación TALENTO [Traducción y lenguaje simplificado del patrimonio para todos. Herramienta de análisis y consulta (PID2020-118775RB-C21)], financiado por el Ministerio Español de Ciencia e Innovación (MICINN).

Abstract

Interlingual Live Subtitling (ILS) finds its foundations in subtitling for the d/Deaf and Hard of Hearing (SDH) and Simultaneous Interpreting (SI) and it is situated at the crossroads between Audiovisual Translation (AVT) and SI (Romero-Fresco & Alonso 2022), providing access to the audio content of live events and TV programmes for audiences with and without hearing impairments and disabilities (Romero-Fresco 2018). It is currently being achieved through different approaches requiring human-mediated translation and automatic language processing systems to different extents and among them, respeaking is triggering growing interest. This paper begins with an introduction to interlingual respeaking, a technique in which a transcript under the form of subtitles is provided thanks to a respeaker working with a speech recognition, and which derives from SI itself according to Carlo Eugeni and Francesca Marchionne (2014). The article shares a training proposal in interlingual respeaking in Spanish to Italian and the results of an experiment (Pagano, 2022a) are then presented, comparing interlingual respeaking to other four different workflows of ILS, with different degrees of human-machine interaction: SI and intralingual respeaking, SI and Automatic Speech Recognition (ASR), intralingual respeaking and Machine Translation (MT), and ASR and MT. Final results on quality of the Live Subtitling (LS) of the experiment show their linguistic accuracy using a model by Pablo Romero-Fresco and Franz Pöchhacker (2017), their delay and broadcasting. Each of the five methods is then described pointing out its strengths and weaknesses to shed some light on which one can be more suited to provide high quality ILS for live events.

Keywords: media accessibility, interlingual respeaking, simultaneous interpreting, automatic speech recognition, Machine Translation

1. Introduction

This research deals with Media Accessibility (MA) for accessible multilingual communication, covering accessibility through respeaking a diamesic translation (Gottlieb, 2007) that transfers an audio input into written output, and from one language into another (interlingual). Interlingual respeaking accounts not only for sensory, such as subtitling for SDH, but also for linguistic barriers where the cross-cultural factors need to be addressed in a similar manner to SI. Integrating Pablo Romero-Fresco’s definition of respeaking, Hayley Dawson (2020) defines it as:

a technique in which one listens to the original sound of a (live) programme or event in one language and respeaks (interprets) it in another language, including punctuation marks and some specific features for an audience who cannot access the sound in its original form, to a speech recognition software, which turns the recognised utterances into text displayed on screen with the shortest possible delay (Romero-Fresco 2011: 1).

Respeaking is a technology-enabled hybrid modality of translation (Davitti & Sandrelli, 2020), that shares a common ground with SI in terms of skills and competences required (Russello, 2013), and with subtitling as well as it was initially carried out only by simultaneous interpreters and subtitlers (Eugeni & Mack, 2006; Szarkowska et al., 2018; Romero-Fresco & Eugeni, 2020).

The second section of this paper (§2) is dedicated to briefly introduce Human-Machine Interaction (HMI), while §3 covers assessment in ILS presenting the NTR model for linguistic accuracy. The following §4 aims at outlining the Spanish to Italian course proposal in interlingual respeaking carried out to train participants for the experiment, while §5 define the research method, with emphasis on participants, tools, and materials. Finally, results of the pilot study are presented (§6), followed by some concluding remarks (§7).

2. Human-Machine Interaction in Interlingual Live Subtitling

This pilot study focuses on ILS produced through different methods that require different degrees of HMI. In 1992, Hewett et al. defined this interdependence with technology as Human-Computer Interaction (HCI), or the implementation of interactive computing systems for human use. As explained by Eugeni (2019), in many fields today we see a rise in technological intervention rather than human, which is gradually being replaced by automated processes. In some fields where ASR and Natural Language Processing (NLP) systems are used, for example, “technological evolution is reducing the place of humans in this interaction to such an extent, that their profession could hardly be possible without it” (Eugeni 2019: 873). Three different categories of HCI can be distinguished in LS (Eugeni 2019), that can be applied to both intra and interlingual subtitling, given that this structure is technology-oriented (Pagano 2020a):

• Computer-Aided ILS, which implies a SI by an interpreter and a live subtitler to transcribe what is being said, namely a human carries out the job and a machine assist them;
• Human-Aided ILS, which is the reverse of Computer-Aided ILS, in which a machine carries out the job and humans assist it, editing the transcript;
• Fully-Automated ILS, without any human assistance. For intralingual LS an ASR system produces the transcript. For ILS, instead, this is the case of an ASR system recognizing the audio input in one language, and connected to an MT software producing the transcription in a second language.

3. Assessment in ILS: linguistic accuracy

Different models have been designed and tested over the years to assess accuracy in ILS (Roberts et al. 2017), varying depending on the method by which subtitles are produced, such as the conceptual measurement IRA (Idea Rendition Assessment) model by Eugeni (2017), based on the main distinction of rendered and non-rendered ideas in the Target Text (TT), and therefore more on communicative and conceptual level of analysis based on ideas, than taking into account formal errors. Other assessment models have been proposed, such as the WIRA (Weighted-Idea-Rendition-Assessment) model (Eichmeyer-Hell, forthcoming), and the NERLE model (Moores 2023).

In this research the NTR model (Number of words, Translation errors, Recognition errors) by Romero-Fresco and Pöchhacker (2017) is used to analyse the sets of subtitles created through the different methods. The model was first applied to test interlingual respeaking feasibility (Dawson 2020), and then different ILS outputs within the SMART (Shaping Multilingual Access Through Respeaking Technology) project[1] experiments (Sandrelli 2020). The model can be applied to assess subtitles produced in and from any language pair (not sign language). As shown in Figure 1 below, the calculation consists in subtracting the sum of translation errors (T) and recognition errors (R) from the total number of words in the subtitles (N), dividing by N again, and then multiplying by 100: the reference threshold rate for a subtitle in order to be deemed acceptable is 98%. In addition, the model takes into account Effective Editions (EEs), namely those reformulations that can be deemed strategic, i.e. that do not lead to a loss of information or content.

Figure 1. The NTR model with error type categorizations (Romero-Fresco & Pöchhacker 2017)

In this model, translation errors refer to the errors made by the respeakers in their performance, while recognition errors refer to misrecognitions by the Speech Recognition (SR) software. Translation errors are divided into error types based on content or form, the former accounting for omissions, additions and substitutions, the latter for style and correction errors. Error grading is threefold and is categorised using the following terminology: minor errors (0.25), major errors (0.5), critical errors (1). The error grading identified for the NTR model entails both translation and recognition processes: minor errors cause only a small loss of content and do not impact comprehension, being mainly recognition errors and relating to punctuation, spelling, and misrecognition of small words such as prepositions or articles. A major error implies depriving the reader of part of content without them noticing: major translation errors would be the omission of a full independent idea unit, while major recognition errors account to those that the reader can identify (either a misrecognition resulting in a nonsensical word, or a sensical new word that can be clearly deemed a misrecognition since it makes no sense in the context). Finally, critical errors introduce new content unrecognizable as such by the reader, constituting misleading information. While recognition errors (R) do not have subcategories, translation errors (T) can relate to both content and form. Content errors account to omissions, additions, and substitutions, while form correctness errors refer to errors of grammar in the TT and form style errors to errors caused by unnatural translation or changes in the register. Lastly, effective editions (EE) can be omissions or substitutions (reformulation, generalization, etc.) that do not lead to loss of content. An EE is spotted when the Source Text (ST) has been modified to a certain extent but without losing relevant information and can therefore be considered an improvement, for example when the respeaker omits orality or reformulates something to make it more readable.

3.1 About delay and the concept of ‘quality’

‘Quality’ is not always a straightforward concept to identify. The ‘high quality’ or ‘low quality’ judgment on almost everything is heavily driven by subjective elements such as personal opinions and preferences, for example, and it is therefore difficult to standardise with clear-cut characteristics. Similarly to translated and interpreted products, quality assessment in LS cannot be restricted to mere linguistic accuracy, but includes other factors (Pöchhacker 2013: 34) such as delay. In SI delay in the interpreting process plays an important role in assessing the overall quality of a SI output. Between accuracy and delay there is an important link, underlined by Romero-Fresco:

[T]he interplay between accuracy and delay constitutes an intrinsic part of live subtitling and is often described as a trade-off: launching the subtitles without prior correction results in smaller delays but less accuracy, while correcting the subtitles before cueing them on air increases accuracy but also delay. (Romero-Fresco 2019b: 99)

Delay in subtitling is often referred to as ‘latency’, namely “the delay between the source speech and real-time target text, [which] will also vary in relation to the output delivery and degree of editing” (Davitti & Sandrelli 2020: 105). As described by Dan McIntyre et al. 2018 (in Moores 2020):

there is an inherent delay or latency in respeaking between a word being spoken and appearing on screen in a subtitle; this results from the time needed for the spoken word to be heard, respoken, recognised and processed through the subtitling software and onto the screen.

Therefore, an overall quality assessment that takes into account not only linguistic accuracy, but also latency is the one that was sought in the analysis of this experiment results, to guarantee the most comprehensive assessment possible.

Having outlined the theoretical framework that lies behind the technique of interlingual respeaking, the following § 4 is dedicated to presenting the training modules of the course.

4. Training course proposal in interlingual respeaking

To train participants to the experiment in intra and interlingual respeaking, they followed a Spanish to Italian respeaking course offered at the University of Genoa in the academic year 2021/22 (Dawson 2019; Dawson & Romero-Fresco 2021; Pagano 2022b; forthcoming). The training lasted 70 hours and included synchronous, distance learning and individual practical exercises, and it was taught over a three-month period. Some materials were taken and adapted from the ILSA (Interlingual Live Subtitling for Access) course, a three-year Erasmus+ project co-financed by the European Union (2017-2020) whose key objective was to bridge the gap between intra and interlingual live subtitling as recognised professional practices (Robert et al. 2019b) by identifying the profile of the interlingual live subtitler and developing the first training course on ILS.

As shown in Figure 2 below, the training at the University of Genoa comprised a theoretical introduction to MA and pre-recorded subtitling, a review of SI with preparatory exercises, presented the use of the ASR systems and, ultimately, proposed intralingual and interlingual respeaking practice.

Figure 2. An overview of the English to Italian interlingual respeaking workshop.

More in detail, Module 1 showed an introduction to MA and AVT, and an overview on fundamentals of subtitling (condensation, line breaks, characters per line, subtitles duration, etc.) and of SDH. Module 2 on SI preparatory exercises consisted of counting, shadowing, paraphrasing and summarizing, and sight translation. Module 3, emphasised the dictation practice and presented the ASR software that was used during the experiment as well, Dragon Naturally Speaking, working on pronunciation, sentences segmentation, punctuation dictation and error edition. Final Modules 4 and 5 were dedicated to practice in both intra and interlingual respeaking.

5. The experiment

5.1 Participants

According to several scholars, interpreters and translators are the preferable option to train professionals in intra and interlingual respeaking, thanks to their background and skills (Russello 2010; Romero-Fresco 2012; Pagano 2022b). The participants to the experiment were, indeed, five students of the master’s degree in Translation and Interpreting of the University of Genoa, the same who were trained through the Spanish to Italian respeaking course. Four females and one male, aged between 21 and 23 years old, they all had previous experience, albeit introductory, in subtitling and SI. They were all Italian native speakers with Spanish as their B language, and with a C1 proficiency level according to the CEFR. None of them had any previous knowledge or training on LS or respeaking. Given their performances were recorded on screen, they were previously informed and agreed with an authorization form on ethical procedure[2].

5.2 Tested methods

The aim of the experiment was to test the participants’ performance in five different modes of ILS by comparing their outputs (see also Dawson 2021; Romero-Fresco & Alonso-Bacigalupe 2022). The methods and the roles assigned to each participant in the experiment are presented in Table 1.

Table 1. Participants and roles needed for each tested method.

On a scale of different extents of HMI, Method 1 is the most human-centered mode, while Method 5 is machine-centered and fully automated. The experiment was conducted on completion of training, during which each participant was trained in SI, intra and interlingual respeaking. The participant who was more fluent in SI during the training was assigned the role of the interpreter, who performed better in interlingual respeaking performed as an interlingual respeaker, and who scored worse results in interlingual respeaking was given the intralingual IT>IT respeaking role. Due to the lack of Spanish native speakers among the participants, participant D carried out the intralingual respeaking task ES>ES despite not being a native. Role’s assignment was based on assessment results of performances during the training and on formative assessment. Roles were also assigned according to the participants’ self-evaluation and self-confidence in performing one or the others. It is important to note that, while total participants to the experiment were five, only data from four of them were taken into account in the final results analysis. Two people carried out the Spanish to Spanish intralingual respeaking task, and only the one performing better was analysed.

5.3 Tools

As for tools, the ASR system used in all the methods was Dragon Naturally Speaking, version 15, the same students were trained with during the course. In Methods 4 and 5 the MT machine used was Google Translate, the neural machine translation service developed by Google, which is one of the most widely used worldwide. Similar experiments conducted in parallel with this pilot study (Romero-Fresco & Alonso-Bacigalupe 2021, 2022; Dawson 2021) also used Google Translate as MT, so it was also chosen to compare results even with different language pairs.

During the experiment, participants also recorded their screens with FlashBack Recorder, exporting all necessary files for the analysis and the delay calculation. Moreover, delay was calculated according to the Spanish Norma UNE 153010: 2012 for Methods 1 and 2, which consists of choosing one sentence-ending each minute of the source video and calculate the lag between the moment in which a specific utterance was spoken, and the moment in which it was displayed as subtitles on screen (Romero-Fresco & Alonso-Bacigalupe 2022).

5.4 Materials

Each participant was asked to interpret or respeak intra or interlingually three short videos chunks, as detailed in Table 2 below, according to their roles. One participant performed as an interlingual respeaker, one as a simultaneous interpreter, one as intralingual respeaker ES>ES, and another IT>IT. The texts were extract of original speeches in Spanish that were not altered (re-read, accelerated or adapted), featuring low levels of information density. The three text chunks were chosen on purpose with different variables – speech rate and difficulty of the topic – in order to be able to observe how the five different methods handled the diverse typologies.

Table 2. Experiment material information.

The first speech had a slow speech rate and was given to the participants to warm up. It was a speech delivered by Pope Francis on the subject of the Earth’s Day and mentioned climate change and Covid. The Pope speaks a native Argentine Spanish that is dissimilar to the Castellano the students were used to. Nevertheless, the speech is delivered clearly and with good intonation and articulation. The second chunk was by former Spanish President Mariano Rajoy, addressing the Senate. It was faster, but shorter in length and did not feature any specific terminology. The third and final video was taken from the Speech Repository by the European Commission[6] and was a video by the European Economic and Social Committee, featuring this time with a higher density of information and with institutional terminology such as dictamen (appraisal), marco jurídico (legal framework), directiva marco (framework directive), propuesta de reglamento (proposal for a regulation), autoridad (authority), gestión del espectro (spectrum management), separación functional (functional separation). Participants attempted their task – interlingual or intralingual respeaking or simultaneous interpreting, according to their roles – only once, and they were not given the opportunity to watch the proposed videos beforehand. A total of 15 outputs were provided with the experiment: 5 (number of tested methods) for each of the 3 videos.

6. Results and discussion

Table 3 below shows the results of the testing for the three videos. The final NTR percentage scores for each of the five methods are displayed individually, and then on average.

Table 3. NTR scores

Together with the percentage score, each output was attributed a grade on a 10-point scale linked to a descriptive classification of the performance (Dawson, 2020), as shown in Table 4.

Table 4. Classification of performances in reference to the NTR model.

While testing the NTR model for linguistic accuracy, subjectivity in the analysis of error gradings was crucial (Romero-Fresco & Pöchhacker 2017). One solution to minimise it is to have more than one evaluator carrying out the assessment (inter-annotators), to monitor any discrepancies between different people’s opinions. In this sense, NTR analyses for all the outputs of the experiments where first carried out by the participants in a self-evaluation process and revised in a peer-evaluation phase by the other course students. Only then, the trainer of the course and author of this contribution worked as a first evaluator; afterwards, a second inter-annotator with previous knowledge and experience of assessment with the NTR model reviewed all the analysis again. If and when discrepancy in error grading between the first evaluator and the inter-annotator was detected (rarely), they would agree on the severity to give to the error after having discussed its impact.

Pre and post-experiment questionnaires were submitted to participants. In the former, all participants expressed some anxiety for the forthcoming experiment, but they also answered they felt adequately prepared for the task. In the latter, when asked to consider how difficult they found the entire test (all three videos), two participants declared ‘neither easy, nor difficult’, and the other three ‘difficult’. They were also asked how difficult they found each text chunk: for the speech by the Pope, all responded between ‘neither easy nor difficult’ to ‘very easy’; the speech by Mariano Rajoy was rated ‘very difficult’, and for the speech “Marco jurídico de las Comunicaciones electrónicas”, one person rated it ‘easy’, two ‘neither easy nor difficult’, and two ‘difficult’, showing a divergence of opinions on the matter. For all the videos, the participants deemed neither the topics nor the terminology to be complex.

6.1 NTR analyses for linguistic accuracy

The final corpus with all the subtitles from the five methods for each of the three videos amounted to 4,200 words, with 258 errors in total. Of these, 45.7% (118) were translation errors, and 54.3% (140) were recognition errors. Out of translation errors, 57.6% were minor, 18.7% were major, and 23.7% critical. Out of recognition errors, 71.4% were minor, 15% major, and 13.6% critical.

Translation and recognition errors for each method are now briefly outlined to observe their frequency. Translation errors in Methods 1, 2, and 3 were only imputable to the interpreter or the respeaker, while in Methods 4 and 5 were all imputable to either the or the MT tool used, or they were recognition error by Dragon.

6.1.1    Translation errors

Before moving on to the detail of error types and their frequency in the following sections, Figure 3 below shows an overview of the different errors for the five methods. The trend shows content omission errors (in blue) frequently occurring as minor errors, followed by major ones. Content additions (in orange) are almost non-existent, while content substitutions (in gray) confirm they pertain almost exclusively to critical errors. Errors of form, both in style and correctness, are confined to minor severity with only one exception.

Figure 3. Overview of the most frequent translation error typologies according to severity

The participant using Method 1 said this was – understandably, as they had to perform two new tasks contrary to all other participants – the most challenging. Still, the results of the analyses from this method are promising since an average of 97.1% was achieved. Translation errors detected totalled 30 (16 minor, 7 major, and 7 critical). Interestingly, the subject using Method 2 (SI + intralingual respeaking) ranked first in linguistic accuracy, far outperforming the other methods. There was a total of 19 translation errors detected: 11 minor, 5 major, and 3. In the whole Method 3 (SI + ASR), 10 translation errors were counted among the three severities: 6 minor, 4 major, and no critical, ranking second in accuracy. Translation errors for Method 4 totalled 35, 10 of which by the Google Translate MT system (8 minor and 2 critical, with no major detected). Fully automated Method 5 (ASR + MT) totalled 24 translation errors.

Methods 4 and 5 in which the final subtitle was product of an automatic process of MT showed some curious examples of mistranslations or misrecognitions that will now be illustrated[7]. The examples display the Spanish ST, the Spanish to Spanish respoken text (RT), the text recognised by Dragon (ASR), and the final translation by the MT for examples 1-3 taken from Method 4. Examples showed from Method 5 display ST, the ASR text and the TT, namely the final output produced by the MT. The back translation of the final output translated by the MT is provided in Italian by the author – please note that any not common or grammatically wrong English expression are willingly left to maintain the hints of the machine-translated input. No back translation is provided for the other utterances because the focus wants to be on the final subtitle only, namely what the audience would read.

Example 1 (Method 4):

ST: “De una crisis no se sale igual. Salimos mejores o peores”.

RT: “De una crisis no se sale igual o se sale mejor o peor”.

ASR: “La Cristina se sale igua o se sale mejor o peor”.

MT: Cristina esce lo stesso o esce meglio o peggio.

Back translation: Cristina goes out any way either she goes out better or worse.

In this example taken from text 1 (speech by Pope Francisco) the ASR software transcribes a sentence that does not make any sense, misrecognizing the dictation of ‘de una crisis’ con ‘La Cristina’, and therefore displaying a proper name that, despite not seeming correct in the given context, could confuse the reader to some extent.

Example 2 (Method 4):

ST: “Gracias a todos y de manera muy especial a mi partido sin el cual nada hubiera sido posible”.

RT: “Gracias a todos y de manera muy especial a mi partido sin el cual nada había sido posible”.

ASR: “Gracias a todos y de manera muy especial a mi partido sin el cual nada había sido posible”.

TT: “Grazie a tutti e in modo specialissimo alla mia festa senza la quale nulla sarebbe stato possibile”.

Back translation: Thanks to everybody and in a super special way to my party without her anything would have been possible.

In example 2 taken from text 2 (speech by former President Mariano Rajoy) both the respoken text and the transcription worked well, however the MT features firstly an issue of style (“de manera muy especial” as “in modo specialissimo”, which in Italian reads informally for a speech given in a parliamentary session), secondly, the translation of “Partido”, which makes sense in the English back translation as a “party” may be both a political party or a celebration, which is not the case for either Spanish or Italian. As an assumption, this could derive from the Google Translate MT model that from Spanish, first translates the utterance into English and not directly in Italian. One positive aspect, however, is that when the respeaker poorly dictated a past conditional tense sentence using “había” instead of “habría”, the software autocorrected it displaying the correct grammatical structure in the Italian TT.

Example 3 (Method 4):

ST: “Y vamos a ser más resilientes cuando trabajemos juntos en lugar de hacerlo solos”.

RT: “Vamos a ser más resilientes cuando trabajemos juntos en lugar de hacerlo solos”.

ASR: “Vamos a ser madres hirientes cuando trabajemos juntos en lugar de hacerlo solos”.

TT: “Saremo madri dolorose quando lavoreremo insieme invece che da sole”.

Back translation: We will be grieving mothers when we will work together instead of alone.

In this last example the misrecognition by Dragon leads to a critical error, since it creates a new meaning for the reader that was not the one in the ST. Interestingl, the MT, despite having translated correctly the second half of the sentence and detected the male gender and plural, consistently to what has wrongly recognised as “madres hirientes”, changed it all into the feminine plural.

Example 4 (Method 5):

ST: “En resumen, la pandemia del COVID nos ha enseñado esta interdependencia […]”.

ASR: “En resumen, la pandemia del obispo nos ha enseñado esta interdependencia […]”.

TT: “Insomma, la pandemia del vescovo ci ha insegnato questa interdipendenza […]”.

Back translation: To sum up, the bishop pandemic taught us this interdependence.

In example 4, taken from text 1, the word ‘Covid’ is misrecognised in “obispo” (bishop). While it would probably be clear that a ‘bishop pandemic’ is not the intended meaning, in this case it is the Pope speaking, so semantically he could be referring to bishops and it can be misleading.

Example 5 (Method 5):

ST: “Si alguien se ha sentido, en esta cámara o fuera de ella, ofendido o perjudicado le pido disculpas. Gracias a todos”.

ASR: “Si alguien se ha sentido en esta cámara oscura, bello, ofendido o perjudicado le pido gracias a todos”.

TT: “Se qualcuno si è sentito bello, offeso o leso in questa camera oscura, chiedo grazie a tutti”.

Back translation: If someone felt good looking, offended, or damaged in this darkroom, I ask thanks to everyone.

There are several problems in this last example taken from the speech by Mariano Rajoy. First, “o fuera”, jointly with the word “cámara” (referring to the Parliament) is misrecognised as “oscura” (darkroom) – as in for photography, which makes very little sense, or even as in a ‘shady room’, which could be particularly misleading given the diplomatic scenario in which the speech is given. Secondly, “de ella” is misrecognised as “bella” (good looking), and then tuned into a masculine by the MT. Lastly, the ST “disculpas” is omitted, which is an exception here since the machine rarely completely misses parts of speech. Together with the non-recognition of the full stop at the end of the sentence, the beginning of the new one is linked to the previous, resulting in misinformation: “pedir gracias” should be “decir gracias” and does not sound natural but could be seen as a stylistic error and not as a double omission (of “disculpas” and of the punctuation mark). Here the subtitling displays a worrying mistranslation.

6.1.2 Recognition errors

Recognition errors for the five tested methods were attributable to the ASR software; a total of 140 errors were detected.

In Method 1, 9 recognition errors were found: 5 were minor, 3 were major, and 1 was critical.

In Method 2, 6 recognition errors were found for the three videos: 5 were minor, 1 major. Method 3 registered the highest number of recognition errors, i.e. 57 in total: 51 minor, 2 major, 4 critical. In Method 4 there was just 1 critical recognition error, while for Method 5, as per Method 3, many more recognition errors were detected since there was no human intervention in either dictation to the SR software or monitoring of the written output, as can be seen in Figure 4 below. A total of 61 recognition errors were detected for the fully automated mode: 36 were minor, 12 were major, and 13 were critical.

Figure 4. Recognition errors in the five methods.

Recognition errors were mainly imputable to lack or shift of punctuation marks throughout the subtitles. As a result, in Methods 3 and 5 in which punctuation was added automatically by the speech recognition software, it is in some parts non-existent. Just for Method 3, approximately 80% of detected errors were punctuation based and out of 5 critical errors in Methods 3 and 5, 3 were punctuation errors, as follows.

Example 6 (Method 3):

ST: “Es el momento de actuar, estamos en el límite. Quisiera repetir un dicho viejo español”.

TT: “È il momento di agire perché siamo al limite secondo un vecchio detto spagnolo”.

Back translation: It is the moment to act because we have no time left according to an old Spanish saying.

Example 7 (Method 3):

ST: “Señoras y señores diputados, seré muy breve. A la vista de lo que todos sabemos […]”.

TT: “Signore, signori deputati, cercherò di essere molto breve come tutti sappiamo […]”.

Back translation: Ladies and gentlemen MPs, I will try to be concise as we all know.

Example 8 (Method 5):

ST: “[…] es bueno recordar cosas que nos decimos mutuamente para que no caigan en el olvido. Desde hace tiempo estamos tomando más conciencia […]”.

TT: “è sempre bene ricordare le cose che ci diciamo nell'oblio di quel tempo stiamo diventando più consapevoli…”.

Back translation: It is always good to remember things that we say in the oblivion of that time we are becoming more and more aware.

In this last example in particular, the sentence does not flow and it is clear that there was a misrecognition since “oblio” (oblivion) is it not relevant there. The lack of a full stop makes it seem as though “desde hace tiempo” refers to the previous sentence, thus giving it a different meaning.

6.2 Delay

The translation process is made of translated idea units and not sentences. To calculate delay, each rendered idea unit could have been considered calculating how many seconds after each idea is given in the ST it appears on screen as a subtitle. According to the Norma UNE (ibid.) an average is to be made of one sentence ending (idea unit ending) per minute of the video. Therefore, delay was calculated by choosing one sentence ending per minute in each ST and assessing the delay of that sentence ending compared to when the relevant text first appeared on screen as a subtitle. Methods 3, 4 and 5, delay calculations were speculative since the whole process was split into different stages during the pilot study. For Method 3, it was only possible to calculate delay for the delay of the simultaneous interpreter while translating, as their .mp3 recording was afterwards fed in Dragon and did not occur live. This was due to the fact the Dragon had no option of adding automatic punctuation, therefore the subtitles would have been a meaningless non-stop flow of words. To the interpreter’s delay an average of 1 extra second was added as per the time the ASR software usually takes to process the audio input and produce the written text. Usually, ASR takes less to process the transcript, but it does take a bit longer when longer utterances are dictated, as it is often the case when interpreting in SI. For Method 4, the intralingual respeaker delay was calculated and an average of 1 extra second was added as per the time the MT software usually takes to produce the translated output. Lastly, for Method 5, on average 2 seconds for the ASR and 1 extra second for the MT were calculated. It is to bear in mind that sometimes it can take less, and that this was a first attempt to compare the delays, without carrying out the five Methods fully live unfortunately. Anyway, delay in Methods 4 and 5 consistently took much less than any other method, with a substantial difference.

Table 5 below shows the delay calculated in each method for each ST, also offering an average among the three calculations and their ranking.

Table 5. Participants’ delay results in seconds.

Concerning delay, an inverse correlation with accuracy was detected: those methods that were faster in production of subtitles were the ones that scored lower accuracy rates, and those that required a longer delay were more accurate. In particular, participants in Method 2, who scored the highest in linguistic accuracy, was the slowest since two subjects were involved in the process. The correlation between accuracy and delay is all the more important and intended as a ‘trade-off’: more accurate subtitles take more time to be produced (and corrected), while less accurate ones are cued with much shorter delay.

Table 6. Accuracy and delay results.

7. Conclusions

By combining linguistic accuracy and delay results for each method, a consistent correlation can be observed: the least accurate method (Method 5) is the fastest in delivering the subtitles, while the most accurate (Method 2) takes longer, up to 11 seconds delay. The highest overall quality this experiment sought is, in fact, determined by the setting and the situation in which the ILS is delivered: the need for faster subtitles is to the detriment of accurate transcription and translation, otherwise better accuracy entails slower subtitles. In terms of overall quality, it is shown that Methods 1, 3, and 4 can be a good compromise, ranging between 6 to 7 seconds of delay, which seems to be acceptable especially for live programmes, also according to broadcasters that use LS such as the BBC. In the trade-off between accuracy and delay, despite being slower, Method 2 does provide a good level of accuracy. Other variables to consider in searching for quality ILS methods are the type of live event setting: Method 3 features too many errors in recognizing and placing punctuation marks, making it difficult for audiences to read the subtitles. The more and fully-automated Methods 4 and 5 leave the final decision on what and how to broadcast the content to the machine, which can be very risky in formal and important situations, especially concerning mistranslations. Nevertheless, these are also the cheapest methods, which need to be considered while evaluating which mode is the preferred, and most usable.

Several limitations have impacted this research, such as the distance setting for the training and testing, its restriction to the language combination Spanish to Italian, and to the use of only few available software, both ASR and MT systems. Moreover, subjectivity in the NTR analyses played a major role in the interpretation of the final results (Romero-Fresco & Pöchhacker, 2017), despite trying to minimise it through the two evaluators’ assessment. The fact that the intralingual respeaking in Spanish for Method 4 was carried out by a non-native speaker could is also a variable to take into account, even if – when listening to the audio recordings of the participant’s performance – phonology was clear and per se did not represent an obstacle in the ASR process.

Though marginally and despite all the limitations, it is hoped that these preliminary results can start and shed some light on which strengths and weaknesses each ILS method entails.

Abstract

English:

In our age of artificial intelligence, technology pervades every single human activity and human-machine interaction is now daily practice. In the area of diamesic translation, this has led to important distinctions in workflows, greatly reducing human intervention in many fields including institutional ones. However, and while it is clear that technology is here to stay, the stages that marked this journey often lie forgotten in end-of-project reports, published online and then removed at the end of the project itself. In an attempt to reconstruct the path that led to the pervasive role of technology in the production phases of translation, from speech to writing, this article suggests a quadripartition of human-machine interaction and illustrates two important Italian research projects, which implemented speech recognition technology in the production of pre-recorded, real-time subtitles. On the basis of these experiments, this article highlights the advancement of automation in the field of accessibility on one hand, which guarantees speed and accuracy=, and on the other underlines the contribution that accessibility has made to society in terms of flexibility and of the transparency of democratic and legislative processes.

Italian:

Nell’era dell’intelligenza artificiale, la tecnologia pervade ormai ogni singola attività umana e l’interazione uomo-macchina è prassi quotidiana. Nell’ambito della traduzione diamesica, essa ha portato a importanti distinzioni dei flussi di lavoro, riducendo notevolmente l’intervento umano in molti ambiti, compreso quello istituzionale. Ma se è chiaro che la tecnologia è qui per restare, le tappe che hanno segnato questo percorso sono spesso dimenticate in relazioni di fine progetto pubblicate online e poi rimosse con la fine del progetto stesso. In un tentativo di ricostruzione del percorso che ha portato la tecnologia a pervadere le fasi produttive della traduzione dal parlato allo scritto nella stessa lingua e in lingua straniera, questo articolo propone una quadripartizione dell’interazione uomo-macchina e illustra due importanti progetti di ricerca italiani, che hanno implementato la tecnologia del riconoscimento del parlato nella produzione di sottotitoli preregistrati e in tempo reale. Sulla base di questi esperimenti volti all’accessibilità, si è potuto notare l’avanzata dell’automazione nell’ambito dell’accessibilità, che riesce anche a garantire rapidità e accuratezza da un lato e il contributo che l’accessibilità ha dato alla società in termini di flessibilità e trasparenza dei processi democratici e legislativi dall’altro.

Keywords: interazione uomo-macchina, resocontazione parlamentare, sottotitoli per sordi, respeaking, traduzione intralinguistica, human-machine interaction, parliamentary reporting, subtitling for the deaf, intralinguistic translation

1. Introduzione

Nell’era dell’intelligenza artificiale, la tecnologia pervade ormai ogni singola attività umana. Hewett et al. definiscono questa interdipendenza tra l’uomo e la tecnologia interazione uomo-macchina e, più specificamente, l’implementazione di sistemi di calcolo interattivi ad uso umano (Hewett et al. 1992). Nell’ambito della traduzione basti pensare ai Computer-Aided Translation tools e alla loro evoluzione in Human-Aided Translation tools e Fully-Automated Translation tools (Eugeni e Gambier 2023), questi ultimi all’origine del Machine Translation Post Editing. Nell’elaborazione del linguaggio naturale, il riconoscimento automatico del parlato ha ridotto notevolmente l’intervento umano negli ambiti del Media Indexing, della resocontazione, della sottotitolazione e perfino dell’interpretariato (Eugeni 2019, Pagano 2020, Spinolo e Amato 2020, Romero-Fresco 2023) dove sono stati introdotti i concetti ibridi di written interpreting (Eugeni e Bernabé 2021) e AI-enhanced computer-assisted interpreting (Fantinuoli 2023). Nonostante la qualità del lavoro della macchina rispetto a quella del lavoro umano sia ancora molto dibattuta, il verdetto è chiaro: la tecnologia è qui per restare e il mercato non può che adeguarsi ad esso. Ma come si è arrivati a questo risultato? Quali sono state le tappe che hanno segnato questo percorso? Qual è stato l’impatto della tecnologia sulla vita di tutti i giorni? Come potrà ulteriormente evolvere l’interazione uomo-macchina in settori affini alla traduzione come la resocontazione parlamentare e la verbalizzazione del processo penale? In particolare, l’accessibilità, spesso beneficiaria della tecnologia e dell’adattamento di testi pensati per un pubblico più ampio, può contribuire a “restituire il favore” alla società, permettendole di capitalizzare sui successi dei progetti per l’accessibilità? Il presente articolo si propone di rispondere a queste domande, elaborando brevemente una ripartizione più dettagliata dell’interazione uomo-macchina (§2). Successivamente, verranno illustrati i tre progetti in cui si sono messi a confronto la produzione del lavoro umano con quello della macchina in due ambiti distinti ma affini, la sottotitolazione per sordi preregistrata (§3) e in tempo reale (§4). Infine, verrà illustrato il contributo in questi ambiti alla resocontazione parlamentare e la verbalizzazione del processo penale, tramite una piattaforma multifunzionale che garantisca accuratezza, trasparenza e accessibilità (§5).

2. Interazione Uomo-Macchina nella traduzione diamesica

Come appena introdotto, l’intervento della tecnologia gradualmente riduce l’intervento umano. Nell’ambito della traduzione diamesica, o traduzione intra-linguistica dal parlato allo scritto sotto forma, per esempio, di sottotitoli per sordi, resoconti parlamentari o trascrizione in tempo reale, “technological investments have reduced the place of humans to such an extent that the profession is impossible without the former, but not without the latter” (Eugeni, 2020: 22). Se ignoriamo il contributo della tecnologia nello sviluppo delle tecniche utilizzate (respeaking e velotipia) per produrre le forme di traduzione diamesica analizzate e ci concentriamo solo sulle fasi di produzione di una traduzione (nel caso dei sottotitoli le fasi sono trascrizione, eventualmente traduzione, correzione e formattazione), risulta chiara la ripartizione dei flussi di lavoro secondo il grado di interazione uomo-macchina presente, indipendentemente dalla tecnologia usata per produrre i testi di arrivo (Eugeni 2019), dalla lingua (Pagano 2020) e dal modo traduttivo (Fantinuoli 2023):

Human-Only, in cui le fasi fondamentali del processo traduttivo sono assegnati nella loro interezza a uno o più professionisti. Nella traduzione diamesica, i professionisti trascrivono, eventualmente traducono e correggono, limitando l’uso della tecnologia alla tecnica per produrre (respeaking, stenotipia, tastiera QWERTY) e visualizzare (software di videoscrittura) i sottotitoli;  

Computer-Aided, in cui il processo traduttivo è gestito dal professionista, che si avvale della traduzione per una minoranza delle fasi principali del processo traduttivo. Nella traduzione diamesica, il professionista trascrive il testo e usa la tecnologia per una o più fasi più facilmente automatizzabili come la correzione, la formattazione o, più spesso la traduzione;

Human-Aided, in cui il professionista vede il suo contribuito diminuire rispetto a quello della macchina nel processo traduttivo. Nella traduzione diamesica, il live editor/scopist si limita a correggere la trascrizione ed eventualmente la traduzione prodotte della macchina;  

Computer-Only, in cui il processo traduttivo è interamente assegnato a uno o più software. Nella traduzione diamesica, un software di riconoscimento automatico del parlato trascrive immediatamente il testo di partenza e un altro eventualmente lo traduce, lo corregge e lo manda in onda, senza che un professionista lo corregga o adegui alle esigenze degli utenti.

3. Interazione Uomo-Macchina nella sottotitolazione preregistrata

Nell’ambito del progetto CLAST[1], sono stati messi a confronto cinque video in lingua italiana, di cui due sottotitolati in italiano per sordi[2] da professionisti e due sottotitolati automaticamente. Anche un quinto video è stato sottotitolato dall’inglese in italiano in modalità automatica. Obiettivo della sperimentazione era la comprensione dei sottotitoli automatici intralinguistici e interlinguistici da parte degli utenti sordi rispetto a quella degli udenti. I video sono stati inizialmente valutati con la tassonomia IRA[3] e sono stati sottoposti a un gruppo di 16 sordi segnanti e a un gruppo di 16 udenti di pari livello socio-culturale. Dopo ogni video, a ciascun partecipante è stato chiesto di rispondere a test di comprensione e a un think-aloud protocol, per poterne valutare la ricezione e la percezione.  

3.1. Materiali

Al fine di valutare la qualità dei sottotitoli prodotti automaticamente tramite la tecnologia sviluppata durante il progetto CLAST, sono stati selezionati cinque video di una durata compresa tra i 2 e i 3 minuti:

• un documentario in italiano divulgativo sulla città di Siena nel Rinascimento;
• un documentario in italiano divulgativo sulla città di Firenze nel Rinascimento;
• un notiziario in italiano tecnico sullo sviluppo economico nell’Italia contemporanea;
• un notiziario in italiano tecnico sull’e-commerce;
• un notiziario in inglese tecnico sui fondamenti dell’economia.

La scelta dei tipi testuali (documentario divulgativo e notiziario economico) è stata dettata dalla necessità di testare i due profili sviluppati nel progetto CLAST: il modello divulgativo e il modello giornalistico. Per poter valutare l’efficacia della sottotitolazione automatica rispetto a quella manuale, dei due video a carattere divulgativo, il primo è stato sottotitolato automaticamente (Computer Made) dall’italiano in italiano (video 1) e l’altro manualmente (Human-Only), sempre dall’italiano in italiano (video 2). La stessa operazione è stata fatta con i primi due notiziari di ambito economico (video 3 e 4). Il quinto video è stato sottotitolato automaticamente (Computer-Only) dall’inglese in italiano (video 5). Non sono state usate forme di interazione uomo-macchina intermedie (Computer-Aided o Human-Aided) per la produzione dei sottotitoli.

Per quanto riguarda la qualità della sottotitolazione, si rende evidente che, applicando la tassonomia concettuale IRA, la qualità del video 1 è pari all’88 per cento, mentre quella del video 2 è pari al 100 per cento. Una differenza inferiore tra i due video è stata riscontrata tra il video 3 e il video 4. Il video 3 ha una qualità del 95 per cento contro il 100 per cento della qualità del video 4. La qualità del video 5 è invece dell’85 per cento. Pur essendo di una qualità inferiore al 95 per cento considerata come accettabile nell’elaborazione della tassonomia IRA, i video 1 e 5 sono comunque stati testati.

I video sono stati sottoposti a tutti i partecipanti con l’audio a un volume udibile a tutti gli udenti e con i sottotitoli. Ogni diverso trattamento tra i due gruppi è stato considerato inutile o fuorviante e pertanto scartato. Va sottolineato che la componente visiva dei primi due video ha una valenza ancillare rispetto alla componente acustica, perché la comprensione delle immagini dipende sempre dalla corretta percezione della componente acustica verbale. La stessa caratteristica è stata riscontrata nel video sottotitolato interlinguisticamente (video 5), per quanto quest’ultimo contenga informazioni di tipo scritto-verbale. La componente visiva dei video 3 e 4, invece, svolge un ruolo diverso, perché auto-esplicativa: fornisce informazioni comprensibili senza fruire la relativa componente acustica verbale.

3.2. Partecipanti

Target principale del progetto CLAST sono stati gli utenti sordi. Per questo motivo si è scelto di collaborare con l’Ente Nazionale Sordi di Trento. Oltre che per motivazioni legate alla vicinanza dalla sede della ricerca, la scelta di testare i sottotitoli su una popolazione segnante è stata dettata dalla necessità di testare l’efficacia dei sottotitoli sul tipo di pubblico che maggiormente si allontana dallo standard linguistico dell’utente medio dei video in questione. In linea con sperimentazioni simili svolte a livello nazionale[4] e internazionale[5], la scelta del gruppo target è ricaduta sui sordi segnanti. Il focus group era composto da 17 sordi segnanti gravi o profondi, di cui 9 donne e 8 uomini, impiegati, di età compresa tra i 30 e i 70 anni e con scolarizzazione tra le scuole medie e le scuole superiori. Dal gruppo iniziale, sono stati esclusi i risultati di un volontario che aveva scelto più opzioni di risposta a quasi tutti i quesiti, alterando così l’esito del test.

Nonostante tutti si fossero dichiarati bilingui, il focus group era composto da soci dell’ENS – sordi prelinguali o perilinguali – la cui lingua materna è la lingua dei segni italiana (LIS). Il control group era composto da altrettante persone (16), scelte sulla base del medesimo profilo socio-culturale dei 16 sordi rimanenti del focus group: 9 donne e 7 uomini impiegati, di età compresa tra i 30 e i 70 anni e scolarizzati fino alle superiori.

A latere, preme ricordare che, da uno studio condotto dall’ente regolatore dei broadcaster britannici Ofcom, che gli utenti del servizio di sottotitolazione dei teletext britannici solo per il 20 per cento sono composti da ipoacusici o cofotici. Il restante 80 per cento è rappresentato principalmente da stranieri (immigrati, studenti, amici di britannici temporaneamente nel Regno Unito), abitanti che lavorano in ambienti rumorosi o di cittadini britannici emigrati all’estero - o dei loro discendenti - che vogliono mantenere un contatto con la lingua materna scritta e parlata (Ofcom 2006, 2013). Pur con differenze di ordine sociale e demografico, lo studio britannico non può essere ignorato nella valutazione della qualità dei sottotitoli automatici oggetto del presente studio di ricerca.

3.3. Questionari

Sei questionari sono stati sottoposti ai partecipanti, a ognuno dei due gruppi che si sono sottoposti al test, in linea con quanto già definito nel progetto DTV4All: il primo questionario è stato proposto ai partecipanti prima della presentazione del lavoro e richiedeva dati riguardanti le generalità principali dei partecipanti (genere, età, scolarizzazione, impiego e metodo di comunicazione), il loro rapporto con la lingua scritta e la lettura (siti web, sottotitoli TV, social network) e le loro preferenze in termini di tipi testuali (film e fiction, documentari divulgativi, notiziari economici, programmi sportivi).

Gli altri cinque questionari sono stati sottoposti ai partecipanti dopo ogni video visionato ed erano composti da sei domande ciascuno, incentrate sulle due grandi componenti semiotiche del testo: acustica e visiva. Le tre domande di ogni questionario incentrate sulla componente acustica riguardavano i macro-argomenti trattati nel video, mentre quelle incentrate sulla componente visiva riguardavano immagini di supporto al testo ma utili alla sua comprensione o immagini auto-esplicative come luoghi descritti, dati (parole e numeri) sovrimpressi o grafici.

Alla fine della compilazione dei questionari si è proceduto alla somministrazione di un Think-Aloud Protocol individuale e poi a un confronto collettivo sulle criticità dell’esperimento. I risultati sono stati raccolti e analizzati sia globalmente, sia nel dettaglio.

3.4. Risultati

Non essendo state rilevate differenze sostanziali dovute al genere, all’età o al livello di  scolarizzazione, si procederà ora ad analizzare i dati secondo le variabili più interessanti. Dopo un’analisi generale dei risultati ottenuti dai sordi e dagli udenti, si entrerà nel dettaglio dei singoli video in termini assoluti (percentuale di risposte esatte) e relativi (percentuale di risposte esatte rispetto alla qualità dei sottotitoli). Poi si prenderanno in considerazione le macro-variabili più interessanti, a partire da quelle di maggiore interesse per il presente studio, come la modalità di produzione del sottotitolo (automatico vs. manuale). Successivamente si procederà all’osservazione dei dati riguardanti il tipo di programma (documentario divulgativo vs. notiziario economico). Questo ci permetterà di capire se e in che misura il tipo testuale dei video influenza la percezione e la ricezione di un video. In seguito, si analizzeranno i due tipi di domande contenute nei questionari: quelle relative alla comprensione delle informazioni veicolate dalla componente acustica e quelle relative alla comprensione delle informazioni veicolate dalla componente visiva. Questo permetterà di comprendere se, indipendentemente dalla qualità, i sottotitoli automatici comunque agevolano la comprensione delle immagini oppure no. Infine, si analizzeranno i dati relativi al video sottotitolato automaticamente dall’inglese in italiano.

3.4.1. Sordi vs Udenti

In generale, le risposte degli udenti sono state migliori di quelle date dai sordi, con una media di 4,2 risposte corrette su 6 per i primi vs 3,8 per i secondi. Questo dimostra che la comprensione dei video da parte di chi riesce contemporaneamente a percepire la componente acustica e quella visiva su due canali distinti è superiore a quella di chi deve utilizzare un unico canale (quello visivo) per percepire informazioni di tipo sia visivo che acustico (cfr. Paivio 1986). Questa operazione comporta una minore possibilità di concentrarsi anche sulle immagini, non riuscendo quindi a percepire correttamente l’interrelazione tra parlato e immagini, contrariamente al pubblico udente. Si consideri inoltre che i partecipanti al focus group non sono madrelingua italiani e i messaggi del video sono tutti veicolati sul canale visivo, che non è comunemente deputato alla comunicazione orale. Per comprendere, quindi, se i sottotitoli automatici hanno influito negativamente su questo risultato, occorre analizzare l’esito delle risposte alle domande di tipo testuale e di tipo visivo.

Tabella 1: Risposte corrette ai questionari suddivise per singolo video

Dai dati esposti nella Tabella 1 emerge quanto espresso precedentemente sulla comprensione generale degli udenti rispetto ai sordi. Sistematicamente, la comprensione degli udenti dei diversi video è maggiore (video 1, 2, 3, 5) rispetto a quella dei sordi. Da una più approfondita analisi, tale divario risulta maggiore nel caso dei video sottotitolati automaticamente (+ 1 nel video 1 e + 0,7 nel video 3), mentre in uno dei due video sottotitolati manualmente il divario è - 0,4 nel video 4, anche se il video 2 non conferma tale costante. Considerata questa eccezione, è necessario controllare i dati più nel dettaglio per comprendere se l’ipotesi che ne deriva – i sottotitoli manuali sono più adeguati alla comprensione rispetto a quelli automatici – è fondata o il dato deriva da altre ragioni (cfr. § 3.4.3).

Un altro dato che salta all’occhio riguarda la maggiore comprensione da parte dei sordi del video 4, ossia il notiziario economico sottotitolato manualmente. Come si vedrà dall’analisi delle risposte alle domande di tipo testuale e le domande di tipo visivo (§ 3.4.5), questo dato è la risultante della maggiore comprensione da parte dei sordi della componente visiva. Visto che il notiziario in questione è il più ricco di dati a video, questo ci porta a tre prime conclusioni:

• anche gli udenti soffrono di sovraccarico cognitivo qualora la componente video veicoli non solo informazioni secondarie (immagini che accompagnano il testo pronunciato oralmente), ma anche primarie (in questo caso i dati sull’e-commerce);
• di contro, la maggiore abilità a usare la vista per comprendere le informazioni offre ai sordi un vantaggio sugli udenti nel comprendere testi con prevalenza di informazioni visive;
• oltre a permettere una maggiore comprensione della componente scritta rispetto ai sottotitoli prodotti automaticamente, i sottotitoli prodotti manualmente permettono di lasciare il tempo agli utenti di ottenere informazioni anche dalla componente visiva.

Altro dato degno di nota che sarà analizzato successivamente (§ 3.4.6) riguarda i dati dei sordi riferiti al video in inglese sottotitolato automaticamente, che registra dei risultati vicini a quelli degli udenti. Questo dato è interessante perché sordi e segnanti sono, in questo caso, posti su un livello di esposizione al video particolarmente più simile rispetto alla ricezione degli altri quattro video, dato che la loro conoscenza dell’inglese parlato è più limitata.

3.4.2. Comprensione assoluta vs. Comprensione relativa

Un altro elemento importante da analizzare riguarda la comprensione relativa. I dati sopra riportati, infatti, riguardano la comprensione assoluta, vale a dire la comprensione del singolo video indipendentemente dalla maggiore o minore qualità dei sottotitoli. Per comprendere quanto la qualità del sottotitolo influisca sulla comprensione generale del prodotto, però, è più interessante conoscere la qualità relativa del sottotitolo. Per qualità relativa si intende quanto i sottotitoli, in base alla loro accuratezza misurata con IRA, riescono a veicolare i messaggi ai quali si riferiscono. La comprensione relativa (CR), ossia il numero relativo di risposte esatte alle domande poste, viene ricavato moltiplicando la comprensione assoluta (CA) per il quoziente qualitativo dei sottotitoli (qqS), che si ottiene dividendo 100 per la qualità dei sottotitoli calcolata da IRA (qIRA), secondo la formula CR = CA * qqS, dove qqS = 100/qIRA, quindi CR = CA * 100/qIRA.

Da questo emerge che la CR è inversamente proporzionale alla qualità dei sottotitoli. Vale a dire che, a parità di CA, minore è la qualità dei sottotitoli e maggiore sarà la CR. Di fatto, si tratta dunque di comprendere quanto i rispondenti abbiano davvero capito del video, nonostante eventuali errori nei sottotitoli. Saranno considerati soltanto i sottotitoli visionati dai sordi, poiché gli udenti, al netto di irrilevanti casi, hanno potuto percepire acusticamente la componente testuale. Calcolare anche per gli udenti la comprensione relativa significherebbe falsare la realtà del processo cognitivo degli udenti, pur riconoscendo l’influenza della presenza dei sottotitoli nel processo di visualizzazione dei video da parte degli udenti stessi (cfr. Romero-Fresco 2105 e Bianchi et al. 2020). Da segnalare è anche la scelta delle componenti da considerare. Analizzare la sola componente testuale vorrebbe dire riconoscere ai sottotitoli un’influenza sulla sola componente testuale. Studi scientifici hanno ampiamente dimostrato che la qualità e la densità dei sottotitoli hanno un impatto decisivo anche sulla visione delle immagini (ibidem). In particolare più è denso il sottotitolo, minore sarà il tempo dedicato alla visione delle immagini da parte degli utenti. Di converso, maggiore sarà la qualità dei sottotitoli maggiore saranno il tempo e l’attenzione dedicati alla visione delle immagini.

Tabella 2: Comprensione assoluta e relativa dei sordi suddivisa per singolo video

I dati sulla CR illustrati in Tabella 2 mostrano una prima evidenza: al relativizzare della comprensione, il divario tra la sottotitolazione automatica e quella manuale si riduce notevolmente. Nel caso dei primi due video, esso si riduce di oltre la metà (da 0,7 a 0,3), mentre nel caso dei secondi due rimane pressoché invariato, per via della maggiore qualità dei sottotitoli automatici del video 3. Sorprende altresì il balzo qualitativo del video 5, che arriva a rivaleggiare con il video 2, sottotitolato manualmente. Questo non significa che i sottotitoli interlinguistici automatici siano efficienti quanto quelli intra-linguistici manuali, tuttavia dimostra il potenziale di sottotitoli prodotti dalla doppia automazione della trascrizione e della traduzione. Con l’evolversi della ricerca in questo ambito, l’impiego efficiente di sottotitoli automatici, intra-linguistici o interlinguistici, è sempre più a portata di mano.

3.4.3. Sottotitoli automatici vs. Sottotitoli manuali

I sottotitoli prodotti manualmente (video 2 e 4) sono più efficienti di quelli prodotti automaticamente (video 1, 3 e 5) nel veicolare messaggi di tipo sia acustico che visivo. Questo risulta non solo dai dati assoluti (§ 3.4.1), ma anche da quelli relativi (§ 3.4.2) e da quelli scorporati per componente semiotica (§ 3.4.4). Nella Tabella 3, i dati riferiti ai sottotitoli automatici comprendono solamente la media delle risposte corrette riguardanti i video sottotitolati dall’italiano in italiano (video 1 e 3), perché il video 5 introduce la variabile della traduzione interlinguistica, che non è presente nei sottotitoli manuali.

Tabella 3: Comprensione dei sordi suddivisa per tipo di sottotitoli

Dai dati della Tabella 3 emergono diverse realtà degne di nota. Innanzitutto, in termini assoluti, risulta chiaro che i sottotitoli prodotti manualmente garantiscono una maggiore comprensibilità dei messaggi veicolati dai video usati nella sperimentazione (+ 0,7 domande), pari a una differenza dell’11,7 per cento Tuttavia, visto che la qualità del riconoscimento automatico del parlato è in progressivo miglioramento e che la qualità della componente testuale impatta anche sulla fruibilità della componente visiva, quindi del video nel suo insieme, si è ritenuto interessante rapportare la comprensione dei sordi alla quantità effettiva di informazioni ricevute dal video sottotitolato automaticamente. In questo caso, le informazioni effettivamente veicolate dai video sottotitolati automaticamente non corrispondono al 100 per cento delle informazioni veicolate nei video sottotitolati manualmente, ma a una percentuale di volta in volta variabile. Da questo calcolo della comprensione relativa dei video sottotitolati automaticamente, emerge che il divario con i video sottotitolati manualmente, prima netto, si rivela molto inferiore (0,4 risposte, pari al 6,7 per cento.

Un altro dato interessante, che si può leggere da due prospettive opposte, riguarda la componente testuale. Le risposte alle domande sulla componente testuale sono quelle che mostrano il maggiore divario tra i video sottotitolati automaticamente (3,3 risposte corrette su 6) e quelli sottotitolati manualmente (4,4 su 6):

• da un lato, 1,1 risposte su 6 dimostrano che l’area di intervento diretto della sottotitolazione risente negativamente dell’automazione rispetto a un trattamento professionale. Questo può sembrare banale, ma è sempre interessante quantificarne la portata, che in questo caso è del 18,3 per cento.
• dall’altro lato, le stesse 1,1 risposte su 6 di differenza dicono che i sottotitoli automatici non impattano negativamente sulla ricezione del video che traducono, perché il divario in termini relativi, ma anche assoluti, è inferiore.

La seconda ipotesi è confermata dai dati riguardanti la comprensione della componente visiva dei video. In termini teorici, una buona sottotitolazione permette agli spettatori di comprendere la componente acustica del video, lasciando loro il tempo di concentrarsi sulle immagini. Di conseguenza, sottotitoli qualitativamente inferiori non solo veicolerebbero meno informazioni, ma comprometterebbero la visione del video nel suo insieme, rallentando la lettura dei sottotitoli e lasciando allo spettatore meno tempo per guardare le immagini e cogliere le informazioni che veicolano. Di fatto, invece, emerge che la relativa bassa qualità dei sottotitoli automatici non comporta un perdita di informazioni altrettanto importante.

3.4.4. Componente acustica vs. Componente visiva

A completamento dell’analisi precedente sui sottotitoli automatici rispetto ai sottotitoli manuali (§3.3), è forse utile controllare non solo i dati riferiti ai singoli video, ma anche i dati riferiti agli utenti udenti. Questa necessità si impone per poter corroborare o confutare le ipotesi fin qui esposte e riferite alle due componenti in questione. Una maggiore presenza delle domande corrette in una delle due categorie potrebbe, infatti, dipendere non solo dalla qualità dei sottotitoli, ma anche da una intrinseca maggiore o minore trasparenza delle informazioni oggetto dei singoli questionari. Neanche in questa analisi sono stati considerati i dati relativi al video 5 (cfr. § 3.4.6), perché alterano il rapporto con la fruibilità del video da parte degli udenti.

Tabella 4: Comprensione suddivisa per tipo di componente semiotica e tipo di utente

Dalla Tabella 4 emerge un dato che parrebbe contraddire quanto esposto: non esiste correlazione alcuna tra qualità del sottotitolo e comprensione testuale. Infatti, i dati riferiti alla comprensione della componente audio rispetto a quelli riferiti alla componente video non sembrano presentare regolarità, né per ognuna delle due categorie di utenti, né all’interno dello stesso video. A voler forzare l’analisi, escludendo quindi i dati contraddittori, è possibile trarre tre parziali conclusioni, non tutte valide, perché in contraddizione le une con le altre:

• se si ignorano i dati riferiti al video 2, è possibile riscontrare una maggiore comprensione da parte dei sordi della componente visiva veicolata dai sottotitoli rispetto alla comprensione della componente acustica. Se questa ipotesi fosse suffragata dai dati, questo significherebbe che il tempo dedicato alla componente visiva è superiore a quello dedicato alla lettura dei sottotitoli, che quindi non monopolizzerebbero l’attenzione dello spettatore, neanche in caso di sottotitoli automatici, che superano in termini quantitativi quelli prodotti manualmente. Questa ipotesi è parzialmente avvalorata da una caratteristica intrinseca dei primi due video: la secondarietà delle immagini rispetto al testo esclusivamente veicolato tramite la componente acustica;
• se si ignorano i dati riferiti al video 3, un’ipotesi immediata ma poco probabile è l’andamento diametralmente opposto della comprensione delle componenti semiotiche da parte dei sordi rispetto alla comprensione degli udenti. Infatti, ad eccezione del video 3, ogni volta che la comprensione della componente acustica da parte dei sordi risulta inferiore rispetto a quella visiva, negli udenti si ha una situazione ribaltata (comprensione della componente acustica superiore a quella visiva). Allo stesso modo, ogni volta che la comprensione della componente acustica da parte dei sordi risulta superiore rispetto a quella visiva, negli udenti la comprensione della componente acustica risulta inferiore a quella visiva. Questa ipotesi potrebbe essere dettata dalla scarsa influenza dei sottotitoli sulla visione dei programmi da parte degli udenti;
• se si ignorano i dati riferiti al video 4, emerge una ipotesi più ragionevole della precedente per quanto banale: la qualità dei sottotitoli è direttamente proporzionale alla comprensione delle informazioni veicolate dai sottotitoli stessi e sul tempo dedicato alla componente video. Se si analizzano i dati dei soli sordi, si noterà, infatti, che i video sottotitolati manualmente comportano una maggiore comprensione della componente acustica contrariamente a quanto accade nei video sottotitolati automaticamente, che contengono errori e/o troppo testo. 

3.4.5. Documentari culturali vs. Notiziari economici

Un’interessante analisi è quella riferita alla comprensione dei due tipi di video (documentario culturale e notiziario economico) in base alle preferenze dei singoli rispondenti, ai quali è stato chiesto di indicare i programmi preferiti. L’ipotesi è che la comprensione dei singoli programmi dipenda dalla maggiore o minore affinità del singolo utente con il tipo di programma in questione.

Tabella 5: Comprensione suddivisa per tipo di programma e tipo di utente

L’ipotesi appare immediatamente e inequivocabilmente suffragata dai dati riportati in Tabella 5. Sia i sordi, sia gli udenti sembrano comprendere meglio sia la componente acustica (veicolata o meno dai sottotitoli), sia la componente visiva del tipo di programma che preferiscono guardare in TV o su Internet. Per quanto riguarda i sordi, questo sembra valere per i programmi sottotitolati manualmente e per i programmi sottotitolati automaticamente.

Questo ci porta a un’ulteriore considerazione: in caso di programma con il quale lo spettatore ha maggiore affinità, la qualità del sottotitolo ha una minore influenza sulla comprensione. L’effetto positivo della maggiore affinità di un utente sordo con il programma sembra esserci anche in termini di tempo di lettura (più rapidi) e di tempo a disposizione per guardare le immagini.

Guardando i dati con maggiore attenzione, si può notare infine che la variazione sia più forte nei sordi. La ragione dipende dal maggiore sforzo cognitivo compiuto dai sordi nell’usare la vista come solo canale sensoriale per accedere a entrambe le componenti dei video. Tuttavia, una domanda sorge spontanea: sordi e udenti comprendono meglio un programma preferito perché hanno davvero colto il senso del filmato o perché conoscevano le risposte ex ante? Certamente una conclusione che può essere tratta è che nel caso di video preferito, la relativa minore qualità dei sottotitoli automatici rispetto a quella dei sottotitoli manuali avrà un effetto minore sulla comprensione, dato che l’utente userà la propria conoscenza per sopperire a eventuali carenze dei sottotitoli. Questo offre ai sottotitoli automatici una più ampia gamma d’impiego.

3.4.6. La sottotitolazione interlinguistica automatica

L’ultima analisi riguarda i dati sulla sottotitolazione automatica dall’inglese in italiano. I dati riguardanti il video 5 sono particolarmente interessanti per diverse ragioni. Innanzitutto perché udenti e sordi sono stati messi su un simile livello[6], quindi tutti i partecipanti hanno utilizzato i sottotitoli automatici per poter rispondere alle domande di comprensione poste nel questionario.

Tabella 6: Comprensione del video 5 suddivisa per tipo di comprensione e tipo di utente

Dai dati della Tabella 6, emerge una sensibile differenza tra la media dei dati riferiti alla comprensione degli altri video (più evidente negli udenti), spiegabile con l’introduzione della doppia automazione (della trascrizione e della traduzione). Da qui emerge una prima conclusione: quando sordi e udenti sono posti su un livello di relativa parità rispetto alla visione di un filmato (perché entrambi i gruppi devono sostanzialmente dipendere dai sottotitoli per comprenderlo), i risultati in termini di comprensione tendono a equivalersi.

É interessante, tuttavia, notare i dati scorporati per tipo di domande al fine di comprendere come, nel caso dei sordi, la comprensione del video sia maggiormente dettata da una prevalenza di risposte corrette alle domande sulla componente visiva, mentre negli udenti prevalgono le risposte alle domande su informazioni veicolate acusticamente. Questo ci porta a tre conclusioni:

• dconsiderato che i sordi sono più abituati a usare la vista per comprendere un video, essi tendono a trarre vantaggio da questa maggiore abilità, quando sono messi su un livello di relativa parità con gli udenti, riuscendo a cogliere più informazioni provenienti dalla componente visiva del video e più agevolmente;
• anche in caso di sottotitoli non perfetti, i sordi si fanno meno distrarre dall’errore e procedono a cogliere il senso delle informazioni veicolate dai sottotitoli per potersi concentrare sulle immagini il più possibile;
• gli udenti sono più disturbati dall’errore nei sottotitoli[7] rispetto ai sordi, perché tendono a concentrarsi più sulla correttezza grammaticale dei sottotitoli e meno sul senso.

Dall’analisi delle risposte al questionario sui dati personali emerge che quasi la metà degli udenti ha usato i sottotitoli per completare le loro limitate conoscenze della lingua inglese. Risulta quindi chiaro il motivo della maggiore prevalenza della comprensione alle domande sulla componente acustica (che maggiormente veicola informazioni di tipo verbale) rispetto a quella visiva.

3.5 Discussione

La sperimentazione che ha coronato il percorso del progetto CLAST ha portato alla luce interessanti risultati degli sforzi compiuti nello sviluppo dei software di trascrizione e traduzione automatiche. La prima evidente e scontata conclusione riguarda la maggiore comprensione dei video da parte degli udenti, specialmente se sottotitolati automaticamente. Questo dato conferma la maggiore difficoltà dei sottotitoli di permettere l’accessibilità di informazioni pensate per un pubblico udente. Tuttavia esso cela un interessante e inaspettato aspetto, cioè la maggiore comprensione dei sordi del video 4, che comporta due ipotesi interessanti:

• la maggiore abilità dei sordi a usare la vista offre loro un vantaggio sugli udenti nella comprensione di programmi la cui componente visiva svolge un ruolo prevalente;
• i sottotitoli automatici non inficiano la comprensione delle informazioni veicolate dalle immagini, neanche qualora queste fossero collegate a sottotitoli non corretti.

Queste considerazioni valgono non solo in termini assoluti, ma anche e soprattutto in termini relativi, cioè quando si rapporta la media delle risposte corrette all’effettiva quantità di informazioni veicolata dai video sottotitolati. Da questo calcolo della CR dei video sottotitolati automaticamente, si deduce che il divario con i video sottotitolati manualmente è meno importante della metà.

Considerando le differenze tra comprensione delle informazioni veicolate dalla componente acustica e informazioni veicolate da quella visiva, emergono altre due conclusioni fondamentali:

• sia i sordi, sia gli udenti comprendono meglio sia la componente acustica (veicolata o meno dai sottotitoli), sia la componente visiva del tipo di programma che preferiscono.
• per quanto riguarda i sordi, questo sembra valere sia per i programmi sottotitolati manualmente, sia per quelli sottotitolati automaticamente.

Pertanto, in caso di programma con il quale lo spettatore ha maggiore affinità, la qualità del sottotitolo ha una minore influenza sulla comprensione dello stesso, dato che userà la propria conoscenza per sopperire a eventuali carenze dei sottotitoli. Per quanto riguarda i sottotitoli interlinguistici prodotti automaticamente, qualora gli udenti ne avessero bisogno per comprendere le informazioni contenute nei video, essi creano una situazione che permette maggiormente di comparare le prestazioni di sordi e udenti. In termini di comprensione tendono, infatti, a equivalersi, mostrando una verità sostanziale: l’abitudine a utilizzare i sottotitoli rende i sordi meno distratti da eventuali errori in essi contenuti. Di contro, gli udenti sono maggiormente disturbati dall’errore nei sottotitoli rispetto ai sordi, perché si concentrano di più sulla loro correttezza formale.

Da questa analisi, i sottotitoli automatici intra-linguistici e interlinguistici, mostrano di assolvere alla loro funzione di veicolo di informazioni, specialmente in caso di video che tratta argomenti affini allo spettatore. Sarà quindi interessante capire quanto lo sviluppo di questa tecnologia abbia influito su altre discipline, come ad esempio la sottotitolazione in tempo reale.

4. Interazione Uomo-Macchina nella sottotitolazione in tempo reale

Nel 2018, la Città Metropolitana di Roma ha avviato il progetto Tirone per l’accessibilità universale delle sedute consiliari: un servizio di sottotitolazione intralinguistica (da Italiano a Italiano) e interpretariato in Lingua dei Segni Italiana (LIS) in tempo reale delle sedute consiliari trasmesse in streaming[8]. L’obiettivo è renderle fruibili al pubblico sordo in primis e a tutte le persone che dovessero trovare utile o necessario accedervi tramite i sottotitoli in italiano o la LIS.

Il principio ispiratore del Progetto Tirone è il concetto di progettazione universale così come interpretato dalla Fondazione ASPHI onlus: “approccio incentrato sull’utente (…), al fine di non proporre una soluzione unica per tutti, piuttosto un prodotto capace di fornire diverse alternative per soddisfare (meglio se automaticamente, apprendendo e adattandosi) l’insieme di abilità, requisiti e preferenze dei singoli utenti[9]. Come mostrato dalla Figura 4, il processo di produzione del servizio di accessibilità del progetto Tirone avviene in più fasi e su più tracce, ma il servizio viene fornito in un unico stream disponibile sul canale YouTube del Comune di Roma.

Il presente paragrafo fornisce un’analisi quali-quantitativa della fase sperimentale del progetto e delle raccomandazioni fornite da una consultazione pubblica con i rappresentanti delle associazioni di sordi oralisti e segnanti, che hanno aggiunto il punto di vista dell’utenza finale ai dati oggettivi. Dopo una breve illustrazione del metodo seguito nel progetto Tirone (§4.1) e dell’analisi quali-quantitativa dei sottotitoli e dell’interpretariato LIS delle sedute consiliari (§4.2), seguiranno le raccomandazioni dei tecnici e il punto di vista dell’utenza finale sorda, raggruppate per garantire una maggiore usabilità del presente lavoro (§4.3) e uno sviluppo della tecnologia in materia (§5).

4.1. Metodo

In questo progetto si è seguito un metodo parzialmente diverso da quello utilizzato nel progetto CLAST. In particolare, sono stati selezionati 3 campioni di 10 minuti ciascuno relativi alla diretta in streaming di tre sedute del consiglio capitolino in maniera del tutto casuale. L’audio originale è stato trascritto e suddiviso in unità concettuali. Le unità concettuali sono state paragonate ai sottotitoli e all’interpretariato LIS. Grazie alla tassonomia IRA già descritta precedentemente, è stato possibile valutare la qualità dei sottotitoli (§4.2.1) e dell’interpretariato (§4.2.2). Rispetto all’analisi precedente, si è proceduto in questo caso ad applicare la tassonomia IRA per intero, permettendo così di comprendere come sono state rese (ripetizione o alterazione, ulteriormente suddivisa in riduzione, correzione o errore marginale) o non rese (omissione o travisamenti, ulteriormente suddivisi in errori grammaticali e lessicali) le unità concettuali in questione. Come nello studio precedente, è stata presa in considerazione la soglia del 95 per cento come soglia minima dell’accuratezza, corrispondente al 98 per cento della tassonomia NER (Romero Fresco e Martínez 2015) usata da Ofcom. Dall’Ofcom è stato ripreso anche il concetto di qualità dei servizi di accessibilità, vale a dire una resa del maggior numero possibile di unità concettuali, utilizzando (nel caso dei sottotitoli) il maggior numero possibile di parole del testo di partenza (Ofcom 2013). Invece di procedere a un test sulla ricezione dei sottotitoli, si è proceduto poi con una consultazione pubblica alla quale hanno preso parte le associazioni di categoria. Durante l’incontro sono stati illustrati i dati e discussi i risultati in termini di punti di forza e opportunità di sviluppo del servizio (§4.3).

4.2. Risultati

In questo paragrafo saranno brevemente illustrati i dati relativi alla qualità oggettiva dei sottotitoli in tempo reale e dell’interpretariato in LIS di tre sedute di un consiglio capitolino risultante dall’applicazione della tassonomia IRA precedentemente illustrata.

4.2.1. La qualità dei sottotitoli intralinguistici

All’interno del progetto Tirone, la piattaforma utilizzata per l’accessibilità delle sedute consiliari prevede una doppia modalità di produzione dei sottotitoli in tempo reale:

• assistita: un respeaker detta i sottotitoli al software di riconoscimento del parlato che li trascrive e un live editor li corregge;
• siretta: il software produce i sottotitoli automaticamente e il live editor li corregge.

Dall’analisi dei sottotitoli prodotti secondo la modalità assistita, emerge subito che la quantità delle unità concettuali rese nei sottotitoli è del 96,2 per cento, superiore al criterio minimo utilizzato come riferimento (Figura 1).

Figura 1: Risultati dell’analisi della qualità dei sottotitoli prodotti dal respeaker

A uno sguardo più approfondito ai risultati dell’analisi condotta (Figura 1), emerge anche che i sottotitoli sono stati prodotti con un numero di ripetizioni molto alto (87 per cento), conformemente alle linee guida Ofcom. Per quanto riguarda le alterazioni, o modifiche rispetto al testo di partenza, esse rappresentano un decimo (9,2 per cento) delle strategie adottate dai sottotitolatori per giungere al risultato finale. Tra queste, preponderanti sono le riduzioni (omissione di parole ridondanti o compressione di pensieri complessi) e le correzioni (principalmente grammaticali) dell’oratore da parte dei sottotitolatori. Trascurabili sono gli errori che non influiscono sulla comprensibilità dei sottotitoli. Quanto alle unità concettuali che non sono passate nei sottotitoli (3,8 per cento), notiamo che sono principalmente costituite dalle omissioni dell’unità concettuale stessa. Si tratta principalmente di unità che garantiscono la transizione da un concetto all’altro con una portata semantica marginale. Tuttavia la coesione testuale risulta essere intaccata, seppur in maniera molto circostanziata. Gli errori riscontrati sono pochi, ma in questo caso compromettono la comprensibilità dei sottotitoli. Infine il ritardo registrato è di 3,9 secondi in media. Nel dettaglio esso è di 3,4 secondi nel caso di sottotitoli contenenti ripetizioni del testo pronunciato e di 4,5 quando il sottotitolatore lo modifica. Tale ritardo è conforme a quanto richiesto dai parametri di qualità previsti dal Televideo – RAI (6 secondi).

Se si mettono questi dati a confronto con quelli dei sottotitoli prodotti tramite la seconda modalità (diretta), emergono alcune conclusioni, di cui alcune saltano subito all’occhio, mentre altre richiedono una maggiore attenzione (Figura 2).

Figura 2: Risultati dell’analisi della qualità dei sottotitoli prodotti dalla macchina

In primis, il numero di ripetizioni prodotte con questa modalità è superiore rispetto a quelle prodotte dalla modalità assistita dal respeaker, nonostante l’intervento del live editor. Questo avviene per due motivi fondamentali: 1) nonostante provi un approccio verbatim alla sottotitolazione, il respeaker non riesce, come invece fa la macchina, a ripetere tutte le parole dell’originale e quindi 2) seleziona quelle essenziali e tralascia quelle ridondanti, agevolando così la leggibilità dei sottotitoli. Il secondo motivo si applica anche per giustificare la maggiore presenza di altre forme di riduzione nella sottotitolazione diretta, come la compressione nei sottotitoli prodotti dal respeaker. Un ulteriore risultato evidente riguarda il maggior numero di errori presenti nella modalità diretta rispetto a quella assistita, dovuti intuitivamente a una minore qualità della trascrizione proveniente dalla macchina. Questa non è chiaramente imputabile alla macchina, ma alla qualità del input: mentre un respeaker è abituato a dettare in maniera professionale alla macchina, il politico che viene trascritto è meno attento alla presenza del software di riconoscimento del parlato e quindi produce un testo meno chiaro sia foneticamente che grammaticalmente. Tuttavia, la differenza tra errori prodotti nella modalità assistita ed errori prodotti nella modalità diretta non è così sostanziale (0,1 per cento nel caso di errori che alterano l’unità concettuale e 0,3 per cento nel caso di errori che non alterano l’unità concettuale). Un dato che invece sembra controintuitivo riguarda le unità concettuali rese. Stante la seconda motivazione appena menzionata, l’ipotesi è che il respeaker produca sottotitoli maggiormente accurati. Se questo è vero in termini di qualità dell’input, i dati dimostrano che la seconda modalità garantisce una resa più alta delle unità concettuali rispetto alla seconda (98,1 per cento e 96,2 per cento rispettivamente). Questo perché il respeaker, come risultante della prima motivazione, non riesce a tenere sempre il passo dell’oratore non solo perché l’oratore parla più velocemente di quanto riesca il respeaker a dettare, ma anche per ragioni che pertengono a una bassa qualità del discorso da sottotitolare (in termini grammaticali e fonetici) e alla stanchezza del respeaker. Visto che ne consegue una riduzione quantitativa del testo iniziale, anche il numero di unità concettuali non rese risulta maggiore, sia che esse comportino una mancata resa dell’unità concettuale in questione (3,2 per cento in modalità assistita contro il 1,2 per cento nella modalità diretta) oppure no (6,1 per cento di riduzioni in modalità assistita contro il 4,6 per cento in diretta).

4.2.2. La qualità dell’interpretariato in LIS

Passando all’analisi dell’interpretariato in LIS, la tassonomia IRA è stata adattata in maniera da contemplare la differenza tra i sottotitoli e l’interpretariato (i primi sono intra-linguistici, cioè dall’italiano all’italiano, la seconda è interlinguistica). A tal fine, le ripetizioni sono state sostituite con le traduzioni complete, cioè traduzioni che non trascurano alcun elemento dell’originale, fatte salve le differenze grammaticali tra le due lingue (Figura 3). L’analisi mostra che la quantità delle unità concettuali rese è superiore al minimo richiesto (95,4 per cento) e che le unità concettuali rese sono state prodotte con un numero di traduzioni complete molto alto (88,1 per cento), conformemente alle linee guida Ofcom sulle ripetizioni.

Figura 3: Risultati dell’analisi della qualità dell’interpretariato LIS

Quanto alle alterazioni rispetto al testo di partenza, esse sono comprensibilmente inferiori rispetto ai sottotitoli (7,3 per cento) e sono composte perlopiù da riduzioni, vale a dire omissione di parole ridondanti o compressione di pensieri complessi. Assenti le correzioni e trascurabili, seppur in numero maggiore, gli errori che non influiscono sulla comprensibilità del segnato. Per quanto riguarda le unità concettuali che non sono passate nell’interpretariato (4,6 per cento), notiamo anche qui che esse sono principalmente costituite dalle omissioni dell’unità concettuale stessa. Anche in questo caso si tratta di unità che garantiscono la transizione da un concetto all’altro con una portata semantica marginale. La sola differenza con i sottotitoli è rappresentato dalla maggior presenza di errori, che potrebbero essere imputabili alla difficoltà di comprensione del testo dell’oratore, alla maggiore complessità dell’operazione di interpretariato rispetto a quella di ripetizione o ancora all’assenza di una strumentazione che permetta l’ascolto in cuffia da parte dell’interprete. Infine il ritardo registrato è di 0,9 secondi in media. Nel dettaglio, il ritardo è di 0,8 secondi quando il testo non presenta difficoltà e di 1,3 quando sono necessari interventi che riducono il testo originale.

4.2.3. La qualità dei sottotitoli interlinguistici

In uno studio simile condotto dalla Federazione Internazionale di Elaborazione dell’Informazione e della Comunicazione Intersteno, si è portato avanti il Communication Project (iniziato nel 2017, cfr. Eugeni et al. 2018), volto alla valutazione del migliore flusso di lavoro possibile in termini di immediatezza, economicità e accuratezza per la produzione di sottotitoli interlinguistici (da inglese a italiano o francese) in tempo reale (ILS) di riunioni di varia natura, quali conferenze, assemblee e consigli. Lo studio ha testato i seguenti flussi di lavoro di produzione di ILS nell’ambito del Congresso Intersteno 2019:

1. un interprete traduce la riunione in un’altra lingua e uno stenotipista trascrive (flusso di lavoro Human-Only);
2. un respeaker interlinguistico produce i sottotitoli direttamente in lingua straniera (flusso di lavoro Human-Only);
3. un professionista trascrive la riunione parola per parola e una macchina traduce (flusso di lavoro Computer-Aided);
4. un professionista intralinguistico trascrive la riunione semplificando la sintassi e una macchina traduce (flusso di lavoro Computer-Aided);
5. un software ASR trascrive la riunione e un live editor corregge eventuali errori (flusso di lavoro Human-Aided).

Si noti che il flusso ILS 3 era stato già testato nel 2017 (Manetti 2018) in un contesto simile. A variare, in questo caso, sono il professionista (nel Communication Project – ILS 3a – un respeaker, nell’altro caso – ILS 3b – un velotipista), la coppia linguistica dei sottotitoli valutata (dall’inglese in italiano in ILS 3a e dall’inglese in francese in ILS 3b), la qualità del software di traduzione automatica intuitivamente superiore in ILS 3a e le variabili testate (in ILS 3b solo il numero di parole al minuto e la qualità). Per ognuno di questi cinque flussi, sono state testate 4 variabili:

• fedeltà relativa dei sottotitoli misurata in numero di parole prodotte rispetto a quello originale (WORDS);
• velocità di scorrimento dei sottotitoli misurata in numero di parole al minuto (WPM);
• accuratezza dei sottotitoli misurata in numero di idee concettuali rese (IRA);
• ritardo dei sottotitoli misurati in secondi tra l’occorrenza di un’idea concettuale e la comparsa a schermo del relativo sottotitolo (DELAY).

Come mostra la Tabella 7, a un primo sguardo risulta chiaro che il flusso più fedele al testo di partenza è ILS 5 (è anche il solo flusso Human-Aided testato); in termini di velocità di lettura, che dipende dal numero di parole al minuto del discorso originale; il flusso che produce il maggior numero di parole e anche una qualità dei sottotitoli minore è ILS 3b, Computer-Aided; mentre quello più facile da leggere perché produce meno parole al minuto è ILS 1, che è Human-Only e anche il più accurato; il flusso con il ritardo minore è ILS 3a, altro flusso Human-Aided. In altre parole, in termini assoluti e senza distinzione tra i vari flussi di lavoro, quelli Human-Only sono più accurati ma anche meno fedeli, quelli Computer-Aided sono più rapidi ma anche meno accurati e quelli Human-Aided sono più fedeli ma anche meno rapidi.

Tabella 7: Valutazione della qualità dei sottotitoli interlinguistici

Se si guardano i dati un po’ più nel dettaglio, ci si può rendere conto che i flussi Human-Only (ILS 1 e ILS 2) sono gli unici due flussi che garantiscono un’accuratezza professionalmente accettabile (97,3 per cento e 95,8 per cento). Per farlo, però, hanno bisogno di un importante lavoro editoriale a opera dei professionisti coinvolti che devono possedere competenze di interpretariato e di sottotitolazione in tempo reale. Questo può risultare nella produzione di un numero di parole di circa il 20 per cento inferiore rispetto a quelle pronunciate del testo originale (79,6 per cento e 82,2 per cento) e quindi una velocità di lettura del sottotitolo inferiore (110 e 114 wpm). Se, da un lato, questo garantisce una maggiore leggibilità del sottotitolo, dall’altro può anche creare dissonanza cognitiva negli utenti. Da segnalare è anche il ritardo maggiore per ILS 1 (4,3”). Tuttavia, se un solo professionista (in ILS2 un respeaker interlinguistico) si occupa sia della traduzione sia della trascrizione, il ritardo si riduce sostanzialmente (1,8”). Un aspetto da considerare, non misurato nella Tabella 7, è l’investimento da parte dell’eventuale cliente, che oltre a una squadra di professionisti per turno, deve anche considerarne una squadra per ogni coppia linguistica di cui dovesse avere bisogno, perché i flussi Human-Only sono chiaramente language dependent. Tra i due flussi, l’uso di un respeaker interlinguistico risulta chiaramente più conveniente per via del numero minore di professionisti coinvolti, che comporta meno ritardo e meno spesa.

Per quanto riguarda i flussi Computer-Aided (ILS 3a, ILS 3b e ILS 4), i dati che saranno qui analizzati riguardano solo ILS 3a e ILS 4, dato che ILS 3b (71,2 per cento) è stato testato in un periodo in cui le tecnologie della trascrizione e della traduzione erano intuitivamente meno avanzate. Essi garantiscono un’accuratezza comunque superiore al 90 per cento (91,6 per cento e 92,1 per cento). Questa qualità, comunque accettabile anche se inferiore al 95 per cento stabilita come soglia minima nella sottotitolazione intralinguistica misurata con IRA, comporta un intervento editoriale minore rispetto ai flussi Human-Only (87,7 per cento e 86 per cento) e quindi una velocità di lettura del sottotitolo superiore, anche se di poco (121 e 118 wpm). Questo crea comunque una buona leggibilità del sottotitolo (stabilita da Rai tra le 120 e le 180 parole al minuto[10]) e un’accuratezza superiore a quella dei flussi Human-only (87,7 per cento e 86 per cento). Quanto al ritardo, ILS 3a (e intuitivamente anche ILS 3b) garantisce il ritardo più basso (1,3”) mentre ILS 4 implica un ritardo maggiore, probabilmente dovuto al fatto che il sottotitolatore non si limita a ripetere il testo di partenza (come in ILS 3a e 3b), ma deve pensare a semplificare il testo di arrivo. Per questo motivo, ci si sarebbe aspettati una produzione di parole rispetto al testo di partenza, e quindi anche di parole al minuto, di molto inferiore a quelle prodotte da ILS 3a. Se la velocità di produzione del testo di partenza può spiegare differenze nella compressione o nel numero di parole al minuto, essa non può giustificare entrambe. Quindi, se risulta abbastanza chiaro che il respeaker in ILS4 non abbia operato una semplificazione del testo di partenza maggiore rispetto a quanto fatto dal respeaker in ILS 3a, le ragioni potrebbero essere molteplici. Tra le più probabili, il respeaker non è riuscito a semplificare il testo di partenza e il testo di partenza non aveva bisogno di ulteriore semplificazione. In questo specifico ambito può risultare utile dare uno sguardo ai dati relativi a ILS 3b che mostrano chiaramente che il testo di arrivo aveva una velocità di produzione di molto superiore a quella degli altri flussi (141 wpm). In termini di spesa, i flussi Computer-Aided somigliano a ILS 2 per economicità ma garantisce una maggiore scalabilità del servizio, dato che non serve una squadra per ogni coppia linguistica di cui dovesse avere bisogno, perché i flussi Computer-Aided sono language independent.

Quanto al solo flusso Human-Aided analizzato (ILS 5), esso garantisce la minore accuratezza di tutti gli altri flussi. Tuttavia, se si guardano i dati di ILS 4, si nota che la qualità, in questo caso è notevolmente superiore, dato particolarmente incoraggiante se si considera che la distanza le variabili tra i due sono non solo la distanza di due anni tra un esperimento e l’altro (2017 e 2019), ma anche il flusso di lavoro, che in ILS 5 non prevede una gestione del flusso stesso da parte del professionista. Un ulteriore dato evidente riguarda la fedeltà del testo di arrivo rispetto a quello di partenza (95,5 per cento). Incrociando questo dato con il precedente, risulta chiaro che l’automazione, da una parte implica una trascrizione e una traduzione del 100 per cento del testo di partenza; e dall’altra implica una serie di errori che vengono corretti o eliminati dal live editor/scopist. Quest’ultimo dato viene anche confermato dall’alto tasso di fedeltà relativa (95,5 per cento) e dal ritardo con cui i sottotitoli vengono rilasciati (5,1”). In un contesto professionale, i sottotitoli prodotti in modalità Human-Aided sono anche intuitivamente migliori di sottotitoli prodotti in modalità Computer-Only, nonostante interessanti sviluppi anche in questo ambito (Romero-Fresco 2015). A far sorgere alcune perplessità sono la qualità dell’input, che può comportare diverse rese qualitative (Pagano 2020), e la conseguente impossibilità del professionista di gestire il flusso di lavoro. In termini economici, se la spessa per i software risulta ammortizzabile sul lungo periodo, la modalità Human-Aided richiede un numero di squadre di sottotitolazione pari a quello delle coppie linguistiche richieste.

4.3. Consultazione pubblica

Dalla tavola rotonda organizzata con il gruppo di esperti e con i rappresentanti dell’utenza sorda, è emerso che il servizio di sottotitolazione e di interpretariato in LIS soddisfa i criteri tecnici previsti dal bando e i tecnici di qualità previsti dall’Ofcom. Tuttavia, è stata proposta una serie di migliorie per aumentare la qualità tecnica dei sottotitoli e dell’interpretariato, così come la fruibilità degli stessi da parte dell’utenza finale. Per quanto riguarda l’interpretariato, si è raccomandato un ingrandimento del riquadro contenente l’interprete LIS e uno sfondo che garantisca una maggiore visibilità dei segni. Inoltre è stato consigliato l’uso delle cuffie da parte degli interpreti per meglio sentire l’originale. Quanto ai sottotitoli, è stato raccomandato l’uso di tre righe per garantire una maggiore leggibilità delle frasi (spesso lunghe) in essi contenute. Tra i desiderata la creazione di un’applicazione per poter seguire più agevolmente le sedute tramite cellulare, che personalizzi il servizio di accessibilità con un’opzione che permetta di scegliere tra interpretariato LIS, sottotitoli, diretta e una combinazione delle tre. L’applicazione dovrebbe anche offrire la possibilità di regolare il ritardo dei sottotitoli e dell’interpretato in LIS in maniera da garantire una maggior sincronizzazione tra il servizio di accessibilità e la diretta streaming. Un’ultima raccomandazione ha riguardato la possibilità di riutilizzare i sottotitoli integrandoli nel processo di resocontazione (§5).

5. Interazione Uomo-Macchina nella resocontazione

Capitalizzando sul concetto di progettazione universale e, tra le altre, la raccomandazione sul riutilizzo dei sottotitoli in tempo reale per la produzione di resoconti, la piattaforma utilizzata per la resa accessibile delle sedute capitoline può essere impiegata in un flusso di lavoro ideale che si rifà al concetto di verbale multimediale, proposto in Italia in seno al gruppo di ricerca del Governo ForumTAL,[11] con l’obiettivo di digitalizzare al massimo il processo penale e garantire così una Giustizia più rapida ed efficacie. Il verbale multimediale crea un flusso di lavoro in cui la verbalizzazione passa da Computer-Aided a Human-Aided e il verbale da oggetto ultimo della verbalizzazione a flessibile strumento del processo penale stesso (Figura 4).  

Figura 4: Flusso di lavoro per un resoconto multimediale

Il flusso di lavoro del resoconto multimediale parte con la registrazione multitraccia della seduta simultaneamente al suo svolgimento (fase A del flusso di lavoro). Oltre alla registrazione, la piattaforma che gestisce il flusso di lavoro che porta - tra gli altri servizi - al resoconto multimediale, procede anche all’elaborazione dei dati in ingresso e alla loro trascrizione automatica tramite software di riconoscimento del parlato (fase B). L’elaborazione dei dati serve da riferimento per la sottotitolazione che passa per un ulteriore postazione di lavoro. Quest’ultima permette alla squadra di sottotitolazione di scegliere tra una delle due modalità di produzione dei sottotitoli summenzionate (assistita o diretta) l’uso diretto della trascrizione prodotta nella fase B o l’inserimento dei sottotitoli da parte del respeaker nel caso in cui la qualità della trascrizione non permettesse una facile e accurata impaginazione dei sottotitoli in tempo reale da parte del live editor (fase C). Queste tre fasi corrono parallele all’interpretariato in LIS (fase D) che viene registrato in un’altra traccia rispetto a quella dei sottotitoli, entrambi trasmessi nella stessa traccia o in tracce diverse rispetto a quella della seduta da rendere accessibile.

Dalle fasi di produzione si arriva alla ricezione dei singoli servizi offerti da parte degli utenti finali, che possono scegliere se attivarli in combinazione con altri servizi:

• i segnanti avranno la possibilità di accedere all’interpretariato in LIS prodotto nella fase D simultaneamente all’incontro, con o senza il fisiologico ritardo tra i due;
• gli oralisti, sordi e non, avranno la possibilità di accedere
  • alla trascrizione automatica prodotta dalla macchina nella fase B,
  • ai sottotitoli intralinguistici prodotti in modalità respeaking diretta o automatica in fase C simultaneamente all’incontro, con o senza il fisiologico ritardo tra i due,
  • ai sottotitoli interlinguistici prodotti automaticamente dopo la modalità diretta o assistita della fase C e simultaneamente all’incontro, con o senza il fisiologico ritardo tra i due, 
  • al resoconto multimediale, contenente il video della seduta e il resoconto, le cui sezioni e parole sono state corrette e indicizzate e sincronizzate con il video stesso.

Conclusioni

L’interazione uomo-macchina nell’ambito della traduzione diamesica ha portato a importanti distinzioni dei flussi di lavoro in Human-Only, Computer-Aided, Human-Aided Translation o Computer-Only. Grazie all’intelligenza artificiale, le soluzioni Computer-Only un tempo neanche considerate, sono già realtà in contesti comunicativi anche molto importanti come le istituzioni europee. Nel tentativo di ricostruire le tappe che hanno portato a questo risultato e per comprendere l’impatto della tecnologia sulla vita di tutti i giorni e il suo potenziale sviluppo, il presente studio di ricerca ha illustrato alcuni progetti condotti in contesti principalmente italofoni. Nell’analizzare e raffrontare sottotitoli preregistrati e in tempo reale, sia intralinguistici che interlinguistici, prodotti principalmente o esclusivamente dall’uomo o dalla macchina, sono emersi interessanti spunti di riflessione non soltanto sulla rapida evoluzione della tecnologia e della sua applicazione nella traduzione diamesica, ma anche in ottica di contributo delle tecnologie assistive alla evoluzione della società nel suo complesso, in una specie di processo inverso rispetto a quanto si è assistito negli ultimi decenni. In particolare, una piattaforma multifunzionale destinata all’accessibilità che riesce anche a garantire rapidità, accuratezza e trasparenza dei processi democratici e legislativi di un Paese è la dimostrazione di quanto l’accessibilità non possa più essere considerata una questione marginale della società, ma intrinsecamente parte della stessa.

Abstract

English:

[abstract and keywords in English required]

Spanish:

Cada vez son más los contenidos que se hacen accesibles para distintos públicos con diferentes necesidades. En particular, centrándonos en la accesibilidad a la imagen, en esta investigación nos ocuparemos de la audiodescripción (AD) aplicada al cómic o novela gráfica, tipología poco explorada hasta la fecha. Como texto multimodal (Taylor 2016), el cómic plantea numerosos interrogantes para la realización de una traducción accesible destinada a personas que por distintas razones no pueden acceder a la imagen. En este trabajo nos proponemos estudiar dicha problemática a partir de dos propuestas y desde la perspectiva de la audiodescripción, cuyo objetivo es hacer ‘visible’ los distintos códigos que componen esta forma expresiva artística, en particular, el contenido icónico (Kress y van Leeuwen, 2002, 2006), es decir, las imágenes junto con el texto escrito en sus distintas dimensiones (narración y símbolos gráficos). A través de un breve recorrido caracterizante de este tipo de obras y de los códigos semióticos implicados y a la luz de las propuestas que se presentan, se avanzará un posible itinerario para la realización de la versión accesible.

Keywords: texto multimodal, cómic, traducción intersemiótica, audiodescripción, accesibilidad

1. Introducción

La audiodescripción (AD) surge en la segunda mitad del siglo XX en EE.UU. para dar respuesta al deseo, por parte de personas con discapacidad visual, de disfrutar de espectáculos teatrales. Posteriormente se desarrolla en Europa en el sector cinematográfico y este hecho la enmarca en el ámbito de la Traducción Audiovisual (TAV) (Gambier 2004, Díaz Cintas 2005, Matamala y Orero 2007). Sin embargo, teniendo en cuenta que esta modalidad de la traducción se ha ido expandiendo a otros contenidos (productos cinematográficos, museos, arte en general, parques, espacios al aire libre, sitios arqueológicos, recorridos turísticos, entre otros), excede ese marco por lo que el área que mejor puede dar cuenta de ella es la que podemos denominar Traducción Intersemiótica Accesible, que veremos a continuación.

A partir de un estudio sobre los distintos elementos que componen el texto multimodal del cómic (contenido icónico y textual), se analizarán dos propuestas sobre la audiodescripción aplicada a este tipo de obras, así mismo se mencionarán otras experimentaciones que van en esa dirección. Se formularán algunas pautas que puedan servir de base para una posible versión accesible así como vías de fruición factibles.

2. La audiodescripción

Queda definida en la guía oficial española (AENOR 2005: 3-4) como “servicio de apoyo a la comunicación que consiste en […] compensar la carencia de captación de la parte visual […] suministrando una adecuada información sonora” de obras audiovisuales, así como “espectáculos en directo […], monumentos […], museos y exposiciones, entornos naturales y espacios temáticos […]“ (2005: 3). Anna Matamala y Pilar Orero (2013: 150) matizan ese horizonte indicando “dynamic or static, i.e. from a guided city tour of Barcelona or a 3D film, to a Picasso painting”.  Por otro lado, Jiménez Hurtado y Seibel (2008: 452) se refieren a la audiodescripción como ‘técnica’ y también como ‘proceso de traducción’. Específicamente la denominan “traducción audiovisual accesible”. Tal y como se la ha situado desde sus inicios, es decir, como modalidad de la Traducción Audiovisual (Díaz Cintas 2005, Matamala y Orero 2007), la audiodescripción responde, siguiendo la taxonomía jakobsoniana (1959), al tipo denominado traducción intersemiótica, pues se trata, en el caso que nos ocupa, de una operación que consiste en formular verbalmente un contenido icónico, es decir, expresar mediante un sistema de signos lingüístico (palabras) lo que se presenta a través de un sistema de signos distinto (imágenes en nuestro caso), de ahí la denominación ‘Traducción Intersemiótica Accesible’. Se hace necesaria esta nueva línea de investigación que expande su acción más allá del marco de la TAV o del llamado Media Access (Greco 2018), ya que se ocupa de contenidos de todo tipo, desde entornos naturales (parques, playas, por citar dos ejemplos), arquitectura, arqueología, recorridos turísticos, en definitiva, cualquier contenido que se perciba a través de la vista. La naturaleza intersemiótica de estos contenidos se manifiesta precisamente por la operación de traducción de sistemas de significación de distinto tipo mediante signos verbales. Por otro lado, con el término ‘accesible’ se subraya la necesidad de que tales contenidos se describan pensando en un destinatario concreto: ciego o con baja visión, por lo tanto, proporcionando la información visual a la que no tiene acceso.

Esta modalidad emergente de la traducción va a cumplir casi tres cuartos de siglo desde su nacimiento al otro lado del Atlántico, pero cuya expansión se ha producido en Europa y Reino Unido, como hemos mencionado más arriba. El principal usuario, o más bien, el más inmediato es un público con distintos grados de discapacidad visual, innata o adquirida. El objetivo de esta tipología de traducción es hacer posible que dicho colectivo pueda tener acceso a contenidos tanto informativos como lúdicos o formativos que utilizan la imagen (Díaz Cintas, 2005), y en este caso al entrelazado binomio imagen-texto, para comunicar.

3. El cómic o historieta

Como el producto fílmico, el cómic es asimismo un texto multimodal puesto que se presenta como un entramado de imagen y texto escrito donde este último aparece bien en el interior de globos (regulares o irregulares -con picos o formas asimétricas-) que dan voz o pensamiento a un personaje, bien fuera del cuadro como voz en off del narrador. La imagen fija presentada a través de un orden secuencial permite percibir el movimiento. Se trata de una característica importante de este tipo de texto que, como en una película, hace posible el desarrollo de la historia. Dicho orden, en la cultura occidental, sigue normalmente una dirección horizontal de izquierda a derecha y de arriba abajo, aunque, por supuesto, pueden darse alteraciones a esta práctica. La narración, pues, se presenta a través de una estructura de acontecimientos concatenados.

Sobre la existencia de un predominio de imagen o texto, hay autores (Vilches 1984, Borodo 2015, Brandon 2018) que consideran que la imagen es prevalente, se mira antes de leer, la lectura del texto escrito se produciría en un segundo momento; mientras que para otros (Eisner 2008, Yuste Frías 2011) los dos medios de expresión tendrían el mismo peso o igual consideración, por lo tanto ninguno estaría restringiendo o limitando al otro, lo que significa que en la traducción no se puede prescindir o no debe limitarse a uno solo de estos códigos, es decir, al lingüístico, como normalmente sucede (Rodríguez Rodríguez 2019). Podemos preguntarnos entonces qué implicaciones tienen estas consideraciones si pensamos en su traducción, como afirma Federico Zanettin (2008:12):

The translation of comics into another language is primarily their translation into another visual culture, so that not only are different natural languages such as English, Japanese, Italian or French involved, but also different cultural traditions and different sets of conventions for comics. In other words, the translation of comics does not only imply the interlinguistic (or intralinguistic) replacement of verbal material. Comics published in other languages may also undergo a number of changes which involve the interpretation of other sign systems, not just ‘translation proper’ between natural languages […].  

Por lo tanto, el traductor, en su tarea, debe ser consciente del hecho de que el significado surge de esa interrelación entre la imagen y la palabra, consideración esencial para la realización de una AD.

3.1 Códigos semióticos del cómic

En este género narrativo se pueden distinguir fundamentalmente, como hemos mencionado, dos códigos: icónico y lingüístico. Los dos modos, verbal y visual, interactúan contribuyendo por igual en la creación de significado.

Michał Borodo (2015: 23 y 24) alude a un estudio de Radan Martinec y Andrew Salway (2005) en el que explican tres formas en que los dos modos semióticos pueden relacionarse. En el primero, el texto haría referencia solo a ciertos aspectos ya presentes en la imagen, por lo que se produce una repetición; en el segundo, el texto iría más allá de la imagen añadiendo información y, en el tercero, el texto se expande en la imagen proporcionando información que tiene que ver con las relaciones espaciales y temporales. En otras palabras, en el primer caso, la información llega al destinatario a través de los dos canales: textual y visual. Se trata de un aspecto relevante para los autores (guionista y dibujante) puesto que los dos insisten en un determinado elemento, trabajan en la misma dirección. Si pensamos en la AD, este elemento debería quedar plasmado, ser ‘visible’, en nuestra opinión porque se respetaría la intencionalidad del autor. En la segunda y tercera forma, se añade información, por lo que ambos modos semióticos se complementan para crear el significado. Este sería el caso que presentamos a continuación, donde la imagen se entreteje con las palabras ofreciendo mayor información. Se trata de la versión en cómic de la obra de Mary Shelley realizada por Lagartos e Iglesias[1]. En la viñeta en cuestión, que ocupa toda una página, asistimos a un momento de la célebre reunión literaria en la que Byron les propone a Mary Shelley y John Polidori que escriban una ‘historia espantosa’. Las sombras que proyectan tanto Shelley como Polidori anticipan las obras que posteriormente escribirán, es decir, Frankenstein y El vampiro.

[image quality too low]

Figura 1: Mary y Frankenstein / John y el vampiro. Fuente. Elaboración propia

A esta forma de ‘narrar’ mediante la imagen, Pablo De Santis (1988:11) la denomina ‘silenciosa’, donde es el dibujante y no el guionista quien ‘habla’ a través de la figura representada.  Estamos pues ante una ilustración que transmite un contenido cultural que debe recogerse en la AD. Si bien es cierto que se trata de obras mundialmente conocidas, no queda tan claro que un usuario medio recuerde o reconozca a sus autores, por lo que resulta útil también para un destinatario en general, con o sin problemas de vista. En este caso no hablamos de intertextualidad puesto que la referencia visual no es a otro texto ya existente, sino precisamente a las obras que Shelley y Polidori escribirán. Se trata más bien de un guiño a un lector distraído.

Siguiendo con los elementos que caracterizan la escritura de un cómic, además de la tipografía (trazo, tamaño y tipo de carácter que se utilice), destaca el uso de los colores que, como explican Liber Cuñarro y José Enrique Finol (2013: 283) “han estado vinculados a distintas emociones y procesos psicológicos”, aunque desde un punto de vista sociocultural no hay acuerdo en lo que connotan.  En este sentido, conviene prestar atención, como exponen Gunther Kress y Theo Van Leeuwen (2006: 3), al hecho de que nos encontramos ante un producto que presenta valores culturales específicos: “Visual language is not –despite assumptions to the contrary – transparent and universally understood; it is culturally specific” y que por lo tanto nuestra comprensión se ve limitada por nuestro horizonte de conocimiento. Es lícito, pues, interrogarse por nuestra capacidad, como lectores, para interpretar adecuadamente dichos códigos, sobre todo el visual para el que no recibimos instrucción[2], muy al contrario de lo que sucede con el verbal para el que se nos alfabetiza desde nuestra tierna infancia y que nos permite leer y escribir. 

Otro elemento importante de la comunicación, presente en el intercambio visual, es el lenguaje no verbal que nos transmite una información esencial para comprender el estado emocional de los personajes, como afirma Will Eisner (2008: 103): “el cuerpo humano, la estilización de su forma, y la codificación de los gestos y posiciones producidos por las emociones son acumulados y archivados en la memoria, conformando un vocabulario no verbal de gestos”. Por lo tanto, en la comprensión de los personajes, importa tanto lo que dicen (su expresión lingüística con todas su variantes) como lo que comunican a través del lenguaje no verbal. A estos, hay que añadir la manera en que expresan lo que quieren decir (mediante un grito, pausadamente, etc.), en otras palabras, los rasgos paralingüísticos que acompañan a su enunciación y que en el cómic se denominan marcas fonografológicas (Sinagra 2014: 90-94, cit. en Rodríguez Rodríguez 2019: 39), estas:

representan distintos volúmenes de voz (gritos, susurros, etc.) o la velocidad de dicción (rapidez o lentitud). La tipografía se agranda, se alarga, cambia de color y de aspecto —tanto dentro como fuera del bocadillo— con objeto de que el lector “oiga” el sonido. Las mayúsculas y las negritas cumplen unas funciones similares. Los signos de puntuación atribuyen un “énfasis de la oralidad”, poniendo de manifiesto la ira, el asombro o la perplejidad de los personajes. Los puntos suspensivos —estrechamente vinculados con las marcas lingüísticas de oralidad — se emplean para indicar las dubitaciones, las frases inconclusas, las interrupciones, etc.

Por otro lado, para la información espacial referida al lugar que ocupan los personajes en la viñeta, es decir, parte de la puesta en escena, Borodo (2015: 24) alude a un estudio de Jens Allwood (2002) en el que explicita su importancia junto con los gestos a los que acabamos de aludir, ya que completan la comunicación verbal y se identifican en gran parte con determinadas culturas:

comic book characters do not merely interact and communicate meanings through speech balloons but, equally importantly, through gesture, posture, eye gaze or facial expression. Protagonists are also positioned in specific ways within panels and both their positioning and body movements may be revealing with regard to the nature of their relationship, be it friendliness, indifference, superiority, suspicion, tension, irritation or hostility. These nonverbal interactions are thus a crucial component of communication, possessing a considerable meaning-making potential activated by the reader/viewer in the process of interpreting comic book panels.

Por último, hay que añadir que, como en el producto cinematográfico, en esa ‘puesta en escena’ observaremos su contenido a través de una serie de planos o encuadres que nos proporcionarán información, como por ejemplo los primeros planos para transmitir dichas emociones y la secuenciación de las viñetas (su montaje) que nos dará indicaciones sobre el desarrollo de la historia, entre otros[3]. En definitiva, todo un reto para el traductor.

4. AD y cómic

Un intento de creación de una obra accesible la encontramos en El libro negro de los colores, de Menena Cottin y Rosanna Faria (ilustradora), publicado en 2006. Presentan las ilustraciones en relieve y asocian los colores a otros sentidos como el olfato y el gusto. Si bien es un intento de acercar el texto a personas con baja visión, lo cierto es que no se trata de un verdadero texto accesible puesto que no hay descripción del contenido icónico, pero utiliza otros recursos interesantes, como la sinestesia, para dar cuenta de distintos códigos. Constituye un punto de reflexión sobre las sugerencias que plantea e integración en una posible versión con AD.

Por lo que se refiere a la audiodescripción y su aplicación al cómic, Rachel Osolen y Leah Brochu (2020) de la National Network for Equitable Library Service (NNEL) con sede en Canadá, han llevado a cabo una experimentación teniendo en cuenta la escritura del cómic. Estos autores proponen una serie de líneas guía basándose en su conocimiento de la AD aplicada al cine fundamentalmente. Siguiendo otra experimentación realizada por Liana Kerr en 2015 (Osolen y Brochu 2020: 112), añaden algunas pautas más. Finalizan su presentación con un estudio de recepción en el que participó un único informante, por lo que el resultado se quedó en lo anecdótico.

En este artículo presentamos dos iniciativas. La primera es una propuesta en línea de Teri Grossman[4] “Techniques for Describing Graphic Novels”[5] (s.f.). La segunda, de Brandon (2018)[6].

Basándonos fundamentalmente en estas dos manifestaciones, nos hemos planteado cómo podría ser una versión accesible de este género.

La AD, en su aplicación al cómic, representa una temática que empieza a despertar el interés de la comunidad científica. De hecho, ha sido la materia de una presentación en un congreso, concretamente de Gian Maria Greco: “On the audio description of comic books” (2017) de cuya publicación no tenemos noticia.

4.1 Propuesta de Christopher Brandon (2018)

En su estudio, después de pasar revista a una serie de publicaciones sobre accesibilidad y la aplicación de la AD a museos donde se exponen prácticas que sugieren distintos recorridos tanto táctiles como auditivos, se pregunta por la eficacia de estas iniciativas. En efecto, sin un análisis semiótico de los componentes característicos del texto origen (sea una obra pictórica, escultura o un cómic, como es nuestro caso) y su estudio de recepción nos quedaremos en una lectura superficial y, consiguientemente, la audiodescripción no podrá cumplir su objetivo o quizá solo parcialmente.

Brandon pone en relación esta tipología de traducción, es decir, la que se está llevando a cabo en museos para una posible aplicación a este género gráfico. En su opinión, el predominio del código visual del cómic no impide su transmisión a través del canal audio para un destinatario con discapacidad visual. En su artículo, presenta dos propuestas con planteamientos diferentes. La primera es un intento de traducir el cómic Daredevil[7] (2011) a través del canal audio. La segunda, Life[8] (2013) o la vida de Philipp Meyer, es un cómic diseñado en origen para que sea accesible tanto a lectores con problemas de visión como sin ella. El autor sugiere que la dependencia de lo visual es más arbitraria de lo que se piensa.

En el primer caso que presenta, habla de un cómic en el que el protagonista es ciego y esta es la justificación o la razón de haber desarrollado otros sentidos como el tacto o el oído a modo de superpoderes. Brandon explica que va a establecer una diferencia entre lo que llama cómic auditivo ‘aural comic’, es decir, entre la versión audio y la versión impresa. De hecho, presta más atención a la dimensión verbal, textual y casi ninguna a la icónica. En consecuencia, eliminando casi por completo la imagen, el texto audio no diferiría de lo que podría escucharse referido a una novela sin imágenes. Operando de esta manera se pierde uno de los modos que identifican al cómic como texto multimodal, afecta propiamente a su naturaleza. En conclusión, la AD en este caso está ausente, puesto que en realidad se transmite solo el contenido textual a través del canal audio, del mismo modo que podría hacerse mediante el sistema braille.

En el segundo ejemplo asistimos a la vida de Philipp Meyer, una historia titulada Life. Se trata de un cómic táctil pensado y diseñado desde su proceso creativo para todos los públicos, independientemente de su condición visual. El texto consta de seis páginas donde se presenta una historia mediante el uso de puntos en relieve similares a los que usa el braille. En dicha narración se cuenta sucintamente la vida de una persona: nacimiento, vida adulta, muerte. Ese recorrido se lleva a cabo, como hemos mencionado, a través de puntos en relieve en círculos de distinto tamaño. La propuesta demuestra que se puede prescindir de la imagen. Sin embargo, este es un caso muy especial y la realidad es que, excluyendo este tipo de prácticas, la casi totalidad de la novela gráfica contiene, en su configuración y por naturaleza, imagen. Por lo tanto, no sería aconsejable o útil pensar en un cómic sin contenido icónico.

4.2 Propuesta de Grossman

La autora presenta una iniciativa de AD de un cómic o novela gráfica en sentido tradicional, es decir, de un texto que responde a las convenciones de la novela gráfica, no se trata de un texto en el que de alguna manera ya está presente o pensada con antelación la discapacidad, como en los casos anteriores que Brandon o Cottin y Faria presentan.

El objetivo de Teri Grossman es hacer realidad la accesibilidad a ese contenido por parte de personas con discapacidad visual. En concreto, está dirigido a estudiantes, la autora declara que su intención es la de “[…] ensure that students who are blind or visually impaired have a comparable experience reading their materials as their sighted classmates¨.

En líneas generales, podemos afirmar que en la elaboración de la AD, la autora sigue los criterios de sobra conocidos para la AD de productos fílmicos. En especial, hay que tener cuidado, como explica Grossman a propósito de lo que denomina General Principles of Description con:

Do not let the description interrupt the story. Description is intended to complement the story not to interfere or to take its place. The describer decides where in each panel the description will go in relation to the narration or the speakers. Sometimes it is places before sometimes after and sometimes in between this [sic] elements.

A continuación mostramos una viñeta del cómic propuesto[9] con AD. En la parte izquierda se sitúan las imágenes y a la derecha el texto: por un lado el que acompaña a la viñeta (Nar:) y por otro el que corresponde a la AD (Des:).

[image quality too low]

Figura 2. Prisoner on the Hell Planet (Art Spiegelman, 1972). Fuente: Elaboración propia

5. Posibles fases para la AD aplicada al cómic

Siguiendo por un lado la exposición de la autora, Grossman y por otro, basándonos en las etapas para la elaboración de una AD para productos audiovisuales (Valero Gisbert, 2021), a continuación reelaboramos y ampliamos esos principios proponiendo las siguientes fases o etapas para la elaboración de un cómic con AD:

1. Documentación sobre el autor y la obra.
2. Lectura del cómic.
3. Análisis semiótico de la obra.
4. Toma de notas sobre lo que sucede, identificación de los protagonistas, los lugares en los que se desarrolla la historia y los acontecimientos que narra.
5. Descripción de la portada y la contraportada de la publicación, dedicatoria si la hay, agradecimientos, y otras informaciones que pudiera contener el aparato paratextual. Se correspondería con los títulos de apertura y cierre en un producto cinematográfico.
6. Describir el marco de cada página y el título si lo hay.
7. Anotaciones del descriptor: en las que se explica cómo ha operado, concretamente:
   - los paneles de la página, número, si varían, si tienen medidas estándares;
   - tipografía: cómo son las letras del texto: mayúsculas, el tipo de carácter, tamaño;
   - qué espacio ocupan: si el texto está en bocadillos, de qué tipo: rectangulares, redondos.
   Bocadillos de pensamiento representado a través de círculos que salen del personaje en una escala de menor a mayor, por ejemplo. Si se trata de un texto que se susurra/ se grita/ se interrumpe y sigue en el panel o viñeta siguiente. Indicación del número de página: si precede al texto, o si se interrumpe el parágrafo.
   -información sobre el código acústico representado (referido a la enunciación: tono de la voz, ritmo, etc., así como a los sonidos en general: ruidos de distinto tipo).
8. Obra de arte: describir la ilustración. Código cromático: la historia está ilustrada en color/blanco y negro.
9. Código espacial: puesta en escena y puesta en cuadro: personajes y objetos, la posición que ocupan en la viñeta, el punto de vista y cómo se muestra.
10. La descripción, en Occidente, sigue recuadros de arriba a abajo de izquierda a derecha, a menos que el autor haya alterado ese desarrollo.
11. No es necesario describir cada viñeta obligatoriamente, si son similares, se aconseja dejar avanzar la historia sin interrupciones.

Como hemos mencionado, es importante la etapa de documentación con información de distinto tipo sobre el cómic, el autor, el tipo de léxico que se empleará en función del contenido de la historia, así como la elaboración de un glosario específico. Esta fase puede enmarcarse en lo que llamamos macroestructura. En la microestructura se procederá al análisis semiótico de los códigos empleados ya que nos servirá para comprender el significado y aislar los elementos relevantes.

En la fase de escritura del guion audiodescrito, desde un punto de vista lingüístico, brevemente recordamos que se recomienda el uso del presente, frases cortas formadas por sujeto/verbo/predicado, por lo tanto se aconseja evitar perífrasis u otros rodeos lingüísticos. Asimismo, se recomienda prestar atención al uso de artículos determinados y/o indeterminados para presentar elementos nuevos. En cuanto a su identificación, se aconseja denominarlos siempre del mismo modo puesto que ayuda a identificarlos y relacionarlos como objetos o personas que han aparecido con anterioridad. Por lo que se refiere al léxico, la precisión y concisión son dos nociones claves. La frase breve y con sentido completo se comprende más fácilmente. A ello contribuye la elección de un vocabulario apropiado, evitando cultismos o términos demasiado especializados.

En lo que respecta a la identificación de los personajes, como en los productos fílmicos, se prestará atención al momento en que podemos llamarlos por su nombre, sea porque alguien los ha nombrado, sea porque se han presentado ellos mismos, o por cualquier otro modo de haberlos reconocido.

Por lo que se refiere a la imagen, como se ha indicado, es necesario realizar un análisis semiótico para comprender lo que transmiten y valorar las prioridades. A modo de ejemplo, no podemos perder de vista el concepto de polisemia, que se aplica también a las imágenes y donde su significado queda determinado por el montaje, es decir, por la combinación de las distintas partes. 

Otros aspectos importantes que hay que considerar derivan de la presencia de intertextualidades, tanto icónicas como gráficas o sonoras, fenómeno que remite a la presencia de otras obras y autores de cualquier momento histórico, Estas cumplen una función específica en la construcción del sentido global de ese texto. Es fundamental identificarlas como fase del proceso traductor para, posteriormente y sobre la base del tipo del destinatario (conocimiento y edad) valorar la oportunidad de hacerlas visibles o no en la AD.

Llamamos la atención sobre un aspecto que señala Grossman y que trata de los sentimientos de los personajes. La autora explica como principios generales de AD que no se describan las emociones y lo justifica aduciendo que no se pueden ver: ¨Do not describe moods, motives or reasoning of the characters. These are unseen emotional states and are not described. Describe what you can see. Trust the reader to grasp the context and allow them to reach their own conclusions about the moods, motives and reasoning¨ (diapositive 16). Se trata de una afirmación de la que disentimos, pues como es práctica habitual en la AD para el cine, la puesta en cuadro en el cómic puede presentar el estado de ánimo de los personajes según el tipo de plano o encuadre que se emplee. Por lo tanto, se trata de un aspecto que en nuestra opinión debe describirse, de lo contrario se perdería una parte importante del sentido del texto.

6. Metodología de accesibilidad

Por lo que se refiere a la propuesta de Grossman, que es la única pensada realmente como producto con AD, no tenemos datos sobre el modo en que el usuario accederá a esta versión. A continuación, hemos hipotizado una posible manera: pensamos que esta propuesta de AD se realiza a través del canal audio donde se identifican la voz del narrador, la de los personajes y la de la AD, los sonidos se incorporan, creemos, a la pista audio. Suponemos que se usan distintas voces para distinguir cada intervención según a lo que corresponda: por un lado, las voces de los personajes, voz del narrador, voz del audiodescriptor, y por otro, los sonidos en general. 

Por lo que se refiere a las restricciones temporales tan importantes en la AD de productos fílmicos, en el cómic no tenemos esa limitación, sin embargo, eso no significa que la descripción pueda extenderse, no solo porque podría cansar, sino sobre todo porque el flujo narrativo se interrumpiría.

Lo que es evidente es que, a diferencia de una película, la expresión oral (los diálogos de los personajes) se percibe a través del texto escrito y por lo tanto constituye otro elemento que requiere una pista audio (que incluya los rasgos paralingüísticos) o quizá escritura braille (más limitada) para permitir su acceso.

Sobre otras posibilidades de hacer accesible el cómic, pensamos en una combinación de tres distintas aplicaciones: canal audio, braille y paneles/páginas táctiles. Estos últimos se utilizarían para dar cuenta de la estructura del cómic (disposición de viñetas, personajes, etc.). 

Las opciones para el código verbal son tres: en un caso se optaría por el audio, un doble canal audio que, por un lado dará cuenta de la imagen y, por otro, de los diálogos de los personajes o voz del narrador. En otro caso, la parte escrita podría estar en braille y el canal audio se utilizaría para la descripción de las imágenes a la que se accedería mediante un QR u otra tecnología emergente. La tercera, recoge lo expuesto para el primer caso: canal audio tanto para la AD como para los diálogos, formando una sola pista integrada que incluiría también los sonidos y paneles táctiles con una breve indicación introductoria sobre la exploración. Creemos que esta opción puede ser la más apropiada para evitar solapamientos que podrían confundir al oyente u otras problemáticas sin registrar.

Quedan por resolver algunas cuestiones, como por ejemplo la de la sincronía, ¿constituye un problema?, ¿cuándo se decide lo que se describe antes?, ¿se escucha la AD de la imagen sincronizándola en el segundo caso con la lectura braille del texto y en el tercero con la exploración del panel táctil? Respuestas y otros muchos interrogantes para los que se requiere experimentación y estudios de recepción.

7. A modo de conclusión

En este trabajo nuestra intención ha sido la de explorar un género narrativo poco conocido en el ámbito de la accesibilidad. Nos hemos centrado en la aplicación de la audiodescripción al cómic partiendo de un estudio breve del género y de dos experimentaciones fundamentalmente. Nos hemos planteado cuáles podían ser las etapas para producir el guion audiodescrito que acompañará al resto de los códigos presentes. A falta de unas línea guía y basándonos en las existentes para otros materiales, en mayor medida de productos cinematográficos y por afinidad a estos, hemos intentando esbozar algunas líneas o fases para un análisis del texto que permita elaborar la descripción y, en segundo lugar, reflexionar sobre la forma de fruición, es decir el componente o medios tecnológicos con que se propone su disfrute.

Queda claro que para la realización de una versión accesible de un texto como el cómic o novela gráfica, de naturaleza multimodal, es necesario realizar un análisis de los distintos códigos que lo conforman, sea para entender lo que denotan como lo que connotan, y por lo tanto realizar un estudio sobre el autor y la realidad sociocultural  en la que se inserta. Decíamos más arriba que el destinatario es un público con discapacidad visual, sin embargo y teniendo en cuenta las referencias culturales, se convierte en una modalidad válida asimismo para personas que no presentan dicha discapacidad, pero que pertenecen a distintas culturas.

Pensamos que una experimentación como la que hemos propuesto donde se dan cita distintos medios para transmitir dicho contenido puede ser una opción con buenas posibilidades, por supuesto, será un estudio de recepción el que nos proporcionará los datos necesarios para validar o modificar la propuesta.

No cabe duda de que se trata de un medio que debe ofrecerse a todos los públicos, independientemente de su condición física y de que es necesario continuar la investigación para conseguir resultados que demuestren su factibilidad y su accesibilidad se convierta en una realidad.

Abstract

English:

Communication can be found in every facet of life. What we do (or do not do) conveys a message, transmits information that will reach the receiver with greater or lesser intensity. In certain circumstances, this information needs to be interpreted by third parties and transferred through different systems for it to be correctly understood and assimilated, a transfer which, as Jakobson (1959: 232-239) points out, gives rise to translation; Jakobson recognises different types of translation depending on the information that is transferred. The intention is to devote some space to the analysis of some means of access to information and culture by the blind or seriously visually impaired community through intersemiotic translation: first we will present the new technology developed by ONCE to facilitate guidance through museum spaces, and then we will present an innovative project in the world of fashion that aims to enable people with sight difficulties to dress in a way that allows them to be aware of the clothes they are wearing.

Although initiatives such as these are a step towards accessibility, it is necessary to work as a society to promote information that encourages personal autonomy and independence.

Spanish:

La comunicación se encuentra presente en todas las facetas de nuestra vida. Lo que hacemos (o no) comunica un mensaje, transmite una información que llegará con mayor o menor intensidad al receptor. En determinadas circunstancias, dicha información necesita ser interpretada por terceros y trasladada a través de diferentes sistemas para su correcta comprensión y asimilación, trasvase que, como señala Jakobson (1959: 232-239), da lugar a una traducción. Jakobson reconoce varios tipos de traducción en función de la información que se traslada. Nuestra intención es dedicar un espacio al análisis de algunos medios de acceso a la información y la cultura por parte de la comunidad ciega o con serios problemas de visión a través de la traducción intersemiótica. En primer lugar presentaremos la nueva tecnología desarrollada por ONCE para facilitar el guiado por los espacios museísticos, y posteriormente daremos a conocer un proyecto innovador en el mundo de la moda que busca permitir que personas con dificultades de visión puedan vestirse conscientes de la ropa que llevan puesta.

Aunque iniciativas como estas constituyen un avance hacia la accesibilidad, es necesario trabajar como sociedad para promover una información que fomente la autonomía e independencia personal.

Keywords: imágenes, accesibilidad, ceguera, tecnología, trasvase, images, accessibility, blindness, technology, transfer

1. Introducción

La plural sociedad actual nos lleva a la presencia de colectivos cuyo acceso a la información se ve limitado por circunstancias personales; para facilitarlo, la traducción e interpretación representa una herramienta de ayuda. De los tres tipos de traducción a los que se refiere Jakobson (1959: 232-239) y que presentaremos con más detenimiento a continuación. En el presente trabajo pondremos el foco en la intersemiótica, que por sus características permite que un mayor número de personas puedan acceder a la información y participar sin barreras en entornos a los que nos referiremos más adelante.

Tras dedicar un espacio a la traducción intersemiótica como herramienta de mediación facilitadora de la comunicación, nos centraremos en la accesibilidad. En este sentido, y tras definir el concepto, presentaremos la normativa más destacada en torno a la misma. Posteriormente procederemos a analizar la audiodescripción, considerada precisamente una forma de traducción intersemiótica, lo cual nos permitirá centrarnos en la manera en que se hace presente en el mundo ¡”tual’para favorecer la accesibilidad, presentando para este fin dos iniciativas en ámbitos muy diferentes entre sí, cuya información se ha obtenido mediante el sistema de entrevista a las personas responsables de los proyectos: por una lado nos referiremos al proyecto AMUSE y el uso de “beepcons”, una innovadora aplicación interactiva y accesible mediante balizas de guiado inteligente por bluetooth para recorrer museos que podría abrir las puertas a su aplicación en otros espacios y derribar así cada vez más barreras en términos de comunicación y accesibilidad. Por otro, nos referiremos a una iniciativa desarrollada en el ámbito de la moda de la mano de la diseñadora ecuatoriana Camila Chiriboga, quien a través de entrevista presenta un innovador diseño de etiquetado en la ropa basado en la audiodescripción y que busca facilitar el día a día de quienes tienen serias dificultades visuales.

2. El traductor audiovisual como mediador y facilitador de la comunicación

La comunicación es parte de nuestra vida; independientemente del formato en que esta se produzca, el traductor deberá trabajar para lograr que su contenido llegue al máximo número de personas, por lo que deberán tenerse en cuenta posibles discapacidades o dificultades del receptor a la hora de recibir y entender la información en cuestión. Esta afirmación se apoya en las palabras de Torop (2002: 1), que sostiene que todos los tipos de comunicación en la cultura pueden presentarse como un proceso de traducción de textos (o fragmentos) en otros textos. Entre estos textos, el autor se refiere a aquellos hechos de una sustancia, como puede ser la verbal, que se traducen a textos hechos de otra sustancia, como puede ser la audiovisual (Torop 2002: 2). Precisamente queremos referirnos a la traducción audiovisual por los cambios que ha experimentado con el paso de los años.

Hasta comienzos del siglo XXI no existían demasiados estudios en torno a la traducción audiovisual en comparación con otras modalidades de traducción como la traducción literaria o la traducción de la Biblia, con décadas de historia, algo que, como apuntaba Chaume en 2004 (114), podía deberse a los siguientes factores:

• La historia de los Estudios de Traducción es relativamente joven, pues constituye una disciplina que no nacería como tal hasta que se llevaron a cabo los primeros estudios en profundidad a mediados del siglo XX.
• Los estudios sobre comunicación y los medios de comunicación también son recientes.
• Parece que la Traducción Audiovisual ha recibido una consideración menor a la de modalidades como la traducción jurídica, la literaria o la científica.
• En el ámbito académico se le ha concedido poco crédito a la traducción y el trabajo del traductor, algo que empeora en el caso de la traducción audiovisual.

No obstante, como ya apuntaba Orrego-Carmona en 2013 (298-299), los grandes avances tecnológicos habrían provocado que los estudios en traducción audiovisual hayan aumentado significativamente durante las últimas décadas. A este respecto, existe un dato que nos permitiría afirmar que la investigación en este campo ha crecido exponencialmente: desde 2004 y hasta el tiempo presente, el buscador de Google Scholar arroja cerca de 19000 investigaciones en torno a la TAV en lengua española, y alrededor de 21500 en lengua inglesa.

Díaz Cintas (2007:9) señala que el auge de la traducción audiovisual se da en los años noventa, a raíz de la investigación que se desarrolla en este sentido y que da lugar a publicaciones, congresos y la creación de cursos que abordan esta modalidad. A este respecto, a finales del siglo xx, Mayoral (1998) señala que la traducción audiovisual incluye diversos tipos de traducción, como son el doblaje subtitulado, las voces superpuestas, la narración, la traducción simultánea o el half-dubbing, los cuales se utilizan en géneros audiovisuales como la ficción, documentales, publicidad o telediarios. Esta visión la comparten a principios del siglo xxi autores como Gambier (2000) o Chaume (2004).

Si bien podríamos pensar que la traducción audiovisual se limita a aquella que tiene lugar en el ámbito cinematográfico o a la que genera productos de vídeo y televisión con el subtitulado y el doblaje como protagonistas, con el paso de los años se incluyen otros tipos de traducción como son los que dan lugar a las 'modalidades de accesibilidad'. Reverter Oliver et al. (2021: XII) destacan la contribución de Chaume (2013), que realiza una nueva propuesta articulada alrededor de dos ejes que identifica como macro-modalidades de TAV. Por un lado, se encuentra el captioning, que incluye, además de la subtitulación, el sobretitulado, el rehablado (o subtitulado en vivo), la subtitulación para sordos y el fan/funsubbing. Por otro, el revoicing, en el cual se incluye nuevas modalidades como la interpretación simultánea, el comentario libre, los fan/fundubs o la audiodescripción.

Creemos que el uso que tradicionalmente se ha hecho de modalidades más convencionales de traducción audiovisual responde a cuestiones culturales, económicas o incluso políticas; sin embargo, la importancia que ha cobrado en los últimos tiempos la accesibilidad ha dado lugar a la utilización de otras técnicas como la subtitulación para sordos o la audiodescripción, modalidad esta última en la que nos centraremos en el presente trabajo.

No resulta simple definir el concepto de traducción, pues su descripción varía en función del autor. Así, nos encontramos con opiniones como la de Steiner (1975/1980:65), quien afirma que la traducción supone “cualquier modelo de comunicación que da pie a un modelo de traslado, de transferencia vertical u horizontal de significado”. Opinión similar sostiene Bettetini, que señala que:

traducir es reescribir, respetando un proyecto de comunicación. Traducir, en el caso del traslado de un texto literario a un texto audiovisual, es producir una nueva máquina semiótica, que intenta repetir por analogía (también en sus relaciones con el usuario) el trabajo de aquella de la que se ha partido. (Bettetini 1986: 102)

En la misma línea, Chaume (2004: 141-142) apunta que traducir es una operación consistente en trasladar un texto de un medio a otro, independientemente de si en esta traslación se produce un cambio de lengua o no. Así, podemos decir que su definición va en la línea de las palabras de autores como Visión diferente tiene Martínez Sierra (2004: 16), quien hace ya dos décadas consideraba que la traducción “ha de implicar, como mínimo, dos lenguas y dos sistemas culturales distintos en mayor o menor grado”.

Por su parte, Jakobson (1959: 232-239), como se ha adelantado, distingue tres tipos de interpretación del signo lingüístico, en función de si su sentido se transfiere, traduce o transpone en otros signos de la misma lengua, en otra lengua, o en un sistema no lingüístico. Nos encontramos en primer lugar con la traducción intralingüística, que supone la interpretación de palabras y mensajes, en definitiva, de signos verbales, y su reexpresión en el mismo idioma mediante sinónimos, neologismos, cambios semánticos o circunloquios que actúan con carácter explicativo (Jakobson 1981: 69).

Por otro lado, la traducción interlingüística se ocupa de trasladar un mensaje entre lenguas diferentes; es la que solemos considerar como traducción propiamente dicha. A este respecto, Santaemilia (2010) considera que, desde la perspectiva de los actuales estudios de la traducción, esta denominación es inexacta, y apunta que, en realidad, hablar de traducción intercultural sería más apropiado, pues en el proceso de traducción no se produce un simple trasvase de unidades lingüísticas, sino una reescritura ideológica, de unidades culturales.

Como apunta Zavala (2008:48), Jakobson fue quien, con su modelo lingüístico, primero destacó la importancia de extrapolar las categorías de la lingüística con el fin de estudiar no solo la traducción que él mismo denominaría interlingüística, esto es, la que se produce de una lengua natural como puede ser el inglés a otra como pudiese ser el español, sino también la traducción intersemiótica, que supone la traducción de textos que pertenecen a distintos lenguajes (ya sean de carácter lingüístico o semiótico), la interpretación de signos, por ejemplo, de un sistema no verbal a uno que sí lo es, o viceversa (Jakobson 1959).  En esta categoría tienen cabida, por ejemplo, las onomatopeyas y las expresiones gestuales y faciales, pero también las imágenes.  

La traducción intersemiótica incluye modalidades de traducción diferentes: audiodescripcion (AD), subtitulación para sordos, interpretación en lengua de signos española (ILSE) y adaptación textual a lectura fácil. Se trata de modalidades que permiten la recepción de una información por parte del usuario que debido a su condición no podría acceder a los productos resultantes de la traducción inter- o intralingüística.

3. Accesibilidad

Ahora que entendemos qué es la traducción intersemiótica y vemos que permite el acceso a una información que de otro modo quedaría inaccesible para parte de la población, analicemos brevemente el concepto de accesiblidad.

Diferentes organismos y asociaciones, entre ellas Fundación ONCE (2009) definen la accesibilidad como la posibilidad de tener acceso, paso o entrada a un lugar o actividad sin limitación alguna por razón de deficiencia, discapacidad, o minusvalía, y distingue entre distintos tipos de accesibilidad:

• Urbanística, referida al medio urbano o físico.
• Arquitectónica, referida a edificios públicos y privados.
• Accesibilidad en el transporte, referida a los medios de transporte públicos.
• Accesibilidad en la comunicación, referida a la información individual y colectiva.

En este sentido, Richart-Marset y Calamita (2020: 32) subrayan precisamente la importancia de la accesibilidad, tanto en el marco de los Estudios de Traducción como en el seno de la traducción audiovisual, dado “su objetivo último, el de la integración social y la eliminación de barreras de dependencia”. Tengamos en cuenta que, en función de las características del entorno, así como de las capacidades de cada persona, la interacción con el medio variará. Si quiere alcanzarse una interacción satisfactoria por parte del máximo de personas con capacidades funcionales diferentes, será necesario trabajar en el diseño del entorno, producto o servicio al que pretenda accederse (Mati 2020).

El traductor debe actualizar sus conocimientos constantemente para realizar las labores propias de su profesión, actuando como mediador lingüístico y cultural (Carlucci  and Seibel 2016), fomentando con su labor la accesibilidad en el ámbito de la comunicación.  El objetivo que se persigue no es otro que cumplir con las exigencias de comunicación globales para facilitar el acceso no solo a la información, sino también al conocimiento y la accesibilidad universales en diferentes ámbitos, que se encuentran entre los siete principios del diseño universal, creados en 1997 en el Center for Universal Design (Carolina del Norte). Este constituye hoy un centro de referencia y de investigación en materia del Diseño para Todos y aplicables al diseño de cualquier elemento para ofrecer a los diseñadores una guía para integrar mejor las características y buscar así resolver las necesidades de tantos usuarios como sea posible (Ibid).

En este sentido, diversos autores abordan la relación entre la traducción y la accesibilidad; destacamos, entre otros, el trabajo de Díaz Cintas (2007), Márquez Linares (2007), Alba Rodríguez (2014), Jiménez Hurtado y Soler Gallego (2015), Talaván et al. (2016) o Romero-Fresco (2018, 2021, 2022).

3.1 Normativa en torno a la accesibilidad

La accesibilidad es una necesidad que debe atenderse y cubrirse si se quiere dar cumplimiento a la legislación existente.  Si bien dedicar un espacio a analizar cada uno de los instrumentos legislativos disponibles en materia de discapacidad y accesibilidad escaparía del ámbito en que queremos centrar nuestro trabajo y ocuparía un espacio del que no disponemos.  Desde el punto de vista concreto de la discapacidad podría afirmarse que el origen de la defensa de la accesibilidad se encuentra en la adopción por parte de Asamblea General de las Naciones Unidas  en 1966 y entrada en vigor una década después del Pacto Internacional de Derechos Civiles y Políticos, si interpretamos en este sentido su artículo 3, que defiende la igualdad entre hombres y mujeres en el disfrute de sus derechos civiles y políticos, para lo cual entendemos como necesario este componente propio de la accesibilidad.

Por su parte, la Convención Internacional sobre la Eliminación de todas las Formas de Discriminación Racial establece en su artículo 5 el derecho “de toda persona a la igualdad ante la ley, sin distinción de raza, color y origen nacional o étnico, particularmente en el goce de entre otros derechos, el de acceso a todos los lugares y servicios destinados al uso público”, así como “a participar, en condiciones de igualdad, en las actividades culturales”.

Más directa o específica en cuanto al tratamiento de la discapacidad es la Convención Internacional sobre los derechos de las personas con discapacidad, aprobada el 13 de diciembre de 2006 por la Asamblea General de las Naciones Unidas (ONU), ratificada por España el 3 de diciembre de 2007 y que entró en vigor el 3 de mayo de 2008.Regula el acceso de las personas con discapacidad a todos los ámbitos de la sociedad, incluidas la Cultura y la Educación,  y su artículo 9, dedicado específicamente a la accesibilidad, establece que los Estados Parte están obligados a adoptar las medidas necesarias para asegurar el acceso de las personas con discapacidad, en igualdad de condiciones con las demás, a la información y a las comunicaciones. 

Otro guiño a la atención a las personas con algún tipo de discapacidad lo encontramos en el artículo 13 del Tratado constitutivo de la Comunidad Europea, que habilita al Consejo para “adoptar acciones adecuadas para luchar contra la discriminación por motivos de sexo, de origen racial o étnico, religión o convicciones, discapacidad, edad u orientación sexual” y da lugar a una serie de directivas para desarrollar esa competencia.

A nivel nacional, España aprueba diversos textos jurídicos que persiguen la defensa de los derechos de las personas con discapacidad a través de la accesibilidad.  Comenzamos con la Constitución Española, que en cierto modo recoge este derecho a la accesibilidad en su artículo 9 al exigir a los poderes públicos que garanticen el máximo bienestar de la ciudadanía facilitando su participación en diferentes ámbitos de la vida. Además, en su artículo 49 se defiende la creación de una política que integre a quienes sufren algún tipo de discapacidad, ya sea física, psíquica o sensorial.

Queremos destacar la Ley 51/2003, de 2 de diciembre, de igualdad de oportunidades, no discriminación y accesibilidad universal de las personas con discapacidad, que en su  artículo 1 define la igualdad de oportunidades como “la ausencia de toda discriminación, igualdad de oportunidades la ausencia de discriminación, directa o indirecta, que tenga su causa en una discapacidad, así como la adopción de medidas de acción positiva orientadas a evitar o compensar las desventajas de una persona con discapacidad para participar plenamente en la vida política, económica, cultural y social” y que se aprueba con la intención de servir como estrategia de lucha contra la discriminación y de “accesibilidad universal”. Esta ley, que supuso un importante avance para la configuración de muchas medidas sobre materia de accesibilidad porque establecía su aplicación también al patrimonio cultural, se verá modificada por la Ley 26/2011, de 1 de agosto, de adaptación normativa a la Convención Internacional sobre los Derechos de las Personas con Discapacidad” para ajustar la definición legal de “persona con discapacidad” a la contenida en la Convención y salvaguardar sus derechos de manera que puedan avanzar hacia la autonomía personal desinstitucionalizada y garantizar su no discriminación.

Por otro lado, y en el ámbito de las tecnologías, nos encontramos con el Real Decreto 1494/2007, de 12 de noviembre, de condiciones básicas para el acceso de las personas con discapacidad a las tecnologías, productos y servicios relacionados con la sociedad de la información y medios de comunicación social, inspirado en los principios establecidos en la Ley 51/2003 y en vigor actualmente.

La Ley 51/2003 quedará derogada por el Real Decreto Legislativo 1/2013, de 29 de noviembre, por el que se aprueba el Texto Refundido de la Ley General de derechos de las personas con discapacidad y su inclusión social. La norma define, por primera vez, los distintos tipos de discapacidad, protege especialmente a quienes sean susceptibles de sufrir "discriminación múltiple” y dedica un título a los derechos de las más de cuatro millones de personas con discapacidad en España, algo que desde el Gobierno español (La Moncloa 2023) no tiene precedentes en nuestra legislación, abordando su protección en todos los ámbitos.

A su vez, dicho Texto Refundido se verá modificado por la Ley 6/2022, de 31 de marzo, de los derechos de las personas con discapacidad y de su inclusión social, que establece y regula la accesibilidad cognitiva y sus condiciones de exigencia y aplicación. Dicha ley consta de un artículo que incluye la modificación y la adición de diversos artículos al Real Decreto Legislativo 1/2013. En él se define la accesibilidad universal como

la condición que deben cumplir los entornos, procesos, bienes, productos y servicios, así como los objetos, instrumentos, herramientas y dispositivos para ser comprensibles, utilizables y practicables por todas las personas en condiciones de seguridad y comodidad y de la forma más autónoma y natural posible.

Y, como adelantábamos, se incluye en la accesibilidad universal la accesibilidad cognitiva para permitir la fácil comprensión, la comunicación e interacción a todas las personas. mediante la lectura fácil, sistemas alternativos y aumentativos de comunicación, pictogramas y otros medios humanos y tecnológicos disponibles para tal fin.

Las instituciones deben adoptar las medidas que garanticen un acceso más igualitario a la información, independientemente del ámbito y formato en que la información se presente. Aunque el cumplimiento de la normativa por parte de algunos sectores está en tela de juicio, contamos en España con organismos como la Organización Nacional de Ciegos Españoles (ONCE) que llevan a cabo acciones para que el patrimonio cultural y natural del país sea accesible y que quienes lo necesiten puedan acceder a los contenidos de las exposiciones de forma visual, táctil y auditivamente. Todo ello bajo la idea de un proyecto integrador y no exclusivo para personas con discapacidad visual.

Seibel et al. (2020: 224) destacan la importancia que adquieren las nuevas modalidades de traducción intersemiótica de imágenes a palabras, en tanto que constituyen una herramienta clave para la accesibilidad universal, e incluyen entre las mismas la subtitulación, la interpretación en lengua de signos española (ILSE) para personas sordas, la adaptación textual a lectura fácil para la diversidad cognitiva y la audiodescripción para personas ciegas, en la cual deseamos centrarnos en este momento.

4. Audiodescripción

Según Nida (1959: 429), y aunque pudiera parecer lógico, al traducir debe alcanzarse la misma respuesta en los receptores meta del texto traducido que la que se produce en los receptores del texto original. Si esto lo aplicamos a la audiodescripción, la imagen, que representaría en este caso el texto origen, deberá ir acompañada de descripciones, que darán lugar al texto meta, de manera que el receptor, en este caso una persona con problemas de visión, pueda acceder al contenido de la imagen en cuestión.

El grupo TRACCE (s.f) la entiende como una “la modalidad de traducción multimodal o intersemiótica en la que se traducen imágenes estáticas o dinámicas a una lengua, entendida como sistema de comunicación verbal, gestual o escrito, propio de una comunidad humana”.

Si bien hablamos de nuevas modalidades, la audiodescripción lleva tiempo entre nosotros; concretamente en España, empieza a utilizarse a finales de los años ochenta  (Hit 2022). En 2005 se publica la norma UNE 153020, que en su párrafo 2.1. define esta práctica como:

el servicio de apoyo a la comunicación que consiste en el conjunto de técnicas y habilidades aplicadas, con objeto de compensar la carencia de captación de la parte visual contenida en cualquier tipo de mensaje, suministrando una adecuada información sonora que la traduce o explica, de manera que el posible receptor discapacitado visual perciba dicho mensaje como un todo armónico y de la forma más parecida a como la percibe una persona que ve. 

Asimismo, establece una serie de requisitos para garantizar la calidad de las producciones audiodescritas, las cuales, si bien no vamos a presentar en el presente trabajo, pueden conocerse consultando dicha norma.

Por otro lado,recientemente, en 2020, se aprueba la Orden CUD/582/2020, de 26 de junio, por la que se establecen las bases reguladoras de las ayudas estatales para la producción de largometrajes y de cortometrajes y regula la estructura del Registro Administrativo de Empresas Cinematográficas y Audiovisuales, que obliga a las producciones que deseen solicitar alguna ayuda pública de las que concede el Ministerio de Cultura y Deporte a través del Instituto de la Cinematografía y de las Artes Audiovisuales de España (ICAA), a incluir el servicio de audiodescripción para personas con discapacidad visual junto con el servicio de subtitulado adaptado para personas con discapacidad auditiva.

Según la Organización Mundial de la Salud (2023), más de dos mil millones de personas en todo el mundo viven con algún tipo de discapacidad visual. A este respecto resulta interesante subrayar que aunque la audiodescripción pueda estar dirigida especialmente a aquellas personas con discapacidad visual, ya se trate de una pérdida de visión total o parcial o baja, también pueden beneficiarse de ella personas mayores y niños con necesidades específicas de aprendizaje, pues por sus características sirve de apoyo en su alfabetización, así como por cualquier persona en general al utilizarse como refuerzo para la comprensión de un mensaje visual (Snyder 2008: 192).

En cuanto a la aplicación de la audiodescripción, su uso no se limita al cine y la televisión, sino que tiene cabida en espectáculos en directo como pueden ser musicales, obras de teatro, exposiciones de arte y museos o la danza, además de en espacios naturales o culturales, entre otros (Rodríguez Domínguez y Domínguez Burrieza 2018).

4.1. Actuaciones en torno a la accesibilidad y la audiodescripción

Para favorecer la accesibilidad, tan importante para intentar alcanzar un trato más igualitario de todos los integrantes de la sociedad, existen ciertos proyectos novedosos cuyo desarrollo quisiéramos dar a conocer. Como adelantábamos, en primer lugar nos referiremos a la audiodescripción en el ámbito museístico para presentar una novedosa tecnología de la mano de Fundación ONCE, y posteriormente daremos a conocer un proyecto de moda que incluye en sus prendas la audiodescripción para así facilitar el día a día de quienes no pueden ver.

4.1.1.  Cultura accesible: la audiodescripción en el entorno museístico

En el ámbito de la cultura, si lo que se persigue es que la persona con discapacidad visual pueda acceder a los contenidos de un evento o colección y conseguir obtener una imagen mental del objeto representado, en ocasiones existe la posibilidad de utilizar el tacto como canal perceptivo, por un lado, y en el caso de que tuviera algo de visión, recurrir a la utilización de contrastes, colores o macrocaracteres. No obstante, como señala Hermida (2016: 17), en no pocas ocasiones la creación de esta imagen mental no es posible a través del tacto; es el caso de cuadros, cerámicas, o materiales delicados, así como si se trata de obras de grandes dimensiones que, por su tamaño, impiden su percepción a través de este sentido. Si bien podemos afirmar que la accesibilidad total para los diferentes tipos de discapacidad es una utopía, puede lograrse un mayor grado de accesibilidad si se ponen a disposición del usuario audioguías adaptadas con toda la información precisa, que describa los contenidos táctilmente accesibles. Una señalización táctil o auditiva y visual adecuada de los accesos, entornos físicos visitables y de esos equipamientos de museos, monumentos o incluso rutas guiadas y disponer de información auditiva mediante paneles, vídeos también cobran gran importancia para convertir estos espacios en lugares accesibles.

Diferentes autores han investigado en torno a la audiodescripción en entornos museísticos; es el caso de Díaz Cintas (2010), Soler y Chica (2014), Álvarez de Morales (2018), Barnés Castaño y Jiménez Hurtado (2020), Carlucci y Seibel (2020), y Ara Gregorio (2021), entre muchos otros. En esta ocasión, si bien permanecemos en el ámbito de los museos, queremos centrarnos no tanto en el acceso a la información relativa a las obras presentes en dichos museos, sino más bien en los medios puestos a disposición del usuario con discapacidad — visual en este caso — para facilitar su guiado por el espacio museístico.

Como muestra de la labor desarrollada en el entorno museístico, presentamos a continuación una iniciativa puesta en marcha en España y que consideramos resulta innovadora y de gran ayuda para quienes necesitan acceder a la información a través de la audiodescripción.

a. El sistema BEEPCONS y AMUSE

El sistema BEEPCONS es un proyecto impulsado por ILUNION Tecnología Accesibilidad (Fundación ONCE), un grupo empresarial de España con empresas de vigilancia, lavandería, desarrollo y accesibilidad, además de con una cadena hotelera. Las empresas del grupo, que trabaja exclusivamente proyectos de accesibilidad sobre todo desde un enfoque tecnológico, prioriza la inclusión de personas discapacitadas. Concretamente el 40% de sus empleados presentan alguna discapacidad.

El sistema de guiado por medio de balizas inteligentes denominadas ‘beepcons’ facilita la localización de objetos dentro de un espacio desconocido, así como la orientación del usuario, y se utiliza en espacios museísticos, restaurantes u hoteles. Para ello, estos dispositivos de señalización utilizan la tecnología Bluetooth Low Energy para enviar información diversa a los móviles que estén cerca: algunos de los datos que se proporcionan son la ubicación, distancia, elementos existentes en el entorno y descripción de los mismos, entre otros. El funcionamiento es el siguiente: en primer lugar, se lleva a cabo su instalación en un punto concreto dentro del espacio museístico. Los usuarios, a través de la app instalada en sus teléfonos, reciben un aviso mediante sonido, vibración o notificación verbal, señal mediante la cual localizan los baños, las puertas de la cafetería, y les permite acercarse al destino elegido. 

En el caso concreto de los museos, los beepcons permiten almacenar en ellos información grabada relativa a lo que hay en esa estancia o toda la información de la obra de arte en cuestión, que será descrita al usuario cuando este la solicite, haciendo que esta se convierta en una experiencia más interactiva más novedosa y divertida, y la visita sea más atractiva. En todo momento se localiza dónde está el usuario; si se detecta que se encuentra en una fase de la ruta que no corresponde, se le puede dirigir, si lo desea, a la información correspondiente, además de facilitársele un guiado entre sala y sala.

Además, el sistema de beepcons tiene una función más allá de las audioguías; puede utilizarse a modo de yincana con preguntas que resolver y puntos que obtener hasta completar la colección que conforma el juego. De este modo, los usuarios tienen la misión de “capturar” artistas y herramientas de estos artistas a lo largo de diferentes museos, fomentando así la movilidad de un museo a otro.

 En 2018, el Museo Tiflológico se convertía en el primer espacio museístico en contar con esta tecnología. Si bien podríamos pensar que esta innovadora aplicación interactiva es apropiada únicamente para las personas con discapacidad visual, y aunque ciertamente es especialmente útil para ellas,  también puede ayudar a aquellas que no presentan problemas de visión sino discapacidad intelectual y que necesitan o desean acceder a una información en un formato más sencillo, lo que la convierte también en adecuada para el público general, que podría preferir escuchar esa información que permite la identificación o localización de los objetos cercanos a tener que leerla.

Y aunque presenta numerosas ventajas, en conversación mantenida en 2022 con el jefe de la Unidad de Información y Accesibilidad de la ONCE, Luis Palomares, este subraya que la importancia a la hora de colocarlos, para lo cual resulta necesario estudiar su propósito, ya que en algunos casos su uso no está recomendado. Supongamos que se colocan pegados los unos a los otros:  si cada persona activa un beepcon a la vez, las señales acústicas confundirían porque se mezclarían entre sí. Asimismo, Palomares nos indica que la ONCE no está de acuerdo con su implementación en entornos cerrados, desaconsejando su instalación especialmente en aquellos lugares en los que debe haber silencio, como puede ser una catedral. No obstante, consideramos que la alternativa en estos casos podría ser la utilización de auriculares conectados al dispositivo móvil, para evitar molestar al resto de los visitantes. En cuanto a su adquisición por parte del ente que quiera emplearlos, su coste no es elevado; empresas, entidades, organismos, pueden comprar este servicio de beepcons, siendo empresas privadas o ayuntamientos quienes suelen financiar su compra.

Podría surgir la duda de lo que sucede con los derechos de propiedad y pensar que, una vez que se accede a la información contenida en los beepcons, esta queda descargada y puede consultarse desde cualquier punto y en cualquier momento. Pues bien, para proteger estos derechos de propiedad, esta tecnología únicamente permite el acceso a la información en el museo, quedando inaccesible una vez fuera de este espacio.

El sistema beepcons se encuentra plenamente relacionado con el proyecto AMUSE, que cuenta con la colaboración de ILUNION, FUNDACIÓN ONCE y el Ministerio de Tecnología y Deporte. Consiste en una plataforma que se aloja en una página web a la que los museos que dispongan de beepcons, entre otros servicios, pueden unirse y gestionar la información que quieren dar a conocer a quienes visiten sus espacios. En otras palabras, AMUSE no constituye una app de cada museo, sino una plataforma en la que estos pueden presentarse. Su carácter genérico permite que, mediante su registro en la plataforma, los museos carguen sus contenidos y personalicen la forma en que su información figura en la plataforma, incluyendo no solo la información de audioguía estándar sino también las características y elementos de accesibilidad de que disponen: audiodescripción en vídeos, si tienen vídeos en lengua de signos, cuál es el nivel de accesibilidad de una determinada ruta, etc. AMUSE permite así llevar el concepto de audioguía a otro nivel de manera que sea más inclusiva e interactiva.

Como dato interesante, hay que señalar que su nombre es la unión de la Accesibilidad (A), por un lado, y la referencia inglesa a las musas que inspiran el arte, por otro (muse).  Además, la designación coincide en inglés con el verbo cuyo significado es “entretener”, lo cual lo convierte en un nombre más que adecuado para este proyecto. AMUSE no está pensada solamente para usuarios con discapacidad; con su creación se pretende ir más allá de la accesibilidad, utilizando esta herramienta para fomentar que el público se anime a visitar los museos. Lo que busca es facilitar que, al entrar en la plataforma, el usuario conozca qué elementos, que a menudo los museos tienen pero son desconocidos, ofrece el museo, además de permitirle saber qué rutas son más adecuadas para sus necesidades, pudiendo escoger una ruta más sencilla.

En entrevista mantenida con Palomares, nos explica que esta plataforma permite que los responsables de cada museo puedan ir actualizando y modificando las preguntas de los juegos, así como el contenido relativo a exposiciones temporales. Si tenemos en cuenta que, en el caso de las exposiciones temporales, normalmente no existe una guía electrónica por el coste que ello conlleva, puede afirmarse que el uso de esta plataforma es ideal en este caso, pues convierte la información en dinámica y cambiante.

En palabras de Palomares, si bien podría decirse que la implantación de AMUSE constituye una solución por delante de la normativa en materia de accesibilidad y que por tanto tiene un componente muy positivo, debe tenerse en cuenta que la normativa relativa a la accesibilidad recoge lo básico. Las medidas que la ley establece llevan, según explica, a la puesta en marcha de mecanismos como este en nuevos espacios para evitar penalizaciones; sin embargo, depende ya de cada uno hasta qué punto se desarrolla su uso. Es decir, su existencia no asegura que alguien sea totalmente autónomo. A este respecto, señala que la norma no especifica aspectos como pueden ser la localización de puntos concretos en un parque, por ejemplo.

 En cuanto a la presencia de beepcons, adelanta que su uso no debe ser masivo para evitar interferencias entre dichas balizas, e indica que además de en espacios museísticos se pueden encontrar en lugares al aire libre como parques y espacios diáfanos, pudiéndose instalar incluso de manera puntual y con motivo de eventos concretos en la ciudad, como puede ser la Navidad, facilitando el acceso a la descripción de los espacios a los usuarios que lo necesiten. En cualquier caso, y puesto que su presencia se da en pequeñas cantidades y en lugares no muy masificados, es reducida la cantidad de gente que puede acceder a la información contenida en los beepcons al mismo tiempo.

En 2022, y según nos informa desde la Dirección de Accesibilidad Universal e Innovación de la Fundación ONCE, Lourdes González Perea, los espacios que ha implementado AMUSE son:

• 2017: Tabacalera–Promoción del Arte, espacio dependiente del Ministerio de Educación, Cultura y Deporte.
• 2018:  Bienal de Arte Contemporáneo y el Museo Tiflológico.
• 2020:  Aquarium de San Sebastián, Museo del Enclave de la Muralla (MUDEM) de Molina de Segura y Museo Aeronáutico Tiflológico (MAT) de San Javier.

Al preguntarle por proyectos a partir de 2021, explica que, con la llegada del COVID, su expansión se paró para dar prioridad a la virtualidad, centrando los esfuerzos en crear un espacio web a través del cual acceder a los diferentes espacios.

b. Moda y audiodescripción

Otra de las áreas en las que la audiodescripción se hace presente es la moda. Se pueden encontrar precedentes en la AD de ópera, donde el vestuario suele ocupar un lugar importante, siendo uno de los elementos que marcan la producción y contribuyen a realzar la magnitud de la obra en cuestión. A esta realidad se refieren Cabeza y Matamala, quienes señalan que el flujo de información puede llegar a ser excesivo, y que puesto que el objetivo es que el usuario se haga una idea general de la ópera, se procura que la información audiodescrita, que incluye no solo información del vestuario, sino también datos históricos y el argumento, dure solamente unos diez minutos[1].

Nuestro interés por conocer otros usos de la audiodescripción en el ámbito de la moda nos empuja a realizar una búsqueda de proyectos que pudieran aunar estos dos ámbitos. En esta ocasión nos centramos en una iniciativa que tiene como protagonista a la diseñadora de moda Camila Chiriboga, de Ecuador, cuya idea consideramos puntera, motivo por el cual se contacta con ella para la realización de una entrevista.

Camila lleva inmersa en este proyecto desde 2018: dedicó un primer año de investigación a conocer las necesidades y forma de funcionar de estas personas con discapacidad visual, además de los tres años que necesitó para perfeccionar la aplicación a través de la tecnología. La diseñadora ha podido trabajar en este proyecto gracias al apoyo económico de profesores, miembros de organizaciones y asociaciones de personas ciegas y discapacidades visuales, así como personas individuales.

A la pregunta de cómo surge su plan, Chiriboga cuenta que todo surge al pasar un tiempo ingresada en el hospital, donde su dolencia y movilidad reducida le llevan a pensar en cómo se las arregla la gente con algún problema grave de salud para vestirse en su día a día. Centrándose en aquellas personas con discapacidad visual, y para entender de la mejor manera las dificultades a las que las mismas se enfrentan en su día a día, recurre en primer lugar a la autoexperimentación, comenzando a vestirse con los ojos cerrados. Asimismo, observando en su entorno se da cuenta de que quienes tienen problemas graves de visión se visten con ropa con más pelo, característica esta que las convierte en prendas más identificables gracias a su suavidad y tacto más reconocible. A su vez, esto hace que estas personas con discapacidad visual a menudo asistan a su lugar de trabajo o a la universidad en pijama sin ser conscientes de que ese no es el código de vestimenta adecuado para esos lugares. Esto sucede obviamente al desconocer qué aspecto tiene un pijama, pues eligen la prenda basándose en su tacto.

Su investigación le permite conocer a personas que han perdido la visión a lo largo de su vida; al no haber nacido con esa condición, se encuentran desorientados, especialmente cuando la ropa es nueva. Todo lo anterior lleva a Camila a querer crear una etiqueta basada en el sistema de braille para que puedan vestirse solos y ganar independencia. Sin embargo, su trabajo en el ámbito del diseño de moda con una profesora ciega le permite conocer, gracias a la experiencia de la última, numerosos aspectos de la vida de las personas con discapacidad visual. En este sentido destaca que solo quienes nacen ciegos aprenden braille, y que quienes pierden la visión en edades adultas no tienen esa sensibilidad para entender y aprender a detectar este sistema de puntos en relieve.

Esto le lleva a crear una etiqueta con tacto y audio, el cual simplemente debe conectarse a un móvil, lo que simplifica el acceso a la información, si tenemos en cuenta que la mayoría de las personas tiene acceso a uno, evitando así tener que adquirir una tecnología nueva. Como explica la joven diseñadora, el Código QR o la identificación por radiofrecuencia (en adelante, RFID (Radio Frequency Identification) por sus siglas en inglés) de la etiqueta permite acceder a la información a través del móvil, que reproducirá un audio con la información sobre la prenda: color, tamaño, cómo usarla, con qué se combina, o cómo se lava, entre otros aspectos. No obstante, señala que el uso del código QR no resulta tan sencillo como la RFID, ya que mientras este último identifica la señal por radiofrecuencia y es suficiente con acercarse a ella, el primero exige que la cámara encaje con el recuadro del QR, lo cual puede resultar complicado para alguien con problemas de visión.

Chiriboga destaca que lo más costoso en todos los sentidos es la patente, si bien la considera necesaria para proteger la idea y compartirla, y actualmente está en contacto con diferentes marcas que trabajan con gente con dificultades de visión con el objetivo de trasladar su idea a dichas prendas.

Para finalizar, al preguntarle su opinión acerca de la audiodescrpción en el ámbito de la moda y del consumo de moda, subraya que si bien cada vez hay más concienciación, todo debería empezar por codificar las webs para ofrecer audiodescripción y lograr así la accesibilidad a mayor escala.

5. Conclusión

De la información del presente artículo se pueden extraer diversas conclusiones: en primer lugar, la tarea clave de los profesionales de la traducción en materia de inclusión. Su papel es fundamental para fomentar el acceso a la información, independientemente de su formato. A este respecto, la audiodescripción, como traducción intersemiótica, constituye una modalidad que favorece la accesibilidad de colectivos que, de otro modo, no podrían acceder a la información.

Como herramienta para la audiodescripción, el uso de la tecnología basada en los beepcons representa una original solución cuya implantación no se limita únicamente al entorno museístico, sino que podría producirse en otros espacios y derribar así cada vez más barreras en términos de comunicación y accesibilidad. Y es que, aunque podría pensarse que constituyen un elemento de ayuda, debería decirse en su lugar que son “elementos necesarios”. La accesibilidad es un derecho universal: ciertamente existe una normativa que obliga a tomar ciertas medidas en este sentido, como en el  plano urbanístico, donde se exige que en el pavimento haya cierta señalética podotáctil para que quienes tienen problemas visuales identifiquen la parada de autobús, o los semáforos sonoros, entre otros; pero como señalábamos anteriormente, la normativa solamente recoge los aspectos más básicos en lo que a la audiodescripción se refiere, y no desarrolla aspectos tan específicos como el uso de esta tecnología.

Si bien estas iniciativas constituyen un avance hacia la accesibilidad, como sociedad es necesario seguir trabajando en este sentido para lograr que quienes necesitan estas herramientas en su vida cotidiana puedan alcanzar la autonomía e independencia personal. Así, coincidimos con la perspectiva de Palomares, que apunta que lo ideal sería que iniciativas como la de los beepcons no tuvieran que existir porque todos los servicios, como puede ser Google Maps, o páginas web de distinta índole incluyeran la audiodescripción, ofreciendo un acceso inclusivo para todos. Y con respecto a la segunda propuesta, ¿no se concluye nada?

Abstract

English:

This article presents the concept of alternative audiodescription, of which musical ekphrasis is a central category. The concept is based on the idea that selected elements of film drama, which escape verbal description due to limited time and space in audiodescription, could be translated into a piece of music. As in the case of ekphrasis set in the literary tradition, in the case of musical ekphrasis too, what is important is not mimetic accuracy and the reproduction of content, but the aesthetic tension that accompanies the reception of the work. Understood in this way, musical ekphrasis is meant to be an impression on selected elements of the language of the film, while audiodescription itself is defined as an approximation - one of many possible attempts to get closer to the original.

Following the theoretical introduction, the article discusses the results of a survey that summarises the 'Spaces of Audiodescription' project carried out between 2021 and 2023 at Adam Mickiewicz University in Poznań. As part of the project, an audiodescription and a musical ekphrasis were created for a film from the resources of the Centre for Blind Children in Owińska ('Spatial Orietation Park and our centre from a bird's eye view'), which provides a musical illustration for the bird's eye view.  After the presentation of the audiodescription film and the musical piece, a survey was conducted among the Centre's students on the reception of the ekphrasis. The results of this survey are presented in the final chapter of the article.

German:

[abstract in German needed]

Keywords: Audiodeskription, musikalische Ekphrasis, Filmsprache, Hermeneutik, audio description, alternative AD, musical ekphrasis, film language, hermeneutics

1. Einführung

Bei der Erstellung der Audiodeskription wird der Beschreiber vor eine besondere Aufgabe gestellt: Einerseits soll er in sein Manuskript so viele der vorhandenen visuellen Informationen wie möglich aufnehmen, damit der blinde oder sehbehinderte Rezipient das künstlerische Schaffen, sei es beispielsweise der Hörfilm oder das Hörmuseum, umfassend erleben und genießen kann. Andererseits aber können aus der Fülle der durch das Bild vermittelten Informationen im Hinblick auf Raum- und Zeitfaktor meist nicht alle berücksichtigt werden. Dieses besondere Spannungsverhältnis, das von Bernd Benecke als „Audiodeskriptionsdilemma“ (Benecke 2014: 2) bezeichnet wird, ist keineswegs neu und wurde bereits im Diskurs um die Theorie und Praxis der Audiodeskription aus vielen Perspektiven diskutiert. Das lässt den Schluss ableiten, dass die Audiodeskription in gewissem Sinne oft in Kategorien einer Verlustbilanz betrachtet wird. Der vorliegende  Beitrag nimmt sich nun vor, der Frage nachzugehen, auf welche Art und Weise diese potentiellen Verluste auszugleichen wären. Das besondere Augenmerk gilt dabei der so genannten alternativen Audiodeskription. Nach der Begriffsbestimmung und nach der Einbettung der Audiodeskription in die Kategorien der intermedialen Transkriptionen im Sinne von Ludwig Jäger wird die Optik der Erwägungen auf  die Problematik der musikalischen Ekphrasis verschoben. Annäherung an die fokussierte Thematik bietet das im Rahmen des Beitrags dargestellte Projekt „Erfahrungsräume der Audiodeskription“, das  in den Jahren 2021-2023 im Institut für Germanische Philologie der Adam-Mickiewicz-Universität Posen durchgeführt wurde.

2. Zum Begriff der alternativen Audiodeskription

Im Laufe der Entwicklungsgeschichte der Audiodeskription werden neue Varianten geboten, die als eine Alternative zur konventionellen, hergebrachten Formen zu verstehen sind und mit deren Hilfe man danach strebt, die erwähnte Verlustbilanz aufzuheben.  

Kennzeichnend für die weltweite Professionalisierung des Audiodeskriptionsverfahrens im 21. Jahrhundert war das Veröffentlichen von Regelwerken für die Praxis. Diese legten großen Wert auf den Grundsatz der Objektivität. Wie Marina Ramos Caro und Ana María Rojo López konstatierten, wurde die Suche nach der Objektivität „nicht nur ein Desideratum“(Ramos Caro und Rojo López 2014: 133), sondern eher „eine der goldenen Regeln“(Ramos Caro und Rojo López 2014: 134) in der professionellen AD-Praxis. Dennoch stößt die Forderung nach der Objektivität und nach der starren Standardisierung der Audiodeskription in den letzten Jahren immer häufiger auf Kritik (vgl. u.a. Schaeffer-Lacroix, Reviers und Di Giovanni 2023). Im Anschluss daran werden in den unten angestellten Überlegungen als alternativ alle Varianten der Audiodeskription verstanden, die sich bewusst in eine Diskussion mit den hergebrachten, konventionellen Formen der Audiodeskription einlassen und eine eigene Konzeption vorlegen, die eine Antwort auf die aus dieser Diskussion resultierenden Unzulänglichkeiten versucht.

Als Sammelbegriff für unterschiedliche Varianten der Audiodeskription, die sich als alternativ zur konventionellen Beschreibungen verstehen, verwenden Agnieszka Chmiel und Iwona Mazur die Bezeichnungen alternative Audiodeskriptionen (alternatywne audiodeskrypcje) (Chmiel und Mazur 2014:58) sowie alternative Gattung bzw. Subgattung der AD (alternatywny gatunek lub podgatunek AD) (Chmiel und Mazur 2014:59). Agnieszka Szarkowska und Izabela Künstler sprechen dagegen in diesem Kontext von der experimentellen Audiodeskription (audiodeskrypcja eksperymentalna) (Szarkowska und Künstler 2012:91).

Wie bereits erwähnt, bilden die oben angeführten Bezeichnungen Oberbegriffe für verschiedene abgewandelte Formen der Audiodeskription. In die Diskussion um die Audiodeskriptionstheorie- und Praxis wurden jedoch auch viele Termini eingeführt, die sich jeweils direkt auf eine konkrete Konzeption beziehen. Bei einem etwas genaueren Einblick in die vorliegenden Entwürfe der alternativen Formen der Audiodeskription wird man schnell zum Schluss kommen, dass es unter ihnen zahlreiche Überschneidungen gibt. Doch es lässt sich alle alternativen Audiodeskriptionsansätze  je nach der einen oder anderen dominierenden Leitidee der Übersichtlichkeit der Ausführungen halber gruppieren, ohne feste Trennungslinien zwischen den einzelnen Kategorien den alternativen Audiodeskription zu ziehen.  Es geht über den Rahmen des vorliegenden Beitrags weit hinaus, einen umfassenden Überblick über alle Konzeptionen der alternativen Audiodeskription zu bieten, deshalb wird an dieser Stelle ausschließlich die Vielfalt der entwickelten Ansätze angedeutet. In grober Vereinfachung kann man die alternativen Formen der Audiodeskription in folgende Ansätze einteilen: narratologische Ansätze (u.a. Fels/Udo 2006; Kruger 2010; Ramos Caro 2016), filmästhetische Ansätze (Künstler/Butkiewicz/Więckowski 2012; Fryer/ Freeman 2012; Szarkowska 2013; Wilken/Kruger 2016; Bardini 2020), sprachexpressive Ansätze (Zabrocka 2017; Walczak/Fryer 2017; Chica 2019), integrative Ansätze (Romero-Fresco 2019; Naraine/ Fels/ Whitfield  2018; Fryer 2018) sowie technologische Ansätze (Lopez/Kearney/Hofstädter 2021; Pantula/ Kuppusamy 2019).

3. Medienimmanenz und transkriptive Intelligenz

Im kommunikativen Kontext stehen die visuellen Inhalte selten allein, sie kommen zumeist mit anderen Zeichensystemen verknüpft vor. Selbst so scheinbar typisch reine Bildgebrauchsdomänen wie zum Beispiel die Gemäldegalerie verwenden  in anderen Zeichensystemen kodierten Erklärungen oder Hinweise, beispielsweise  Bildlegenden oder Künstlerbiografien. Dasselbe geschieht auf dem Gebiet anderer semiotischer Systeme. Partituren werden von Künstlern interpretiert und in tönende Musik überführt, komplexe Sprachtexte werden mit Kommentaren versehen, literarische Stoffe werden einer filmischen Adaptation unterzogen. Auf „diesen Zustand der inhärenten quasi unvermeidbaren und zwingenden Multimodalität“ (Stöckl 2011: 46) weist verstärkt Ludwig Jäger hin (Jäger 2002), der  von der so genannten „Medienimmanenz“ (Jäger 2002:35) ausgeht. Die Medienimmanenz besagt, dass die Welt immer nur dadurch zugänglich wird, dass man sie sich durch kommunikative Entäußerung erst aneignet. Der dazu notwendige Zeichen- und Mediengebrauch beschränkt sich aber in der Auffassung von Jäger nicht auf ein Symbolsystem oder ein Medium, sondern ist essentiell von Transkriptionen gekennzeichnet. Bedeutungen werden ständig von einem Zeichensystem in ein anderes überführt, was offensichtlich in der Beschränktheit einer einzelnen semiotischen Ressource begründet ist. In der Auffassung von Jäger ergeben unsere Umwelt und unsere sozialen Praktiken offenbar nur dann einen Sinn, wenn man in einer Zeichenmodalität kodierte Inhalte in einer anderen kommentiert, expliziert und paraphrasiert. Nach Jäger wird die Welt erst durch intra- und intermediale Transkriptionen lesbar und in dieser Notwendigkeit sieht er das Grundprinzip kultureller Semantik. Die Fähigkeit zu derartiger Transkription bezeichnet er als „transkriptive Intelligenz“ (Jäger 2002: 35). Wie im Folgenden gezeigt wird, eröffnet sich gerade im spezifischen Feld der transkriptiven  Intelligenz ein weites Forschungsgeld für die Weiterentwicklung der alternativen Audiodeskription.

4. Musikekphrasis in der Audiodeskription

Die intra- und intermediale Transkriptionen im Sinne von Jäher haben in der Programmmusik eine sehr lange Tradition. So vermag dieses Kapitel das aufzuzeigen, was für den vorliegenden Beitrag eine zentrale Fragestellung bedeutet: das Potenzial der Musik für das Übersetzen der visuellen Inhalte.

4.1  Programmmusik

Die Programmmusik zählt zu einer der umstrittensten musikalischen Gattungen. Wie Constantin Floros feststellt (Floros 1983: 10), fehlt es nicht nur „an einer allgemein akzeptierten Definition, sondern es lässt sich auch oft zwischen den konträren Auffassungen gar nicht vermitteln“. Man gebraucht den Begriff in verschiedenen Sinnvarianten, und die Kriterien für die Begriffsbestimmung sind dabei jeweils unterschiedlich. Da eine ausführliche Diskussion rund um die Begriffsauslegung der Programmmusik weit über den Rahmen des vorliegenden Aufsatzes hinausgeht, wird die vereinfachte Definition von Kurt Westphal (1965) angeführt. Als „Programmmusik“ definiert Westphal (Westphal 1965: 3) Musik, „die zu anderen Gegenstandsbereichen und Künsten in Beziehung tritt. Sie ist nicht eigenständige, sondern kommentierende, nicht aussagende, sondern erzählende und malende Musik“. Die Definierung der Programmmusik erfolgt oft durch die Abgrenzung von der absoluten Musik. Während absolute Musik eine Gattung ist, „die frei ist von außermusikalischen Bindungen, losgelöst von den Realitäten der Erscheinungs- und Erfahrungswelt“, stellt Programmmusik „selbstständige Instrumentalmusik mit einem ausdrücklich benannten außermusikalischen Bezug“ (Lang 2014: 8) dar.

Innerhalb der Programmmusik unterscheidet man drei Formen:

1. die abbildende
2. die erzählende
3. und die versinnbildlichende Musik.

Die abbildende Musik, auch als Tonmalerei bezeichnet, spiegelt die Eindrücke der Außenwelt mit den Mitteln des Klanges wider, zum Beispiel den Flug des Schmetterlings oder das Heulen des Sturms. Die erzählende Programmmusik greift in die Sphäre der Dichtung über. So wie ein Roman stellt sie eine Handlung, einen Vorgang, ein Ereignis dar, wobei diese Handlung sowohl äußerer als auch psychologisch innerer Art sein kann. Solche Musikstücke sind gleichsam „Klangfilme, Filme ohne Bilder“ (Westphal 1965:8), wie zum Beispiel die „Biblischen Historien“ von Bachs Leipziger Amtsvorgänger Johann Kuhnau oder die Mehrheit aller Programmsymphonien und symphonischen Dichtungen auf der Linie Berlioz-Liszt-Strauss (vgl. Westphal 1965: 8).  Mit der versinnbildlichenden Musik ist die symbolische Dimension der Musik gemeint – z.B. die Sechsstimmigkeit des Sanctus in der H-moll-Messe. Westphal konstatiert (Westphal 1965: 34):

Der abbildende Programmmusiker rivalisiert mit dem Maler, der erzählende Programmmusiker mit dem Dichter. Der Tonmaler fängt ein Bild in Tönen, Rhythmen und Klängen auf, mag es ein optisches oder schon von Natur aus ein akustisches sein. Der Tondichter will Vorgänge schildern, Geschehnisse über das Ohr vor Augen führen. Die Übergänge sind fließend.

4.2 Musikekphrasis und Bedeutungspotenzial des Films

Es darf nicht übersehen werden, dass alle Kategorien der Programmmusik zwar immer einen ausdrücklich benannten außermusikalischen Bezug haben, jedoch eine selbständige Instrumentalmusik bilden. Diesen Aspekt hebt auch Siglind Bruhn (Bruhn 2000) in ihren Erwägungen hervor und führt in den Diskurs um die Programmmusik die Kategorie der „Musikekphrasis“ (musical ekphrasis) ein. Der aus der Literaturwissenschaft stammende Terminus bedeutet im weiteren Sinne die detaillierte Beschreibung von Personen, Gegenständen oder Ereignissen, deren Ziel es ist, ein „anschauliches Vor-Augen-Stellen“ (Port 2007: 182) der beschriebenen Sache. Im engeren Sinne wird mit der Ekphrasis die Beschreibung von Werken der bildenden Kunst gemeint. In der Auffassung von Bruhn (Bruhn 2000: 28-9) sind folgende Merkmale für die Abgrenzung der Musikekphrasis von der Programmmusik distinktiv:

The two genres belong to the same general species: both denote purely instrumental music that has its raison d’être in a definite referential, narrative, or pictorial scheme; both have variously been described as „illustrative“ or „representative“ music. […] One way of approaching the difference is to ask whose fictional reality is being represented. „Program music“ narrates or paints, suggests or represents scenes or stories  (and, by extension, evens or characters) that may or may not exist out there but enter the music from the comopsers own‘s mind. […] Musical ekphrasis, by contrast, narrates or paints a fictional reality created by an artist, other than the composer of the music: by a painter or a poet. Also, musical ekphrasis usually relates not only to the content of the poetically or pictorially conveyed fictional reality, but also to the form and style of the representation in which this content was cast in its primary medium.

Die von Bruhn erwähnte fiktive Realität, die von einem Künstler geschaffen wird, der nicht der Musikkomponist ist, bezieht sich bei der präsentierten Konzeption der alternativen Audiodeskription auf die fiktive Realität des Films.  Die Auffassung von Bruhn betont den engen Zusammenhang auf der Inhalts- und Formebene zwischen dem Ausgangsmaterial bzw. dem Ausgangswerk und dem Musikstück, und schließt sich somit an die hermeneutische Definition des Filmtextes an – Filme als übersummative und multiperspektivische Ganzheiten. Diese Prämissen bilden eine gut fundierte Grundlage für die potentielle Anwendung der Musikekphrasis in der Audiodeskription. Bruhn kontrastiert literarische Ekphrasis mit der Musikekphrasis und geht von der dreistufigen Struktur der Wirklichkeit und ihrer künstlerischen Umsetzung aus. Bei der literarischen Ekphrasis sind das:

1. ein realer oder fiktiver Text, der als Quelle für die künstlerische Darstellung dient;
2. eine primäre Darstellung dieses Textes in visueller Form (als Gemälde, Zeichnung, Fotografie, Schnitzerei, Skulptur usw.) oder als Film oder Tanz;
3. eine Neudarstellung dieser ersten Darstellung in poetischer Sprache;

Die poetische Wiedergabe geht über das Aufzählen der visuellen Details und Charakteristik ihrer räumlichen Position hinaus.  Charakteristisch ist, dass sie Interpretationen oder zusätzliche Bedeutungsebenen vermittelt sowie den Blick des Betrachters auf Details und Zusammenhänge lenkt, die uns sonst entgehen würden.

Dementsprechend umfasst die dreistufige Struktur der Musikekphrasis folgende Ebenen:

1. ein realer oder fiktiver Text, der als Quelle für die künstlerische Darstellung dient;
2. eine primäre Darstellung dieses Textes in visueller oder verbaler Form; und
3. eine Repräsentation dieser ersten (visuellen oder verbalen) Darstellung in musikalischer Sprache (vgl. Bruhn 2001: 6).

Erhellend für das Potenzial der musikalischen Ekphrasis in der Audiodeskription ist die lange Rezeptionsgeschichte der Programmmusik und die in dieser Geschichte dokumentierte Kunst, Bilder in Tönen einzufangen. 30 Jahre lang – in der Ära des Stummfilms - fand die Kinematographie Wege, alles Wichtige ohne Tonanwendung auszudrücken, und bildete eine hoch entwickelte Bildkultur heraus.  Im Laufe der Zeit, als man im Filmkunstwerk immer feinere Inhalte, komplizierte Konfliktnuancen mit allen psychologischen Schattierungen darstellen wollte, stellte es sich als unmöglich heraus, die Hörbarkeit der Welt zu ignorieren, und so begann die Existenz  von Geräuschen, Musik und der gesprochenen Sprache neben dem Bild. Ihre Funktionen sind in den bewegten Bildern zahlreich. Sie können Emotionen provozieren, Sympathien lenken, die Aufmerksamkeit steuern, Spannung und eine bestimmte Atmosphäre erzeugen sowie als Leitmotiv fungieren. In der im Rahmen dieses Beitrags vorgeschlagenen Konzeption der alternativen Audiodeskription zielt man darauf ab, bestimmte Elemente der Filmsprache, bestimmte Momente, die die Atmosphäre im Film mitgestalten, Emotionen der Figuren und emotionelle Momente ins musikalische Medium zu übertragen.  Diese Kompositionen können dann als zusätzliche Tonspur zum Hörfilm auf DVD oder bei den Filmen, die per Streaming zugänglich sind, zur Verfügung gestellt werden.

Das Interesse daran, komplexe visuelle Inhalte des Films in Musik zu übertragen, legt zwingend eine Klärung der möglichen Interpretationsansätze des Films nahe. Die Filmwissenschaft bietet ein ganzes Spektrum verschiedener Analysemethoden, aber, worauf  Anke-Marie Lohmeier hinweist, neben den herkömmlichen Modellen der Filmanalyse, die insbesondere quantitative Analyseverfahren entwickelt haben, gibt es vor allem semiotische Modelle der Sprach- und Literaturwissenschaft (vgl. Lohmeier 1996: XIV). Die quantitative Filmanalyse legt ein Inventar beschreibender Begriffe vor, die der Erfassung der spezifisch kinematographischen Abbildungsverfahren dienen sollen, und präsentiert Modelle der statistischen Auswertung filmischer, vor allem kameraspezifischer Daten. Sie ist von der Überzeugung geleitet, dass den verschiedenen Gestaltungsmitteln des Films substantielle Bedeutungsinhalte zuzuordnen seien, die noch vor Berücksichtigung ihrer Beziehung zum je konkreten Bildinhalt als gegeben vorauszusetzen wären. Auf dem Gebiet der Semiotik haben sich dagegen Theorien über Zeichenstrukturen und Kodesysteme mit universalem Geltungsanspruch für alle Bereiche der Kommunikation entwickelt, sie sind aber in praxi – außerhalb des engeren linguistischen Arbeitsfeldes – noch nicht wesentlich über die Konstruktion hypothetischer Modelle hinausgelangt. In Anbetracht der erwähnten Tatsachen muss man zusammen mit Lohmeier  zu dem Schluss kommen (Lohmeier 1996: XII),

dass einerseits das Methodeninventar der herkömmlichen (quantitativen) Filmanalyse nicht  für die Lösung komplexer Interpretationsprobleme ausreicht, und dass andererseits die filmsemiotischen Modelle nicht in der Lage sind, Forschungen zu fundieren, die sich weniger auf die Dekodierung filmischer Zeichen als vielmehr auf das Verstehen der mit diesen Zeichen erzeugten filmischen Texte und dessen Begründung konzentrieren. 

Von der Annahme ausgehend, dass ein filmischer Text mehr ist als die Summe seiner Zeichen und Zeichenrelationen und dass sein Verstehen Anspruch auf eine textüberschreitende Perspektive erhebt, scheint eine hermeneutische Herangehensweise an den Film sinnvoll.  Ein hermeneutisches Modell der Filminterpretation schließt die Fixierung semantischer Funktionen einzelner kinematographischer Ausdrucksmittel aus und lässt Filmtexte als Ganzheiten betrachten (vgl. Korycińska-Wegner 201).

Der oben erwähnte Vorschlag, das Bedeutungspotential der Bilder samt Elementen der Filmsprache in Musikstücke zu übersetzen, basiert auf der hermeneutischen Herangehensweise an den Film und an den Übersetzungsprozess. Im Mittelpunkt steht dabei die Definition des (Film)textes als einer übersummativen und multiperspektivischen Ganzheit (vgl. Paepcke 1986: 103-4). Dies impliziert, dass es sich bei der Übertragung der filmischen Bilder in Musik nicht darum handelt, den einzelnen visuellen Gestaltungsmitteln des Films substantielle Bedeutungsinhalte – einzelne Noten und Töne zuzuordnen oder für die einzelnen visuellen Inhalte Eins-zu-eins-Entsprechungen zu finden, sondern vielmehr darum, Bilder samt sie konstituierenden Mitteln der Filmsprache in Musikimpressionen einzufangen. Eine derartige Herangehensweise an die Übertragung der Bilder in Musik rückt die Kategorie der erzählenden Programmmusik in den Mittelpunkt.

Für einen ekphrastischen Text ist nicht eine mimetische Genauigkeit von Bedeutung, sondern das ästhetische Erlebnis, die ästhetische Spannung, die im Kontakt mit dem Kunstwerk evoziert wird. Und, wie Robert Więckowski, der Mitbegründer der Stiftung „Kulturen ohne Barrieren“[1] und Konsultant für Audiodeskription betont,  gerade in diesen Merkmalen der Ekphrasis sieht man das Potenzial für die Weiterentwicklung der Audiodeskription  (vgl. Więckowski 2014: 118). Die Stimme von Więckowski  schließt sich der aktuellen Tendenz an, in der die Audiodeskription über die Rolle eines neutralen Werkzeugs der Barrierefreiheit hinausgeht und stattdessen  eine künstlerische Antwort auf das Ausgangsmaterial wird (vgl. u.a. Cavallo 2015; Cavallo und Fryer 2018; Thompson 2018; Kleege 2018; Romero und Dangerfield 2022).

5. Projekt „Erfahrungsräume der Audiodeskription“

In den Jahren 2021-2023 wurde an der Adam-Mickiewicz-Universität in Posen im Institut für Germanische Philologie unter der Leitung der Autorin des vorliegenden Beitrags  das Projekt „Erfahrungsräume der Audiodeskription“ durchgeführt. Im Rahmen des Projekts wurden Audiodeskriptionen für die Filme aus der Sammlung des Bildungszentrums für Blinde in Owińska bei Posen bearbeitet.

Zu einem  der Filme - „Park der Orientierungssinne und unser Bildungszentrum aus der Vogelperspektive“, entstand außer dem Manuskript eine Musikekphrasis, die eine mentale Illustration der Vogelperspektive evozieren soll. Die Komposition wurde von Grzech Piotrowski geschaffen.

Grzech Piotrowski ist ein polnischer Musiker, Komponist, Musikproduzent und Schöpfer der so genannten World Orchestra – eines internationalen Musikprojekts, im Rahmen dessen unter seiner Leitung unterschiedliche Musikgattungen aus weit entfernen Ecken der Erde miteinander verbunden werden.

5.1 Projekt – Umfrage

Den Kulminationspunkt des Projekts bildete die im Dezember 2023 unter den Schülern (24 Befragte) des Bildungszentrums in Owińska durchgeführte Umfrage in Bezug auf die Rezeption der alternativen Audiodeskription – Hörfilm samt Ekphrasis, die die Vogelperspektive illustrierte.

Den Schülern des Bildungszentrums für Blinde in Owińska wurden folgende Fragen gestellt:

1. Was hast du dir während des Hörens des präsentierten Musikstücks vorgestellt?
2. Welche Beschreibung gibt besser den Charakter des Musikstücks wieder?

1. Ich gehe den Gartenweg unter den Pflanzen entlang. Ich gehe an unterschiedlichen Sträuchen, Blumen vorbei.
2. Ich gehe den Gartenweg entlang. Ich schwinge mich auf. Ich bin hoch, immer höher.

3. Welche Wörter passen besser zum präsentierten Musikstück?

1. Last, Begrenztheit, Erde,
2. Leichtigkeit, offener Raum, Luft.

5.2 Die Ergebnisse des   Projekts

Die Fragen wurden von den Schülern folgendermaßen beantwortet:

1. Was hast du dir während des Hörens des präsentierten Musikstücks vorgestellt?

Ich hatte den Eindruck, als ob ich fliegen würde….

Sommer in unserem Park

Wolken, Wolken, Wolken…

Fliegen

Ich spürte Kitzeln im Bauch

Freiheit

Leichtigkeit

Ich schwebte auf wie ein Luftballon

Beruhigung, Entspannung, Musik reguliert die Atmung

Eine gehaltvolle Musik, ruhig und doch bewegend

Ich sehe Berggipfel

Ich  schwebe im Wasser

Ruhe, Spaziergang, Natur

2. Welche Beschreibung gibt besser den Charakter des Musikstücks wieder?

1. Ich gehe den Gartenweg unter den Pflanzen entlang. Ich gehe an unterschiedlichen Sträuchen, Blumen vorbei.
2. Ich gehe den Gartenweg entlang. Ich schwinge mich auf. Ich bin hoch, immer höher.

Ergebnisse der Umfrage:

3. Welche Wörter passen besser zum präsentierten Musikstück?

1. Last, Begrenztheit, Erde,
2. Leichtigkeit, offener Raum, Luft;

Ergebnisse der Umfrage

Andere Antwort – „weder A noch B“

6. Fazit

Wie aus den Ergebnissen der Umfrage hervorgeht, hat die alternative Audiodeskription ein großes Potenzial. Man könnte die Meinung riskieren, dass diese zwei Varianten der Audiodeskription – die konventionelle und alternative Audiodeskription auf dem Gebiet der zugänglichen Filmkultur koexistieren könnten und dadurch das Angebot für blinde und sehbehinderte Rezipienten erweitert werden könnte.

Es handelt sich nicht darum, dass die konventionelle Audiodeskription durch die alternative Variante ersetzt wird: Aus Rücksicht auf den subjektiven Charakter der alternativen Audiodeskription wäre es angebracht, dass die blinden Rezipienten selbst entscheiden können, ob sie die zusätzliche Tonspur mit den Musikekphrasen, die die bestimmten Einzelbilder, Momente oder Protagonisten des Films illustrieren, hören oder auch nicht. Wie Robert Więckowski zu Recht bemerkt (Więckowski 2021: 78): „Jede  Audiodeskription ist eine Art Approximation, eine auf dem visuellen Werk beruhende Variation – jede, indem sie ein Werk der bildenden Kunst der sehbehinderten Person näherbringen will, beschreibt es und zugleich ein bisschen verfälscht, weil sie durch die subjektive Deutung des Übersetzers gekennzeichnet ist”. Somit kann man mit Więckowski zu dem Schluss kommen, dass die Weiterentwicklung der barrierefreien Kultur und das maximale Näherbringen des jeweiligen Werks dann erfolgen könnte, wenn man den blinden und sehbehinderten Rezipienten  mehrere Varianten der Audiodeskription zu demselben Werk zur Verfügung stellen würde (vgl. Więckowski 2021: 78).

Abstract

The audio description service in Lithuania is still relatively new phenomenon. Although there are general guidelines for audio description valid worldwide, it is undeniable that the specificities of audio description also depend on the language or culture of a particular nation. To maximise audience satisfaction with audiovisual content tailored to their needs, research is being carried out to discover the experiences and needs of blind and partially sighted people regarding access to cinema, theatre, and other cultural content. This article aims to internationally present one of these studies conducted in Lithuania.

Keywords: accessibility, audio description, cinema, theatre, blind, partially sighted, Lithuania

1. Introduction

Audio description can generally be understood as verbal commentaries of the visual layer of an audiovisual (or statical visual) product. The Lithuanian audience of blind and partially sighted people experienced a notable surge in the availability of audiovisual content with audio description only in recent years. Previously, cinema or theatre screenings with commentaries covering visual information were rare occurrences, rather than the norm in our country. The first screening of a film with live Lithuanian audio description dates to 2000, and only in 2018 audio description made their debut on Lithuanian television, marking a significant milestone in the development of this field in Lithuania. Although the target audience has a relatively limited experience in viewing the content with audio description, a recent investigation has revealed interesting and valuable trends, which this article aims to reflect on.

The aim of this paper is to discuss the experiences and needs of Lithuanian audience with visual impairments, as revealed by the research conducted in Lithuania during the time span from the beginning of 2021 to the end of 2022. The investigation was related to accessibility of cinema and theatre as well as the application of audio description (hereinafter AD) in this field in Lithuania.

The objectives are as follows:

• to give a brief overview of the history of film and theatre AD in the world and in Lithuania;
• to convey the results of the implemented project “Kultūrinė įtrauktis: audiovizualinių produktų pritaikomumo klausos ir regos neįgaliesiems tyrimas” [Inclusive culture: the study on accessibility of audiovisual products for the visually and hearing impaired] (hereinafter abbreviation “KlaRega”) regarding the needs and experiences of the blind and partially sighted audience[1] in relation to theatre and cinema AD in Lithuania.

The article links critical content analysis with diachronic and statistical analysis as methods for research. The research set out in the project “KlaRega” used a complex methodology. In this article the anonymous survey on the Google Forms platform is relevant, which consisted of several groups of questions and helped both to identify a typical visually impaired user profile and to identify very individual experiences and needs regarding access to cinema and theatre. A direct unstructured interview-discussion method was also used during the research. This step was necessary to find out about the experiences and needs of those blind and partially sighted people who, for some reason, were not able to fill in the above-mentioned electronic questionnaire. This paper explores the main aspects and trends that emerged from the survey and from the live discussion with the blind and partially sighted.

The article consists of several structurally important and logically interlinking parts: a brief historical overview of the Lithuanian film, theatre, and television AD in comparison with foreign countries, a presentation of the previous research on the adaptation of audiovisual content for blind and visually impaired audiences in Lithuania, and the analysis of the process and the results of the newest project “KlaRega”.

2. Audio Description as a mode of Audiovisual Translation

Talking about audiovisual (hereinafter AV) content we clearly understand that the processes of globalisation force the creators to very “traditional” transformation of the products created by them – to the interlingual translation. Sound and image form the unity of an AV work, and if a person does not have access to one of these essential components, the content of the AV product may be extremely limited or inaccessible. In such cases, translation from one sign system to another is used – intersemiotic translation, where, in the strict sense, non-linguistic signs are transformed into linguistic signs. This mode of AV translation is called AD and defined by Bernd Benecke (2004: 78) as “the technique used for making theatre, movies and TV programmes accessible to blind and visually impaired people: an additional narration describes the action, body language, facial expressions, scenery and costumes. The description fits in between the dialogue and does not interfere with important sound and music effects”. For blind and partially sighted people, the dialogue / monologue / polylogue, music and sounds that accompany the action help them understand a film, play or TV programme, but this is usually not enough. The objective verbal descriptions (the AD[2]) of the visual layer of movie / theatre performance (its’ action, place of action, appearance and body language of characters, camera angles, colours and lighting, on-screen texts, etc.) are crucial for the qualitative understanding of AV products.

The style and content of the AD is determined by information about the audience: it is not only the age, education, or level of proficiency in mother tongue and foreign languages that matters. The nature of the visual impairment is a key parameter. The list of visual impairments is extensive, and the causes of poor eyesight or sight loss are wide-ranging: from birth defects, genetic disorders, disease to trauma. It is therefore particularly important to recognise that the target audience of AD is also heterogeneous: some people were born blind, while others lost their sight at an early age and simply do not remember seeing. In fact, they try to transmit visual information to their other senses, associating certain visual things (such as colours) with other senses, for example, with smell, taste, sound, and touch of the object or other familiar objects, associated with that colour. The second important group is the audience, who are currently blind, but still have visual memories (can remember images, colours). The third category is partially sighted: they still remember images and have residual sight. This group is large compared to the blind: according to the European Blind Union (EBU 2022), there are four times more partially sighted than blind people. The diversity of the last group is exceptional: those people may be partially sighted from birth, or they may become visually impaired later in life for various reasons (disease, injury, aging, and so on). These factors determine not only their personal health histories, but also their attitudes towards and access to the environment around them including the cultural content (for example, AV material). This was also revealed by the research carried out during the project “KlaRega”, focusing on AV content with Lithuanian AD.

3. History of filmic Audio Description

In world history, the AD methodology has been used since the beginning of the 20th century. As witnessed by the National Institute for the Blind’s journal, “The Beacon”[3], the first official attempt to audio describe the film took place in the UK in January 1917. In Spain AD of films started to be provided after the Civil War, in the late 1940s. The commentaries provided by journalist Gerardo Esteban were transmitted on radio frequencies once per week (Orero 2007: 179). As Marc Vighetti states, the first known attempt to audio describe cinema in France was documented in 1988, and consequently in 1989 the first official presentation of the process of AD at the Cannes Film Festival took place (Vighetti 2021). Inspired by the festival, a group of German enthusiasts known as “Münchner Filmbeschreibergruppe” [Munich film describers’ group] also prepared commentaries for some American comedy movies in the same year in Germany (Kirf 2008: 19–20; Arma 2011: 45). According to Saveria Arma, the first screening of a film with AD in UK was organised in 2002 (Arma 2011: 44). In Baltic states Estonia and Latvia, who together with Lithuania belong to the Baltic Sea region and share relevant historical moments, AD services for films started accordingly in 2009[4] and in 2022[5].

Any AD events were officially recorded until the very end of 20th century in Lithuania. The first documented live AD attempt of a film and of AD in general dates back to the year 2000 – in April the film “Tylos ir tamsos šalis” [Land of Silence and Darkness] (1971, directed by German film director Werner Herzog) was live audio described by the Lithuanian filmmaker Audrius Stonys at the cinema Lietuva in Vilnius (Mykolaitytė 2000). In 2006 the second attempt of cinema AD followed – the film “Black Sun”[6]. In 2012, the first Lithuanian film with pre-recorded AD was screened in the former Lithuanian Library for the Blind in several Lithuanian towns: this film – “Anarchija Žirmūnuose” [Anarchy in Žirmūnai] by Saulius Drunga – can also be viewed on DVD, giving the opportunity to experience the adapted product repeatedly, and not only, as before, to attend a one-off screening. To date, 8 audio described films (7 of them with pre-recorded AD) have been screened at Lithuanian film festivals (for more details read Niedzviegienė 2023: 7–8).

Already in year 2000 Lithuania was aware of the progress of other countries and intended to introduce AD into television, but it was not enough for the television authorities to take the initiative alone. The availability of improved consumer electronics such as TVs with stereo sound was essential to meet the needs of viewers (Mykolaitytė 2000). Until autumn 2012, Lithuanian television was analogue. This was the main obstacle preventing TV broadcasting with AD in Lithuania although some other countries counted the history of TV programs with AD for more than 10 years. The introduction of digital TV in Lithuania did not change anything for the next 5 years (2012–17): our TV content remained without AD until the beginning of 2018, when the first attempt of the AD of a TV product took place – the film “Gražuolė” [The Beautiful Girl] (directed by Arūnas Žebriūnas) was audio described and made public. The very detailed overview of the TV production with AD in Lithuania can be found in the other article of the author (see Niedzviegienė 2023). Only in the last years, Lithuania has made significant advancements in its efforts to offer real-time ADs for movies. It should be highlighted that there is a growing availability of films through Lithuanian National Radio and Television (hereinafter LRT), as they now broadcast movies with AD (about 25 hours annually).

4. Audio described theatre performances

Till 2000, the technical possibility of providing live commentary in Lithuanian theatres using devices for simultaneous interpretation did exist. During this time, the Lithuanian blind and partially sighted community harboured a desire to experience at least one audio described theatre performance annually. At the time, the process of adapting such performance seemed easier, requiring fewer technical resources, and relying more on human skill (Mykolaitytė 2000). However, there has been no tangible action.

Meanwhile, in other countries, theatre AD has been performed for some time. The official start of theatre AD can be traced back to the USA: during a briefing on theatre accessibility held in 1980 by Wayne White, the manager of the Arena Stage in Washington, D.C., an important development took place. Margaret Pfanstiehl, who was blind and managed the Metropolitan Washington Ear (a radio reading service for the blind), along with her husband Cody, initiated the world’s first AD program for performing arts as a result of this briefing. Remarkably, this program is still active today (Snyder 2008: 192). According to Arma, theatrical performances with AD took place in the UK in the mid-1980s for the first time (Arma 2011: 43–45). In Baltic state Estonia AD services for theatre performances started in 2010; no data about theatre AD in Latvia found.

Since 2012, the Lithuanian Union of the Blind and Visually Impaired has been running the project “Audio Description in Theatre”, which aims to adapt the cultural environment for people with visual impairments and to provide them with the opportunity to experience professional theatre art: in 2012, for the first time in Lithuania a performance with live AD – Chekhov’s “Uncle Vanya” – was presented for the blind audience broadcasting the AD through wireless headphones. The audio describer of this performance was the Lithuanian actress Adrija Čepaitė. In direct cooperation between Vilnius theatres and the Lithuanian Union of the Blind and Visually Impaired, a total of 6 performances were adapted. Five theatre performances were shown in cooperation with the Lithuanian National Drama Theatre (Vilnius): the above mentioned “Uncle Vanya” (author A. Chekhov, directed by E. Lacascade), “An Enemy of the People” (author H. Ibsen, directed by J. Vaitkus), “Liliom” (author F. Molnár, directed by L. Bagossy), “Cathedral” (author J. Marcinkevičius, directed by O. Koršunovas), “Divine Comedy” (author D. Alighieri, directed by E. Nekrošius). One another performance with AD was shown at the Russian Drama Theatre in Vilnius – the tragedy of V. Shakespeare “King Lear” (directed by J. Vaitkus). All above mentioned ADs were performed by the actress Adrija Čepaitė.

This was followed by the involvement of the Kaunas Faculty of Vilnius University in 2015, which is the only educational institution in Lithuania that teaches and researches the intricacies of AD. All other performances so far adapted for the blind and partially sighted audience in Lithuania appeared in collaboration between the Lithuanian Union of the Blind and Visually Impaired and Vilnius University Kaunas Faculty members or alumni (for systemised information see Table 1).

Table 1: The statistics of theatre performances with AD in Lithuania[13]

Until beginning of 2018 theatre AD was the most developed AD field in our country with more than ten audio described performances since 2012. The national broadcaster LRT later came to dominate with audio described films.

At this point switches the article from practical activities of AD to a scientific context and seeks to overview the research on the topic of cinema and theatre accessibility for the Lithuanian blind and partially sighted audience carried out by various national cultural and educational institutions.

5. Cinema and theatre Audio Description research in Lithuania

Research on the accessibility of AV material in Lithuania has been increasing recently. Several small-scale studies were carried out by the Lithuanian Audiosensory Library (formerly the Lithuanian Library for the Blind) in 2018[14]. In the same year, 2018, a large-scale study focusing on the needs of persons with disabilities (including the needs of the blind and visually impaired community) in the field of culture was carried out by the program “Kurk Lietuvai” [Create Lithuania][15]. Vilnius University, the largest and oldest university in Lithuania, has also been involved in practical and scientific activities and research related to AD since 2014: research on cinema and theatre AD is carried out here by BA and MA level students, as well as by lecturers. To date, a number of bachelor’s and master’s theses have been successfully defended, and several scientific articles have been published in Lithuanian and English, for example, thematising the adaptation of AV content to the visually impaired audience (Niedzviegienė and Kirejeva 2015), discussing the development of AD in Lithuania and in German-speaking countries (Alosevičienė and Niedzviegienė 2017), analysing the history and peculiarities of theatre AD with special regard to audio introduction guidelines (Niedzviegienė 2017), and extensively overviewing audio described cinema and TV production in Lithuania (Niedzviegienė 2023). The concept for the project “KlaRega” originated exactly at the above-mentioned Vilnius University, and it was subsequently executed with success by two lecturers and researchers of this educational institution.

It is stated that linguistic and cultural peculiarities of each country, as well as the specific needs of the target audience determine partially the methodology of AD. This has been proven during a scientific study conducted from February 2021 to December 2022 in framework of the Lithuanian project “KlaRega”, funded by the Research Council of Lithuania [Lietuvos mokslo taryba]. The qualitative and quantitative research conducted by Assoc. Prof. Dr L. Niedzviegienė (Vilnius University, specialising in AD) and Assoc. Prof. Dr Jurgita Kerevičienė (Vilnius University, specialising in SDH) has reflected the experiences and needs of the audience as well as demonstrated the importance of national AD guidelines. As a result, national AD and SDH guidelines for accessible films and theatre performances were prepared (see Kerevičienė and Niedzviegienė 2022a) and publicly presented during the national seminar to the Lithuanian cultural and art authorities (film and theatre makers and professionals involved in its adaptation process or specialists of film and theatre AD). In addition, a scientific study has been written (see Kerevičienė and Niedzviegienė 2022b). The most relevant findings and outcomes of the research related to the visually impaired audience will be presented in this article.

6. The scope of the study presented in this article

This paper will analyse data gathered through online questionnaire[16] (consisting of 50 questions) on the framework of the project “KlaRega” to get insights from blind and partially sighted individuals regarding their needs and experiences related to the adaptation of AV products in Lithuania.

The main part of the questionnaire was related to the qualitative and quantitative analysis of AV products adapted to the audience and available in Lithuanian theatres, as well as on TV and various film platforms. This part presented short (up to one minute) AV product examples adapted for the target audience according to the general rules of the AD methodology. The AD texts of the selected theatre play and films were created by the author of this article in cooperation with some co-authors. The screening of the excerpts should have allowed an assessment of audience needs, preferences, and interests. Based on these, guidelines for the adaptation of AV products for the target audience should have been draw up.

The research inquired about participants’ experiences and opinions regarding the dissemination of information about adapted AV products as well as common challenges related to the physical and social environment. However, issues of accessibility of the information (cinema or theatre websites, social media, etc.), physical environment (in terms of physical location and building of particular cinema or theatre) and social environment (the skills and knowledge of the cinema / theatre stuff about how to communicate with the blind of partially sighted and how to help them) will not be discussed directly and extensive in this article. For obvious reasons, this paper is limited to an analysis of the issue of accessibility of AV content through AD.

Links to the online questionnaires were shared with the Lithuanian Union of the Blind and Visually Impaired and the former Lithuanian Library for the Blind as well as through the particular social network Facebook group of Lithuanian blind and partially sighted[17]. Additional live meetings-discussions were held to collect additional feedback. However, the data from these meetings (with over 20 participants) will not be addressed here. Instead, only the data from online surveys will be discussed, as these surveys follow a structured format with fixed questions. The number of people involved in the online questionnaire was very small – 37 people. It has been observed in practice that the community tends to be passive in contributing to AD research. In addition, there are not many people in Lithuania who officially[18] consider themselves part of the blind and partially sighted community.

Having explained where the limits of this research lie, it is now possible to turn to the core of the analysis.

7. Lithuanian audience needs and experiences: films

The survey for people with visual impairments asked about their general information, habits, experiences, and preferences related to watching films and attending theatre. It also investigated how easy it is for them to access cinema and theatre by giving examples of movies and plays with AD and asking for their feedback. The last section of the questionnaire allowed participants to share additional insights.

The questionnaire’s initial section addressed demographics, education, and disability levels. Questions about specialised education for the blind aimed to gauge adaptability to the modern information environment and explore the correlation between social adaptation and AV product use. The general section of the survey enabled us to construct a profile of a typical participant, revealing that the survey was mainly attended by individuals who were over 50 years old (45.9 per cent), predominantly women (78.4 per cent), and mainly residents of the capital city, Vilnius (40.5 per cent). Additionally, respondents were uniformly Lithuanian speakers, with the majority (56.8 per cent) having attended regular schools for sighted individuals and possessing higher education qualifications (59.5 per cent). A significant portion of participants were partially sighted (54.1 per cent), visually impaired from birth (43.2 per cent), and raised in sighted families (94.6 per cent) (Kerevičienė and Niedzviegienė 2022b: 51–52).

While a considerable proportion confirmed that they like watching films (67.6 per cent), a majority rarely visited cinemas, with 45.9 per cent doing so only every few years. The TV content was more popular among blind and partially sighted individuals, with 37.8 per cent watching TV for more than 2 hours daily, favouring informative programs (40.5 per cent). Feature films, social documentaries, talk shows or educational shows as most loved genre were mentioned by three participants each (for each genre 8.1 per cent of the respondents).

Regarding viewing habits related to AV production equipment, certain trends emerged. A notable 37.8 per cent reported not watching films on a computer at all, and 40.5 per cent never used online platforms for film and program viewing. Among those who did use such platforms, the LRT media library[19] was the most popular, mentioned by 43.2 per cent of respondents (Kerevičienė and Niedzviegienė 2022b: 52–53).

Summarising the second survey section on film-related habits and preferences, around two-thirds (67.6 per cent) of respondents enjoy films, but almost half (45.9 per cent) rarely go to the cinema, opting for occasional visits every few years. Daily television consumption exceeds two hours for 37.8 per cent of participants, with a preference for informational programs (40.5 per cent). Additionally, 43.2 per cent access films and recordings through the LRT media library, while 40.5 per cent do not use online platforms. Foreign platforms attract less interest, with only 18.9 per cent expressing curiosity. Approximately 37.8 per cent do not watch films on computers, while 29.7 per cent spend 1–2 hours on it daily. A significant 78.4 per cent have experienced films with AD, with 43.2 per cent watching them at least once a year, eliciting positive emotions for 75.7 per cent of the respondents. Furthermore, 73 per cent are content with the pace of AD in already watched films. 56.8 per cent of all the participants value detailed information about characters’ body language provided by the AD text for learning and discovery (Kerevičienė and Niedzviegienė 2022b: 55–56).

8. Lithuanian audience needs and experiences: theatre

As far as the theatre AD is concerned, it should be mentioned that this genre is subject to many of the same rules and norms that apply to cinema or other visual products of a dynamic nature. However, there are also some very specific aspects that have been explored in the third part of the questionnaire. In addition, an example of an audio described theatre performance recording included in the survey[20] gave the opportunity for the respondents to express their opinion on the AD of the excerpt.

Among the respondents there were two who do not like and do not watch plays. As many as 73 per cent like watching plays. 40.5 per cent go to the theatre several times a year, 27 per cent very rarely, only every few years, 18.9 per cent go once a year, and 8.1 per cent once a month. There were also some who have never been to the theatre or go several times a month.

The largest share of the audience prefers dramatic plays (37.8 per cent), followed by comedies (16.2 per cent), musical plays (10.8 per cent), and plays dealing with social issues (8.1 per cent). The majority (56.8 per cent) do not watch recordings of performances online. Almost a third said they watch freely available recordings online (29.7 per cent). The rest were not aware of this option or said they pay to watch a recording or live stream of a performance. As many as 62.2 per cent of the respondents have seen performances with AD (live or online), and almost a third (29.7 per cent) have not yet watched theatre with AD and are planning to do so. 8.1 per cent of the participants said they did not know audio described performances exist. When asked about the frequency of watching performances with AD, two trends emerged: 40.5 per cent watch at least once a year, while almost a quarter (24.3 per cent) watch it once every few years. 27 per cent said they had never seen audio described performances before.

In terms of their experience with audio described performances, nearly two-thirds (59.5 per cent) expressed solely positive emotions. About 32.4 per cent mentioned not having had the chance to attend such a performance but expressed a desire to do so. Some participants felt the AD format might be challenging, while others, with residual vision, found issues with premature AD or discrepancies between the video and AD. Regarding the pace of AD, 48.6 per cent were satisfied with it during their past experiences. 32.4 per cent had no opinion due to not having witnessed an AD performance. Approximately 13.5 per cent reported occasional difficulty hearing everything but stated it did not impact their understanding. Several responses expressed confusion due to information overload or stated that sometimes the AD text is not what it should be about.

It turned out that almost half of the respondents (45.9 per cent) would be / are happy to attend in the audio introduction (a pre-show presentation of the actors and scenery), if possible. Only 16.2 per cent of the respondents do not find the audio introduction important, only the performance. When asked about the importance of stage lighting information in AD text, 43.2 per cent of respondents said it is important and should be included. 13.5 per cent felt stage lighting is unimportant and can be excluded from AD.

As mentioned before, this section deals with audience needs and experiences specifically related to theatre AD, that is, those elements that are not covered in cinema AD. For example, a noisy soundtrack is less of an issue for cinema, as it is easily solved by adjusting the soundtracks, as their parameters can be easily changed. Meanwhile, the volume of the music or other sounds in the performance cannot be changed because, in our experience so far, the directors are reluctant to do so. Extremely noisy episodes with intense background music or noises pose challenges in AD. However, 40.5 per cent of respondents believe that even in the presence of loud noise or music, it is necessary to convey crucial stage action information. This indicates that the audience both needs and expects an AD in various situations during a performance.

9. The preferred style of Audio Description

The questionnaire proceeded to inquire about which style of AD the viewer prefers: neutral or subjective, detailed or more abstract. Some questions regarding the specific text elements that make one style more effective or engaging than others were included, encouraging respondents to reflect on their personal experiences with different formats of AD.

The survey responses show that it is not always easy for blind viewers to identify characters from the voice alone: while a significant proportion (37.8 per cent) said that they have no difficulty in identifying characters in this way, as many as 43.2 per cent are happy to hear clues to a character’s identity in the AD text. 29.7 per cent stated sometimes having difficulties by understanding which character is speaking.

The issue of audio describing body language is highly debated from both scientific and practical perspectives, making the audience’s responses particularly valuable: 54.1 per cent said that they like to receive detailed information about characters’ body language because in this way they can learn new things. 24.3 per cent most prefer when a character’s body language is not described and instead their mental state and emotions are directly named. One admitted that he is annoyed by the descriptions of the body language of the characters, because he does not understand them anyway. Another said that there should be as much talk about body language as is necessary to reveal the plot. Some of the respondents could not answer the question because they had not seen this type of performance or had no opinion.

The naming of colours in the AD text is also often debated. Some partially sighted people said they are happy to get some information about colours (27 per cent). Another 18.9 per cent have residual vision that helps them to see colours, so they do not think it is necessary to mention colours in the text. 16.2 per cent said they have been blind since birth, but colours are important to them and they like it when colours are mentioned in the text. Several other respondents, who have not been able to see since adulthood or have lost their sight recently, confirmed the same. Only one respondent said that he / she has been blind since birth, so colours do not mean anything to him / her, and colour’ names are annoying in text. One respondent stated that colours have an archetypal meaning and can therefore be significant in a film or a play. Thus, the results of the survey revealed that blind and partially sighted people are in favour of the mention of colours in the AD text.

After the general questions asking about the preferred AD style, the questionnaire included links to four short extracts of AV products. Here, both about general trends and specifics related to the audio described excerpts of the films and of the performance were asked. The questions about the identification of characters, colours or pace of the AD revealed similar trends as in the previous parts of the survey. However, there were some additional questions, the answers to which are relevant in the context of this article and therefore will be discussed below.

When demonstrating excerpt from selected film[21] and discussing how much AD text should be included and how much of the original film soundtrack should be left uncovered, the audience was sharply divided: 35.1 per cent of respondents would prefer the silent scenes to be audio described as fully as possible, filling all the time when the actors are not speaking. 21.6 per cent opposed this position, arguing that silent scenes should also have silent pauses during which the film’s music or other sounds could be heard. 37.8 per cent are in favour of describing silent scenes in as much detail as possible, but not overwhelming important sounds in the original soundtrack.

10. The relevance of the sound layer

In addition to the visual layer of a product, the audio layer is of particular importance for AD. Here we should mention both the peculiarities of the original soundtrack (for example, a very noisy film track that interferes with the AD and must be adjusted) and the fact that the original music and background sounds of the film must be largely preserved in order not to lose the character of the product. In addition, some film soundtracks are not monolingual and foreign interludes are not always translated, so the sound layer in the film’s soundtrack also poses a challenge. The voice of the narrator, who will voice the AD soundtrack or give live AD in the theatre, also seems to be important.

In AV products, the atmosphere is frequently enhanced through the incorporation of background music. However, it is crucial not to ruin the overall atmosphere with AD. Sometimes these layers intersect, so it was important to find out the needs of the Lithuanian audience in terms of matching the AD text to the music and songs in films and performances. Almost half of the respondents (45.9 per cent) said that only music without lyrics can be covered with AD. Their reasoning was that lyrics are important, and adding an AD track over them could interfere with the audience’s ability to hear and understand the lyrics. Just over a quarter (27 per cent) took a similar view, stating that it would be great if as little background music as possible could be covered with AD text. The same proportion of respondents (27 per cent) feel that it is important to get as much information as possible and are therefore in favour of using the musical background to insert AD.

Voice quality is a crucial element of audio content. The audience of blind and partially sighted people had very clear expectations about the voice of the person reading the AD text. The most popular choices in the questionnaire were:

• It is important to me that the voice is pleasing to the ear (81.1 per cent).
• I only want to hear correct diction and accents (32.4 per cent).
• The gender of the reader is not important (29.7 per cent).

The other options given are less frequently selected:

• The most important thing is to have as many products with AD as possible and the quality of the voice is not important (18.9 per cent).
• The gender of the reader of the AD text should be the same as that of the main character (16.2 per cent).
• It does not matter that the diction is poor, and the reading is not properly accented, it does not bother me (10.8 per cent).

In the open-ended answer box, respondents mentioned that they had not seen such performances (one respondent), that the reader should stand out from the other voices in the performance “so that you don’t mix it with the troupe of actors”. Another respondent modified one of the suggested answers and said that both – the quality of the voice and the maximum number of products with AD – are important.

Another selected and demonstrated excerpt from the film[22] contained multilingual inserts (when Russian and German are spoken). 52.8 per cent of viewers appreciated having a voice reading translations for these parts, as it helped them understand. 33.3 per cent were convinced that they knew Russian well enough not to need a translation, and that the voice-over only interfered. A total of 12 respondents chose this answer option: two of them are in the 31–40 age group, one is in the 41–50 age group, and as many as nine are in the 50+ age group. 5.6 per cent of the respondents stated that they speak German well enough so that translation is not necessary, and that the voice-over interferes. These persons belong to the age group 31–40 years.

11. Features of Audio Description for children

The interviewees were also introduced to an excerpt from the children’s film[23]. Minors did not take part in the survey, only adult respondents, so responses related to AD products for children should be treated with caution as they are not the responses of children or adolescents. However, some of today’s visually impaired people have had disabilities since birth or childhood, so their insights are also valuable. The most significant statement for the largest proportion of respondents (40.5 per cent) was that the pace of reading of AD text should be similar for both children and adults. 32.4 per cent of respondents confirmed that children’s film AD can be moderated in the use of more complex words, as this allows them to learn new things. Almost the same number of respondents (29.7 per cent) believed the opposite, that the AD text of children’s films should avoid any more complex words or terms. As regards the conveyance of emotions, 29.7 per cent of all respondents believe that short and clear commentaries, no matter how subjective, are more useful for young audiences, so, here again, we could conclude that the audience prefers the direct naming of an emotion to the description of body language. Just over a fifth (21.6 per cent) chose the statement that in children’s films the gender of the person reading the AD text is not important. However, 8.1 per cent considered the gender of the narrator reading AD text to be important and tended to agree that the gender of the main character and of the AD voice should be different in children’s films, as this makes it easier for children to understand where the characters are speaking and where the AD is.

As already mentioned, the full study covered not only the audience’s experiences regarding the accessibility of AV products through AD, but also the attitudes towards the accessibility of the informational, physical, and social environment. The discoveries from the research should give an impetus to the growth of cultural accessibility to the visually impaired community in Lithuania.

12. Summary and conclusion

Visually impaired individuals in Lithuania still face challenges when it comes to accessing AV content. Previously, AD in theatre productions was the most developed area until January 2019. Currently, two to three theatre performances with AD are shown annually in Lithuania for the blind and partially sighted. With the airing of the TV series “Laisvės kaina. Disidentai” [The Price of Freedom. Dissidents] on the national broadcaster LRT in January 2019, the number of TV programs with AD has surpassed audio described theatre performances. The LRT is improving accessibility by broadcasting cinema productions with AD. They currently broadcast around 25 hours of blind-friendly content per year, which includes full-length films or TV series episodes on a weekly or bi-weekly basis. However, achieving a certain amount of legally defined AV output with AD is not enough to maximise audience satisfaction. It is necessary to ensure that the adaptation meets the audience’s expectations, and this requires an understanding of audience needs.

The research, conducted in Lithuania in 2021–2022 by implementing the project “KlaRega” (the full title “Kultūrinė įtrauktis: audiovizualinių produktų pritaikomumo klausos ir regos neįgaliesiems tyrimas” [Inclusive culture: the study on accessibility of audiovisual products for the visually and hearing impaired]), revealed that there is a lack of suitable AV productions with AD, especially for children, and that the number of audio described AV material from foreign producers is also extremely limited. Foreign films were for a long time not audio described in Lithuania due to restrictions on editing foreign films and the short validity period of licenses obtained for foreign films. This made it economically unviable as the audio described movie cannot be viewed once the license expires. Therefore, only classic Lithuanian films or films by contemporary Lithuanian directors received AD for visually impaired viewers on our national television, which was almost the only provider of AD service for several years. Now is the situation changing, because of private VOD platforms and film festivals, they became main agents ensuring that foreign films reach Lithuania’s blind and partially sighted.

Two respondents expressed a clear belief that high-quality ADs, characterised by being fast, accurate, and not too loud, are not only beneficial for visually impaired individuals but also do not disturb those without visual impairments. Conversely, slow-paced descriptions were regarded as frustrating, even for the main target audience. It is gratifying that audiences are satisfied with the current quality of the products already adapted.

The interviewees reaffirmed their statements that AD is most essential during silent scenes with minimal or no dialogue. In other cases, the progression of the film or performance can be mainly understood through the characters’ speech or sounds. For 35.1 per cent of the respondents it is important that silent scenes are audio described in as much detail as possible, filling all the time before the actors speak. One participant has stated that explanations of actions obvious from the main soundtrack, such as the sound of a squeaky door closing, can be annoying for the visually impaired audience and has recommended to utilise the available time for naming the action or the characters’ emotion rather than explaining the source of the sound.

Feedback from discussions on film excerpts revealed that Lithuanian visually impaired audience appreciates detailed descriptions of characters, their body language, movements, and visual details of shown object such as colours and fabric patterns. The respondents mostly prefer subjective naming of characters’ emotions in audio described content. They also favour information that reveals the identities of the characters. Attention should be paid to the complexity and expressiveness of the language used in the AD texts. This is particularly important in AV products audio described for children.

When theatre content is complex, viewers prepare by seeking primary sources, like the original book or drama script, to aid comprehension. Context is crucial, and audiences appreciate audio introductions before the performance. During the show, visually impaired viewers desire ongoing AD of stage lighting, the mention of character names, and crucial actions, even in the presence of loud noises or music.

The quality of the AD voice is important to the audience: pleasant timbre, clear diction, and proper accentuation matter, unlike gender of the AD voice. Some respondents expressed dissatisfaction with the sound quality of some previously watched products, stating that the AD voice was sometimes unclear (described as “muffled”). The comfortable AD pace is also crucial, allowing the audience to comprehend the spoken text. According to the audience, in certain cases, the AD may partially cover the musical background, if deemed necessary. Audio subtitling for the multilingual fragments were found useful to clarify AV content.

These useful opinions and insights will enable AD practitioners to better tailor AV productions to the visually impaired community in Lithuania.

Abstract

This paper delves into the analysis of the adaptation of Atwood’s novel The Handmaid’s Tale for the screen. Bruce Miller’s TV adaptation of the book on Hulu in 2017 inspired a revival of interest in Atwood’s work, topping the world’s bestseller lists and winning a Golden Globe and eight Primetime Emmy Awards. In this study, aimed at investigating the quality of the Italian subtitled version of the TV show, an approach inspired by a set of taxonomies within Descriptive Translation Studies (DTS) related to domestication and foreignization (Venuti 1995), translation equivalence (Nida 1964; Toury 1995), subtitling quality (Gottlieb 2009), and good subtitling practice (Carroll and Ivarsson 1998) was employed in order to identify the main categories adopted, while Remael and Robert’s quality parameters in subtitling (2016) also contributed to the discussion of the results of our study. Based on the results of the quality assessment, it was determined that the Italian subtitled version of The Handmaid’s Tale exhibited high quality in terms of content and transfer (e.g., accuracy, completeness, logic), and style and form (e.g., grammar, spelling, and punctuation), with only a few minor errors observed in terms of formatting, some major issues related to the technical dimension (e.g., spotting), and adherence to the dubbing script, a peculiar phenomenon we have extensively described.

Keywords: audiovisual translation, streaming video on demand, subtitling, Translation Studies, TV adaptation

Introduction^[1]

Since the 1970s, scholars have written profusely about Margaret Atwood; yet most research has solely focused on her literary production, rather than on the screen adaptations of her books. More recently, the regained popularity of Atwood’s dystopian novel, The Handmaid’s Tale (1985), produced by Bruce Miller’s TV adaptation on Hulu in 2017, has sparked a resurgence of interest in her work, topping bestseller lists around the world. The second and third seasons, an elaboration of Atwood’s vision that covers the story beyond the ending of the novel (Kröller 2021), followed in 2018 and 2019. While season 4, announced for 2020, was delayed because of the COVID-19 pandemic and then released in 2021, season 5 was broadcast in 2022. Therefore, the present article will cover seasons 1-5, while a final season of the show (6) is currently being shot with a slight delay due to the strike of the Writers Guild of America (WGA) supported by the Screen Actors Guild – American Federation of Television and Radio Artists (SAG-AFTRA). 

After premiering on 26 April 2017 with a cameo by Atwood herself, the series became a global phenomenon. It concurred with Donald Trump’s election as 45th US president and the rise of the #MeToo movement (referenced in episode 9 season 2), ad-hoc timing to showcase dystopian America narrated from a female perspective and ruled by the theocratic Republic of Gilead (Bertrand 2019). The great success of Miller’s adaptation with its most iconic markers, the maiden’s red coat and white wings, symbols of female empowerment, has enthused Atwood to respond by transforming Offred’s story from a novel of repression into a narrative of rebellion with her 2019 sequel, The Testaments (Danneil 2022).

The popularity of The Handmaid’s Tale goes well beyond the realm of entertainment: in 2018, a group of feminists dressed as Handmaids lined up to protest the U.S. Supreme Court confirmation of Brett Kavanaugh following allegations of sexual misconduct (Moeggenberg and Solomon, 2018), and the same year, a supportive letter from Atwood was read at a rally of Handmaids in Argentina.

The ‘visual impact of the Handmaids’ red dresses and winged white caps was also widely acknowledged as “powerful and important” when women began to appear in the costume at rallies across the United States, the United Kingdom, Ireland, Poland, Croatia, Costa Rica, Australia, and Argentina in protest against evidence of systemic misogyny. (Kröller 2021: 196)

The visual appeal of The Handmaid’s Tale is largely derived from the faithful adaptation of some key passages into ‘painterly, cinematic, and other intertextual allusions’ (ibid.: 194). Such aspects of the adaptation of The Handmaid’s Tales for the media will be included in our analysis along with a list of categories, such as the treatment of religious Newspeak (Orwell 1949) – a peculiarity in the exchanges of the people of Gilead – conveying the imbalanced relationships between men and women through specific gendered expressions, as well as the register and style of the show, and the powerful influence of the Italian dubbed version on the creation of the subtitled version of The Handmaid’s Tale, a rather unusual practice in interlingual subtitling.

This study is focused on the analysis of the Italian subtitled version of The Handmaid’s Tale Seasons 1-5 available on Amazon Prime Video in order to evaluate the transposition of content across a specific set of categories. This paper will ultimately report on the assessment of the overall quality of the linguistic and technical dimensions of the subtitled version through a series of examples.

1. The Handmaid's Tale (1985)

Atwood's dystopian novel The Handmaid's Tale (1985) takes place in an imaginary future United States after a coup d'état led by a group of men known as the Sons of Jacob. The newly established ruling class has successfully instituted restrictions, abrogated essential individual liberties, and supplanted the former democratic framework with a dystopian regime that subjugates women and deprives them of their autonomy, all while enforcing Christian legislation and claiming divine authority.    

The primary goal of the new regime is officially to increase natality in order to reverse the decline in birth rates that is allegedly caused by the Western world's prioritisation of women's professions over parenthood and pollution chemicals. As a result, the Sons of Jacob prohibits women from participating in paid labour and classified them based on their social status and fertility. Young women from wealthy families are assigned the role of Commanders' Wives, whereas those from poorer backgrounds are married to labourers and relegated to the status of Econowives. Infertile women can either become Marthas, who serve as household servants for the Commanders, or Unwomen, who are sent to the Colonies to dispose of radioactive waste. Women who are still fertile but have had abortions or were previously married to divorced men, are considered offenders, and are given the opportunity to atone for their sins by becoming Handmaids (incubators for the Commanders). Young women who refuse to obey and defy authority, are forced to work in a state-run brothel for Commanders known as the Jezebel’s, or declared Unwomen. A distinct group of women is referred to as the Aunts. They collaborate with the regime to control and coerce other women in return for the assurance of not being deported to the colonies or compelled a to assume the status of sexual objects. They are also allowed to read and enjoy some degree of power.

The narrator of The Handmaid's Tale is Offred, the central character, who has been deprived of all agency, identity, and even her name. In a society that values women exclusively based on their reproductive capabilities, she is diminished to a mere instrument of procreation. Offred describes her transformation into a Handmaid, including a period of ‘education’ at the Rachel and Leah Centre. Aunt Lydia and the other Aunts at this facility teach women how to become Handmaids in the biblical sense, as described in the book of Genesis, where Rachel begs her husband to give her a child through her handmaid, Bilah. This biblical passage is read prior to the Ceremony, a strictly regulated monthly ritual (during the Handmaids' fertile days) in which the Commander rapes the Handmaid while his wife holds her down.

Offred recalls her previous life with her husband, Luke, and their daughter, whom she wishes to rescue after being kidnapped. Before the coup, she had a career, personal funds, and access to information. This, however, is no longer the case. She is left with nothing, not even a proper name, as her given name, Offred, comes from her Commander, Fred Waterford, whose wife, Serena Joy, coerces her into having sexual relations with Nick, the Commander's driver, out of fear that Fred is sterile. Offred agrees, understanding that if she does not conceive, she may be sent to the colonies. As a result, Offred forms a covert partnership with Nick, and they conceive a child together. A black van arrives at the end of the novel to pick  Offred up. Nick informs her that the people in the van have arrived to her aid; she accepts his word and travels to either her place of confinement or her place of release.

Atwood's own take on the novel has evolved. In 1985, she assured her readers that her story was realistic, delving into details that had already happened in our world. However, in 2003, she stated that the novel possesses a universal appeal because  “the inclination towards tyranny, the wielding of absolute power by the few over the many, knows no ideological boundaries and is not confined to one time or space" (Atwood, 2003: online). Similarly, Hutcheon affirms that reading The Handmaid's Tale is always unique and rewarding because of the novel's ability to match "the varied social, historical, political, and aesthetic contexts in which it occurred" (2021: 252) to those of the reader. As a result, following the airing of season 1, which retells the novel's plot, there was a desire for a more empowering ending for Offred's (dis)adventures — and those of her fellow Handmaids — and the new series were written to describe Offred's attempts to protect her daughters and her  becoming the leader of the rebellion against Gilead. In the television series, Offred and her daughters regain their names and identities: Offred was previously known as June Osbourne and her first daughter as Hannah, and the child she bore with Nick as Nichole. June eventually succeeds in smuggling Nichole to Canada, where Luke and Moira look after her. Later, June escapes to Canada and is granted asylum. She even manages to bring Fred and Serena to court with the help of Marc Tuello, who represents the United States and tries to help refugees in Canada while also coordinating the resistance in Gilead.

Miller authored the initial five seasons in close collaboration with Atwood as consulting producer, as he himself has stated in numerous interviews (see Wigler 2018; Feldman 2019). The final series will be written by Yahlin Chang and Eric Tuchman, who collaborated with Miller on previous seasons and will co-showrun The Handmaid’s Tale season six.

2. Subtitling The Handmaid's Tale for an Italian Audience

The practice of subtitling poses a set of key challenges due to its peculiar nature, particularly when it comes to the adaptation of a novel such as The Handmaid’s Tale for the screen. Interlingual subtitling – from a source language into a target language – is a mode of transfer in which a range of technical, linguistic, and cultural features are enmeshed to convey the original message through a twofold approach, a mix of verbal and non-verbal content, and through a double channel, the audio-visual. As a result, the set of limitations subtitlers must face when working on an audiovisual program is rather challenging in terms of spatial and temporal constraints.

Díaz Cintas and Remael argue that “subtitling [is] an unusual form of translation because it is added to the source text [and] it also stands out as a unique translational genre because it renders speech in writing, in a counter-movement from film dialogue, which is written to be spoken” (2020: 88). It is also peculiar in that both the original dialogue, or source text (ST), and the subtitled version, or target text (TT), coexist on the screen during the viewing process, as opposed to dubbing, for example, which replaces the original soundtrack altogether, zeroing the chance of being exposed to criticism: the original cannot be heard, therefore the audience cannot compare it to the translation provided. Among the main norms and conventions to be followed as per Amazon Prime Video style guides, subtitlers working from English into Italian have a specific number of lines (2), characters per line (42) and positioning (centre-justified at the bottom of the screen). The minimum duration of a subtitle event is five-sixths of a second (20 frames for 24fps), and the maximum duration is set to 7 seconds. The reading speed for adult programs is 17 characters per second, the average reading speed commonly used on other streaming platforms such as Netflix and Disney Plus nowadays. However, in the Italian subtitled version of The Handmaid’s Tale that we have examined, the aforementioned conventions are not consistently adhered to,  especially regarding synchronisation and line break treatment.

In the following section, the methodology implemented for this research study will be provided along with its related theoretical framework that enabled the creation of a set of macro and micro categories described below.

2.1 Methodology, data collection and theoretical framework

This section is focused on the methodology and theoretical framework devised prior to the analysis of the Italian subtitled version of The Handmaid’s Tale seasons 1-5. As far as the methodology is concerned, the audiovisual material was retrieved on Amazon Prime Video from both a British and an Italian account to verify that the Italian subtitles provided were exactly the same. Once verified, we noted that while on the Italian Amazon Prime Video account all the episodes were available for free, on the UK account season 4 was not included in the Prime membership and had to be purchased. We also realised that for season 4 UK, no Italian subtitles were provided, while this option was instead available on the Italian account.

The first phase of the data collection entailed the viewing of each episode in English with Italian subtitles. It was a guided process organised into a set of tasks: 1. collecting data on the linguistic and extra-linguistic aspects of the Italian subtitles (e.g., containing grammatical, morphological or spelling mistakes, mistranslations, etc.); 2. when an issue was found, the player was paused to switch the audio from English to Italian as a way to check whether the same issue or incongruence was retrievable in the dubbed version; 3. if that was the case, taking notes of the chunk of text of the Italian dubbing script related to the specific subtitle under scrutiny; 4. progressively populating a pre-prepared Excel file divided into a set of categories containing all the notes as described in task 1, 2 and 3 (see fig. 1 below).

Fig. 1 – Data collection phase

Before embarking on the note-taking part, we looked closely at the technical aspects of the subtitled version of each episode. A screenshot was taken whenever the screen presented a visible anomaly, for example the inappropriateness of the line breaks produced, untranslated information manifested into missing subtitles, or the incorrect implementation of standard conventions. The screenshots were subsequently stored in an online repository for the purpose of illustrating specific examples about the multi-semiotic nature of subtitling. 

The analysis of the data collected led us to devise four macro categories namely, 1. linguistic and extra-linguistic aspects, 2. technical dimension, 3. influence of dubbing, 4. religious newspeak, which report on the linguistic and technical issues  encountered during the data collection phase. They were selected according  to the specific areas of interest identified during the post-viewing phase of this study which are at the core of Audiovisual Translation: the linguistic and technical aspects of this distinct field of Translation Studies. More specifically, a framework of reference was devised, and a first macro category (MC1) identified for the translation of linguistic issues and extra-linguistic references related to customs and traditions, food, institutions, people, places etc. (Pedersen 2011) belonging to the dystopian TV show. A second category (MC2) was created to address a range of technical issues, such as line break treatment and standard conventions, sometimes of a more visual nature, singled out during the analysis of the information collected and the screenshots saved during the data collection process. A third one, (MC3), was employed to categorise the influence exerted by the Italian dubbed version of the show.

A fourth and final category (MC4) was devised for the analysis of Gilead’s Religious Newspeak (GRN), a set of greetings, formulaic and gendered expressions, which Margaret Atwood created, reminiscent of Orwell's 1984 (1949), as a powerful means to depict the dystopian society of Gilead. This linguistic device, which intertwines biblical passages with fabricated everyday religious expressions typical of the characters' dialogues in both the novel and its television adaptation, not only aligns with the aims of Gilead's dictatorship but also fortifies its substantial control over linguistic and historical narratives. This reveals the regime's duplicitous strategies to manipulate the populace and stifle dissent, thereby being pivotal to the study of both the novel and the television series.

Table 1 – Macro and micro categories

The 12 micro categories (mC) selected were identified at a later stage and were based both on the typologies of error collected and in line with the theoretical framework adopted for the research study, keeping in mind that the main objective of the investigation was the assessment of the overall quality of the Italian subtitled version. Thus, the resulting structure shows a total of 12 mCs inspired by a set of  taxonomies of translation and subtitling strategies within Descriptive Translation Studies (DTS) aiming at “describing the phenomena of translation” (Assis Rosa 2010: 94) and by the approach on subtitling quality proposed by Remael and Roberts (2016), which have contributed to discussing the results of the analysis carried out. Remael and Roberts’ (2016) established four groups of translation quality parameters: (1) content and meaning transfer (including accuracy, completeness, logic); (2) grammar, spelling, and punctuation; (3) readability (i.e. ease of comprehension and coherence between individual subtitles); (4) appropriateness (socio-cultural features of the audience). We have retained the parameters 1-2-3 and left out the fourth as not relevant to the study and have added a fourth layer related to the technical dimension, which was not envisaged in their approach as not related to AVT specifically but translation in general, and thus of paramount relevance in the present study.

3 The Handmaid's Tale subtitled version into Italian: a selection of case studies

This section is meant to provide a selection of examples based on MC 1-4 and related to 12 micro categories (mCs) as discussed in section 2 of this paper. Each section will shed light on a set of specific case studies based on four topics featured in each MC (linguistic and extra-linguistic references; technical issues; influence exerted by Italian dubbing; religious newspeak) in order to offer clear-cut examples able to show how a specific linguistic item has been translated, the impact the issue encountered had on the overall viewing experience, along with the strategies employed by subtitlers for a particular linguistic or technical solution. Each case study also discusses potential options that could have been adopted to improve the quality of the overall output.

3.1 Linguistic and cultural references

Within the first macro category (MC1), we have identified the following micro categories: mistranslations, profanities, generalisation, idioms, slang and fixed expressions, calques, grammar, acronyms, additions, songs and rhymes, untranslated words, and register (see tab. 1 for a summary of both macro and micro categories).

The first example of the section will examine a mistranslation issue found in “Faithful” (season 1, episode 5). June and Luke, her future husband, have just met. In this scene, they are involved in a romantic exchange where they imagine their first sexual encounter. In order to understand the context of the excerpt under analysis, please see table 2 below:

Table 2 - Scene I.

Right after these exchanges, the subtitles appearing on screen are as follows (tab. 3):

Table 3 – Case study 1, example 1.

While in the ST exchange Luke makes a joke referring to the time needed to “get a restraining order” (in case June changes her mind and regards him as a stalker rather than a lover), the Italian subtitle  renders this as “for the perfect alibi”, a solution that manages to keep the linguistic domain of the English expression but changes the joke suggesting the possibility that he will be murdered as a result of their romantic encounter. It is not clear why the expression was so heavily manipulated to achieve this level of mistranslation. One might believe that the reading speed would not allow for a faithful translation, but this is not the case: the subtitle could have been literally rendered as “per un ordine restrittivo” which respects the constraints imposed by the medium. Yet, the most problematic aspect of this choice is that the following subtitle, “Yes, file the papers” is translated as “fill in the papers”, which has a totally different meaning. The subtitler seems to have misinterpreted the scene, resulting in a cascading effect of multiple translation errors stemming from the initial misunderstanding.

In the second scene selected from season 1, episode 2, “Birth Day”, we propose an exchange between June and a fellow Handmaid, Ofglen, who tells her (tab. 4 below):

Table 4 – Case study 1, example 2.

The Italian version shows a rather strong manipulation of the original expression with: “È pericoloso fidarsi di chiunque, specialmente di una lesbica traditrice” (back translation: it is dangerous to trust anyone, especially a lesbian traitor). The reference to “carpet-munching” (a person who performs cunnilingus) in Italian is totally absent, toning down the strong expression, a choice that ends up spoiling the dialogue of the original flavour and style of the show. Nevertheless, in line with this tendency, in the third scene selected from season 1, episode 4, “Other Women”, we provide an example of how profanities are generally treated in the Italian subtitled version of The Handmaid’s Tale (see tab. 5). Here, June is in chains, detained in the Red Centre after she has just been captured for trying to escape. Aunt Lydia comes into her room and turns to June, saying: “Blessed be the fruit. I see you’ve eaten well, Offred.” referring to her name as Handmaid (that is ‘property of Fred Waterford’, her commander). June then replies:

Table 5 – Case study 1, example 3.

The Italian subtitle reads “You know my name very well” with no trace of the profanity in the original. This tendency to censor or tone down swear words, or strong and loaded expression (Bucaria 2024), is maintained throughout the seasons in the Italian version although we know from Amazon Prime Video guidelines that subtitlers should not censor or manipulate the text and are required to render the dialogue as faithfully as possible into the target language. It may be a case of self-censorship on the part of subtitlers, a phenomenon that seems, at times, to be still rather frequent, at various levels, in some adaptations and subtitled versions for the screen from English into Italian. However, little do we know about all the LSPs selected for the subtitling of The Handmaid’s Tale seasons 1-5. The only information available is related to the dubbing studios: Studio Hamburg Synchron GMSH (German), Candiani Dubbing Studio (Spanish); MG Estudios (Brazilian) and the many professionals credited at the end of the episodes (see table 6).

Table 6 – Translators’ credits

3.2 Technical issues

In the seasons analysed in this study (1-5), some instances of asynchrony can be retrieved. They mainly involve the IN-time of the Italian subtitles provided by Amazon Prime Video: more specifically in season 2, episode 1, 3 and 4 and season 4, episode 10. While on the whole said instances might be regarded as minor issues sparsely found in the show, it is worth noting that season 2, episode 3 stands out for the almost total lack of synchrony of the Italian subtitled version.

Season 2, episode 3, “Baggage”, opens with the protagonist, June, hiding from her commander after escaping. She is left alone in a building where she is jogging and thinking to herself. Her inner dialogue starts  as she remembers her mother. From the first couple of minutes, one realises that the IN-time of the Italian version is completely out of synchrony: specifically, the IN of the subtitles comes rather later than due, roughly 7-10 frames. However, the asynchrony is not consistent throughout the episode: it can be appreciated for a long portion of the video, then the synchrony returns to normal for a while (roughly 1-3 frames) to return to complete asynchrony.

The technical issue seems only to affect the Italian subtitles, while the English SDH subtitles provided on the streaming platform are in perfect harmony with the video. There are many reasons for these issues to happen. While we were unable to get in touch with our contacts at Amazon Prime when we reached out to them, we have reflected on what might have gone wrong in the post-production phase. Allegedly, the subtitle version provided for season 2, episode 3 was produced for a video file format with a different frame rate, and or video encoding and extension, resulting in a visible offset for the audience. The subtitle file under analysis, belonging to a different video file/format, would actually be perfect but only when associated with the correct video content, yet not the one available on the streaming platform. We propose this interpretation with some degree of certainty as the IN-time is almost systematically set 7-10 frames after the correct IN-time, creating a consistent pattern in the asynchrony which shows that it is not a singularity in the spotting, but rather a matter of offset that could have been fixed during the Quality Check (QC). As for the QC, in this case it was either not performed or not performed accurately enough to avoid such a blatant mistake.

Another category of technical issues we have encountered during our analysis regards the line break treatment in the Italian subtitled version throughout the whole show. As shown in tab.1, 32 instances of wrong line breaks were retrieved, a rather high number considering the relevance and popularity of the TV series. In season 1, episode 6, “A Woman’s Place”, shown in tab.6, the subtitle reads:

Table 7 – Case study 2, example 1.

Fig. 2 below shows how the last word of the first line contains the article “una” (BT: “one”) and the second line contains the noun referred to it “di quelle purghe anti lesbiche”  (BT:  of the dyke purges). Subtitling conventions establish that the article and noun are not to be separated as “it is advisable for each subtitle [line] to make sense in itself” (Díaz Cintas and Remael 2020: 171), to allow the viewers to read comfortably.

              Fig. 2 – Example 1 (S01-E01)

A similar issue was found in season 2, episode 3 (fig.3) where the second line is visibly much shorter than the first, the lines are not self-contained, and the line break could have easily been improved as per Amazon Prime Video’s Italian (Italy) Timed Text Style Guide (as well as Netflix’s Timed Text Guide) which clearly state that line breaks should ideally match a logical and grammatical break in the dialogue. This norm is systematically broken in specific sets of subtitles throughout season 1-5.

Fig. 3 - Example 2 (S02-E03)

The third category devised within the technical dimension of this study is related to the use of italics. Italics in subtitling serve a specific purpose, and the conventions related to its use are rather unanimous in the localization industry. In line with Díaz Cintas and Remael (2020), Amazon Prime Video’s Italian (Italy) Timed Text Style Guide clearly states that dialogue heard through electronic media, such as a phone, radio, or television should be displayed in italics. In the example provided below from season 2, episode 8, “Women’s Work” (fig. 3), Moira and Luke are at home in Canada watching TV as the Waterford are travelling to Canada for the first time after the coup. The Italian subtitles should have been displayed in italics as they recount the voice of the journalist during the news. Yet, as shown in fig. 4, the font of the subtitles is regular as if the voices were coming from the characters rather than from the TV set.

Fig. 4 – Example 3 (S02E08)

3.3  Influence exerted by dubbing

Interestingly, MC1 and MC2 are intertwined with MC3, as some of the mistakes found are sometimes due to a strict adherence to the dubbing script created for The Handmaid’s Tale, a pattern that can be observed throughout seasons 1-3. It is a rather peculiar feature of the Italian subtitled version of the show, the almost total reliance on the dubbing script created for Amazon Prime Video. It appears as if the subtitled version was not performed from scratch but rather as a deliberate manipulation of the Italian dubbed version, a modus operandi simply not aligned with good subtitling practices.

Dubbing, famously referred to as traduction totale (Cary 1960), is a mode of transfer presenting a set of constraints that greatly differ from subtitling limitations (see section 2), as they serve a different purpose: “dubbing’s ultimate fabrication is to make viewers believe that the characters on screen share the same language as the viewer” (Díaz Cintas and Remael 2020: 8). In order to create this illusion, dubbing replaces the original soundtrack altogether while respecting three types of synchrony:

1. isochrony, to ensure that the source and the target text dialogue coincide in length;

2. kinetic synchrony, to make sure that the target text is in harmony with the action and characters’ gestures;

3. lip synchrony, to match the translated sounds to the original ones, especially when shown in close-up.

As a result, using the dubbing script of The Handmaid’s Tale as a guide to perform the subtitled version has generated a plethora of errors, mistranslations, and utter translation mistakes. In table 1, for each micro category we have reported the errors due to the reliance on the dubbing script: additions (1 out of 2); generalisation (4 out of 10); mistranslations (8 out 36); profanities (1 out of 9); songs and rhymes (1 out of 2); idioms, slang & fixed expressions (1 out 9); calques ( 1 out 6).

The reasons behind this peculiar practice may lie in the way professionals were hired for the subtitling project in question. Interestingly, the national contract regulating dubbing practices in Italy^[2] also regulates subtitling rates (fig. 5) which amount to 50% of the total rate provided for dubbing adapters. Such a low rate would potentially call for extreme measures on the part of professionals working in the dubbing industry, such as reworking and or modifying the translation used for the dubbing script instead of producing a subtitled version from scratch.

Fig. 5 – Subtitling rates

The only instance in which such a practice works in the audiovisual translation (AVT) industry is when producing intralingual subtitles for the deaf and hard of hearing (SDH) for a dubbed version: in this case the subtitles have to match the Italian soundtrack, as the main skopos of the translation (Vermeer 1989) is shifted to meet the needs of a specific audience. In the case of The Handmaid’s Tale (seasons 1-3), the practice employed is pointless in terms of efficiency and efficacy overall, as shown in the examples below (tab. 8).

Tab. 8 Case study 3, example 1.

However, from season 3 episode 9 on, the reliance on the dubbing script suddenly stops as the dubbing script solutions are disregarded suggesting that the subtitling of the show might have been outsourced and commissioned to a different Language Service Provider (LSP) the name and nature of whom still remains unknown.

3.4 Gilead’s Religious Newspeak

The fourth and final category (MC4) investigates the rendering of Gilead's Religious Newspeak (GRN), a set of greetings, formulaic and gendered expressions unique to the novel and TV show, in which biblical passages are mixed with made-up day-to-day religious formulas and utterances that characterise the characters' exchanges. The main features of Gilead’s Religious Newspeak (GRN) in both the television series adaptation and the novel encompass, for instance, the employment of neologisms (e.g., ‘Birthmobile’), standardised expressions (e.g., ‘we've been sent good weather’), and the dehumanising categorisation of individuals (e.g., Marthas, Wives, Eyes). In addition, the employment of gender-specific terms (e.g., Aunts, Handmaids), as well as the introduction of possessive forms (e.g., Offred and Ofglen), strongly emphasise the sexism of the regime and the control exerted by the dictatorship over women. Through the reuse of religious expressions (e.g., Ceremonies, Salvagings), symbolism and allusions to the Bible (e.g., Jezebel), the government efficiently establishes its power over its subjects, impacting not just their speech but also their own perception of reality and identity.

The process of translating these items into subtitles was the focus of our study in the 4th macro category (MC4). An evident breach of translation rules is observed in the rendering of the Gileadean name of the Handmaids; for example, Offred becomes ‘Difred’ in both the Italian subtitles and dubbing script, while proper names of people are typically not translated. This apparent violation further emphasises the effect of Atwood’s invention: by translating a first name, it is made transparent that, despite being capitalised, it is not a proper name intended to identify an individual but rather a common name used to refer to all the Handmaids who belong to a certain commander. Afterall, Offred is just a patronymic derived from the preposition ‘of’ and the Commander’s name, Fred. Therefore, it is accurately rendered as ‘Difred’ into Italian.

The translation of neologisms, shown in table 9 below, must have posed particular difficulties for the subtitlers, who had to deal with the limitations imposed by the medium and ensure that the Italian renditions would convey a similar effect in the target language. These neologisms are capitalised both in the English version of the novel and in the Italian, in order to signify their uniqueness in the Gileadian world.

Tab. 9 - Case study 4, Example 1 – Neologisms.

The cases of ‘Birthmobile’ and ‘Particicution’ are particularly significant. The Italian version for ‘Birthmobile’ is ‘Partomobile’, a morphological calque formed by combining ‘birth’ (parto) and ‘mobile’ (mobile). The employed word order is essential, as in Italian, the premodifier typically follows the noun, resulting in connotations of variation. In this case, however, the translation is effective as it aligns with other compound terms ending in ‘mobile’, which are  loanwords from English, including automobile, and Batmobile, for example. The only difference with the novel is that in the subtitled version, ‘partomobile’ – as well as other Newspeak words - is not capitalized. ‘Particicution’ refers to the act of executing someone accused of heinous crimes, such as rape, by means of a circle of Handmaids. This distressing moment, reminiscent of the Orwellian 'Two Minutes Hate' is defined in English by a word that combines participation (partecipazione) and execution (esecuzione), thus lessening the inherent intensity of the second term’s lexical root that exposes the brutality of the activity.

The Italian translation used in the TV series, both in the dubbed and subtitled version, is ‘partecicuzione’. This translation is, once again, a morphological calque. However, the Italian translations of the term in Atwood’s novels differ: while in Il racconto dell’Ancella (2017 [1985] Milano: Ponte alle Grazie translated by Camillo Pennati) it is translated as ‘Partecipazione’, in I Testamenti (2019 Milano: Ponte alle Grazie translated by Guido Calza), written and subsequently translated after the release of the first series, it is rendered as ‘Particicuzione’, carrying a more unsettling connotation, and suggesting a potential etymology composed of "party" + "esecuzione" (execution), although it could simply be an adaptation of the English spelling.

The primary characteristic of GRN is the inhibition of independent thought through the use of standardised language, frequently derived from Christian religious terminology, seemingly intended to emphasise the purity of the speaker, of which a sample is given in the table here below.

Tab. 10 - Case study 4, example 2 – Formulae.

A number of biblically derived phrases are deftly altered in the television adaptation to emphasise that the regime's doublethink is of the same distorted nature as the language it uses. Among the instances listed in table 10 above, ‘Nolite te Bastardes Carborundorum’ stands out. This fabricated expression in mock Latin, which has assumed an autonomous existence as a feminist rallying cry for women, could be literally translated as “don’t let the bastards grind you down.” But as Atwood herself said, various forms of the phrase go back much further than The Handmaid’s Tale itself, as the motto was already a joke during her academic years (Bradley 2017). Differently from the other formulae, and quite curiously considering its relevance in the novel, in the TV series, and in the outside world, ‘Nolite te Bastardes Carborundorum'’ has not been translated consistently throughout the seasons. In series 1, episode 4, “The Wilderness",  when Fred translates the motto for the first time to Offred, both in the subtitles and in the dubbed version, it is rendered as “non permettere ai bastardi di schiacciarti” (BT: don’t let the bastards crush you). On the contrary, in series 5, episode 1, when Mark Tuello says it to June in Canada while trying to console her, it is translated with a more empathetic and gentler “non si abbatta per quei bastardi” (BT: don't beat yourself up over those bastards), indirectly empowering and making her the protagonist of the action. This translation is not influenced by the dubbing script (“non si lasci sottomettere dai bastardi”) or by the novel “che i bastardi non ti schiaccino.”

The first time the motto appears in the TV series (S01 E04) is when Offred finds it written in Latin inside her wardrobe. No translation is given into English or Italian. The same happens in series 2, episode 13 when Offred is running away from the Waterfords home with baby Nichole. But, in series 4, episode 10, where it appears as a graffiti under the dead body of Fred Waterford, killed in a sort of Particicution by former Handmaids and other women who had been captive in Gilead, it is translated as “non permettere che i bastardi ti annientino” (BT: don't let the bastards annihilate you). The verb ‘’annihilate” (destroy, nullify) is much more intense than the former translations.

4. Discussion

A revised version of Remael and Roberts’ approach (2016) was employed for the quality assessment made up of four groups of parameters: (1) content and meaning transfer (including accuracy, completeness, and logic); (2) grammar, spelling, and punctuation; (3) readability (i.e., ease of comprehension and coherence between individual subtitles) and (4) technical dimension (e.g., spotting, line break). Against this background, we argue that of the 12 micro categories related to the 4 macro categories emerging from the data collection phase, the majority of errors encountered belong to MC1 (linguistic and extra-linguistic issues) related to Remael and Roberts’ first and second categories: a total of 36 severe mistakes. To MC2 (technical issues and conventions) are attributed 43 mistakes overall: we have aligned MC2 to a new category as Remael and Roberts have not devised a specific parameter to evaluate technical issues. All in all, seven micro categories show a range of issues belonging to MC3 (influence of dubbing) complicating the picture even further. As for the fourth MC (religious newspeak), we have selected a range of peculiarities that we do not categorise as issues tout court and to which we have devoted a separate section.

Interestingly, MC1, and MC2 are intertwined with MC3 as some of the mistakes found, at times, are due to a strict adherence to the dubbing script, something that can be retrieved rather frequently throughout seasons 1-3. However, from season 3 episode 9 on, the modus operandi appears to be changed and the dubbing script solutions are disregarded, which suggests that the subtitling of the show might have been outsourced and commissioned to a different Language Service Provider (LSP). As for MC2 related to parameter 4, it is important to stress that most of the issues logged in this study, are mainly found in season 2, episode 3, where both the IN-time and OUT-time of the Italian subtitles are utterly out of synchrony, suggesting that the spotting was not produced following the minimum standards of accuracy and quality expected from the streaming platform and as per Amazon Prime Style Guide. From episode 4 on, the technical issues seemingly disappear to reappear in a scattered and rather contained fashion, throughout the rest of the seasons.

5. Conclusion

This research provides insight into the complex difficulty of translating a work that is nuanced while still maintaining the tone and intention of the original. To this end, we have carried out a qualitative analysis of the Italian subtitled version of the TV adaptation of Atwood's The Handmaid's Tale, originally broadcast on Hulu and later on various other streaming platforms and among them Amazon Prime Video, which is the version selected for the analysis.

The framework devised for this study falls within DTS and is inspired by a variety of approaches related to domestication and foreignization (Venuti 1995), translation equivalence (Nida 1964; Toury 1995), subtitling quality (Gottlieb 2009) and good subtitling practice (Carroll and Ivarsson 1998). Their taxonomies have inspired the creation of a system of macro- and micro-categories able to identify the main issues found in the collection data phase, both linguistic and technical in nature, while Remael and Robert’s revised approach on quality parameters in subtitling (2016) served as our benchmark for the discussion of the results.

While the subtitling into Italian for The Handmaid's Tale is of very good quality overall, the sets of examples provided illustrate the challenges that translators encountered and how some of the solutions offered could have been improved to ensure that the translation complies with the norms and conventions to be followed both linguistically and technically speaking. In addition, it is worth noting that, as shown in table 6, the number of professionals involved in the translation of seasons 1-5 is significant: in seasons 3 and 4, ten different people in total worked at the translation of The Handmaid's Tale. These numbers suggest that the workflow was particularly inefficient having so many hands on the project at the same time, which hinders translation fluency, style and terminological standardisation. However, it is also true that season 5 of The Handmaid's Tale was solely translated by one individual, Giovanna Noce, and it shows in the overall quality of the Italian subtitled version of the TV series, which is exceptional as compared to the rest of the seasons analysed. As for season 1-3, of which we have less information as regards the translators involved, the influence exerted by the dubbed version of the show played a central role in the overall lower quality of the subtitled version analysed, as discussed in this paper (section 3.3).

About the author(s)

Simonetta Falchi (PhD) is currently Associate Professor of English Language and Translation at the University of Sassari, Sardinia (Italy). In 2012–2017, she was a researcher of English Literature and Culture at the University of Sassari, Sardinia (Italy). In 2008–2009, she was an academic visitor in Cambridge at the Faculty of English and Lucy Cavendish College. She has been a member of the editorial board of the Journal of English Studies, The Grove (University of Jaen), and the cross-cultural journal Annali della Facoltà di Lingue e Letterature Straniere (Università di Sassari). She is a member of the Governing Board of the I-Land Research Center and the National Correspondent for the ESSE Gender Studies Network. Her main research interest is the intertwining of technology, language, literature, and translation. On these topics, she has so far published three books and several articles.

Serenella Massidda is Senior Researcher at the University of Chieti Pescara (IT) and Honorary Research Fellow at University College London (UK). Professional translator and subtitler, she holds a European Doctorate in Audiovisual Translation (AVT). She is executive board member of the European Association for Studies in Screen Translation (ESIST), Media for All Conference member of Scientific Committee, and Editorial Board Member of Linguistica Antverpiensia (LANS – TTS). She is the author of Audiovisual Translation in the Digital Age, the Italian Fansubbing Phenomenon, and has published widely on audiovisual translation and subtitling. She has carried out international research projects and academic consultancies with the media localization industry on subtitling quality, technologies and professional development such as the Netflix Hermes Project (2016–17), the Isub Project (2020–22), and the AVT Pro Subtitling Certification Project currently under development (2022–).

Email: [please login or register to view author's email address]

Abstract

Keywords:

1. Cross-cultural variation through translation

Understanding cross-cultural variation in human communication is highly important and has attracted the attention of scholars in contrastive linguistics and translation studies (Hatim and Mason 1990). It has beneficial social effects in that it helps appreciate the ‘other’ and avoid misunderstandings in social interaction. A field which cuts across cross-cultural variation is cross-cultural pragmatics, namely, the study of implied meaning in what we say and how we say it, cross-culturally. The assumption, in this special issue, is that intercultural variation may be fruitfully studied through translation practice and inform both translation training and EFL. This is because translator insight, which adjusts messages to target context conventions, seems to be a rich resource for identifying and researching intercultural variation.

Translation and pragmatics, or pragmatics through translation, has attracted the attention of scholars early enough (Hickey 1998) and the interest proliferated (Tipton and Desilla 2019, Locher and Sidiropoulou 2021, Sidiropoulou 2021, Desilla 2024) as new areas of intercultural transfer came under the lens of pragmatics (for instance, oral-to-oral interpreting, written to oral interpreting [sight translation], oral-to-gesture [sign interpreting], subtitling, dubbing etc., Dayter, Locher and Messerli 2023).

Recent accounts of the interaction between translation and pragmatics talk about ‘cultural filters’ which result from contrastive analysis (House 2018). House (2021) suggests that the importance of context for both translation and pragmatics is evidence of the close relation between them. Likewise, House and Kádár (2021, 2022) include translation studies in the areas which fall within the cross-cultural pragmatics paradigm and outline its principles, which are presented here in order to show the relevance of the articles in this special issue to cross-cultural pragmatics.

(1) Bottom-up research (House and Kádár 2022: 152): all articles in this special issue take a bottom-up approach to the data.

(2) A “multimethod approach to researching language use” (House and Kádár 2022: 152). Αll articles in this special issue use both an etic approach to the data (describing the analyst’s view) and an emic approach (other speakers’ view) through questionnaires. Troy McConachy and Helen Spencer-Oatey (2021) highlight the significance of metapragmatic comments in sociopragmatic research, namely, the evaluative and explanatory comments that language users make in relation to particular features of interaction. Such comments reveal ideologies and may be received through questionnaires, interviews, newspaper articles and media reports. Questionnaires in these papers usually ask respondents to explain their choices, for the insight they can offer to the research perspectives.

(3) “Relying on interrelated but distinct units of analysis and finite typologies of these units […] the researcher is advised to identify a particular unit of analysis to examine linguaculturally embedded data” (House and Kádár 2022: 154). All articles in this special issue focus on a pragmatic phenomenon and examine how it is realized cross-culturally through translation, namely, what may be threatening cross-culturally.

(4) “Cross-cultural pragmaticians may pursue interest in intracultural and intralinguistic variations of languages” (House and Kádár 2022: 154). When the articles examine more than one version of the data within the same language, an intracultural comparison occurs between discursive options used diachronically.  Regarding the cross-cultural dimension, House and Kádár (2022) suggest that “[t]he more typologically distant these languages are, the more challenging it may be to contrastively examine them” (2022: 154-55). The special issue contrasts English and Greek, which favour different politeness orientations (English: negative, Greek: positive in certain genres) and brings Russian, which differentiates itself in many ways, into the picture (as a source or target language).

(5) “Relying on corpora and the Principle of Comparability”. All articles use mini data sets to draw their conclusions.

(6) The last principle of cross-cultural pragmatics refers to “[u]sing linguistically-based terminology”, in contrast to terms like “ideology”, “values”, and “identity” formation (House and Kádár 2022: 156). When the articles in this special issue refer to variation in “ideology”, “values” and “identity”, variation is described through linguistically-based terminology.

Another recent account of the interaction between translation and pragmatics, and what matters most in studying it, points to three layers of consideration, which may allow safe conclusions in the study of translation and pragmatics. These are the level of (a) ‘mediality’ (how the medium may affect the way transfer is performed), (b) ‘participant roles’ (addressees, speakers, bystanders, overhearers etc.) and (c) ‘relational work’ in the framework of interpersonal pragmatics: “Relational work is the linguistic and multimodal “work” individuals invest in negotiating relationships with others” (Locher & Watts, 2005: 10).

The special issue takes into consideration mediality, in that it examines print or online information and how it is transferred, occasionally multimodally. The whole of the special issue is about relational work, that is, how text producers negotiate their relationship with audiences in non-/fictional texts (where the intention is to avoid or implicitly enhance threat).

Parallel corpora, like the diachronic mini-corpora samples in this special issue, compare and contrast regularities of languages (Zanettin 2014) and are “a reasonably reliable repository of all the features of a language” Tognini Bonelli 2010: 20). In answering the question ‘what can corpora tell us about pragmatics?’, Rühlemann (2010) points to discourse markers and speech act expressions. The parallel data sets in this special issue ‘compare and contrast regularities’ which pertain to interpersonal dynamics, in non-/threatening situations.

For instance, threat may need to be enhanced in a target version for achieving appropriateness; it may arise unintentionally in a target context, if a source context prefers enhanced aggression and the transfer is ‘ST oriented’ (Toury 1995); it may arise out of standard  terminology for a social minority, academic or medical ST conventions may be threatening if transferred intact in another cultural context; threat may appear because it may be intended in a target context; the relational dynamics between interlocutors, in time y, may be threatening in time x, because of societal change in the meantime; enhanced threat in a target version, shaping the identity of a female figure by a male translator, may be too offensive in the ears of a female translator, out of solidarity for women. These are quite a few instances of threat the special issue raises awareness of, in translation practice.

2. The individual contributions

Genre is of utmost importance in shaping discourse (Trosborg 1997). The articles in this special issue examine a variety of translated genres, both fictional (Part II) and non-fictional (Part I).

2.1 Non-Fiction

The article ‘Manipulating Τhreat in Μedical Αdvertising’ advances perception of medical discourse norms cross-culturally. Following the traditional concern about the doctor-patient relationship (Pendleton and Hasler 1983, Heath 1986/2006, Von Raffler-Engel 1989) and how discourse is structured, Daphne Charalampopoulou examines the relational dynamics between expert/doctor and potential patient, in medical leaflets of a private hospital, in Athens, advertising services the hospital offers for various medical conditions. The paper concludes that the interpersonal distance (high-power distance) favoured in the Greek version of the data, between hospital experts and potential patients, would have been offensive if transferred in English, which instead favours lower power-distance features (than Greek) in certain contexts. The paper draws attention to variation in the make-up of medical advertising across English-Greek, in pragmatically relevant ways, which need to be taken into consideration in translation practice.

The special issue goes on with examining im/politeness in how EU English-Greek legal texts referring to disability shape the identity of disabled people. Aimilia Papadopoulou and Maria Sidiropoulou (‘Representing Disability in English and Greek Legal Discourse’) suggest that the ‘medical’ model of disability is gradually abandoned in the Greek version of discourses on disability, in favour of the ‘social’ model, despite the fact that the formality and high-power distance between doctor-patient in Greek is rather highly appreciated and may impede adhering to the social model. Questionnaire respondents seemed to appreciate disability terms (as polite) which, however, are not favoured in the Greek version of EU disability discourses. For instance, the AμεΑ abbreviation in Greek (for ‘Persons with Disabilities’) avoids potential offensive overtones sedimented within everyday terms of disability. The article advances understanding of potential challenges in managing offensiveness associated with disability terms across English-Greek and pairs with the article “Portraying Intellectual Disability through Translating Fiction” by Vasiliki Papaconstantinou in Part II of this issue in that they both focus on the identity of disabled persons, in different genres.

The next article, ‘Translating Academia: Shaping the Academic Author’ by Chrysoula Gatsiou, deals with the identity of the academic author as shaped by the English source and Greek target version of an academic coursebook on history. The author examines (a) author-expert and reader-student interpersonal distance in academic interaction and (b) the degree of certainty with which author views are presented. In Hofstede, Hofstede and Minkov’s (2010) framework these two features correspond to the dimensions of high-/low ‘power distance’ and high-/low- ‘uncertainty avoidance’. Discursive options in Greek are supposed to heighten the social status of the academic author, as a highly esteemed member of the community in Greek. A low-power distance approach (as in English) would have been perceived as threatening in Greek and a low uncertainty avoidance perspective would not have been persuasive enough in Greek. The paper reveals discoursal features which improve texture in a target context and should be taken into account for the transfer to be successful.

The paper ‘Shaping Political Ideologies in the UK BBC and the Russian BBC News Service’ examines implicatures following from BBC English and BBC Russian articles on Liz Truss and Rishi Sunak. Pigi Haidouli examines multimodal material appearing in the two outlets and shows how ‘transediting’ processes in the news attack the face of Liz Truss in the Russian version, while protecting the face of Rishi Sunak. The paper shows how news discourses undergo two mediation processes (Chouliaraki 2012), one in interpreting events in the source language and manipulating threat as intended, and another one where manipulation of threat through translation may yield varied connotations.

2.2 Fiction

Studying pragmatics through fiction has already attracted the attention of scholars (Locher and Jucker 2017), and translation seems to be another arena where the pragmatics of fiction may be studied cross-culturally. Part II of the special issue examines:

- two Greek versions of a piece of Russian literature for children with respect to how im/politeness is rendered intra-culturally,

- two Greek versions of a horror novel and what narratives of im/politeness and threat they give voice to,

- how im/politeness and threat plays out in shaping disability in two Greek versions in fiction,

- two versions of two Shakespearian plays and how translators handle im/politeness and threat in them diachronically, and

- how a Greek version of Russian love poetry at the beginning of the 20^th century reshapes the interpersonal dynamics between the poet lover and the beloved nowadays, as contrasted to three English versions of the Russian love poem.

Eleni Piperidou’s article ‘Translating Threat and Power Distance in Pushkin’s The Fisherman and the Goldfish’ examines two Greek target versions and two English target versions of Alexander Pushkin’s story ‘The Fisherman and the Goldfish’ (1833). The two pairs of versions allow examination of threat manipulated intra-culturally and cross-culturally in order to appeal to relevant audiences.  Results show that, as time goes by, threat and aggression are enhanced in both English and Greek contexts, with the visual material showing instances of minimized interpersonal distance and hierarchical relations, in agreement with the verbal material. The article shows diachronic modification in the relationship between the fisherman and his wife and a higher level of aggression and threat on the wife’s part.

The article ‘Carmilla into Greek: Translating Horror and Queerness’ by Maria Episkopou examines how the Gothic subculture plays out in two Greek versions of Joseph Sheridan Le Fanu’s novel ‘Carmilla’ (1872) and how threatening queerness may be to target Greek audiences. Results show that the earlier Greek version rather focused on the supernatural aspect of the story and silenced gothic horror and queer sexuality as too offensive for its audiences; the latest version tends to focus on psychological and more violent aspects of the self, highlighting queerness, which appears not as threatening as it earlier was. The article shows diachronic variation in what may be assumed threatening in the universe of the novel, which translators are expected to evaluate the appropriateness of.

The article ‘Portraying Intellectual Disability through Translating Fiction’ by Vasiliki Papaconstantinou is a diachronical analysis of whether and how fiction translation may register societal attitudes to disability. The study focuses on the character of Lennie, in two Greek target versions (1961, 2010) of Steinbeck’s novella ‘Of Mice and Men’ (1937). The 1961 version enhanced the weak traits of Lennie, constructing a vivid image of his disabled self, alluding to a medical model of disability. By contrast, the 2010 version portrayed a more powerful image of disabled Lennie, rather assuming a social model of disability.

In the article ‘Translating Threat in Greek Versions of Othello’, Stavroula Apostolopoulou examines how two Greek translators manipulate im/politeness and threat expressed in the discourse of characters in ‘Othello’, which give voice to current interpretations of the play, allowing racist, sexist and misogynic narratives and identities. Garcés-Conejos Blitvich and Georgakopoulou (2021) suggest that the study of identity should take centre stage, along with the notion of agency, in situated practices. The article highlights intra-cultural variation manifested through translation, along with heightened aggression and threat in present-day discourses. It confirms that translation practice is another arena where intracultural variation manifests itself through rendition of impoliteness and threat.

The article ‘Rendering Patriarchy through Gendered Translator Gaze in Romeo and Juliet’ by Dionysia Nikoloudaki examines patriarchy in four Greek versions of ‘Romeo and Juliet’ by assessing the foul language translators use in shaping the representation of female figures in the play. Questionnaire respondents suggest that offensiveness (and threat) is raised in the later versions, though not in the latest.  The questionnaire was produced by a female translator, evidently out of female solidarity with the female characters. The study suggests that gendered gaze may be a factor which affects the way patriarchal aggression and threat may be rendered.

Love poetry translation is another context where emotion and the interpersonal relationship between a poet and a beloved can be shaped. In ‘Explicitness as Threat in Love Poetry Translation’, Eleni Sichidi examines how disillusionment is rendered in a lyric love poem by Sergei Alexandrovich Yesenin (1925), when the poem is translated into Greek and into English. She examines primarily the Greek version, which she contrasts with three recent English versions of the poem. She shows that disillusionment is shaped differently in the Greek version, along with assumptions of interpersonal distance between the poet and the beloved and a different perception of what love is. Unlike the English versions, which are more loyal to the Russian source version, the Greek version portrays a closer relationship between the ex-lovers and more suffering emanating from the separation.

The article ‘Subtitling and Dubbing Intimacy and Threat: ‘Harry Potter’ in Greek” by Maria-Nikoleta Blana and Maria Sidiropoulou tackles audiovisual translation and examines how intimacy/friendliness and threat/aggression may be transferred in two audiovisual modalities, subtitling and dubbing. Digitally mediated communication (Xie and Yus 2021) has had its own rules and conventions which are worth examining through AVT modalities (Guillot 2020). Results show that dubbing is more active in heightening awareness of threat/aggression and intimacy, in contrast to subtitling which favours the strategy of ‘retention’. The paper prominently shows that mediality (Dayter, Locher and Messerli 2023) is a factor which significantly affects cross-cultural message rendition.

The special issue shows how people from different cultures interpret and understand each other’s im/polite speech acts and behaviours or/and what may be considered threatening in the speaker-addressee relationship or in the relationship of a speaker with the entity referred to. It examines fiction and non-fiction parallel data, in the English-Greek-Russian and the Russian-Greek and English paradigms. The non-fiction genres include parallel press data, EU legal discourse, academic writing, medical advertising. Fiction genres investigate cross-cultural variation of threat in versions of plays, novels, children’s literature and poetry.

As “[t]he search for regularities in corpora of translations has been mostly carried out with the aim of investigating universals” (Zanettin 2012: 23), this special issue uses data sets to monitor manipulation of relational dynamics between non-/fictional interlocutors and how translators avoid threat (if not intended) or enhance it to meet the expectations of a target community of practice. As a ‘corpus-based’ approach is a distinct paradigm in translation studies (Laviosa 1998), the special issue provides a mini-‘[c]orpus view of similarity and difference in translation’ (Baker 2004) with reference to manipulation of relational work (Locher and Watts 2005), and in un-/intended threat situations.

Acknowledgements

A lot of people have worked for completing this project. I am extremely grateful to my colleagues at the Language and Linguistics Division, of the Department of English Language and Literature, and to collaborators from the School of Philosophy, National and Kapodistrian University of Athens. They kindly accepted to anonymously peer-review the papers in this Special Issue, which immensely improved the texture and potential of the individual papers. Thus, a special thanks goes to the following Professors, Associate Professors and Assistant Professors for contributing their expertise and making crucial suggestions for improvement (alphabetically): Dr. Olga Alexandropoulou, Dr. Tatiana Borisova, Dr. Anna Hatzidaki, Dr. Elly Ifantidou, Dr. Evdokia Karava, Dr. Nikolaos Lavidas, Dr. Bessie Mitsikopoulou, Dr. Anna Piata, Dr. Nikolaos Sifakis, Dr. Aggeliki Tzanne. Last but not least, I am indebted to Ms. Alfia Khusainova for editing the Russian examples.

Their contribution has been invaluable and highly appreciated.

About the author(s)

Maria Sidiropoulou is Professor Emerita of the Department of English Language and Literature, National and Kapodistrian University of Athens (specialization Translation Studies and Pragmatics). She was Chair of the Department (Dec 1, 2017- Aug. 31, 2020), Head of the Interuniversity and Interdepartmental Co-ordinating Committee of the 'Translation-Translatology' MA Programme of the University of Athens (2009-2011), Director of the MA Programme 'English Language, Linguistics and Translation' (2020-2022), Chair of the Interdepartmental MA Programme 'Translation: Greek, English, Russian' (2020-2022). She was Director of the 'Translation Studies and Interpreting LABORATORY, META-FRASEIS' (2020-2023) and founding Director of the META-FRASEIS TRANSLATION PROGRAMME (2007-2020). Her publications (books, co-/edited volumes, articles) and talks deal with intercultural issues manifested through Greek-English translation in the press, in advertising, in academic discourse, in EU documentation, in tourism, in literature, on stage and screen.

Email: [please login or register to view author's email address]

Abstract

Medical discourse has attracted the attention of scholars internationally, both in monolingual research and in translation. A topic which has attracted particular attention is the relational dynamics between interlocutors, in this case between expert/doctor and patient. When medical discourse is to be transferred cross-culturally, the question arises whether the interpersonal dynamics between expert and patient shift or are invariable. The aim of the study is to research potential variation in the relationship between hospital/expert-patient through Greek-English translated medical advertising. The study examined twenty medical leaflets of the ‘Hygeia’ private hospital, in Athens, advertising services the hospital offers, for various medical conditions. The analysis utilizes im/politeness theory to examine manipulation of threat in the communicative situation and power distance between expert and potential patient. Findings show that the Greek version of the data heightens power distance and assumes higher threat awareness, which seems to be confirmed by a questionnaire addressing 15 bilingual respondents. The significance of the research lies in that it draws attention to cross-cultural variation in medical communication, suggesting that translation data is another platform where medical communication may be fruitfully researched

Keywords: medical discourse, impoliteness translation, hospital leaflets, advertising, expert-patient communication

About the author(s)

Daphne Charalampopoulou has a BA in ‘French Language and Literature’ and holds a MA in ‘English Language, Linguistics and Translation’ (specialization ‘Translation Studies and Interpreting’) from the Department of English Language and Literature, National and Kapodistrian University of Athens. She has attended seminars specialised in subtitling and has worked in the field of localisation. Her interests lie in medical translation and interpreting, as well as in psycholinguistics focused on disability issues.

Email: [please login or register to view author's email address]

Abstract

The study examines international, European and Greek sample data and focuses on threat in naming disabled people. Findings show a gradual shift in the representation of disabled people, in legal texts. The shift relates to the implementation of medical and social mo­dels of disability, which seem to regulate discursive representation of the disabled. The further back the analy­sis goes in time, the more manifestations of the medical model seem to appear, with the social model taking over as the analysis moves forward in time. Greece seems to have its own manner to adopt current trends in representing the disabled, as questionnaire results reveal. The AμεΑ (AmeA) abbreviation, widely used in the Greek context nowadays, seems to set anew the representation of the disabled, toning down the offensive value of previous terms. This is probably because Greek may have additional barriers in abandoning the medical model, because of culture-specific communication style features it prefers. The significa­nce of the study lies in that it advances understanding of cross-cultural perceptions of disability, which – in Greek – involves manipulation of the terms’ threatening potential.

Keywords: disability, legal discourse, medicalsocial models, european union, bilingual texts

1. Introduction: A fuzzy term

Disability studies is an expanding field and a highly interdisciplinary one cutting across literature, linguistics, philosophy, technology etc. (Solvang 2000, Barnes, Barton and Oliver 2002, Mabbett 2002, Siebers 2011). Disability is nowadays treated as a social phenomenon that affects societal and financial growth. It is seen as a barrier that needs to be surpassed and eliminated in the long run, for allowing an unimpeded operation of state apparatuses and realization of long-term goals for sustainable development. Disability is a rather fuzzy term, which institutions have attempted to clarify and discursively represent the disabled in various ways. For instance, in English, a part of disability studies has been theorizing about the terms ‘impairment’ and ‘disability’ and their potential difference and interplay. The Anglocentric point of view on disability, as mainly expressed in the Disabled People’s International (DPI) terminology, ‘impairment’ is seen as the functional limitation within an individual caused by physical, sensory or mental impairment, while ‘disability’ is seen as the loss or limitation of opportunities on the part of the disabled to take part in the normal life of the community, on an equal level with others, due to physical and social barriers (Goodley and Runswick-Cole 2011).

Another pair of terms in English is the one between the items ‘persons with disability’ vs. ‘people with disability’. In disability studies, ‘people with disabilities’ has been contested as implying that disability is something inherent to an individual, thus undermining the political nature of disability. Major representative organizations, throughout the world, prefer the term disabled persons, for (a) emphasi­zing the social inducement of disability, i.e., that people become disabled by society, rather than accidentally residing in it and (b) avoiding to highlight the individual character of the problem (Whalley Hammel 2006). People focuses on the individual rather than on the societal aspect and runs contrary to broadening the societal perception of disability.

2. Literature review

Αs mentioned, perceptions of disability vary and the interpretations assumed seem to be ideologically different (Oliver 2009, Oliver and Barnes 2010). A model of disability is important because it may be the basis on which states, institutional agents or regulatory authorities shape their policies on disability and their support measures; in this sense it is important for authorities and citizens to critically monitor the model of disability they conform to, in interaction and law-making. The medical model was followed by the social welfare model, which the following subsections elaborate on.

2.1 The medical view

The medical view sees “medical impairments as automatically resulting in disa­dvantage and exclusion, which can be ameliorated by cash benefits and other social welfare benefits” (Mabbett 2002:20); “people with different disabilities are the object of treatment and care […] while their disability remains their most conspicuous defining attribute” (Stopa 2012:146). Hence, people see themselves as disabled and marked in terms of the impairment identity. The prevalence of this model however has rather led to some marginalization of people with disabilities instead of promoting their integration in society (in labour, in education). As expert evidence shows, their inclusion is now more crucial than ever before, considering the economic circumstances prevailing in this era of recess and uncertainty. OECD seems very much aware of the financial deficit a marginalization of the disabled may entail and argues in favour of integrating them in the labour market:

integrating more fully into the labour market people with disability is essential in meeting economic and social challenges arising from these broad drivers of change. The recent economic downturn is further reinforcing this urgent need, as people with disability have been hard hit by job losses and the reduction in job vacancies. This may push them to the margin of the labour market, raising the risk of further structural increases in the disability beneficiary caseload (OECD 2010:21).

The medical model treats disability on the basis of its individual aspects and not as a multi-faceted social issue:

This characteristic confirms that we are facing a clear manifestation of the medical paradigm of treatment of disability, a circumstance that is incoherent, as other authors have already pointed out, if we take into account how other institutions of the European Union have been strong advocates of the implementation of the social paradigm of treatment of disability since 1996 (Gutiérrez Colominas 2017: 3).

2.2 The social/affirmation model

The social model sees disability as a social phenomenon rather than a medical one. The social model emerged in the 1980s, and was followed by the affirmation model. In enhancing the affirmative model, Cameron (2013) explains that “it also establishes their rights [of the disabled] to enjoy being who they are as people with impairments rather than regarding impairment as a cloud overshadowing their existence” (2013:28) and focuses on their activity as social actors, thus in the frame of a comprehensive relationship.

The affirmation model (Swain and French 2000) caters for actual inclusion in society: the beginning was marked with the social model which gave the disabled persons voice and occupied a space for them in the public arena: now this space is about to be filled and understood in terms of integration and interconnection, through an ‘affirmation’ of disability as an aspect of normalcy (Thomas 2004).

3. Methodology

As suggested, the aim of the study was to examine how the representation of disability plays out in the global/European/Greek context diachronically. The scope of the research can be huge, so the study selectively examines (a) early interna­tion­al dis­cou­rses on disability (1971-1981, with their official Greek versions), (b) European discourses on disability (1995-2014, with their official Greek versions) and (c) Greek discourses on disability between 1990-2018 (originally drafted in Greek).

The selection of data types may be justified on the grounds that if translated data show tendencies in naming disability in Greek, the monolingual data may show preferences in the Greek context more intuitively, without influence from another version. Table 1 summarizes selection of data in terms of the time span they derive from.

Table 1. Periods and types of data selection

The study, thus, examines both parallel data and Greek original (monolingual) legal discourse in order to trace tendencies in rendition of the terms ‘disabled’/’disability’ (see appendix for the list of data sources and word-count). The total word-count of the data set is 271,584 words.

After the authors’ analysis of the data (etic perspective), the study adopted an emic perspective by addressing 43 Greek respondents; the intention was to elicit evidence on the non-/threatening value of certain disability terms, asking respondents to seek appropriate renditions into Greek, by using their intuition in Greek. Results showed that there are terms which are not used in the EU context, which however seem to resonate with current native preferences.

4. Data analysis

The section presents parallel data samples which display variation in the transfer of disability issues, organized in three groups, i.e., from the international (for instance, UN discourses), European (EU discourses) and Greek context.

4.1 The international context

The General Assembly of the United Nations has prepared a declaration on the rights of persons with mental disorder in order to promote higher standards of living, full employment and conditions of economic and social progress and development. Sample texts derive from

• The Declaration on the Rights of Mentally Retarded Persons. Proclai­med by General Assembly, resolution 2856 (XXVI) of 20 December 1971.
• The Sundberg Declaration 1981.

Example 1 comes from the Sundberg Declaration, 1981. The World Conference on Actions and Strategies for Education, Prevention and Integration organized by the Spanish Government in co-operation with UNESCO (1981), set anew the goals of the ‘Universal Declaration of Human Rights’ and highlighted the need for steps to be taken for (a) rehabilitation and other support to reduce the handicapping effects of disability and (b) meeting the needs of disabled persons for full participation in social life:

Example 1

Τhe assumption is that the two texts (the EN and GR version), which are indicative of the language used by  various international instruments, issued at that point, suggest that the disabled people are somewhat separate from the remaining population. Τhis is conveyed by the item ‘normal individual’ which actually segregates the disabled persons from the remaining populace, by further implying that their condition is not normal. The English version may gloss over disability by the term ‘disabled persons’, but the Greek version seems more aware of a medical interpretation of disability, by explaining the range of types of disabilities referred to, in terms of options like ‘σοβαρό­τητα των μειονεκτη­μά­των και ανικανο­τήτων τους’ (the severity of their disadvantages and inabilities). Besides, the term 'mentally retarded' (literally rendered in Greek) is highly offensive and impolite towards the disabled, attacking their positive face (Brown and Levinson 1978/1987)[2]. Another shift between the two versions is the shift of the deictic adjective (EN ‘those’ → GR ‘these’): in English see ‘those rights’ assuming some distance and,  in Greek, see ‘these persons’, which implies proximity: interpersonal proximity is a positive politeness trait and Greek nowadays tends to show a tendency for positive politeness strategies, in interaction (Sifianou 1992), in contrast to English which shows a tendency for using negative politeness strategies, indicating that it favours interpersonal distance.

In theoretical terms, this point of view aligns with the morals of medical sociology, a theory that was dominant in the decades preceding the 1990’s (when the social model of disability emerged) and focused on the medicalization of disability that treats disability as a deviance from the norm, as a merely individual matter that impedes the persons due to their medical impairments. As Linton (1998) and Linton and Bérubé (1998) suggest addressing disability as a merely personal condition does not offer a comprehensive perception of disability:

Briefly, the medicalization of disability casts human variation as devi­an­ce from the norm, as pathological condition, as deficit and, signifi­can­tly, as an individual burden and personal tragedy. Society, in agreeing to assign medical meaning to disability, colludes to keep the issue within the purview of the medical establishment, to keep it a personal matter and “treat” the condition and the person with the condition rather than “treating the social processes and policies that constrict disabled peoples’ lives (Linton 1998:162).

This section offered a sample of discourse which implements a rather medical inter­pre­tation of disability, with the Greek version enforcing the medical inter­pre­tation even further, by favouring the specific (‘whatever the origin, nature and the severity of their disadvantages and inabilities’) and avoiding vagueness. Favouring the specific in Greek is manifested in quite a few contexts, in various ways, and has been grammaticalized through a frequent use of the definite article, even with proper names and generics (Sidiropoulou 2019). This is also manifested in example 2, where ‘All disabled persons’ is rendered in Greek as ‘Every disabled person’.

 4.2 The language of the European legal instruments

This section examines rendition of disability related terms at a more recent time span, 1995-2014 (see appendix for European data sources and word count), to trace ren­dition of disability-related terms in the official Greek versions of the selected terms.

Back in 2003, the European Union officially acknowledged that equality had not yet been achieved for the disabled and declared the year 2003 as the European Year of Persons with Disabilities suggesting that a turn should be made in the European frame towards removing all impediments that create difficulties for disabled people:

2003 is the European Year of People with Disabilities. The year will drive progress towards achieving equal rights for people with disabilities. Across Europe attention will be focused on the many areas of European society where barriers and discrimination still exist for the one in ten Europeans with a disability (The European Year of People with Disabilities 2003:8).

The ‘European Strategy on Disability 2010-2020’ brings to the fore notions like accessibility, participation, equality, employment, education and training, social pro­te­ction, and health. The Greek version does not seem content with its readily available option for the English term ‘disability’, in Greek ‘αναπηρία’ (disability); the drafter complements the term with the term ‘ειδικές ανάγκες’ (special needs) as if the term ‘disability’ is not adequate or appropriate. Example 2 shows an instance of this.

Example 2

The question that arises at this point concerns how prevalent these options may be in the data set, in terms of frequency. Table 2 shows number of occurrences per term (counted manually) in the parallel versions, Table 3 shows frequency percentages of certain terms in the English and Greek version of the data.