For the last 18 months, LIBER has worked with partners from around Europe to advance digital language equality. In this blog post, we look back at the European Language Equality (ELE) project, reflect on what has been achieved, and outline the impact on research libraries.
The ELE project and LIBER’s role
Europe is home to 24 official languages and over 60 regional and minority languages. All contribute to the continent’s rich and diverse linguistic make-up, each uniquely adding to our shared heritage and culture. This multilingualism is a cornerstone of Europe and is a key signifier of what it means to be and feel European.
However, language barriers still hamper communication in physical and virtual spaces. As our lives are becoming more reliant on digital spaces, some languages are digitally left behind, with a lack of online resources, and poor translation support. Such problems are underpinned by the fact that only 22% of the world’s population speaks English, but English is the language of over half the internet’s content.
Aims and outcomes of the project
The European Language Equality (ELE) project, funded by the European Commission, is set to address some of the challenges working on language and how it is supported by technology, with a goal to level the playing field for European languages. The project consortium was comprised of 52 partner institutions, representing national languages, users and consumers of language technology, and industry experts.
ELE began 18 months ago with the clear and ambitious aim of forming strategic research, innovation, and implementation agenda as well as a roadmap for achieving full digital language equality in Europe by 2030. The project came about in response to the 2018 European parliament resolution, ‘Language equality in the digital age’.
As the project ends, an impressive 66 deliverables have been published in the last year and a half. Within these reports a range of important content can be found, investigating the state of specific languages, technological capacity for languages at present, and reviews of current policy. From this robust base of information, the consortium can drive efforts toward evidence-based recommendations on how digital language equality within Europe can be realised.
LIBER proudly represented research libraries within this project. A core part of this was the creation of a report which surveyed and interviewed our Participants. In addition, we hosted a workshop in which experts from the ELE consortium were invited to discuss language technologies and their implications for research libraries (watch a recording of this session). Our participation in this project ensured that the voice of European research libraries is heard as language technologies develop and great equality between all European languages becomes a reality.
Final conference and project resources
On Tuesday the 9th of June, ELE hosted a concluding conference in Brussels, bringing the project to an end by presenting work across the different work packages and announcing the initial findings of the strategic agenda (publication forthcoming).
The conference acted as a clear reminder of the scale of the project, with its ambitious aim to survey the landscape of language technologies and each European language, collating this information to draw up proposals on how to move forward – not leaving languages behind. Speakers took to the stage sharing insights from their respective European countries, with Dutch, Irish, Icelandic, and Bulgarian being spoken, true to the project’s multi-lingual values.
This event served as a reminder of the valuable resources the project has produced which can be utilised going forward:
- Reports: An overview of the deliverables was presented and summarised in Brussels. The contents of these reports could provide a good starting point for anyone interested in language equality and language technology. Between 500-750 people were surveyed and interviewed in the production of this material. In these documents, it is possible to dive into assessments of 35 European languages, an assessment of language technologies, and a review of existing policy and strategy in this field. Finally, ELE will publish its accompanying book.
- Language Equality Dashboard: Together with ELE’s sister project, the European Language Grid (ELG), an interactive dashboard has been developed. This tool allows users to compare the levels of technological support for different European languages based on a Digital language equality (DLE) metric. Calculations for this index are done based on the project’s research and are continuously updated, producing a DLE score to show a language’s readiness for technologically embedded multilingualism through technological and contextual factors.
Language technology and research libraries — the way forward
AI and machine learning in the field of language will increasingly affect the work of libraries and the world of information in which they operate. While this project has achieved much, the threat of digital extinction for many European languages still looms. And language technologies will continue to become more powerful – for some languages, at least. It is unlikely these trends will be reversed any time soon and both will cause unknown challenges and opportunities.
Though the ELE project has ended, there are no clear answers to what the future development of language technologies will hold, the possibility of European language equality, and what repercussions there could be for libraries and the wider research community. After this project, some key questions for research libraries arise. We would like to highlight three areas – research, technological development, and trust and protecting cultural heritage – to reflect on language technologies and research libraries.
- Do libraries have a role to play in developing and gaining expertise in language technologies?
Vast resources have been poured into digitising collections. Now that information exists in this format it could be used to train language models and help develop language tools. The legal and ethical repercussions of doing so would have to be considered, with careful attention paid to copyright. But historically librarians have been trusted to provide researchers with knowledge for their work, understanding which sources are trusted, and what represents the state-of-the-art. Why could this skill not be transferred to training algorithms? Could libraries provide software engineers and technicians with good data sources to use to develop language models? And how would this shape the resulting language tools? For more detail on this point, read this blog post from 2017 which encompasses such ideas.
- Translation tools would improve the accessibility and openness of research
When work and findings are available in multiple languages, it is automatically more accessible to a wider audience. How could researchers be incentivised to publish research in multiple languages, improving the reach of their work, increasing its visibility to more readers? How could this be incorporated into publishing processes and who should take responsibility for this? These steps could also go some way to preserving language, as the need to publish in English could be less essential.
Furthermore, how could more knowledge about translation technologies – what is available and trusted, and how they are best used – help researchers in their work? Should libraries be acquiring expertise in these areas to help educate research staff to ensure their work is executed to the highest standard, properly utilising all available tools?
- At present, the main drivers of language technologies and translation tools are a handful of technology companies based in the United States and Asia
At the ELE conference, the audience heard from an Icelandic representative who explained how a national delegation, headed by the country’s president, had travelled to California to discuss increased digital support for the Iceland language with tech companies. This situation leads to questions about how a drive for profit serves the interests of a global digital community. What happens if they withdraw certain technologies, or eventually switch to paid models for products which are currently free? How is data being used which is used by leading translation services? Within all this, transparency and trust are key issues.
Public trust is particularly high in libraries as institutions, which could be harnessed for them to play a role in forming other paths to language technology. As institutions invested in heritage and cultural preservation, would this not lead to a more equitable playing field for the languages?
As ELE’s co-coordinator, Prof. Dr Georg Rehm (German Research Centre for Artificial Intelligence, DFKI; Berlin, Germany) explains: “We have been working on the topic of technology-enabled multilingualism for many years, stressing the importance of developing and also providing high-quality and robust technologies for all European languages. As Europeans, we must not ignore a more and more crucial, all-encompassing piece of our digital infrastructure, delegating this to hyperscaler companies from other continents. We need language technologies developed in Europe for Europe.”
An unavoidable fact is that the effects of these technologies will be felt by libraries and librarians whether they play an active role in the development of the tools or not. Projects like ELE allow us to participate in these conversations and shape the outcomes, so they work for the benefit of the sector and the research community. The future is unclear, but it is there to be shaped.
The future of ELE
The project will continue with a slimmed-down consortium for an additional year. The five core partners plus two organisations representing the European language communities will continue to take steps towards European language equality, seeking support and advice from the initial project’s larger consortium.
Read more about ELE here.