A few years ago, Nelson Mandela nailed the importance of multilingualism in two sentences: "If you talk to a man in a language he understands, that goes to his head. If you talk to him in his own language, that goes to his heart."



Director-General for Translation Rytis Martikonis

Authors: Roberto Viola (Director General DG Connect) / Rytis Martikonis (Director General DG Translation)

However, what exactly is multilingualism in practice and how much does it influence our lives? What can digital dreamers and European policymakers do to avoid getting  'lost in translation'? Why do we go through such lengths in the EU to operate in all our 24 official languages?

In his essay on languages, the linguist Wilhelm von Humboldt states that 'the character and structure of a language expresses the inner life and knowledge of its speakers, and that languages must differ from one another in the same way and to the same degree as those who use them.' (Muriel Mirak Weissbach (1999). "Wilhelm von Humboldt's Study of the Kawi Language: The Proof of the Existence of the Malayan-Polynesian Language Culture". Fidelio Magazine. VIII (1). Archived from the original on 12 July 2014.)

This means that language is not only a means of communication but the expression of the inner world of its speakers. We are all intrinsically different, and our languages express our differences of thought and meaning. The result is an incredible richness of forms of expression, proof of our infinite human variety. This is true for European nations and its languages; we are proud of this richness.

Because it respects and values these differences, the European Union has not chosen to operate with a language policy that selects a few of its languages, but it supports all 24 offical languages. This principle is enshrined in the EU Charter (Article 22) and in the Treaty on European Union (art. 3(3) TEU).

Translation as a public service

EU citizens have a right to address the EU institutions in any of the EU’s official languages and to receive a reply in that language (Treaty on European Union (art. 3(3) TEU). Indeed, translation was the subject of the very first regulation that the Commission adopted, Regulation No 1 of 1958, which sets rules governing the languages of the EU’s institutions and require that all binding legal acts are drafted in all official EU languages.

To enable citizens to exercise this right, the EU institutions have translation services that play an indispensable role in multilingual law making to deliver all 24 equally authentic language versions of legislation. All language versions must convey the same meaning so that texts can be interpreted and applied in a uniform way and produce the same legal effect in all languages. This puts the principle of multilingualism into practice and brings transparency to the law-making process.

Languages online

Europeans are proud of their languages and, in addition to having access to legislation in their own language, they should be able to continue using them in the online world too. We view the linguistic diversity of the EU as an asset and a great opportunity for the digital single market. Digital technology can build bridges to connect people speaking different languages. So how can we make this possible, what tools can we use?  

In addition to human translation for legislation, the EU institutions also provide machine translation to enable users to get an immediate grasp of other types of texts written in languages they don't understand. The European Commission's machine translation system, MT@EC, is a secure service suitable for handling confidential material. It's tailored to translate EC documents, i.e. the language it uses, the terminology and phrases, were extracted from a wealth of high-quality human translations produced by the EC over the past decades. The service covers all official EU languages - a total of 552 language pairs! It powers several European digital service platforms, such as the Online Dispute Resolution platform, European Open Data Portal and the European e-Justice portal.

In November, we plan to roll out the next generation machine translation service, eTranslation. It will be based on MT@EC, which is already open to EU institutions and national public administrations. eTranslation will become the automated translation platform of the Connecting Europe Facility.  The aim is to improve the speed and quality of the service and support a greater variety of subject domains. Over the next few years, the number of digital service infrastructures using eTranslation will be extended to cover social security, eProcurement and eHealth.

We encourage Member States to get involved in improving the platform.  We have started talking with national authorities at meetings on European Language Resource Coordination on how we can pool our language resources for the common good of building bridges across language communities.

Language technologies

Over the last seven years, the European Commission has invested in research and innovation in language technologies such as machine translation, speech recognition and data analytics. Several very promising initiatives are growing out of this investment.

Matecat is a Computer-Aided Translation tool developed as part of an EU-funded project.  It integrates the largest collaborative translation memory and the best machine translation. It's becoming increasingly popular among language service providers, such as Welocalize and big commercial players such as eBay.

Under the EU's 7th Framework programme for research,  we funded a research project called Parlance. The project team developed an interactive web search service designed to be used on mobile phones. It uses exclusively speech, both as input and output, meaning that you speak to the device and get a spoken reply in your chosen language.

But many of the technologies currently available do not support all European languages. One of the reasons is that they are morphologically too complex for the current solutions. One project, QT21, aims to address this by developing 'improved statistical and machine-learning based translation models for challenging languages and resource scenarios.' They design deep learning solutions that identify patterns from data to predict or classify future requests. It's starting to have a major impact in language technologies: these 'neural' machine translation services seem now able to produce better quality translations than statistical-based systems.  

The future of multilingualism is bright. Big data, Cloud and Supercomputing have the potential to raise the quality of automated translation, but there is still plenty of work to do. Machines still cannot understand context or bring the real life knowledge that enables human translators to choose the right words in the right contexts.  

The pace of change in technology is rapid. We think that, thanks to language technologies, in future we will be able to shop online wherever in Europe we choose and language will no longer be a barrier. Small businesses will be able to market their products and services to a much bigger market and scale up. Researchers will be able to use scientific information written in different languages more quickly. Member States will be able to work better together on common issues like cross-border crime or terrorism. The list is long and the potential benefits are enormous.

By using digital solutions we can bridge language barriers if we consider our diversity as an opportunity rather than an obstacle. This is how our ambitious goal to create a Digital Single Market fits in the bigger picture. It helps to make our motto reality: Europe, united in diversity.