Dear reader,

With December on our hands, we take the time to look back at META-FORUM 2021, the tenth edition of our conference in the middle of November which saw record numbers in participants, many exciting sessions on European Language Technology and a successful virtual project expo. We collected a full recap and links to the session recordings and slides in the ELG section below – it’s worth to scroll on!

Other great news reached us from the North: Our ELE partners from the Language Bank of Finland were awarded for the accessibility and reusability of their language resources. To learn about why they received the Open Science Data Award 2021, have a look at the ELE section further down.

Finally, this fourth ELT newsletter includes an interview with Inaki Irazabalbeitia, a distinguished representative of the Basque language technology community. See for yourself and enjoy some positive news in this final month of the year!

With best regards

Georg Rehm

• “Tech companies blasted for inaction on Spanish-language disinformation” – NBC News, 4 November 2021
• "The Neuroscience of… Language” – Technology Networks, 8 November 2021
• “Albanian Language at Risk of Loss Due to ‘Txt Spk’” – Exit News, 9 November 2021
• “French AI strategy: Tech sector to receive over €2 bln in next 5 years” – Euractiv, 9 November 2021
• “New AI brings the power of natural language processing to African languages” – TechXplore, 9 November 2021
• “IBM enhances Watson Discovery’s natural language processing capabilities” – AI News, 10 November 2021
• “OpenAI rival Cohere launches language model API” – Venture Beat, 15 November 2021
• “Natural Language Processing Takes Off in State Government” – StateTech, 19 November 2021
• “Machine learning improves Arabic speech transcription capabilities” – Technology Review, 24 November 2021
• “Could artificial intelligence have written this story?” – The Washington Post, 26 November 2021
• “The Downside of English’s Dominance” – The Wall Street Journal, 27 November 2021
• “Weird word origins: Why slogans are Irish, muscles are mice and monkeys are monks” – The Irish Times, 29 November 2021

• 3-year-old Moraima tries talking to Siri in Galician and finds out why we need Digital Language Equality in Europe.
• Steve Nouri presents the capabilities of NVIDIAs new GauGAN 2 software.
• How large is the percentage of users on the internet that are native English speakers?
• Expert.AI explains why you shouldn’t mistake NLU for NLP.

As it is often the case in computer science, numbers paint the clearest picture: More than 400 participants joined META-FORUM 2021, the second virtual, third ELG-focussed and tenth overall conference on powerful and innovative Language Technology for the multilingual European information society. At peak times, around 300 people watched one of the seven sessions, which saw presentations on the European Language Grid, language-centric Artificial Intelligence, reports from industry as well as EU-funded projects in the field of Language Technology. 

At the virtual expo, which opened on day one, 36 projects presented their research plans and first results at their digital booths in Gathertown. The final session referred directly to this year’s conference title, “Using the European Language Grid”: In a workshop, the technical ELG team, tuning in from Germany, Greece and the UK, explained different ways of using and contributing to the ELG platform.

Our upcoming ELG blog article will present the Berlin-based company Coreon, a participant in our ELG pilot project call. If you’re already curious to learn more, have a look at Jochen Hummel’s contribution to the European language technology industry track at META-FORUM 2021.

• Croatian tweet sentiment analysis – A sentiment classifier for the Croatian language, trained on labeled Croatian tweets, using the BERTić language model. The tool classifies sentences (or short paragraphs) into three predefined classes (positive, negative or neutral), based on the sentiment/stance of the input text. The running tool is available in ELG, therefore everybody can directly try it out with an example sentence. The tool was added by the Faculty of Computer and Information Science of the University of Ljubljana on 09 July 2021 and was updated on 5 November 2021.
• Multi-Attributed Structured Text-to-face Dataset – A new data consolidation. The motivation is to have a large corpus of high-quality face images with fine-grained and attribute-focussed annotations. This has the benefits of the attribute oriented approach as well as the semantics in a textual description. The resources were automatically harvested from Zenodo.
• A Novel Approach to a Semantically-Aware Representation of Items (NASARI) – The dataset contains semantic vector representations for BabelNet synsets and Wikipedia pages in several languages: English, Spanish, French, German and Italian. Currently, there are the three vector types lexical, unified and embedded available. The conceptual resource was added by the Department of Computer Science of the Sapienza University of Rome on 12 November 2021.

"Digital Language Equality is the state of affairs in which all languages have the technological support and situational context necessary for them to continue to exist and prosper as living languages in the digital age." During the session on the European Language Equality project on Day 3 of META-FORUM 2021, Federico Gaspari from the ADAPT Centre at the Dublin City University (DCU) shared the preliminary working definition of Digital Language Equality – one of the key ingredients of the Scientific Research and Innovation Agenda and Roadmap. In the session, Federico and other project partners from the UK, Greece, the Basque Country and Germany gave insight into the current state of the European Language Equality project and its many connections to the European Language Grid. The recording is worth looking into for anybody interested in multilingualism in the digital age.

What would Language Technology be without open data? The importance of accessibility and reusability of scientific data is recognized by the University of Helsinki, which awarded its annual Open Science Award 2021 to the Language Bank of Finland, partner of the ELE project. Since 1995, the Language Bank provides text and speech corpora, tools, training, analysis and data management and is involved in campaigns such as Donate Speech, in which 10,000 hours of Finnish speech are collected – an effort worth awarding.

Inaki Irazabalbeitia played an important role in bringing about the European Language Equality project. He was a member of the European Parliament 2013-2014 and has been active in the world of language technology and politics. We interviewed Mr. Irazabalbeitia about the Basque language, its ongoing journey to equality and the importance of available representation. Find out why he considers the current situation of the Basque language “simultaneously an opportunity and a threat” – in our first interview on the ELE blog!

Basque Centre for Language Technology – HiTZ 

HiTZ is a multidisciplinary research center at the University of the Basque Country that specializes in Language-centric Artificial Intelligence. The objective of the center is to investigate language and speech technologies as well as to transfer knowledge and technology to companies. It comprises two research groups, Aholab and Ixa, that bring together members from seven departments at the university. Both have extensive experience since 1993 in conducting research, creating linguistic resources and tools, and launching several commercial products. HiTZ is also a founding member of CLARIN-K Center, a member of CLAIRE and a full member of BDVA and DAIRO. Through CLAIRE and BDVA, HiTZ also participates in the European Public Private Partnership on Artificial Intelligence, Data and Robotics.

At the moment, HiTZ has more than 60 members, including computer scientists, linguists and research technicians. In the last five years, HiTZ researchers have published more than 200 scientific publications. The group is a leader in applying deep learning techniques to language processing and its recent work in the area has been cited more than 4,000 times in the last two years. During their history, the groups have participated in more than 200 research projects ranging from regional to European projects. It has also participated in more than 100 industrial contracts with the aim of transferring technology into the industry. HiTZ members have been advisors in the creation of the National Plan for Spanish Language Technologies.

HiTZ is also a member of the Erasmus Mundus+ European Masters Programme in Language and Communication Technologies (LCT). It is designed to meet the demands of industry and research in the rapidly growing field of language technology. HiTZ also offers a Doctoral Programme in Language Analysis and Processing.

The next ELT newsletter will be sent out on 4 January 2022. Until then, follow our ELT social media accounts (as linked below) for the latest news!