Academia • By Maria Stasimioti On

Here’s What Happened at the ‘New Trends in Translation and Technology’ Conference

NeTTT Rhodes Machine Translation Conference 2022

After the two-year wait caused by a global pandemic, the first New Trends in Translation and Technology (NeTTT) international conference took place on the island of Rhodes, Greece, from July 4–6, 2022.

NeTTT was preceded by a two-day summer school on neural machine translation (NMT), a workshop on translation technology for creative domains, and concluded on July 7 with a series of tutorials.

The main conference, which welcomed more than 170 delegates, featured 5 keynote speeches, 65 oral presentations, and 14 poster presentations.

NeTTT brought together academia, industry, and EU institutions and served as a platform for discussing all emerging developments and practices in the field. The main takeaway: “translation, technology, and terminology go hand in hand for the future of all linguistic professions,” as Rodolfo Maslias, Head of Terminology Coordination at the European Parliament, tweeted.

The conference also attracted several sponsors, Slator among them. Supporting NeTTT as a silver sponsor, Slator was represented by Conference Producer, Deppy Nikolaidou, and Maria Stasimioti, Research Analyst.

NMT Summer School

During the first day of the summer school, the participants had the chance to attend a lecture on NMT by Mikel Forcada, Full Professor at the Universitat d’Alacant, and “play” with pre-trained models.

That same day, Sheila Castilho, IRC Research Fellow at ADAPT Center, talked about the essentials of machine translation (MT) evaluation, highlighted the complexity of the task, and familiarized participants with PET, a tool for post-editing and assessing machine translation.

The second day of the summer school started with a lecture by Marcello Federico, Principal Applied Scientist at AWS AI Labs, who explained the foundations of neural networks.

During the afternoon session, Tharindu Ranasinghe, Lecturer at the University of Wolverhampton, introduced sequence-to-sequence models and explained how to build an NMT system. “When someone says ‘I’m training an MT system,’ I want you to have the confidence to say ‘Ah, now I can do it too,’” Ranasinghe said.

The last speaker, Panos Kanavos, founder of Lexorama, talked about NMT specialization and domain control. During the practical session, Kanavos introduced NeuralDesktop, a customizable, domain-aware, and user-configurable NMT system.

The participants were also offered a workshop on translation technology for creative domains. The focus of the presentations was on post-editese in literary translations, personalized MT systems for literary translation, and Computer-Assisted Literary Translation (CALT).

New Paths to Translation Augmentation

The main conference kicked off with a keynote speech by Sharon O’Brien, Professor at Dublin City University, on augmented translation. 

According to O’Brien, augmentation should lead not only to better and speedier solutions, but also the possibility of finding solutions to problems that before seemed insoluble. “Translation has already been augmented,” O’Brien said, “but there is certainly potential for considering new real pathways to augmentation.”

Three parallel sessions on different topics followed — such as machine translation, machine translation evaluation, post-editing, translation platforms, translation workflows and automations, speech translation, audiovisual translation, natural language processing (NLP) support for translation, ethics in MT, gender and incluse MT, translator training, and translation reception.

More keynote speakers followed, such as Valter Mavrič, Director-General of the Translation Service (DG TRAD) at the European Parliament (EP), and And Merit-Ene Ilja, Translation Director at the Directorate-General forTranslation (DGT) of the European Commission.

Mavrič illustrated how the translator role at the EP has evolved over the last years into that of a versatile language professional. Ilja talked about the DGT’s translation ecosystem, notably how “DGT is open to embrace what technologies can offer.”

Yves Champollion, founder and Chief Architect of Wordfast, discussed the new features of the translation-memory tool and how “the next disruption in translation technology will be voice dictation combined with MT-enabled predictive text.”

Summer school lecturers Federico and Forcada also spoked on using context information in MT and MT usage rights of language data, respectively.

MultiTraiNMT Workshop

After the end of the main conference, the workshop “Machine Translation for Everyone” organized by the MultiTrainNMT team took place. MultiTrainNMT is an EU-funded project that aims to teach NMT technologies to non-technical students and more general audiences. The goal is to effectively empower future users of AI-powered MT with a better understanding of how it works and should be used.

The program of this event featured six 10-minute talks where attendees presented their experiences in teaching MT to translation or language students. In addition, the MultiTrainNMT team informed the attendees that the book “Machine Translation for Everyone: Empowering Users in the Age of Artificial Intelligence” was out. As Miguel A. Jimenez-Crespo, Associate Professor at Rutgers University, tweeted, the book is “required reading for anyone interested in the topic.”

Tutorials 

July 7 was a full day of tutorials, starting with Jorge Dias Cintas, Professor at University College London, teaching participants subtitling on the cloud using the OOONA platform. 

Then, Ricardo Munoz, Full Professor at the University of Bologna, introduced Inputlog, a logging tool for  researching writing processes. The final presentation was given by Anna Zaretskaya, Director of MT Strategy and Implementation at TransPerfect, who talked about the post-editing for media and subtitles. Zaretskaya also introduced TransPerfect’s AI-driven subtitling workflows through the MediaNEXT platform.