LinguaTech: The Ghost in the Translation Machine

In line with the European Commission’s priority to create ‘A Europe fit for the Digital Age’ and the Digital Education Action Plan (2021–2027), the project aims to integrate practical applications of (neural) machine translation (NMT) and Generative AI (GenAI) into language competence assessment and translator training.

Initially, LinguaTech will develop a thorough understanding of the additional language (i.e., foreign/second language, AL) competence of translators as language experts. In view of the of recent advancements in language technologies, this development should be tailored to the requirements of NMT post-editing and supported by recent innovations in GenAI. Professional translation is a much more complex cognitive task than predicting the next sequence of characters in a string based on prior (and often limited in scope and language coverage, as well as occasionally biased and incorrect) text-based training data. Thus, expecting GenAI to replace human translation in all use-cases carries several ethical and practical concerns. On the other hand, however, constant development of NMT and GenAI brings valuable opportunities that should be exploited provided the risks are carefully mitigated. Essentially, the project will provide insight, tools and guidelines that will allow translators’ AL teachers to blend the latest language technology developments into their training, and thus make their students better able to meet the linguistic demands of NMT post-editing (PEMT).

The project combines corpus and linguistic analysis with qualitative interviews and machine translation quality evalua­tion to:

• examine the AL competence of translators as language experts and develop and validate AL competence descriptors aligned with a common language assessment framework (CEFR);
• compare and validate these findings in cooperation with RWS, a secondment partner and major translation solutions provider, by assessing the linguistic demands of post-editing of machine translation (PEMT), and to translate these insights into practical guidelines for AL teaching;
• design an open-access online platform for AL teachers on translation programmes with pedagogical resources that integrate the potential of Generative AI and neural machine translation (NMT), while also addressing their limitations.

Project team: Mag. Melita Koletnik, PhD; Univ.-Prof. Dragoș Ciobanu, PhD

Project duration: 1/06/25 → 31/05/27