LT3 Newsletter 2026

These projects helped strengthen the LT3 team with five brand new PhD students (Eleni, Sadegh, Marie, Rose and Aurora) and ensured that Florian and Joni can continue their predoc careers. We also happily obtained two HPC Tier 1 grants in 2025 to cover most of our GPU needs.

In March, Jasper took up a new role as a postdoctoral research assistant at the department with a special focus on generative AI for language learning and remains affiliated with LT3. After their respective PhD defenses (see more later), Aaron and Colin were able to join new projects as senior researchers, and Pranay managed to obtain a BOF postdoc grant. 

In academia, goodbyes are also part of the landscape as colleagues move on to new institutions, new projects or entirely new careers. The same was true for LT3 in 2025, we celebrated Margot’s and Sofie’s PhD Defense, and organised a goodbye party for Serafina and Liesbeth. Liesbeth will be working on her postdoctoral FWO project Advancing Language Models with Human Social Reasoning Skills for Socially-Aware Decision-Making at UGent’s AIDA group. Sofie has joined Harvard Business School on a BAEF-scholarship to work with Amit Goldenberg on the topic of emotion regulation. As of October 2025 Ellen became full-time CEO of AlfaSent and what better way to celebrate her time with us than to sing karaoke together one last time. Michael, who was our developer for several years, also decided that it was time for a new challenge. Luckily, we were able to quickly find someone new for the job, welcome Simon!

As you probably could already derive from the previous section, a record number of five junior researchers defended their PhD in 2025!

Margot defended her work on translation tool-assisted second language writing in Swedish. Pranay’s dissertation covered multilingual language models for LOTE (languages other than English). Colin worked on modelling semantic similarity in Byzantine Greek poetry. Aaron’s dissertation discussed computational irony modelling, and Sofie defended her work on data-driven modelling of emotional processes in social interaction. 

• Fonteyne, M. (2025). One Tool Fits All? The Role of Tool Type and Language Proficiency in Translation Tool-Assisted Second Language Writing Processes. Ghent University. Faculty of Arts and Philosophy (Ghent, Belgium).
• Singh, P. (2025) Bridging the Gap: Adapting English & Multilingual Language Models for Languages Other than English. Ghent University. Faculty of Arts and Philosophy (Ghent, Belgium). 
• Swaelens, C. (2025). Verse by Verse: Modelling Semantic Similarity in Byzantine Greek Poetry. Ghent University. Faculty of Arts and Philosophy (Ghent, Belgium). 
• Maladry, A. (2025). When Meaning is Implicit: In Pursuit of World Knowledge in Computational Irony Modelling. Ghent University. Faculty of Arts and Philosophy (Ghent, Belgium).
• Labat, S. (2025). Affective Threads: Data-Driven Modeling of Emotional Processes in Social Interaction. Ghent University. Faculty of Arts and Philosophy (Ghent, Belgium).

Public PhD defenses can be a prime opportunity to gain a deeper understanding of the field, which translates into workshop time! This year, we organised three such end-of-project workshops:

• Colin's PhD defense was preceded by a workshop on Ancient Language Processing, with presentations by Thea Sommershield (Durham University) and Barbara McGillivray (King’s College London).
• Aaron organised a pre-defense workshop where Roman Klinger (University of Bamberg) shared insights on modelling emotion analysis and appraisals, followed by Senja Pollak’s (Jožef Stefan Institute) presentation on computational analysis of migration discourse.
• Sofie’s PhD defense was also preceded by a workshop, with Bennett Kleinberg (Tilburg University) giving a talk on empathetic AI and An-Sofie Claeys (Ghent University) presenting on humour in crisis communication.  

Congratulations again to all five new doctors!

Over the past year, the LT3 research team has had the pleasure of hosting four talented visiting PhD students. Each student spent between three and twelve months fully integrated into the team's collaborative environment. 

• Delu Kong (Tongji University) working on post-editese and machine translationese 
• Ran Zhang (University of Mannheim) exploring LLMs for literary translation  
• Shijie Liu (Maritime University of Shanghai) investigating LLMs for domain-specific terminology extraction 
• Xiaoye Li (Hunan University) studying the impact of text type and source (neural machine translation & human translation) on translation effort and quality; LLM-feedback for translation revision

As is the case every year, conference season was fairly busy, with many team members involved in the organisation of conferences, workshops and shared tasks. 

• Our TransTech team was well-represented at MT Summit 2025 in Geneva: Joke was track chair for the Translators and Users track, Joke and Janiça co-organised GITT 2025, and Lieve co-organised CTT. 
• Sofie & Véronique co-organised NLPSI, the First Workshop on Integrating NLP and Psychology to Study Social Interactions.
• In August many LT3 colleagues were in Vienna for ACL, where Els  was handbook chair and session chair, Pranay and Colin both had a paper accepted in findings, and Alessandra, Jasper, Quanqi, Natalia, Florian and Janiça all presented at workshops.
• Closer to home, Francesca co-organised the ninth edition of the Linghentian Doctorials.

Several keynotes were delivered by:

• Véronique on “The different faces of emotion: Navigating complex emotions in natural language and across modalities” at AItLA-2025. 
• Els on “From Manuscripts to Neural Networks: Unlocking the Past with Natural Language Processing” at the CHR 2025 conference.
• Véronique on “Flemish Language Models” at the KVAB conference.

LT3 colleagues also presented their most recent work at many different conferences and workshops throughout the year: CLIN35, NITS, NLP4CALL, ICTIC, LTRC, DHBenelux, ICLC11, EST11, PSLT2025, MT Summit 2025, SLA Research Seminar, ICWSM, ACL and EMNLP. Please refer to LT3’s 2025 publication list if you want to learn more about our latest research.

In addition to conferences, we also had the honour of welcoming various (international) guests at our LT3 office in Ghent to discuss research and possible collaborations:

• Mae Sosto (CWI Amsterdam) on how LLMs reflect societal norms on gender and sexuality in sentence completion tasks.
• Martina Rosola (University of Barcelona) on Gender-fair language rewriting in Italian.
• Gaurav Kamath (McGill University and Mila) on Measuring Word Meaning Change Across Time and Speaker Age.
• Ruben Janssens (UGent-imec) on Evaluating social robot dialogue with visual input.

We would also like to put the spotlight on:

• Arda who gave his inaugural speech at our faculty.
• Natalia, Loic, Véronique and Els, congratulations on winning the best paper award at the 12th Workshop on Argument Mining for Stance-aware Definition Generation for Argumentative Texts.
• Janiça and Eleni, with their Honorable mention at the GeBNLP workshop (ACL).
• Our book “Taaltechnologie Ontrafeld” of which more than a 1,000 copies have already been sold!

In 2025 we also kicked off the second year of the new Language Technology track in the Applied Linguistics Bachelor! Instead of choosing two foreign languages, students can choose one foreign language and focus on Language Technology. Please help to continue spreading the word about this unique educational track where students can combine their passion for languages with STEM!

This year, 10  students are following the CALM postgraduate programme. One of the main strengths of this programme is our collaboration with trusted partners, such as Yamagata Europe, where our students were invited again for a company visit in the first semester.

We also hosted another edition of the Methods in Language Sciences (MiLS) summer school and welcomed 19 international students to the course on Natural Language Processing with Python (taught by Cynthia, Colin and Joni). 

With agentic AI being one of the main trends and buzzwords in 2025 and even selected as marketing word of the year by the ANA (though maybe not unsurprisingly, there was a tie with the word “authenticity”), we have been happy to further contribute to a better understanding of AI to the broader public in 2025. And please, rest assured, authenticity is what we strive for in all our activities.

For our outreach activities we gave keynotes, presentations, prepared webinars, videos and workshops and participated in debates. All this, for a variety of audiences, including high-school students, (language) teachers, researchers and professionals in both the language industry and beyond.

Vlaams Talenplatform, Stad Gent, BQTA, Culturele Kring Gent, KTV Kennisnet, VDAB, VRT, European Parliament, European Court of Justice, Wetenschap aan Zee, World Expo Osaka, Jonge Academie, Faculty Engineering & Architecture, Faculty LW, EduMa Talen, Humanities Academy

Schools where we gave workshops at or of which pupils came to us:

• 'AI nascholing voor onderwijsprofessionals" (GO! Busleyden Atheneum Mechelen)
• Ondertiteling: Hoe begin je eraan? Wat betekent automatisering in het domein? (College Melle, GO! Atheneum Voskenslaan, Leiepoort campus Sint-Hendrik Deinze, VLOT Campus Sint-Lodewijk, SiVi - Sint-Rembert Torhout, Sint-Franciscus-Xaveriusinstituut Brugge, Emmaüs secundair Aalter, Bertinus Collectief Poperinge)

Research:

• MultiGEC - dataset for Multilingual Grammatical Error Correction in 12 European languages (Czech, English, Estonian, German, Greek, Icelandic, Italian, Latvian, Russian, Slovene, Swedish and Ukrainian) compiled by the CompSLA working group and over 20 external data providers in the context of MultiGEC-2025, the first text-level GEC shared task. 

For the broader public:

• Orphée contributed to the online learning module that was created for all university students on 'Generative AI: Concepts, Creations, Research and Classroom Practice'
• The podcast “Dwars Door de Klas” featured an episode on language and technology with Cynthia and Orphée. 
• The MOOC AI for Social Professionals which was created by the Vives University of Applied Sciences went live in May 2025 and by the end of 2025 over 4,000 users had already enrolled. Contributions were made by Véronique, Lieve, Orphée & Cynthia.

As we continue to show through these newsletters: we are all about a good balance at LT3. This means that we are still coming up with creative PhD and/or goodbye gifts:

As well as creative teambuildings at the office:

And we can also enjoy a conference from time to time:

As promised, we could not just let this fifth edition pass without making it a bit special, and so we decided to go back in time with a small crossword to test your knowledge from the last five years of LT3. If you’re stuck, it might help to go check in the previous newsletters, all information can be found there.

1. During these last years, some of our colleagues launched a spinoff, what is it called?
2. What is the name of Marzieh’s project that she started in 2024 under the supervision of Els and Véronique?
3. Who was the program chair for LREC-COLING in 2024 (first name)?
4. In 2024, Arda was involved in organising a first edition of a workshop during the EAMT conference, what was it called?
5. In 2022, only one LT3 researcher defended their PhD, what is their surname? 
6. How many LT3 researchers defended their PhD in 2021?
7. Who worked on a project regarding Translation Tool-Assisted Second Language Writing and defended their PhD in May of 2025 (first name)?
8. The book that was published by Cynthia, Véronique and many other colleagues from LT3 in 2024 is titled: ‘Taaltechnologie …’?
9. For which workshop did Els, Pranay and Aaron visit ACL in Bangkok in 2024? It was co-organised by Orphée.