DeMeVa workshop 3

The third DeMeVa workshop, organized by Massimo Poesio, took place in the unusual setting of the UU Botanic Gardens’ biggest greenhouse. The choice of the location turned out to be fitting, as the motto of the day was preservation — not of natural species, but of dissenting voices in natural language data. This goal has been gaining more importance given the drift towards language unification as embodied by omni-capable language models. In the meantime, the talks by Manfred Stede, Frances Yung, Matthias Orlikowski, Rupak Sarkar, Beiduo Chen, who kindly contributed their work, as well as by the DeMeVa project team itself showed that the compromise between linguistic diversity and technology is possible and can take a variety of forms. The workshop shed light on some of these forms: subjectivity-aware models, diversified generated explanations, and many others; meanwhile, many more are yet to be discovered, and we hope to hear about them in a year, in Workshop 4 (2026).