Konstantinos Kogkalidis wins E.W. Beth Dissertation Prize
The Association for Language, Logic and Information (FoLLI) has awarded Utrecht PhD graduate Konstantinos Kogkalidis the E. W. Beth Dissertation Prize for his thesis ‘Dependency as Modality, Parsing as Permutation. A Neurosymbolic Perspective on Categorial Grammars’. The thesis was developed in the Institute for Language Sciences under the supervision of Michael Moortgat and Richard Moot.
“A remarkable synthesis”
The FoLLI jury committee praised the dissertation for its interdisciplinary approach, describing it as “a remarkable synthesis of formal linguistics, type theory, and machine learning.” They highlighted its rigor and innovative strength, noting that it develops methods where “logic and neural learning interact meaningfully”. The work was recognised as a “successful implementation of explainable AI” and “a milestone in the development of type-theoretic grammars”.
“I am obviously very happy,” says Kogkalidis. “Over the last few years, the summer schools organised by FoLLI have consistently been a breath of fresh air in a field contaminated by LLM hype and benchmarking frenzy. Being recognised by the association means a great deal to me – I couldn’t have hoped for a better conclusion to my PhD.”
Mathematical logic meets machine learning
Konstantinos’ thesis modernises type-logical grammars, a framework that connects language, logic, and computation. It introduces a new approach to thinking about sentence structure and meaning, combining mathematical logic with modern neural networks. By creating tools that can formally and reliably analyse and process language, the work demonstrates the practical potential of these grammars.
The research includes the development of a large dataset for the Dutch language and the creation of advanced systems that can learn to interpret sentences with higher precision and detail. The thesis shows how sophisticated theories of language can be applied to real-world language processing tasks.
About Kogkalidis and FoLLI
Before his PhD, Kogkalidis graduated from Utrecht University with a Master’s degree in AI. He is currently a post-doctoral researcher at Aalto University in Finland and a visiting researcher at the University of Bologna in Italy.
Τhe Association for Logic, Language, and Information (FoLLI) was founded in 1991 to advance the practicing of research and education on the interfaces between logic, linguistics, computer science and related disciplines. It has been awarding the annual E.W. Beth Dissertation Prize to outstanding PhD dissertations in logic, language, and information since 1998. The prize consists of a certificate, a monetary donation, and an invitation to submit the dissertation for publication as a volume in the FoLLI book series.