Research in the Natural Language Processing and Digital Humanities unit focuses on automated analysis, interpretation and generation of human language and their extension towards language technology. Our work encompasses a range of topics within natural language processing (NLP), such as syntactic parsing, computational semantics and pragmatics, discourse processing, dialogue modelling, machine translation and multilingual NLP.

Our interdisciplinary focus, incorporating insights from linguistics, cognitive science, psychology and machine learning, gives our group’s research a unique profile, having led to numerous distinctive contributions over four decades. Whilst well-known for its influential research in the areas of statistical parsing, syntax based machine translation and semantic role labeling, recently the group has pioneered methods for interpretability of neural models, graph neural networks for NLP and few-shot learning applied to NLP tasks.

Another prominent research direction focuses on the development of societally-oriented and responsible NLP technology, as well as applications in digital humanities, media studies and computational social science. To this end, the group has explored how statistical and neural models can retrieve information from text to help answer questions in the humanities, ranging from history to philosophy, and aid large-scale data-driven analysis of cultural artifacts.