Broadly, my research uses computational methods to study how socially meaningful concepts change over time, and how those shifts reflect wider cultural and societal dynamics. At the intersection of psychology, linguistics, and natural language processing, I develop theory-driven approaches for tracing conceptual change in historical text corpora using contextual embeddings, large language models, lexical resources, and statistical modelling. My current work focuses especially on mental health-related concepts and how their meanings and representations shift across scientific, media, and everyday language. This research program develops new ways of modelling conceptual change, evaluating computational methods, and applying them to important societal and cultural questions across languages, domains and disciplines.

Key Contributions:

• SIBling: A multidimensional framework for modelling lexical semantic change across three core dimensions: Sentiment, Intensity, and Breadth (SIB). [Paper]
• SIB Toolkit: A computational implementation of SIBling for measuring semantic change across SIB and related features such as salience and thematic content across scientific, media, and everyday domains.
• LSC-Eval: A benchmarking framework that uses LLM-generated synthetic historical data to evaluate methods for detecting lexical semantic change and identify dimension- and domain-sensitive approaches. [Paper]
• Sense-tracking pipeline, for estimating the prevalence of word senses in historical text corpora, to assists with LSC interpretability. [Paper])
• Applications: Case studies of mental health-related concepts, such as schizophrenia and autism, examining broader cultural trends including concept creep, pathologisation, and stigmatisation.