Bastien Liétard

I am a post-doctoral researcher in Computational Linguistics, within the COALA project at King's College London. This interdisciplinary project, lead by Dr Barbara McGillivray, aims at building a large-scale sense-annotated corpus of Latin texts spanning over 2000 years of history, and at analysing how word meanings have evolved over time in this corpus. In particular, my role consists in developping computational methods to automatically sense-annotate the final, large-scale corpus based on a smaller amount of manually annotated data. I am also working on data-driven studies of mechanisms of semantic change.

In early 2026, I graduated from my PhD at INRIA MAGNET, affiliated with University of Lille, conducted under the supervision of Marc Tommasi, Pascal Denis and Mikaela Keller (INRIA MAGNET team) and Anne Carlier (STIH unit, Paris Sorbonne University) .

In my PhD thesis, titled Concept-Aware Computational Models of Lexical Semantic Change, I explored data-driven approaches to the modeling of word-meaning associations and their evolution through the lens of induced semantic concepts. This contributed towards an onomasiological (i.e. meaning-to-words) study of these diachronic processes, alongside the more common semasiological perspective (i.e. word-to-meanings). I worked mostly to English and French historical data.

In 2021 I did a research intership in the CoAStAL group at the University of Copenhagen, under the supervision of Anders Søgaard. My reseach was about probing Language Models for geographic knowledge.

I also worked on Language Modeling and Semantics of adjectives, with a focus on so-called privative adjectives. This could be ranged under a specific type of adjective-noun compositionality.

Please feel free to reach out if any of the following research topic is of interest to you and you wish to collaborate:

• Structure of language models' representational space with respect to lexical meaning.
• Data-driven studies of the relations between Prototype Theory and the evolution of word meanings.
• Computational studies of onomasiology (meaning-to-word relations), especially in a diachronic perspective.
• Proposing and/or verifying laws and patterns of Lexical Semantic Change in historical data using computational approaches.
• Building interpretable representations of lexical meaning (e.g. feature-based vectors, definitions, ...) and their applications.
• Unsupervised learning (including but not limited to clustering) approaches to extract lexical meaning representations from word embeddings.
• Applications of computational models of Lexical Semantic Change to Romance languages.
• Original applications of computational models of lexical semantic change.
• Data-driven studies of specific kinds of compositionality, e.g. privative adjectives.

I am writing tabletop RPGs (both game design and narrative design) and even my own novel - at a very irregular pace.

I am generally interested in fantasy and sci-fi (especially cyberpunk) cultures, poetry, and 13th century european history.

I can also provide a lot of music recommendations, if I'm asked to.