Personal profile
Research interests
I’m Harish Tayyar Madabushi, Associate Professor (Senior Lecturer) in Artificial Intelligence at the University of Bath. I work at the intersection of Natural Language Processing, reasoning, linguistics and AI safety, asking a simple but stubbornly hard question: how do large language models actually learn to use language and reason, and what does that mean for how we deploy them in the real world?
My research treats usage-based learning as the foundation for how language competence and bounded reasoning emerge in large language models. On the linguistic side, I draw on Construction Grammar (CxG) to understand what kinds of constructions models really acquire; on the modelling side, I develop Context-Directed Extrapolation, a framework for explaining so-called “emergent abilities” not as magic or AGI, but as models extrapolating from statistical priors when guided by context. This work has reshaped parts of the debate around LLM capabilities, influencing research, policy discussions (including the UK AI Safety Summit), and industry practice.
I organise my work into three strands:
-
Emergent reasoning and limits of LLMs: mapping what models can and cannot do, and when scaling stops being a silver bullet.
-
Linguistically grounded evaluation: using insights from theoretical linguistics, specifically construction grammar to probe abstraction, generalisation and comprehension in LLMs.
-
Applications in sensitive domains: from debiasing speech systems and supporting public services, to studying how LLMs may nudge users towards extremist content, and designing scaffolds (like divergent chain-of-thought and frame-based retrieval) that make models more reliable, interpretable and standard-aligned.
Before starting his PhD in automated question answering at the University of Birmingham, Dr. Tayyar Madabushi founded and headed a social media data analytics company based out of Singapore.
Publication record on Google Scholar.
Read more about my work on my website.
Willing to supervise doctoral students
I am looking for highly motivated students for a PhD in Natural Language Processing at the University of Bath.
If you are interested in pursuing a PhD in NLP, please get in touch.
External positions
Honorary Research Fellow, University of Birmingham
15 Mar 2021 → 14 Mar 2024
Fingerprint
- 1 Similar Profiles
Collaborations and top research areas from the last five years
Projects
- 1 Finished
-
Wyser: Reducing Bias in ASR
Tayyar Madabushi, H. (PI)
Innovate UK, Innovate UK Business Connect
1/02/24 → 31/03/25
Project: Central government, health and local authorities
Research output
-
Are Emergent Abilities in Large Language Models just In-Context Learning?
Lu, S., Bigoulaeva, I., Sachdeva, R. S., Tayyar Madabushi, H. & Gurevych, I., 31 Aug 2024, Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics. Ku, L.-W., Martins, A. F. T. & Srikumar, V. (eds.). Long Papers ed. Bangkok, Thailand: Association for Computational Linguistics, Vol. 1. p. 5098–5139 42 p. 2024.acl-long.279. (Proceedings of the Annual Meeting of the Association for Computational Linguistics; vol. 1).Research output: Chapter or section in a book/report/conference proceeding › Chapter in a published conference proceeding
Open Access22 Link opens in a new tab Citations (SciVal) -
CxGBERT: BERT meets Construction Grammar
Tayyar Madabushi, H., Romain, L., Divjak, D. & Milin, P., 1 Dec 2020, Proceedings of the 28th International Conference on Computational Linguistics. Barcelona, Spain (Online): International Committee on Computational Linguistics (ICCL), p. 4020-4032 13 p. 2020.coling-main.355Research output: Chapter or section in a book/report/conference proceeding › Chapter in a published conference proceeding
Open Access31 Link opens in a new tab Citations (SciVal) -
Improving the understandability of clinical guidelines: development and evaluation of a GPT-4–based pipeline
Jones, M. D., Torgbi, M. & Tayyar Madabushi, H., 23 Feb 2026, In: Journal of Medical Internet Research. 28, 11 p., e81915.Research output: Contribution to journal › Article › peer-review
Open Access1 Link opens in a new tab Citation (SciVal) -
Adapting Whisper for Regional Dialects: Enhancing Public Services for Vulnerable Populations in the United Kingdom
Torgbi, M., Clayman, A., Speight, J. J. & Madabushi, H. T., 19 Jan 2025, VarDial 2025 - 12th Workshop on NLP for Similar Languages, Varieties and Dialects, Proceedings of the Workshop. Scherrer, Y., Jauhiainen, T., Ljubesic, N., Nakov, P., Tiedemann, J. & Zampieri, M. (eds.). New York, U. S. A.: Association for Computational Linguistics (ACL), p. 29-38 10 p. (VarDial 2025 - 12th Workshop on NLP for Similar Languages, Varieties and Dialects, Proceedings of the Workshop).Research output: Chapter or section in a book/report/conference proceeding › Chapter in a published conference proceeding
Open Access2 Link opens in a new tab Citations (SciVal) -
Beyond Memorization: Assessing Semantic Generalization in Large Language Models Using Phrasal Constructions
Scivetti, W., Torgbi, M., Blodgett, A., Shichman, M., Pellegrin, T., Bonial, C. & Tayyar Madabushi, H., 31 Dec 2025, IJCNLP-AACL 2025: The 4th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics and the 14th International Joint Conference on Natural Language Processing.Research output: Chapter or section in a book/report/conference proceeding › Chapter in a published conference proceeding
Open Access
Activities
- 1 Keynote presentation or prestigious invited talk
-
Keynote: The Joint Workshop on Multiword Expressions and Universal Dependencies (MWE-UD) @LREC-COLING 2024
Tayyar Madabushi, H. (Speaker)
25 May 2024Activity: Academic conferences and events (excluding conference publications) › Keynote presentation or prestigious invited talk