Language Models for Language Science
(WS26/27)
(WS26/27)
Large Language Models (LLMs) are neural networks trained on large corpora of texts to predict and generate linguistic content. The language they generate is strikingly similar to human language, and this raises some fundamental questions about what they learn from the linguistic input they are exposed to and what they can tell us about the nature of human language and human language processing.
In this seminar, we will look at whether and how LLMs can be used as tools for empirical research in language science, and vice versa, whether insights from language science can inform our understanding and development of these technologies.
After a short and accessible introduction to LLMs, we will focus on the following topics:
Linguistic Structure: Do LMs implicitly learn linguistic structure in ways that are comparable to those proposed by linguistic theories? Can LLM behavior provide evidence for or against specific theoretical claims?
Language Learning: How does the massive amount of text required to train an LLM compare to the linguistic input available to children during language acquisition? What can differences between LLMs and children in learning trajectories tell us about human language learning mechanisms?
Language Processing: Can LLM surprisal or next-word prediction serve as a proxy for human processing difficulty, for example in predicting reading times? Where do LLM and human processing patterns diverge, and what might these differences reveal about the mechanisms underlying human sentence processing?
Language in the Brain: To what extent do the internal representations of language models correspond to neural activity recorded during human language comprehension, for example with fMRI or EEG? Can representations learned by artificial neural networks provide useful models of human neurocognitive language processing?
Language of instruction: The course readings will consist of research articles in English. Students should therefore feel comfortable reading scientific literature in English. The seminar will be taught primarily in English but student presentations may be given in German if preferred.
Start date: 16 October
Time: Fridays, 10:15 - 11:45
Classroom: Building C7 3, Seminarraum 1.14
Contact: delogu@lst.uni-saarland.de
Amouyal, S., Meltzer-Asscher, A., & Berant, J. (2024, March). Large language models for psycholinguistic plausibility pretesting. In Findings of the Association for Computational Linguistics: EACL 2024 (pp. 166-181).
Chomsky, N., Roberts, I., & Watumull, J. (2023). Noam chomsky: The false promise of chatgpt. The New York Times, 8(4), 177-179.
Connell, L., & Lynott, D. (2024). What can language models tell us about human cognition?. Current Directions in Psychological Science, 33(3), 181-189.
Contreras Kallens, P., Kristensen‐McLachlan, R. D., & Christiansen, M. H. (2023). Large language models demonstrate the potential of statistical learning in language. Cognitive Science, 47(3), e13256.
Cuskley, C., Woods, R., & Flaherty, M. (2024). The limitations of large language models for understanding human language and cognition. Open Mind, 8, 1058-1083.
Huang, K. J., Arehalli, S., Kugemoto, M., Muxica, C., Prasad, G., Dillon, B., & Linzen, T. (2024). Large-scale benchmark yields no evidence that language model surprisal explains syntactic disambiguation difficulty. Journal of Memory and Language, 137, 104510.
Hwang, H. (2026). Can large language models be viewed as a cognitive model of human language? Not yet, regardless of reasoning capability and size. Acta Psychologica, 267, 107025.
Krieger, B., Brouwer, H., Aurnhammer, C., & Crocker, M. W. (2025). On the limits of LLM surprisal as a functional explanation of the N400 and P600. Brain Research, 149841.
Lang, C., Kretzschmar, F., & Hansen, S. (2025, September). Using LLMs for experimental stimulus pretests in linguistics. Evidence from semantic associations between words and social gender. In Proceedings of the 21st Conference on Natural Language Processing (KONVENS 2025): Long and Short Papers (pp. 326-332).
Levy, R., Kim, Y., & Fox, D. (2025). The science of language in the era of generative ai. https://doi.org/10.21428/e4baedd9.f6a0052d
Mahowald, K., Ivanova, A. A., Blank, I. A., Kanwisher, N., Tenenbaum, J. B., & Fedorenko, E. (2024). Dissociating language and thought in large language models: a cognitive perspective. Trends in Cognitive Sciences.
Millière, R. (2024). Language models as models of language. arXiv preprint arXiv:2408.07144.
Pissani, L., Jobanputra, M., & Demberg, V. (2026). LLMs replicate metaphor norms based on word co-occurrence but struggle with topic-vehicle mappings. Frontiers in Language Sciences, 5, 1756514.
Futrell, R., & Mahowald, K. (2026). How linguistics learned to stop worrying and love the language models. Behavioral and Brain Sciences, 49, e198.
Piantadosi, S. (2023). Modern language models refute Chomsky’s approach to language. Lingbuzz 7180.
Staub, A. (2025). Predictability in language comprehension: Prospects and problems for surprisal. Annual Review of Linguistics, 11(1), 17-34.
Timkey, W., Dillon, B., & Linzen, T. (2026). Why are language models less surprised than humans? Testing the Parse Multiplicity Mismatch Hypothesis. arXiv preprint arXiv:2605.15440.
Tuckute, G., Kanwisher, N., & Fedorenko, E. (2024). Language in brains, minds, and machines. Annual Review of Neuroscience, 47(1), 277-301.