Yiming Liang
Postdoctoral researcher | Department of Linguistics | Universiteit Gent
Postdoctoral researcher | Department of Linguistics | Universiteit Gent
I am a postdoctoral researcher in linguistics at Universiteit Gent (Ghent, Belgium), affiliated with the ERC Causality project led by Alexandra Simonenko. My current focus is on the investigation of causal relationships among various diachronic changes in the morpho-syntax of West Germanic languages.
I completed my PhD in Linguistics from Université Paris Cité (Paris, France) and Laboratoire de Linguistique Formelle (LLF), under the supervision of Heather Burnett and Pascal Amsili. My thesis, funded by the ERC SMIC project (PI: Heather Burnett), employed generative syntax and information theory to explain preferential patterns in morpho-syntactic variation in oral French.
My research interests
My research aims to understand how and why languages vary and change, by interacting formal and quantitative syntax, information theory, large language models (LLMs) and efficiency-based accounts. For example, I investigate why languages differ in how they express grammatical subjects: some, like Spanish, allow subjects to be omitted ([ella] está contenta, ‘she is happy’), while others, like English, generally require them (she is happy, but not is happy). My research has demonstrated how probabilistic and numerical approaches can refine parameters in formal syntax and improve their capacity to account for gradience, variability, and universals in languages.
I have a solid background in computational linguistics. I have eight years of extensive experience in Python programming and corpus linguistics. I have also developed substantial experience in training and fine-tuning LLMs, including training a transformer-based multilingual parser for historical languages and extracting probability estimates from LLMs to investigate how efficiency shapes linguistic variation. I am particularly interested in developing information-theoretic metrics to quantify linguistic notions using LLMs, as well as understanding the internal computations of LLMs through probing and mechanistic intervention methods.
Below is an example using bigram models to estimate the information density (approximated by surprisal) of a complement clause (CC) given the main verb as context. The results show that when the CC is highly predictable, speakers tend to omit que to raise information density toward the average. When the CC is highly unpredictable, speakers prefer to realize que to lower information density toward the average. This suggests that que-omission functions as a strategy to avoid peaks and troughs in the information-density profile, supporting the Uniform Information Density hypothesis.
Omission of "que" (red line) brings the information density at the beginning of complement clause (CC) closer to the average, in contrast with the version with "que" (blue line). The omission of "que" is predicted to be preferred when the information density at the beginning of CC is low, as the omission of "que" avoids the trough in the information density profile of the sentence.
Omission of "que" (red line) deviates the information density at the beginning of CC from the average, compared with the version with "que" (blue line). The omission of "que" is predicted to be dispreferred when the information density at the beginning of CC is high, as the realization of "que" avoids the peak in the information density profile of the sentence.
Liang, Y., & Zhao, F. (2026). Lexicalized Constituency Parsing for Middle Dutch: Low-resource Training and Cross-Domain Generalization. In Proceedings of the Fifteenth Language Resources and Evaluation Conference (LREC 2026) (pp. 10276–10290). European Language Resources Association (ELRA). https://doi.org/10.63317/3bg3jhcbj8in.
Liang, Y., Burnett, H. & Amsili, P. (2026). The Polarity Effect in Future Temporal Reference: new insights from Parisian French. Revue canadienne de linguistique/Canadian Journal of Linguistics. Published online 2026:1-14. https://doi.org/10.1017/cnj.2025.6.
Liang, Y., Amsili, P. & Burnett, H. (2025). Soft Locality Restrictions in Negative Concord: Evidence from the French Future Polarity Effect. Natural Language & Linguistic Theory 43, 1731-1769. https://doi.org/10.1007/s11049-024-09650-4
Liang, Y., Donati C., & Burnett H. (2024). French Subject Doubling: A Third Path. RLLT23, eds. Lisa Brunetti, Ioana Chitoran & Alexandru Mardale. Special Issue of Isogloss. Open Journal of Romance Linguistics, 10(7)/12, 1-29. https://doi.org/10.5565/rev/isogloss.420
Liang, Y., Amsili, P., Burnett, H. & Demberg, V. (2024). Uniform information density explains subject doubling in French. Proceedings of the Annual Meeting of the Cognitive Science Society, 46. Retrieved from https://escholarship.org/uc/item/645673fs
Liang, Y., Amsili, P. & Burnett, H. (2023). Uniformité de la densité informationnelle : le cas du redoublement du sujet. In Actes de la 30e Conférence sur le Traitement Automatique des Langues Naturelles (TALN), volume 1 : travaux de recherche originaux – articles longs, pp. 164–176. https://aclanthology.org/2023.jeptalnrecital-long.13
Liang, Y., Amsili, P. & Burnett, H. (2021). New ways of analyzing complementizer drop in Montréal French: Exploration of cognitive factors. Language Variation and Change, 33(3), 359-385. DOI: https://doi.org/10.1017/S0954394521000223
Chen, S., Amsili, P., & Liang, Y. (2021). Inter-clausal Anaphora in Chinese Conditionals: a Multi-factorial Analysis. In Hu, K., Kim, J. B., Zong, C., & Chersoni, E. (Eds.), Proceedings of the 35th Pacific Asia Conference on Language, Information and Computation (pp. 140-148). arXiv: https://aclanthology.org/2021.paclic-1.15
Liang, Y. & Simonenko, A. “Verbal agreement syncretism and null subjects in Historical Icelandic”. Diachronic Generative Syntax (DiGS27), June 16th-19th, 2026, Padova, Italy.
Liang, Y. & Simonenko, A. (2024). “Quantify syncretism in subject-verb agreement marking”. New Ways of Analyzing Variation 52 (NWAV52), October 7th-9th, 2024, Miami Beach, United States.
Liang, Y., Amsili, P., Burnett, H. & Demberg, V. (2024). “Uniform information density explains subject doubling in French”. The 46th Annual Conference of the Cognitive Science Society (Cogsci), July 24th-27th, 2024, Rotterdam, the Netherlands. (acceptance rate for talks: 19%)
Liang, Y., Donati, C. & Burnett, H. (2023). “French clitics: A paradox from variation”. Workshop on Variation and Syntactic Theory at New Ways of Analyzing Variation 51 (NWAV 51), October 14th, 2023, New York, United States.
Liang, Y., Amsili, P. & Burnett, H. (2023). “Uniformité de la densité informationnelle : le cas du redoublement du sujet (The Uniform Information Density Hypothesis: the case of subjet-doubling in French).” Traitement automatique des langues naturelles (TALN) 2023, June 5th-9th, Paris, France.
Liang, Y., Donati, C. & Burnett, H. (2023). “French Subject Doubling: A Third Path.” 53rd Linguistic Symposium on Romance Languages (LSRL 53), June 26th–30th, 2023, Paris, France.
Martin, A., Abbou, J., Dénigot, Q., Liang, Y. & Burnett, H. (2022). “The power of the written word: A re-analysis of the social meanings of variable liaison.” Sociolinguistics Symposium 24, July 13th–16th, 2022, Ghent, Belgium.
Martin, A., Liang, Y., Abbou, J., Dénigot, Q., Amsili, P. & Burnett H. (2021). “Tu n’utiliseras pas le futur proche: a corpus and experimental study of the effect of polarity on future variation in Metropolitan French.” New Ways of Analyzing Variation 49 (NWAV49), October 19th-24th, 2021, Austin, United States (online).
Liang, Y., Amsili, P. & Burnett H. (2021). “Exploring the polarity effect on future variation in Parisian French.” UK Language Variation and Change 13 (UKLVC13), September 8th-10th, 2021, Glasgow, United Kingdom (online).
Chen, S., Liang, Y. & Amsili, P. (2021). “Chinese Inter-clausal Anaphora in Conditionals: A Mixed Effects Logistic Regression Study.” 34th Journées de Linguistique Asie Orientale (JLAO34), July 7th-9th, 2021, Paris, France (online).
Liang, Y., Amsili, P. & Burnett H. (2021). “Revisiter l'omission du complémenteur que en français montréalais: exploration de facteurs cognitifs.” Annual conference of the Canadian Linguistic Association 2021, June 4th-7th, 2021, Canada (online).
Liang, Y. & Zhao, F. “Lexicalized Constituency Parsing for Middle Dutch: Low-resource Training and Cross-Domain Generalization”. Poster presented at the 15th Language Resources and Evaluation Conference (LREC2026), May 11th-16th, 2026, Palma de Mallorca, Spain.
Liang, Y. & Simonenko, A. “Revisiting Taraldsen’s Generalization: A New Metric for Quantifying Syncretism”. Poster presented at Diachronic Generative Syntax (DiGS26), June 23th-26th, 2025, Oxford, United Kingdom.
Liang, Y., Abbou, J., Amsili, P., Dénigot, Q., Martin, A., & Burnett, H. (2022). “Soft Intervention Effects in Sociolinguistic Variation''. Poster presented at Generative Linguistics in the Old World 45 (GLOW45), April 27th–29th, 2022, London, United Kingdom.
Liang, Y., Amsili, P. & Burnett H. (2019). “Faut je t'apporte quelque chose: Complementizer drop in Quebec French''. Poster presented at Colloque de Syntaxe et Sémantique à Paris 19 (CSSP19), October 2nd-4th, 2019, Paris, France. [PDF]
Chen, S., Yuan, C. & Liang, Y. (2019). “Interclausal Anaphora in Chinese: The Case of Conditional and Causal Sentences''. Poster presented at Summer School Conditionals in Paris – logic, linguistics and psychology, June 3rd-7th, 2019, Inalco, Paris, France.
Co-lecturer (Medelesgever) for General lingusitics: Language Variation and Change, Department of Linguistics, Universiteit Gent, Ghent, Belgium
Final year of a Bachelor's course (Responsible lecturer: Professor Alexandra Simonenko)
Lectures in Sociolinguistics, historical sociolinguistics and typology.
Taught in English
Responsible lecturer (Verantwoordelijke lesgever) for Generative Grammar, Department of Linguistics, Universiteit Gent, Ghent, Belgium
Master's level course
Lectures in Generative Syntax and Formal Semantics
Taught in English
Co-lecturer (Medelesgever) for Generative Grammar, Department of Linguistics, Universiteit Gent, Ghent, Belgium
Master's level course (Responsible lecturer: Professor Alexandra Simonenko)
6 lectures in Generative Syntax
Taught in English
Teaching assistant (Chargée de TD) for LING101: Introduction to Linguistics, Department of Cognitive Sciences, Cogmaster, École Normale Supérieure/EHESS/Univeristé Paris Cité, Paris, France
Master's level course (Responsible lecturer: Professor Salvador Mascarenhas) [syllabus]
Led weekly discussion sessions, grading, office hours
Taught in English
Teaching assistant (Chargée de TD) for Semantics, Department of Linguistics, Université Paris Cité, Paris, France
Final year of a Bachelor's course (Responsible lecturer: Professor Lucia M. Tovena)
Led weekly discussion sessions, grading, designed quizzes
Taught in French
If you wish to get in touch with me regarding my research or potential collaborations, please feel free to contact me at the following address:
flavie.liang at gmail.com
yiming.liang at ugent.be