Sungkyunkwan University professor explores limits of AI language models in new book
Summary
Lee Han-jung, a Sungkyunkwan University professor, is set to publish an English-language book next month examining how AI language models handle linguistic meaning. The book presents a methodology combining linguistic frameworks and artificial neural networks and analyzes the limits of large language models, including ChatGPT. Lee’s research found that widely used models tend to reduce word meanings to generalized representations based on frequent patterns. The university said follow-up research will be presented at a linguistics conference in San Francisco in January 2027.
Key Facts
- The book, titled “Polysemy in Semantics with Contextualized Language Models,” will be published by De Gruyter Brill.
- The book joins the “Language and Computers” series, launched in 1988, which selects digital linguistics works through peer review.
- Lee’s research team combined large-scale textual data analysis, psychological experiments on human word comprehension and investigations into AI systems’ internal semantic representations.
- Lee presented the findings in a keynote lecture at the 2026 Seoul International Conference on Linguistics, held in Seoul in August.
- The follow-up research is scheduled for presentation at the 2027 Annual Meeting of the Linguistic Society of America.
Book introduces new approach to understanding context-dependent word meanings

Lee Han-jung, a professor in Sungkyunkwan University’s Department of English Language and Literature / Courtesy of Sungkyunkwan University
Lee Han-jung, a professor in the Department of English Language and Literature at Sungkyunkwan University, is set to publish an English-language book titled “Polysemy in Semantics with Contextualized Language Models” next month, the university said.
The book will be published by De Gruyter Brill, a renowned international academic publisher. It presents a new artificial intelligence (AI)-based methodology for studying linguistic meaning.
It is the latest addition to “Language and Computers,” an academic book series launched in 1988 that publishes outstanding works in digital linguistics selected through a rigorous peer-review process.
The book is particularly significant in that it bridges lexical semantics, traditionally rooted in the humanities, with computational models powered by advanced AI technologies.
It offers a scientific analysis of the fundamental limitations of large language models (LLMs), including ChatGPT, while exploring ways to advance their future development.
“Through this book, I sought to explore both the potential and limitations of AI language models and shed light on how AI affects human language and thought,” Lee said. “I will continue to pursue interdisciplinary research that bridges linguistics, AI and cognitive science, expanding the horizons of English studies and the humanities.”
Humans can naturally distinguish among the different meanings of a word based on its context. For example, Korean speakers can easily determine from context whether the word “nun” refers to an eye or snow.
For AI language models, however, accurately identifying context-dependent meanings and resolving lexical ambiguity remain major challenges.

The front and back covers of “Polysemy in Semantics with Contextualized Language Models, ” a book by Lee Han-jung, a professor in Sungkyunkwan University’s department of English language and literature / Courtesy of Sungkyunkwan University
Lee has developed a new approach to understanding how word meanings vary depending on context by integrating linguistic frameworks with artificial neural networks. Such networks allow computers to process language by representing words and expressions as numerical data.
Her research team also conducted a comprehensive study combining large-scale textual data analysis, psychological experiments on human word comprehension and in-depth investigations into the internal semantic representations of AI systems.
The findings provided scientific evidence that even widely used, state-of-the-art AI language models fail to fully capture the richness and complexity of human language.
Instead, the models tend to reduce word meanings to generalized representations based on frequently occurring patterns, limiting their ability to capture the full range of meanings.
Lee presented the findings in a keynote lecture to an international audience of linguists at the 2026 Seoul International Conference on Linguistics, held in Seoul in August.
The university said follow-up research will be presented at the 2027 Annual Meeting of the Linguistic Society of America, one of the world’s leading linguistics conferences, in San Francisco in January.
Explore More
- Q.
- Q.
- Q.