This interdisciplinary study explores subtle semantic variations in identical words, using computational linguistics methods to analyze semantic differences between South Korean conservative (Dong A) and progressive (Oh My News) newspapers. I selected 36 target words, abstract and concrete, and extracted sentences containing each from both newspapers. Employing Latent Dirichlet Allocation (LDA) Topic Modeling, I investigated favored discussion contexts. Latent Semantic Analysis (LSA) identified the 100 closest words to each target word, revealing related concepts. Lastly, Sentiment Analysis demonstrated emotional portrayals. The results showed that word meanings can be influenced by context, with quantifiable differences. The conservative newspaper often linked words to national security, while the progressive newspaper covered broader social issues, reflecting each outlet's media frame (Goffman, 1974). For example, ‘peace’ in conservative media suggests ‘peace against North Korean war threats,’ while in progressive media, it signifies ‘a societal state achieved through communication'. This contextual redefinition of words has implications for theoretical linguistics, validating fluid semantics and highlighting language's role in social cognition and ideology.

Authors:
Hana Jee, York St John University, United Kingdom

About the Presenter(s)
Dr. Hana Jee is currently a lecturer in Languages and Linguistics at York St John University. Her research interests lie in the relation between culture and language from a cognitive linguistic view.

Connect on Linkedin
https://www.linkedin.com/in/hana-jee-895b03253

See this presentation on the full schedule – On Demand Schedule

Bookmark