Frequency Accelerates Semantic Change: Evidence from 500 Years of Korean

2026-06-25 · Source: Paper Index on ACL Anthology · Field: Science & Research — Social Sciences & Behavioral Studies, Research Methodology & Innovation · Depth: Expert, quick

Summary

A study analyzing 500 years of Korean language, specifically from the 15th to the 20th centuries, challenges the "law of conformity" which posits that frequent words are semantically stable. Using diachronic word embeddings trained on historical corpora, researchers found a robust positive correlation between word frequency and semantic shift in Korean. High-frequency Korean words demonstrated greater semantic change, contrary to the established law. This pattern was confirmed through six robustness controls and validated by an English replication. Partial correlation analysis further revealed that the mediating role of polysemy in the frequency–change relationship is not constant, varying with time resolution and corpus homogeneity. The findings suggest this reversal is connected to frequency-driven reductive processes prevalent in Korean, such as grammaticalization, semantic bleaching, and domain shift, indicating that the frequency–change relationship is not a universal regularity but depends on language typology and analytical conditions.

Key takeaway

For research scientists analyzing language evolution or developing diachronic NLP models, you should critically re-evaluate the assumption that high-frequency words are semantically stable. This study demonstrates that for languages like Korean, frequency can accelerate semantic change, necessitating a more nuanced approach. Your models and analyses should account for language-specific typological features and the dynamic influence of factors like polysemy, rather than assuming universal linguistic regularities.

Key insights

Frequent Korean words undergo more semantic change, challenging the universal "law of conformity" and highlighting language typology's role.

Principles

• The "law of conformity" is not universal.
• Frequency can accelerate semantic shift.
• Language typology influences change patterns.

Method

Diachronic word embeddings trained on historical corpora (15th–20th century Korean) were used to measure semantic shift and correlate it with word frequency.

In practice

• Re-evaluate semantic stability assumptions.
• Consider language typology in diachronic studies.
• Analyze polysemy's dynamic role.

Topics

• Diachronic Word Embeddings
• Semantic Change
• Korean Language
• Language Typology
• Historical Linguistics
• Law of Conformity

Best for: NLP Engineer, AI Scientist, Research Scientist

Related on AIssential

• The Dynamics of Human and AI-Generated Language: How Semantics Fluctuates across Different Timescales — The ACW-0 semantic timescale analysis effectively distinguishes temporal organization of generic versus specific content in human and AI speech.
• From sunblock to softblock: Analyzing the correlates of neology in published writing and on social media — Neology correlates similarly across published writing and social media, though topic popularity's influence varies.
• Multilinguality of Large Language Models From a Structural Perspective — LLMs' multilinguality involves structural differences, with low-resource languages diverging more from English.
• Evaluating Cross-lingual Knowledge Consistency in Code-Mixed vis-a-vis Indian Languages using IndicKLAR — Code-mixing inputs significantly improves large language model knowledge recall in Indian languages, nearly matching English performance.
• Modelling the Morphology of Verbal Paradigms: A Case Study in the Tokenization of Turkish and Hebrew — Tokenization strategy critically impacts Transformer models' ability to represent complex morphology, especially in non-concatenative languages.
• Tracking Representation Dynamics in Large Language Models with Persistent Homology — Persistent homology uncovers internal representation dynamics in LLMs during fine-tuning, showing early topological reorganization and objective-specific trajectories.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Paper Index on ACL Anthology.