Perceived-Severity Hierarchy and Rating Disagreement in Korean Unethical Utterances: Harmfulness Judgments Are Not Reducible to Sentiment

Toxic language detection for large language model safety has largely been framed as binary classification, yet unethical expressions differ in severity across types and human judgments of severity are far from uniform. We analyze a large-scale Korean ethics-verification corpus with eight type labels (seven unethical types plus neutral) and the individual intensity votes (0–3) of five raters per sentence (363,154 sentences; 1,001,030 votes). First, the seven unethical types form a three-tier perceived-severity hierarchy—censure (1.42) < discrimination, hate, and abuse (1.53–1.59) < sexual, violence and crime incitement (1.74–1.75)—supported by post-hoc tests: all between-tier differences are significant with substantial effect sizes (Holm-corrected p < 0.05; |Cliff’s δ| = 0.14–0.45), while within-tier effects are negligible. Second, automatically estimated sentiment valence barely explains the hierarchy (Spearman ρ = −0.142; AUC = 0.621; replicated with a Korean-specific sentiment model and with preceding dialogue context included in the input); the two most severe types show the most neutral surface valence, so sentiment signals capture best precisely the least severe type. Third, inter-rater agreement is low (ordinal Krippendorff’s α = 0.373), yet the disagreement is not explained by the basic demographics the corpus records—rater gender and age (equivalence tests bound the pooled gender effect at |d|≤0.013 and the age correlation at |ρ|≤0.002; unrecorded rater characteristics remain untested)—and instead grows with item severity (sentence-level ρ = 0.248). Harmfulness judgment appears to be a dimension largely independent of sentiment, whose annotation may benefit from continuous scales, multi-rater averaging, and severity-proportional rating resources.

Authors

Institutions

Publication Details

Journal
Electronics
Published
2026-11-19
DOI
https://doi.org/10.3390/electronics15225394
Primary Topic
Hate Speech and Cyberbullying Detection
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Perceived-Severity Hierarchy and Rating Disagreement in Korean Unethical Utterances: Harmfulness Judgments Are Not Reducible to Sentiment

Seong-Yoon Shin, Yunam Cheong, Kwang-Seong Shin
Electronics
Hate Speech and Cyberbullying Detection
article

Perceived-Severity Hierarchy and Rating Disagreement in Korean Unethical Utterances: Harmfulness Judgments Are Not Reducible to Sentiment

Seong-Yoon Shin, Yunam Cheong, Kwang-Seong Shin
article en

Abstract

Toxic language detection for large language model safety has largely been framed as binary classification, yet unethical expressions differ in severity across types and human judgments of severity are far from uniform. We analyze a large-scale Korean ethics-verification corpus with eight type labels (seven unethical types plus neutral) and the individual intensity votes (0–3) of five raters per sentence (363,154 sentences; 1,001,030 votes). First, the seven unethical types form a three-tier perceived-severity hierarchy—censure (1.42) < discrimination, hate, and abuse (1.53–1.59) < sexual, violence and crime incitement (1.74–1.75)—supported by post-hoc tests: all between-tier differences are significant with substantial effect sizes (Holm-corrected p < 0.05; |Cliff’s δ| = 0.14–0.45), while within-tier effects are negligible. Second, automatically estimated sentiment valence barely explains the hierarchy (Spearman ρ = −0.142; AUC = 0.621; replicated with a Korean-specific sentiment model and with preceding dialogue context included in the input); the two most severe types show the most neutral surface valence, so sentiment signals capture best precisely the least severe type. Third, inter-rater agreement is low (ordinal Krippendorff’s α = 0.373), yet the disagreement is not explained by the basic demographics the corpus records—rater gender and age (equivalence tests bound the pooled gender effect at |d|≤0.013 and the age correlation at |ρ|≤0.002; unrecorded rater characteristics remain untested)—and instead grows with item severity (sentence-level ρ = 0.248). Harmfulness judgment appears to be a dimension largely independent of sentiment, whose annotation may benefit from continuous scales, multi-rater averaging, and severity-proportional rating resources.

ElectronicsVol. 15(22)
Kunsan National University (KR), Sunchon National University (KR)
Hate Speech and Cyberbullying Detection
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon