Semantic Markers in Judicial Reasoning: A Case Study of Toxic Dog Meat Judgments in China

Abstract
The sale of poisoned dogs poses food safety risks, and how such behavior is criminalized reflects prevailing social values. This study analyzes 53 Chinese criminal judgments (2014–2024) involving toxic dog meat to examine how linguistic expressions in judicial reasoning encode moral and institutional meanings. We manually annotated three binary semantic indicators—“poison needle,” “nickname,” and “live dog”—representing respectivelytechnical, organizational, and de‑animalizing linguistic features. Using Ordinary Least Squares (OLS)models with heteroskedasticity‑consistent (HC3) standard errors, we find significant positive effects of “poison needle” (b = 62.50, p = 0.020) and “nickname” (b = 131.16, p = 0.035) on sentence length, whereas “live dog” is not significant (p = 0.241). The findings suggest that courts respond more to semantic cues reflecting intent and organization than to moral expressions aboutthe value of life. This points to a hybrid paradigm in which moral and legal logics coexist: Traditional dog‑eating culture and emerging animal‑protection values run in parallel within legal language. By linking semantic indicators to sentencing outcomes, the study demonstrateshow language analysis can revealjudicial cognition and provides insightsforinterpretinglegal information systems. Keywords - Semantic Indicators, Judicial Reasoning, Toxic Food Crime, Moral Governance, China