The Polysemy of “Bias” in Contemporary English

A Corpus-Based Disambiguation Analysis Across Social, Statistical, Juridical, Technical, and Everyday Registers

Abstract

In contemporary English, “bias” demonstrates several meanings that change with each register, as the same form of “bias” indicates different meanings. It is a social prejudice, a procedural ground to disqualify a judge, or a systematic error in research. Therefore, this study explores the distribution of these senses in political, legal, and daily English contexts, as well as what that distribution implies for public comprehension. Combining etymological reconstruction with corpus analysis, it operationalized the eight Cambridge Dictionary senses as a coding inventory and assembled eight sub-corpora from the english-corpora.org family, supplemented with United States court opinions and a daily-usage benchmark. Collocates within an L5 to R5 window were ranked by Log-Likelihood, with Mutual Information as a supplementary measure, and sense membership was tested through a chi-square test of independence with Cramér’s V and Monte Carlo confirmation. The seven sense profiles proved statistically separable, χ²(305, N = 6,922) = 13,559.35, p < .001, Cramér’s V = .62, the sharpest separation falling in the statistical sense and the weakest between the prejudice and political-legal senses. In everyday usage the preference reading predominated while the physical senses receded, concentrating comprehension risk where the civic stakes of “bias” are highest.