Linguistic Evaluability: Preserving Independent Judgment in Artificial Intelligence-Mediated Communication

Abstract

Artificial intelligence increasingly mediates the language through which societies communicate, interpret, and evaluate knowledge. Existing AI governance approaches emphasize transparency, accountability, documentation, verification, and explanation, yet these mechanisms depend upon communication that preserves the relationships necessary for independent evaluation. This paper introduces linguistic evaluability as a philosophical dimension of communication concerned with whether language preserves sufficient structure for later evaluators to reconstruct how judgments emerge. The investigation employs E-Prime as a philosophical probe rather than as a proposed linguistic reform. By constraining familiar patterns of expression, the analysis examines which evaluative relationships become more visible, including agency, evidence, criteria, context, qualification, and uncertainty. The resulting framework extends beyond E-Prime by identifying linguistic functions that support reconstructability across diverse communicative settings. The paper argues that communication does more than transmit conclusions. It preserves or compresses the relationships through which conclusions remain open to examination. As artificial intelligence increasingly participates in producing institutional, scientific, educational, and public communication, linguistic evaluability provides an additional perspective for examining how societies preserve the conditions necessary for independent human judgment.