Linguagem judicial e garantia de acesso à informação: uma análise linguística a partir de decisões em ações sobre crédito consignado

Abstract

This thesis investigates the role of legal language in ensuring access to information and justice, with an emphasis on court decisions involving fraudulent payroll-deductible loans targeting elderly individuals. The research adopts a critical, empirical, and interdisciplinary approach, integrating Corpus Linguistics and Systemic-Functional Linguistics with a focus on legal and forensic communication. The analysis of decisions from the Court of Justice of Minas Gerais (TJMG) reveals a judicial language characterized by excessive formalism and a recurrent use of authoritative arguments, which hinders laypeople’s understanding and reinforces inequalities in access to justice. Artificial Intelligence (AI), particularly ChatGPT, was used as a diagnostic and experimental tool for textual simplification. The results show significant improvements in clarity and readability without compromising legal reasoning, although limitations of AI – such as the loss of contextual nuances – are acknowledged. The thesis concludes that the true challenge lies not in adopting artificial intelligence as a definitive solution, but in reformulating legal language as a communicative practice. Overcoming hermetic technicality requires recognizing language as an instrument of power and exclusion, and proposing more accessible and democratic models without renouncing legal reasoning and technique. It is a matter of abandoning the fetish of jargon and embracing clarity as a pathway to the effective exercise of the right to information and citizenship.