Directive Speech Acts in Chinese University Regulations: a Pragmatic Analysis of Human and AI-generated Normative Language

This dissertation investigates directive speech acts in Chinese university regulations and compares their realisation in authentic institutional texts with texts generated by ChatGPT. The study examines whether generated regulations reproduce the directive force and deontic organisation of authentic institutional discourse. The thesis is divided into five chapters. Chapter One outlines the theoretical framework, presenting speech act theory, directive functions, dialogic engagement and research on stereotypicality in AI-generated language. Chapter Two describes the research context and corpus, explaining the selection of admission and student administration documents from Capital Normal University and the generation of comparable texts through bilingual prompting. Chapter Three illustrates the comparative methodology, including the segmentation of regulatory provisions and annotation of directive categories, linguistic forms and illocutionary strength. It also describes the prompting procedure and the criteria used to analyse the generated texts. Chapter Four presents the analysis of directive categories, strength, markers and realisation types across the two corpora and three prompting conditions. The generated texts reproduce recognisable regulatory forms but contain no implicit structural constraints of the kind identified in the authentic corpus. Chapter Five discusses the findings, their contribution and the limitations of the study. The dissertation ends with a bibliography and sitography.