LLMs in Discourse Analysis and Production: A Systematic Review through the Lens of Appraisal Theory
Year: 2026
Type: review
Study design: Systematic Review
Abstract: The application of Large Language Models (LLMs) and Generative Artificial Intelligence (GenAI) to the study of Martin and White’s Appraisal Theory offers promising opportunities for automated discourse analysis. However, it faces substantial challenges due to the conceptual complexity of the framework and the models' limitations in handling indirect, interpersonal meanings. Despite growing interest, no systematic overview of existing research currently exists that focuses on: (1) the effectiveness of LLMs as tools for the automatic annotation of linguistic evaluative resources and (2) the comparative production of evaluative discourse between humans and machines. To address this gap, a systematic literature review was conducted following the PRISMA 2020 methodology. Nine peer-reviewed studies were identified and analyzed, focusing on the above-mentioned two points. The review shows that current LLMs perform well when evaluation is explicit, structurally marked, and supported by clear contextual cues. Their weaknesses become visible when tasks require recognition of invoked meaning, cultural knowledge, and dialogic positioning. In production tasks, LLMs frequently rely on safe impersonal patterns, often reducing interpersonal warmth and expressive range. Future developments should focus on creating richer, context-based resources and developing "human-in-the-loop" approaches to fully realize the potential of these technologies in deep interpersonal analysis and interpretation.
Institution: Hellenic Open University
Language: English
Subfield: Linguistics and Language
Field: Social Sciences
Domain: Social Sciences
Open Access status: hybrid
OpenAlex ID: W7214599059