Creative Convergence or Imitation? Genre-Specific Homogeneity in LLM-Generated Chinese Literature

arXiv cs.CL / 3/17/2026

💬 OpinionIdeas & Deep AnalysisModels & Research

共有:

Key Points

The paper proposes a framework that combines Proppian narratology with a set of 34 narrative functions to analyze LLM-generated narratives in Chinese web literature.
It extends Propp's theory for modern web storytelling and builds a human-annotated corpus to study narrative structures in LLM output.
Experimental results show prominent homogenization in LLM-generated narratives, driven by rigid generation paradigms and a lack of correct understanding of narrative functions.
The work highlights implications for improving AI storytelling systems by addressing narrative-function semantics to diversify generated plots and resolutions.

Abstract

Large Language Models (LLMs) have demonstrated remarkable capabilities in narrative generation. However, they often produce structurally homogenized stories, frequently following repetitive arrangements and combinations of plot events along with stereotypical resolutions. In this paper, we propose a novel theoretical framework for analysis by incorporating Proppian narratology and narrative functions. This framework is used to analyze the composition of narrative texts generated by LLMs to uncover their underlying narrative logic. Taking Chinese web literature as our research focus, we extend Propp's narrative theory, defining 34 narrative functions suited to modern web narrative structures. We further construct a human-annotated corpus to support the analysis of narrative structures within LLM-generated text. Experiments reveal that the primary reasons for the singular narrative logic and severe homogenization in generated texts are that current LLMs are unable to correctly comprehend the meanings of narrative functions and instead adhere to rigid narrative generation paradigms.