Specific linguistic tropes and repetitive plot structures dominate stories produced by Large Language Models. Researchers identified a persistent lack of narrative tension and a tendency toward moralistic endings. These idiosyncrasies reveal systemic limitations in long-form coherence. Practitioners can use these findings to refine prompting strategies and improve narrative arc consistency.