Melancholy clouds, mud-seeking robots, and Arum lily cave choirs: One-sentence multimodal stories with text-to-image diffusion models and student-generated declarative sentence machines