Automatic storytelling is a difficult NLP task that poses a challenge even for state-of-the-art large language models. This paper proposes a pipeline for interactive fairy tale generation in a mixed-initiative setting. Our approach introduces a story goal as a stopping condition, imposes minimal structure on the narrative in the form of a simple emotional arc, and controls the transition between the stages of the story via system prompt engineering. The resulting framework reconciles creating a structured and complete short-form narrative with retaining player agency and allowing users to influence the storyline through their input. We evaluate our approach with several proprietary and open-source language models and examine its transferability to different languages, specifically English and Russian.

Within the currently dominant Minimalist framework for syntax (Chomsky, 1995, 2000), it is not uncommon to encounter multiple proposals for the same natural language pattern in the literature. We investigate the possibility of evaluating and comparing analyses of syntax phenomena, implemented as minimalist grammars (Stabler, 1997), from a quantitative point of view. This paper introduces a principled way of making linguistic generalizations by detecting and eliminating syntactic and phonological redundancies in the data. As proof of concept, we first provide a small step-by-step example transforming a naive grammar over unsegmented words into a linguistically motivated grammar over morphemes, and then discuss a description of the English auxiliary system, passives, and raising verbs produced by a prototype implementation of a procedure for automated grammar optimization.

Probabilistic approaches have proven themselves well in learning phonological structure. In contrast, theoretical linguistics usually works with deterministic generalizations. The goal of this paper is to explore possible interactions between information-theoretic methods and deterministic linguistic knowledge and to examine some ways in which both can be used in tandem to extract phonological and morphophonological patterns from a small annotated dataset. Local and nonlocal processes in Mishar Tatar (Turkic/Kipchak) are examined as a case study.