Agent-based models for the evolution of morphological alternation patterns

The article presents a multi-agent simulation framework for studying the emergence of morphological alternation patterns in languages, specifically focusing on how forms like "go" and "went" can evolve through agent interactions. It features an AI Historical Linguist, a Large Language Model-based tool for evaluating the realism of generated morphologies against real-world data, and allows for extensive customization with realistic phonological rules and large lexicons. This work is significant for practitioners as it provides a robust method for exploring language evolution and morphological changes, with implications for understanding linguistic dynamics in AI language models.