Work

Emergent Syntax Discovery via Evolutionary Graph Grammars

Year: 2026
Type: preprint
Abstract: This research investigates the potential for discovering novel syntactic structures within human languages through an evolutionary approach utilizing graph grammars. The core hypothesis posits that complex linguistic phenomena, often challenging to model with traditional rule-based systems, emerge from a process of gradual, adaptive change. This work proposes a system where graph grammars are subjected to an evolutionary process, driven by a fitness function that assesses the coherence and expressiveness of the generated graph structures. A key innovation is the introduction of a mutation operator designed to emulate linguistic drift, introducing small, random variations into the grammar rules. This process aims to mimic the natural evolution of language, potentially leading to the discovery of entirely new, computationally-viable grammars. The study explores the feasibility of this approach and discusses its implications for understanding the origins and development of linguistic syntax. The primary goal is to demonstrate that an automated, evolutionary system can generate grammars capable of representing complex syntactic patterns, suggesting a pathway for computational models to better capture the emergent nature of human language. This work utilizes graph grammars as a framework for simulating this process, offering a novel perspective on syntactic discovery.
Language: English
Open Access status: green
OpenAlex ID: W7204575827
Cites:
Cited by:
Related to: 10