Discovering Properties of Inflectional Morphology in Neural Emergent Communication

2025-08-07

AI Summary

Current emergent communication (EmCom) research overlooks inflectional morphology—particularly duality of patterning, compositionality, and fusion—fundamental to natural language. Method: We propose a phonologically constrained communication framework wherein agents engage in a vocabulary-limited attribute-value reconstruction game to emulate duality of patterning; we further introduce quantitative metrics for fusion and compositionality, enabling neural agents to spontaneously encode multiple grammatical features within single morphemes—a hallmark of human inflectional morphology. Contribution/Results: Experiments demonstrate that phonological constraints significantly enhance compositional morpheme generation, and the emergent languages exhibit inflectional properties highly consistent with natural languages—including systematic fusion of syntactic features. This work establishes a novel paradigm for investigating how neural communication systems evolve toward language-like structure and provides interpretable, morphology-aware evaluation tools for emergent linguistic behavior.

Technology Category

• Natural Language Processing: Lexical Semantics and Morphology
• Multiagent Systems: Agent Communication
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

Emergent communication (EmCom) with deep neural network-based agents promises to yield insights into the nature of human language, but remains focused primarily on a few subfield-specific goals and metrics that prioritize communication schemes which represent attributes with unique characters one-to-one and compose them syntactically. We thus reinterpret a common EmCom setting, the attribute-value reconstruction game, by imposing a small-vocabulary constraint to simulate double articulation, and formulating a novel setting analogous to naturalistic inflectional morphology (enabling meaningful comparison to natural language communication schemes). We develop new metrics and explore variations of this game motivated by real properties of inflectional morphology: concatenativity and fusionality. Through our experiments, we discover that simulated phonological constraints encourage concatenative morphology, and emergent languages replicate the tendency of natural languages to fuse grammatical attributes.

Problem

• Study neural emergent communication's inflectional morphology properties
• Develop metrics for naturalistic inflectional morphology comparison
• Explore phonological constraints' impact on concatenative morphology

Innovation

• Small-vocabulary constraint simulates double articulation
• Novel setting analogous to inflectional morphology
• Phonological constraints encourage concatenative morphology