How Network Structure Shapes Languages: Disentangling the Factors Driving Variation in Communicative Agents

Area of Science:

• Linguistics
• Computational Social Science
• Complexity Science

Background:

• Language variability exists among speakers, but the role of communicative network structure is not well understood.
• Previous research focused on language change, leaving gaps in understanding network effects on language variability.

Purpose of the Study:

• To investigate how communicative network structure influences language variability.
• To differentiate the effects of network structure on language change versus language emergence.

Main Methods:

• Developed a Bayesian agent-based model for language evolution.
• Conducted thousands of simulations to isolate the effects of global network metrics.
• Contrasted scenarios of language change and language emergence.

Main Results:

• Global network characteristics critically shape interindividual language variation.
• Intraindividual language variation is minimally affected by network structure.
• Path length and clustering coefficient are key drivers of interindividual variation, not size or density.
• Variation is more probable in less connected populations and smaller communities.

Conclusions:

• Network structure, particularly path length and clustering, is a primary determinant of interindividual language variation.
• Findings challenge traditional views emphasizing network size and density.
• The study offers theoretical insights into the mechanisms driving language variation.