NeLLCom-X: A Comprehensive Neural-Agent Framework to Simulate Language Learning and Group Communication

TL;DR

This work validate NeLLCom-X by replicating key findings from prior research simulating the emergence of a word-order/case-marking trade-off and investigates how interaction affects linguistic convergence and emergence of the trade-off.

Abstract

Recent advances in computational linguistics include simulating the emergence of human-like languages with interacting neural network agents, starting from sets of random symbols. The recently introduced NeLLCom framework (Lian et al., 2023) allows agents to first learn an artificial language and then use it to communicate, with the aim of studying the emergence of specific linguistics properties. We extend this framework (NeLLCom-X) by introducing more realistic role-alternating agents and group communication in order to investigate the interplay between language learnability, communication pressures, and group size effects. We validate NeLLCom-X by replicating key findings from prior research simulating the emergence of a word-order/case-marking trade-off. Next, we investigate how interaction affects linguistic convergence and emergence of the trade-off. The novel framework facilitates future simulations of diverse linguistic aspects, emphasizing the importance of interaction and group dynamics in language evolution.

Figures (8)

• Figure 1: Overview of the NeLLCom-X framework.
• Figure 2: Two populations of 50 agents engaging in self-play (no interaction) after having learned two flexible-order, optional-marker languages: one with 67% the other with 50% marking. Left column: Average communication success across self-play turns. Right column: Production preferences: solid diamonds mark the initial language; each empty circle denotes a full-fledged agent at the end of self-play; solid circles are the average of all agents, with error bars showing standard deviation.
• Figure 3: Interactive communication between different language speakers. The first agent is always trained on 50s+50m ($\alpha_{b}$). Each experiment is repeated with 50 agent pairs.
• Figure 4: Impact of self-play during interaction in pairs of agents speaking 80s+20m and 20s+20m respectively. Each experiment is repeated with 20 agent pairs, and the average communication per turn is shown.
• Figure 5: Interactive communication in groups of same-language speakers (50s+50m). Right column: Group-level production preferences (each point is a group) and Spearman's correlation $\rho$ between marker use and order entropy.