Searching for Structure: Investigating Emergent Communication with Large Language Models
This work extends experimental findings on language evolution by using LLMs as simulation tools, offering insights for future human-machine experiments, though it is incremental in applying existing methods to new models.
The paper investigates whether artificial languages optimized for Large Language Models (LLMs) develop structure similar to human languages through simulated referential games, finding that initially unstructured languages gain structural properties enabling successful communication between LLM agents, with generational transmission improving learnability but sometimes leading to non-humanlike vocabularies.
Human languages have evolved to be structured through repeated language learning and use. These processes introduce biases that operate during language acquisition and shape linguistic systems toward communicative efficiency. In this paper, we investigate whether the same happens if artificial languages are optimised for implicit biases of Large Language Models (LLMs). To this end, we simulate a classical referential game in which LLMs learn and use artificial languages. Our results show that initially unstructured holistic languages are indeed shaped to have some structural properties that allow two LLM agents to communicate successfully. Similar to observations in human experiments, generational transmission increases the learnability of languages, but can at the same time result in non-humanlike degenerate vocabularies. Taken together, this work extends experimental findings, shows that LLMs can be used as tools in simulations of language evolution, and opens possibilities for future human-machine experiments in this field.