Abstract
The field of emergent language represents a novel area of research within thedomain of artificial intelligence, particularly within the context ofmulti-agent reinforcement learning. Although the concept of studying languageemergence is not new, early approaches were primarily concerned with explaininghuman language formation, with little consideration given to its potentialutility for artificial agents. In contrast, studies based on reinforcementlearning aim to develop communicative capabilities in agents that arecomparable to or even superior to human language. Thus, they extend beyond thelearned statistical representations that are common in natural languageprocessing research. This gives rise to a number of fundamental questions, fromthe prerequisites for language emergence to the criteria for measuring itssuccess. This paper addresses these questions by providing a comprehensivereview of 181 scientific publications on emergent language in artificialintelligence. Its objective is to serve as a reference for researchersinterested in or proficient in the field. Consequently, the main contributionsare the definition and overview of the prevailing terminology, the analysis ofexisting evaluation methods and metrics, and the description of the identifiedresearch gaps.