Polysemy and metaphorical extension in an artificial language learning experiment

Abstract

This study investigates the emergence of metaphorical meanings in abstract visual symbols within an ‘alien language’ artificial language learning game. Findings reveal that participants extend concrete meanings to abstract meanings based on metaphorical mappings proposed by Conceptual Metaphor Theory, but not consistently across all cases. When multiple salient mappings are possible, participants’ choices exhibit greater variability. Certain mappings, such as SAD IS DOWN and ANGER IS FIRE, are particularly salient and chosen frequently, whereas others, like POWERFUL IS UP, show less consistency. Additionally, some unexpected mappings were observed. While some could be explained via post hoc analysis, others remain unexplained. Experimental manipulations to enhance the salience of metaphorical mappings did not significantly alter participant responses. Overall, results suggest that participants spontaneously engage in semantic and metaphorical meaning extensions, supporting the idea that these cognitive mechanisms underpin polysemy and the conventionalisation of new word senses. Participants make choices for semantic extensions in a motivated manner, with some extensions being based on conceptual metaphorical mappings, and others on other types of salient associative mappings. This study contributes to theories of language evolution that highlight the central role of metaphor in shaping meaning extensions and the cultural emergence of linguistic structure.