Proceedings of the Annual Meeting of the Cognitive Science Society

Language evolves through cultural transmission, with certain linguistic patterns being passed on over generations while others disappear. Various factors have been proposed to explain which patterns persist, but one factor receiving little empirical attention is aesthetic appeal, i.e. how beautiful language users perceive particular forms to be. We hypothesized that appealing words would stabilize in the lexicon, whereas unappealing words would disappear. We designed an iterated learning experiment, in which participants learned novel names for novel shapes. Each shape was paired with two labels, one appealing and one unappealing. After learning all labels, participants labeled each shape twice. Their responses were passed on to the next participant in the chain, who learned only the labels produced by their predecessor. We found that, indeed, participants preferentially used the appealing labels, and that over successive generations, appealing names dominated the language. This suggests that aesthetic preferences constitute cognitive biases in language evolution.