Proceedings of the Annual Meeting of the Cognitive Science Society

We address competing perspectives on how social-indexicalmeaning is learned in language, using data from artificial lan-guage learning experiments and two studies in small-scalesocieties. Our results indicate that learning social-indexicalmeaning is primarily allocentric as opposed to egocentric:speaker success in learning a social-indexical meaning patterndepends on overall exposure to the pattern more than the pat-tern’s relative importance to the speaker. We base these claimson data from American English-speaking adults, Datooga-speaking children, as well as adults and children speakingMurrinhpatha. The results highlight the importance of widen-ing the sample of methods and data sources in studying howvariation in language is learned and maintained.