Gender bias through production about and memory for names

Abstract

Two online experiments investigate how alternative forms for reference to individuals—first, last, and full names—affect inferences about the gender of the referent. Given distributional correspondences between first names and gender, first and full names provide probabilistic information about an individual's gender, whereas last names do not. We asked what inferences people make about a referent’s gender based on whether they are referenced using their first, last, or full fame, as well as distributional characteristics of the first name. Results of a sentence continuation task (Experiment 1) revealed that when a character is introduced by last name alone, participants overwhelmingly chose he/him over she/her pronouns to subsequently reference the character. This bias was also apparent when the character was introduced using a first or full name, as participants were less likely to use she/her pronouns than the gender distribution of the first names would predict. When explicitly queried about the referent’s gender (Experiment 2), characters introduced by last name alone were again preferentially inferred as male, and this bias persisted even when first and full names were used. We discuss implications for models of how world knowledge is linked to language use.