Identifying User Demographic Traits Through Virtual-World Language Use

The paper presents approaches for identifying real-world demographic attributes based on language use in the virtual world. We apply features developed from the classic literature on sociolinguistics and sound symbolism to data collected from virtual-world chat and avatar naming to determine participants’ age and gender. We also examine participants’ use of avatar names across virtual worlds and how these names are employed to project a consistent identity across environments, which we call “traveling characteristics.”