Foreign-language effects in cross-cultural behavioral research: Evidence from the Tanzanian Hadza

Abstract

Abstract Behavioral research in traditional subsistence populations is often conducted in a non-native language. Recent studies show that non-native language-use systematically influences behavior, including in widely used methodologies. However, such studies are largely conducted in rich, industrialized societies, using at least one European language. This study expands sample diversity. We presented four standard tasks—a “dictator” game, two sacrificial dilemmas, a wager task, and five Likert-risk tolerance measures—to 129 Hadza participants. We randomly varied study languages—Hadzane and Kiswahili—between participants. We report a moderate impact of study language on wager decisions, alongside a substantial effect on dilemma decisions and responses to Likert-assessments of risk. As expected, non-native languages fostered utilitarian choices in sacrificial dilemmas. Unlike previous studies, non-native-language-use decreased risk preference in wager and Likert-tasks. We consider alternative explanatory mechanisms to account for this reversal, including linguistic relativity and cultural context. Given the strength of the effects reported here, we recommend, where possible, that future cross-cultural research should be conducted in participants’ first language.