Is throwing garbage into a bin an instance of preserving nature? Previous research showed that a connection between a universal human value (such as nature preservation) and a concrete action (such as throwing garbage into a bin) that is seen as typical in one culture can be seen as much less typical in a different culture. Therefore, if two people both care for the same human value and one of them thinks an action should be done to attain a value, the other might disagree, as they do not see the action as an instantiation of the value and thus do not think the value can be attained by performing the action. To compare value-driven behavior across cultures, one thus has to ask: What are the typical instantiations for a value in a culture?
Using current AI technology, we make a step towards answering this question from large-scale behavioral data. Using the ValuesML dataset—which contains more than 2500 news articles and political manifestos in nine languages, coded for human value mentions by trained scholars from respective countries—we apply AI approaches from text clustering and language modeling to identify typical instantiations per country/culture that the data originates from and then contrast these typical instantiations between the data subsets. We make our approach available online to the research community in order to enable analyses using different and larger data in the future.