In working with the topics of racial discrimination and household energy reductions it is clear that just measuring attitudes is no longer sufficient. We need a fuller social context to understand what is said about these topics, as the same words can be used for many reasons. I am working with four new methods for "sampling community discourses" on topics such as racial prejudice, to replace or supplement the measurement of attitudes. The initial analysis is whether the talk functions to get someone to do something directly, to get someone to believe the 'facts', or whether the talk functions for the establishment or maintenance of social relationships. Many problems remain, especially the representativeness of the sampling and keeping as much of the context of the talk as possible.