The history of sociolinguistic research leads us to expect certain patterns of variation by social class, age and gender, but as the field expands to new societies with different social structures one often encounters unexpected results. Such findings can have great value in leading to higher level generalizations. Cases are cited of unexpected gender variation in studies of rural communities in Spain, Egypt, central India and south China where ethnographic observation leads to a better understanding of what seemed at first to be anomalous results.