It would be interesting to see a study of place names testing the question of whether or not toponyms are culturally selected. Place names might be demonstrated to be chosen in conformity with cultural expectations about ways in which names normatively describe places to which they are applied.
A similar study was done on personal names (see Hahn and Bentley's 2003 "Drift as a Mechanism for Cultural Change: An Example From Baby Names" in Proceedings of the Royal Society B), concluding that fashions in personal naming (i.e. proper first names for individuals chosen at birth) are subject to random drift, and thus are not selected based on past frequency of occurrence.
Essential to my (hypothetical) study of conformity in place naming practices would be the use of GIS, which could facilitate a 'spatialization' of the phenomenon in question (i.e. naming practices). This top-down approach could be balanced and made critical by the production of in situ ethnographic case studies of selected named places. Such a qualitative/critical GIS project would use a mixed methods approach to fill in gaps left open by either a strictly quantitative or a strictly qualitative approach.
So, this is a call for proposals (probably aimed at myself, but also at interested collaborators) to find a method or procedure for carrying out such a study. My paper Toponymic Constraints in Wemindji gives some idea of where I'm coming from, and hopefully where I'm going, with this idea.