Paper short abstract:

Researchers of ethnology and folklore are always challenged not to “cherry-pick” the specific examples they closely analyze. The huge volumes of vernacular expression available online makes the problem worse. A computational approach to this problem offers a surprising view of online gun-lovers.

Paper long abstract:

One challenge that has long been faced by researchers of ethnology and folklore is how to fairly pick the specific examples that we locate, contextualize, and closely analyze. Do we attempt to represent a whole community based only a few exemplary cases? Do we focus on the most active bearers of the traditions we are interested in? Or do we seek to find "average" actors to represent what some idealized "average" community member experiences?

In face-to-face ethnography, these decisions are often made for us by factors on the ground that we cannot control: whom we have time to talk to or who will talk to us for example. In many online contexts, however, incomprehensibly huge volumes of vernacular expression are often publicly available for us to examine. In those cases, how do we avoid "cherry-picking" the examples that are the most interesting to us or those that potentially best support our presuppositions?

In the case of research on online gun discourse, its important to consider specific communication events that seem to foster a disturbing paranoia and even potentially some sort of violence. Those communications, however, emerged associated with many others. Looking just at individual posts expressing fears of conspiracy or rumors of wars, we would miss the fact that we are dealing with a folk culture that at least allows and maybe even encourages reasonable discourse.