What if computational methods could be made to be as thought-provoking as ethnographic defamiliarisation?

read here

In this paper, we present a foray into the computational study of anthropological texts. Drawing on a corpus of approximately 2,500 articles published in the Journal of the Royal Anthropological Institute (formerly Man) from 1950 to 2018, we discuss selected findings from the deployment of two methods for computational text analysis, namely word-frequency approaches, and topic modelling. The aim of the paper is exploratory: it asks how computational methods can contribute to how anthropologists usually think about the history of their discipline. Computational approaches have several limitations, but we argue they can also help to defamiliarize a corpus in ways which are not unlike the defamiliarizing effects of the ethnographic encounter. In conclusion, we reflect on the possibilities – and pitfalls – of this and other ongoing attempts to use computational methods to map trends within anthropology and cognate disciplines.