Abstract

With the emergence of large digitized collections of historical texts, scholarship in the humanities has increasingly turned to studying texts as data. This article argues that seeing text as data is particularly apt for the study of conceptual history. The quantitative perspective allows for rethinking the analytical terminology used to study the transformation of political and social terminology. Further, quantitative conceptual history requires re-evaluation on four levels. First, it forces scholars of conceptual history to reconsider the role of reception in the spread and lexicalization of linguistic innovations. Second, it forces them to assess how to interpret quantitative analyses of linguistic change. Third, the use of quantitative methods calls for clarity in describing what is being measured and what is being interpreted based on quantitative analyses. Fourth, the use of machine-learning methods for conceptual history should remain careful. They can be very useful for exploration, but should be combined with count-based methods to provide concrete proof.