X.1.1: The Tuberculosis Corpus

The tuberculosis corpus—the collection of books and other historical materials used to make the arguments in this dissertation—was developed using materials made available on HathiTrust, an online archive where universities host their digital collections for public use. It was originally developed in fall 2021, using a series of title searches with the words “tuberculosis” and “consumption” in English language publications. I chose to omit the inclusion of non-tuberculosis specific journals like The Lancet, because I did not have time to go through each journal looking for unique articles to include in the corpus.¹ The problem was that I would have to comb through every publication for unique articles that specifically address tuberculosis, because they would only cover that topic infrequently. This original selection of books was then narrowed by a number of factors. The final collection which comprises the corpus can be found on HathiTrust.

To make this collection of books, I first removed any text that used the term “consumption” in an economic sense. These books ranged from Marxist-adjacent titles, to economics reports, and had nothing to do with human health. Second, I removed any publications that emphasized animal health as opposed to human health. While there was significant overlap in the discourses regarding human tuberculosis and tuberculosis of other animals, especially considering Robert Koch’s original bacteriological studies (2.1.1; 2.1.4), the discourses were distinct enough to warrant separation. This is especially important because the United Kingdom published a series of reports on bovine tuberculosis in the early 1900’s which, because of their length and multiple volumes, would have altered the composition of the corpus in a notable way. Third, I further winnowed the selected texts to only include publications made between 1880 - 1926. This timeframe was chosen because of Robert Koch’s notable research in 1882, and the availability of the public domain in 2021 when I began the project.

These decisions were made under the assumption that the corpus would be used for more intricate text analysis work. Text analysis and its large language model influenced approaches like topic modelling, and sentiment analysis would have benefited from a well-defined collection of books, and it was my intention early in the project to have a large, if not exhaustive, corpus with which to work.

While I ran some simple counts of terms early in the dissertation and did some topic modelling to double check some of my work, the textual analysis part of this project ended up being much less significant.