Year: 2024
Type: book-chapter
Abstract: The purpose of this chapter is to investigate the rhetoric of revolution during the Age of Revolutions and the continuities and differences in that discourse in different contexts.To begin the investigation, we acquired a large corpus of texts from HathiTrust covering all works in English published between the years 1750 and 1875. 1 This original corpus represented 626,167 pieces of writing.The date range was selected to give about a twenty-five-year buffer on either side of what might strictly be considered the Age of Revolutions so that nascent and trailing revolutionary speech would be included.After cleaning and assembling metadata on the English-language corpus, we were left with about 165,000 works that we felt would be appropriate to include in the analysis, representing a total of approximately 50,000 authors. 2 "Appropriate" in this case meant that the works were out of copyright, we had exact dates of publication, and the OCR was acceptably clean.3 Because we cast such a wide net in corpus acquisition, this still left us with an exceptionally large object of study, encompassing works of fiction and nonfiction tracts, pamphlets, government documents, newspapers, and myriad other artifacts, which could yield insights into global trends in political discourse in the English language.For context, in the field of quantitative literary analysis, many major studies use corpora consisting of several dozen, several hundred, or, very rarely, several thousand works.See, for instance, the sixty-four-work corpus used by Dennis Yi Tenen in "Toward a Computational Archaeology of Fictional Space" (which uses the same corpus as "Extracting Social Networks from Literary Fiction," the 352-work corpus created and used by Mark Algee-Hewitt and Mark McGurl in "Between Canon and Corpus: Six Perspectives on 20th-Century (COUNTER-)REVOLUTIONARY DISCOURSE IN THE AGE OF REVOLUTIONS Brad Rittenhouse, Christian Boylston, and Afshawn Lotfi159 160 ChAptEr 9Novels," or the 1,327-work corpus used in "The Emotional Arcs of Stories are Dominated by Six Basic Shapes.") 4 There are some notable exceptions, particularly from interdisciplinary research teams and interlocutors from computer science and related fields, including Smitha Milli and David Bamman, but overwhelmingly, quantitative text analysis in the humanities has not often met the potential for scale in digital studies.5 Of course, that is not to say that bigger is better in this type of study: often large corpora prove unwieldy and invalidating errors can abound.In this case, though, the inclusion of so many works should provide insight into trends, conflicts, and continuities that may have been overlooked by human researchers, or allow us to discover discourse that may have been excluded or censored from the historical record, and thus was not included in a more curated corpus.Studies of large corpora like this, however, will never be comprehensive.A close reading of 165,000 documents could not be done in 3,000 pages, let alone the 30 allotted here.What we have done is use quantitative measures to find pertinent works in the corpus, explore those works more closely, and connect those works to larger trends that we have also measured quantitatively.We develop a larger argument around what we think is a surprising feature of the corpus, but there are many other equally valid stories that could be told about these works.
Source: Cornell University Press eBooks
Language: English
Subfield: History
Field: Arts and Humanities
Domain: Social Sciences
Open Access status: hybrid
OpenAlex ID: W4401653663