Gleanings from Digital History for the Reading of Sources

Many of the conclusions I drew from my research on the centenary commemoration of the War of 1812 came from examining elements that have been part of the traditional archive for political, social and cultural historians for years: annual reports, newsletters and other organizational records.  However, many of the leads I followed were drawn from the tools and logic of digital humanities (DH).  As a result of this experience, I have now begun to think differently as a social historian, particularly in how I read sources. What I want to discuss here are three examples of how the tools of the digital humanities can help scholars to formulate new interpretations of their sources even if they are reluctant to turn fully to the digital realm.

1)      Use as many sources and documents as possible in a corpus study, rather than sticking to a narrow document-based case-study approach.

This issue forms a crucial part of a debate within literary criticism on the merits of close and distant reading. Close reading is reading one text in its own context and drawing out details or metaphors.  This reading is based on analog or paper ways of production and comprehension and in this process often focuses on elements like syntax.   Any conclusions drawn from analysis may be extended then to the author’s work as a whole.

Matthew G. Kirschenbaum, a proponent of distant reading and a practitioner of the digital humanities, advocates approaching authors’ work in the full context of the corpus of their work, using programming like his own MONK program  to connect publication dates and other small details that are often deemed unimportant in the close reading of single works.  He is also a proponent of data mining analysis, discussed later on in this post.  An article on this that outlines the MONK program is located through this Zotero reference here. Kirschenbaum draws  on the earlier work of the  Stanford University literary scholar Franco Morelli, who coined the term “distant reading” and was an advocate for  “using statistical, quantitative methods to ‘read’ large volumes of text at a distance, using ‘graphs, maps, and trees’ as forms of abstract representation that enable the study of patterns over time.” [1] In works like “Graphs, Maps and Trees,” Morelli  has sought to look beyond established scholarship and ask wider questions, addressing, for example, the rise of the novel over the nineteenth and twentieth century.  One of his articles can be read here.

Digital tools such as Voyant and Mallet allow scholars to examine multiple sources on the same subject relatively quickly, and to investigate runs of writing in greater detail than can be done using more established techniques. Such DH tools can help one discover trends from multiple sources and conclusions that balance out some of the biases that occur when you use fewer sources.  Scholars can learn from this by moving away from a close reading of just a few documents, enabling them to draw wider conclusions about an author’s life or perspective or a set time period by including a wider corpus of works.  This level of analysis should not replace close reading of these documents, but can supplement that work.

2)      Breaking up the text into smaller parts does not invalidate analysis.

History and literary studies tend to view texts as connected and focus on placing them in a larger context, which does not necessarily give us a complete picture.  Breaking up the text into pairings or individual words removes the context and allows the scholar to step back and re-examine some assumptions. Although DH work in text mining through Voyant might allow for some problematic conclusions to be drawn by over-emphasizing the importance of connectors or personal nouns, it can allow scholars to examine the use of some words to see if they might be more important than previously realized.   The use of simple works like “Empire” can challenge assumptions, as I have discovered by text mining of two years of the annual reports of the Ontario Historical Society.  Over time ‘Empire” decreased being used in the context of “United Empire Loyalists” in favor of “British Empire,” suggesting a possible rise and decline of influences from 1889 to 1902.  The frequency of the word and its immediate context can suggest the increasing or decreasing influence of certain ideas, a conclusion that may not be apparent without text mining.  In her work A Midwife’s Tale, Laurel Thatcher Ulrich foreshadowed this approach by manually examining word use in Martha Ballard’s diary, a project that the DH has undertaken with similar results through topic modeling and Mallet.

3)      Benign detail or lists can be compelling sources for scholarship.

Scholars in the humanities could benefit by breaking down the methodological boundaries that separate humanities from social sciences.  By looking at the executive lists of the organizations in my own work, I found patterns that both confirmed previous scholarship and also suggested new interpretations and angles of research. By using digital tools, I was able to re-examine sources by looking for simple patterns in names.  Ulrich’s book A Midwife’s Tale: The Life of Martha Ballard, Based on Her Diary, 1785-1812 and the PBS film  both show how Martha Ballard’s diary was ignored by scholars due both to the obscurity of its author and the apparently benign details of the weather and other tasks she provided.  Ulrich was able to take these details and connect them to provide clues that helped to enrich her interpretation of social history and to uncover new narratives.  Cameron Blevins has attempted the same task as Ulrich by applying digital techniques and shows how the digital analytical methods can find a quite similar answer to what Ulrich found by hand.  This suggests how methods promoted by the digital humanities can actually be based on work done without digital tools, indicating how both DH and non-DH approaches can enrich each other.

These three examples demonstrate how those who are intrigued by the possibilities offered by digital humanities but are still reluctant to use digital tools can incorporate some of the methods into their own approaches.

[1] Mathew Kirschenbaum “The Remaking of Reading: Data Mining and the Digital Humanities”, 2

Creative Commons Licence
Reading the Archive by Peter W. Holdsworth is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported License.