Matthew Wilkens’s “The Geographic Imagination of Civil War-Era American Fiction” seeks to draw out the significance of space in a literary period that would seemingly be about nothing else. The years mapped out by the corpus used for the project (1851-1875) cover a time in America’s history that is defined by exploration, geographic upheaval, and war. Wilkens looks at the mentions of place names as a meter for the significance of those locations relative to their populations at the time of their mentioning. He does this to gain a more comprehensive perspective on the nature of regionalism as it becomes a popular literary form. The analysis does much to validate (initially) common assumptions about the significant places of the time period. New York and the US are the two most mentioned cities and countries respectively and Paris and London command a great deal of influence as cities of literary import.
In my view, the most important contribution “Geographic Imagination” makes is to complicate the regionalist hypothesis by revealing the underlying cosmopolitanism of the novels in Wilkens’s corpus. He writes:
[T]he world is not undifferentiatedly other in this set of formative American texts, which make consistent use of a huge array of foreign locales; references to London and Paris alone were plainly not sufficient to explore the larger world, nor to assess the many potential American relations to it. This is to say that even in those cases in which foreign locations were used for what we might call domestic ends, the range of such locations is sufficiently large as to suggest meaningful variety in both the nature of those ends and the geographic–imaginary frameworks through which they are achieved (821).
Here the digitally-furnished spatial data that Wilkens puts to use provides a perspective on the literature of this purportedly-regional period that reveals the connections America shared with the other countries in those texts. The project’s value here lies in the ability to work beyond what close reading and established literary history reveal about pre and post-Civil War texts.
Having briefly accounted for what Wilkens does in the essay, I want to look a bit closer at the particular method he applies to draw his conclusions and speculate on the ways in which they might be applied outside of this context. I will start with the discussion of weighted appearances that takes place in the “Raw Data” section of the essay. This is where the labor of getting past those initial, predictable models I mention in the first paragraph of my analysis happens. To be more specific, Wilkens first conjures maps of the US and the globe that mark off the focal points on the east and west coast of the American continent and Western Europe. He then remarks that these graphs are deceptive because they do not account for population growth relative to the rate at which mentions occur. This is why the middle of the country remains relatively blank while New England and California are shaded solidly in. The application of the Dunning-log method to the data allows for the raw number of mentions to be sorted into a more nuanced and comprehensive representation. The method separates the mentions by time period and measures the statistical likelihood that they are over or under-represented. As seen in Figure 3, the states that initially slip beneath the radar are those that experience a great deal of population growth (and therefore become more spatially significant within the context of literary mention) due to initially low populations. This means that those states in the Midwest and on the frontier that see new cities and settlements founded during the time period covered in Wilkens’s corpus are under-represented because they would need to greatly over-represented just to keep up with the rate of population growth (815).
What I would be interested in seeing is the Dunning-log method applied to a similar spatialization of global literary trends over a larger time frame. As is often the case with digital methods, spatialization offers a great opportunity to refute the centralized nature of literary history. If one could apply this method to the contributions of every country in (and this is really just an example) the Atlantic community without starting from the same European and American perspectives that dominate the ways in which the great texts are enumerated and labelled, a great deal might be revealed about the significance of literature that appears as a minor player but is actually proportionally quite significant. This is also not limited to spatial methods or even to the task of uncovering the significance of previously-marginal texts. One could use the Dunning-log method within a corpus of texts to operationalize the significance of characters and events that run parallel to the central action or plot. It is difficult to quantify exactly why and how a subplot or supporting character affects a work. By weighting mentions and scenes against the development of plot significance, one might be able to get past the centralizing nature of traditional literary structures (those adopted to undertake the simple and direct analyses Wilkens refers to in his essay) and look at bit players and locales with a more democratized lens.