The results of this modeling offered particularly interesting insights when contextualized against Benedict Anderson’s Imagined Communities. The Topic list of, “told young time manner replied till room love words home behaviour give found company received ill account gentleman lover power,” might corroborate or even expand upon the Anderson’s claims that the early novel provided “the technical means” for adopting of “‘homogenous empty time’ in which simultaneity is, as it were, transverse, cross-time, marked not by prefiguring and fulfillment, but by temporal coincidence, and measured by clock and calendar” (423). While Anderson considers Dynastic sexual politics in his analysis, our topic modeling might suggest there would also be merit in examining how the etiquette of more ordinary might have contributed to these conceptions of time. The grouping of “Time, love and lover” suggests that the marriage plot might be linked to rise of empty time. In addition to offering a possible link topic modeling offers us a list of texts from which we might examine this connection in greater detail.
Despite its benefits, using topic modeling is not necessarily intuitive and I find Ross’s lament initially relatable. The parameters are delicate and require a decent amount of tinkering to get at helpful information. At the scale of 50 topics, studying a wide swarth of texts by their most common word clusters can be dizzying, and some often include misspellings (although perhaps these are a product of the digitization of texts and actually present in the bodies of work.) Conversely at the scale 10 topics, they begin to reflect the formal conventions of the time with words lists like, “dear letter lady…” reflecting the epistolary novel and “king people england english war prince power army time earl ... “ reflecting the history. Consequently this tool presents a bit of paradox in that it is most rigorous when applied to a large data set, but perhaps most helpful when exploring a specific set of questions.