Toward a Demography of Literary Forms: Building on Moretti's Graphs

Toward a Demography of Literary Forms: Building on Moretti's Graphs
Riddell, Allen B., Duke University, allen.riddell@duke.edu
Why do novelistic genres end? Why do we see gothic and industrial novels, Bildungsromane and mysteries, all disappear after periods of popularity during the 18th and 19th centuries? How literary forms—novelistic genres in particular—come and cease to be has long been an area of inquiry, and the work of Franco Moretti (2005) in Graphs, Maps, Trees has given the topic new energy.

The years since the publication of “Graphs” in 2003 have seen the resources available for investigating patterns in 19th century literary production expand immeasurably. For example, scans of over 7,800 volumes of 19th century British novels are now available from the University of Illinois-Urbana-Champaign’s collection alone. Those interested in extending or challenging Moretti’s observation that novelistic genres tend to arrive in “bursts” linked to social generations—or, in general, in contributing to Moretti’s proposed “sociology of literary form”—now have access to a wealth of new data. Given the continuing interest that Moretti’s work has generated among students of the human, social, and natural sciences, this represents an important opportunity.

My contribution explores further the prospects for a “demography of literary forms,” building on Moretti’s proposal for research at the intersection of literary history and sociology.I owe the suggestion for a demographic approach to Shalizi (2006)’s review of Graphs, Maps, Tress: “There is a demography of businesses, of interest groups, even of medieval manuscripts of classical works, and so why not one of literary texts?” First, I consider opportunities to improve Moretti’s original “generational model” of cycles in literary forms, offering new methods to remedy identified evidential gaps. For example, Moretti’s periodizations of genres—e.g. Courtship Novel, 1740-1820—have been criticized as too neatly falling on certain “focal dates” such as years falling at the end of a decade (Shalizi, 2006). I present new evidence in support of this criticism and demonstrate a method for making periodizations reproducible by others. Being able to reproduce Moretti’s results will hopefully make the research program itself more open to experimentation and invite collaboration. My method uses bibliographic databases to establish the period during which the vast majority (~90%) of the novels in a given genre were published. This provides a reproducible periodization suited to an inquiry into social history.The bibliographic databases used are Garside’s British Fiction Database, 1800-1829 with 2,272 titles and Bassett’s At the Circulating Library, 1837-1901 with 7335 titles I also briefly discuss the application of classification algorithms from machine learning to identify possibly overlooked genres in Moretti’s dataset. The new method for periodization is demonstrated in detail for two of the forty-four genres in Moretti’s dataset, the silver fork and Newgate novels.The methods presented are designed to accommodate changes to the underlying dataset and can be used by others who may disagree with classifications of particular novels and even with the bibliographic records provided by Garside and Bassett.

Second, in order to offer an alternative to the generational model, I argue for and attempt to test the hypothesis that the observed clustering of genre appearances and disappearances can also be explained by positing a “carrying capacity,” an upper limit on the number of established novelistic genres able to be supported by writers, readers, and publishers in any given year.

Finally, I explore the suggestion that generational changes in “mental climate” might manifest themselves not (only) in changes in novelistic genres, as suggested by Moretti, but rather in topical changes within novelistic genres. In studying a topic model (Steyvers et al., 2007) of the 7,800 volumes in UIUC’s 19th century novels collection, I observe some evidence for topical trends cutting across multiple genres. For example, starting in the mid-19th century there is a proportional rise in a cluster of words suggestive of farming and rural life.

Moretti’s work has been a touchstone for numerous discussions connected to the digital humanities—for example, the fate of “close reading”. Perhaps more significant in the long-run is the interest his work has gained from scholars in the social and natural sciences, with examples ranging from formal reviews like that of statistician Cosma Shalizi (2006) to more informal commentary on quantitative literary history from social scientists like Henry Farrell (2010) and Andrew Gelman (2010). Finding scholars outside of literary studies and literary history publicly engaging with research is an important development. Given the desirability of new models of research in the humanities, Moretti’s program can claim a interdisciplinary following that makes revisiting and extending his work of particular significance.

References:
Bassett, Troy J. 2010 “At the Circulating Library, ” (link)

Farrell, Henry 2010 “Hugo Awards II, ” (link)

Garside,P. D., J. E.Belanger and S. A. Ragaz 2004 “British Fiction,1800–1829: A Database of Production, Circulation & Reception, ” (link)

Gelman,Andrew 2010 “The Triumph of the Thriller, ” (link)

Moretti, Franco 2000 “Conjectures on World Literature, ” New Lew Review, 1

Moretti, Franco 2003 “Graphs, Maps, Trees, 1., ” New Lew Review, 24 (link)

Moretti, Franco 2005 Graphs, Maps, Trees: Abstract Models for a Literary History, London Verso

Shalizi, Cosma 2006 “Graphs, Trees, Materialism, Fishing, ” (link)

Steyvers, Mark Tom Griffiths T Landauer D Mcnamara S Dennis W Kintsch. 2007 “Probabilistic Topic Models, ” In Handbook of Latent Semantic Analysis, Lawrence Erlbaum Associates