The Longue Durée of Literary Prestige

TL;DR: The authors studied the stylistic differences associated with literary prominence across a century and found that there is a steady tendency for new volumes of poetry to change by slightly exaggerating certain features that defined prestige in the recent past.

Abstract: A history of literary prestige needs to study both works that achieved distinction and the mass of volumes from which they were distinguished. To understand how those patterns of preference changed across a century, we gathered two samples of English-language poetry from the period 1820–1919: one drawn from volumes reviewed in prominent periodicals and one selected at random from a large digital library (in which the majority of authors are relatively obscure). The stylistic differences associated with literary prominence turn out to be quite stable: a statistical model trained to distinguish reviewed from random volumes in any quarter of this century can make predictions almost as accurate about the rest of the period. The “poetic revolutions” described by many histories are not visible in this model; instead, there is a steady tendency for new volumes of poetry to change by slightly exaggerating certain features that defined prestige in the recent past.

Summary

The Plan of the Experiment

• The authors could have checked where the books were reviewed; they didn't need computers to guess for us, and they didn't really care whether computers were good at guessing.
• So although the authors recorded reviewers' sentiments when they were clear, this article places more emphasis on the fact that an author was reviewed at all.
• The Athenaeum was influential but reviewed so many novels that it was not a sign of great distinction to be included there.
• The authors also needed a sample that contained books reviewed less often.

A Model of Reception

• The authors goal here is to assess the strength of the relationship between poetic language and reception.
• It could be largely accidental; in effect, the algorithm has only "memorized" the quirks of particular examples.
• If the authors consider publication date as a factor and use the slanted black line to divide the data set, the model is 79.2 percent accurate.
• Scholars propose different dates, but almost everyone agrees that poetic standards changed dramatically in the nineteenth century.
• If a single list of words can predict literary prestige across that distance, some aspect of reception must be more stable than the authors anticipated.

The Logic of Poetic Distinction

• Since the canonical literary tradition seems too diverse to produce this kind of stable boundary, the authors suspected at first that the source of stability must be located in their random sample.
• A list of the top ten words that individually have the largest effect on the model's predictions might not tell us very much.
• If it were actually judging poems, the model would be wrong about that line, by the way: the dissolution of imagined immediacy into painful memory at the end is beautiful in context, and it is apt that a poem called "Echo" ends with repetition.
• So these are the broad patterns that leap out immediately from a model of poetic reception 1820-1919: a preference for concrete language and a relatively dark tone (or at least not a sentimentally uplifting one) (fig. 2 ).
• One advantage of distant reading is that it can be more patient with historicism, revealing even slow changes as historical phenomena.

How Quickly Does Reception Change?

• Both articles contrast prominent and obscure works to discover a system of differences that defines literary success.
• If their models predict that boundary reliably, the authors know that they have captured something important about reception; if one model can predict the boundary reasonably well across a century, they know that some important aspects of reception changed slowly.
• Poetry itself may have changed in other ways: D. H. Lawrence writes things about the sex lives of whales that would have made Alfred Tennyson blush.
• In practice, the textual differences associated with success seem to have changed slowly.
• The authors could train a model only on volumes from one quarter century but ask it to make predictions about the rest of the century.

Synchronic Distinction and Diachronic Change

• Nevertheless there were many changes, some of them even visible in the figures above.
• In all these models, the median probability that a volume will be reviewed appears to increase across time.
• And this is not the only kind of change that can happen in literary history: many different changes are always happening, and many of them won't be captured by a model of distinction.
• The authors just suggest that, whenever scholars do define a linguistic proxy for social distinction in a given period, they will find that change relative to that axis moves in an upward direction during the period itself.
• There are ways for us to untangle this causal knot.

Gender and Nationality

• The methodology the authors are using is close to social science, and they need to be alert for the interactions between variables that preoccupy social scientists.
• If women were less likely to be reviewed, their model might confound literary prestige with masculinity.
• The model's predictions for women are just as accurate as those for men, and if the authors run the modeling process on a data set restricted to women, it works just as well.
• Concrete, troubling images still make poets more successful.
• American authors are overrepresented in the random sample, and their works are probably more obscure, on the whole, than randomly selected works by British writers.

What Became of Our Original Hypothesis?

• The original goal of this experiment was to test whether reviewed and random samples would become easier to differentiate as time passed.
• Critical tradition suggested that distinctions between popular and elite poetic culture had hardened "over the course of the nineteenth century, as the increasingly centralized media and entertainment industries interacted with the growth of education" (Gray 2001: 347) .
• The authors had planned to begin in 1840, because they didn't expect the style associated with elite taste to be clearly distinct in the period 1840-59 (perhaps a model would be only 60 percent accurate).
• Figure 1 shows that reviewed and random volumes are more evenly mixed before 1840; accuracy in that section of the timeline is only 66.7 percent.
• In other words, the early part of the timeline is not just organized by a stylistic boundary different from the one established later; there really appears to be less consensus about stylistic prestige in the first twenty years.

Conclusion

• A lot of descriptive work remains to be done in literary history, because the authors still know relatively little about patterns above the scale of a few hundred books.
• Literary historians have often generalized about the pace of change, for instance-contrasting epochs of relative stability to the "revolutions" that separate them (Fallis 1976; Greenblatt et al. 2006 Greenblatt et al. : 1834)) .
• But those claims are based on limited evidence.
• The explanation the authors need may have to cover not just poetry but nineteenthcentury literary prestige more generally.
• 16 Literary scholars haven't described the history of reception very fully yet; it shouldn't be surprising that the authors cannot yet fully explain it.

Figures

Figure 2. The same model, with outliers and interesting cases labeled

Figure 3. Amodel trained only on volumes from1845 to 1869makes predictions about the whole century

Table 1. Periodicals sampled

Figure 1. A model of literary prestige from 1820 to 1919