Digital Humanities and Its Application in The Study of Literature and Culture

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

University of Oregon

Digital Humanities and Its Application in the Study of Literature and Culture
Author(s): MATTHEW WILKENS
Source: Comparative Literature, Vol. 67, No. 1 (March 2015), pp. 11-20
Published by: Duke University Press on behalf of the University of Oregon
Stable URL: https://www.jstor.org/stable/24694545
Accessed: 01-01-2024 10:18 +00:00

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

University of Oregon, Duke University Press are collaborating with JSTOR to digitize,
preserve and extend access to Comparative Literature

This content downloaded from 210.31.15.200 on Mon, 01 Jan 2024 10:18:19 +00:00
All use subject to https://about.jstor.org/terms
MATTHEW WILKENS

Digital Humanities and


Its Application in the St
of Literature and Culture

COMPUTATIONAL
number of people METHODS are
who work in notthe
even taking over the
expansively humanities.
defined Th
digital human
ities is modest, and only a small fraction of those conduct primarily data-drive
research. Fewer still are comparatists of any stripe. But it's not hard to see wh
there is intense interest in the benefits and alleged dangers of quantification
within literary disciplines that have long understood their projects in almost
exclusively qualitative terms, yet often confront bafflingly great masses of source
material. What digital humanities in general—and computationally assisted lit
erary studies in particular—offer is a new set of methods for dealing with suc
abundance. These methods produce new types of evidence that can be used in
combination with existing approaches in order to pursue humanistic work in
richer, more inclusive ways. Computational work has already begun to deliver o
this promise, and the prospects for future advances are especially bright despit
both social and technological obstacles.
What computational methods offer most directly is help identifying and asses
ing literary patterns at scales from the individual text to whole fields and systems
of cultural production. That's not the only thing that computers are good for, o
course, but it's a useful place to start because it makes clear the basic continuity
between older and newer approaches. Literary scholars often underestimate,
think, the extent to which their claims are implicitly quantitative, pattern-based,
and dependent on reductive models of the texts they treat. If you argue, for
instance (and as I have), that certain late modernist novels use allegory and ency
clopedism to respond to the untenability of high modernist representational tec
niques in a postwar cultural environment, you need to be able to show that thes
features are plausibly present in the texts, that they differ in meaningful way
from other (absent or deemphasized) aspects that you might have expected to
find, and—-if you want some large-scale stakes for your reading—that the features
you have found are in sortie sense typical or symptomatic of important change
taking place at the time. In doing so, you'll ignore nearly all of the features of your
object texts in favor of a sharply restricted subset that captures those aspects of th
problem you want to address. You likely won't frame any of these pieces of you

Comparative Literature 67:1


DOI 10.1215/00104124-2861911 © 2015 by University of Oregon

This content downloaded from 210.31.15.200 on Mon, 01 Jan 2024 10:18:19 +00:00
All use subject to https://about.jstor.org/terms
COMPARATIVE LITERATURE / 12

argument in numerical terms, but they depend on a defensible sense of the distri
bution of specific features within and across a range of relevant texts. You will, in
short, have built an abstractly quantifiable model of your problem domain and o
your texts' place within it. My thesis is that there are many cases in which it's bet
ter, both conceptually and evidentially, for this process to be explicit rather than
(solely) implicit and that the need for quantitative approaches to literature is thus
great indeed. In fact, there now exist diverse achieved results in literary and cu
tural studies that make this point convincingly, primarily by producing new kinds
of evidence and bringing it to bear on socio-literary analysis.1
It also seems clear that many of the large-scale questions in literary and cu
tural studies that are best suited to computational analysis—including those fa
ing under the headings of world literature and longue duree literary history—are
similar to those raised by systems theory. That there has not been greater cros
pollination between the two fields is attributable perhaps both to the relative lack
of attention to systems theory as such within the Anglophone North America
institutions where digital humanities has recently found greatest purchase and to
the modest number of digital humanities practitioners in departments of com
parative literature (which have been most receptive to the work of Immanue
Wallerstein, Niklas Luhmann, Siegfried J. Schmidt, and the like). But the socio
logical and systemic orientation of much of the best computational work, inclu
ing the focus on networks of textual production, reception, and circulation to b
explored below, suggests that a more explicit engagement with systems theory will
be an almost inevitable consequence of the rise of digital humanities.2

Text Mining and Deformance


An example of recent work that employs sophisticated computational method
in support of readily recognizable literary ends is Andrew Piper and Mark Alge
Hewitt's research on the linguistic signature of Goethe's The Sorrows of Young
Werther ncross the whole of Goethe's writings. Piper and Algee-Hewitt call the
method "topological reading," by which they mean that it is attuned to the pa
terns by which groups of words co-occur in different texts.3 Their object, of course,
is not simply to quantify these patterns, but to use them to trace the afterlife of
Werther in Goethe's later work.
The technical details of their procedure are straightforward despite the appar
ent complexity of its output (see figure 1) and the nuance of their argument. They

1 In addition to the scholars cited elsewhere in the article, interested readers might see, for exam
ple, Elson, Dames, and McKeown; Michel and Liberman Aiden; Tangherlini and Leonard. Moretti
work remains, rightly, the starting point for many.
2 For orientation, obvious touchstones on the literary side of systems-style thinking include W
Chee Dimock, Paul Giles, David Damrosch, Pascale Casanova, and, of course, Franco Moretti.

3 In their own words, "topology attends to the recurrence of words, the way language repeats itself
at a distance [...]. Instead of understanding language in significatory terms (what it says), topology
allows us to think more about language in agential terms (what it does). It shows us how the patterns
of lexical repetition within texts produce meanings that are not localized in or inherent to those pat
terns. Meaning is not a function of signification in a topology but organization" (157). Full method
ological details are available in Piper and Algee-Hewitt. For a more theoretically oriented treatment
of topology itself, see Piper.

This content downloaded from 210.31.15.200 on Mon, 01 Jan 2024 10:18:19 +00:00
All use subject to https://about.jstor.org/terms
ACLA FORUM / 13

Figure 1. Goethe's collected works arranged according to words common in Werther.


(From Piper and Algee-Hewitt 158; reproduced with permission.)

count the occurrences of each word in Werther, remove high-frequency function


words like der, die, and und that carry little semantic content of their own, and then
compare the frequencies of the most common remaining words (ninety-one of
them, to be exact) in each of Goethe's other published works. To provide a single
measure of similarity between the topology of Werther and that of any other text
they compute the Euclidean distance between them in the resulting ninety-one
dimensional space. Happily, that's not as hard as it sounds; it's just a matter of
comparing the frequency of each word in two texts and adding up their differ
ences.4 Texts that use words common in Werther at similar rates will be close to one

another by this measure, while texts with different word frequency patterns will be
further apart. Finally, Piper and Algee-Hewitt collapse all the pairwise distanc
measures into a two-dimensional Voronoi diagram like the one in figure 1. (Thin
of this by analogy to flattening a globe into a wall map, only with more latitud

4 OK, it's not quite that simple; you're squaring the differences, summing them, and taking th
square root. What you're really doing is computing the Pythagorean equation in ninety-one dim
sions rather than the two you probably learned in elementary school. So where the dista
between points A and B in two-dimensional space = V(x2 + y2), with x the distance between A and
in one dimension and y the distance between them in the second dimension, Piper and Al
Hewitt perform the same calculation on xv x2, . . . , xm with each the difference between any tw
texts in frequency of one of the ninety-one most common words in Werther (not counting t
removed function words).

This content downloaded from 210.31.15.200 on Mon, 01 Jan 2024 10:18:19 +00:00
All use subject to https://about.jstor.org/terms
COMPARATIVE LITERATURE / 14

about where things end up.) Each tile in figure 1 corresponds to a single text of
Goethe's. Tiles that are closer together are more similar in the rates at which they
use the ninety-one Wertherwords, though those words are not necessarily uniquely
"Wertherian" in the sense of occurring often in Werther and infrequently els
where (they occur often in Werther, but the method does not differentiate between
words that are common in that text alone and those that are common everywhere,
apart from removing stopwords from all the documents). Large tiles are less sim
lar to the "average" text in Goethe's corpus than are small tiles (again, by the fr
quency-distribution distance metric in question).
What Piper and Algee-Hewitt do with their data is not what you might expect
They're modestly interested in Werther s centrality (or lack thereof; it's the white tile
near the left edge of figure 1) and the long-posited break between Goethe's earl
style and his later writings (not much of a break, when assessed in this way), both of
which are cast in a slightly different light by their computational evaluation. But
most of their analysis springs from what Stephen Ramsay, following Lisa Samuels
and Jerome McGann, has called the "deformative" nature of computational cri
cism (Ramsay 32) and what the authors name the "variation engine" (169) of top
logical reading. By this I mean that they spend much of their time reading new
assemblages of page-size chunks within two top-level clusters produced by their
algorithm. To do that, they break the texts clustered near Werther into pieces (each
about 200 words), then subject those pieces to the same procedure originally
applied to the full corpus, producing subclusters of similar and dissimilar pages
Then they perform the same calculation on a cluster of texts further away from th
Werther-centered group. (Both of these clusters are outlined in black on figure 1.)
They do this not to locate some calculated essence of Wertherness, but as an oppor
tunity (one might have said excuse with no rebuke implied) to encounter a diffe
ent type of source material and to subject it to a much more conventional-looking
close reading. So, for instance, they spend the length of a short article within their
longer essay offering a reading of the hand as a unifying figure tied to artistic pro
duction across the "pages" of a new text built from those pieces that cluster most
tightly within the Werther group. They do the same for another such text bui
within the top-level cluster further removed from Werther; arguing that it repre
sents a move from specifically artistic knowledge to a more general consideration
of epistemology as grounded in observation.
Piper and Algee-Hewitt's most striking claim is that there's a strong underlying
continuity between Goethe's early and late work as evinced by the coherence an
interpretability of the clusters they produce by way of words derived from Werther
itself. It's possible that this slightly overstates the linking role of Werther, since the
high-frequency words in that novel are not necessarily uniquely characteristic of it
(a possibility acknowledged at the end of their essay as an area of future work). But
their method reflects an entirely defensible interpretive choice to emphasize th
role of difference within a framework of likeness, a fact that also serves as a
reminder of the inseparability of interpretation from evidence gathering in quan
titative work. In any case, no matter the outcome of this specific question, Piper
and Algee-Hewitt's analysis demonstrates one of the major benefits of computa
tionally assisted literary criticism. Their method allows them to reconfigure
Goethe's corpus around a new set of premises and provides them the opportunity

This content downloaded from 210.31.15.200 on Mon, 01 Jan 2024 10:18:19 +00:00
All use subject to https://about.jstor.org/terms
ACLA FORUM / 15

to read that corpus as a newly estranged object. For critics working on major
authors, texts, and movements—objects about which our critical narratives are in
general deeply entrenched—such a defamiliarizing prompt will often be a rare
and valuable event.

Network Analysis and Literary Sociology

At the opposite extreme of dependence on close textual analysis are projects


aimed primarily at the sociology of literary production and consumption. Much of
Franco Moretti's recent work falls under this heading, but its imperatives and
potential are especially clear in Richard Jean So and Hoyt Long's research on
modernist literary networks in the United States, Japan, and China. So and Long
are interested in the structure of literary networks both as a means by which t
understand the development of global modernism in the early twentieth century
and as a way of thinking differently about the relationship between influence and
poetic form. The ground they cover is, at that level and as their own literatur
review makes clear, not new, but their approach is based on a deep reconceptua
ization of the kinds of evidence that bear on it.

So and Long work with a corpus not of literary texts but of metadata abou
such texts in the form of publication records for hundreds of poetic and literary
journals in the years 1915 to 1930 (for the U.S.), 1920-1944 (Japan), and 1911-19
(China). They subject this data to network analysis, counting the number of
poems each author published in anyjournal by year and linking authors to jou
nals according to the number of poems published in that journal.' They visuali
their data in diagrams of the increasingly familiar sort shown in figure 2.
Their findings are significant in two ways. In their comparative descriptions of
magazine-based modernist poetics, they observe important differences amon
their three national contexts. In the U.S. the network is dominated by the journal
Poetry, to which many of the period's leading writers were linked. But there a
also numerous secondary and tertiary clusters, each organizing its own constel
tion of poets; clusters are linked, in general, by a small number of writers wh
published more eclectically. In Japan, the network is essentially divided into tw
large camps bridged almost solely by the journal Shishin, suggesting both a mo
polarized poetic climate and a central role for Shishin in the development ofja
nese modernism. In China—where the data are more sparse — it appears tha
journals developed much more exclusive coteries as evinced by the tiny number of
poets who published work in multiple outlets. This suggests a literary landscape of
extreme balkanization, one with far fewer opportunities to develop an integrat
or standardized version of modernist forms.

Beyond this new evidence of differently evolving social structures, So and Long
argue for the critical relevance of what they call, borrowing from the sociologi
cal literature, "brokers" in the poetic field. These are the poets, often drawn from
the rungs now thought to be just below the period's most influential or successful

' To use the technical terms, authors and journals are nodes linked by undirected edges repre
senting poems by a poet published in a journal. Edges are weighted according to the number of
poems published (generally broken down by year).

This content downloaded from 210.31.15.200 on Mon, 01 Jan 2024 10:18:19 +00:00
All use subject to https://about.jstor.org/terms
COMPARATIVE LITERATURE / 16

«i Dial Group
■ Poetry Circle
JohnCroAR^«P
Louis<jfc»i>*rg
■ Greenwich Village
©
—©™s -© 13 Leftist

o
o
ED Fugitives
0»n»vi»ftTj5^r ■ Harlem Renaissance

^ VaoM®nd>ay
$j) JohnGs^t^l*":h*r

^ f*U©r»n4 PWShUroo®.ndar»on4i Wii)iamCj#sWim^^ W


<& V -—Ci:Voi^dburg
7* Vach-^&nasay

W®am@*8»nst Edn*5!.'v®9ntMfl.jy , <id JohnGo^^ktch®'

J _ f Poetry \
J© ©
HLkWfencf^ Sh«nyoo^nd«rsonQ
MwiflLoy
— v Carios^^tdburg
iaSl.'v^P«ntMilljy
ezra©bund

c
R <*\*-<fi3difv3!on Wa>ia.^*v«ns
pK»y
-J®
r.3ffii>o(

0
RichardSfeington

jMji^RsdBtonPauMt

American Poets, 1917-1918


iamMWh««of^hn*on
Marion'V®Du!hb»it

Figure 2. Network structure of American poetry, 1917-18.


(From So and Long 150; reproduced with permission.)

writers, who served to connect otherwise disparate coteries. Brokers such


Lowell and Countee Cullen, they note, appear to have been those wr
were at least partially excluded from any single inner circle and thus d
influence and prestige accruing to a coterie's leaders, who tended t
exclusively in the flagship-like journal of their group.6 Attention to th
brokerage suggests new relevance for the study of poets such as Lowell
len in the U.S. or Kond5 Azuma, Takenaka Iku, Kitagawa Fuyuhiko,
Kamenosuke in Japan. More generally, it implies the continuing need f
approaches attuned to the function of social-structural positions as such
this move signals a renewed openness to broadly structuralist investiga
and Long's work brings a comparatist and historicist awareness to the pr
distinguishes it from classical forms of social and linguistic structuralism

Clustering and Mapping

Somewhere between textual deformance and sociological structure lie


tational approaches that seek to characterize fields of literary produ
information extracted directly from very large bodies of text. Som
approaches are akin to the one used by Piper and Algee-Hewitt insofar as

6 So and Long note that the literary brokerage they discuss obviously differs from soc
age, the latter being more frequently dominated by high-prestige writers such as Pound

This content downloaded from 210.31.15.200 on Mon, 01 Jan 2024 10:18:19 +00:00
All use subject to https://about.jstor.org/terms
ACLA FORUM / 17

Figure 3. Global locations in American fiction, 1851-75.


(Reproduced from Wilkens "Geographic" 809.)

on identifying patterns of word co-occurrence. Under this heading fall techniques


such as topic modeling and other text classification algorithms that have been
used to good effect by a range of scholars working in mostly single-language disci
plines. For example, in Macroanalysis Matthew Jockers has constructed an elabo
rate map of something that resembles literary influence in nineteenth-century
Anglo-American fiction through a combination of function word counting, topic
modeling, and distance metrics of the type employed by Piper and Algee-Hewitt.
Jocker's book—the lone recent attempt to explore computational methods of lit
erary study at monograph length—is well worth a read for this specific result and
for the range of quantitative methods it presents. The romanticist Ted Under
wood has used related approaches to assess the evolution of literary diction in the
eighteenth and nineteenth centuries, changes in twentieth-century critical prac
tices, and the generic composition of the massive HathiTrust digital archive.7
Geographic space and patterns of literary attention are areas of particular
interest to comparatists. Among the highest-profile digital humanities projects is
Stanford's "Mapping the Republic of Letters," which seeks to extend the work of
Pascale Casanova and others on the same object through a database of period cor
respondence and a series of customizable and interactive cartographic visualiza
tions. It shares with So and Long an emphasis on metadata rather than on compu
tational content analysis.
My own research has used natural language processing techniques to identify
place names in literary texts and to associate those names with specific geographic
coordinates, as shown in figure 3. In the case of U.S. fiction published around the
Civil War, this method provides evidence in support of a significantly transatlantic
and international literary-geographic investment among American writers, with
more than forty percent of all location mentions falling outside the boundaries of
the United States. It also demonstrates the overwhelming extent to which U.S.
domestic locations fell outside New England, tracking westward population shifts

7 On literary diction, see Underwood and Sellers; for professional practice in critical journals,
Goldstone and Underwood; on genres, Underwood, Black, Auvil, and Capitanu.

This content downloaded from 210.31.15.200 on Mon, 01 Jan 2024 10:18:19 +00:00
All use subject to https://about.jstor.org/terms
COMPARATIVE LITERATURE / 18

and southern political events rather than remaining rooted in the older strong
holds of the Puritan imagination. The data thus support the increasingly transn
tional critical view of nineteenth-century American literature while pointing t
the salience of domestic demographic and economic developments tied to migra
tion and urbanization as drivers of literary attention in the period.
In all of these cases—from Piper and Algee-Hewitt's study of Goethe, to So an
Long's modernist networks, to Jockers' influence diagrams, Underwood's critica
genealogies and genre analyses, and my literary maps—the intellectual contribu
tions are to be found in two distinct domains. They present interventions in estab
lished or emerging critical debates, which is to say that they make specific claims
about the nature and function of their literary objects. They are achieved pieces of
computationally assisted literary criticism, meant to stand on their own, even a
they often represent just one step in a larger interpretive program. At the sam
time, they provide broad and genuinely new contextual information about the
configuration of whole literary fields. This information is qualitatively different
from that provided by nearly all existing scholarship in literary history or sociol
ogy, not because it is necessarily better for using numbers, but because, as Moretti
has noted, fields as such "cannot be understood by stitching together separate bits
of knowledge about individual cases" (4). Systems of literary production may b
made of books, but they are not themselves books any more than an elephant is a
very large pile of cells. Computational methods that provide information abou
entire fields thus help even the most traditional critics to understand both the lit
erary systems that have sometimes been their professed objects as well as the larger
contexts in which their individual texts appeared.

Issues and Future Directions

As rewarding as are the computational methods described here, t


important barriers to their widespread adoption in comparative an
erary studies. Setting aside these disciplines' strong methodological
(often framed in grand terms as a desire to "defend the humaniti
putative positivism of any approach that might supplement close
largest issues are three. One is legal and technical: it can be difficu
suitable corpora for computational analysis, especially in the era af
many texts remain in copyright. This is one of the reasons that so mu
computational work has focused on eighteenth- and nineteenth-ce
which can be digitized freely (if not always easily). But collections from
can be both sparse and "dirty," lacking high-quality metadata or
numerous transcription errors, often the result of performing automa
ognition on irregularly printed documents. In any case, it's the twenti
where the volume of literary production is well and truly overwhelmi
where computational methods are most urgently needed. Help is on the
form of the HathiTrust digital library and its associated research c
hold more than fourteen million digitized volumes (in an array of l
inally published from the 1500s to the present) derived from library s
ects including the massive Google Books. There are obviously meta
scription errors in the HathiTrust collection, and the details of access
extensive in-copyright holdings are still being worked out, but it appe

This content downloaded from 210.31.15.200 on Mon, 01 Jan 2024 10:18:19 +00:00
All use subject to https://about.jstor.org/terms
AC LA FORUM / 19

will be the single best source for truly large-scale literary work in the years to
come. Still, there will be an ongoing need for more specialized and higher quality
corpora for those projects that depend on pristine input or absolutely complete
coverage within a period or area, and curating these corpora will remain relatively
resource intensive.

A second issue has to do with people. At the moment, there are few scholars i
the humanities who have been trained in the skills and methods necessary f
computational work. This is changing for reasons both intellectual and marke
based, but it will be many years before every graduate student will be expected to
have at least a baseline competence in quantitative methods. Comparative liter
ture may suffer in this regard from its small size, although comparatists' tendency
to work across multiple departments and national literary traditions increases the
opportunities to gain training wherever it may be found. Fortunately, ours are not
the first disciplines to discover the use value of quantitative methods. We stand to
benefit from the large body of existing expertise in many corners of the university
even as we inevitably find it necessary to adapt others' techniques to our uniqu
concerns.

Finally, comparative literature and comparative cu


will need to deal with the challenges of working acr
techniques discussed here vary in their attachment
them depend in one way or another on counting wo
terns of occurrence. These can be applied to texts in
ally difficult to compare the results of the same pr
tions in different languages. One of the advantages of
is that it relies on publication metadata rather than fu
them to compare quite directly systems of modern
lish, Japanese, and Chinese (and in any other langua
suitable data). But this is not a disabling difficulty: com
and collectively, the advantages that come with hav
texts and processes their objects of study. Consider
phenomena examined by the projects I discuss: glob
romanticism, transatlantic literary formation, the R
large, multilingual, globe-spanning, yet nationally unev
why they're especially in need of the new types of evid
can offer. But they're also, and for the same reason, am
comparative study. Digital humanities needs compara
need computational methods.

University of Notre Dame

Works Cited

Casanova, Pascale. The World Republic of Letters. Trans. M.B. DeBevoise. Cambridge: Harvard
2004. Print.

Damrosch, David. What Is World Literature? Princeton: Princeton UP, 2003. Print.

Dimock, Wai Chee. Through Other Continents: American Literature across Deep Time. Princeton: Prince
ton UP, 2006. Print.

This content downloaded from 210.31.15.200 on Mon, 01 Jan 2024 10:18:19 +00:00
All use subject to https://about.jstor.org/terms
COMPARATIVE LITERATURE / 20

Elson, David K., Nicholas Dames, and Kathleen R. McKeown. "Extracting Social Networks
Literary Fiction." Proceedings of the 48th Annual Meeting of the Association for Computational Lin
tics. Ed. Jan Hajic, Sandra Carberry, Stephen Clark, andjoakim Nivre. Uppsala: Association
Computational Linguistics, 2010. 138-47. Print.
Giles, Paul. The Global Remapping of American Literature. Princeton: Princeton UP, 2011. Print.
Goldstone, Andrew, and Ted Underwood. "The Quiet Transformations of Literary Studies: W
Thirteen Thousand Scholars Could Tell Us." New Literary History (2014): Forthcoming.
Jockers, Matthew L. Macroanalysis: Digital Methods and Literary History. Champaign: U of Illino
2013. Print.

Luhmann, Niklas. Social Systems. Trans. John Bednarzjr. and Dirk Baecker. Stanford: Stanford UP,
1995. Print.

Michel, Jean-Baptiste, and Erez Liberman Aiden. "Quantitative Analysis of Culture Using Millions of
Digitized Books." Science331.6014 (2011): 176-82. Print.
Moretti, Franco. Maps, Graphs, Trees: Abstract Models for a Literaiy History. New York: Verso, 2005. Print.
Piper, Andrew. "Reading's Refrain: From Bibliography to Topology." ELH80.2 (2013): 373-99. Print.
Piper, Andrew, and Mark Algee-Hewitt. "The Werther Effect I: Goethe Topologically." Distant Read
ings/Descriptive Turns: Topologies of German Culture in the Long Nineteenth Century. Ed. Matt Erlin and
Lynn Tatlock. Rochester: Camden House, 2014. 155-84. Print.
Ramsay, Stephen. Reading Machines: Toward an Algorithmic Criticism. Urbana: U of Illinois P, 2011.
Print.

Samuels, Lisa, and Jerome McGann. "Deformance and Interpretation." New Literary History 30.1
(1999): 25-56. Print.
Schmidt, Siegfried J. "Literary Studies from Hermeneutics to Media Culture Studies." CLCWeb: Com
parative Literature and Culture 12.1 (2010): <http://dx.doi.org/10.777l/1481-4374.1569>.
So, Richard Jean, and Hoyt Long. "Network Analysis and the Sociology of Modernism." boundary 2
40.2 (2013): 147-82. Print.
Tangherlini, Timothy R., and Peter Leonard. "Trawling in the Sea of the Great Unread: Sub-Corpus
Topic Modeling and Humanities Research." Poetics 41.6 (2013): 725-49. Print.
Underwood, Ted, and Jordan Sellers. "The Emergence of Literary Diction ."Journal of Digital Humani
ties 2.1 (2012): <http://journalofdigitalhumanities.org/l-2/the-emergence-of-literary-diction-by
ted-underwood-and-jordan-sellers/>.
Underwood, Ted, Michael L. Black, Loretta Auvil, and Boris Capitanu. "Mapping Mutable Genres in
Structurally Complex Volumes." Proceedings of IEEE Big Data. Santa Clara, California (18 Sept.
2013): <http://arxiv.org/abs/1309.3323>.
Wallerstein, Immanuel, ed. The Modern World System in the "LongueDuree" London: Paradigm, 2004.
Print.

Wilkens, Matthew. "Nothing as He Thought It Would Be: William Gaddis and American Postwar Fic
tion." Contemporary Literature51.3 (2010): 596-628. Print.
. "The Geographic Imagination of Civil War-Era American Fiction." American Literary History
25.4 (2013): 803-40. Print.

This content downloaded from 210.31.15.200 on Mon, 01 Jan 2024 10:18:19 +00:00
All use subject to https://about.jstor.org/terms

You might also like