Junot Díaz’s writing actively questions the boundaries between genre and “literary” fiction, aesthetics and politics, and English and Spanish, using a framework of multiple linguistic, formal and cultural registers to establish an authorial presence that defies critical categorization. Díaz arrived explosively on the U.S. literary scene with his second book, the novel The Brief Wondrous Life of Oscar Wao (2007). The multi-generational story of a Dominican American family overshadowed by a brutal dictatorship and the challenges of forging a new life in the United States earned many accolades, including the Pulitzer Prize for Fiction and the National Book Critics Circle Award. Three years after its publication in 2008 it remained at the top of Amazon's bestseller list for the “Hispanic American Literature & Fiction” category and is well on its way to central canonical status. The “galvanic” prose of Oscar Wao [1] traverses “tropical magic realism, punk-rock feminism, hip-hop machismo, post-postmodern pyrotechnics and enough polymorphous multiculturalism to fill up an Introduction to Cultural Studies syllabus,” to quote two of the novel’s reviews from The New York Times, and it illuminates the rules of contemporary canonicity through the incandescent exceptions Díaz has carved out [Scott 2007] [Kakutani 2007]. I will read Díaz’s transgressive blending of genre and linguistic registers as a “reverse colonization” that calls into question the demarcations of American ethnicity as well as the racial politics of nerds (a term I define below). My argument uses Díaz as both an object of study and a paradigm for the potential of a hybrid digital humanities methodology. The complex cultural translation that Díaz asks his readers to perform creates a middle ground where Caribbean history, language politics and the class and ethnic tensions of immigration collide with the nerdy core of the mainstream American imagination.

My methodology in pursuing these claims is to define a framework for “the literary” in contemporary American fiction by asking how books are contextualized and discussed not just among critics and scholars but also among a general readership online. Digital traces of book culture (by which I mean user reviews, ratings and the algorithmic trails that our browsing and purchasing actions leave online) allow us to make claims about relatively large groups of readers and consumers of books, creating opportunities for the ‘distant reading’ of literary fame, but without losing the specificity of individual texts and authors.[2] As we shall see, this approach attempts to find a middle ground between the trend for large scale, data-driven digital humanities research and the kinds of analysis more comfortable to traditional literary scholars. My research here incorporates book reviews from professional critics and everyday users of the websites Amazon and LibraryThing as well as automated recommendation networks operating on both of those sites. By considering these digital traces of reading through the conceptual frame of the network, this article explores literary culture as a process of contextualization and elevation. My approach draws from John Guillory’s work on cultural capital and his argument that the syllabus and the list are central to the process of canon-making [Guillory 1995, 28–38]. The list, expanded into multiple dimensions, becomes the network. These new dimensions are visible articulations of taste: both book reviews and website recommendations make certain cultural assertions about the relationships between various texts and authors. By recasting these assertions as links and the names and titles as nodes, we can begin to see where books cluster and where they remain solitary. This is a way of studying the social forces influencing books and readers at their intersection with literary production itself, and it draws on related approaches that James English and Mark McGurl have pioneered in their work on contemporary literary culture, and more distantly from Bruno Latour and Pierre Bourdieu [Bourdieu 1993] [Bourdieu 1996] [Bourdieu 2007] [English 2005] [Latour 2005] [McGurl 2009]. In short, I propose a turn in the digital humanities to incorporate not just algorithms but algorithmic culture into our critical framework and to recognize that humans and machines are learning to read in new ways.[3]

Network analysis allows us to explore the middle ground between the individual text and the distant reading of thousands of texts, exploring the emerging space of cultural distinction for a particular author’s work at a particular time.[4] It is this combination of specificity (in terms of time and authors) and generality (in the sense that hundreds or thousands of data-points might influence a particular configuration between books) that offers us a window into a changing system of literary prestige, a glimpse at the middle ground where readers and critics make their judgments within constrained horizons. Professional book reviewing and commercial consumption are two major intersections between the worlds of art and commerce, zones of exchange where popular canonicity and literary prestige are forged. As Pierre Bourdieu noted, “the boundary has never been as blurred between the experimental work and the bestseller,” a truth that can hardly be better demonstrated than with the runaway commercial and critical success of Díaz’s Oscar Wao [Bourdieu 1996, 347]. The novel’s deliberate complication of genre and ethnic literary boundaries has led Díaz to an almost defiant acknowledgement of his mainstream success: “They’re so happy to claim me as literature because it makes them all look better. They don’t want to relegate me to areas of ethnic studies”  [Céspedes and Torres-Saillant 2000, 905].

This position has led scholarly reception of Díaz to its own complex multivalent space, with critics from the Latino/a, Caribbean, African Diaspora and Dominican literary traditions all interpreting his success (Díaz acknowledged each of these traditions in the interview quoted above). Daniel Bautista agrees that Díaz is “somewhere both inside and outside of the ‘‘mainstream,’’” writing fiction that defies a simple binary between ethnic studies and the mass market [Bautista 2009, 89]. Indeed, Elena Machado Sáez suggests that “the picture emerging from the criticism of Oscar Wao is of a superheroic literary text…that breaks through oppression to posit an ideally marginal but resistant diasporic subject”  [Sáez 2011, 525]. A reaction to this superheroic reading argues instead that the novel is a darker meditation on the violence inherent in narrative. For example, the text’s own charming but dictatorial narrator (Yunior, a version of a recurrent protagonist from Díaz’s earlier stories) serves as a troubling proxy for its primary narrative engine, the dictatorship of Rafael Trujillo in the Dominican Republic from the 1930s–1960s [Hanna 2010]. Reading the novel as a “historiographic battle royale” clearly aimed at a United States reader, as Monica Hanna does, neatly encapsulates the challenges Díaz’s work poses to the established categories of genre[5] and comfortable patterns of reading. Its gripping style conceals a refusal to translate along multiple axes: the novel does not gloss its Spanish and Spanglish slang for that mainstream reader, nor does it contextualize its references to Caribbean literary figures like Edouard Glissant any more than it explicates its allusions to comic and fantasy characters like Darkseid or Sauron. This “weird English,” as Evelyn Nien-Ming Ch’ien calls it, demands that its readers bridge the gap, confronting the untranslatable and interpolating around it, grappling with Díaz’s rebellious idiom.

By insisting on blending codes and discourses beyond Dominican and “standard” English, Díaz alienates himself from particular communities in the tradition of commercially successful postcolonial writers around the globe [Brouillette 2007, 60]. In applying the same technique to subcultures within the mainstream United States, he makes a subtler point about racial politics. As an immigrant, Díaz did not take solace in his identity as a Dominican American, since that identity was continually pressured by poverty and racial bias. Instead, he claims “what saved my life was being a nerd, watching all those bad science fiction movies and reading cartoons”  [Ch’ien 2004, 221]. Díaz doubly rejects normative behavior, embracing neither the standards of his own ethnic community nor those of mainstream American culture. This renunciation is standard nerd practice, but the staging is usually quite different. The nerd is a complex and shifting figure in contemporary culture with its roots in what Mary Bucholtz calls “hyperwhite” performance, including “superstandard English” (like Oscar Wao with his “flash words”) and the rejection of “cool” culture [Bucholtz 2001, 86, 87] [Díaz 2007, 50]. I extend this frame to assert that the nerd is defined in relation not merely to “superstandard” language but highly specific dialects — the mastery of technoscientific skills, obscure popular genres and, increasingly, a socially impermissible obsession with almost any arcane subject. For example wine nerds, XML nerds and comic book nerds are all likely to be among the readers of this article. In this sense, the nerd as I deploy the term here is an extreme version of the “fan” or “fanboy,” a phrase Díaz also uses and to which I will return below.

With Díaz as just one example, it is also important to note that the racial categorization of the nerd is becoming increasingly nuanced, with Michael Hanson defining the “Afrogeek” categories of “thug-nerds” and “rap-geeks” [Hanson 2007]. Oscar’s status as an ambiguously dark-skinned Dominican completes his own hybrid role as a “GhettoNerd” [Díaz 2007, 32, 11]. In fact, as Ron Eglash notes, the figure of the nerd is not so much a bastion of white male elitism as it is a “potential paradox that might allow greater amounts of gender and race diversity into the potent locations of technoscience” because it is not racially but performatively white, or “implicitly white”  [Eglash 2002, 50, 60]. While the future of nerds may be defined more by Barack Obama than The Simpsons’ Comic Book Guy, whiteness continues to function as a powerful stigma and defining attribute of the category, leading African American students who excel in science and technology to be accused of “‘acting white’” even as it sustains a form of “hyperwhiteness” performed by nerds of all kinds [Eglash 2002, 59]. Díaz’s choice to deploy the dialects of comic book nerds, science fiction nerds and others establishes a correspondence between ethnic boundaries and the deep, nerdy interior of the mainstream white imagination.

Allow me to develop this point through my own nerd self-affirmation as I discuss my methodology. The data I present here is drawn from literary culture “in the wild” — newspaper and magazine book reviews along with traces of digital consumption (consumer reviews and automated recommendations) from the websites Amazon and LibraryThing. Each of these sources links us to only a fraction of Díaz’s public readership, but they are communities of literary practice that illuminate how his stylistic project has played out in the mainstream. I will use two basic sets of data to build these empirical arguments. First, book reviews: I have collected and analyzed a set of professional reviews, author profiles and other literary press from nationally recognized newspapers and magazines, as well as a complementary database of consumer book reviews from Amazon.[6] I identify the proper nouns, particularly authors and texts that are mentioned together in these book reviews and create networks based on the number of times different terms co-occur in the same paragraph.[7] Over the course of this research project I assembled a hand-tooled dictionary of proper nouns and used a Perl script to seek potential matches in individual documents from this corpus (repeating the process numerous times as I extended the dictionary). The script created a series of basic XML files with proper nouns tagged as nodes that I checked for errors before analyzing for co-occurrences using another set of Perl scripts that indexed those nodes by paragraph. I stored the resulting connections in a MySQL database and output co-occurrences in GraphML format for viewing with yEd, a network graph editor [yWorks GmbH 2011].

The result is a basic network of primary literary entities — authors and texts — that are explicitly named[8] in particular paragraphs of Díaz’s professional and consumer book reviews. As a brief example, if we were to apply this system to a one-sentence review that said “Díaz reminds me of Frank Herbert, but also of Julia Alvarez,” we would get the following:

The co-occurrence links all three nodes in an equal, non-directional way. The method treats each paragraph as a “bag of words” with no attention paid to syntax or nuance. Replacing the review above with the following would generate the same diagram: “Díaz is nothing like Frank Herbert or Julia Alvarez.” This may seem counterintuitive, but I posit that the fact of connection is at least as important as its tenor. As I have argued elsewhere, the canonizing influence of a reference or a list in a piece of criticism serves to put the books in question into the same context, even if the reference itself is negative [Finn 2011, 16]. Another way of making this point is to consider the peculiar economy of the book review, a piece of publicly oriented secondary literature challenged with keeping the reader’s attention in a compressed format that is by definition a metanarrative. The set of major reviewing publications yielded 22 articles on Díaz, with an average word length of 1,294. By contrast I collected 613 Amazon reviews, averaging just under 166 words each. Reviewers deploy references with purpose in this context, and a negative comparison carries as much cultural weight for the reader as a positive one, establishing in both cases a contextual framework for the subject of the review.

As the average lengths indicate, this economy is especially true for reviewers on Amazon, where the competition for literary attention is sharpened by multiple reviews of the same book and the status of the arbiter is always in question. My methodology compares these two fields of literary discourse through the framework of proper noun networks. Tracing these webs of allusion, reference and context allows us to explore both sides of that great literary event horizon, the purchasing of a book. While there are of course exceptions, people tend to read professional critics before they buy something, and they tend to write Amazon reviews afterwards. If book critics are “gatekeepers” to the literary marketplace, readers on Amazon report back from beyond the turnstiles, telling us why they went through. Below we will see that as the balance of power between the two groups shifts, everyday readers are coming to define the “literary” in new ways.

The project’s second data set, recommendations, focuses on this literary exchange as a market: I have collected data from Amazon, the world’s largest bookseller, and LibraryThing, a leading “social reading” community site, regarding how consumers on Amazon associate books through their shopping patterns and readers on LibraryThing do the same by listing books together in their virtual libraries. Amazon and LibraryThing present this data in recommendations such as the “Customers Who Bought This Item Also Bought” mechanism on Amazon, providing an algorithmically driven feedback loop between accumulated acts of cultural distinction and specific kinds of literary desire. In this instance, I used Perl scripts to record an expanding web of recommendations on each site from a given start page (arbitrarily selected by me as the author in question’s most prestigious book). For Amazon, the script would start at Oscar Wao and record the first ten[9] “Customers Who Bought This Item Also Bought” links, and then move on to record the first ten links on each of those pages, continuing to three levels of depth. The same process was repeated for LibraryThing based on the first ten “LibraryThing recommendations.” As a brief example of how these recommendations are diagrammed, consider this excerpt from Figure 4:

In this data, the lines between nodes have directionality: browsers on the Oscar Wao page will see a link pointing them towards Kavalier & Clay. Likewise, those starting out on the Chabon page will see a link back to Oscar Wao. In this case, Amazon believes that readers of one book will enjoy the other, and the two texts end up mutually supporting each other in the larger system of recommendations.

The mutual link is based, Amazon claims, primarily on purchase data: enough people who bought one book went on to purchase the other that Amazon codified the connection, hoping to entice further sales from users making a similar cultural calculus. The fact that this feedback loop is maintained, manipulated and studied by Amazon speaks both to our increasingly corporate cultural lives and to the importance of recognizing the contingency of data — all data. Amazon sells books and manipulates its recommendations in order to sell them more efficiently, so it is possible that the mutual connection above is shaped by other factors, such as marketing campaigns, but we can rest assured that the connection is productive, or it would be replaced by another, more productive one.[10] That is to say that given Amazon’s dominance, we can assume it is reasonably deft at this particular cultural game: acting as an algorithmic mirror to its users’ literary desires, inflecting and reflecting taste through countless computations and statistical judgments.

A full analysis of Amazon’s growing dominance in literary production, from author contracts to sales, is beyond the scope of this article, but I want to briefly sketch out its role and my reasons for studying it. At this stage the relevance of Amazon to mainstream literary life is hard to deny, if we give any credence to critics like John O’Brien, publisher of the Dalkey Archive Press, who believes “the greatest threat to book publishing in the United States right now is Amazon” (quoted in [White 2011]). Whatever role Amazon plays in our readerly futures, its role in the present is to influence a huge number of acts of literary distinction, and to study these acts is to explore a core sample of mainstream American literary culture. And it is mainstream in all the ways Díaz inflected that term above: its users are better-educated, wealthier and less likely to be Hispanic than the U.S. Internet user average [Quantcast 2011]. Reading Díaz through Amazon is to read him through a broad American public still rooted in a white mainstream, unveiling a cultural politics of authenticity and success that surfaces constantly in the data below. That contest of literacies is mirrored by my methodology’s contest for what Ted Striphas calls “algorithmic literacy” [Striphas 2011]. Just as I am “reading” Amazon here, Amazon is “reading” its users and perfecting algorithms to perform that reading ever more efficiently and more closely.

For his part, Díaz confronts the question of literacies with a culturally charged mixture of the familiar and the exotic (an effect that varies in specifics but not in nature depending on whether the reader identifies most readily as Latino/a or white, an English professor or a science fiction nerd), and my approach does something similar by challenging our notions of reading on several levels. “Reading,” in this framework, is a fundamentally social action, bound up in collectively negotiated conceptions of language and inflected at every stage by the sociocultural influences that lead us to books and shape our thinking before, during and after our direct engagements with them. It is also, as I noted above, a form of collective interpretation that can be performed by Amazon’s servers as well as human beings. As more of our literary cultural actions, from browsing to discussion, migrate to screens and digital media, reading takes place increasingly online. “Readers,” then, are the users visible through their aggregate choices in the methodology just described.[11] It should also be noted that professional critics and Amazon reviewers are not necessarily distinct, though I have yet to encounter a professional critic who also voluntarily (or openly) reviews books on the site. But of course there are overlaps — this is a marketplace, and professional critics influence markets, just as markets influence critics — perhaps the most obvious being that many critics purchase books on Amazon. The middle ground perspective that I adopt here balances out the impact of such individual overlaps by looking at reviews and recommendations in aggregate, exploring the remarkable public literary spaces created by thousands of readers online.