Virality and (Speculative) Bibliography
From a textual standpoint, virality is, we would propose, another name under which to discuss bibliography, as both seek to describe how often a given text was repeated and in what venues. Unlike an enumerative bibliography, however (which focuses primarily on accounting for witnesses), or a descriptive bibliography (which attempts to account for the textual differences among witnesses), accounts of a text’s virality focus on its social life and rhetorical power. How far and in what forms did it spread? In which communities did it circulate? How was this text modified, remixed, responded to, or commented upon? To what extent did this text saturate a given network? How does the spread of this text compare with that of others? And, finally but perhaps most importantly, what textual, thematic, or stylistic features allowed this text to be easily shared? The lens of virality privileges questions of circulation, reception, and comparison over questions of textual authority; it elevates the social over the stemma. As an interpretive frame, virality also requires a more capacious understanding of “the text,” including in its ambit both bibliographic witnesses and a penumbra of related, often fragmentary texts that speak to the social, literary, or historical impact of the original item. As described above, our uses of “edition” and “cluster” in this book speak to this collective sense of the text, not as a singular object but as a set of texts moving through space and time, or a series of cultural gestures.
In this book and the larger Viral Texts project, we largely focus on novel and extended forms of repetition and not, for instance, on short quotations from more canonical sources. Lincoln Mullen’s America’s Public Bible project draws from the same newspaper corpora as Viral Texts, but attends to the reuse of biblical quotations in order to ask whether particular invocations of Bible verses are “typical or exceptional” in particular periods.33 Mullen’s project demonstrates how historical practices of excepting and reprinting allow a single—if large and heterogenous—text to be deployed for widely diverse social, political, and rhetorical ends. Where Mullen seeks to identify the wide, steady, and granular distribution of a single, hyper-canonical text through the newspaper, we seek to identify more temporally-demarcated bursts of editorial and readerly activity around longer textual selections. While we do note many texts going the rounds multiple times over the course of decades, in most cases we observe distinct waves of attention and fallowness.
A viral theory of textuality extends the notion of “the social text,” first proposed by D. F. McKenzie in Bibliography and the Sociology of Texts and further developed by Jerome McGann through projects such as the Rossetti Archive. As McGann notes, when McKenzie first proposed the “social text” as an editorial procedure, prominent bibliographers “remarked that while McKenzie’s ideas had a certain theoretical appeal, they could not be practically implemented” because “critical editing—as opposed to facsimile and diplomatic editing—was designed to investigate texts, which are linguistic forms, not books, which are social events.”34 In the Rossetti Archive, McGann sought to represent the “social text” by including all editions of a given work in an online archive, rather than simply the “Reading Text” and “Variorum Text” of the standard critical edition. However, even the social text model remains focused on discrete works—books, most often, though also stories or poems—that can be collated and compared as distinct entities. Viral textuality is messier still, aligning fragmentary texts and textual echoes not only through books but also through ephemeral and hybrid media, the latter of these exemplified in this book by the nineteenth-century newspaper. The “viral text” of a particular poem would comprise official and unofficial reprintings, but also parodies, quotations, reviews, paraphrases, allusions, and more: all the echoes that testify to that text’s reception, the event of that text resonating through media.
As texts moved through nineteenth-century newspaper networks, they assumed new forms and sometimes sparked wider viral events—flurries of responding textual activity. The metaphor of virality, while imperfect for describing the circulation of cultural ideas and artifacts, nonetheless proves useful for framing the social text: its vectors of transmission, its diverse audiences, its many and varied modes of expression. Perhaps most importantly, virality narrows our attention to cultural repetition and remediation, to modeling the social diffusion of content. The terms we borrow in this book—virality, memes, viral events, rhetorical velocity—are in essence new ways of framing bibliographic description—what was printed, where, when, and how often?—within a textual system that makes definitive or singular answers to those questions difficult to pin down. Virality helps us better reckon with a distributed and diffuse print culture(s) that thrived well beyond the east coast, urban centers such as New York, Boston, and Philadelphia that tend to disproportionately dominate literary and historical scholarship. In addition to those enumerative questions, virality also prompts us to ask “why”: what qualities, internal or external, drove editors to reprint texts, or led readers to read (and sometimes save) them? Finally, virality prompts us to ask how we might understand selection and editing as expressive, even literary, practices? A theory of viral textuality must wrestle with unusually capacious ideas of the text, including in its purview the continually shifting penumbrae of readers’ responses that testify to that text’s life within culture and its mass media.
Practically, by bringing together our analyses with the code that generated and an evolving set of speculative bibliographies, as we will describe further in the next chapter, our project and this book take up the challenge McGann identifies with implementing McKenzie’s framework of the social text. Our analysis aligns both discrete and fragmentary texts that evidence texts authorized and fragmented lives within culture. Our speculative editions offer a critical structure for locating specific witnesses within dynamic print systems, or for measuring the rhetorical velocity of social texts. Our bibliographies will continue to evolve even after this book is printed, as our sources expand and we hone our computational methods. The social text, in other words, comprises both the trajectories of these newspaper selections in the nineteenth century and the trajectories of our historical ßunderstanding in the twenty-first.
Book Terminology and Citations
To reflect these capacious notions of textuality, we use several specific technical terms to refer to the objects of our study in this book: the texts reprinted in nineteenth-century newspapers and often excerpted, fragmented, or dramatically reconfigured. We refer to an individual reprinting using witness, a term of art in the field of bibliography that designates a single material instance of a given work. A particular copy of the first edition of Harriet Beecher Stowe’s novel Uncle Tom’s Cabin would be considered a bibliographic witness, for example, whereas an edition of the novel would comprise all copies, or witnesses, produced from a single setting of type. By this definition, each newspaper reprinting of a given text could be considered a new edition, since it was produced from a new, local setting of type; if they reprinted the same poem, the compositors for the Boon’s Lick Times and those for the Volcano Weekly Ledger would each have set that poem in type from an exemplar.
In a strict bibliographic sense, then, each newspaper reprinting of a text is a distinct edition. However, as we discuss in greater length later in the next chapter, the proliferation and fungibility of newspaper texts makes these bibliographic distinctions fundamentally messy in ways that do not perfectly align with book production. While the bibliographic edition stems from material realities of production, it also captures something temporal: copies of books created around the same time, stemming from the same production event and circulating to contemporaneous readers. In this book we attend not to whole publications but to fragments: small sections of text, printed on particular newspaper pages that appear in close temporal proximity to similar selections printed in sections of other newspaper pages that might be geographically quite dispersed.
A single newspaper text might have proliferated through hundreds of editions over the course of weeks or months, then, but that book-bound term becomes unwieldy when what we really want is to capture the event of a text’s rapid circulation and reception through the periodical press. Because of these issues, we use witness to refer to a specific instance of a reprinted text, as printed on a particular day in a particular newspaper, and use edition more collectively, to designate the set of genetically-related texts reprinted around the same time across multiple newspapers. This definition of edition is, we confess, a bit baggy, but helps capture the near-simultaneous multiplicity of the texts at the center of this study. To put this in terms of research about twenty-first century, online virality, what we call an edition composes the viral event of a particular text’s rapid circulation and popular attention.
Finally, the term cluster is in most ways quite similar to edition above. Cluster derives not from bibliography but instead from the computational methods we use to determine textual similarity, which are described in detail in the next chapter. Cluster is distinct from edition insofar as it is, theoretically if not always in practice, a broader term, including all texts that were designated by our reprint-detection algorithm as matching. Clusters sometimes include false positives: texts that are not reprints of each other, but perhaps all include a longer, shared quotation from a text such as the Bible. Similarly, clusters sometimes gather bibliographically distinct texts that nonetheless share enough language to be clustered together, such as parodies clustered with their originals—common for the newspaper poetry in our study—or even political speeches that share enough stock phrases that our reprint-detection methods match them. While these are all species of textual duplication they are not, as we will argue later in this chapter, editions of reprinted newspaper selections. Finally, cluster is often a temporally broader term, and in many cases we might mark multiple editions of a single selection within a cluster. A certain recipe might have circulated rapidly between 1852-1855, then disappeared, only to be republished and recirculated between 1876-1880. Our computational model might gather witnesses from across these periods of circulation and fallowness into a single cluster, but as textual scholars we can recognize each wave of publication as a viral edition of the text. Even here, however, the lines are not always clear; between two distinct waves of reprinting we might observe occasional individual witnesses that may be separate editions, or may be connected to the observed editions by reprints in newspapers not included in our data. Our use of edition, then, is closely tied to editors’ and compositors’ collective work, their understanding that a particular selection was currently “going the rounds” of other papers.
In this book, then, we conceive of cluster, edition, and witness as closely linked terms moving from a computational model of textual similarity, to genetically and temporally linked sets of a given text, to specific instances of a given text. We might consider these terms as a series of nesting dolls, ordered from most capacious to most narrow: cluster → edition → witness. All three terms work together to help model and describe the chaotic, dynamic textuality of nineteenth-century newspaper exchanges. The clusters that we publish with this book are our speculative bibliographies—our editions—of nineteenth-century newspaper reprinting, collected through a computational model rather than individually by hand.
Our references to reprinted texts in this book reflect our commitment to working within, rather than around, the messiness of nineteenth-century newspaper reprinting. When we write about a particular text, we may cite the textual or paratextual details of a particular witness, for example describing how one editor introduced the text in their newspaper. However, our primary referent is always the edition or cluster, rather than any particular witness. We want readers to hold the full event of each selection’s circulation in mind and have ready access to it. Practically, this commitment manifests in the IDs associated with each cited reprint in this book, which direct readers to all clusters associated with that text in our database of Viral Texts reprints. We describe the theory behind this practice in more detail in the following chapter, but in brief: while reading this book online, each time we cite a particular reprinted text, one transcribed witness will be available for quick reference in the Manifold reading platform. This transcription serves also as a bibliographic anchor text, identified by a unique ID that links directly to a database of associated clusters identified by our reprint-detection algorithm. When our algorithm clusters likely reprints based on textual similarity, it identifies any clusters that are internally similar to our transcriptions for this book project and associates those clusters with those anchor texts. Importantly, while the transcription’s ID remains stable, a search term that will return all clusters associated with that anchor text, the database will continue to be updated as we refine our methods and incorporate new data into the project. The full set of clusters associated with any particular anchor text—and the size and contents of those clusters—will continue to evolve as the Viral Texts project continues its work. If you are reading our account of the popular poem “Beautiful Snow” in chapter three today, for example, you will find 268 witnesses in the linked cluster, but if you return in two years you might find a different number as we add more newspaper corpora to our study and refine our reprint detection methods.
Just as we have found transcribing anchor texts useful for tethering our own historical evidence within an iterative computational project, this practice has proved useful for external scholars looking to identify clusters of interest within our data. While researchers can use keyword search to seek out texts related to their work in our Proteus database, keyword search can be unreliable due to textual variations introduced by historical reprinting practices in the nineteenth-century and optical character recognition (OCR) in the twentieth and twenty-first centuries.35 We have worked with several scholars who submitted plain-text transcriptions of their own anchor texts, which we can then incorporate into our workflow. After we next run our reprint-detection algorithm, those scholars can find all clusters associated with their transcriptions in Proteus, or we can send tabular data. With the launch of this book on Manifold, we can solicit transcriptions from all readers through the annotation features of the platform. If you would like to learn more about the newspaper reprinting history of a particular text, you can submit a transcription for incorporation into Viral Texts workflow.