The Frankenstein Variorum Challenge: Finding a Clearer View of Change Over Time

No cultural artifact lasts forever in a single, static, universal form. Texts change with editors and audiences, and tracking their changes challenges our ways of reading and understanding, heightening awareness of mediations, censorships, interventions, reinventions. Yet, telling the story of change over time in a variorum edition can be tedious, disruptive of joy in the reading experience, alien to the popular experience of “getting lost in a book.” As Brett D. Hirsch and Hugh Craig have discussed of Shakespeare editions, the print variorum has yet to be surpassed for its “highly developed economy of presentation.” (Hirsch and Craig, 2014: 26-27) Can we find a humane way—attuned to readerly impatience and excitement in the reading process—to build an edition that teaches a non-specialist how to read change over time, without sacrificing the standards of textual scholarship? Must digital methods that comprehensively document variation at their best mimic the printed critical edition in burying the experience of change over time to the footnote or the end space? Or, can we design a variorum edition that engages not only scholars but also fans and new readers in the dynamic complexity of a changing text?
Since the early days of hypertext, Mary Shelley’s novel
Frankenstein has inspired experiments using electronic texts that invite new ways of reading. The novel about assembling a new monstrously superior life from dead bodies easily becomes, in the age of computers, a metaphor for attempting to unify a super-text complexly assembled from multiple variant pieces, since the process of computationally comparing texts necessitates editors to identify the smallest meaningful units to align and process. Like Victor Frankenstein's assembly of the Creature from multiple corpses, variorum editors aspire to animate a new composite reading experience. Such was the experiment to make the 1818 and 1831 editions of
Frankenstein simultaneously readable in the Pennsylvania Electronic Edition (PAEE) produced by Stuart Curran and Jack Lynch in the mid-1990s. Designed in a framework of small pieces like the hypercard novel popular at the time, the PAEE encouraged the reader to wander between two editions, explore hyperlinked ancillary material, and resequence the novel at will (see
http://knarf.english.upenn.edu/Colv1/f1101.html
). The transfer of this edition to Romantic Circles, a scholarly web resource in British Romanticism, conserved its textual data without exceeding its early intertextual vision of a complex, particulated reading experience with many reading paths always available (see
https://romantic-circles.org/editions/frankenstein).

Our project to design a new Variorum of
Frankenstein began with an encounter with these earlier digital editions. Originally called the “Pittsburgh Digital
Frankenstein project,” and inspired by the worldwide bicentennial celebrations of
Frankenstein’s first publication, we started work in October 2016 as a voluntary collaboration of Pittsburgh-area digital humanists with the Maryland Institute of Technology in the Humanities to update the
Frankenstein edition on Romantic Circles. The original goal was to revise and update the TEI code of the Romantic Circles edition to find ways to connect it with the

Shelley-Godwin Archive

's

(S-GA)

diplomatic edition of the

Frankenstein

notebooks
. The mission shifted to design a variorum edition to collate five editions of the novel produced in the author’s lifetime, rather than simply to update and interlink the two most famous earlier editions. Our team from the University of Pittsburgh, Carnegie Mellon University, and the University of Maryland are now producing a new variorum edition of the novel encoded and published in TEI XML. Inspired in part by Barbara Bordalejo’s
Online Variorum of Darwin’s Origin of Species, the
Frankenstein Variorum incorporates annotations and visualizations that highlight change over time to guide readers to discover the "hotspot" moments of most intense variation in the novel.

The
Frankenstein Variorum is developing a scholarly edition that augments but does not replace previous print and digital editions. Our project pulls earlier editions into a new comparative perspective, to view the elaborately encoded S-GA edition of the manuscript through a variorum interface designed to illuminate where the manuscript contains material lost in later editions, and where later editions preserve or adapt its original content (see Fig.1). At DH2018 and at the Balisage Markup Conference in June and July 2018, Beshero-Bondar and Viglianti presented on the first phase of these efforts: preparing the differently-encoded source documents for computed collation with CollateX and building the foundational structure of the Variorum as a "spine" of standoff pointers into the separate edition files (Beshero-Bondar and Viglianti, 2018). This DH2019 paper addresses a new phase of work: developing a browser-based user interface that incorporates accessibility and responsive design features, inviting visitors to read specific editions, review variants, support edition-specific and cross-edition annotations, and visualize GIS data of the movements of characters, author(s) and co-editors in space and time.

One of the most complex elements of the
Frankenstein Variorum is its application of stand-off XML to point to rather than recreate the diplomatic TEI encoding of the 1816 manuscripts in the S-GA. Our interface should guide readers in navigating the complexity of the edition, how to find its most interesting moments of variation, and how to read the richly encoded textual information about the writing process available in the markup. To help guide the reader, the annotations team on the
Frankenstein Variorum seeks to balance contextual annotations (informing about sciences, gender and political contexts, literary allusions relevant to the novel) with a more innovative commentary designed specifically for navigating the Variorum: a set of “tunneling” annotations to highlight internal changes across the five editions. These annotations will augment previous scholarship on the persistence of various hands in the composition, emendation, and revision process. Incorporating these annotations with the variant apparatus poses a challenge for designing our interface that we look forward to discussing in our paper.

The
Frankenstein Variorum must support traditional, linear readings of individual editions along with nonlinear explorations of textual comparison through collation units and “tunneling” multi-edition annotations. To achieve this, we build our web interfaces directly on the TEI without transforming it to HTML first. We use CETEIcean, a JavaScript library for publishing TEI documents in the browser, to display the texts, visualize the linkages among
Frankenstein’s sources, and scholarly annotations to the TEI-encoded texts (Cayless and Viglianti, 2018). We aspire to the principles of Universal Design (UD) to increase access for those with visual limitations or who require a less cluttered and more responsive interface. We acknowledge the near-impossibility of addressing the variety of user needs and devices used to access web-based content (Godden and Hsy, 2016); yet, if we cannot meet the ideal, we can nevertheless draw on UD principles to expand our audience and inform our creation of vision-sensitive color palettes and adaptive stylesheets for those with difficulty differentiating particular colors or low visual acuity. The UI will be optimized for those using tablets and larger displays, applying Responsive Design to prioritize tablet delivery, limit navigation clutter, and highlight primary features including the text reader, variations table, and annotation window.

We are refining a method of signaling the level of “heat” or variance of a selection across editions, based on Levenshtein distance (or edit distance), the mathematical calculation of the minimum number of edits required to change one text string into another. We are aware of the limitations of Levenshtein for evaluating intensity of variation, since a change of a single letter or mark may totally change the meaning of a passage, such as the edit distance of 1 between the words “went” and “wept”. Nevertheless, as a convenient method for surveying intensity of edits on a macro scale, edit-distance values prove helpful. The tunneling annotations and overrides to the computed color codes will help to intervene where we discover the Levenshtein values to be less meaningful. Our color-coding system applies the maximum edit-distance based on pairwise comparison of each set of divergent witnesses at given variant locations in the novel, as represented in Figure 1 below.

Fig. 1: Prototype view of the Frankenstein Variorum interface.
Features Depicted in Fig 1:

A single text reading window with identifying text and navigation elements for edition, volume, chapter, and page. (Left side)
Alternate reading frame to review collation sections with navigation by collation unit and source. (Left side, alternate)
Variations frame holds a table that displays the strings from each edition in a collation table.
An annotation frame will display relevant annotations for the text selection, including those shared by multiple editions,
Color intensity indicates intensity of the variance across a human-mediated ten-value range. Words and passages are encoded using this scale.

At DH2019 we will share a more complete view of the interface, its annotation apparatus, and visualizations that will help guide the reader to sections where the most intensive editing took place. We look forward to conducting usability testing to enhance the interactivity and responsiveness of the digital interface in the coming months. We hope to provide a user experience that invites Frankenstein fans, new readers, and scholars alike to take pleasure not only in reading multiple
Frankensteins but also in reading the story of how this novel changed over time.

Bibliography

Beshero-Bondar, Elisa E., and Raffaele Viglianti. (2018). “Stand-off Bridges in the Frankenstein Variorum Project: Interchange and Interoperability within TEI Markup Ecosystems.”
Proceedings of Balisage: The Markup Conference 2018. Balisage Series on Markup Technologies, vol. 21.
https://doi.org/10.4242/BalisageVol21.Beshero-Bondar01.

Cayless, Hugh, and Raffaele Viglianti. (2018). “CETEIcean: TEI in the Browser.” In
Proceedings of Balisage: The Markup Conference 2018. Balisage Series on Markup Technologies, vol. 21.
https://doi.org/10.4242/BalisageVol21.Cayless01.

Godden, Rick and Jonathan Hsy. (2016). “Universal Design and Its Discontents.” MLA position paper in
Disrupting the Digital Humanities. Eds. Dorothy Kim and Jesse Stommel.
http://www.disruptingdh.com/universal-design-and-its-discontents/

Hirsch, Brett D. and Hugh Craig. (2014). “‘Mingled Yarn’: The State of Computing in Shakespeare 2.0.”
Digital Shakespeares: Innovations, Interventions, Mediations. Ed. Brett D. Hirsch and Hugh Craig. Special issue of
The Shakespearean International Yearbook 14: 3-35.
https://hcommons.org/deposits/objects/hc:10794/datastreams/CONTENT/content

Williams, George H. (2012). “Disability, Universal Design, and the Digital Humanities” in
Debates in the Digital Humanities.
http://dhdebates.gc.cuny.edu/debates/text/44