Networking a Scholarly Edition—Networking Ulysses as a Digital Text and Research Site

Author: Hans Walter Gabler (Ludwig-Maximilians-Universität München)

Assuming a graded distinction between information sites, knowledge sites, and research sites,[1] what is it that specifically marks out the digital research site? The types of site are all repositories of content. The content they comprise and harvest is substantial knowledge. The research site in particular is structured by nodes for content enrichment. Assembling, recording, and holding knowledge, the research site serves simultaneously to modify, revise, and increase it. In research practice, this happens in a man-to-machine query:response interaction. Utilizing the dynamic potential of the digital medium, the digital research site brings to the fore the dynamic process nature of knowledge. The constant dialogic negotiation of its content and substance is what renders the digital site a research site.

Enmeshed as we are today in the digital medium, we use the internet comprehensively as an information-cum-knowledge-cum-research site. In the net environment, we commonly devise dedicated sites ourselves. These may be given over to themes and research questions generally conceived, and for which the knowledge substance may be profitably drawn from, as well as enrichingly committed back to, the data cloud in cyberspace. Alternatively, they may be constructed over foundations of material heritages. This is comprehensively the case in the humanities disciplines drawing on the cultural heritage of texts. Here, the locus for preserving and mediating texts of every genre has of old been the scholarly edition in print. With editions in print, the medium deployed has been identical with that of the initial materialization of the texts they preserved and mediated. Today, while texts live as of yore to be read in books, the locus for the scholarly edition is the digital medium.

Written testimony of the past provides the foundation for the scholarly edition mediating the cultural heritage inscribed in texts. The disciplines span widely that produce and rely on scholarly editions. Today, under the aegis of the medial shift, they have in common their outreach towards methodological innovation. As an interdisciplinary grouping—to name only history, literature, musicology, art history, philosophy, jurisprudence (in terms of its texts and their history)—they significantly contribute to defining and lending substance to the term and notion of the Digital Humanities. In the individual humanities’ disciplines severally, it is true, longstanding traditions of convention and procedure have variously shaped scholarly editions. Generically, nonetheless, the scholarly edition constitutes a significant bracket across the disciplines. This means today no less than that the scholarly edition, in terms of digital methodology and procedure, constitutes common ground for the Digital Humanities. Even pre-digitally, scholarly editions always already constituted nodes for research. A scholarly edition in the digital medium is hence predestined to be—or, realistically speaking, predestined to be developed into—a site for dynamic interactive research: a distinct subtype of the digital research site.

Although scholarly editions in the respective disciplines have over the past decades largely already gone digital, the medial migration has not been accompanied comprehensively as yet by in-depth reconceptualizations of method, format, or purpose. Predictably instead, first impulses have been to shape digital editions to patterns adapted from the printed book. (The digital medium reveals its full range of potential only gradually, after all, as we deploy it.) Conveniences of organization and presentation that the digital environment offers have been taken advantage of, it is true. But the fundamental notion is by and large still adhered to, namely that the scholarly edition establishes, and by establishing decrees a text, surrounds it with auxiliary material, and lastly goes public only as an end product, as the crowning achievement of the text-critical and editorial labor invested. What results from such premises will doubtless be sound by the conventional standards for scholarly editions. Yet even though the result under present-day circumstances be in fact a digitally born edition, it will not possess the dimensions, quality, and potential of a digital research site. The scholarly edition gone digital is simply not (yet) such a site. Instead: to innovate the scholarly edition as a digital resource, it needs to be rethought in terms of the essential qualities of the digital medium, its process dynamics, and dialogic interactivity.

In consequence, I wish to class the digital scholarly edition as a mode, a distinctly definable subtype, of the digital research site. It is distinct in that it does not primarily reach out centrifugally from a kernel of theme or research question into the data cloud in cyberspace. Instead, holding its core within itself, it is centripetal. It stores the comprehensive record of the material evidence of its research object at its own digital center. In incorporating in digital transcription all extant traces and bodies of writing, text, and text transmission of that research object, it is commensurate with the scholarly edition in the pre-digital paperbound medium. It differs from editions in book presentation, however, in that as a (type of) digital research site it lives in the digital medium.

How, then, would one network the material substance from writing, text, and text transmission as, and into, a digital research site? What structures and features would need to be newly invented and implemented for the purpose? To what extent and in what selection, at the same time, would one accept inheriting conceptions and practices from the proven traditions of the material scholarly edition?

To set a framework for the innovation adumbrated, it is useful to ask initially what indeed it is that, by convention, the paperbound scholarly edition actually edits. Do scholarly editions in print, have they commonly edited (a) text, or a work? Where, and as long as, editions were bound within the same material medium that carried and transmitted their objects, in the first place, one could afford to leave this question open, or not even to ask it. To make available in print a text under a given title meant essentially to present the work of that title. Behind this equation lay weightily, moreover, a theory definition of the “work” as an entity of creation achieved at a moment of closure of its text—so achieved analogously by authors ending their writing in acts of publication, and, within Anglo-American textual criticism and editing in particular, of editors “closing in” on texts of authors’ (final) intention. This theory position has in literary criticism long been superseded in favour of a conception of the essential openness of the work of art in language. Its openness depends on its potential to mean. This potential in turn is a function of the ceaselessly dynamic articulating power of language.

To respond to the articulating power of language and to actualize the potentials of meaning requires that writing and the written be read and interpreted. It is for reading and interpretation throughout that writing in our cultural tradition has been brought before the public in print, be it in editions “straight from” their authors and publishers, or in critical editions established by textual critics and editors. What is, or has been, simultaneously assumed, moreover, is that editions (whether “straight” or editorially mediated) give us the work whose title is inscribed on the edition’s title page.[2] Strictly and materially speaking, however, what they provide is text—commonly one text for the work in question, unembellished in straight editions, multiply embedded in critical editions. The one text is on all accounts frozen in closure. Text materialized on paper thus manifestly stands before us as a product. Yet by the power of language out of which it is made, text always kindles and re-kindles processes of interpretation—dialogic, and thus hermeneutic processes triggered through reading. The hermeneutic processes are predicated on the processuality of the text. If product is one, then process is the other side of the nature of texts. As product, a text is inertly uncommunicative. It is from out of its process potential that it speaks. By interpretatively actualizing this potential of the text, we moreover also discern the work for which (by its paratextual title) the given text stands. But if it is true to say that it is only through interpreting the text which an edition offers for reading that we arrive at an interpretive understanding of the work, then this defines “work” as immaterial, as an entity but adumbrated hermeneutically. “Text,” by contrast, is its materially manifest foundation. Assuming the validity of this distinction, we gain, for one thing, a clearer sense of how to answer the question as to whether an edition edits (a) text or a work: it is precisely through editing text that it edits a work. Moreover, and more essentially for our purpose, we are put in a better position to explore by what bearings to set up a digital research site expressly for a work, and whether and to what extent to do this on premises of the scholarly edition, or how indeed to modify and extend that format.

Let us approach this complex and its problematics centripetally, and firstly therefore from the reader’s and user’s perspective. In editions, whether “straight” or critically mediated, we encounter text frozen in closure as product. Acts of reading and interpretation, taking their departure from the product presentation, reanimate the text as processual. The process of interpretation corresponds to the processuality of the text. Interpretation in turn generates response texts: interpretations articulated as texts in limitless succession and proliferation, and expressing whatever sense is being made of the text, and of the work whose material substratum the text interpreted is. Related to the work, since generated from the text interpreted, the response texts—a potential galaxy, or data cloud, of them—constitute content material for a digital research site. That content material connects, in multiply specific correlation, to the site’s substantial core—that is, to the site’s text representation of the work interpreted. This text representation must consequently be enabled to function as the reference grid for all response texts attached to or otherwise associated with it.

So modelling the correlation between the data cloud of response texts and the core text of an edition as digital research site, we recover in a nutshell a main functionality of scholarly and readers’ editions of the Gutenberg and pre-Gutenberg eras. Books, and before them medieval manuscripts, while of necessity materially focused on the texts they reproduced and presented, were equally designed as instruments to mediate the works, and their ranges of meaning, for which these texts stood—or, in abbreviation: their function was to render understandable the texts they provided. The means by which they did this was annotation and commentary. All the more are scholarly editions in the mode of digital research sites apt platforms for the mediation of text(s) and work. But in the digital environment, the provision of the manifold bodies of associated text needs to be significantly reconceived. In books, characteristically, annotation and commentary cumulate additively and are, besides, provided for receptive consumption only. Grouping and cross-referencing, if attempted at all, is achieved at most via explicitly individualized page(.line) references and indexing. Attaching and associating response texts in the environment of a digital research site, by contrast, means comprehensively to enmesh these texts with the core text and to network them among each other: to link them singly or by group accession to every instance in the core text to which they apply, as well as to crosslink the body of associated texts in itself.

If the provision of annotation and commentary in the paperbound medium was additively incremental, the association and attachment of response texts in a scholarly edition qua digital research site is thus throughout to be relationally organized, and comprehensively to be coordinated by means of the reference system and grid of the edited text. It is by such stable referential structuring and organization of its content data that the digital scholarly edition is enabled to function as a digital research site. Furthermore, and perhaps all-importantly: the digital scholarly edition operating through a digital research site is not uploaded on the net as the sum and end product of the scholarly labour invested in it. Instead, with as much innovative inventiveness and care put into the structuring and organization of its content as of its forms of visualizing interfaces, it is made publicly accessible as a launching platform for community use and incremental enrichment.

The digital scholarly edition as a shared, dynamically progressive research site is, admittedly, a vision rather than an anywhere as yet accomplished reality. Nor have I here yet given sufficient attention to what I termed the core of this type of research site: the text. I have posited it as an edited text, and have already defined its reference function to hold together and coordinate the site in all its parts. To specify the edited text in substance, let it here suffice to assume that textual criticism and editorial scholarship be professionally exercised in the editing. For naturally, the standards to secure the site’s material foundations in their digital conversion must equal those traditionally deployed to prepare scholarly editions in book. At the same time, however, the editorial goal, as well as key premises of the exercise, must be freshly conceived. A critically established text within a digital research site differs in one significant respect from an edited text in a conventional book edition: it is not a text the editor or the edition decrees. It is instead an open offer of a text, initially both text-critically and critically considered, to be sure, and then given over to be critically engaged with in turn. The advocated community use of, and dynamic response to, the digital research site crucially involves the site’s readers’ and users’ engagement with the text.

The difference, thus, between a scholarly edition of old and a digital edition operating through a digital research site is fundamental. Since, as posited, the research site in the mode of a digital scholarly edition is designed to enable and support the dynamically progressive engagement with the work, the edited text the site comprises (if it does) serves but as a portal of entry to explorations of the work’s text in a comprehensive sense—that is, as found in evidence materially, and consequently transcribed digitally into the site, in manifold variegated instantiations. With the scholarly edition in the format of a digital research site, the days are over of the “critical edition” of Anglo-American, or the “historical-critical edition” of German-school observance. At the core of the digital research site is no longer the decreed edition but the dialogically arguable edition.

We already clarified at the outset above that a text made public, be it in a “not-edited” or an edited format (and hence either originating relatively unmediated from the author, or issuing through the mediation of an editor), represents in each case but one text instantiation of the given work. In the real world, in fact, and its vast realms of material textual transmission, there commonly exist multiple texts representing the work, which are, moreover, very likely to be multiply variant. It has consequently always been a basic exercise of textual criticism to correlate extant text instantiations. If now the digital research site is to fulfil its function to mediate the work, it goes without saying that it, too, must make the multiplicity of text instantiations transparent. In terms of its structuring as so far outlined, it needs to weave for them as well a relational network referenced to the site’s core text. The net so woven to hold and correlate the body of text that variantly instances the work must be designed, too, for interactively dynamic increase. Its reference base will be the same as that for the response texts of annotation and commentary—and naturally so, not only to observe reference consistency throughout the research site, but importantly also because the exploration of variation between text instantiations of the work is likely to generate interpretive commentary. Hence, the two nets posited as structured into the digital research site should be enabled to interact.

To see differences between texts not in the first place text-critically (and under the prejudicial assumption of error)—namely, as correct versus corrupt readings—but instead to engage with them critically to interpret the work represented in its texts: this is a novel approach to “the scholarly edition.” It requires conceptual rethinking, and significant reformatting in consequence. Comprehensively, this involves including, in the digital research site laid out for a work, all extant prepublication material of writing and texting pertaining to it. To justify the inclusion, both the theoretical and the practical positioning of the prepublication materials within the site need to be considered. We have already established the theoretical grounds. As said above: if product is one, then process is the other side of the nature of texts. By transcribing a series, maybe a multiplicity of text instantiations into the digital research site, we accumulate texts of the work as products. The moment we correlate these product states and versions by way of identifying their variants and interpretatively responding to them, we (re-)animate the texts’ processuality.

The process nature of texts takes effect from the moment they are first written down, and from then on the process of the text is materially observable and analyzable in the processes of writing. (Abstractly speaking, of course, the processuality of texts was operative much before the earliest writing traces, but we simply lack the evidence of the phases of mental composition preceding any written records that ever existed, or have survived.) Even as writing sets in to articulate text, it near-simultaneously also responds to the texting it materially produces. In the material evidence of composition before our eyes, the process of the text grows out of the process of the writing. In and through the processes of writing as they materialize, we are enabled to discern materially the processuality of the text, progressively lending sense and shape, meaning and form to the work.

From an author’s point-of-view, it is true, a work stands defined at its time of publication. In general cultural awareness, too, publication is a significant event. Yet, as we have maintained, publication at the same time imposes closure on the work’s text progression. It freezes the work’s text in a synchronous state. Text itself, by contrast, always extends in time. The work’s text therefore reaches back to the earliest germs of composition, and it stretches out from there across the full range of its instantiations—be it in the form of notes, or sketches, or drafts, or fair copies, or states of publication. On the strength of the processuality of its text, the work is in essence diachronic. Art formed out of language into works is a time art. On such understanding, in consequence, a digital research site dedicated to a work must comprise all text manifestations pertaining to the work and must be rendered capable of simulating the work’s diachronicity in the digital medium.

To bring in all material evidence for the work from the time span before its first publication cannot be accomplished by textual transcription only. It requires a double digitized input. The progression of the text in its initial phases is inextricably tied to the progression of the writing. What needs to be made accessible for the research site, therefore, are both high-quality digital facsimiles of the original documents of writing and, together with them, digital transcriptions of equally high professional standard of these drafts. Drafts are a sui generis type of document, both allographic—that is, providing reproducible text—and autographic, meaning they are paper-space supports of visual originality for laying out the traces of inscription of compositional writing. This double nature of drafts makes it essential to incorporate them both as images and as transcriptions of the writing that fills them. In consequence, too, the digital research site should provide a dedicated space to bring this double input into a state-of-the-art correlation of image and transcription for the purpose of researching the processes of composition and revision as such. At the next level of networking, the manuscript-image plus draft-transcription partition must also be enmeshed comprehensively with the site’s reference framework. Once more, the reference grid provided by the research site’s core (edited) text will need to be relied upon to support queries into, and analyses of the work’s diachronicity throughout, reaching back to the seminal processes of writing as processes of its text-in-progress and following them up through all states of text along the work’s entire time-line. This is where a digital scholarly edition, as distinct from the traditional scholarly edition, is set categorically apart from the edition in book. Holding in evidence the material substance of the text’s and the work’s progression in time (provided of course that instantiations of the work in composition and revision survive at all to be incorporated), it can be designed to replay in the virtuality of the medium the entire process of text into work. It so acquires the format of a genuine genetic digital edition. This adds distinctly to the potential of the digital research site as a whole within which the digital scholarly edition is framed. As the multi-dimensional research platform it is—or, as I as yet imagine it to be—it can lend support, too, to genetic criticism.


Against this theorized background and systematic mapping of a digital research site in the mode of a digital scholarly edition, I append here the vision of such a research site for James Joyce’s Ulysses, as it is presently under discussion and in its very early phases of realization.

To conceive of a digital research site for Ulysses means to imagine a many-text relational network. In one respect, the site will comprise multiple instantiations of text for the work Ulysses itself. These instantiations will vary greatly in both extension and substance: from notes to textual snippets to sketches to drafts to typescripts to prepublications to book proofs to published editions. Complementarily, the research site will provide ranges of other texts, importantly so any amount of prose entering it as contexts and response texts (traditionally termed “annotation and commentary”). The overriding function of a digital Ulysses research site will be to con-textualize in multiple dimensions the conception, writing, composing, revising, and publishing, as well as the critical assessing of the work.

The Ulysses Critical and Synoptic Edition of 1984 will be made use of as a point of departure.[3] Tangibly, it is a three-volume (near-)conventional edition in book. In this identity it marks, and is marked by, its historical moment. As the book triple-decker it is, it constitutes in traditional ways a product of editorial scholarship. Yet this book edition already rests on a digital record of the processes that went into constituting it: the authorial processes of writing and revising and altogether shaping Ulysses, the processes of its publication, and as well the processes of realizing the edition. The edition was from the ground up both assembled and shaped digitally. From the digital record too, consequently, the production in book form was generated in tripartite organization and triple display, as

1. a synopsis of the assembled text’s genetic progression from fair copy to first edition, with emendation notes appended page by page to record the assembled text’s critical editing (these interlinked elements are provided on the book edition’s left-hand pages);

2. a reading text of the end result of the text assembly plus critical editing (this reading text is extrapolated onto the book edition’s right-hand pages);

3. a Historical Collation to record a full complement of variant readings in the main author-connected life-time printed editions of Ulysses (this is lemmatized in lists appended to the edition as a whole).

The reading text integral to the 1984 edition—which incidentally has been available since 1986 separately also in the commercial hardback/paperback editions from Random House, New York, and The Bodley Head, now Random House UK, London—falls into a historical line of published text instantiations for the work. Their main strand commences with the incomplete prepublication serialized in The Little Review (begun in 1918 and banned in 1920 through censorship after 13 of what would become 18 chapters, or episodes) and runs further from the first publication of 1922 onwards through the subsequent 1926, 1932, 1934, 1936/37, 1939, 1960, and 1961 book editions. Following in the wake of these, the 1984 reading text evidently, too, critically relates to them. Moreover, the three-volume critical and synoptic edition incorporates them all, for by means of its system of apparatuses it coordinates textually all the reading text antecedents in book editions.

That system of apparatuses in turn is a sub-system within the edition’s total structures of coordination pivoted on the processes of Joyce’s writing and shaping Ulysses—authorial processes that the edition in its turn coordinates under its process-oriented perspective of correlation and editorial shaping. It is in and through its comprehensive network of correlation and coordination that the critical and synoptic editing of Ulysses lives in its digital record. Accessible as it is today in XML-TEI-P5 encoding, this digital record holds an undiminished potential for unlocking and exploring the substance, processes, and correlations keyed into it. The digital research site for Ulysses envisaged will be grounded on this potential through actualizing it into born-digital processes of access and exploration. Through these processes as they stand to be developed, the research site will significantly grow from the seed-bed of its initial grounding.

What ranges of information are encoded in the digital record’s mark-up? Firstly, the mark-up identifies the text’s progressive layering as assembled from the fair-copy to first-edition material documents. The yield is a “raw text,” or “continuous manuscript text,” behind the edition text; the layering as such is in the book edition displayed synoptically. Secondly, the mark-up identifies the editorial modification of the text assembly (in other words, the editor’s critical emendation of the document inscriptions); and thirdly, it relates selected document conditions (such as transmissional departures from Joycean document inscriptions) to the acts of emendation. Emendations and selected document conditions are footnoted in the book edition. In the edition’s total digital record, fourthly, is also stored the textual variation from a main selection of post-first-editions of Ulysses. This variation is printed as Historical Collation in the book edition.

Thus, by the range and classes of the mark-up of the digital record it is possible to specify and isolate, for any number of analytical queries, individual levels of the textual development as straight texts or in variant relations to or between one another. Successive variation being thus “stackable” allows then (for instance) to identify text that—within the range of the writing process captured—was once under consideration, or valid, but did not “make it” into the published text. The record further allows dating textual operations, and thus also establishing time-lines.

Furthermore, from the edition’s text+footnotes+historical-collation materials may be generated, say, the Little Review—that is, the pre-first-edition serialization—text; the 1922 first-edition text; or, for instance, the 1934 or 1961 Random House texts. These, as the recognized historical “milestones” they are in the publication history of Ulysses, would not need to be fed into a digital research site by separate ingestion. They are already contained in, and thus available for extraction from, the synoptic edition.

From the synoptic edition’s digital record, moreover, two textual states of especial interest may be generated. These are the fair copy (Rosenbach manuscript) text and the “final working draft” text.[4] In view of a digital research site, the reason why the fair-copy and “final working draft” texts are of especial interest is that they are in the position—precisely, in the genetic position—to provide the anchoring ground for the incorporation into the research site of the documentation of writing and text preceding James Joyce’s fair copy of Ulysses. This documentation comprises all pre-fair-copy materials (notes, drafts, and so on) which the critical and synoptic edition does not yet incorporate. To a large degree, the material in question is found in documents that were in fact not even known to exist when the critical and synoptic edition in book was established. Over and above a minority already in the relevant Joyce collections, the majority of the material resurfaced only just over a decade ago.[5]

For the digital research site, these materials await editing in their own right according to state-of-the-art methodologies for genetic editing of today (and tomorrow).[6] Then, to be integrated into the Ulysses site, their genetic representation will require linking into the genetic development already digitally recorded. The writing and text for Ulysses throughout its progression crosses document border after document border. The border line between the pre-fair-copy genesis and what is already captured and stratified in the synoptic edition is constituted by the anchoring ground of fair-copy text, and/or final working draft text.

Incorporating all extant materially documented writing and text for Ulysses in the digital research site requires designing an encompassing reference framework. To this end, the reference grid of the critical and synoptic edition, which already organizes, too, the digital record, defines a point of departure. In print, what guides the use of the edition is the episode.line(.word) numbering established for its reading text. This has been identically maintained in all printed editions and has become the standard for referencing Ulysses. To design a reference framework and grid for the digital research site, this reference template could/should be adopted and carefully adapted so as to be rendered capable of coordinating and multi-directionally cross-referencing the digital research site entire. To organize and structure both input and access for interactive, dynamic and accretive use of the research site, it will on all accounts be necessary to conceive from the ground up a reference system stable and robust in itself as well as versatile in use applications. Should this require “backgrounding” rather than foregrounding the printed edition’s episode.line numbering, that numbering should nonetheless remain alive and re-generable so that, in principle, every stage of composition and prepublication revision becomes relatable to it, all pre- or post-publication stages become correlatable, and all the site’s response-text content accessible or attachable, through it. The internal reference system-to-be-devised will render the research site explorable; at the same time, it is the reference template of the episode.line numbering, though of print-medium origin, that will remain the cross-medial access key to the site for both the general and the specialist user.

On the foundation of an encompassing and stable reference grid it will be possible to ingest and attach to the reference base as constituted by the Ulysses text (and textual materials) all manner of external material: digital document images (of autographs, typescripts, proof-sheets, auxiliary instructions), textual notes, natural-language response texts in divers groupings (historical and factual; discursive; critical; multi-medial; as well as dynamically interactive dialogical contributions). With all functions and content seminally in place, the site should within a mid-term foreseeable future be launchable for dynamically progressive interactive use.

[1] As earlier suggested in Hans Walter Gabler, “Thoughts on Scholarly Editing,” review of Securing the Past: Conservation in Art, Architecture, and Literature, by Paul Eggert, JLT Online (2011), Reprint, Ecdotica 7 (2010), 105-127.

[2] According to Gérard Genette’s taxonomy of the elements of a book, interestingly enough, the title signaling the work for which the ensuing text stands, is itself a threshold into the book itself, and therefore belongs among the book’s paratexts, not to its text itself. See Genette, Seuils (Paris: Éditions du Seuil, 1987).

[3] James Joyce, Ulysses: A Critical and Synoptic Edition, ed. Hans Walter Gabler with Wolfhard Steppe and Claus Melchior, 3 vols. (New York: Garland Publishing Inc., 1984).

[4] In terms of traditional editorial conceptions and procedure, the “final working draft” is an “archetype” behind the fair copy and the typescript, and thus not a materially extant document. Its text however is inferable as a state-of-text jointly from the fair-copy and typescript texts for those episodes or stretches of episodes where the fair copy and the typescript inscriptions collaterally derive from it.

[5] See Michael Groden, “The National Library of Ireland’s New Joyce Manuscripts: An Outline and Archive Comparisons,” Joyce Studies Annual 14 (Summer 2003): 5-17.

[6] They are likely to provide rich testing ground for the TEI’s “Encoding Model for Genetic Editions.”