The Genetic Dossier in the Web of Data

1 Introduction

Archivists and genetic scholars differ considerably in their methodological frameworks, which may explain why they

are largely not taking part in the same conversations, not speaking the same conceptual languages, and not benefiting from each other’s insights. (Caswell 2016, 2)

The digital turn in archival preservation and scholarly editing presents a window of opportunity to bring the two fields together. As more GLAM institutions responsible for preserving and providing access to literary archives of modern or contemporary authors undertake large-scale digitisation of material under their custody, new possibilities emerge for digital scholarly projects to reconnect those resources within a Semantic Web ecosystem, fostering closer collaborations between genetic scholars and archival institutions. However, the Web of Data also presents specific challenges that warrant further research and attention.

After comparing differing organising approaches to literary archives in recordkeeping and genetic criticism, this article will consider how Semantic Web technologies may bridge the methodological frameworks of both fields, while outlining two current obstacles to this pursuit: some limitations of LOD in representing genetic dossiers in informative ways and a series of legal issues that prevent digital scholarly archives of genetic orientation from realising their full potential in the Web of Data.

2 Literary Archives in Recordkeeping vs. Genetic Criticism

Typically comprising the working and personal papers of modern or contemporary authors, literary archives are characterised by a wide variety of material,¹ whose significance and appreciation set them apart from most other types of archival repositories. Increasingly cherished by literary enthusiasts and collectors since the beginning of the Romantic Geniezeit,² the relevance of literary archives lies in “the insights they give into the act of creation”, which translates into “a higher financial value” and may explain why an author’s holographs end up being “scattered in diverse locations” worldwide (Sutton 2014, 295‑6):

Literary archives […] tend to travel much further than other types of papers and to be housed in unpredictable locations – often […] determined by market forces rather than by internal archival logic. […L]iterary papers are usually found […] to be divided between several collecting institutions. This phenomenon, which we have come to call ‘split collections’ […] represent[s] an essential part of the world of literary manuscripts. (Sutton 2018, 7‑8)

Indeed, very few writers have their manuscripts entirely preserved in a single institution, whether a public or university library, a state archive, a private foundation, a literary house, or a museum. Not only do authors themselves disseminate documentary evidence of their writing, offering drafts as a memento or gift of friendship (Boie 1993, 42‑43) to those with whom they correspond throughout their lives, but, after passing away, their estates are also frequently divided among heirs and subject to the sort of posthumous plunder that is implicit in the Latin root of the Portuguese term designating literary archives: espólio (from the Latin spolia, ‘spoils’ or ‘stolen treasures’).

While archival studies and literary genetic criticism approach the spoils of an author with very different methodologies, we believe their perspectives and understandings “overlap and can be brought together” (Bunn, Rayner 2019, 360).

Archival studies is a subfield of information science dedicated to “the nature, management, and uses of records”, defined as “persistent representations of activities, created by participants or observers” (Caswell 2016, 3, 5). Traditionally, recordkeepers curating literary archives are guided in their work by a foundational principle of “respect for the fonds” (Muller, Feith, Fruin 2003, 54), which implies preserving the so-called “provenance” of archival documents that bear an organic relationship to one another. In practice, this means that documentary pieces shall not be aggregated by subject or any other interpretative criteria; instead, records created by different individuals must be “kept separately”, with their “original order” and context preserved (Caswell 2016, 7), while classified according to genre, type, or material support (Lopes 2007, 55).

This traditional understanding of “provenance” as an organising principle of archival studies contrasts with the “speculative approach” (Drucker, Nowviskie 2004, 431) of scholarly editors with a “genetic orientation” (Van Hulle, Shillingsburg 2015, 36), who, conversely, view records not as a “fixed product” (Bunn, Rayner 2019, 369), but as “dynamic objects in motion” (Caswell 2016, 6).

Originating in France during the 1960s, in connection with the Centre d’Analyse des Manuscrits in Paris (later evolved in the current ITEM – Institut des Textes et Manuscrits Modernes), genetic criticism succeeded in adding a temporal, paradigmatic dimension to the literary text, regarded as a process rather than a product, by drawing attention to its variations in draft form and all the transformations that result from the author’s writing or rewriting activity over time. This type of “archaeology of the manuscript” (Van Mierlo 2013) relies on literary archives, mainly from the nineteenth century onward, to provide insight into the compositional development of a literary work and expand the interpretation possibilities of the text:

manuscripts […] offer up new and unseen material, and also suggest, in their very physicality, the writing methods and processes unique to the subject of study. They can further ‘solve factual problems like the dating of a poem or establishing an accurate text’ and ‘illuminate the broader meanings of a literary work’ (Gioia 2004: 36). Beyond this, archival materials offer us other conduits of research and knowledge, […revealing], as Cook argues, the ‘context behind the text, the power relationships shaping the documentary heritage, and indeed the document’s form and content’. (Stead 2016, 4)

For that to be in place, scholars must compile a genetic dossier (Grésillon [1994] 2016, 286) comprising all the physically dispersed documents of an author’s writing project that “bear witness to the evolution of the work” (De Biasi 2004, 38). This may include the version records that preceded publication (e.g. notes, drafts, revised manuscripts, typescripts, print proofs), as well as other correlated evidence of the broader interpersonal networks that contribute to the author’s creative process, such as his library and correspondence. The methodology involves not only collecting all extant genetic documentation (recensio) but also comparing the respective textual variants to infer the genealogical relationships among the collected pieces and organising the work’s avant-texte (Bellemin-Noël 1972) according to the writing chronology.

Whereas recordkeepers use systematic guidelines in their archival practice, based on such principles as provenance, collective control, or original order of records, genetic scholars aggregate different authorial material of various provenance (e.g. marginalia in books, notebooks, draft manuscripts, typescripts, letters exchanged with other people), subjectively organising the jigsaw pieces into speculative archives – the genetic dossier – aimed at reconstructing the author’s creative process.

In recent years, the digital medium has significantly facilitated the constitution of these interpretative scholarly archives,³ allowing researchers to aggregate facsimiles and transcriptions of material scattered across different institutions and model their textual relations within a dedicated virtual environment for specific academic purposes.

3 Transitioning Genetic Dossiers into the Web of Data

Over the past decade, scholarly editorial initiatives, such as the Samuel Beckett Digital Manuscript Project,⁴ the Shelley-Godwin Archive,⁵ or the Gustave Roud: Textes & Archives,⁶ have succeeded in digitally reuniting dispersed documentation of modern and contemporary authors, facilitating the examination of the genetic dossier of their works.

Although Wout Dillen has rightly noted that many of these DH projects call themselves archives,⁷ their “archival impulse” (Eggert 2019) diverges from the principles followed by librarians and recordkeepers, who are invariably bemused by the term used in this context.⁸ Instead of attending to the provenance of documents, digital scholarly archives of genetic orientation are “hermeneutical instruments” (Ramsay, Rockwell 2012, 79) aimed at organising a “purposeful collection of surrogates” (Price 2008) to reveal interpretative connections between dispersed textual witnesses. Presenting themselves as a “work-site” (Eggert 2005, 433), a “knowledge site” (Shillingsburg 2006, 88), or a “platform for learning” (Theimer 2014, 146), those projects actively engage with the dynamics of variation through a range of digital tools to allow readers to navigate multiple versions of a text and follow the author’s compositional development over a more or less extended period of experimentation and revision.

In addition to XML-TEI markup and algorithmic collation, participatory editorial projects, such as the LdoD Archive,⁹ have been exploring social editing functionalities, supported by structured databases and Web 2.0 environments that enable users to create virtual editions of the documentation.¹⁰ Scholars are invited to manipulate the “dynamic layer of the archive” (Portela 2022, 191), either performing editorial script acts such as annotations, or reconfiguring the writing sequence of texts, in what some have also been labelling as new interactive “forms of analysis and creativity”, in line with the poststucturalist “esthetic of the possible” (Gooding et al. 2019, 386, 376).

More recently, digital scholarly archives of genetic orientation have also been drawing inspiration from advancements in Linked Open Data (LOD) and other technologies that follow up on Berners-Lee’s vision of a Semantic Web of Data (Berners-Lee et al. 2001; Berners-Lee 2006), to reveal and enhance the complex network of relationships devised among various documents of a genetic dossier. In a nutshell, these projects use persistent URIs to identify resources and apply web ontologies to formally represent relationships or the underlying logic among different nodes in the documentary network. Resource Description Framework (RDF) datasets, represented as subject-predicate-object triples, will model a graph structure that computers can interpret, while users can interact via a SPARQL endpoint with a graphical interface that leverages the semantic layer for querying and manipulating the graph database.¹¹

Among the projects that have been applying Semantic Web technologies to genetic dossiers,¹² the Gustave Roud: Textes & Archives (Jaquier, Maggetti 2022) deserves special mention, as the team designed a new data model, formalised in the Web Ontology Language, specifically for Genetic Criticism.¹³ This GeNO ontology effectively describes the interwoven networks within and outside an author’s genetic dossier and can be queried using cURL and Gravsearch, a virtual graph search based on SPARQL that allows researchers to find, for instance, which diary entry of the author ended up in his fictional work.

The Shelley-Godwin Archive (Fraistat et al. 2013-present) is another initiative worth mentioning, as it builds on linked data principles and the Shared Canvas data model to support a participatory platform where anyone on the web can describe, discuss, and reuse facsimiles and transcriptions of archival material, within a global, interconnected network of information that aligns with the 5S model¹⁴ and an interdisciplinary vision of “Linked Research” (Capadisli 2016). Interestingly, the project stems from a partnership with several public and university libraries, expanding still-rare collaborations between literary scholars and recordkeepers¹⁵ into the Web of Data and opening up possibilities for further cooperation.

In fact, as more GLAM institutions lead large-scale digitisation projects that adhere to protocols such as the International Image Interoperability Framework (IIIF) and the Text Encoding Initiative (TEI), new opportunities emerge for archivists to incorporate digital scholarly archives into local descriptions, enhancing or contextualising their records. Conversely, scholars should also be able to connect genetic dossiers to the authors’ archival repositories and personal libraries,¹⁶ allowing users to navigate the virtual research interface without losing contact with the provenance and archival order of the material records. But while steps have been taken along the path, the vision of a global web of literary archives remains “far away on the utopian horizon” (Fordham, cited in Anderson et al. 2021, 7), hindered so far by at least two main obstacles.

The first issue that stands out is the lack of shared vocabularies, ontologies, and “good human-usable interfaces for the Semantic Web” (Brown, Simpson 2015). Recent initiatives promoting Linked Open Data vocabularies for the description of manuscripts¹⁷ and textual variation¹⁸ are promising contributions to addressing the problem, but so far, digital scholarly projects experimenting with ontologies to express textual relationships across different authorial materials have designed their data models independently of the archival records underpinning the projects.¹⁹ One suggestion to overcome the current disconnect between literary archives and genetic dossiers, enabling rich, interlinked data to be shared and repurposed by third-party applications, could involve open knowledge bases such as Wikidata, as well as Solid Pods specifically designed for GLAM institutions to share archival records and promote decentralised data networks, as this kind of resource enables different web APIs to provide new views into the knowledge graph.²⁰ Still, while knowledge graph visualisation for the Solid ecosystem is making progress and paving the way for further research,²¹ experts in graph technologies recognise that network graphs in general do not present complex datasets of textual information in a clear and intelligible manner, mainly because Linked Data is a machine-readable format not intended for humans,²² and “[m]any levels of discursive mediation are needed for the methods of close and distant reading to productively inform one another” (Stoyanova 2023, 39). The Gustave Roud: Textes & Archives, for instance, drew inspiration from celestial maps to explore the centrality of the author’s diary within the genetic dossier, achieving positive usage test results among experts (Elli et al. 2023, 32, 36), but interpreting graph representations of such complex data sets is incredibly difficult for literary scholars without the technical knowledge for querying the ontology.²³ As such, we need archivists, genetic scholars, and data scientists to come together and codevelop graphic user interfaces that make graph network visualisation more accessible and “informative for a wider audience”.²⁴

In the case of modern and contemporary literary archives, however, another major obstacle to reconnecting authorial material within a Semantic Web ecosystem stands out, due to a fundamental conflict between the free availability of distributed data, implicit in the concept of LOD,²⁵ and a series of legal restrictions affecting authorial repositories, particularly in countries with a legal tradition of Droit d’Auteur. Despite usually being deposited in institutions funded by public resources, manuscripts and other archival material of twentieth and twenty-first-century writers is subject to a series of copyright and non-copyright restrictions that protect the privacy and moral rights of authors, forcing GLAM institutions to “curtail the widespread digitisation of whole collections” (Anderson et al. 2021, 7) and restrict access to “a minority of researchers who have the time and funding” to view the documents on site (Jaillant 2019, 290). In fact, those willing to use contemporary literary materials for research purposes often find themselves in a never-ending maze of bureaucracy, including formal authorisations from both copyright owners and custodians of the material, which implies dealing with different propriety, authority, dependency, and privacy restrictions:

The owner of copyright for material in the Manuscripts Collection is the writer or creator of the material, or the creator’s legal heir(s). Note that the donor of the material is not always the copyright owner. In addition, many collections contain a variety of letters, diaries, documents owned by multiple copyright owners. […] Should you wish to publish material from the Library’s Manuscript Collection, you will need to: declare your intention to the Library as custodian of the material, obtain copyright clearance from the copyright holder(s).²⁶

26 National Library of Australia, “Rights and the Manuscripts Collection”. https://www.library.gov.au/services/copyright-library-collections/rights-and-manuscripts-collection.

As demonstrated in a previous article dedicated to major legal obstacles encountered by European genetic scholars, recent exceptions introduced by the CDSM Directive did not meet the requirements of ongoing advancements in digital humanities,²⁷ making the path towards publishing and providing online international access to contemporary literary archives difficult to navigate, especially for unpublished works, where “the waters are particularly muddy” (Dillen, Neyt 2016, 788). Before taking further steps towards a much-anticipated vision of genetic dossiers in the Web of Data, scholars investigating twentieth and twenty-first-century authors therefore need bold policy-making adjustments to ensure that their work will not be rendered worthless by someone refusing publication permission:

we need to extend the scope of the available exceptions […] to allow for scholarly publication in the digital age – or otherwise, a legal license designed with scholarship in mind so that academic researchers may work with published texts and holographic materials in public archive libraries, disclosing research results (in person, on paper, and online) without interference from heirs or successors. Moreover, we also need national or European management systems led by independent copyright boards to facilitate the clearance of orphan works for different uses and reduce the randomness of our current authorisation system. (Pereira 2023, 523‑4).

4 Conclusion and Outlook

In his 2014 book on memory and scholarship in the age of digital reproduction, Jerome McGann argued that to study literary creativity, scholars needed “cultural records to be comprehensive, stable, and accessible” while being able to augment “that record with our own contributions” (McGann 2014, 131‑2). The proposal implied shifting the idea of archival records as fixed informational resources to embrace the digital scholarly perspective on the term, regarded as “a complex system inhabited by all the different agents involved in the production of academic work” (Bunn, Rayner 2019, 369).

In this regard, Semantic Web technologies present a window of opportunity to enhance interdisciplinary collaboration among archivists, textual scholars, and genetic critics, reapproaching their methodological frameworks to think anew about the working methods of prominent writers. While the persistence of practical obstacles to this pursuit leaves the full potential of genetic dossiers in the Web of Data untapped, much work has been done to overcome those shortcomings and interdisciplinary working groups must be formed to carry on the efforts, co-designing archival software for the Semantic Web ecosystem and simultaneously allowing for so-called “digital forensic work” on literary archives, i.e.:

multiple modes of ordering and interpreting while also, at the same time, securing the collections that underpin this innovative work. (Gooding et al. 2019, 376)

Meanwhile, scholars and recordkeepers must come together and exert pressure on legislators to introduce the policy changes necessary to allow greater freedom in using copyrighted works for the preservation of cultural heritage in the Digital Single Market. The recent (albeit insufficient) TDM exception introduced into European legislation demonstrated that only through sustained commitment can we achieve the legal measures necessary to enable ongoing developments in computational literary studies. Allowing contemporary documentary collections and genetic dossiers to transition into the Web of Data should be our next goal.