To the Edge, and Back Again: Redeveloping a Dying Database, Women Writers – SHEWROTE

As we enter an era in which more and more DH projects are ageing, long-term sustainability looms large on DH agendas. Over the past decade, studies have variously analysed factors hampering project sustainability (Dombrowski 2014, 2019, 2020), prompting project PIs to think strategically about where to preserve data, and in which form, including end-of-project planning and project embalming (Rockwell et al. 2014). These factors loom large in the  choice between repurposing or restoration to give legacy projects a useful afterlife (Antonini et al. 2020).

How then does one revive an older database once the software it was developed in is no longer supported, and after project funding has run out? How does one preserve the work of past scholars while making the project  fit for new types of scholarship and future generations? This post discusses the experience of the Women Writers in History DARIAH Working Group which has redeveloped a database that had been developed over three decades at several Dutch institutions. Initially built at the University of Amsterdam and released in 2001, as Women Writers, at Utrecht University, the database was revamped in 2014 at the Huygens Institute as NEWW: New Approaches to European Women’s Writing, and is now being redeveloped at Radboud University as SHEWROTE. Over several technical overhauls, hosting at four institutions, and research questions that evolved significantly between 2001 and today, this complex history posed multiple challenges for the database’s redevelopment. 

Database Breakdown: From one to Multiple Databases

The first challenge pertained to the data itself. The Women Writers database provided an exceptionally rich dataset detailing the reception of women writers and their works from Antiquity to circa 1945. Data had been created and curated over the preceding decades by a distributed network of close to 300 researchers across Europe and beyond. Given the diffuse, diverse nature of this network, data-editing took place not in a top-down way, i.e. from a predesigned data model, but bottom-up, with researchers heuristically describing material as they encountered it. However, one of the consequences of this organizational structure was that as the project matured, it increasingly grew to encompass different standards, definitions, research questions, and degrees of data comprehensiveness, as well as of documentation. 

A second challenge was changing levels of institutional support. In 2016, after almost two decades of continuous funding, this ended when the HERA-funded Travelling Texts 1790-1914 project (2013-2016) ran out. At this juncture, by becoming a DARIAH Working Group, Women Writers in History, the scholars still actively using the database were able to ensure both a modicum of institutional recognition, and a formal structure within which to continue work, punctuated by DARIAH’s Annual Event and other meetings. If the database itself was now heading toward a state of technical decline at least the human community around it continued to function.

The end of funding, indeed, coincided with the breakdown of the software upon which the database had been built. With database maintenance now lagging, it became increasingly difficult for researchers to edit data in the existing tool. Additionally, the underlying model proved too static to accommodate new types of source material. This accelerated the process of creating  spin-off databases created for specific projects with their own project funding. These included the the Serbian-funded Knjiženstvo database Theory and History of Women’s Writing in Serbian until 1915 headed by Biljana Dojčinovič (2014-2020), the ERC-funded RECIRC database (Reception and Circulation of Early Modern Women’s Writing, 1550-1700) headed by Marie-Louise Coolahan (2018-2020), the Finnish-Russian Texts on the move database headed by Viola Parente-Čapková (2018-2020), the ERC-funded MEDIATE database (Measuring Enlightenment: Disseminating Ideas, Authors, and Texts in Europe, 1665 – 1830) headed by Alicia Montoya (2016-2022). 

Database Failure: A Perfect Storm 

Then, in 2021, outright database failure looked imminent. There were three components to this failure: institutional, technical, and conceptual. Institutionally, the end of project funding and subsequent end of database maintenance was compounded by the retirement of the database’s original creator, Suzan van Dijk, and colleagues moving on to new institutions and responsibilities; this led the Huygens Institute to end database maintenance. Moreover, the original in-house software with which the database had been developed was no longer supported; since the database had been developed in an inappropriate non-SQL format, in a SORL platform originally intended for text editing purposes rather than databases, this effectively hampered researchers’ work and limited their autonomy. Finally, there were conceptual issues. Conceived in the 1990s, the project no longer aligned with current questions and analytic desiderata  in women’s history and DH. Fading memories and incomplete documentation, finally, added to the problems. 

The dilemma faced was, put bluntly, whether the database was worth reviving.  How to fix the technical issues, i.e. deprecated software and inappropriate structure? How to respond to changes in the field(s), new questions, concepts, and tools? How to ensure the database’s continuing sustainability? And just as importantly, what should be its aim? Was it to simply restore the database, i.e. to ‘implement remedial actions that “update” the project to new contexts to preserve its function and role (e.g. extending its data structure to address new questions)’ (Antonini et al. 2020)? Or was it rather to repurpose the database, i.e. ‘implement actions that rethink the value of the project by finding it new purposes, functions and roles in new contexts (e.g. defining new questions to be addressed with its existing data)’ (Antonini et al. 2020)? Or should we, instead, abandon it altogether, making the data available in a repository ‘in a suitable open casket, ready for reanimation or reuse’ (Rockwell et al. 2014)? 

Rebuilding the Database: New Technologies, New Data Model

Because the database was still used by the audience it was designed to serve, we were not yet ready for the last option. At this point, DARIAH proved crucial. Because of our DARIAH status as a functioning working group, DARIAH provided a platform from which to approach institutions to find new hosting for the project. This happily paid off, as the Radboud Institute for Culture and History, then directed by feminist scholar Liedeke Plate, generously stepped in, guaranteeing necessary funding and technical support to rebuild the database  at the institute’s Humanities Lab. This meant a migration to a fourth institution, but also an opportunity to redesign the software with the above-mentioned issues in mind. 

In 2021 and 2022, Working Group co-Chair Amelia Sanz and I hammered out a new data model, fundamentally rethinking the database structure. In doing this, we drew inspiration from the richly granular data model created by Marie-Louise Coolahan’s team for the  RECIRC database, one of the offshoots of Women Writers. Further insights came from examining other, innovative databases on women writers, such as the Women’s Print History Project. These then fed into our new data model, for example through the addition of new fields such as Edition and Circulation. In this way, interoperability with other databases was built into the new data model from the outset, anticipating a future, vast ecosystem of interlinked databases. The new data model thus encompassed new ontologies to reflect research questions that had emerged in our field in the intervening decades since Women Writers had initially been conceived.

The database was then rebuilt  by developers Micha Hulsbosch and Dennis Bor. The data was migrated from the previous SORL platform to a more appropriate SQL structure, using a Django database framework. To emphasize that this was not only a new iteration of the database, but a fundamentally reconceived tool, we gave it a new name: SHEWROTE (Studying Historical Early Women’s Reception: Oeuvres, Texts, Engagements).