Encoding the haunting of an object catalogue: on the potential of digital technologies to perpetuate or subvert the silence and bias of the early-modern archive

Abstract

The subjectivities that shape data collection and management have received extensive criticism, especially with regards to the digitization projects and digital archives of galleries, libraries, archives and museums (GLAM institutions). The role of digital methods for recovering data absences is increasingly receiving attention too. Conceptualizing the absence of non-hegemonic individuals from the catalogues of Sir Hans Sloane as an instance of textual haunting, this article will ask: to what extent do data-driven approaches further entrench archival absences and silences? Can digital approaches be used to highlight or recover absent data? This article will give a decisive overview of relevant literature and projects so as to examine how digital tools are being realigned to recover, or more modestly acknowledge, the vast, undocumented network of individuals who have been omitted from canonical histories. Drawing on the example of Sloane, this article will reiterate the importance of a more rigorous ethics of digital practice, and propose recommendations for the management and representation of historical data, so cultural heritage institutions and digital humanists may better inform users of the absences and subjectivities that shape digital datasets and archives. This article is built on a comprehensive survey of digital humanities’ current algorithmic approaches to absence and bias. It also presents reflections on how we, the authors, grappled with unforeseen questions of absence and bias during a Leverhulme-funded collaboration between the British Museum and University College London (UCL), entitled ‘Enlightenment Architectures: Sir Hans Sloane’s Catalogues of his collections’.

What do they make us see: a comparative study of cultural bias in online databases of two large museums

“…Users' expectations to see a variety of contexts are sustained by the quality of museum information, a diversity of art objects selected for the first pages of the user interface, and by the abovementioned missions of the two museums aimed at multi-cultural representation. However, the composition and distribution of museum collections as well as silences and omissions in the collections' data, may create cultural bias (Bode, 2020;Kizhner et al, 2021;Ortolja-Baird and Nyhan, 2021;Boyd-Davis et al, 2021).…”

“…Bias can exist at any stage of information retrieval, from the stage of selecting data for the online system and dataset, to the stage at which the information retrieval algorithm is selected, and through its presentation in the online systems (Mowshowitz and Kawaguchi, 2002). Although most linguistic corpora or collections of cultural heritage objects are inherently biased (Chomsky, 1957;Clear, 1992;Bode, 2020;Bagga and Piper, 2020), this is rarely explicitly announced by the editors of digital collections or platforms (Hajibayova, 2018;Bode, 2020;Kizhner et al, 2021;Ortolja-Baird and Nyhan, 2021). As a result, trustful sources of documented information, such as cultural heritage institutions, face epistemological problems as knowledge creators and disseminators (Boyd-Davis et al, 2021).…”

“…Bias assessment in online systems dates back to the time when Mowshowitz and Kawaguchi (2002) demonstrated the importance of having a representative and balanced collection of results when retrieving information online. One of many reasons for bias in online digital collections of cultural heritage relates to colonial practices prevailing in displaying or publishing important records of the past (Kizhner et al, 2021;Ortolja-Baird and Cultural bias in online databases Nyhan, 2021). In this context, the question is how can the level of bias be measured?…”

“…In a similar vein, although for a very different collection, Ortolja-Baird and Nyhan (2021) compare place and person names, showing the rates at which place names for the countries of Henry Sloane's travels correlate with Western names as sources of objects, and the depth of absence of indigenous people from objects' descriptions. The emphasis on omissions in place and person names shows the importance of assessing geographical distribution as a simple but an essential step in measuring bias (Ortolja-Baird and Nyhan, 2021). In a different field, Raja et al (2022) find sampling biases in paleontological knowledge production.…”

“…However, the composition and distribution of museum collections as well as silences and omissions in the collections' data, may create cultural bias (Bode, 2020; Kizhner et al. , 2021; Ortolja-Baird and Nyhan, 2021; Boyd-Davis et al. , 2021).…”