Old Books and Digital Publishing: Eighteenth-Century Collections Online

Abstract

with his point that the term 'texts' should go beyond the printed text to encompass the very broadest set of human communication media and evenmost vital for us -'computer-stored information'; there is, he argues, 'no evading the challenge which those new forms have created' (McKenzie, 1999: 13). The discipline of bibliography 'studies texts as recorded forms, and the processes of their transmission, including their production and reception' (12). Perhaps the strongest argument for my history is that a 'sociology of texts' should allow 'us to describe not only the technical but the social processes of their transmission', and it 'directs us to consider the human motives and interactions which texts involve at every stage of their production, transmission, and consumption. It alerts us to the role of institutions, and their own complex structures, in affecting the forms of social discourse, past and present' (McKenzie, 1999: 13, 15). Sociocultural forces, institutions, technology, and human agency all play their part in this history.Eighteenth Century Collections Onlineor any digital entityis not a static or an unchanging entity: it has a history. The rapidity with which commercial publishing technology supersedes older versions of itself has meant that some circumstances of its development are now obscure and others are irrecoverable. Tellingly, a part of ECCO will become invisible from 2020 when its original interface is scheduled to be turned off: it will literally be history. So this book is partly an act of recovery. The third chapter, 'Beginnings', turns to the development of ECCO itself. In the first section, I examine the immediate contexts that shaped how ECCO was to work and to be sold. It was decisively influenced by the downward movement of the academic publishing market and the emerging so-called disruptive technologies in the 1990s (Bower & Christensen, 1995). I focus on the techno-commercial choices facing Gale by illustrating contemporary digital resources created by two of its key commercial competitors in academic publishing of the time: Chadwyck-Healey and ProQuest. In addition, some aspects of how ECCO works 'under the hood' arein common with many digital productssimply invisible to the public. The chapter goes on to explain Gale's digitisation of the microfilm collection, discussing the problems created by the use of optical character recognition (OCR) software to

Reflections

“…24 My book did go beyond the screen to discuss the Anglocentric biases of ECCO and the wider debate about the whiteness of digital archives highlighted in work by Roopika Risam, Amy Earhart, Tim Hitchcock, Adeline Koh, and Martha Nell Smith. 25 However, it underplayed ECCO's place within the neocolonial division of labour in the global information business and what Alex Gil recently called the "uber whiteness" of the leadership of "knowledge cartels" such as EBSCO and ProQuest, and which is equally visible in the make-up of Gale's own executive leadership. 26…”

Recent approaches to paratext studies in eighteenth‐century literature

“…Addressing more contemporary issues of paratextual presentation, anthologies such as Siemens and Schreibman (2007) and Burrows and Roe (2020) have begun to explore how textual metadata is recorded and stored within digital repositories. For eighteenth‐century studies in particular, these ideas are crystallised by Gregg (2020), who provides a timely and detailed history of the development of the digital archive and the decisions which influence how literary texts are documented and reproduced for the age of the internet. The considerable potential for archival metadata to shape our understanding and experience of historical texts merits scholarly attention, and Gregg makes a convincing ‘argument for recovering, reading, and researching digital resources critically’.…”

OCR quality and the resilience of algorithmic identification of linguistic register features in Eighteenth Century Collections Online

“…However, a major source of large-scale low-quality textual humanities data, particularly in fields such as historical linguistics and computational history, are texts which have been turned from scans of physical documents into machine-readable format using optical character recognition (OCR) technology. For instance, the OCR quality of Eighteenth Century Collections Online (ECCO) 1 , a collection of over 200,000 mostly English-language works published in the United Kingdom during the 18th century [see Tolonen et al, 2021] and a central resource for DH scholars [Gregg, 2020], is extremely variable. A main contributing factor of the often low OCR quality for ECCO is the OCR being run on bitonal scans of microfilms with OCR algorithms which have not been fine-tuned for eighteenth-century typefaces or trained to recognize e.g.…”