Richard Hadden: Digital scholarly editing and mass digitisation

Hi, I’m Richard Hadden (ESR1). I’ve just started my work as a DiXiT fellow at the National University of Ireland Maynooth (NUIM), with Susan Schreibman. I previously studied for an MA in Electronic Communication & Publishing at UCL (London), and a BA in French, Spanish and Italian at Durham (latter languages need a little brushing up on, but keen to practise). I’ve also dabbled in some programming and web design, and my head works in the syntax of Ruby.

My research for the next three years will look at the differing methodologies of digital scholarly editing and mass digitisation (actually, I think, the rather important philosophical differences that derive from these processes). If scholarly editing sets out to make a useful edition from documents, mass digitisation seems to be digitisation for its own sake, with very little forethought (or, some might say, rigour): in the case of Google Books, this feels very much like the means (and the end). However, mass digitisation, as I hope to show through my research, can be a valid approach to what may — or may not be — termed scholarly editing.

My research will be primarily the Letters of 1916 project, now hosted by NUIM. A crowd-sourced and crowd-transcribed project, Letters sets out to digitise letters written or sent within Dublin in 1916, the year of the Easter Rising. Whilst a relatively small-scale project, its wide scope has led to the adoption of methodologies that would be considered primarily as mass-digitisation: particularly, a kind of catch–what–you–can–and–digitise–it–first approach, without the kind of specific rationale that would underlie the creation of a scholarly edition. If, as Hans Walter Gabler says of scholarly editing, “the horse of the document [should be placed] properly before the cart of its eventually emerging text,” in the Letters project, this is being deliberately ignored.

The obvious reasons for this are the scale of the project; the inherent heterogeneity of the documents (when manually and deliberately choosing documents for an edition, there is a tendency towards homogenisation: after all, they must sit comfortably together in the edition); and, above all, the unknown: who knows what letters are in attics in biscuit tins? The purpose of mass digitisation, from the point of my research, is therefore not mass digitisation for its own sake, but digitisation and analysis as a means to rationalise a large and diverse set of textual evidence.

But — here is my research — how can this be reconciled with the tenets of scholarly editing? Firstly, by “transcribing first and asking questions later”, the fundamental link to the original document has been lost (Gabler’s horse has been loaded onto the cart, and given a good shove down a hill). Secondly, with such a large collation and the difficulties of reading and making sense of it all, what ‘scholarship’ can be done on it? How far ‘back’ towards the document and the edition can algorithmic analysis and machine learning, concept analysis and automated mark-up, take us? And is there an end result that can, without too much semantic wrangling, be called an edition?

Welcome to my PhD.

A slightly more crazed version of this post exists here. Thanks to fellow DiXiT fellows for their interest and suggestions and piscine tolerance.

For any other suggestions, tips, advice, constructive criticism, grounding in statistical theory: please contact at @oculardexterity.