About eight months ago, I spent a long, blustery afternoon developing a new work-flow for my research. I had hoped, naively, that this detailed methodology would make my research more efficient and, in large measure, standardise the data I was collecting. For many weeks, I diligently followed this predetermined path. For many months thereafter, I struggled and staggered along, attributing my slow progress to mounting teaching pressures and upcoming research and conference trips. As summer dawned, teaching dispersed, and conferences faded, I found that my ascent had halted completely. My well crafted methodology was simply not fit for purpose.
Upon reflection and analysis, there are several key reasons why this work-flow (as well as my ad hoc adjustments) were flawed. Although my system was designed to promote efficiency, guiding a single article through the work-flow took far longer than I had originally anticipated. Despite maintaining an above-average typing speed, I soon found that my estimate of 500 words per article was grossly inaccurate. Although articles of 400-600 words are common, the average article on Australian settlement (my main focus for the spring) was well over 1000 words, and many approached 2500.
This greater length has had several knock-on effects. First, the initial transcriptions and corrections take twice if not four times as long as originally expected. Second, this extended length is usually in the form of breadth rather than depth. With such eclectic content, sufficient and precise tagging become much more time-consuming. Yet, these additional outlays of time and effort were designed to make the overall process more efficient by removing the need to re-read articles during a secondary ‘cataloguing’ process. In theory. In fact, shepherding a single article through my work-flow has proven devastatingly time-consuming owing to five nasty little words: Your session has timed out.
Although the miracles of digitisation have made this project possible by allowing remote access to thousands of pages of periodical material, there is a temporal price to be paid. Even very short articles take far longer to transcribe and catalogue than the British Library and other subscription-based digital archives deem appropriate. After a seemingly brief period of idleness, the system closes my session, forcing me to return to the institutional log-in page, re-enter my details, re-submit my sometimes very specific search parameters, and scroll through pages of results in order pick up where I had left off. This occurs not only when transcribing but sometimes while simply obtaining a new cup of tea, greatly and unnecessarily lengthening my research time and frustration.
In order to compensate for this problem (being unable to efficiently ‘trick’ the library into keeping my session open), I returned to my former habit of assembly-line transcription. I read through an entire title of a publication (within the relevant chronological period) making digital copies as I found relevant pieces. Afterwards (sometimes weeks later), I went through the transcription and cataloguing process.
Unfortunately, owing to pressing publication and speaking deadlines, full transcription had to be abandoned in favour of selective transcription of those articles most relevant to upcoming pieces. This meant quickly (re-)reading through my pieces, sorting them into relevant stacks, and attaching rudimentary tags (fit only for my immediate needs). These pieces were then transcribed and run through my discourse analysis work-flow.
This short-circuited the workflow and several new problems manifested themselves, namely that I failed to attach a complete set of meta data to some articles. Several had only partial dates or were missing page numbers while others were truncated, losing the final line or column of text. This meant re-trawling through the electronic database (whose OCR search function was often ill-equipped for finding specific phrases rather than words) in order to obtain the missing information. Moreover, I now have several stacks of half-catalogued, half-transcribed pieces, many of which may have added additional light and nuance to recently presented works.
I am thus left with a methodological dilemma. The handicraft system was unworkable with pay-wall archives. The assembly-line method was massively inefficient and destructive to scholarly rigour. The answer, of course, is some form of hybrid. Yet, this must be devised with reflection if the inefficiencies of my ad hoc system are to be removed. Moreover, it must be flexible for a variety of digital and hard-copy environments. Will such a system be devised in time for next week post? We can only hope.
*Image courtesy of urbangarden