There is a recent thread on the archives listserv that talks about transcriptions – specifically for small projects or those that have little financial support. There is even a case in which there is no easy OCR answer due to the state of the digitized microfilm records.
One of the suggestions was to use some combination of human effort to read the documents – either into a program that would transcribe them, or to another human who would do the typing. It made me wonder what it would look like to make a place online where people who wanted to could volunteer their transcription time. In the case where the records are already digitized and viewable, this seems like an interesting approach.
Something like this already exists for the genealogy world over at the USGenWeb Archives Project. They have a long list of different projects listed here. Though the interface is a bit confusing, the spirit of the effort is clear – many hands make light work. Precious genealogical resources can be digitized, transcribed and added to this archive to support the research of many by anyone – anywhere in the world.
Of course in the case of transcribing archival records there are challenges to be overcome. How do you validate what is transcribed? How do you provide guidance and training for people working from anywhere in the world? If I have figured out that a particular shape is a capital S in a specific set of documents, that could help me (or an OCR program) as I progress through the documents, but if I only see one page from a series – I will have to puzzle through that one page without the support of my past experience. Perhaps that would encourage people to keep helping with a specific set of records? Maybe you give people a few sample pages with validated translations to practice with? And many records won’t be that hard to read – easy for a human’s eye but still a challenge for an OCR program.
The optimist in me hopes that it could be a tempting task for those who want to volunteer but don’t have time to come in during the normal working day. Transcribing digitized records can be done in the middle of the night in your pajamas from anywhere in the world. Talk about increasing your pool of possible volunteers! I would think that it could even be an interesting project for high school and college students – a chance to work with primary sources. With careful design, I can even imagine providing an option to select from a preordained set of subjects or tags (or in Folksonomy friendly environment, the option to add any tags that the transcriber deems appropriate) – though that may be another topic worthy of its own exploration independent of transcription.
The initial investment for a project like this would come from building a framework to support a distributed group of volunteers. You would need an easy way to serve up a record or group of records to a volunteer and prevent duplication of effort – but this is an old problem with good solutions from the configuration management world of software development and other collaboration work environments.
It makes a nice picture in my mind – a slow, but steady, team effort to transcribe collections like the Colorado River Bed Case (2,125 pages of digitized microfilm at the University of Utah’s J. Willard Marriott Library) – mostly done from people’s homes on their personal computers in the middle of the night. A central website for managing digitized archival transcriptions could give the research community the ability to vote on the next collection that warrants attention. Admit it – you would type a page or two yourself, wouldn’t you?
- reCAPTCHA: crowdsourcing transcription comes to life
- Squirl.info – an interesting option for putting collections online
- THATCamp 2008: Crowdsourced Transcription and Collaborative Annotation
- Harnessing The Power of We: Transcription, Acquisition and Tagging