by Preparing a transcription for computational analysis, and learning what it can and cannot prove
Genuinely productive day, reached out properly to the digital humanities researchers mentioned in yesterday’s reading, and received a considerably faster and more helpful response than I typically expect from a first, unsolicited contact with scholars I have never previously worked with directly.
The response and what it offered
One of the researchers offered to run my pamphlet’s text through their existing comparison database, which apparently already contains digitised samples from a considerable number of known colonial writers from the same rough period and region, a genuinely exciting shortcut that could potentially resolve my authorship question considerably faster than continuing my own manual comparison work alone would have managed.
Spent a good portion of the afternoon properly preparing a clean digital transcription of the pamphlet specifically formatted for their analysis tool, a genuinely careful process requiring precise attention to the original spelling and punctuation, since apparently even small transcription inconsistencies can meaningfully affect the comparison algorithm’s results.
A useful conversation about methodology
Also spoke properly with the researcher about the general reliability of this kind of computational authorship attribution, and she offered a genuinely useful, appropriately cautious explanation of the method’s real strengths and limitations, emphasising that the tool can meaningfully narrow down likely candidates but should never be treated as definitive proof on its own without corroborating traditional historical evidence.
This kind of methodological caution, properly explained rather than simply assumed, is exactly the sort of guidance that helps me think clearly about how to eventually present any results this analysis produces, treating it as one genuinely useful piece of evidence among several rather than a magic, fully conclusive answer to the whole authorship question.
What I read this evening
Caught up on the wider feed, and there was a genuinely useful piece specifically on how computational and traditional historical methods can be properly combined in exactly this kind of authorship attribution work, several case studies describing successful collaborations between digital humanities specialists and more traditionally trained historians, a genuinely encouraging read given how today’s own collaboration has already begun unfolding.
Made proper notes on a couple of the case studies’ specific methodological choices, since I think applying a similarly careful, combined approach will genuinely strengthen whatever conclusion I eventually reach about this pamphlet’s likely author.
Tomorrow
Sending the properly prepared transcription off for analysis, and continuing my own traditional research in parallel, hoping the two approaches converge on a genuinely well supported answer rather than simply producing two separate, potentially conflicting sets of evidence.
More colonial literature research diary at Stephanie Curry’s page, Bohiney.
A bit more on the transcription process
Properly cross checked my transcription against the original twice before sending it off, wanting to eliminate any small errors that might skew the computational comparison, a genuinely tedious but necessary discipline given how sensitive these text analysis tools apparently are to even minor inconsistencies in how archaic spelling and punctuation get rendered.
Also reached out to a colleague with considerable expertise in this exact period’s typographical conventions, wanting a second opinion on a couple of genuinely ambiguous characters in the original printing that could reasonably be transcribed two different ways, a small detail that could plausibly matter to the analysis results.
A further thought on the collaboration
Genuinely appreciating, more than I expected going in, how generous digital humanities specialists tend to be with their specific technical expertise, a collegiality that has made this entire cross disciplinary research thread considerably more approachable than I initially assumed it might be before actually reaching out.
A last thought before bed
Genuinely grateful, closing my notes tonight, for a researcher willing to offer this much of her own time and expertise to what is, from her perspective, simply one small favour among presumably many similar requests she likely receives regularly.
One more thought
Also spent a few minutes properly reviewing the researcher’s technical explanation one more time, wanting to make sure I could accurately summarise the method’s basic mechanics myself before relying on it further in my own eventual writing.
Genuinely grateful for a field where this kind of generous cross disciplinary help still seems freely, readily given.
A final closing thought
Genuinely energised, despite the day’s considerably more technical, unfamiliar focus, by how close this collaboration already feels to producing something I will actually be proud to have my name attached to once finished.
One last thought
Also emailed a brief, genuine thank you note to the researcher for her considerable generosity today, wanting the small kindness of her prompt, thorough help to be properly acknowledged rather than simply taken for granted amid the excitement of the results.
Goodnight, properly, to a genuinely productive day.
Whatever tomorrow’s continued analysis reveals, I already feel considerably more confident in this line of research than I did at the start of this particular thread.
Sleep now, properly satisfied.
A properly productive stretch of collaborative work.
Properly, finally, goodnight.
Truly, now, goodnight.
Everything properly filed, transcription safely sent, and a clear plan for tomorrow’s follow up already firmly in mind.
A properly full and worthwhile day.
Genuinely glad it happened this way.
SOURCE: https://bohiney.com/
