what's red and invisible? no tomatoes

May 10, 2023 7:44 AM

Nothing survives transcription, nothing doesn’t survive transcription: a talk (or the text thereof, and, yes yes, you're very clever, now shut up and read it) about the fundamental inability of transcription to capture that which it is transcribing, by Mefi's Own Allison Parrish.
posted by cortex (7 comments total)

Wow, that was super interesting! Of course nothing doesn't survive transcription!

Thanks so much for posting. I never would have read this otherwise.
posted by hippybear at 8:24 AM on May 10, 2023

THIS TALK IS SO GOOD. I am biased because Allison is a friend of mine but still!
posted by brainwane at 8:57 AM on May 10, 2023 [1 favorite]

this is an excellent, thought-provoking essay (with bonus Emily Dickinson-related art & original MSS.)
thanks for posting!
posted by chavenet at 10:16 AM on May 10, 2023

Terrific post! William Blake is another poet whose work cannot be transcribed, as not only are the poems inextricable from the art, but also poems in the "same" book look radically different from copy to copy (e.g., "The Lamb"--click on the matrix view).
posted by thomas j wise at 11:25 AM on May 10, 2023 [1 favorite]

mmmm.
posted by rrrrrrrrrt at 12:24 PM on May 10, 2023

Really interesting text! This is a big issue for historians because of the ever-present problem of archive digitization. You'd think it'd be a really straightforward question, but it brings in a lot of factors most people don't consider. How high should the quality be on stored archival files, for instance? Do you optimize for loading times or accuracy? How do you replicate the organization of a file in a digital format? I've recently been working with a lot of documents archived in the nineteenth century in which there are all kinds of sheafs and bundles that may represent different levels of organization or may not, or levels that were once relevant but no longer are.

For eighteenth century documents, a classic problem I've been having recently: someone sends someone a letter (one full page folded in half hamburger-style, with writing on all four of the resulting pages). The letter contains two enclosures, such that the last two pages of the letter come after the enclosures. Do you digitize the entire letter, then the enclosures? In that case it's not necessarily clear they're being enclosed. Do you digitize the first half of the letter, the enclosures, and then the second half? In that case the textual continuity between the two halves risks getting lost, especially if the files are split up somehow. In either case, you want to reproduce not just the page with the text but also the surrounding surface to get a sense of the relative sizes of the letter and the enclosures, and you want to use color to convey differences in the shade and inks of the paper. Unfortunately, many digitized archives--including the ones I make quick and dirty for my own research--are optimized for readability and portability, which often means removing background and color.

Robert Darnton has another classic example--I think it was a colleague of his who was studying plague quarantines and would sniff the pages of documents to see if they'd been treated with a sanitizing compound! Digitize THAT!
posted by derrinyet at 7:21 AM on May 12, 2023 [2 favorites]

Digitize THAT!

I mean, a spectrometer will do this. But I get what you mean.
posted by jaduncan at 1:36 PM on May 14, 2023