Adventures in the Kuyper archive: Column5: ‘Deciphering the illegible – Transkribus and Kuyper’
This week, a column by teaching assistants Jens van Lingen and Romée Neef. They would like to tell you more about Transkribus and what it can do.
04/12/2021 | 3:41 PM
Abraham Kuyper led a very busy life. When he wasn't writing a newspaper column, he was lecturing at the university. And alongside all his professional activities, Kuyper also managed to maintain a constant flow of correspondence. He found the time to write letters to his wife, children, friends and colleagues. In some of those letters he expressed his personal feelings. During his tour of the USA from August to December 1898, Kuyper wrote letters home complaining about his experiences. For example, on 8 October in New York he wrote that he was suffering from stomach problems again and that he was finding evening meals with American colleagues to be boring.
Often, though, Kuyper was in too much of a hurry to make sure his letters were neat and tidy – he had a boat to catch, a lecture to teach or a council of ministers to attend. As a result, his letters are often messy, difficult to follow and full of abbreviations. The longer letters, during which he probably lost his patience, can be particularly hard to decipher.
Transkribus in action
The University Library is trying to solve this problem using Transkribus. Transkribus makes it possible to decipher different writing styles automatically. To do this, you need to create a model that is able to read that person’s handwriting for itself. By adding multiple letters by the same person and in the same handwriting, and then adding the relevant transcriptions, Transkribus can train itself to read letters, numbers and words.
We are currently training a model that will help us to decipher Kuyper's letters in the future. The Abraham Kuyper archive contains almost 9000 letters written by and to Kuyper. And to this day, new and hitherto unknown letters from Kuyper are being added to the collection. By training a Kuyper model, we hope to make the Kuyper archive more accessible. This will allow more and better research into Kuyper and his correspondence in the future. Perhaps the model will work so well that we will be able to read even the most difficult of Kuyper's letters, and make some new and interesting discoveries.