It was a cold, sunny, winter morning, and just like every other morning I was standing on a platform of Giessen’s train station. But this time I wasn’t headed to the IEG in Mainz, where I work, but to Innsbruck, Austria. No, the anticipation I felt wasn’t for an exciting skiing weekend on the Austrian Alps (I can’t ski, anyway!), but for the Transkribus User Conference 2020. How can one be excited for a conference on a Handwriting Text Recognition Software, you ask? Well, as a cultural historian, a humanist, I find it always intriguing when I get to deal with something “technical”, something “scientific”. Perhaps, that is one of the reasons why I took on a research project in Jewish Studies, my original field, which uses digital methods.
You can read more about my research project here. For now, it will suffice to say that I am analyzing 19th-century Italian printed prayer books and utilizing them as historical sources. Basically, in spite of being canonical texts that were more or less crystallized by the nineteenth century, prayer books were nonetheless often reprinted in many different editions, with minimal changes. I’m exactly looking for these small changes in the different editions, in order to study them and offer an interpretation of their historical meaning. Considering that my corpus consists of about 150 books with an average of 400 pages each, one can easily grasp why this quest for discerning differences needs to employ a digital methodology. Luckily enough, nowadays there are several options I can choose from, in terms of software. After a few attempts, in collaboration with our fantastic Digital Humanities Lab at the IEG, I now opted to try Transkribus. However, I am still in the experimental phase and there is no certainty that this will be my final choice.
This leads me back to Innsbruck. I decided to take part in this conference precisely because I needed to get more acquainted with existing tools. As I said, I’m a cultural historian. A large hall of the beautiful Faculty of Catholic Theology welcomed me on the first day of this two-day workshop. This time I was only a listener. I could sit back, relax, and learn. The Conference was organized in a smart and useful way. The two days were divided in morning and afternoon sessions: the mornings were devoted to general presentations about Transkribus and its newest features and to more specific insights into current projects utilizing this software. The exemplary projects presented were mostly large digitization and transcription enterprises carried out by libraries and archives throughout Europe.
The most interesting sessions for me were the afternoon ones. They consisted of specific workshops with hand-on sessions that dealt with practical aspects of Transkribus. I was particularly interested in two of them: 1) the Workshop for beginners; 2) the Workshop for right-to-left writings, which of course included Hebrew. I found both workshops very useful and interesting for a beginner like me. One has to be aware, however, that, if starting from scratch, these Workshops can only give you an initial familiarity with the software, they do not make you an expert user. For that, you need time and practice.
As regards my specific research needs, I learnt some rather useful things. 1) Although I am dealing with printed material, Transkribus can be an optimal tool, in spite of the fact that it is originally programmed for hand-written texts, because it can reach an even higher reliability when applied to printed texts. According to some, it can reach a 98% transcription accuracy. Furthermore, nineteenth-century Jewish prayer books, although printed, often included the so-called Rashi script, a type of script that resembles hand-written Hebrew, thus requiring higher accuracy. 2) As I have learnt, Transkribus can very well be applied to right-to-left languages, including Hebrew. 3) Users have already shared on Transkribus some tested models applied to Hebrew texts, which I could partly re-use. 4) It is a 100% open source. Whether these characteristics will be sufficient to make Transkribus “the winner”, remains to be seen. I mainly need to verify its functionality in terms of text comparison, as I do not need to transcribe entire texts, but rather to compare specific parts of several texts.
All in all, it was a great experience, both useful in research terms and fun. What more could you wish for? Two very productive workdays immersed in a magical winter scenario!
Alessandro Grazi is a Research Associate at the Leibniz Institute of European History, Mainz. He’s currently working on a research project called “Minhag Italia: Variations of Jewishness in the 19th Century reflected in Italian prayer books. A digital analysis”. When he’s not travelling or playing with his daughter, he enjoys reflecting on modern Jewish history.
Cite this article as: Alessandro Grazi, "Initial familiarity with Transkribus: The User Conference 2020," in Digital Humanities Lab, 29/05/2020, https://dhlab.hypotheses.org/?p=1441.
Featured Image: Innsbruck, A. Grazi
OpenEdition schlägt Ihnen vor, diesen Beitrag wie folgt zu zitieren:
Thorsten Wübbena (29. Mai 2020). Initial familiarity with Transkribus: The User Conference 2020. Digital Humanities Lab. Abgerufen am 5. Oktober 2024 von https://doi.org/10.58079/nl5a