Transkribus in the Classroom. Students Find Their Way to Manuscripts and Handwritten Text Recognition (HTR) #dhiha8

Pia Eckhart (University of Freiburg) and Ina Serif (University of Basel)

At University Library Basel, having a look at Cod. A IX 23a (picture: Pia Eckhart, pe).

In the fall of 2018 a workshop on the possibilities of integrating digital tools and computational methods into historical research was held at the chair for Medieval History II at the University of Freiburg. Over coffee, the idea came up to organize a hands-on Transkribus seminar for students. Our interest in this tool, which offers an environment for the (semi)automated transcription of handwritten and printed documents, had been primarily research driven so far: Ina had already used Transkribus in several projects, and Pia was in the process of evaluating the role the software could play in a new research project. Now we discovered our common interest: Ina was about to publish a Transkribus tutorial and Pia was planning to integrate the tool into her next paleography course, so we decided to put our heads together and come up with a joined plan.

Goals

Our goal was to show students how to take a first step into the digital realm of history, without having to learn how to code or pay for expensive software, and to teach them how to integrate different tools, resources and methods into their own research. At the end of the course, the students are expected to be able to choose a source and enhance it with transcribed text in the Transkribus environment, using the software’s different functionalities, such as applying an existing text recognition model and training it with new material. Moreover, they should be able to reflect on the expected outcome, evaluate the actual results, and master common difficulties typically arising in the process of transcribing a source, be it in an analog or digital environment.

Screenshot of a transcribed page of Freiburg, Universitätsbibliothek, Hs. 497-3, working with Transkribus.

The syllabus

In April 2019 we proposed to the students the following work schedule: Pick out two manuscripts from Freiburg University Library, train our own HTR data models step by step utilizing Ina’s Transkribus tutorial as a guideline, then scan manuscripts with similar handwriting at Basel University Library using ScanTents and a scan app for smartphones (both side-products of Transkribus), and test our own models on these manuscripts. To get credit points, the students will publish several blog articles: a Transkribus trial run from the students’ point of view for the Mittelalter – Medieval Studies blog, and two articles on the Basel manuscripts for the Basel University Library blog.

Digitising a manuscript with the ScanTent and the DocScan app at University Library Basel (Basel, Cod. A IX 23a; picture: pe).

Student responses

We were happily surprised when the fourteen students at Bachelor, Master and PhD level all got on board with our proposal without hesitation. Some of them had already worked with Transkribus for their own research projects, others engaged with the challenges of studying manuscripts for the first time. Whereas Pia and Ina both specialize in premodern history, there were also some Kurrent lovers in the classroom. The heterogeneity of the group regarding experience and interest should ultimately work to the project’s advantage, since the students mostly work in teams and learn from each other.

Work in progress

Six weeks in, things look promising. In addition to Transkribus, we introduced Ad fontes, a well established online learning program which not only offers paleographical tutorials combined with exercises, but also provides useful resources and interesting links for the study of historical documents. The students also got to know several online repositories for medieval and modern manuscripts and early prints as well as databases that provide an overview over digitized sources.[1] Using the Ad fontes e-learning program, we showed the beginners how to develop their paleographic skills and they fearlessly went ahead with it. The group then picked two digitized manuscripts from Freiburg University Library. Our choices, based on the students interests, were a German Kurrent from ca. 1800, Hs. 497-3, and a gothic bookhand from the later 15th century, Hs. 1500,13. The group worked out individual transcription rules for both scripts and is now in the process of producing correct diplomatic transcription as a ground truth for model training.

Chosen manuscripts from University Library Freiburg: Hs. 497-3 (on the left) and Hs. 1500,13 (licence: CC BY-SA).

We took two field trips to Basel University Library to work with ScanTents and the DocScan app on a sample of manuscripts there, A VII 2, A IX 23a and A XIII 97. In small groups, the students established a workflow while scanning the manuscripts, and took notes on bugs or practical issues of the app for future feature requests.

Digitising a manuscript with the ScanTent and the help of scan fingers at Basel University Library (Basel, Cod. A XIII 97; picture: pe).
Upload of the pictures taken with the DocScan app into Transkribus (picture: pe).

The last part of the course will be devoted to writing and editing the blog articles. We hope that this will enable the students to gather first experiences with scholarly blogging and publishing one’s findings in general.

Closing reflections

The seminar has already succeeded in bringing together students and institutions: the University Libraries Basel and Freiburg showed interest early on and provided substantial assistance for the project. Institutions and students can both profit from hands-on seminars like this: Libraries and archives get insights into academic teaching and current research, whereas students acquaint themselves with the holdings of different institutions, and – probably most importantly – realize that they are welcome to work with these sources.

Furthermore, through the practical application of programs and apps, students become skilled and competent users and even begin to see themselves as an active part of a scholarly community that helps improving research software. By being (so far) free to use as well as platform independent, and by providing a graphical user interface as well as an active and growing user community and direct support, programs like Transkribus have the potential to close the gap between analog and digital history, enabling users to enhance a digitized source on different levels and in different ways, depending on their research goals and technical knowledge.[2]


Suggested citation: Pia Eckhart/Ina Serif: Transkribus in the Classroom. Students Find Their Way to Manuscripts and Handwritten Text Recognition (HTR) #dhiha8, in: Digital Humanities am DHIP, 07/06/2019, https://dhdhi.hypotheses.org/5985.

[1] E.g. e-codices, e-manuscripta, e-rara, Fragmentarium, Handschriftencensus; a good and up-to-date-overview is provided by Klaus Graf, https://archivalia.hypotheses.org/5487. All links were checked on June 5, 2019.

[2] We would like to thank Monika Studer at Basel University Library, Noah Regenass at UB Basel Blog, Hannah Busch at Mittelalter – Medieval Studies Blog, Marcus Schröter at Freiburg University Library and Tobias Hodel, University Zurich/State Archives Zurich.

Ina Serif

Nach dem Studium der Fächer Geschichte, Germanistik und Italienisch an der Universität Freiburg und der Università degli Studi Roma Tre war sie Wissenschaftliche Mitarbeiterin am Lehrstuhl für Mittelalterliche Geschichte II an der Universität Freiburg, wo sie ihre Doktorarbeit über städtische Geschichtsschreibung und deren Vernetzung, Aneignung und (Re-)Funktionalisierung verfasst hat. Zurzeit ist sie Wissenschaftliche Mitarbeiterin am Lehrstuhl für Geschichte der Renaissance und Frühe Neuzeit am Departement Geschichte der Universität Basel und im Projekt Stadt.Geschichte.Basel.

More Posts

Follow Me:
Twitter


Das könnte Dich auch interessieren …

1 Antwort

  1. 27/06/2019

    […] Pia Eckhart, Ina Serif, Transkribus in the Classroom. Students Find Their Way to Manuscripts and Handwritten Text Recognition (HTR) #dhiha8, in: Digital Humanities am DHIP, June 7, 2019, https://dhdhi.hypotheses.org/5985. […]

Schreibe einen Kommentar

Deine E-Mail-Adresse wird nicht veröffentlicht.

Diese Website verwendet Akismet, um Spam zu reduzieren. Erfahre mehr darüber, wie deine Kommentardaten verarbeitet werden.