Material
Building upon our previous work training the first public Syriac HTR models on Transkribus: the Vienna Syriac Gospels Serto model and the Syriac New Testament of Jerusalem (Estrangelo - Melkite) model, which reached an impressive 3.53% Character Error Rate, this year the Syriac group will expand into new manuscript corpora. We will focus on one of the ancient New Testament Syriac manuscripts from the British Library collection.
We will work on the British Library Syriac manuscript Add MS 14470, a 5th or 6th-century Peshitta New Testament written in an elegant Edessene Estrangelo script. The manuscript likely originated in Syria or Mesopotamia before reaching the famous Syrian Monastery (Dayr al-Suryān) in the Nitrian Desert (Wadi al-Natrun, Egypt) around the 10th century.
Tasks for the Participants
Prepare and refine line segmentation for assigned Syriac manuscript folios in Transkribus.
Engage in a two-stage quality assurance workflow (manually proofreading and upgrading page status under editorial supervision) to produce high-precision Ground Truth.
Contribute to building and refining open-access Syriac HTR models and datasets for the international research community.
Transcription Tools and Guidelines
The group will work in Transkribus, optimizing workflows for Syriac's right-to-left (RTL) script. We follow a diplomatic transcription protocol: preserving letterforms, word boundaries, and structural diacritics (such as Syome and homograph-distinguishing dots) as attested in the manuscripts, while establishing standardized handling for line segmentation and punctuation.
Requirements
- Some knowledge of Syriac palaeography (specifically Estrangelo) as you will be involved in transcribing and correcting handwritten texts.
- Sufficient knowledge of the Syriac language to identify and correct automatic transcriptions, recognize Syriac words, and understand basic grammatical structures.
- No prior experience with HTR or Transkribus is required.
Supervisors
Ephrem Aboud Ishac
Institute for Medieval Research, Austrian Academy of Sciences / Central European University
Christine Roughan
Center for Digital Humanities, Princeton University