LiViTo: Linguistic and Visual Features Tool for Assisted Analysis of Historic Manuscripts

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 885–890 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 885 LiViTo: linguistic and visual features tool for assisted analysis of historic manuscripts Klaus Müller1, Aleksej Tikhonov2, Roland Meyer3 1Musterfabrik Berlin/Fraunhofer IPK, Brunnenstr. 156, 10115 Berlin, Germany 2,3Department of West-Slavic Languages, Humboldt University Berlin, Unter den Linden 6, 10099 Berlin, Germany klaus.mueller@musterfabrik-fabrik.de, tikhonal@hu-berlin.de, roland.meyer@hu-berlin.de Abstract We propose a mixed methods approach to the identification of scribes and authors in handwritten documents, and present LiViTo, a software tool which combines linguistic insights and computer vision techniques in order to assist researchers in the analysis of handwritten historical documents. Our research shows that it is feasible to train neural networks for the automatic transcription of handwritten documents and to use these transcriptions as input for further learning processes. Hypotheses about scribes can be tested effectively by extracting visual handwriting features and clustering them appropriately. Methods from linguistics and from computer vision research integrate into a mixed methods system, with benefits on both sides. LiViTo was trained with historical Czech texts by 18th century immigrants to Berlin, a total of 564 pages from a corpus of about 5000 handwritten pages without indication of author or scribe. We provide an overview of the three-year development of LiViTo and an introduction into its methodology and its functions. We then present our findings concerning the corpus of Berlin Czech manuscripts and discuss possible further usage scenarios. Keywords: mixed methods, digital humanities, machine learning, linguistics, manuscripts, authorship, Czech, Slavic, Slavonic, handwriting 1. Introduction Historical manuscripts are an essential source for all humanities. In line with the apparently rising interest in everyday culture (Meyer, 2007), new research stimuli can be expected from hitherto unknown texts in small private or parochial archives. They may contain reports by personal witnesses, new information on everyday culture, and samples of everyday language. For a long time, many communities have been cherishing a purely handwritten culture that can be fully understood only by assessing personal manuscripts. The present project is devoted to exploring handwritten sources of a small, but linguistically, culturally and historically extraordinary community of migrants to 18th c. Berlin: the "Czech brethren" (aka Moravian Church, Herrnhuter Brüdergemeine), who fled from anti-Protestant persecution in the Czech lands to Saxony and Prussia. Research questions that arise in this context are: 1) Are the manuscripts originals or handwritten copies? 2) Are the originator/author and the scribe the same person? 3) How many authors and scribes worked on the manuscripts? 4) Can these authors and scribes be found in other manuscripts? 5) Are there revisions in the manuscripts and where are they? For the identification of scribes and authors in manuscripts, methods from classical linguistic analysis are combined with modern computer vision approaches, such as neural networks. This approach enhances the knowledge discovery process and knowledge representation. The idea of automatically discovering knowledge from manuscripts is a very attractive and challenging task. The main motivation for applying neural …

Content truncated to 500 words