after a well-deserved summer break in cooler climate, it was time to once again dedicate ourselves more fully to courtly duties and set off on visits of representation to various principalities.
My entire court and I traveled (fortunately not by carriage) to the imperial city of Cologne for the annual conference of the German Society for Music Research (September 11–14). There, we provided exclusive insights into the current work of the project and exchanged views with envoys from other courts. My primary and very responsible task was to represent.
Why this noble “conference” took place in Cologne, however, is quite incomprehensible from the perspective of a Viennese court pug—after all, there isn’t even a proper court there. Apparently, much has changed in recent centuries, and French was even spoken in the city at some point. Additionally, I discovered that the “stagione di carnevale” is highly valued in this city—unfortunately, not in September. I had even brought my crown especially for the occasion.
A crucial step in creating the digital edition is transcribing the sources. In our case, we’re dealing with 18th-century manuscripts in French, which are fortunately relatively readable, both for humans and machines. However, there are at least two, possibly more, different scribes with their quirks, and we’ve found that even the machine is still having some difficulties.
For machine text recognition, also known as Handwritten Text Recognition (HTR), we use the application Transkribus. Transkribus is based on artificial intelligence and allows for automated transcriptions of handwritten or printed sources (more on that here).
Wn_Mus.Hs.34580b f 142r: continuous text scribe 1Wn_Mus.Hs.34580c f. 25v: table structure schribe 2
Before we could start transcribing, we had to prepare the source for Transkribus because most of the chronicles are recorded in tables, which poses some challenges for handwriting recognition. Therefore, Christina first trained a table recognition model so that each page gets the correct layout and all lines are recognized accurately. Meanwhile, Véronique and Jakob began transcribing the continuous texts and discovered a few weaknesses in the existing language model.
Working with Transkribus is both exciting and challenging. On one hand, the software significantly eases the transcription process by using machine learning, saving us a lot of time since we don’t have to type everything manually. On the other hand, manual post-processing still requires a lot of time and care, especially for structurally more complex pages. The inconsistent use of uppercase and lowercase letters for example (due to the lack of contemporary rules) especially makes our heads spin.
beforeafterTranscription of US-CAh-MS_Thr_248/p.50 pre and post manual correction
Correcting the automatic transcriptions however, is an essential step for further working with Transkribus. Based on these corrected pages, known as “Ground Truth,” Christina can train a specific handwriting recognition model for each scribe. This improves the recognition of letters, numbers, and superscripts, meaning future transcriptions will require fewer corrections and hopefully less time.
At the end of this post, Rudie would like to have his say:
Thank you very much! Now that my court… erm… the team has taken a well-deserved summer break, I have also decided to enjoy my summer retreat and perform my representative duties by the pool.
Rudie’s holiday greetings (AI-generated)
Je vous souhaite à tous un bel été! „Le très-humble, et très obéissant serviteur”,
Why it is important to examine historical sources in person sometimes
With the rapid advancement of digitization, it is now easier than ever to access information from all over the world at the click of a button. On her first visit to the Music Collection of the ÖNB (Österreichische Nationalbibliothek) in Vienna, our project employee Véronique was initially greeted by a staff member with a friendly “You do know that the Gumpenhuber volumes are also available as scans, right?” Of course, we were aware of this, after all, Christina was already diligently trying at that time to transcribe said scans using the software Transkribus. Nevertheless, there are many arguments in favor of examining historical sources in their original form at least once, especially for a good source description (which is indispensable for a detailed edition).
As Ingeborg has already stated in part 1 (https://gudieen.hypotheses.org/284), many visual conditions cannot be correctly captured by the scan and would therefore be inevitably lost in the source description. In addition, pages are often cut off in the digital copy (due to the still stable binding of the volumes), so that whole words disappear, which are however fully captured in the original sources – a far greater problem in the Viennese volumes than in the US volumes. Véronique is still working on the collection of these incomplete passages at the moment.
(Example of an incomplete passage: Mus.Hs.34580/b, p. 38v – left: View scan / right: View original)
In addition, touching the original sources can enhance the overall understanding of the source and its context. Holding an old document in your hands allows you to appreciate its physical characteristics, such as the weight and texture of the paper, the fading of the ink, and any imperfections or damage that may have occurred over time.
The personal examination of historical sources can be a more immersive and engaging experience than digital reading and therefore offers a unique and valuable way to connect with the past. By seeing, touching, and engaging with original documents, you get an overall better understanding of the document being passed down through the hands of countless others over the years and can therefore form a tangible connection to the past. In contrast, digital copies, while convenient and accessible, can sometimes feel sterile and detached. They lack the physicality and tactility that can bring a historical source to life. While digital tools can be valuable for research and analysis, they should not replace the experience of seeing and holding original sources whenever possible.
First New Insights
Similar to the US chronicles (https://gudieen.hypotheses.org/284), the Vienna volumes also feature gilded pages and a green silk ribbon (although this is unfortunately torn off in vol. 2 and therefore no longer present).
Another surprising discovery was that vols. 2 and 3, while having a similar page count, appear to be different thicknesses when viewed visually. This difference in thickness is likely due to the variation in paper stock used in the two volumes. Upon initial examination, the paper in vol. 2 appears to be significantly thicker than that used in vol. 3.
Many thanks once again to the helpful staff of the Music Collection in Vienna, who has also provided Véronique with additional working materials (such as a light pad for examining watermarks, etc.) during her ongoing visits!
Before we enter the world of digital handwriting analysis in the next few blog posts, here is a brief insight into the pre-processing of the data.
As mentioned in the last blog post, there are four volumes of Philipp Gumpenhuber’s chronicles in the Theatre Collection of the Houghton Library at Harvard University (https://library.harvard.edu/collections/harvard-theatre-collection). So far, these are linked by a single IIIF manifest. Each volume is a separate entity and holds a mosaic of events and performances that deserves to be presented and analyzed individually.
But first, what is IIIF? IIIF (International Image Interoperability Framework) provides a standardized method for presenting materials from libraries and archives worldwide, enabling, for example, the examination of high-resolution digital copies of documents. More information about IIIF can be found on the official website at https://iiif.io/get-started/how-iiif-works/.
Thanks to the Python programming language, we are able to process each work individually and thus give it back its digital independence. This division is important for detailed analyses using Transkribus (https://www.transkribus.org/de). Through this careful separation, we have created four autonomous digital manifests that allow us to faithfully capture and validate individual works.
Therefore, Christina Dittmann created Python scripts that split a large IIIF manifest into individual, autonomous manifests. Through this process, each volume regains its digital identity and becomes ready for detailed transcription using Transkribus.
The process can be divided into four key steps: (1) the integration of the overall manifest into Python, (2) the division of the manifest into individual volumes based on the existing structure, (3) the creation and formatting of standalone IIIF manifests for each individual volume, and (4) a validation to ensure interoperability and correct function.
The entire process and associated code are publicly available in the GitHub repository at https://github.com/dittm/iiif-manifest-splitter.git, providing a resource for anyone interested in working at the intersection of history and digital technology.
A specific feature of Philipp Gumpenhuber’s preserved theater chronicles is that the collection is currently located on both sides of the Atlantic.
One part of the collection is held in the Theatre Collection at Houghton Library (Harvard University) (https://library.harvard.edu/collections/harvard-theatre-collection), the other part is preserved in the music collection of the Austrian National Library in Vienna.
Accordingly, Ingeborg Zechner traveled to Cambridge, Massachusetts to search for watermarks and other specifics of the source that are not visible on the scans.
It turned out, among other things, that the chronicles have a splendid presentation with gilded book edges and index ribbons made of green silk. In addition, paper from different paper manufacturers was compiled for the chronicles.
It is still unclear why the collection was originally split up – the digital edition of the chronicle forms an important task in reuniting the preserved volumes.
In order to fulfill her representative duties as principal investigator, Ingeborg Zechner met the dedicatee of the chronicles, Giacomo Durazzo and his wife Ernestine Aloisia Ungnade von Weissenwolff, for tea in the baroque salon of the Metropolitan Museum of Art (https://www.metmuseum.org/art/collection/search/437177).
Project pug Rudie was delighted that the Durazzo family was apparently also very fond of four-legged companions (Havanese and hunting dogs).
‘Excellences’… oh, please forgive me, this is an English-language blog.
Allow me to introduce myself briefly: I’m Rudie – ‘conseiller intime, et Directeur de menus plaisirs’ of the GuDiE research project.
Rudie
As a pug of aristocratic descent, I have many years of experience in courtly representation and will support my court…er… the GuDiE project team with my expertise. I will keep you, Excellences, informed about current project developments here in the coming years.
To test the skills of the GuDiE team, I had my staff locked in the castle dungeon as part of an aptitude test: The fact that they were able to escape gives me hope for the rest of the project. I took a photo to prove it:
Kickoff event: project team in the escape room (Jakob Leitner, Christina Dittmann, Véronique Braquet, Selina Galka, Ingeborg Zechner)
So that you not only get to know me, but also the entire GuDiE court, allow me to briefly introduce the individual members at the start of the project:
Ingeborg Zechner is a musicologist at the University of Graz. Her main areas of research include music theatre (eighteenth-twentieth century) and film music. She deals with questions of mediality and migration/mobility, the social, cultural and reception history of music as well as digital musicology.
Véronique Braquet is currently writing her master’s thesis in musicology at the University of Vienna. Her research focuses on a wide range of topics, including the revolutionary upheavals of the eighteenth and nineteenth centuries (with a thematic focus on church music, Marian devotion and national music, as well as general questions of social and cultural history), the late Middle Ages and African-American music.
Christina Dittmann is a Master’s student of Digital Humanities at the University of Graz. She holds a Bachelor’s degree in Sociology and English Studies from the University of Heidelberg. Her research interests focus on literary and cultural studies as well as digital methods and technologies. She is particularly interested in digital editions as well as XML, XSLT and web technologies.
Selina Galka is a research associate at the University of Graz at the Institute Centre for Information Modelling. Her research focuses on digital editions and data modelling. After completing her Master’s degree in ‘German Philology of the Middle Ages and Early Modern Period’ and ‘Digital Humanities’, she is currently a PhD candidate in the field of Digital Humanities.
Jakob Leitner is studying for a Master’s degree in Historical Musicology at the University of Graz and the University of Music and Performing Arts Graz. His research interests lie in the field of musicological gender research with a focus on female composers as well as in digital humanities and digital musicology.
‘Le très-humble, et très obéissant serviteur’,
Rudie (project pug)
PS: Since I want to take modern democratic tendencies into account (I’m not stuck in the eighteenth century), I will also let the individual project members have their say in this blog from time to time.
In the eighteenth century, the Viennese court was a center of European social and cultural life. Accordingly, it maintained an extensive network for representation, in which the Viennese court theatres (and their personnel) functioned as important institutions of cultural agency. For the years 1758 to 1763, the cultural activities of the Viennese court have been documented in theatre chronicles that have been compiled by the ballet master Philipp Gumpenhuber. Gumpenhuber’s theatre chronicles represent an important and unique source for eighteenth-century music history. However, these chronicles provide not only insights into the Viennese court’s various theatrical ventures, but also into the broader social and political court life.
A-Wn, Mus.Hs.34580b, f. 40
Only recently has research on eighteenth-century music theatre employed an expanded concept of the music theatre as an artwork, which perceives music theatre as a “performance” rather than a “text”. Equally this musical performance is shaped by various “agents” in conjunction with their sociocultural contexts. This so-called “performative turn”, based on concepts from cultural and theatre studies, has permanently changed the study of music theatre history. It offers a variety of interdisciplinary interconnections – also for the project “Digital Edition of Philipp Gumpenhuber’s Theatre Chronicles (1758–1763)”.
The project, carried out at the University of Graz with Ingeborg Zechner as principal investigator, pursues the goal of developing a digital edition (GuDiE) that makes Gumpenhuber’s theatre chronicles accessible to both scholars and the interested public. The digital edition GuDiE generates fundamental, scientifically curated data. This “linked data” consecutively forms the starting point for further digital projects on eighteenth-century European cultural history. GuDiE also provides valuable data for existing digital historical research projects on the Viennese court. Through the methodological interaction of the disciplines musicology, history and digital humanities, GuDiE creates significant insights into the complex interplay of socio-cultural, political, and economic factors influencing the practices of court theatres in the eighteenth century.
GuDiE is carried out in cooperation with the Center for Information Modeling at the University of Graz, the Austrian National Library, the Houghton Library at Harvard University, the Austrian Academy of Sciences (projects theadok.at and VieCPro), and the Institute Public Sector Transformation at the Bern University of Applied Sciences.
