Copenhagen: Digital Applications in Assyriology

Introduction

Today marked the official end of my final summer school before commencing the Fall 2024 academic semester at Hopkins. The Digital Applications in Assyriology Nordic Summer School in Copenhagen, Denmark was held in person from August 12-16 with a final online session on August 30. This was certainly one of the more educational and rewarding programs I attended all summer. The team of instructors included Seraina Nett, Lena Tambs, Gustav Rybert Smidt, Rune Rattenborg, Aleksi Sahala, Saana Svärd, and Émilie Pagé-Perron, all of whom practice and publish in the digital humanities.

Copenhagen Summer School: August 12-16

Each day of the summer school was packed with lectures and practical sessions for learning tools and software associated with the application of computational methods in digital Assyriology. The lectures often revolved around the instructors’ own research projects, such as an overview of the various aspects of the Cuneiform Digital Library Initiative (Pagé-Perron), using Python for computational linguistics (Sahala), visualizing data (Nett and Rattenborg), and Social Network Analysis and Lexical Network Analysis (Svärd; Tambs).

Monday was dedicated to data collection and cleaning, covering file types like TXT and CSV, and utilizing tools such as OpenRefine and Excel to organize and clean data. The afternoon’s practical session involved working with cuneiform catalogues from the British Museum and CDLI. Tuesday was a crash course in Python, including text analysis, DataFrames, and visualizations. On Wednesday, we learned about database management and visualization software, especially in the domain of geo-mapping (QGIS, Google Earth Pro). Thursday was all about networks, especially Social Network Analysis using Gephi. Finally, Friday was devoted to working on our final projects which we presented to the class via Zoom this morning.

Final Project: August 30

For my final project, I worked in collaboration with Adel Oubraham, a French archaeology student at École du Louvre. The title of our project is “Treasures of the Ancient Near East: A Comparative Analysis of the British Museum and Louvre Cuneiform Collections.” In this paper, we aimed to analyze the cuneiform collections of the British Museum and the Louvre using computational tools applied to metadata from the Cuneiform Digital Library Initiative (CDLI). We were interested in comparing these collections from a museological perspective with regard to the chronological, linguistic, and geographical composition of these collections. We had three overarching research questions: (1) What is the linguistic makeup of the collection? (2) What is the provenience distribution? (3) How has the size of each museum’s collection evolved over time vis-à-vis British and French colonial and archaeological interests in the Middle East? Ultimately, given that we’re dealing with over 100,000 cuneiform objects between the two museums, computational methods were necessary to even approach an answer.

Our workflow was straightforward. We used the CDLI API to download all the available metadata about these collections. Then we uploaded the CSV files into data cleaning and analysis software such as Excel and OpenRefine in order to identify inconsistencies, remove duplicates, and standardize the data for accurate analysis and further processing. Finally, we used Tableau, Gephi, and QGIS to create various visualizations. I then wrote up the results in a short 10-page report.

Throughout the process, we recognized there were several biases and limitations in our study. First, we were at the mercy of CDLI metadata which could be inaccurate or incomplete in ways beyond our control. On the one hand, the Louvre was missing a lot of metadata with regard to linguistic and provenance information; on the other hand, CDLI is missing several thousand texts in the British Museum. Second, there proved to be inconsistencies in cataloging and documentation across the two museums over the past two centuries which had the potential for skewing the analysis, such as whether unidentifiable or unprovenanced fragments were grouped together.

Nevertheless, there are some interesting results. First, the cuneiform collection at the British Museum is approximately five or six times larger than that of the Louvre, despite the fact that the Louvre is a much larger museum. According to the available CDLI metadata, the Louvre comprises approximately 14,575 artifacts compared the British Museum’s 75,959 artifacts (again, CDLI is incomplete, as they are closer to 20,000 and 120,000, respectively). Second, while both museum collections emphasize Sumerian and Akkadian texts, the Louvre collection is actually more diverse with a more significant number of Elamite, Hittite, and Ugaritic texts in proportion to its overall size. Third, the provenience of cuneiform artifacts from the British Museum primarily traces back to core Mesopotamian urban centers, especially Nineveh and Girsu, whereas the Louvre’s collection includes a significant number of artifacts from a wider range of locations, including Anatolia, the Levant, and especially Susa. This is influenced by British and French archaeological and museological interests in the region, e.g. Austin Henry Layard’s excavations and discovery of the Library of Assurbanipal while French archaeologists Marcel-Auguste Dieulafoy and Jacques de Morgan were excavating Susa.

While I’m not including any of the visualizations that we created here (in case we use them in a more formal context later), I will leave you with three prospects for future research: First, we want to enlarge our dataset to other museums included on CDLI, especially the National Museums of Iraq and Syria. Second, we need to standardize the metadata to harmonize fields and formats, which would enhance data analysis. Finally, with better and cleaner data, we can use AI and machine learning algorithms to confirm our analysis and identify other patterns in the data.

Postscript

I want to thank the application committee for accepting my application to participate in the program as well as the instructors from whom I learned so much, especially Seraina Nett.

About The Author

Matthew Saunders

Matthew Saunders is a PhD student in the Department of Near Eastern Studies at Johns Hopkins University. He researches the languages and literatures of the ancient Near East, especially Aramaic Studies, Ugaritic Studies, and Comparative Semitics.

See author's posts