Going Public: Professor Emily Kadens Joins AI and Archival Research Through the Launch of the Egerton Model

Thanks to artificial intelligence and the work of a team led by Emily Kadens, Edna B. and Ednyfed H. Williams Memorial Professor of Law and Associate Dean of Research & Strategic Initiatives, researchers can now automatically generate highly accurate transcriptions of early modern English records written in a particularly challenging script called secretary hand.

“Egerton,” the model built by Kadens and her team, recently became available for public use after she and her team trained the AI program to successfully transcribe more than 1 million words from challenging 16th- and 17th-century legal court records.

“This has been a three-year project involving a tremendous amount of work put in by a small team.” Kadens says. “Personally, I’m looking forward to turning from model building to using the model for my own scholarship purposes. There has been a lot of interest in the Egerton Model in the last six months, and a surge of interest when we made it public.”

A legal historian, Kadens specializes in commercial fraud during the period these documents were created. She primarily focuses on the records of equity courts—royal courts known for their discretionary jurisdiction that gained prominence in the 16th century. She notes that equity court documents, mostly written in secretary hand, offer far more detail about disputes compared to common law court records, which relied heavily on oral procedures.

After myriad trips to The National Archives in London involving countless manual transcriptions by hand, Kadens wondered if there was a more efficient way to achieve the same results. That’s when she discovered Transkribus, the AI technology program used to power the Egerton Model.

“The Egerton Model enables people to do really large-scale archival research on secretary hand material of the 16th and 17th centuries, which is a difficult hand to transcribe manually,” Kadens says. “If you can have the model provide a transcription that is extraordinarily accurate, well, then this changes things significantly.”

One of the early adopters of the Egerton Model, Daniel Gosling, Principal Legal Records Specialist at The National Archives, shared his insights on the model’s impact.

“I’ve been following Professor Kadens’ work with interest for a number of years. The Egerton Model was trained on some of the most difficult legal records we work with,” he says. “The main barrier to these records—both from a research and a cataloging perspective—is the challenging hand they are written in. The Egerton Model can solve this problem by providing transcriptions of the majority of these records accurately enough to be understood by non-specialized researchers and catalogers, which could be transformative in opening up these rich but underused records to a wide variety of people.”

Kadens notes that the Egerton Model can benefit a wide range of audiences. Not only can legal historians access previously difficult-to-read documents, unlocking new avenues for research, but graduate students will also find it a valuable educational tool, helping them develop paleography skills while interacting with historical texts. Additionally, genealogists and family historians can utilize the rich data contained within equity court materials to construct detailed family histories and gain insights into property ownership.

The Egerton Model is already garnering interest from various institutions, including family history organizations, showcasing its potential to modernize how archival materials are cataloged and accessed.

“This changes things enormously,” Kadens says. “With the push of a button, researchers can now obtain workable transcriptions, making it easier to explore economic history and other significant themes.”

The future of archival research looks promising with the public launch of the Egerton Model, and its transformative potential is just beginning to be realized.