A Monument in Slips: The Middle English Dictionary at Michigan, continued

Jim to ChatGPT: Could you have created an MED? If so, how long might it take and would it rival the actual MED in comprehensiveness?

ChatGPT: Great question—and one that gets at the heart of how tools like me relate to deep, long-term human scholarship.

Could I have created the Middle English Dictionary?
In short: I could assist in creating something like the MED today, but not without major help from human experts, and it likely wouldn’t rival the actual MED in accuracy, depth, or scholarly authority. Here’s why:

What I can do:

  • Process large volumes of text quickly: I can scan and extract patterns from Middle English texts at scale, identifying word usage, variants, and contexts far faster than any human team.
  • Draft preliminary entries: Given a corpus of Middle English texts, I can suggest meanings, spellings, and examples, much like a first-pass digital assistant.
  • Sort and group meanings: I could help categorize senses of a word by context, frequency, and grammar, saving scholars time.

What I can’t fully replace:

  • Source expertise: Middle English has enormous variation in spelling, grammar, and regional usage. Scholars who spent decades immersed in these texts bring a depth of judgment that no AI can replicate yet.
  • Citation and nuance: The MED isn’t just a word list—it’s a historical record. Each definition is based on specific, dated sources, selected with care. I don’t currently have access to all medieval manuscripts or to the exact quotations used in the MED unless they are part of a digitized and accessible corpus.
  • Scholarly consensus: The MED reflects decades of peer-reviewed academic labor. I can generate hypotheses, but not the kind of scholarly authority that comes from human vetting.

Jim: How long would it take today?

ChatGPT: If a modern team used today’s AI tools—including me—to accelerate slip-gathering, sorting, and drafting, they might build a first-pass version of the MED in a few years, depending on corpus access and team size. But getting to the actual MED’s level—meticulous source-checking, contextual nuance, etymology, and consistency—would still likely take decades, just more efficiently than before.

ChatGPT’s Bottom line:

I could help expedite the making of a dictionary like the MED, especially as a collaborator with scholars. But I wouldn’t rival the MED in comprehensiveness or reliability without deep human oversight. The MED remains a triumph of collective human intelligence—painstaking, analog, and enduring.

———

Jim’s anecdotes about the MED.

  • In the mid 1970’s Jim met with then editor Sherman Kuhn about a grant application to the Andrew Mellon Foundation asking for funding to accelerate completion of the MED. The application was quite short – perhaps 7 pages of text – and was tailored to be sent by U-M President Robben Fleming directly to the Foundation’s President, Nathan Pusey, the former President of Harvard. The application focused on the word grace and its multiple meanings in medieval literature (Pusey’s scholarly work was in English literature and ancient history.) A very short time later, a brief letter was received from Dr. Pusey and it enclosed a check for $950,000.
  • Leading up to this application, Jim met regularly with Dr. Kuhn and his staff. Jim asked Dr. Kuhn if on a summer Friday afternoon, was he ever tempted to ignore one final interpretation of a word and go home early, knowing that in all likelihood, no one would ever know the difference. His response was that he had that feeling nearly every Friday afternoon but stayed to write down that last interpretation anyway.
  • Some long-time staff at the MED spent entire careers working on just several letters of the dictionary.
  • Data collection started at Cornell University before the entire project moved to U-M: Beginning around 1925, Cornell University Professor C. S. Northup directed a project under the auspices of the Modern Language Association of America to collect Middle English citations. Cornell had inherited a significant number of citation slips related to Chaucer from Ewald Flügel. In 1930, when the University of Michigan formally began the Middle English Dictionary project, Cornell donated its collection of slips to serve as a major part of the new dictionary’s corpus.
  • Prior to the external grant funding, the project was funded by the Horace H. Rackham Foundation, the Office of the Vice President for Research, and the English Department. In later years, the National Endowment for the Humanities joined with the Mellon Foundation to find completion of the project.
  • In 2001, Jim was gifted 8 of the 115 fascicles of the MED, spelling out his last name: RANDOLPH. They remain in a cherished spot in his home.