It has been more than six decades since Warren Weaver, a pioneer in automated language translation, suggested applying code-breaking techniques to the challenge of interpreting a foreign language.
That insight led to a generation of statistics-based language programs like Google Translate - and, not so incidentally, to new tools for breaking codes that go back to the Middle Ages.
Now a team of Swedish and American linguists has applied statistics-based translation techniques to crack one of the most stubborn of codes: the Copiale Cipher, a hand-lettered 105-page manuscript that appears to date from the late 1700s.
Discovered in an academic archive in the former East Germany, the elaborately bound volume holds 75,000 characters, a perplexing mix of mysterious symbols and Roman letters.
Kevin Knight, a computer scientist at the Information Sciences Institute at the University of Southern California, collaborated with colleagues in Sweden to decipher the first 16 pages. They turn out to be a detailed description of a ritual from a secret society that apparently had a fascination with eye surgery and ophthalmology.
It began as a weekend project this year, Knight said in an interview, adding: "I don't have much experience in cryptography. My background is primarily in computational linguistics and machine translation."
Eventually the researchers concluded that the Roman letters were so-called nulls, meant to mislead the code breaker, and that the letters represented spaces between words made up of elaborate symbols.
The Copiale Cipher has value to historians who are trying to understand the spread of political ideas. Secret societies were all the rage in the 18th century, Knight said, and they had an influence on both the American and French Revolutions.
Modern examples of challenging ciphers include the "Kryptos" sculpture, commissioned for the CIA headquarters, which has been only partly decoded.
But the white whale of the code-breaking world is the Voynich manuscript. Comprising 240 lavishly illustrated vellum pages, it has defied the world's best code breakers. Though cryptographers have long wondered if it is a hoax, it was recently dated to the early 1400s.











