Ancient wisdom has collided with modern information technology to produce the ultimate classics hypertext resource, Mike Holderness reports
It is the best surprise I have yet found on the web. Stuck while writing a paper on virtual reality, I decide I need to read for myself the story of Plato's cave, whose unfortunate prisoners take shadows cast on the walls for an entire universe. After a bit of poking around the AltaVista search engine and people's links pages, I am delivered to the Perseus database of the classics. There is a translation, with copious annotations and links to commentaries and glossaries. After downloading an ancient Greek font, I can use a version in which every word is linked to a definition, frequency statistics and other contexts where the word is used, the ultimate hypertext.
As soon as learning ancient Greek comes to the top of my to-do list, I will have in Perseus the perfect guide and ball of string. Some of humanity's most ancient texts and newest technology have collided, and the result has utterly changed both.
The editor-in-chief of the Perseus project is Gregory Crane, professor of classics at Tufts University. He is a one-man whirlwind of ideas about text, hypertext and the possibilities for constructing a densely linked web of practically everything scholarly. Indeed, whereas the mythical Perseus slew the Gorgon, this one seems to be building a fine head of snakes without need of mirrors.
Having completed the Greek classics, the project moved on. It has made a start on Latin, with large chunks of Caesar, Catullus, Cicero, Hirtius, Horace, Livy, Ovid, Plautus, Servius and Virgil. In July it received a grant of $2.7 million (Pounds 1.6 million) from the US Digital Libraries initiative. One of the projects this will fund is a variorum edition of Shakespeare, with the Modern Languages Association, presenting the various texts and editorial interpolations side-by-side, starting with Antony and Cleopatra. You can get a flavour of the possibilities from the edition of Christopher Marlowe's Dr Faustus which is already on the web site.
Appropriately, a virtual-reality "cave" will be the centrepiece of another new project, presenting ancient Egyptian texts. Users will be able to enter virtual tombs from which they can get full details of every inscription and every hieroglyph. Another project involves the collection of the Boston Museum of Fine Arts; and yet another collating histories of the City of London in the 18th century with maps. In this, the system will be able to infer what date is being discussed even in sections without explicit dates, and link to appropriate background material.
To "mark up" all these texts and images by hand would be a Herculean task, if not a Sisyphean one; push it as we will, new work constantly rebounds the stone of scholarship from the brink of completion. Imagine working through Shakespeare, inserting all those references word by word. Designing software to do the mark-up automatically takes us, however, straight back to deep problems in formal semantics, discourse and use-in-context. These are, of course, the problems which the completed hypertext promises to be so helpful in exploring.
Crane gives the problem of place-names as an example. "The Simpsons," he notes, "is based on the premise that there's a Springfield in practically every state of the US." How could software determine which one was referred to in a text? Searching Perseus for "Perseus" provides another example. Ideally, we "need to figure some way to isolate the primary thread of a narrative from side references". Machines are not (yet) going to provide any such deep analysis but they can "find all the places where the characters are going to or from a place-name". He is intrigued, too, about the possibilities for working with projects like Wordnet, which aim to build databases of words and their definitions, usage and implications.
From that deeply textual note our conversation flies off to other expressions of culture. "Take Thomas Morley's A Plain and Easy Introduction to Practical Music of 1597. How," Crane asks, "do you code it?" Midi, the computer language used in making modern electronic music, seems an obvious candidate for the centrepiece of an online edition, offering the possibility of a "search by riff" facility. From this, a Perseus edition of Morley could link to modern musical notation, the notation of the time, facsimiles of the first edition, scholarship about music in Shakespeare and so forth.
No wonder, given this effervescence of ideas, that Crane was the star turn at the conference on Information and Communication Technology in the Classics, in Oxford at the end of July. Organiser Julian Morgan believes that Perseus will radically change teaching of the classics.
Students "are still going to have to learn a large body of factual knowledge" but, he suggests, "we may end up teaching a different kind of grammar". As long as students understand the mechanisms of infinitives and participles, the database can guide them through the details. By the way, if anyone can find the $200,000 which Crane estimates it would take to mark up the rest of the Latin literature, I among others would be most grateful.
WordNet search: www.notredame.ac.jp/cgi-bin/wn
Joint Association of Classical Teachers