Literary game of drafts

Umberto Eco on... technology and ghosts in the machine

Saturday March 30, 2002
The Guardian

Philologists are often interested in seeing how an author goes from the first to the last draft of a text, and they love to look at its various versions. This activity is often called "notebook criticism". To be a good critic of notebook scribbles, or of versions of texts, it's crucial that an author has left behind various handwritten phases of his work.

For example, we have the various phases through which Alessandro Manzoni's La Pentecoste passed, and it's very interesting to follow the changes of heart, the substantial upheavals and the minimal variations that the author made to his text. Similarly, it is moving to see at the National library of Naples, the phases through which some of Giacomo Leopardi's most beautiful poems became those that we know today. And we come to understand how a minor correction radically changed the magic of a verse.

To make sense of the changes to a text, it is necessary, of course, for an author to have left behind indications of what was changed from the original manuscript. If we are dealing, for example, with an author such as Dante Alighieri - not even the manuscript of The Divine Comedy remains - then the game is over before it begins.

The question of manuscript changes is very important for literary criticism, the psychology of creation and other aspects of the study of literature. So it makes sense that the Institute of Texts and Modern Manuscripts at the National Centre of Scientific Research in Paris dedicates many conventions and seminars to the subject.

A recent seminar focused on a question that's often looked at the wrong way: isn't it the case that the common practice today of writing texts directly on a computer - so that there is only one definitive printed version - kills the study of changes? Now, let's assume that an author drafts the first version of his text, and let's call this version A. To simplify things, let's assume that the author wrote it directly on a computer, or that if he had made any handwritten notes, they have disappeared.

This version A is printed, and at that point, the author begins correcting it by hand. In this way, we get version B, which in turn is transferred to computer, where again it is cleaned up and printed anew and becomes version C. In turn, this version is altered by hand and again recopied as version D on the computer, from which a new version will be created: version E. Since computers encourage corrections and reconstructions, this is how the process can give rise to - if the author does not throw the intermediate steps in the wastepaper basket - a series of versions, let's say from A to Z.

So that's good news for the philologists, who in theory should have more to work with, not less.

But the matter does not end here. Let's go back to version B, which was version A, printed and corrected by hand, and let's imagine it was quite tortured. In transferring it to the computer, does the author reproduce it word for word? Almost never.

Just think about the common practice of composing a simple letter, when we are liable to do a draft, erase or rewrite. In transcribing, new versions are introduced, and perhaps we write down something that we had changed but then regret it, erase it and take another crack at it.

And here - when we print out the version again - we do not have that version C, which was supposed to faithfully reproduce version B. Instead, out comes a version that we will call X, but between B and X there are "ghost" versions, each one different from the other.

It could be the case, though rare, that the author - narcissistic and fanatical about his own changes, and using some kind of special computer program - has kept somewhere, inside the memory of the machine, all these intermediate changes. But usually this does not happen. Those "ghost" copies have vanished; they are erased as soon as the work is finished.

And so the work of the philologists of the future will be based on conjecture, on what those "ghost" copies might have contained - and who knows how many great texts and other erudite publications will be born from that conjecture? To outsiders, they might seem like problems suited only for college exams. But the discussion shows that the use of mechanical systems for writing doesn't necessarily simplify and thereby mechanise the creative activity, but rather can make it that much more shaded and complex.

For example, who says that the possibility of endlessly correcting a text ad infinitum necessarily improves the work? Well, we all know that the best is the enemy of the good. Or, it is true that with a writing program, one can determine (even with a text of hundreds of pages) how many times the same word is repeated and decide to substitute it with synonyms or paraphrasing?

But we know, that, for example, Manzoni's vocabulary was very poor, and the word "good" in his novel I Promessi Sposi appears, to some at least, to have been used excessively. Would Manzoni have benefited from having a computer, eliminating all these repetitions, or would he have made his prose more baroque and less limpid?

© Umberto Eco/New York Times