Interview with John Tranter
by Ramona Koval
Ramona Koval: Tell me about the text analysis program 'brekdown', how it works and how you have been using it to write.
John Tranter: I think it was originally called 'breakdown' but because of the eight letter restriction on DOS file names that had to be constricted to eight letters and so the best way to do that is to spell it 'brekdown'. I call it breakdown. It was developed from a mathematical theory. In 1948 the English mathematician Claude Shannon developed a way of analysing letter group frequencies, perhaps working from decoding algorithms. But in the early 1980s Hugh Kenner, the American Professor of English and literary critic, worked with a computer programmer to build a program called 'Travesty'. This program was based upon Shannon's work: it analysed a particular text in terms of the frequency of the letter groups that appeared in the texts, and it gave a statistical weighting to all of the letter groups that appeared in the text. For example, the letter group 'space bar, t, h, e, space bar' would occur frequently in texts in English. You might give that a frequency of eight out of ten. Whereas the letter group 'z, q, space bar, space bar, space bar' would never occur at all, so you'd give that a weighting of zero. Once you'd built up a frequency table and an index, you can ask a computer to reconstruct the text based on that frequency table.That's where it gets interesting because then you can watch a computer apparently writing pieces of literature.
RK: So you are saying that an individual writer, whether it's Louisa May Alcott or Carlos Castaneda, has a different kind of frequency for different kinds of letter groups and this will give you a taste of that person's writing?
JT: Yes, that's what I'm saying actually. That happens. It's an odd thing too, because that kind of analysis of a work can be applied to literature or writing of any kind, even a newspaper or the weather report. Each piece of writing has characteristics that derive from the language it works in, for a start. There are letters in Russian that you don't get in English, for example, and also different writers writing on different topics use different groups of letters because they use different words. It is interesting that a program that takes no account of spelling, grammar, meaning or paragraph construction, can turn out a garbled text, but one that has that feel of something that somebody was trying to write.
RK:So you actually scan a text of the original author into the computer, then it does its calculations and it spurts something out. What exactly do you get out of the computer?
JT: Well, what you get out of the computer is gibberish really and this is the interesting point about this technique: it's not that it provides you with a text that you can use as a writer, but only the roughest kind of first draft. In fact, it provides you with a text that is absolute gibberish, and to get that into a state where it means something interesting as a work of literature is very hard work. The work I've done on these particular texts -- I've done about eight now and they're each about five pages long -- that's the hardest writing I've ever done. It's just exhausting.You can ask a computer to construct a text based upon the work of a particular text that you have scanned in. You can also ask it to provide you with a raw text from a blend of two or three or more texts that you have scanned in. That's what I'm particularly interested in at the moment because then you get the appearance of a work of writing that was written by two different authors and they tend to struggle with one another behind the text -- or at least this is what the reader feels. In one case I scanned in two texts, a part of a book by Captain Johns called Biggles Defies the Swastika and part of a text called The Well of Loneliness, written by Radclyffe Hall, published in the twenties I think, and the first lesbian novel that had a serious intent. These two texts were scanned into the computer and produced some raw text like this:
That some words when I left this creations terribly Stephen glance that Stephen not to discuss their eyes and you marry me Stephen's all wrong I'm mad sometimes endowed with a mutter of machine and ended at Angela withheld out of his hand...
This goes on for twenty pages.
RK: So what are you going to do with it? How can you make it sensible?
JT: That's the interesting thing -- you can't really. It's an almost impossible task. What I do with these twenty pages of text is just comb through them and throw out all the bits that don't look interesting. Then I do that again and again, until I'm left with about eight pages of text that consist of interesting fragments. Then I look through them to see if there's a pattern in the fragments that I can lump together so that I can get two or three groups of things that feel a bit the same, and then I try to construct a story out of them: this is the interesting part. When you are reading prose you can't really read it for very long unless you feel there is something happening behind it, some kind of narrative, a story happening. And of course there isn't in any of this material to start with -- there's no story whatsoever. Constructing that story brings you to the real point of all this work, and that is: it allows your unconscious mind to get to work on the writing you're doing and to create work you would never imagine doing yourself. This is really why I use the technique, to help me find things to say that I would never think of saying of my own accord.
RK: You described the process as rather like jazz improvisation.
JT: Yes, it's a bit like that, except that it's really the other way around. It's a bit like jazz improvisation back to front. With jazz improvisation you get a standard tune and often when you hear a classic jazz piece the group plays the standard tune first and then one of them improvises on it. As the piece develops the improvisations become more oblique and they go further away from the melody of the original song. What I am doing is beginning with that obscure, abstract, oblique melody and trying to find my way back to an original that may have existed behind it. The technique has something in common with what the OLIPO group does in Paris. The word OLIPO stands for oeuvre du littérateur de potentiale, and it means a workshop where experimental works are constructed. This group of writers and mathematicians and musicians -- about eighteen of them, I think -- develop constrictive forms of writing. For example, the best known of these works of art that have been developed from these constrictive forms is a novel by Georges Perec. It's a lipogram and it's a work of literature constructed without the letter `e'. That is a bizarre restriction, but restrictions in writing have been around for thousands of years. All of the complicated verse forms were originally a very complex and tight restrictive way of making you write. Writing is different from ordinary speech because of its formal constrictions. That's what makes a poem a poem, and a bit of prose a bit of prose, and a bit of talk a bit of talk.
RK: So that piece to do with Biggles and Radclyffe Hall, have you an example of what you've done with that?
JT: Yes, I have a piece called 'Lonely Chaps' which is six pages long, about two thousand words. Eventually I got a kind of story out of all that mess of material.
RK: Biggles and Radclyffe Hall, and then you've combined Allan Ginsberg and the Bobbsie Twins, Carlos Castaneda and Louisa May Alcott, this is rather mischievous of you isn't it?
JT: It's very mischievous, and it's fun. The point is about literature if you're doing a lot of writing and you're going to spend the rest of your life doing it, it's got to be fun.
RK: You've used another program called 'Namegram' to call up the name of the new author of this thing, because it's not John Tranter (but it is).
JT: Yes, I began doing this work years ago when I used the 'Brekdown' program to analyse the work of a poet and then reconstruct in that 'tone of voice'. I did that to the work of Matthew Arnold and then I did another example based on the work of John Ashberry and the two poems that resulted from that exercise were published in Meanjin in 1991 alongside the articles that explained the technique. I wanted to publish the poems as though they were written by imaginary poets, so to get names for them I used this 'Namegram' program which found anagrams of their real names. John Ashberry became Joy H. Breshen. The name Matthew Arnold generated some three and a half thousand anagrams, believe it or not...
RK: If I found something that you'd worked on and then discovered that it wasn't actually written by a person but by computer, I'd feel that it was quite interesting, but I would feel cheated, and dismiss it.
JT: It lacks authenticity, doesn't it? That's what people want, they want authenticity. That's why they were so distressed when they discovered Helen Demidenko was not actually a Ukrainian person whose relatives had actually experienced these events. I think authenticity is a real bugbear in literature. I guess if you're reading a diary that says I actually was there, I felt and experienced these things, come and share them with me, then you would want to believe these things had actually happened. But not all of literature is like that. Henry James is not like that. When we read a historical novel by David Malouf say, we don't believe that David Malouf actually experienced these events that occurred a hundred years ago: we know he made it all up.
RK: But if we are reading a text that makes us feel a certain way and calls up certain emotions in us, we like to think that that manipulation of our emotions is coming via a human mind rather than a computer model.
JT: I suppose people do, it is interesting. T.S. Eliot developed the concept of the 'objective correlative', and I think that's an interesting point for writers to think about. He said if a writer wants to call up in a reader a particular emotion, you don't just dump that emotion on the page. If you want the reader to feel distressed and sad you don't say in your work of art 'I feel awful, I feel sad'. What you do is you think of some event that will make the reader feel that way. So, for example, if you are writing a story about a puppy dog, you have the puppy dog run over and killed and that makes the reader terribly distressed. So you find an 'objective correlative' of the emotional state you wish to provoke and present that to the reader. T.S. Eliot also had a wonderful image about the superficial meaning of a work of writing. He said the meaning on the surface is like the lump of meat that the burglar throws to the guard dog, to distract the guard dog, while the burglar sneaks around the back of the house and breaks in. That idea of deceit, darkness, breaking in, is very interesting in psychological terms I think...
(incomplete)