User notes for the
Adjunct Travesty
How it works (if it works)
This page last updated 5th November 2001.
The Adjunct Travesty is modelled on the original TRAVESTY program, written by Hugh Kenner and Joseph O'Rourke. Here is Kenner's description, quoted without permission from the Afterword to the Book Sentences by Charles O Hartman and Hugh Kenner (Los Angeles, Sun & Moon Press 1995):
In the November 1984 BYTE Hugh Kenner and Joseph O'Rourke published a program called TRAVESTY (based on a long-ago idea from the Father of Information Theory, Claude Shannon), which in working its way through a text gives us something recognizably similar yet oddly skewed. You supply TRAVESTY with an "Order"; if the order is 4, it seeks out every occurrence of the initial 4-character sequence and records the character that comes after. It then chooses at random from its notepad one character to append, moves forward one place, repeats the whole process.... (If you buy many lottery tickets your chances improve. Likewise, a most-frequent-character in the notepad is the likeliest to be chosen.)
Thus if the text begins with the word "Sentences" and the order is 4, TRAVESTY....after inspecting every instance of /Sent/ (e.g. "present", "sentience", "consent")....might well find a high proportion of trailing spaces and choose to append a space: /Sent/ . Spacing forward, it would then inspect every instance of /ent / with the same numb purpose...and on and on until delivery of as much output we'd asked for. (p.77f.)
The version of TRAVESTY on this site was reconstructed in JavaScript from the above description, without access to the original code. There are still bugs: the script is presented simply for fun and as a possible resource for writers. Here's a list of known problems and tips for better use:
- The script has been fully tested in IE5.5 and Netscape 4.6 on a PC. It also works in Opera 5 and Netscape 6.1, though the status line countdown doesn't work in these two browsers, making it difficult to know whether the script is working or not.
- Start off by playing with the Adjunct extract provided, before trying a text of your own.
- When you first try a text of your own, start by generating a 1000 character Travesty. This should be generated reasonably quickly, and will give you an idea of whether or not the text is going to work.
- The script sometimes has problems with short source texts. If you find that your text hangs the browser or generates lettriste gibberish, try pasting in the same text several times over.
- Repetitive letter sequences in the original text can cause the script to fail. E.g. if the text contains a line of hyphens ------------------------- used as a divider, this will cause problems. The same thing can happen if the text contains transcribed laughter ("ha ha ha ha ha ha") or similar...
- The status-line countdown should proceed relatively quickly. If it slows to a crawl, this is usually a sign that something's gone wrong.
- You're welcome to download the page to your own hard disk for offline use.
- The complete text of Adjunct is available on UbuWeb -- try using it as a source for the Travesty.
- The HTML-formatted output does not take account of the line-breaks in lineated/poetic source texts -- however, if you view the source code (IE: right click on the page and select "View source") of the output page, you should find some lineation preserved there.
- You may have to use ctrl+alt+delete (on a PC) to close down your browser if it appears to hang. Be warned!
- I'm still working on it.
Return to the Adjunct Travesty
Return to 