Having just weakened your will to live by exposing you to the word heteroscedasticity , I thought I’d now throw some more paraffin onto your wordy fires. Is the Voynich Manuscript…
…an “ergodic text”?
According to Espen Aarseth [as discussed on the Grand Text Auto website], ergodic literature is where “nontrivial effort is required to allow the reader to traverse the text”… yup, I think we’ve all read a fair few books like that.
Personally, I think this would mean that the Voynich Manuscript – which has had its bifolios shuffled several times during its lifetime – has ended up as “unintentionally ergodic literature”, because nobody knows how to properly traverse its pages, let alone read its text. What’s worse is that I’m reasonably sure that nobody even knows how to parse its letters, increasing its, errrm, ergodicity yet further. The Voynich is, quite literally, hard work to read.
All the same, one nice thing about authors in the kind of cybertext-y tradition Aarseth belongs to is that they do a good job of collecting a whole load of bizarre textual oddities to muse upon. Not novels without the letter ‘E’ or where every sentence is a question *sigh*, but stories with intentionally shufflable pages, or even Ayn Rand’s play where the audience votes for the ending they want… you know, mad stuff like that.
…or an “aleatoric text”?
What is relevant to us is that this stuff overlaps strongly with stochastic (randomly generated) or aleatoric texts, where there is an element of chance in the way that they are written (the Latin word “alea” means “dice”). As is fairly well known, plenty of people have posited the notion that the Voynich Manuscript’s ‘Voynichese’ text was entirely generated using some kind of cleverly randomising process. Most notably, Gordon Rugg suggests a tricksy arrangement with multiple Cardan grille-like tables of Voynichese word parts to achieve such a miracle. If they’re right, the VMs would be – horror of horrors – an aleatoric hoax (or, more precisely, a stochastic simulacrum of a ergodic text).
Given that this is a much-repeated claim, I thought I really ought to dip my toes in the early history of generated texts. Incidentally, there’s a parallel literature on the (mainly modern) tradition of aleatoric music (John Cage, Charles Ives, even Marcel Duchamp), which claims as a parent the 15th century “catholicon” genre, which was a universal musical genre which could be played in any mode or scale. However, I don’t personally see the catholicon as having any real randomness as opposed to just potential multimodality: linking it to John Cage seems a fairly spurious idea.
…or a “generated text”?
But as for generated texts, that’s a different story. A glimpse at your email inbox or even a typical websearch should quickly reveal that the world is now awash with such glibly generated texts. Increasingly, the Internet is populated more with plausible-looking generated text than real text. But when in history did all this awfulness actually begin? Might the big secret of the Voynich Manuscript be that its author was been the world’s first spammer?
Of course, the all-pervading layer of spam that threatens to drown us all is built atop computational linguistics, where computer programmers find ways of sequencing text that appears moderately sensible. In fact, the first documented computationally stochastic text came about in 1959 when Theo Lutz programmed a Zuse Z22 computer to mash up fragments from Franz Kafka’s “The Castle” in a grammatically plausible way.
…or a “permutational text”?
Naturally, we’re looking for something much earlier here: and thanks to determined researchers such as Florian Cramer, you can find plenty of stuff on “permutational texts”, texts that typically allow the reader or performer to swap things around arbitrarily. If you want to try some of these out for yourself, there’s an excellent selection on the Permutationen site.
But there’s a problem with this: these texts are all about permuting words, playing with meaning, synonymity, antonymity, association, conceptual linkages, linguistic simulation: whereas Voynich hoax theorists are looking for non-meaning, obscurity, grammatical simulation – all of which are elements occupying a completely different scale, a far tighter granularity.
…or a “combinatoric text”?
Even Raymond Llull‘s combinatorics (which I mentioned recently) were avowedly combinations of concepts, not letters: the paper machine he described in his “Ars Magna et Ultima” was comprised of multiple concentric word disks, using logical combination as a tool to try to reach ultimate truth.
However, the fascinating thing about this is that it has been claimed (and here’s a link to a particularly nice presentation) that:-
It is believed that Llull’s inspiration for the Ars magna came from observing a device called a zairja, which was used by medieval Arab astrologers to calculate ideas by mechanical means. It used the 28 letters of the Arabic alphabet to signify 28 categories of philosophic thought. By combining number values associated with the letters and categories, new paths of insight and thought were created.
If you’re now suddenly filled with an urge to find out about zairjas, here’s the Wikipedia page on them: further, David Link says in a fascinating article that “Taking into account the moment in time of the enquiry, [a zairja] generated a rhymed answer to any question posed”.
However, the (very) short version of all this is that Llull seems to have taken the zairja’s circular diagrams and done his own thing with them, very much as Leon Battista Alberti did with Llull’s in turn. But all in all, that’s quite a different tradition from what we’re talking about here.
…or none of the above?
Though I’ve searched and searched, I simply haven’t found anything in the history of any of these literatures conceptually similar to the way hoax theorists claim the Voynich “must have” been constructed. Is there some kind of link there to be found? Right now, I really don’t think there is, sorry! But please let me know if you think I’m wrong!