Artikelen van Erik Weijers

Krabbé's random word

[This is a translation of the Dutch article]

I once gave myself the assignment of inventing a completely random word. Completely random, is that possible? And all of a sudden, there it was: Battoowoo Greekgreek (Tim Krabbé, The Rider)

The word Battoowoo Greekgreek (in Dutch: Batüwü Griekgriek, which will be the word under consideration in this article) has, since the day the Krabbé entrusted it to paper, acquired a modest fame. Hundreds of thousands of readers of The Rider cherish the word. Maarten Ducrot confessed, when he was still a professional rider: ‘Whenever I hit absolute rock bottom I always think of those immortal words from The Rider by Tim Krabbé - Batüwü Griekgriek - and everything seems just fine again’. So the word has prospered pretty well for a modest random word. About time to examine if it is random at all! I want to stress first that the result of this examination could, of course, never blemish the magical, almost divine quality of the word. That should go without saying.

Still, there is something at stake here. The random word is within Krabbé’s oeuvre a miniature, so to speak, of an important theme. Many of his characters want to escape the regularity of their existence by doing something random or unpredictable. Not just as a revolt against ‘society’ but moreover as a revolt against being under submission of their own instincts and drives - or even more fundamental forces in this universe. The main character of The Vanishing jumps from a third-story balcony as an adolescent, to prove he is no slave of his own fear of heights. He wants to be able to do anything, according to the 'arbitrariness of his will'. Compared to this, the creation of a random word is a pretty modest deed of anarchy. Has the writer Krabbé accomplished which he lets his characters do successfully?

A first fundamental problem emerges: the randomness of an isolated, short sequence of numbers or letters is hard to assess. Only with longer sequence can anything be said about it in a statistically significant way. Compare an extreme case: of the binary sequence 01 it cannot be said if it is generated by randomness of an underlying systematicity. Fortunately, Batüwü Griekgriek is not such an extreme case, and it can be assessed to a certain degree.

According to the mathematical definition, randomness means absence of pattern. A series of numbers is random if it has no regularities. This implies that the series can not be abbreviated or rewritten to a shorter series. The technical term for this is informational incompressibility. A series that systematically contains repetitions (111111), alternations (010101) or symmetries (001100) – to name but a few elementary examples of pattern - is therefore not random. So Krabbé’s word seems to disqualify: it contains a repetition of the letter sequence: griek (greek), which is even an existing word in its own right.

The main question is from which ‘word container the word is drawn’. Which properties has the set of possible words? In case of number series we assume that all digits from 0 till 9 are allowed, in every possible order. With words this is more complicated. The smallest parts might not be separate letters, but syllables. Related to this is a more or less stringent criterion of pronounceability. A letter sequence like jklaafgyy lkolj does not qualify as a word candidate. Let’s assume that Krabbé’s word container was restricted to what is pronounceable. Note that the set of pronounceable words is bigger that the set of representative words, which by the way would be far more difficult to define. Regardless of exact definition, common sense of native Dutch speakers tells that Batüwü Griekgriek is not a representative Dutch word.

So in the container of random words, no total letter anarchy prevails. Considering the above, the repetition ‘griek’ is already less unlikely: it could be a single repetition and not a – very unlikely – fivefold repetition on letter level. So is the repetition just as likely as the repetition of, for example, the number 6 in a decimal series? No, not even close, because there are many more syllables than decimal numbers. But the exact probability depends, again, on the composition of the ‘word container’. Is the repetition frequency of syllables therein a faithful reflection of the repetition frequency of syllables in dutch words (I guess those are predominantly onomatopoeias and ‘primal words’ like mama). Or is the likelihood of repetition a blind ‘one devided by the number of syllables’ in the container? In the latter case the chances of a repetition of ‘griek’ would be very small, since there are thousands of syllables.

Table: Randomly generated binary sequences with a different alternation factor (0.5 is random)

So because of the repetition, Batüwü Griekgriek is not a very representative random word. I think Krabbé knows this and I suspect that he has chosen for a repetition for a different reason. Research shows the following: if people are asked to invent a random sequence of binary numbers, they shun repetition more than they should. The chance that they write a 0 after a 1, is 0.6 instead of 0.5. (see table above, which shows computer-generated, pseudo random series). This is a manifestation of what is called the gambler’s fallacy: the mistaken notion that the roulette wheel has a bias towards variation/alternation. I think that Krabbé has deliberately not wanted to fall in this trap. And he has even added a little extra on top of his flagrant repetition. He seems to say provocatively: ‘I know, this word contains a repetition. If you guys think it couldn’t be random, go lose your money in the casino’.

But maybe I am on the wrong track. Maybe the writer has truly written the first word that came to his mind on a bicycle tour. That is what he suggests in the quote. This spontaneous realization explains the repetitions in a different manner. Whenever people invent a fantasy rime or a nonsensical little song, repetition and alternation occur very often (tatee tom tom tatee tom). To put it in a different way: these words are in a container that has a lot of repetition and alternation in it. So in this context of spontaneous language production, the word Batüwü Griekgriek ís a pretty representative random word.

Was Krabbé honest when he wrote: ‘And all of a sudden, there it was.’? Wouldn’t it be to cheat his readers if he had considered many word candidates, altered a letter here and there, until he had finally found a word that met his standards? On the other hand, the ultimate deed of Krabbé-esque randomness is not one that comes easy. It often takes his characters careful planning. And isn’t it also essential to the process of writing? To ponder and rewrite endlessly, until finally something apparently spontaneous is on paper?


October 2006



  • is a site that explains about randomness and lets you generate true (not pseudo) random sequences by using atmospheric noise as input
Powered By Website Baker