La Disparition, by George Pérec, and conditional probabilities

As mentioned here (about dust on keyboards), E is undoubtedly the most important letter in French… So, what else do we use if we do not use the E ?

Here is the empirical frequency we obtained on two books, by Mauss and Durkheim (which seem to be representative of what we usually observe)

So if we look at the conditional distribution we have without letter E, we obtain

But does this conditional distribution can be interpreted ? For instance, if on your keyboard E is broken, will the distribution of letters used be like that ? We might think that the answer is no, because if we cannot use a letter, we adapt ourselves, and try to use words that do not contain E.

Actually it is possible to see how we would write if we try to avoid letter E : in 1969, Georges Pérec published a book “la disparition” where the challenge was to write a novel (with a story inside) without using letter E. If we look at the distribution of letters in that book, we have

We can see that we still need to use vowels: if we look at the difference between Pérec, and the conditional distribution obtained from Mauss and Durkheim, we have

i.e. if we cannot used E then we use much more A or I or even U.It is clearly a substitution effect: the French language needs a constant dose of vowels. In Mauss and Durkheim, 37% of the letters are vowels (17% for E and 20% for A, I,O, U and Y). In Pérec, A, I, O, U and Y represent now 30% of the letters…

Cite this blog post
Arthur Charpentier (2010, November 23). La Disparition, by George Pérec, and conditional probabilities. Freakonometrics. Retrieved May 25, 2024, from

3 thoughts on “La Disparition, by George Pérec, and conditional probabilities”

  1. Actually, Christian, Perec also wrote a univocalic novella: “Les revenentes” (written only with E’s as a vowel). To achieve his goal, he however used a rather liberal spelling (e.g. “également” becomes “hegelement” or “sendweech” instead of “sandwich”). Eunoia by C Bok might also be interesting in this respect…

    REPONSE: indeed, but he wrote it afterwards… in “La disparition”, he changed expressions but not words, so in some sense, my exercice was legitime here…. I don’t think it would be on “Les revenentes” (which is too artificial).

  2. Does your observation hold also in the opposite case? I mean that: if one wishes to write only with words containing the letter E, is the frequency of other vowels lower? You could try to observe it considering the other book by Pérec on the topic, Les Revenantes.


    PS: how did you manage to have the occurrence of each letters in a book? I was also asking myself about it when reading your post on Marc Lévy…

    PPS: about words and books, do you known the problem of the Federalist Papers? It could be also a nice subject for your blog..

    REPONSE: the R code for words was given in the post about Marc Levy. For letter, it is not that difficult…. you simply use

    textfile=scan("georges perec la disparition.txt", what="char", sep=" ") textfile<-tolower(textfile) M=NA for(i in 1:length(textfile)){ line=textfile[i] M=c(M,strsplit(as.character(line),"")[[1]]) } T=table(M)

    Please send me references about the Federalist Papers, I am interested….

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.