and back in the early 80's, Scientific American -- then still a pretty good amateur science magazine, had a computer column.

One of the columns was devoted to writing in BASIC– a mock language generator.

the idea was we all –well make that most– sort of recognize a language– even if we can't read it.. French looks different than English, and Latin and Italian. Why?

The rule of each language where analyzed.. (All words in English must have a vowel– a, e,i,o,u are acceptable vowels.. We use th, ch, sh, ght, kn, sm, cl, and other letter combinations. But don't use ts (used in japanese and greek) or schl (except in names from German)– and we don't use cz (many slavic languages..) The average word in english is 6.x letter long, there are no two letter words with the letter u (but a, e, i and o work for two letter words, as does y. ) and sentences had on average 7 words– but could be as short as one... There were lots of rules.

once you had the rules, you could set up a random text generator–and make it follow all the rules.. about every third word was a real english word.. But it was non-sense.

the columnist actually gave the code for a basic program to generate "almost english", and gave some of the parameters for almost latin.. And almost french. I don't know if Scientific American has all of its archives available on its web page.. But I remember typing in the code and playing with program.


my other obsession