Comment by mmatants

Apparently, Peter Naur (the N in BNF) wrote this up nicely back in 1985 in "Programming as Theory Building": http://pages.cs.wisc.edu/~remzi/Naur.pdf