I agree with the main point of the article but there are parser generators like OMeta http://tinlizzie.org/ometa/ that help (more than ANTLR/Lex/Yacc) to think in the grammar from a higher level point of view without paying a lot of attention about ambiguities and grammar restrictions. Sure, OMeta is slow but it offers some solutions to the problems presented in the article.
I was so impressed with OMeta when I read about it the first time that I decided to write my own implementation for Python. I got a first version released[1] but I sort of ran out of steam before I could take it any further. Luckily, there's another OMeta-based parser library for Python called Parsley which seems to be better-maintained (it's already hit v1.0!)
