In fact, the idea is craftily clever. Their stemmer and parser can only stem and parse simple sentences, so if it can't process the sentence with a sufficiently high certainty, they flag it as too complex :-)
I don't know what technology they use in the editor. Also, I quit academia almost ten years ago, so things might have moved a bit since I worked on similar stuff.
But generally, English is one of the easier languages to process. Its morphology is simple (almost no declension, simple conjugation) and the training data for statistical methods are huge.
map{substr$_->[0],$_->[1]||0,1}[\*||{},3],[[]],[ref qr-1,-,-1],[{}],[sub{}^*ARGV,3]
| [reply]
[d/l] |
| [reply] |
Ever tried saying it to an average English speaker?
The lack of grammar keeps related phrases closer to each other which helps parsing a lot.
For free word order languages, grammar seems to help, but due to homonymy (or homography) you usually don't have a solid foundation to base the grammar on.
The most advanced system nowadays are based on Machine Learning, so there's no grammar involved at all, you just need large training data.
map{substr$_->[0],$_->[1]||0,1}[\*||{},3],[[]],[ref qr-1,-,-1],[{}],[sub{}^*ARGV,3]
| [reply]
[d/l] |
| [reply] |
