Right, mine is a pretty specific use case. However, if a thesaurus lookup can figure out that nah is close enough to no that it should just normalize the respon

Right, mine is a pretty specific use case. However, if a thesaurus lookup can figure out that nah is close enough to no that it should just normalize the response to no, that's a win for the user. (Scraping urban dictionary to act like a thesaurus would be a decent way to decipher txt spk and euphemisms as well.) Yes/No questions could be special cased to accept more oddball responses off the bat..'for sure', 'totally', 'no way', etc.