Comprehensively identifying all reserved keywords in each dialect, then handling them, isn't something we really have the bandwidth to pull off...