Correct and incorrect sounds are handled by semantics. So far correct/incorrect
are all just one 'concept', but with multiple sounds so they will be selected
at random. 

Eventually, these concepts should be broken up into separate concepts, but be
sure to keep the same semantics.