"Purple air sounds like rage." Is grammatically correct and passes a wide range of linguistic filters, yet for some reason it has zero meaning. Still read it a few times, it feels like there should be some meaning there. This get's at a core problem with language. Language is based around the assumption that something meaningful is trying to be communicated so we rarely bother with much precision. You can refer to ideas as physical objects but parsing into something meaningful requires a lot of context.
Redness seems to refer to some sort of external stimulus, but both redness and red are really internal qualifications of some external stimulus. So, I can point you to a red filter that looks at color codes and compares them to FF0000. If I then run that filter on FF0000 I get an internal state of 100%. But what that does not create something with emotions and your linking them in that sentence so what's wrong? I would suggest it's not the fact that the code does not produce emotion that's the problem it's your assumption that anything that understands redness must include an emotional context. Still, if you want both a test for redness and an emotional response you could code a simple AI with competing goals who responds emotionally to increasing amounts of red in it's environment.
Granted, you could then argue that such a simplistic representation of internal state is to simplistic a response to cover the full range of emotions. But a that point your not talking about 'redness of red' in the abstract but getting into mimicking your personal emotional responses. Which could be done to some reasonable level of accuracy.
PS: I could also have completely missed what you where trying to communicate, but I hope that line of reasoning demonstrated what I was talking about. Namely that in the context of AI you can translate abstract philosophical ideas into a meaningful context but they stop being interesting somewhere in the process.