Their perceptions ARE real, as real as ours are, with our poor nonsynesthetic senses.

From what I've read, kids with this "condition" learn pretty fast not to talk about it. The "curlicues in the chicken" guy had a friend who was a researcher in synesthesia who didn't know that he (the speaker) was synesthetic until he served him the too-lemony chicken and got that involuntary statement.

The researcher I mentioned earlier was theorizing that everybody is synesthetic to some degree, it's just one of the things we learn to ignore as infants, like shape changes of the same thing ("constancy") and so on. He was testing non-synesthetes ("normals") and finding strong correlations, like if you ask 100 people what color the letter E is, you get a lot of dark blue and very little pale yellow or anything.

Then there is the ten ten ten experiment. Try this on a group of people: tell them you're going to ask them a series of questions and they should answer as quickly as they can, without thinking. Then spend at least several minutes asking quesions all of whose answers are "ten" (this only works in English as far as I know, I'd like to replicate it in French!). What we do is ask what's 9+1? 8+2? 7+3? 6+4? 5+5? 4+6? and so on, being very careful not to ask something the answer to which is NOT "ten" - then after all that, really about 3-4 minutes of saying "ten ten ten" (since the questions are short) - ask them to name a vegetable. They will name an ORANGE vegetable. You'll get a lot of carrots, pumpkins, and maybe some squashes. VERY VERY interesting. I've seen this done several times, it replicates nicely with native speakers of English.