Actually, the Franklin et al. PNAS study shows no such thing. All their work shows is that adults have a left-hemisphere bias for color categorization while infants 4-6 months have a right hemisphere bias. They go on to speculate that because language processing occurs in the left hemisphere, language acquisition forces a categorization scheme upon us that didn't exist prior to language. And from that, they further speculate that language-using adults may have different color experience than pre-linguistic infants.

This is all mess, though. Basically the authors are claiming that learning color words creates color categories that aren't innate. Okay, but why think that linguistic categorizations are radically different? Do the infants sort colors themselves differently (some red as blue, say) and not simply use a different part of the brain to track the same color properties out in the world? Infants and non-linguistic animals successfully track color properties, which probably means that they arguably possess some non-linguistic (non-conceptual?) category scheme. Language mainly lets us talk about things when they're not around and in more detail; why conclude that language always changes what is already there?

More importantly, why think that differences in where information gets processed in the brain translate into different conscious experiences — that is, that babies (consciously) see pure colors while adults' (conscious) color experiences are "refracted" through language simply because language happens on the left side of our heads? And what would it mean to (consciously) see colors purely or As They Really Are? After all, humans are trichromats and have eyes sensitive to only certain wavelengths of light unlike, say, some birds that are tetrachromats. Which means it's pretty tough, if not in principle impossible, to know what a pure color — meaning, presumably, color apart from any perceiver — would be. Add to that all sorts of research showing that humans can track all sorts of things nonconsciously, which means that evidence about tracking bias reveals little about the nature of conscious tracking experience — or the difference between being sensitive to differences in color as opposed to what it's like for one to see colors. (The phenomenon called "blindsight" is an incontrovertibly cool example of nonconscious tracking.) Last, how could they determine whether infants experienced colors differently than adults (or whether two adults experienced colors differently, for that matter)?

Sorry for the long comment, but it strikes me that both the Wired guy and the PNAS folks don't really have a clear idea what they're talking about mainly because they haven't clearly and cleanly drawn their own categories — conscious v. nonconscious seeing, color properties v. color experience, linguistic v. non-linguistic tracking/sorting, and so on. To be fair, the article itself is far less sexy and provocative than the Wired write up of it (as is usually the case), but it's still muddled in many important respects.