Famous psychology experiment used to demonstrate the automaticity of reading. Subjects were shown words for colors printed in different-colored inks, e.g. green (this doesn't work if your hyperlinks are green, in which case I apologize) and asked to say the color of the ink (not the word printed in that color). On average, it took significantly longer for subjects to identify the colors of the inks used for the color words than they did at identifying the colors of random shapes or nonsense words printed in colored inks (although the latter caused interference as well). The delay in response and related errors are attributed to a phenomenon called "Stroop interference" (the semantic content of the words, which are read automatically, interferes with correct identification of the colors).
In a bilingual Stroop test, the color words are printed in one language and the subject must say the ink color in another language, i.e. the correct response to paars would be whatever the color your hyperlinks are in, although "paars" is the word for "purple" in Dutch. Interference still occurs in this case: its effects are greatest when subjects are equally fluent in both languages and, especially when color words are similar across languages (i.e. "red" and "rouge" in English and French, respectively).
My cognitive psychology and bilingualism classes; I can't find a reference for the original Stroop test but the first article I read on the subject was Preston and Lambert's "Interlingual Interference in a Bilingual Version of the Stroop Color-Word Task", reprinted in Experimenting With the Mind: readings in cognitive psychology, edited by Lloyd K. Komatsu (Brooks-Cole Publishing Company, Pacific Grove, CA: 1994).