Friday's paper carried an AP story about researchers who found that opera singers, and especially sopranos, mangled the pronunciation of words in the effort to hit the notes. This comes as no surprise to most of us, and in fact you can show that there's bound to be a conflict between making the note pure and making the pronunciation clear.
Look at the Fourier transform of a vowel sound. Each has a distinctive shape; but the thing to notice is that none of them are a single frequency. A pure note is a single frequency. When a singer sings the vowel U at some note it sounds like that note because the bulk of that aforementioned shape centers around the pitch desired--but it isn't a pure note. It can't be, and still have the recognizable structure of a vowel.
Most of the time we don't care--unless the singer is at some extreme. Singing extremely high or low notes, or extremely loud ones, make it very hard to keep the vowel structure clear--by the definition of extreme. Unfortunately, opera singers have to sing loudly, and often with quite a range . . .