Singing is really hard. You have to do all the things an instrumental player does and this diction thing, too. Not only that but diction functions in various ways.
1. Correctness. Each singer must pronounce the language as a native speaker would. Some languages are a lot fussier than others. The French are very fussy. In German there is a whole language to represent correctness: Hoch Deutsch. When I lived in Germany, all stage actors and television personalities spoke hoch Deutsch. Except in Switzerland. Living in Ulm allowed me to view TV from Austria, Germany and Switzerland from the comfort of my living room. This all leads to the idea of diction police who only worry about this part.
2. Understandability. One should never allow oneself to mistake these two things. Elisabeth Schwarzkopf was always a target of the diction police, but when I hear her singing, for me every word is easily understandable. For me this is far more important than item 1. One of the reasons I love Jonas Kaufmann's Winterreise so much is because it is so easy to understand.
3. Vocal technique. The sound and placement of the vowels and ones ability to move smoothly and easily through the consonants is a powerful force in developing proper technique. This is too hard to explain in a blog post.
4. Phrasing. Diction is also a tool in the phrasing tool box.
The objective is to achieve all four at the same time, not an easy task.
When you make vowels in any language, your tongue mounds up to separate the lips from the pharynx and separate your throat into two chambers front and back. Where the tongue makes its mound determines which vowel it is. Other things happen. You might open the throat into the nasal cavity in back to make a French nasal vowel. You might do different things with the tip of your tongue and your lips. All these things can change the vowel.
As a gross generalization I would say that the back chamber controls the color of the voice and the front chamber controls the clarity of the language you are singing. You can do both at the same time, but it isn't easy to learn.
In 2005 I wrote:
"Sometimes when you see photographs of recording sessions, there is a language coach sitting there with the singers. In the conflict between tone and correct pronunciation, tone should win. There is a school of vocal technique that bases its methods on vowel modification. So having someone there correcting your vowels could actually throw the whole thing off. So where is the vocal coach who is correcting the correcting of the language coach?"
Opera companies can have language coaches. I think this is possibly the explanation for why the singers at the Bayerische Staatsoper are the easiest to understand in German.
The article about Jonas Kaufmann on the diction police blog is fun to read. He actually seems to conclude that music perhaps on occasion tops diction. Here is my explanation of the article:
[ɛ]/[e] EH/AY without the diphthong; [œ]/[ø] UH slightly rounded lips/UH much more rounded French sound; [ɔ]/[o] AW/OH. Closed means narrower lips. He discusses only the first pair of sounds in the word “réveiller” [re vɛ je] RAY VEH YAY again all without diphthongs. A diphthong in IPA would include an additional symbol. In the second performance he sings RAY VAY YAY. The guy likes the second version, thus screwing up all his previous opinions. I always remember my conversation with the French people on Twitter and how in French they like Jonas Kaufmann better than anyone. Better than Sophie Koch who actually is French. This problem is created because it's Jonas. Which makes it funny.
I notice listening to the second version [this refers to films embedded in the article] that the aria has “réveiller” two times in succession. He sings the first one as described and the second one is like the other film. Curious. He gets to the end of his argument and concludes that music wins. Music does win.