Well, there are different opinions on this, but for cinematic and other regular medias, one rule of thumb is to never break the illusion. And one way to keep from that is not to let dialogue stray from the center, be it the center speaker or the entire front row of speakers doesn't matter, unless absolutely necessary. That's the panning aspect in angles.
When it comes to frequency response, it's a lot harder to say a certain method always does the trick. Theoretically, every mix would preferably be mixed with perfect respect to distance, reverbation, and sometimes even atmospheric moist, but practically it's often not a good idea. Mainly for two reasons, but in essence it all comes down to intelligibility. When I work I mostly divide the dialogue in forground, intelligible background, and ambient background. In the foreground I have no perspective whatsoever, and how far that stretches depends totally on the framing. The intelligible background is actually more of a mid, and here I have a special system to keep clarity though three dimensional placement, mostly based on reverb and a frequency treatment which is next to impossible to explain here right now, but basically I keep all processing as sparse as possible, and I rarely use panning. In the ambient background I couldn't care less! Gotten muffled? So much the better! Less risk of distraction! And I use full panning, reverb and EQ as I see fit for the very scene :-)
The first reason I mentioned is that you never really know where people will see/listen to your material. A mix that may sound absolutely marvelous with everything in perfect distance to each other and perfect clarity in one system might very well sound absolute crap in a cheaper system, with dialogue muffled, thin, or in other ways indistinct. The most important part of the soundtrack, with absolutely no competition, is the dialogue and that demands clarity. For the dialogue that actually is part of the story that is, for ambient voices I have total freedom to do as both I and the director wants, as long as it still sounds decent on an artistic level on smaller systems! But actually, the issue with frequency response doesn't really have to be a problem, it becomes a problem because of the source material used.
Which leads to reason according to me number two, unless a total ADR or dub from a very good studio, there are no way in hell the sync dialogue will keep good enough consistency to allow for this kind of treatment. As long as the sync dialogue is in perfect focus with perfect distance and angle these things are quite easy, but as location is basically a warzone, it never is. No matter how good a location recordist one may be, you still have to keep out of frame, keep from shadowing anything in picture, avoid reflections and ambient noises, not to mention you might have to keep moving while on take. In respect to all this, there are LOTS of things that has to be matched up as it is to work at all, and all this processing is only inaudible to a certain point.
A good example was actually from this summer! I had a gig recording a feature, and this day we where at a theater for some extremely important scenes. At this...place, the ceiling was FILLED with lights, and as the play that was scheduled after us already had done all the lights and noone knew how to (or dared) save the settings, we only had TWO options; power and intensity. First actor; With a whole xxxxxxx ceiling flooded in light (actually the actors looked absolutely great, but to try to fit a boom there...), two cameramen, a floor with several squeaky zones I had to keep track of, ventilation, and the actors constantly moving over the entire scene...and there was no way whatsoever to do any ADR later on, I actually pulled it off and it sounded absolutely smashing! To the day I die, I will still say: I did it fanfare! It was the absolutely most physically demanding thing I have ever done on film, and thanks to all the years as a karateka, it was doable.
Then came the next actor. And the one after that. Eight actors all in all as I remember. I nailed it all, but the problem was that without exceptions every actor had totally different conditions from the others. They all sounded great, but they all also had different sounds depending on distance and position, and that's why this is such a good example of this. In editing, all the voices MUST sound the same, or it will fail completely. That will mean several of the voices will need a lot of EQ and processing, and that takes time. Should I also give the voices perspective, I will have to make a second set of processing for every voice in the scene, and I will need a way to pan between the two sets. That takes a lot of time, and time ain't exactly cheap, so a good middle way is to just automate a send to a good reverb, and you get depth with minimal risk of bad response!
Jeez, looking at this reply I'm actually quite thankful you didn't ask about sound effects and ambiance as well, I treat those things differently as well ;-)