I didn't mean to suggest that singers, when improvising harmonies real-time, overtly think about theory along the lines I mentioned above, any more than people explicitly think about nouns and adjectives when they speak. It's more an innate sense of what works harmonically and what doesn't; of when to follow the predictable and when to break away; of when to space the parts closely, farther apart or at unisons or octaves; of when a line needs more melodic or rhythmic interest. These things happen either intuitively or as brief flashes of conscious deliberation--for those who have developed that sense. Ian described the actual process in better terms, and Shelley's deconstruction illustrates some of the overall thought processes for giving an arrangement dynamic shape. But Michael did ask for some theory, so I obliged.

As Shelley's lyrics indicate, when planning a set arrangement, singers do give conscious deliberation to "theoretical" matters such as harmonic structure, parallel movements (or, as Ian remarked, deliberately non-parallel movements, which shows the same awareness), voice proximity and crossing, placement of dissonances and suspensions. They just understand the theory (musical conventions and behavior) more directly and convey what they mean more directly either by concrete example or by informal rather than theoretical terms.

As very young children, we learn language my emulating the sound patterns we hear around us, making copious mistakes and finally arriving at ever more successful patterns that work. As adults, we learn languages more efficiently by combining emulation and experimentation with a more conscious analysis of the grammar rules described by others. Learning to harmonize and improvise happens in the same way; understanding the theory can reduce the bungling and allow you to talk about musical concepts with more precision. But you don't have to learn the theoretical descriptions of harmony in order to "get" how harmony works. Most of us picked up our intrinsic understanding of harmony at the same time we picked up our first language. And most of us still haven't formally studied music theory to much depth, so trying to talk to groups using theoretical terms may be like delivering a lecture in Latin--a strain for both sides, and largely counter-productive.

We also need to remember that music theory has great limitations; it can describe the most usual harmonic structures and progressions we encounter, and a bit of why we respond to them as we do, but it presents a very incomplete picture. It's a meta-description of music, and no substitute for experiencing the music itself.