Why study non-repetition ?
My aim is not to show that some sounds behave according to a cooccurrence constraint, described with OCP or OT or some other refined technique (which often are efficient, admirable and good devices). I want to show that all linguistic elements, sounds, clusters, syllables, words, syntagmas, etc., have a tendency to follow a non-repetitive principle, in a formula it could be xx -> xy /yx |T|, where 'T' stands for tendency. Both variants of the process xx -> yx |T| and xx -> xy |T are possible and in practice found in many languages.
It might seem more likely that the element to be changed should be the second of two identical elements because the mind would not react before both items present themselves to the mind. In most cases however, as it seems, during the elaboration of the phrase to be said, the identity of the two elements presents itself before it is actually enounced - and consequently the identity is avoided, changing anyone of the two elements, the first or the second. This aspect is more complicated than it would seem in the first place.
This way to put the issue is important for more than one reason. One is that you could sometimes want to get hold of some facts about language without first establishing a specific hypothetic mental construction, "theory", about language, and without using linguistic structures to demonstrate non-linguistic ideas, e.g. psychological theories, or in order to show that during the production of speech the "morphological processes" don't "have access" to phonological information or vice versa. I believe that such studies are very important, but not that those points of view be always obligatory, and certainly not the only interesting ones.
Another reason is that, studying a huge amount of processes from a large number of languages, I try to understand to which degree non-repetition is universal, to establish some types of non-repetition processes, and present to the colleagues these facts, so that they, each with her/his specific competence, can correct what needs correction in my work, and further investigate into the single concrete non-repetitive functions.