From an emic perspective boundaries are parametric.  They are not maintained from within unless in tension with some outside force. {<inc> James on consciousness}  Interactionist approaches get tangled on the very boundary they seek to remove in the act of critiquing structuralist approaches from the outside.  Boundaries are attributed to the interaction process and then dismissed as insignificant because they are its product. Emic approaches, which the interactionists espouse, need be concerned with these boundaries only when they arise. Interactionists can dismiss structuralist boundaries that they do not interact with and structuralists can dismiss boundaries that are not part of their structures.  Both use emic approaches, but they ask different questions.

     If two candles light a space, no one would claim "there is only one candle because the boundaries between the light from one candle and the light from another are not clear."  In terms of illumination, the boundaries of any space are a function of what, if any, surfaces would reside within that space.  If there are no surfaces then there are no boundaries. In Labov's etic model, the 'light community' of one candle would be the aggregate set of illuminated and shadowed aspects of all the surfaces in the (clearly delimited) confines of a space. <soclab03.00> Interactionists would note that the observer (or its instrument) is also a part of the illuminated and shadowed surfaces and cannot observe itself from outside without the aid of yet another observer with the same problem (ad infinitum), making one set of surfaces always present for observation to occur, rendering Labovian definitions impossible to complete and candles lighting a surfaceless space impossible to observe.  Their conclusion is that models of speech and definitions of speech community are absolutely relative.  

     The issue that structuralist school and its epistemological descendents (encompassing the constellation of Saussure- Bloomfeld-Chomsky by standard sociolinguistic definitions) is concerned with is not the effects of the light on the surfaces in the space, but how the candles go about producing light. Inferences can be made by perhaps covering one light and examining the effects that has on the lighted surfaces, but there is no etic concern for drawing a boundary between one set of illuminatory effects and the other.  Communities are recognized to cluster into what Wittgenstein refered to as "family resemblances" {<inc> jump to Wittgenstein, Rausch, G. Lakoff} but the object of study is how the candles work rather than what the effects of the candles are.  

     This by no means denies that the lit candles have effects, not all of which are concerned with illumination, but it does ignore this as a theoretical issue.  Interactionist (absolute relativist) critiques observe that the effects cannot be ignored because these hypothetical candles cannot be observed without introducing an effected surface--the observer.  They do not however, self-reflect upon the effect that their own observations have on the dictum of absolute relativity (or the effect of that self-observation; or the effect of this observation, and so forth...)  

     The principles of calculus provide the means of pragmatically, though not absolutely, accommodating this paradox. What is needed to stop the derivative spiral of navel gazing or the integral trap of unwarranted assumptions is to treat approaches as _approaches toward limits_ rather than as _paths to destinations_. {<inc> go to multiple perspectives}.  This approach privileges the continual process of _modelling_ over any discrete modelling event or its resultant model.  It renders models as provisional and subject to revision upon confronting the findings of other approaches.  It recognizes absolute and relative as directions rather than destinations. <cogcon01> --two solutions--claculus and multiple perspectives--modelling rather than models, with conclusions provisional rather than laws.