However, this assumption is not correct, since the appearance of parasitic gaps in domains relatively inaccessible to ordinary extraction may remedy and, at the same time, eliminate a stipulation to place the constructions into these various categories. Conversely, the natural general principle that will subsume this case raises serious doubts about the strong generative capacity of the theory. It may be, then, that most of the methodological work in modern linguistics appears to correlate rather closely with an abstract underlying order. This suggests that this selectionally introduced contextual feature is not subject to irrelevant intervening contexts in selectional rules. Suppose, for instance, that a descriptively adequate grammar suffices to account for a corpus of utterance tokens upon which conformity has been defined by the paired utterance test.
see also: WikiPedia -- Chomskybot