A consequence of the approach just outlined is that the fundamental error of regarding functional notions as categorial is, apparently, determined by a descriptive fact. Furthermore, this selectionally introduced contextual feature delimits problems of phonemic and morphological analysis. If the position of the trace in (99c) were only relatively inaccessible to movement, the speaker-hearer's linguistic intuition may remedy and, at the same time, eliminate the strong generative capacity of the theory. So far, the appearance of parasitic gaps in domains relatively inaccessible to ordinary extraction is not to be considered in determining the requirement that branching is not tolerated within the dominance scope of a complex symbol. Thus the systematic use of complex symbols raises serious doubts about the extended c-command discussed in connection with (34).
see also: WikiPedia -- Chomskybot