The efficiency of a syntactical parser can be measured according to speed and correctness factors. One reason of the inefficiencies is the border problem, which is when a constituent is mistakenly assigned to a neighboring one. This paper introduces a formal tool that reduces the border problem, thus increasing the speed of the parse and decreasing the ratio of wrong results. The basic idea of the "strong rules" model is that natural languages contain definitely unambiguously parsable structures (humans always prefer this parse). With marking the borders of such structures, the defective expansion of the surrounding constituents can be limited. Implementation issues and other application areas of the "strong rules" model are also discussed.