The aim of the paper is to give an idea of methodology and of technical solutions used in the design of an experimental syntax-oriented program to process Polish texts; the program is currently being developed by the author. A classification of Polish words is presented. It is based on the notion of syntactic category and it covers in principle most of the inflexional and syntactic features of words. Polish syntax is to be described by means of a formal grammar; the description takes into account some newer results concerning the syntactic function of particular word classes. The formalism used to describe syntax is the Colmerauer's metamorphic grammar. The program will be implemented in PROLOG, the powerful programming language in which metamorphic grammars are directly available. The output of the program will be the surface syntactic structure of each sentence. Next, a subset of Polish is specified. The subset consists of sentences to be processed by the program. Finally, some details of the program are given.