A COMPUTER PROCEDURE IS DESCRIBED FOR PARSING RUSSIAN SENTENCES WITH A CONTEXT-FREE RECOGNITION GRAMMAR. THIS IS THE FIRST PROJECT UNDER A PROGRAM FOR THE INVESTIGATION OF SEVERAL ASPECTS OF NATURAL LANGUAGE DATA PROCESSING BY FORMALIZED METHODS TO DETERMINE THE USEFULNESS OF FORMALIZED LINGUISTIC TECHNIQUES IN PRACTICAL LANGUAGE DATA PROCESSING APPLICATIONS. EVERY HYPOTHESIS WAS TESTED AS A RUNNING COMPUTATIONAL PROCEDURE BEFORE ACCEPTANCE AS A WORKING PRINCIPLE. THE HARDWARE USED INCLUDED A BUNKER-RAMO MODEL 130 (AN/VYK-1) COMPUTER. THE PROGRAMING LANGUAGE WAS A VERSION OF FORTRAN IV. AN IMPORTANT CHANGE IN THE GRAMMAR AND ALGORITHM OF THE SYSTEM HAS BEEN THE INTRODUCTION OF "GRAMMATICAL VARIABLES" AS COMPONENTS OF GRAMMATICAL LABELS. THE TERM "VECTOR-SYMBOL PHRASE GRAMMAR" IS USED TO DISTINGUISH IT FROM THE MORE USUAL PHRASE "STRUCTURE GRAMMAR." THE BASIC ALGORITHM AND THE FORM AND FUNCTION OF THE GRAMMATICAL VARIABLES ARE DISCUSSED IN DETAIL. THE RESULTS INDICATE THAT THE MODIFICATION OF THE "VECTOR-SYMBOL PHRASE GRAMMAR," WITH NODE SUPPRESSION, MAKES IT POSSIBLE TO UNDERTAKE WRITING A PHRASE STRUCTURE GRAMMAR FOR WRITTEN RUSSIAN SUITABLE FOR DATA PROCESSING APPLICATIONS. (KL)