A model of a generalized document storage and retrieval system is proposed. The model consists of six subsystems (or blocks): logical processor, selector, descriptor file, locator, document file and analysis block. These subsystems function in a partial environment defined by the user and data blocks. Proceeding from a verbal description, a functional representation of each subsystem is developed. The functional representation describes not only what is done but also, to some degree, how tasks are accomplished within each subsystem. An immediate result of the functional representation is the definition of a metalanguage for identifying some necessary characteristics of higher level languages used in the implementation of information storage and retrieval systems. To illustrate the usefulness of the metalanguage, a comparison is made of three languages - FORTRAN, PL/l, and SNOBOL - for implementing document storage and retrieval systems. The functional differences among the blocks of the system are apparent, and the implementation of efficient systems appears to require a multi-language approach.