The proposed research aims to develop and test a corpus-based method to automatic generation of learning materials in English and German. The study was conducted on the authentic language material of G. Orwell, A. Christie and E. M. Remarque fiction works. The paper describes the main research stages including the analysis of the lingvodidactic characteristics of the materials being developed, the study of technical specifications for a software solution, writing programming code of the modules for the corpus manager and testing the program. A wide range of scientific and specific methods are used such as analysis, modeling, programming for specific purposes, experimental implementation. The software development is based on Python programming language, PyQt graphic library and spaCy NLP-library. The procedure of Moodle XML files generation is noted. In the course of the research new functional units of the corpus manager have been created. Based on the developed program experiment results the paper considers new options which allow the user to create a list of lexical units automatically, generate exercises for such parts of speech as articles and verbs. New and successful results of corpus-based learning materials automatic generation are presented.