Evaluation and Generation of Language Tests Based on the Predictability of Words in Context

Task of automatic test generation is not new and became popular both among computational linguists and teachers of foreign languages. Fair amount of time is constantly spent by teachers on creating tests. During the past two decadeso in order to minimize the effort and become a helpful tool for teachers some automatic language tests generators have been already created. These instruments vary in different ways, for instance in terms of supported languages, effectiveness and skills for practice. However, just a few of them are freely available and most of them are one language oriented. This project is aimed to create an instrument that will generate language tests for a given text. The advantage of this instrument is that it is easily scalable to another language when language corpora and morphological parser of a good quality is given. This instrument would be useful both for teachers and students learning foreign languages themselves, for example for the ones who want to practice specific skill.