Automatic thesaurus extraction techniques are applied to computer-generated related word vocabulary questions. These questions assess and provide practice for an aspect of word knowledge found to be important for language learning. Automatic generation of such questions reduces the need for human authoring of practice materials. In evaluations with real teachers, most of the generated questions were considered to be usable in real classrooms. Also, performance of native and nonnative speakers on these automatically generated questions was similar to their performance on manually generated questions for the same words. This application of natural language processing techniques to English as a Second language education is a promising step toward automatically producing vocabulary practice and assessment materials for language learners.