A corpus-based grammar for ELT Dieter Mindt (Berlin, Germany) The year 2000 will see the publication of the book An empirical grammar of the English verb system. The grammar is fully based on machine-readable corpora containing more than 240 million words. It is learner and teacher oriented. Original research was carried out for the description of each item. All examples provided in the grammar are authentic. The approach is inductive: from language data to grammatical generalization. Wherever possible, frequency data are supplied. This new feature is especially important for the language learner, the language teacher, and the author of teaching materials. Learning and teaching can thus be directed to what is central and important in language use. The grammar makes use of a number of grammatical categories which evolved from the analysis of the data. There is a new paradigm of English verb forms as well as a new definition of catenative verbs which leads to a novel distinction between finite and non-finite verb phrases. The description is based on a new analysis of the English verb phrase. All verb patterns are described as exponents of a coherent structural description of the English verb phrase. Within this model there are no exceptions. All instances of verbs and verb phrases can be explained as cases of rule-governed grammatical behaviour.