the study reports the results of the exploration of a machine-readable corpus of brazilian portuguese. the corpus was collected from news distributed on the internet. the news items themselves consisted of excerpts from newspaper stories and tv transcripts. the focus of the paper is on the description of selected language features needed for the production of teaching materials for private portuguese classes in britain. several lexical and grammatical items are described using corpus linguistics tools in what amounts to pioneering work on corpus analysis of portuguese. the paper concludes that guidance provided by existing reference materials such as textbooks, grammars and dictionaries are inadequate since these sources are not based on samples of authentic language.