Abstract. In this paper, we propose a top-down approach for converting business processes information from corporate documents into controlled language. This proposal is achieved with a multi-level methodology. We first characterize document structure by using rhetorical analysis to determine relevant sections for information extraction. Then, a verb-centered event analysis is performed to start defining the typical patterns featured by business processes information. Lastly, morpho-syntactic and dependency parsing is carried out for extracting this information. This multi-level knowledge is used to define rules for converting the extracted sentences into a controlled language, which is intended to be used in software requirements elicitation.