Natural language is the main presentation means in industrial requirements documents. This leads to the fact that requirements documents are often incomplete and inconsistent. Despite the fact that documents are mostly written in natural language, natural language processing (NLP) is barely used in industrial requirements engineering. The presented paper shows, how a natural language processing (NLP) approach can be integrated in a CASE tool. This enables the integrated tool to learn on the fly, which grammatical construction represents which model element. A valuable side effect of the proposed integration is tracing between the textual document and the constructed model. The presented integrated tool was evaluated in several case studies that have shown practical applicability of the tool. Keywords-requirements analysis, system modeling, natural language processing