In this paper standard tagging mechanisms are discussed and partly criticized. We propose a semantically and morphosyntactically enriched mechanism in which many of the shortcomings of standard POS-taggers are eliminated. Therefore rule based disambiguation steps are presented. They include mainly a specification of contextually motivated verbal subcategories. We need this fine grained system for better preprocessing of requirements texts which have to be highly explicit and non ambiguous. This linguistic preprocessor can support an interpretation tool used to extract semantic concepts and relations for the requirements engineering step in software development.