In textual knowledge management, statistical methods prevail. Nonetheless, some difficulties cannot be overcome by these methodologies. I propose a symbolic approach using a complete textual analysis to identify which analysis level can improve the the answers provided by a system. The approach identifies word senses and relation between words and generates as many rephrasings as possible. Using synonyms and derivative, the system provides new utterances without changing the original meaning of the sentences. Such a way, an information can be retrieved whatever the question or answer’s wording may be.