I want my agent to be modular, of course. But I cannot just start to build one module and then the next. For language recognition, in particular, it is not smart to start building a syntactic parser, based on some theory, and only when that is finished, to start thinking about a semantic analyzer.  There may not be a theory of semantics following your chosen theory of syntax.

I am now looking for knowledge about:

• A natural language parser (for which a Dutch grammar is available)
• The output of this parser should either contain semantic information in itself, or should serve as the input of a semantic analyzer (if that is the name for it) from the same school of thought.