Hi Paul,
I've spent more time on these type of questions then is good for me... both for your question and in the past ;) .
If your just interested how it works I'd recommend to have a look at LUKE (code.google.com/.../luke/).
It's an open source viewer for Lucene indexes. (NOTE: you need to have Java running for this).
apart from it being a quick tool to search your index; check what's actually in your index it also helps explains your syntax.
I've tried about a dozen combinations :), but I think it keeps omitting special characters. I think it has something to do with the Analyzer type set on the indexed fields, which ignores special characters.
Since the NodeAliasPath is a base Kentico field I'm not sure you can/want to update that. Kentico's default search probably removes paths from the results instead of the in the query itself.