N3 rules: literals as subject and object in built-in functions

http://www.w3.org/2004/12/rules-ws/paper/94/
In "Experience with N3 rules", it says built-in functions "are simply represented as RDF properties",
and gives the example:
   { ex:d test:point ?x.  ?x math:sin ?y } =>  {...}

Presumably ?x and ?y for math:sin would be a number literal.  But an RDF graph only
allows the object to be a literal, not a subject.
1. Am I right that RDF does not allow the subject of a triple to be a literal?
2. If so, how can a math built-in function that uses literals be the predicate of an RDF triple?

Thanks,
- Jeff