plain literals without language tag compare xsd:string in RDF

Is there any utility to having these being disjoint classes? It would  
seem to me that it would be more sensible to say that any string that  
doesn't have a  language type or a datatype is inferred to be of type  
xsd:string.

Did this situation come about because it was easier to make the RDF  
semantics look cleaner, or was there some principled reason for  
making the distinction?

-Alan