Unparsed entities must be gathered from the XML input document at the time
  when the DOM is built. To achieve this our parser must parse the document DTD
  (if any) and store all data of type NDATA (not XML data) in a
  hashtable in the AbstractTranslet object. All the compiled code for this
  function needs to do is call a method in the translet for retrieving the
  value for the requested element:
  |  |  |  | 
|  | 
    public String AbstractTranslet.getUnparsedEntity(String entityName);
   |  | 
|  |  |  | 
The translet will use the supplied entityName to look up the
  value in the hashtable and then leave the string value on the stack for the
  element that called lang().