CDATA sections aren’t represented in the XPath/XQuery data model - the information is discarded by the XML parser. You can’t tell whether data was originally written in a CDATA section or not, any more than you can tell whether a space was written as a true space or as the character reference . Essentially, CDATA sections (like character references) are regarded as a keyboard shortcut when entering XML, not as something that carries information.
This has always been true of XPath 1.0 and it is equally true of XQuery 1.0. If this information is significant to applications, you will have to find some other way of representing it in your XML markup.