Tape a 30 second
conversation between a husband and a wife, and there are no "headers" or
"pages". It's a different ball game.
Almost all forms of communication begin as structured content in the form of thoughts. You mentally structure what you want to say into sentences, you want parts of those sentences to be emphasized, etc. Then, depending on the medium you want to present those thoughts in (speech, literature, etc), you convert those abstract concepts into things like inflection and pauses for speech, and periods and italics for literature. In my understanding, XHTML/XML is a way of recording that pure structured information before limiting it to the constraints of a specific medium. It is not to record that information after it has been constrained to speech.
Also, grouping headers with pages is flawed logic. Headers have a semantic meaning, while pages are, once again, a constraint of certain presentation mediums.
