My example was actually of the existing WordML format for Word 2003. The docs can be downloaded here:
[link|http://www.microsoft.com/downloads/details.aspx?FamilyId=FE118952-3547-420A-A412-00A2662442D9&displaylang=en|http://www.microsoft...D9&displaylang=en]
There are binary chunks in that format (it is a single xml format unline OO.org); things like graphics, VBA projects, OLE objects, and some other special Word features, but most of it is actual individual elements and attributes with actual primitive data types in text form.
Below is a preview of the schemas for the new default document formats for Office 12:
[link|http://www.microsoft.com/downloads/details.aspx?FamilyId=15805380-F2C0-4B80-9AD1-2CB0C300AEF9&displaylang=en|http://www.microsoft...F9&displaylang=en]
Note that these formats are much like the OO.org ones. They are a number of XML files bundled with graphics (in standard formats) and some binary chunks (like OLE embedded objects and VBA projects) compressed into a ZIP file.