To provide a short answer to your question, version control systems can version virtually any time of file, including rich text documents. Your ability to effectively inspect the diff depends on the internal representation of the document in the versioned file.
"Rich Text" documents can be versioned in Git, like any other file. The right question is probably "how easy is to visually inspect the diff between two versions".
The answer depends on how the file is encoded and stored. If it's a plain binary file, such as an executable, then git will be able to store the different revisions, but the diff view will be meaningless.
In order to take advantage of text diffs, the files must be stored as some kind of text file. It's important to keep in mind that using a text file doesn't necessary mean you are limited to provide "Rich Text". After all, the HTML markup is a good demonstration.
As far as I remember, the latest Office versions adopt a storage format (docx) that in reality is a Zip package. If you unzip the content, part of the information are stored in XML files (I may be wrong, it has been a while since I used one of these files).