I could have sworn I once saw a query published for this.  Is there a way to
just get the list of words per document with no counts?

So if I want to create a word count histogram for all of our documents I
have to do something like:

select * from OPENROWSET('MSIDXS', 'catalog';'';'', 'SELECT PATH FROM
SCOPE()')

and then for each document returned, ivoke the correct iFilter for each
document, process the chunks and extract the words and compute the counts
myself?  In this case is there any way to find out for each word what
'forms-of' evaluates to?