Non-lexical language trees

As has been pointed out here, language trees based on
vocabulary lists, like the Grey-Atkinson one under discussion, miss
significant relationships like the satem-kentum split.
I presume phonological and morphological phenomena could be
described as discrete entities amenable to corrresponding
computation, although it would be a lot tougher job than just
looking at Swadesh word lists. The list of items for I.E languages
certainly couldn't match the 2,449 lexical items of G & A, but would
there be enough to make computation meaningful?
Dan Milton