Knots!
I was fascinated by this article, but find it pretty hard to visualize what the khipus look like, so I looked for photos. They don't really look like what I would expect, plus they look like they would be very hard to read!
Using a physical object to represent language is (or at least should be) in many ways more intuitive than inventing not only an encoding language (system of symbols, i.e. letters), but also a 2-d surface and tool to write these symbols on. I guess if you think of cave drawings as the origin of writing, then it would make sense why man elaborated written expression and invented new 2-d surfaces on which to write on, but i guess what I am trying to say is that I do not find it surprising that there was an attempt to achieve this with three-dimensional objects.