Cool idea, but I don't think synonyms can be graphed because graphing as you have done above represents synonyms as linear combinations of the words on the axes and fails to capture the finer nuances of word distinctions.

Of course, you could always represents these finer nuances by additional dimensions in your space, but orthogonality would be an issue. That is, I don't think you can come up with a set of orthogonal dimensions in which all words could be adequately mapped. I could be wrong, though. It would be interesting to see these mapped in 3D-space or some higher dimensional space that is projected onto 3D or 2D.

Perhaps it would be like this:
Synonyms
denotation = vector(connotation1, connotation2,... connotationN)
I think nuances are limited to a finite number

Concept words
concept-X = f(concept-Y)
values on dimensions like scores on a psychological scale["rate this parameter from 1 - 10"]
graph is locus of points
time represented by animation

Graph of Locus defined as new word