Vignette Rendering Bug
For some reason, WordsEye chose the wrong vignette for this scene.
A jumping kangaroo is on the ground.
I think it's because you named it "jumping kangaroo" and it ignored the word "jumping" as a modifier and as a result just found a regular kangaroo. Vignette names should be single words (or a lexicalized phrase like "grizzly bear") so that sentences referencing them can be parsed normally.
One possible way around this (in the future) would be to allow multi word vignette names if they are enclosed in brackets. That way they could be treated as a single unit when parsed. E.g. "A [jumping kangaroo] is on the ground." But, for now, it has to be a single word.
Wordseye recognizes some verbs in vignettes (like "burning"), but not "jumping." Why does this happen?
That's because "burning bush" is lexicalized (i.e. treated as a single word). It's a particular species of bush...similar to how "grizzly bear" is a particular type of bear. To clear up the confusion, we can add a warning when you name vignettes if a multiword name is not recognized.
Note: It might be possible for vignette attributes (e.g. "jumping") to be parsed themselves when assigned. And then those attributes could be recognized and the corresponding vignette used when the input text for the scene matches it. But it doesn't do that now.
I have written an "adaptive grammar" that could solve this problem, but it's written in Prolog instead of Common Lisp. https://github.com/jarble/adaptive_parser
Type your own scene