When there's an unknown token, there are two options: to say that something should be there instead (span = 0..1), or to say that something else was expected (span = 0..0). By default, Chumsky generates the first in all its internal parsers, which seems to me incorrect, it probably depends on the task.