Currently, the only parser tests check that correct source text is parsed successfully. We should also have tests verifying that incorrect source text produces expected parse errors. These "negative" tests make a natural companion to the positive testing done so far.