Traditionally, mathematics assessments are graded numerically. This tends to favour the correctness of answers over their quality and can lead to poorly understood or non-constructive feedback.