We can list a number of concerns about unseen written exams. These include:

• They do not do much to increase students' desire to learn: many students avoid modules that involve exams, going instead for options they may be less interested in but which are continuously assessed.
• They are not a good way of alerting students to what they really need to learn: students often have to guess at the assessment agenda, and their preparation and learning can therefore be unfocused.
• The amount of feedback that students receive about exams is not optimal: scripts are usually regarded as confidential documents, not to be shown to students.
• Scripts usually have to be marked in a rush and markers are often tired and bored when they get to, say, their 50^th script.
• Writing, marking, second-marking and moderating exams is a very time-consuming business.
• They tend to favour candidates who happen to be skilled at doing exams: some of the skills involved have little to do with language ability.
• They force students into adopting surface rather than deep approaches to learning and into rapidly clearing their minds of previous knowledge when preparing for the next exam.

(Adapted from Race, 2001a: 37-39.)

On the other hand, exams do have some clear advantages compared with a diet of pure coursework. We can perhaps summarize these as follows:

• They treat all candidates the same and thus avoid the problems of plagiarism.
• They force/encourage students to sit down and learn material thoroughly.
• If students' work is mostly continuously assessed, there is a tendency for them to concentrate only on assessed pieces to the detriment of their overall learning; this is less likely to happen if assessment is exam-based.
• When continuously assessed work is returned, there is the strong possibility that students will pay more attention to their marks than to the formative feedback provided by the teacher; with non-assessed coursework there is a greater likelihood of students concentrating on and learning from feedback.
• Continuous assessment, especially if it involves in-class tests, can seriously eat into teaching time which in some research-led universities is already under considerable pressure; end-of-module exams free up more of the semester for teaching and learning.

On balance, the flexibility of continuous assessment, the opportunity to introduce more varied tasks with potential for positive washback on classroom learning, probably tips the balance in its favour. Furthermore, since some candidates are known to perform below their ability in exam conditions, diversifying assessment allows all students to perform to their maximum potential. However, there is clearly a role for exams both as a summative assessment of language achievement and as a means of assessing students on an equal basis. For this reason most language modules nowadays rightly feature a mixture of the two approaches.