Finally, we return to the question of assessment's links with learning outcomes. We must ensure that what we test is not only relevant to what we have taught and what students have learnt, but also that the assessment process serves to promote that learning. When assessment has beneficial effects on teaching and learning, it is common to talk about 'positive washback', whereas if assessment fails to further the learning outcomes, one would see this as 'negative washback' (Alderson and Wall, 1993).

For example, if the final exam in a course on developing reading knowledge in Italian involved students in a role play of a job interview in an Italian firm, preparation for this specific exam task would do little to develop students' capacity to access articles and reference works in Italian - presumably the main aim of the course. Similarly, if an IWLP course promoting students' communicative ability in German were assessed purely via translation of a newspaper article on current affairs, students' efforts to learn relevant vocabulary would do little to help their functional use of German. If, on the other hand, in a course on the French political system, text and oral work in French led to student oral presentations on the pros and cons of the parliamentary system, then the assessment task would dovetail with classroom work, would provide a logical connection with and would follow on naturally from the process of learning and teaching, while serving to motivate student learning towards a clearly identified, relevant and attainable goal. This is an example of positive washback. (See also section 13.1.3 on learning outcomes.)

To summarize, the key things to consider in ensuring positive washback are:

• Test the knowledge and skills you want to encourage, not just those that are easiest to assess.
• Test the learning outcomes directly: if we teach students to write summaries, then they must be tested on that skill.
• Make sure students know what is expected of them in the exam or test.
• Test a representative sample of the syllabus by choosing items from a wide range of topics, rather than just one narrow area.

The holy grail of assessment is for all of these features to coincide, for assessment to be valid, reliable, transparent, practical and to promote positive washback. Sadly, it is rarely possible to satisfy all these demands simultaneously. Objectively scored placement tests (see Module 14, section 14.1.1.2) that assess students' writing and speaking proficiency and their command of vocabulary are likely to be valid, transparent and reliable but will almost certainly prove too time-consuming in the very hectic first days of the academic year. At best, we might give students a multiple-choice grammar test to provide a rough idea of which group we should allocate students to. Similarly, tests that promote positive washback, like the French oral presentations mentioned above, may prove to be unreliable and to lack transparency if students are given different sub-topics of variable difficulty or accessibility (eg the role of parliamentary committees or the electoral system might prove for whatever reason to be more challenging than discussion of the role of the president). In practice, compromise is likely to characterize all our endeavours to reconcile the five key features of assessment, but we must nevertheless always remain aware of them in devising tests and exams.