Key points are not available for this paper at this time.
For many years it was asserted that language tests had a negative impact on teach ing and thereby on learning/learners. This impact has become widely known in language testing as 'washback' ('backwash'). A theory of washback is emerging and its domain being delineated; in this article the theory of washback is linked with the broader concept of 'impact' in educational measurement, and thence to the recent debate on construct validity associated with Messick.
Liz Hamp‐Lyons (1997) studied this question.