Introduction

In the previous chapters, we saw that performance on grammar tests can be influenced by many other factors besides grammatical ability. In fact, test scores can vary as a result of the personal attributes of testtakers such as their age (Farhady, 1983; Zeidner, 1987), gender (Kunnan, 1990; Sunderland, 1995) and language background (Zeidner, 1986, 1987). They can also fluctuate due to their strategy use (Cohen, 1994; Purpura, 1999), motivation (Gardner, 1985) and level of anxiety (Gardner, Lalonde, Moorcroft and Evans, 1987). However, some of the most important factors that affect grammar-test scores, aside from grammatical ability, are the characteristics of the test itself. In fact, anyone who has ever taken a grammar test, or any test for that matter, knows that the types of questions on the test can severely impact performance. For example, some test-takers perform better on multiple-choice tasks than on oral interview tasks; others do better on essays than on cloze tasks; and still others score better if asked to write a letter than if asked to interpret a graph. Each of these tasks has a set of unique characteristics, called test-task characteristics. These characteristics can potentially interact with the characteristics of the examinee (e.g., his or her grammatical knowledge, personal attributes, topical knowledge, affective schemata) to influence test performance. Given the potential impact of test-task characteristics on performance, it is important for test developers to understand the individual characteristics of the tasks they use and to follow systematic procedures for designing and developing tasks that will elicit the best possible manifestations of grammatical ability.