I have been teaching both English as a second language and English literature courses for several years. In this period I have noticed that sometimes, although by no means always, students who had excellent performance in class didn’t do well in tests, whereas those who were not active or were not very willing to participate in discussions in class have better results than might have been expected. I have analyzed the reasons for this occurrence, and have come to certain conclusions that I would like to emphasize in this article.

In teaching English as a second language, the textbooks used in many schools in Macedonia are structured in such a way that the focus is on vocabulary and grammar, whereas in the end of each lesson there is also writing practice to check whether the new vocabulary and grammatical rules, learned in the lesson, can be used by students in more comprehensive ways. When I started teaching, I used samples of tests that were already prepared by past generations of teachers. Most of them consisted of one, or, at the most, two sections, each section including one type of questions, for example, either the whole test consisted of yes/no question, or multiple choice questions, or a third case was a test including a combination of both of them.

From the results of such tests it was evident that even students, whose knowledge of the English language was not on the level prescribed by the course, could do well and get an excellent grade. In this way, the best and the worst students had more or less the same results, so obviously such tests were not reliable and accurate.

In order to improve the testing, I then decided to make it harder, so that it really reflects whether students have learned for it or not. Constructing a test that would give students the task to write and essay on a given topic seemed like the right decision. However, there was now a different problem – some superb students did not get the results they otherwise deserved, according to my assessments of their overall class performance. Apparently, they did not have sufficient knowledge of the topic given or were not keen on writing about it. Giving several topics to choose from improved the reliability of this type of test.

Now there seemed to be another problem – the students lacking skills for essay writing, but with a relatively good grasp of the language, at least such required on the level of the course they attended, did worse than they deserved, according to estimations based on long-term observations.

The next test included a combination of different sections, that is, different types of exercises to address variety of skills, and, in the next few semesters, this proved to be the best choice – it showed the most accurate results. For example, some students have outstanding knowledge of the grammatical rules of the English language, rich vocabulary, excellent pronunciation, show a high level of understanding of given texts and ability to discuss them, but are not so skilled in essay writing. That is why a combination of at least three or more kinds of exercises, which measure different aspects of the command of a certain language makes the test more reliable, more accurate and more valid.

Another point for the advantages of this type of testing is that in the classes where English is taught as a foreign language, the objective of the course is to help the students learn the language rather than make them experts in essay writing, which could be a problem even in the mother tongue. An essay that is not well structured, in the sense of not having a well defined introduction, body or conclusion, does not reflect lack of knowledge of the foreign language. That is why, in creating tests, it is essential to have in mind the purpose of the course, and, consequently the purpose of the test – what it strives to demonstrate.

This is precisely why I mentioned the literature courses in the beginning of this article. A literature course has very different aims and objectives in comparison to an ESL course. It is precisely the writing skills that are measured here. The reading of literary works and attempts to interpret them entail close analysis and providing arguments on the part of the students. To discuss a literary work, it is not enough to only speak the language in which the work was written, but it also requires ability to scrutinize it and, possibly, evaluate it. That is why a test consisting of yes/no questions will not do for this type of a course. It is the way that the given essay is structured and composed that counts as a valid measure in this type of courses. Nevertheless, the choice of a few topics, rather than just one, will be more reliable because students feel more comfortable, and also more encouraged, when they see that they have a part in the decision-making during the test.

This does not mean that only by giving essay writing tasks we can see how well the students perform. Experience has taught me that there are such students who love literature and read carefully all literary works presented in the syllable and even additional reading materials, but have some difficulties in writing a lengthy essay. In order not to neglect the skills of such students by only giving an essay task in the examination process, it would increase the accuracy of the test to include other types of questions, such as, for instance, give extracts and ask students to discuss them in a page or two, which is less than the quantity of written material required for an essay.

That is why I am strongly in favor of tests with several types of exercises in them. Such tests tend to be more reliable because they measure different skills. This would given all students equal opportunities to show their stronger skills. It is also of an essential importance to adjust the test to the requirements of the course. If we want to measure language skills, multiple choice questions or reading comprehension questions regarding a given text might be better indicators for knowledge than an essay. On the other hand, if we want to know how well someone is writing then we cannot get an accurate measure of their ability with yes/no questions.