The advantages of this approach are that it usually ensures brief student responses and thus means the tester can cover more of the text(s) in the test. Unlike when answering multiple-choice questions (see section 14.4.5), the student has to seek the answer himself/herself so that it is more likely the correct answer will be arrived at for reasons of comprehension than any other reason (eg guessing in multiple-choice exercises). The approach allows the skills of inferencing, sequencing and comparing disparate parts of the text to feature in the testing process. The only real drawback is that unless care is taken in formulating questions, marking can be made more difficult and even unreliable by the presence of more than one acceptable answer, especially where students have to infer the answer.

Short questions can be answered in either students' L1 or the FL. It is best not to mix the language of question and answer, so as to avoid translation impinging on the task and confusing the purpose of assessment (see also 14.4.4). Clearly, FL answers to FL questions are more challenging for students if, as should be the case, they are required to avoid simply repeating verbatim the original text. Such questions might thus be thought better for more advanced levels of proficiency, while L1 answers to L1 questions would be more appropriate to lower-level assessment.

Answers in the FL also raise issues for the tutor marking the work: is it the factual content alone that is being marked and is therefore the quality of the language to be ignored where it does not prevent communication of the message? Or are some marks to be awarded for the quality of language used as well as for content? If the latter, it is only fair to let students know exactly how their answers will be marked, and to make clear to them if, and how, they will be penalized for simply reproducing stimulus material, ie copying chunks from the text. On balance, introducing a quality of language mark threatens to confuse the purpose of the assessment, that is to measure comprehension, and it is better not to do it.

Occasionally, under the misnomer of comprehension, one comes across the curious practice of students being asked to supply FL answers to questions on an L1 text. This requires a lot of groundwork on FL vocabulary and structures and is a very challenging task. In effect, it is a variation on prose translation, albeit one that allows students greater freedom in the language they choose to express their meaning; clearly, it has nothing to do with FL comprehension.