Testing listening involves a variety of skills. At the lowest level, it involves discrimination among sounds, discrimination among intonation and stress patterns, and comprehension of short and long listening texts. While the first two are part of listening, they are, of course, not sufficient.
Items with full sentences have the drawback that testees can make use of not just phoneme discrimination but also knowledge of grammar and lexical items. If one of the words that the alternatives does not fit grammatically or semantically in the sentence, then testees who realize that have an advantage.
This type of discrimination item is one that can be used for diagnostic purposes to see whether students have particular problems with distinguishing between phonemes. However, it does not give the teacher any information about the testees' ability to comprehend spoken English.
Ability to understand the meaning of difference in intonation can be tested by having the testees listen to a statement and choose from three interpretations of the statement. For example, testees might be given the statement "Vera is a wonderful musician" and be asked to decide whether the speaker is making a straightforward statement, a sarcastic statement, or a question. Since the context is neutral, however, it is sometimes difficult to avoid ambiguity. In real communication, listeners make use of their background knowledge, the context, etc., as well as the intonation to help them interpret the communicative meaning of an utterance.
Spoken:
I had hoped to visit you while I was in New York.
Written:
A. I was in New York but did not visit you.
B. I will be in New York and hope to visit you.
C. I visited you in New York and hope to again.
D. I am in New York and would like to visit you.
Another type of item is one in which the testees listen to an utterance and choose from among four responses the most appropriate response. In that case, the testees are not being asked directly what the meaning of the utterance is. Rather they are being asked to show that they know what it means by showing that they recognize an appropriate response. This tests both the testees' listening ability and their knowledge of appropriate second pair parts of adjacency pairs.
An example of this type of item is as follows.
Spoken: Would you mind if I visited you next time I came to New York?
Written:
A. Yes, of course. I'd love to visit New York.
B. No, I don't really think that much of New York.
C. Yes, I would. You can come any time.
D. No, not at all. I'd really love to have you.
(At a slightly higher level, both the first statement and the responses can be spoken, but in that case, it might be better to have only three responses, since it would be difficult to keep all four responses in mind.)
In this example, the testees need to know that "Would you mind if I..." is a form used for asking permission, and that a positive response begins with "no (I don't mind)." Because this type of item requires two different types of information, there is a certain amount of controversy about it. Some theorists argue that it is not a good item type, because it requires these two types of knowledge. Testees could possibly understand the utterance perfectly well but not know how to respond to it. Also, since the utterances are presented in isolation and out of context, the situation is not realistic. However, this type of item can be useful if these limitations are kept in mind. It is a more communicative type of task than many listening tasks, so it may have beneficial backwash effects, and it is relatively easy to administer.
The advantage of using visual materials in this way is that they can be used to test listening alone without involving other skills very much (if the tester considers that an advantage)--though it is impossible to entirely eliminate the use of other skills. However, they do not reflect the sorts of listening done in the real world.
One way of using talks in listening tests is to have the testees listen to the talk and then fill in the blanks in a written summary of the talk. The words chosen from blanks should be ones that the testees cannot figure out from the context of the summary, without listening to the talk, but they should also be ones that are related to the main idea of the talk, so that filling in the blanks does not require remembering small details of the talk.
Another way of doing this involves giving testees questions to answer as they listen to the talk. These questions can be short answer/completion, multiple choice, or true/false. The difficulty with using short answer or completion questions is that they require the testees to both read and write while they are listening, something that can be difficult even for native speakers. Multiple choice questions may require a lot of reading, something that may also be a problem. True/false questions may be the best type for this type of task, since they require relatively little reading compared to multiple choice questions and relatively little writing compared to short answer/completion questions. Also, true/false questions can also have a "no information available" option, meaning that the information required to answer the question is not included in the talk. This decreases the amount of guessing and reduces the element of chance.
Another type of task that can be used is a chart that the testees fill in while listening to the talk. Answers in some of the blanks in the chart may be filled in for the testees. Testees should be given the chart in advance of hearing the talk, and be given time to familiarize themselves with it and make sure they understand what is expected of them.
Finally, testees can be allowed to take notes while listening to the talk, and then use the notes to answer questions after the talk is over. Depending on the length and complexity of the talk, testees might be given a list of the major topics included in the talk in order to help them in taking notes.