Playing the Recording Once or Twice

Arguments for and Against

There is an ongoing debate whether to play the listening text once or twice. There are strong arguments in favor of each. Generally, I think the arguments for playing text only once are stronger.

Note that discussion refers to non-collaborative listening situations, which are the norm on most listening tests.

Below is a summary of the arguments I have heard. The reasons for playing the text twice are presented first, in ascending order of importance, and below each point is my response. In the second section, I present a number of additional reasons for playing the text only once.

Arguments For Playing the Listening Text Twice

Point One: From teachers, one often hears that if they allow students to listen a second time in class, then the test should do the same.

I find this argument very unconvincing: teaching is teaching, and testing is testing. We are trying to measure proficiency, not trying to help students to learn.

Point Two: Playing the recording only once places an undue psychological stress on the test-taker. As a result, many test users demand that the recording be played twice.

This argument has some merit, but it is a minor consideration. Psychology tells us that there is both facilitating anxiety and debilitating anxiety. We do not know whether the extra stress is debilitating—it might even enhance performance, even though test takers find it uncomfortable.

Point Three: In interactive listening situations, the listener usually has the opportunity to ask the speaker for repetition or clarification, but in listening tests, it is not possible to do that, and so we should play the recording twice to compensate for that.

Since we are talking about non-collaborative listening—such as radio talks or lectures. —this does not really apply. Most non-interactive listening texts build in redundancies, repetition and reformulation to take account of this, and we need to ensure that we do too.
Further, note that when listeners ask for repetition, they virtually never get that. Speakers rarely repeat what they have said, but generally re-state things in a different manner. And even in the rare case where the same words are repeated, the intonation is always totally different. In fact, in the real world, listeners never really get the same text twice.

Point Four: Listening test items are too hard if test takers hear  the text only once.

This is a serious concern, but it is not a reason to play texts twice. Rather it is an argument to consider text repetition as an important variable in item difficulty.
Research shows that playing the text a second time makes the test task much easier, and further, that showing the question to the test taker after the first listening, but before the repetition, makes the task even easier still.
Playing the text twice is a useful way of making fairly difficult tasks easier for lower-ability test takers. For example, when we want to use authentic texts that were really too challenging for the test takers.

Point Five: In most target language use situations it is not so vital that the listener understands exactly what is said, and comprehension is normally much more approximate than we realize. The testing situation is unnatural in demanding that the listener comprehend with a much greater degree of precision than is normal. Given this, playing the text a second time does not appear such an unnatural thing.

This is an important point, and should be taken seriously, since it relates directly to how we define and operationalize our the listening construct. However, I would deal with this issue by ensuring that the tasks are appropriate for the listening situation. Our items should focus on the main points of the discussion, rather than small detail.
Texts played once should have redundancy built in, so that test-takers will have more than one source of the information necessary to answer. As test development practice, we should ensure that any point tested appears in the text more than once, and we should not write items that depend on understanding one small detail, mentioned only once.

Point Six: In many testing situations there is a room full of people, possibly with inadequate sound quality, and there is always a chance of a noise disturbing the listener at exactly the wrong moment.

Again this is a very important point, and this argument should be taken very seriously. Obviously, if a test item requires identifying an item of information mentioned just once, then it might be a good idea to play the text twice—not just because of ambient noise, but because attention wanders and people get distracted.
I think the key is to ensure that the information targeted by the item occurs in more than one place within the text.

Arguments For Playing the Listening Text Once

Point One: This is how things are done in the real world, and the more we can replicate real-world performance, the more likely the test results will predict general proficiency.

Point Two: Listening is a cognitive process that has been widely researched, but we do not know whether listening to a text a second time involves the same processing skills as listening the first time. It may require different skills, and hence be testing a somewhat different construct.

Point Three: One of the most important of all listening skills is the ability to process the text quickly and automatically. I believe that this is the most important characteristic of listening comprehension as a separate language skill. The degree of automaticity is the single most important variable in second language performance testing. Further, listening is the only one of the four skills that directly assesses the degree to which language knowledge and language processing is automatized. Thus, I believe we should only play the text once, and if we play it a second time, we may lose a significant part of our construct.

Point Four: Another important listening skill is the ability to make inferences. In real world listening, we never get fully complete information, so we have to make inferences by applying our knowledge (of the language, the co-text, the context of situation and our general background knowledge) to the incomplete information we have. When listeners fail to catch something, they need to make inferences to fill in these comprehension gaps—this is a fundamentally important listening skill. If we play the tape twice, we will reduce the effects of inferencing on our tests.

Point Five: Finally, it takes extra time to play the text more than once, and that makes our tests less efficient, since there are less items in any one period of time. Playing the texts a second time increases testing time by about 10% to 20%. It is better to play texts once, and have students listen to more texts and answer more items. This increases both the content coverage of the test and the reliability.

In Conclusion: Although there are some very good reasons to play the text twice, and many reputable testing programs do so, I think the balance of the argument is clear. All other things being equal, we will get better information, and more of it in the same time, if we play the our texts only once.

However, we do need to take care when creating our tests. Playing the text only once does have dangers, and test developers need to take account of those when developing their tests.