This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.
The testing of listening comprehension has undergone considerable changes in recent years as a result of a greater understanding of the processes involved in listening (Brown, 1990). There has been a movement away from the belief that listening comprehension is a one-way bottom-up process towards the theory that it requires a more complex combination of top-down and bottom-up processing. In other words, listening is not just the consecutive processing of sub skills starting from the lowest level to the highest. It requires processing at a number of different levels to which no fixed order can be attributed. The listener utilizes not only his knowledge of the language being spoken, that is, his knowledge of the grammar and vocabulary of the language, but also his knowledge of the outside world and how it relates to the topic at hand, as well as his interpretation of what has been said so far, in order to comprehend the message. The listener predicts about what the speaker will say next and uses his knowledge of the culture to help him understand linguistic complexities. He will therefore rely heavily on context and it is with regard to context that tests of listening comprehension have changed dramatically. There is now a recognition of the fact that "if we ask students to decode short decontextualized sentences, we are not testing listening comprehension at all but asking students to engage in a very unnatural activity which seems to be confined largely to the second language classroom", (Buck, 1988: 22) . As a result, it is becoming less and less common to see decontextualized utterances being used in tests of listening comprehension.
What appears not to have changed is the belief that testers should strive towards designing tests that are 'pure'. Tests that require students to respond in speech or writing are considered 'contaminated' because they are testing more than the skills of listening. Buck (1988: 33), for example, cautions against mistaking intervening variables and suggests ways to overcome the 'problems' of contamination. Yet if we look at the skill of listening as it is used in real-life, it will soon become clear that we rarely listen solely for pleasure or enjoyment. More often than not the listener is required to 'do' something with what he has heard: to respond to it in speech, take notes on it, or reproduce it in written form. Researchers are beginning to assert the fact that students should not only be tested on their level of comprehension, but also on their ability to use the information they have heard, even though they recognize that such more integrative tests may affect the reliability of the measure.
The Issue and its Solution.
The issue here is that to create a valid and reliable listening comprehension test, testers should therefore, sacrifice test purity in order to maximize task authenticity as there is really no way of testing listening without translating the feedback into a non-listening form. Hence, the following conundrum rings in each teacher of ESL's heads, 'How can I design a listening comprehension test which is both valid and reliable in a L2 setting?'
The solution lies upon the definition of second language listening ability. Unfortunately, researchers have as yet been unable to formulate an agreed-upon and workable definition of L2 listening ability. This is partly because the processes involved in L2 listening comprehension are so dependent on the context of the listening situation that the establishment of a global, comprehensive definition is probably impossible. However, Buck (2001) has created what he calls a "default listening construct". Buck defines L2 listening ability as the ability:
â€¢ To process extended samples of realistic spoken language, automatically and in real time
â€¢ To understand the linguistic information that is unequivocally included in the text
â€¢ To make whatever inferences are unambiguously implicated by the content of the passage
(Buck, 2001, p. 114).
Using this definition of L2 listening ability allows teachers and testers to create valid and reliable listening tests.
To process extended samples of realistic spoken language, automatically and in real time
In order to assess whether a language learner has the ability to process and comprehend spoken language, it is necessary to provide realistic spoken texts that are representative of real life language. Providing spoken texts that are representative of the real life language that test takers will encounter adds authenticity to the test, and can contribute to construct validity. But what exactly is "realistic" spoken language? Real life language is rapid and mostly unplanned, and includes false starts, fillers, pauses, and connected speech. Unfortunately, such language has generally not been used in L2 listening assessments. The language learner is usually presented with listening texts that have planned, scripted, and polished speech that is read aloud with precise and clear pronunciation on audio tapes specially prepared for foreign language classrooms and tests. This disconnect between the spoken language tested (and taught) in the foreign language classroom, and the spoken language experienced in real life person-to-person encounters is probably the greatest shortcoming of the way L2 listening ability is now tested. How can this shortcoming be addressed? As testers, it is imperative to include representative, authentic spoken speech on tests of L2 listening ability. This might require including a number of different listening tasks, with different types of listening texts of varying lengths and of different genres, which are representative of the types of spoken language the test taker is expected to be able to comprehend.
In addition, L2 listening tests should demonstrate that the test taker has the ability to process language automatically, in real time. Because of limited processing capacity, a less proficient L2 listener will often be overwhelmed by incoming aural input, and a breakdown in comprehension can occur. Thus, there is a need for the listener to automatize the listening process, and consequently there is a need to assess if the listener can indeed comprehend spoken language automatically in real time. This presents a dilemma for testers, however, in determining the number of times that an audio text should be presented in a testing situation. Ideally, the audio text should be presented one time to the test takers, because this is generally how authentic spoken language is encountered in real life. Traditionally, however, audio texts have often been played two (or sometimes even more) times for listeners on language tests, for some very legitimate reasons. In real life, a listener can ask for clarification, or ask the speaker to repeat what has been said if the listener has not understood. In addition, in a testing situation the listener does not have the breadth of contextual knowledge that he or she would have in a real life situation, and playing the text a second time (after the listener has gained important contextual knowledge from the first listening) allows the listener to overcome this shortcoming. There are also memory issues involved, in that a listening test often requires the listener to remember more things than often required in real life language use, and thus a second playing of the text alleviates some of these memory issues. And finally, testers need to take into account the issue of test takers' feelings and attitudes about different aspects of the listening tests, which can have an impact on their performance on that test. Invariably, test takers report that hearing the text a second time is helpful for them, and they also feel that it is fairer in a testing situation than playing the text only once. There is no easy answer to this issue, although one way around it is to vary the number of times the text is played. Having a number of different listening texts on a test allows the tester to present some of the texts one time, and some of the texts two times.
To Understand the Linguistic Information that Is Unequivocally Included In the Text
This is the least problematic aspect of testing L2 listening ability. Traditionally, tests have included items that assess the test taker's ability to comprehend specifically stated information presented in audio texts. These types of items are usually easy to create, and are generally very reliable and unambiguous. However, some listening tests have included only these types of items, which is problematic, and a threat to test validity.
To Make Whatever Inferences Are Unambiguously Implicated By the Content of the
Unlike testing a listener's ability to comprehend explicitly stated information, assessing a listener's ability to make inferences is a much less clear-cut issue. Although some very basic word-, phrase-, and sometimes even sentence-level inferences are basically unambiguous, more global inferences that include pragmatic inferences about the message a speaker is trying to convey are much more problematic. Unfortunately, such inferences are rarely, if ever, totally "unambiguous." Different listeners will make different but still appropriate inferences from the same text. Obviously, this presents serious problems for testers. If there is not one correct answer, how can the ability to make appropriate inferences be tested? This dilemma probably explains why the writers of many listening tests have avoided testing listener's ability to make meaningful inferences. However, the process of listening is dependent on the listener making inferences in order to construct an interpretation of the incoming text. Brown (1995) states that more than 90% of what might be stated in a spoken text need not be stated, but can be assumed to be inferred by the listener. In many ways the process of listening is the ability to make correct and appropriate inferences. Therefore, even though it is often quite difficult to create reliable and unambiguous items that test a listener's ability to make inferences, it is necessary to include these types of items. Although such items often force the tester to rely on expert judgment to score and interpret items, thus reducing the objectivity of the test, these types of items are necessary if one wants to create a valid L2 listening test.
In a nutshell, creating reliable and valid L2 listening comprehension tests is not an easy process. But because of the importance of listening in language learning and communication, it is imperative that teachers and testers invest the resources needed to make quality tests. By including a number of representative listening texts of different types and genres, and by including items that indicate the test taker's ability to not only comprehend explicitly stated information, but also the ability to make meaningful and appropriate inferences, testers can create useful L2 listening comprehension tests.
Brown, G. (1990). Listening to Spoken English . Longman: London and New York.
Brown, G. (1995). Dimensions of difficulty in listening comprehension. In D. Mendelshon & J.Rubin, (Eds.), A Guide for the Teaching of Second Language Listening (pp. 59-73). San Diego: Dominie Press.
Buck, G. (1988).'Testing Listening Comprehension in Japanese University Entrance Examinations', Japanese Association of Language Teachers Journal 10/1 & 2: 15-42.
Buck, G. (2001). Assessing Listening. Cambridge: Cambridge University Press.