CHAPTER TWO LITERATURE REVIEW
This chapter reviews the issues in listening texts and listening strategies first and then the related studies. The issues in listening texts, including variations of listening texts, test effects, and text effects, will be discussed in section 2.1. The issues in listening strategies, including the classification of listening strategies and the studies of listening strategy use, will be addressed in section 2.2. The related studies will be introduced in section 2.3. Finally, a summary of the chapter will be given in section 2.4.
2.1 Issues in Listening Texts 2.1.1 Variations of Listening Texts
In fact, quite a few researchers mentioned the variations of listening texts. Some of them focused on selecting appropriate listening passages. For instance, to increase the awareness of testing L2 listening comprehension, Thompson (1995) offered a lot of useful information about choosing listening texts. According to her, some features of aural passages should be kept in mind because these features might affect the comprehension of listening texts. These features included oral extent, visual support, length, content familiarity, speech rate, pauses, and other dynamic speech phenomena.
Similarly but more elaborately, Dunkel et al. (1993) claimed that the leveling variables of text difficulty included content imagery, contextual support, syntactic complexity, length, topic familiarity, video support and organization1. T. C. Lin (2000) also made some suggestions about how to choose listening texts. In his opinion, the things that should be paid attention to were difficulty levels, interest, relevance, oral extent, sentence structure, vocabulary, and speech rate. Among them, difficulty levels,
1 Besides the variables listed above, Dunkel et al. (1993) also mentioned other variables such as redundancy, repetition, propositional density, cultural proximity, dialectal variation, lexical difficulty, clarity, speed, participation (opportunity to modify input), and social frequency.
interest, and relevance were the major concerns. He further indicated some factors affecting difficulty levels of listening texts. Visual support, for instance, was particularly helpful for lower-proficiency learners (Mueller, 1980). Besides, texts differed in the degree of drawing listeners’ attention. Some text types like dramatic segments containing both actions and conversations could keep listeners alert longer than news reports, speeches, or lectures. Moreover, content familiarity was also found to affect text difficulty. As to sentence structure, there was still no conclusion about its effects. However, he mentioned that “Glisan (1985) found that longer, modified sentences were actually better understood than shorter unmodified ones. He also found in his advanced Spanish class that word order may affect the comprehension of speech” (p.230). In terms of vocabulary, texts with familiar words were easier than those with technical terms, idioms, and cultural allusions. He reminded that while choosing texts for lower-level students, one thing had to be kept in mind was that the key words should be recognizable or could be inferred from context. However, it was noticed that even frequently used words could become difficult to comprehend when used in a different context.
Similar to T. C. Lin (2000), Anderson (1977) and his colleagues also pointed out that interest played an importance role in listening comprehension. They looked into the effects of interest and background on listening comprehension. It was proven that
“listeners’ interests and past experience strongly affected the way in which they interpreted the passage” (cited in Dunkel, 1986, pp.102-103).
On the other hand, Shohamy & Inbar (1991) mentioned that according to Rubin
& Raforth (1984), the oral extent was a criterion in choosing listenable materials. That is, the more orally-oriented a text was and the easier it was to comprehend. Chaudron
& Richards (1986) also found that listening texts with more conversational styles and discourse markers signaling major propositions, transitions, and emphases were easier
to comprehend. In addition, Thompson (1995) stated that the oral extent of a listening text affected comprehension. Oral texts were featured by short clauses, loosely strung idea units, repetitions, and pauses, while written texts contained longer, more complex, and more densely packed idea units. She contended that “It has been demonstrated that text closer to the oral end of the continuum yield higher scores on listening comprehension tests than passages closer to the written end” (p.24). She found that the comprehension performance in conversations was better than that in expository passages on the listening section of the ETS Comprehensive Russian Proficiency Test.
As T. C. Lin (2000) summarized, oral texts for listening comprehension tended to receive higher scores on listening comprehension tests than texts closer to written language.
The features of oral texts were mentioned by Heaton (1989) as well. He stated that spoken language was featured by redundancy, repetition, hesitation, pause, and false starts. Unlike written language, spoken language was generally organized in clauses instead of sentences, made more use of connecting words such as and, but and so, and seldom contained complex sentences. Besides, in comprehending listening
texts, it was general meaning that was remembered instead of the actual words. Thus, he pointed out that it was not necessary for listeners to feel frustrated simply because of the failure in understanding every word. The important thing was to follow the gist.
Additionally, he claimed that context played a crucial role in comprehending spoken language because “we usually use language in a certain situation for a particular purpose” (Heaton, 1989, p.43). In other words, the comprehension of listening texts relied heavily on contextual information.
Similarly, Thompson (1995) pointed out the importance of context and purpose of listening comprehension. According to her, listening usually appeared in context in the real world, which could reduce ambiguous interpretations and help infer the
meaning. Besides, most of the time, listeners listened with purposes. Thus, they could decide what to pay attention to, and the ways to listen. Likewise, Dunkel (1986) stated the importance of context in comprehension. She stressed that “effective communication depends on whether the listener and speaker share a common
‘semantic field’” (p.103). In other words, getting listeners acquainted with the context was prerequisite for success in listening comprehension. Brown (1990) also contended the significance of contextual information to comprehension because context gave clues to make proper guessing and minimized the range of prediction.
Different from the studies mentioned above, some researchers looked at the factors affecting listening comprehension from a schema-theoretic perspective. For example, Long (1989) stated that “Anderson and his colleagues have identified two types of schemata essential to comprehension. Textual schemata, or ‘knowledge of the discourse-level convention of text’(p.2), relate to the general format or outline followed by specific types of texts….Content schemata, on the other hand, are derived from the individual’s life experiences…” (p.33). Similar remarks were made by Dirven et al. (1985). According to them, the previous knowledge required for listening comprehension consisted of interpersonal matters, world knowledge, and knowledge of other texts. They used the term “intertextuality” to refer to the fact that the
“production and reception of a text involved knowledge of other texts” (p.3).
Still some researchers specifically pointed out that text type would influence listening comprehension. Meanwhile, they tried to classify the types of listening texts.
According to Dunkel et al. (1993), text type was an important component of listening comprehension assessment. In their opinions, there were various types of listening texts, including academic lectures, conversations (casual exchanges, interviews, panel
discussions, telephone conversations), debates, speeches, newscasts, narrative stories2, etc. These text types could further be classified according to rhetorical function, transactional versus interactional language focus, or levels of formality. In addition, Dirven & Taylor (1985) stated that there were three major text types: descriptions, narrations, and argumentations. Each text type represented a particular cognitive schema. In descriptions, there was descriptive schema; in narrations, there was action schema; in argumentations, there was persuasion schema. Lack of the knowledge about these schemata might affect listening comprehension to a certain degree.
What’s more, Anderson & Lynch (1988) gave their opinions about how texts affect listening comprehension and the classification of text types. As they pointed out, these factors affecting listening comprehension could be divided into three categories:
the language type, the task or purpose, and the context. As far as the language type was concerned, there were four major features determining the difficulty levels including the organization of the information, topic familiarity, explicitness of the information, and input type. In terms of the organization of the information, to be specific, the order of events was influential for narrative texts, while the offer of an informative title and the sequence of illustrations following main points were helpful for the comprehension of expository texts. As for topic familiarity, it was recommended to choose texts on familiar topics, but a text which highly interested listeners did not necessarily guarantee good listening performance. When it came to the explicitness of information, it had been found that redundancy benefited more advanced listeners while short and simple texts were more suitable for listeners of lower proficiency. Besides, texts requiring inference would be more difficult for listeners than those with all necessary messages. With regard to input type, Brown &
2 There were still other types of listening texts such as instructions/directions, advertisements, announcements, songs, jokes, poetic lyrics, scripted dramatic pieces, and sermons.
Yule (1983a) proposed a framework for classifying spoken texts based on their experimental findings with teenagers. In their framework, listening input could be classified into three broad types: static, dynamic, and abstract3, with the difficulty levels increasing from left to right. That is, the difficulty levels of spoken texts would vary according to the “complexity of relationships between the things, people, events, and ideas referred to by a speaker.” (cited in Anderson & Lynch, 1988, p.54). Besides, within each text type, difficulty levels might vary with the amount of confusable elements.
These statements mentioned above were all supported by the findings of Brown et al. (1985). They worked with L1 English-speaking students and found that the organization and types of listening texts did affect listening comprehension4. Since the present study aims to explore test effects (picture-format tests vs. non-picture-format tests) and text effects (statements vs. dialogues), in the following section I will further review the issues.
2.1.2 Test Effects
Picture-format tests are different from non-picture-format tests in several ways.
First, picture-format tests can avoid involving other language skills and simply test listening ability. In fact, many researchers (cf. Dry, 1986; Tapia, 1996; Thompson, 1995) claimed that while developing listening tests, listening comprehension should be tested only and not to involve other skills. For instance, Thompson (1995) mentioned that “From a psychometric perspective, the language of response is a source of measurement error because we cannot determine how much of the variation
3 A static text refers to the text in which the relationship between items tends to be fixed such as description and instruction. A dynamic text is a text in which there might be changes in the scene, time, people, or their relationship such as storytelling. An abstract text is the text in which the expression of opinions or concepts is the main concern.
4 There are two major findings in the study. One was that narrative texts were easier to listen to and recall than expository texts. The other was that the narratives in which the events were arranged in chronological order were easier to recall than those with disrupted sequences or flashbacks.
in the scores is attributable to listening comprehension, and how much to writing or speaking ability” (p.28). Therefore, she suggested the use of non-verbal responses and recommended that they should be particularly useful for listeners at the lower proficiency level. Also, Dry (1986) stated that it was suggested that listening tests using picture-choice formats be better than those using reading-choice formats so that listening performance would not be distorted by reading ability. Tapia (1996) emphasized the important role that pictures played in comprehension and strongly recommended the use of pictures for checking comprehension. He stated that
“comprehension does not go any further than understanding a piece of language, whereas answering comprehension questions requires production and entails different strategies” (p.41). His own experiences showed that pictures could avoid problems resulting from production.
Picture-format tests can also more readily provide contextual information to facilitate comprehension (Huang, 2004; Oxford, 1993; Wright, 1989). According to Wright (1989), pictures contributed to interest, motivation, a sense of the context of the language, and a specific reference point or stimulus. Ruhe (1996) stated that it had been established that visual contextual clues such as pictures and video helped activate background knowledge and thus improved comprehension. What’s more, in the comparative analysis of the listening test items of GEPT, STEP, and PETS by Huang (2004), it was explicitly indicated that picture-format listening tests could contribute to better performance and that listening tests without the aid of contexts could result in poorer performance.
Moreover, picture-format tests can get listeners more motivated and attentive and lower their level of anxiety. Cheng (2000) mentioned that “…research has shown that visual messages can raise listeners’ motivation and attention levels and improve their comprehension of gist (Baltoca, 1994; Rubin, 1995)” (p.29). Wright (1989) also
agreed the contribution of pictures in increasing interest and motivation. Besides, according to Tapia (1996), pictures appeared more motivating than traditional measurements for comprehension. In addition, in the study of the validity and reliability of listening tests for entering college, Wu et al. (1999) stated that picture-format items were designed as a warm-up exercise to reduce the anxiety of listeners and it was proven that such a text type scored the best among the four text types.
However, picture-format tests are often criticized for their limitations. First, it is hard to totally get rid of pictorial ambiguity. Dry (1986) stated that “the degree to which naturalistic or realistic language may be sampled at upper levels of the test may be constrained. Harris (1969) points out in this regard that unless the picture items are on the easy side, they tend to be extremely difficult to illustrate simply and unambiguously” (pp. 21-22). That is, picture-format items may carry more information than is intended by the speaker or even mislead the interpretation. Second, such a format may have the limitation in testing some structures. For example, Dry (1986) mentioned that “…certain structures such as yes-no questions, and certain contrasts such as the difference between “could” and “would” cannot be easily tested”
(p.22). Third, pictures can sometimes be distractors to listening comprehension. For instance, the subjects of Chung (1994) reflected that multiple still pictures were somewhat distracting. Fourth, the use of visual materials in testing listening does not
“reflect the sorts of listening done in the real world” (cited in Kitao & Kitao, 1996, p.3). Last but not the least, to create proper pictures may be quite demanding or time-consuming for many test-designers (Cheng, 2000).
2.1.3 Text Effects
A common classification of listening texts is based on the number of speakers. If there is only one speaker involved, such texts are more like transactional talks,
containing more explicit language, complex sentence structure, no interaction and less oral features. These texts can take in many forms such as statements, narration, speech, lectures, news broadcast, instruction, and announcement. Among them, statements differ from others in length. Statements are short compared to the rest of listening texts. The other forms of listening texts can be termed as monologues. On the other hand, if there are more than one speaker involved, such texts tend to be more like interpersonal talks, containing more implicit language, simple sentence structure, more interaction and more oral features. These texts include conversations, interviews, movies, dramas, debates, discussions, etc. Such texts can be termed as dialogues.
Considering the interest and the research purpose of the present study, I will mainly review in this section the related studies of statements and dialogues to see what the findings are.
First of all, Kitao & Kitao (1996) made some remarks on how to test statements.
In their opinions, testing statements was a good way to test the comprehension of a sentence. Their method was to have listeners choose the most proper response among four options after listening to a statement. As they pointed out, the advantages included its being more communicative than many listening tasks, having beneficial backwash effects, being easy to administer. However, on the other hand, it required not only the understanding of the utterance but also the ability to respond to it properly. In fact, listeners might be able to catch the meaning of the statement but failed to give a suitable response. In addition, lack of context added to the difficulty in the interpretation of statements. Consequently, there is still some controversy about the item type.
The problem of lack of context is also mentioned by Weir (1990) and Yang (1995). Weir (1990) pointed out that the lack of contextual clues such as “the visual element, the wealth of normal exophoric reference, and paralinguistic information” (p.
54) increased the difficulties in comprehending listening input and probably led to the underestimation of real listening ability. Yang (1995) discussed this issue in his article on how to test listening by comparing TOEFL and G-TELP. According to him, statements in TOEFL seem relatively easy but in fact they are quite demanding for listeners because of the lack of contextual clues and redundancy. In his opinion, contextual cues play an indispensable role in comprehension because they help listeners build “frame of expectation” and contribute to effective comprehension.
However, in listening to statements, listeners only get a single sentence without the aid of contextual information, so they may feel stressful for having to pay full attention. He thus claimed that a listening test lacking contextual cues failed to reflect the real listening ability. Besides, statements are too concise and lack redundancy.
Under these limitations, listeners can do nothing but listen for each word and focus more on the form than on the meaning. As pointed out by Yang, Bachman and Savingnon (1986) also criticized statements for the lack of context. Bachman and Savingnon proposed the canceling of the text type. Yang considered that dialogues were better than statements for they provided more contextual cues. However, the contextual cues given in dialogues come only from the text itself. Since there are only two sentences, such clues are still very limited. Besides, there are always two speakers, each of whom speaks one sentence. This makes the conversation less natural and true to life. To improve dialogues, he suggested that simple descriptions about the context should be given before listening, and that the length of the dialogues, numbers of talking, gender and number of speakers, and accents should be more flexible.
Wu et al. (1999) also mentioned the testing of statements and dialogues. They arranged the texts in the order of difficulty levels. According to them, dialogues were harder than statements. They indicated that the two kinds of texts differed in purpose.
The purpose of testing statements was to see if listeners could comprehend daily
sentences and give proper responses while the purpose of testing dialogues was to evaluate listeners’ comprehension ability of continual talks and the ability to infer.
Regarding testing statements, according to the results of the pretest, listeners were familiar with wh-questions except questions initiated with “what” while yes-no questions seemed very difficult for low-proficiency listeners. Besides, they also found that on the whole, statements were a difficult text type for low-proficiency listeners because they not only had to understand the meaning of the utterances but also had to respond properly. On the other hand, the assessment of dialogues mainly examined whether listeners could grasp the theme, main ideas, and framework of utterances, find out the purpose of talking, location, relationship between speakers, locate important details, and make inference. They further mentioned the factors affecting the performance of dialogues, including the sentence structure, vocabulary, amount of supporting details, repetition, and idiomatic phrases. Among them, the sentence structure in dialogues was often presented in simplified forms, requiring listeners to integrate other information before they reached comprehension. Such requirements could be very demanding for low-proficiency listeners. Besides, repetition seemed to benefit higher-proficiency listeners more. Moreover, they found that low-proficiency listeners seemed to have poor handling of idiomatic phrases. Similarly, Yen (1987) mentioned that in general, dialogues were more complicated than single statements because listeners not only had to understand what each speaker spoke but also to pay attention to the interactions and connections between their words. Despite the existence of some contextual clues in dialogues, listeners may not necessarily have the time and ability to make use of these clues for they have to do many things during such a short time.
Interestingly, Wu (1999) looked at dialogues from a different perspective. While
explaining the reasons why dialogues scored the best among the three text types5 in her study, Wu (1999) stated that despite the lack of the features commonly seen in face-to-face communication, dialogues contained more elements reflecting the real-life interaction compared to the other two text types. Therefore, dialogues helped to activate listeners’ relevant experiences. Another reason was that dialogues in listening comprehension tests were syntactically easier. In other words, Wu (1999) thought that the common interaction and simpler sentence structure contribute to the better performance in dialogues than the other two text types. Likewise, in the study of Shohamy and Inbar (1991), dialogues were found to be the easiest one6. Since dialogues were the most oral-oriented type, the study concluded that text type affected listening performance according to its position along the oral/literal continuum.
More information about the features of dialogues was offered by Dirven &
Taylor (1985). While discussing the variations in listening materials, Dirven & Taylor (1985) mentioned that “Spontaneous conversation is often extremely difficult to understand, since the socio-cultural setting, the topic, the informal or casual style and the sentence structure may all be very new to the listening foreign learner; also the learner is forced to listen in to a conversation to which (s)he is not a participant”
(pp.10-11). They also indicated that generally speaking, there was no much mutual knowledge between listeners and speakers in spontaneous conversations and therefore such texts could cause listening difficulties. Thus, they suggested that test designers should choose specific dialogues, which enabled them to follow the texts easily even
5 The three text types in the study of Wu (1999) included statements, dialogues, and passages.
6 There were three text types in the study of Shohamy and Inbar (1991). The news broadcast was a prewritten monologue, containing explicit vocabulary, complex sentences and the least oral features.
No redundancies, repetition, or pauses were found in it. The lecturette was a monologue based on written notes, which contained the interaction between the speaker and the audience, less explicit language, both simple and complex sentences, plenty of repetitions, pauses, redundancies, and monitoring of information flow. The consultative dialogue was spontaneous and contained constant interaction, much shared knowledge, feedback, involvement of both speakers, implicit and
context –dependent language, many redundancies, repetitions, interruptions, pauses and mostly simple sentences.
as non-participants.
In conclusion, in section 2.1, I have discussed the variables of listening texts and the findings concerning test and text effects. First, it was believed that a variety of text variables had certain effects on listening comprehension such as visual cues, contextual cues, oral extent, length, and text type. Second, in terms of Test Effects, picture-format tests were found to have the following advantages: testing listening ability only and not involving other language skills (cf. Dry, 1986; Tapia, 1996;
Thompson, 1995), offering contextual information for comprehension and inference (cf. Huang, 2004; Oxford, 1993; Wright, 1989), enhancing attention (cf. Cheng, 2000), arousing interest (cf. Tapia, 1996; Wright, 1989) and lowering anxiety (cf. Wu et al., 1999). On the other hand, non-picture-format tests have the advantages of avoiding pictorial ambiguity (cf. Harris, 1969), having more flexibility and range in testing sentence structures (cf. Dry, 1986), avoiding distraction by pictures (cf. Chung, 1994), reflecting real-life listening (cf. Kitao & Kitao, 1996), and avoiding the trouble finding or drawing suitable pictures (cf. Cheng, 2000). Third, in terms of text effects, statements were found to have the advantages such as being more communicative than many listening tasks, having beneficial backwash effects, and being easy to administer (cf. Kitao & Kitao, 1996). However, statements have often been criticized for requiring both the ability to understand and respond properly (cf. Kitao & Kitao, 1996; Wu et al., 1999) and for lack of contextual clues (cf. Kitao & Kitao, 1996; Yang, 1995). Compared with statements, dialogues were found to have simpler sentence structures (cf. Wu, 1999), more interaction (cf. Wu, 1999; Yen, 1987), more implicit and informal language (cf. Dirven & Taylor, 1985; Wu et al., 1999), and more oral features (cf. Shohamy and Inbar, 1991; Wu, 1999).
2.2 Issues in Listening Strategies
2.2.1 Classification of Listening Strategies
Since the 1970s, more and more researchers (cf. Bialystok, 1979; Brown and Palinscar, 1982; Chamot & Kupper, 1989; Hosenfeld, 1976; Naiman et al., 1978;
O’Malley et al., 1985; Oxford, 1990; Rubin, 1975, 1981; Stern, 1975) shifted their attention from teachers to learners. They became interested in the strategies used by language learners and attempted to identify useful strategies and teach them to the less successful learners. By doing so, they hoped to help students become independent language learners.
Influenced and inspired by the findings of language learning strategies, gradually, some researchers (cf. Murphy, 1985; O’Malley et al., 1989; Vandergrift, 1996, 1997) started to explore learning strategies in different language skills. In terms of listening strategies, however, the related investigations that have ever been made so far are still at very initial stages. In fact, there is still no agreement on the classification of language learning strategies, not to mention listening strategies. Therefore, many researchers (cf. Ku, 1998; Lee, 1997; Ozaki, 2000; Teng, 1996, 1997, 1998;
Vandergrift, 1996, 1997; Wang, 2002) took the findings of language learning strategies as their reference and research basis to explore listening strategies. Among the various classifications, the taxonomies of Oxford (1990) (see Appendix A) and O’Malley & Chamot (1990) (see Appendix B) were applied by many studies. For instance, some studies (Ku, 1998; Teng, 1996, 1997, 1998) adopted the taxonomy of Oxford (1990); some (Lee, 1997; Ozeki, 2000; Vandergrift, 1996, 1997; Wang, 2002) adopted the taxonomy of O’Malley & Chamot (1990).
2.2.2 Studies of Listening Strategy Use
A majority of researchers probed into listening strategy use. Some explored the overall strategy use of the subjects (cf. Bacon, 1992; Chang et al., 1995; Chao, 1996,
1999; Laviosa, 1991; Lee, 1997; Ozeki, 2000; Vandergrift, 1996; Vogely, 1995). Some examined and compared the strategy use of good and poor listeners (cf. Bacon, 1992;
Chao, 1996, 1999; Chien & Li, 1997; DeFillipis, 1980; Fujita, 1984; Ku, 1998; S. H.
Lin, 2000; Murphy, 1985; O’Malley et al., 1989; Ozeki, 2000). It is hoped that in this way, the useful listening strategies can be found and taught to poorer listeners. In fact, most studies reported differences in strategy use between listeners of different levels.
But researchers still have divergent opinions about such differences (cf. Bacon, 1992;
Chao, 1999; Chien & Li, 1997; Ku, 1998; Laviosa, 1991; S. H. Lin, 2000; Murphy, 1985; O’Malley et al., 1989; Ozeki, 2000). Some found such differences existing in both quantity and quality of using strategy (cf. Chien & Li, 1997; S. H. Lin, 2000;
Ozeki, 2000); some found the differences mainly existed in quality rather than quantity (cf. Bacon, 1992; Fujita, 1984; Murphy, 1985). Due to the variations in the background of the chosen subjects and experimental design, it is hard to compare the findings, let alone conclusive remarks. However, if we take a look at these findings, we may get a better understanding of listening strategy use. In the following section, I will report the overall strategy use first, and then compare the strategy use between effective and less effective listeners.
2.2.2.1 Studies of the Overall Strategy Use
Various methods were adopted to explore overall listening strategy use such as think-aloud procedure (cf. Bacon, 1992), immediate recall protocols (cf. Chao, 1996, 1999; Chien & Li, 1997; Laviosa, 1991), questionnaires (cf. Chao, 1996, 1999; Chien
& Li, 1997; S. H. Lin, 2000), listening journals (cf. Katchen, 1996), and interviews (cf.
Bacon, 1992; Chao, 1996, 1999; Chien & Li, 1997). Even the questionnaires they used differed from one another in classification, number of items, or content. The diversity in their methodology resulted in the difficulties in generalizing the findings in the issue. In the following section, I will present the findings of the studies which
differ in their methodology first and then look at those adopting similar methodology.
Chao (1996) found a positive relationship between strategy use and listening
comprehension. In the study of Bacon (1992), cognitive strategies were most frequently reported and among them bottom-up strategies were the most frequently used ones. Bottom-up strategies were also found in Chao (1996) and S. H. Lin (2000).
S. H. Lin (2000) found that the subjects often employed bottom-up processing and relied heavily on word decoding when listening to the English-learning programs on the radio. However, Chien & Li (1997) found that the word-bound habit (listening to each single word and translating) was harmful for listening comprehension. On the other hand, top-down strategies were reported in the study of Chao (1996, 1999).
Chao (1999) even pointed out their significant role in listening comprehension. As for metacognitive strategies, Vandergrift (1996) found the use of metacognitive strategies increased with the course level. Bacon (1992) found they were less reported by the subjects than cognitive strategies and the most frequently used one was monitoring.
Chao (1999) further found that directed attention and self-evaluation played a crucial role in listening comprehension.
In comparison to cognitive and metacognitive strategies, affective and social strategies were much less mentioned. In the study of Bacon (1992), the existence of affective strategies was mentioned but the subjects seldom reported the use of them.
Chien & Li (1997) found that a positive attitude, never giving in easily, and requesting help from the more proficient learners were found facilitating to comprehension and they suggested that TESOL educators should not ignore the affective factors. Social strategies were found in Bacon (1992), and Chao (1996, 1999). However, they were seldom used. According to Chao (1996), they were used less often than self-reliance strategies.
In terms of individual strategies, Laviosa (1991) observed the existence of seven
strategies7, Chien & Li (1997) identified 30 strategies8, Katchen (1996) found some strategies9 used to watch English TV programs, and Chao (1999) even reported that there were seven strategies that could best predict listening comprehension10. Comparing the strategies mentioned in these studies, a generalization was found. That is, two strategies (making contextual inference, connecting prior knowledge and experiences with new information) were consistently mentioned in them.
The above studies differ from one another in methodology more or less.
Interestingly, the following three studies used questionnaires with a similar classification system; however, their results were inconsistent. Their classification included three categories: cognitive strategies, metacognitive strategies, and social-affective strategies. But the results differed a lot from one another in the ranking of three categories. Vandergrift (1996) found that cognitive strategies were reported most, followed by metacognitive strategies, and then socio-affective strategies regardless of course level. However, Ozeki (2000) found that the subjects used cognitive strategies most, seldom used social/affective strategies, and rarely applied metacognitive strategies. Unlike the previous two studies, Lee (1997) found that the social/affective category ranked first, then the metacognitive category and finally the cognitive category.
7 There were seven strategies identified, including contextual inferring, seeking confirmation, using background knowledge, association, selecting information, vocalization, visualization, using cognates.
8 Chien & Li (1997) identified thirty strategies and further grouped them into three categories:
linguistic, cognitive and extra-linguistic. In the linguistic category, a word-bound habit (e.g. listening to every single word, translating) was found to be harmful for comprehension; in the cognitive category, the ability to infer and the ability to connect the existing knowledge to the new information were the two most effective strategies; in the extra-linguistic category, a positive attitude, never giving in easily, and requesting help from the more proficient learners were found facilitating to comprehension.
9 The findings of Katchen (1996) were somewhat trivial; therefore, I just listed in the present study some of them such as repeated listening, making use of visual cues, using subtitles, guessing by contextual information, paying attention to pronunciation and speakers’ mouths and using previous knowledge.
10 The seven strategies included grasping the overall meaning of the text, keeping up with the speed, relating to personal experience and knowledge, focusing attention on the listening, guessing the meaning based on the context, and using the linguistic knowledge (Chao, 1999).
2.2.2.2 Studies of Strategy Use of Effective and Less Effective Listeners
Besides the studies on the overall strategy use, some researchers were interested in exploring how effective and less effective listeners applied listening strategies (cf.
Chang et al., 1995; Chao, 1996, 1999; Chien & Li, 1997; DeFillipis, 1980; Fujita, 1984; Ku, 1998; Murphy, 1985; O’Malley et al., 1989; Ozeki, 2000; Vandergrift, 1997). Their findings are as follows. Chang et al. (1995) found the effective and less effective listeners had similar listening difficulties and learner strategies. Since the top 10% and bottom 10 % students scored quite differently but showed great similarity in strategy use in both classroom and natural situations, they concluded that listening strategies did not seem to play an influential role in listening comprehension.
However, since there were only ten choices offered in the learner strategy part in the questionnaire, there might be other factors needed to be further explored. Similar but more detailed findings were obtained by DeFillipis (1980)11. It was found that more similarity than difference was shown in the listening strategies used by the effective and less effective listeners. Both groups used the same varieties of strategies and almost equal total numbers of strategies. However, they differed in the rank order of the strategies in terms of frequency. That is, the effective listeners used key word strategy most frequently, followed by contextual inferencing strategy, and then grammar strategy while the three top frequencies of strategies used by the less effective listeners were key word strategy, translation strategy, and contexual inferencing strategy. Besides, the effective listeners reported using visualization strategy, cognate strategy, and role identification strategy more often than the less effective listeners and they were better at organizing aural information for recall. The
11 DeFillipis (1980) chose second-semester elementary French students as the sample to identify listening strategies used by effective and less effective listeners and compared the strategy use of the two groups. After taking a listening test, the top 13 and bottom 13 students were identified as the effective and less effective listeners respectively. They were asked to finish the aural comprehension tasks and then self-reported their strategies.
less effective listeners were found to use translation strategy and key word strategy more frequently than the effective listeners did.
Compared to the two studies mentioned above, more studies found there did exist some differences between effective and less effective listeners. Some found such differences mainly in the frequency of using strategies. For example, Ku (1998) worked with 126 freshmen and 40 juniors in a university in North Taiwan to explore listening learning strategies12. The results showed that juniors used every strategy more frequently than freshmen. Besides, for freshmen, six strategies13 were positively related to listening comprehension, while for juniors, it was the cognitive strategy (i.e., writing notes or reports in English). Besides, Chao (1999) even found that the effective listeners used listening strategies significantly more frequently than the less effective ones.
On the other hand, some studies found there were differences both in the quality and quantity of using strategies. For instance, S. H. Lin (2000) found that the effective listeners used listening strategies significantly more frequently and more flexibly than the less effective listeners. The effective listeners preferred comprehension monitoring and problem identification whereas the less effective listeners seemed to depend on clues from their mother language. Similarly, Chien & Li (1997) found that the effective listeners employed more strategies and three categories of strategies (linguistic strategies, cognitive strategies, extra-linguistic strategies) simultaneously.
Likewise, Ozeki (2000) found that the effective listeners used much more cognitive, metacognitive, and social/affective strategies than the less effective listeners. They
12 The students were asked to self-report the frequency of strategy use in the Strategy Inventory for Language Learning (SILL) of Oxford (1989).
13 The six strategies include three cognitive strategies, two compensation strategies, and one metacognitive strategy (i.e., trying to talk like native speakers, practicing English sounds, word analysis, making up new words, reading English without looking up every word, and finding ways to use English).
also differed greatly in the quality of their strategy use.
However, some researchers claimed that the differences between effective and less effective listeners mainly existed in quality, not quantities (cf. Bacon, 1992; Fujita, 1984; Murphy, 1985; O’Malley et al., 1989). For example, Fujita (1984) found that the effective and less effective listeners differed mainly in their ways of strategy use, not the varieties of strategies. To be specific, the effective listeners would listen selectively and actively, seek confirmation and revise hypotheses, listen for main structures, listen for whole sentences, not just words, focus on meaning rather than form, feel relaxed and confident, and take notes. Furthermore, Murphy (1985) found that the effective listeners tended to use the strategies more flexibly and openly while the less effective listeners relied more on the text and a consistent use of paraphrase.
Besides, the effective listeners elaborated from their own knowledge, made inference, drew conclusions, self-described, anticipated more frequently. Similar to the findings of Murphy (1985), the study by O’Malley et al. (1989) found that the effective listeners differed from the less effective listeners significantly (p<.05) in the three strategies (self-monitoring, elaboration, inferencing). In addition, it was reported that the effective listeners would actively apply learning strategies during the three phases of listening comprehension (perceptual processing, parsing, utilization). In perceptual processing stage, the effective listeners were aware of their loss of attention and would redirect attention consciously; however, the less effective listeners simply stopped listening. At the parsing stage, the effective listeners mainly used top-down processing and applied bottom-up processing only when needed, while the less effective listeners consistently depended on bottom-up processing. At the utilization stage, the effective listeners were found to connect new information with prior knowledge often, whereas the less effective listeners did much less. Consistently, Bacon (1992) concluded that the effective listeners showed greater flexibility and
range in strategy use and were “more realistic in their self-evaluation of comprehension and strategy use” (p. 331).
The following two studies on how effective and less effective listeners employed listening strategies again proved that metacognitive strategies were positively related to listening comprehension. Chao (1996) found that the effective listeners used metacognitive strategies more frequently than the less effective ones. Similarly, Vandergrift (1997) pointed out that the effective listeners were distinguished from the less effective listeners mainly by the use of metacognitive strategies, especially comprehension monitoring and problem identification. It was strongly suggested that more emphasis should be put on training monitoring and evaluating. In another study of Vandergrift (1999), he again mentioned that based on the findings of the studies on the differences in strategy use between the effective and less effective listeners, metacognitive strategies had great potential in improving listening performance.
To sum up, in this section, I have reviewed not only the background and classification of listening strategies but also the findings of listening strategy use.
Despite the diversity in the experimental design, background of subjects, and findings, previous researchers tended to show more agreement on the following things than disagreements. First, effective listeners resorted to top-down strategies more and converted to bottom-up strategies while tackling more difficult texts; in contrast, less effective listeners relied heavily on bottom-up strategies. Second, in general, effective listeners tended to use strategies more frequently and flexibly than less effective listeners. Third, some strategies like elaboration played facilitating roles to listening comprehension. However, some strategies like listening word by word and translation seemed more a hindrance than help. Last but not least, top-down and metacognitive strategies seemed to have great potential in enhancing listening performance.
2.3 Related Studies of Test Effects, Text Effects and Listening Strategies
In this section, I will review five related studies: two looked at test effects (picture-format tests vs. non-picture-format tests) (Cheng, 2000; Chen & Luo, 2001), one examined text effects (statements vs. dialogues) (Wu, 2000), and two investigated listening strategies (Teng, 1996; Wang, 2002).
2.3.1 Cheng (2000)
Cheng (2000) investigated the effects of pictures cues on listening comprehension assessment by using 302 second-year junior high school students in Taiwan. There were three instruments, including listening tests, a questionnaire, and an interview. The listening tests consisted of five picture-format tests and five non-picture-format tests. There were two question types: multiple-choice items and match items, including listening to one sentence, short talks, monologues, and conversations.
The findings are as follows. First of all, it was found that the subjects performed better on picture-format tests than on non-picture-format tests though the difference did not reach a significant level. In addition, the results obtained from the questionnaire and interview showed that most subjects held a positive attitude toward picture-format tests for the following reasons: the provision of context, the replacement of new or unfamiliar words, less time required in answering questions, and better grades expected to get. On the contrary, the subjects reported more time needed to answer questions, and feel more anxious when taking non-picture-format tests. Moreover, the effective listeners received significantly higher scores in picture-format tests than the less effective listeners. In fact, the t-test results showed that the effective listeners seemed to benefit more than the less effective listeners from both picture-format and non-picture-format tests. Compared to the effective listeners, the less effective listeners spent more time taking non-picture-format tests. It was also
shown that the effective and less effective listeners both spent less time finishing picture-format tests than non-picture-format tests and thus they consistently recommended the use of picture-format tests. Based on these findings, Cheng (2000) suggested that listening tests should incorporate pictures in junior high school in the future.
2.3.2 Chen & Luo (2001)
Chen & Luo (2001) probed into the effects of visual cues on listening comprehension of elementary school students in Taiwan. Thirty-six fifth graders were asked to participate in the experiment. The instruments were a questionnaire, and three listening comprehension tests. The three listening tests were tests with picture cues only, tests with both picture and textual cues, and tests with textual cues only respectively. For the experimental control, the topics in the three tests were the same.
It was found that tests with textual cues only scored the highest (M=82.8), followed by tests with both visual and textual cues (M=82.5), and then tests with visual cues only (M=81.4). However, no significant difference was found among the average scores on the three kinds of tests. The result indicated that picture cues were not a crucial factor to listening comprehension. Chen & Luo found that advanced learners could take advantage of both picture and textual cues in listening process. On the other hand, according to the number of the years of English learning, the subjects were divided into four groups. The results again showed no significant difference among the three tests, meaning that picture cues did not have great influence on listening performance. Moreover, Chen & Luo also tried to compare the performance according to the subjects’ frequencies of listening to English tapes. It was found that the subjects who seldom listened to English tapes at home performed better on the tests with picture cues than on the tests with textual cues only. Those who listened often or sometimes scored the highest on the tests with textual cues only. The findings
indicated that the lower-level listeners tended to rely more on visual cues while advanced learners might benefit more from the textual cues. In terms of the subjects’
attitude toward different test types, the results showed that the tests with both visual and textual cues ranked the first, followed by the tests with textual cues only, and then the tests with visual cues only. Finally, the subjects with the training in using picture and textual cues before were found to perform slightly better than those without the training though the difference didn’t reach a significant level.
2.3.3 Wu (2000)
Wu (2000) investigated the effects of para-linguistic factors on EFL listening performance. One of the research questions was whether text types were significant factors affecting students’ listening comprehension performance. The subjects were 340 first-year senior high school students from four schools in Chia-yi and Kaohsiung with the equal number of male and female students.
The instruments included three listening comprehension tests designed by Wu on the basis of six junior high school English textbooks and one new version of senior high school English textbook. The scheme of each test was developed with references to that in the listening comprehension section of the Test of English as a Foreign Language (TOEFL) and each test was made up of three parts: Part A (statements), Part B (dialogues), and Part C (passages)14.
The results obtained from both one-way ANOVA and an LSD post hoc comparison showed that dialogues were the easiest for the subjects, followed by passages, with statements the most difficult part and such difference actually reached a significant difference (p< .01). Wu thus concluded that text types did play a significant role in listening performance.
14 Part A included 10 items set at the single sentence level. Part B consisted of 10 short dialogues (A: B:
or A: B: A: B:). Part C contained 10 questions from narration with four levels of text length (about 50, 75, 100, 150 words respectively). There was only one item type (i.e., the multiple choice).
2.3.4 Teng (1996)
Teng (1996) investigated the EFL listening comprehension strategies used by college freshmen in Taiwan. The subjects were 124 students from the freshmen at National Yunlin Institute of Technology. For the lack of standardized and reliable listening assessments, Teng used their performance in the first semester freshman course “English Listening & Speaking Practice” to divide them into the effective and less effective listeners. The instruments included one questionnaire on the learning background of EFL listening comprehension and one questionnaire on the EFL listening comprehension strategies. The former was designed according to the researcher’s personal teaching and research experiences. The latter, which included 51 Likert-type items, was developed according to the listening comprehension strategy model of Oxford (1990).
The results showed that in terms of the overall strategy use, the rank of the six categories according to the frequency from high to low was: compensation, cognitive, metacognitive, memory, social and affective. As to the use of individual strategies, three most frequently used strategies were translating, followed by repeating, and then transferring; three least frequently used strategies included writing a language learning diary, followed by cooperating with proficient users of the new language, and then using laughter. On the other hand, the findings about the strategy use of the effective and less effective listeners are as follows. First, the effective listeners employed strategies significantly more frequently than the less effective listeners did in all the six categories. Second, concerning the use of six categories, the effective listeners employed compensation strategies most frequently, followed by cognitive strategies, then metacognitive strategies, then memory strategies, then social strategies and finally affective strategies. The less effective listeners employed cognitive strategies most frequently, followed by compensation strategies, then metacognitive
strategies, then memory strategies, then social strategies and finally affective strategies. In general, the effective and less effective listeners showed a great similarity in the ranking of the six categories. Third, with regard to the use of individual strategies, the three strategies most frequently used by the effective listeners from high to low were creating structures by highlighting, formally practicing with sounds and writing systems, and repeating. As for the less effective listeners, the three strategies most frequently used from high to low were repeating, translation, using resources such as dictionaries and grammar books. The strategy least frequently used for the effective listeners was working with native speakers, and that for the less effective listeners was writing listening diaries.
2.3.5 Wang (2002)
Wang (2002) attempted to find out the listening strategy use by Taiwanese senior high school students as well as the effects of gender and proficiency on listening strategy use. The subjects were 301 second-year students from three senior high schools in Chunghua. Finally, 297 valid data were collected, including 153 males and 144 females. The instruments were a listening comprehension intermediate test of the General English Proficiency Test (GEPT), a listening comprehension strategy questionnaire, and an interview. The listening comprehension strategy questionnaire15 was adopted from Lee (1997) and modified by Wang on the basis of previous studies (Bacon, 1992b; Chao, 1999; O’Malley & Chamot, 1990; Teng, 1996; Vandergrift, 1997).
Several findings were obtained in the study. First, among the categories of listening strategies, metacognitive strategies were used most frequently, followed by social/affective strategies and then cognitive strategies. Second, with regard to the
15 The questionnaire consisted of two parts: the background information and the listening
comprehension strategy use. The second part included 45 items in total and the listening strategies were divided into three main categories: cognitive, metacognitive, and social/affective strategies.
individual listening strategies, the subjects used translation, elaboration, monitoring, bottom-up, self-management, recombination very frequently (from high to low frequency); the strategies with low frequency of usage from low to high were transfer, deduction, top-down, self-encouragement, and voice inferencing. Third, it was found that proficiency affected strategy use to a significant extent. That is, the effective listeners used listening strategies significantly more frequently than the less effective listeners. Besides, the results showed that the effective listeners used every category significantly more frequently than the less effective listeners. The ranking of the three categories for both the effective and less effective groups was the same:
meta-cognitive strategies the first, social/affective strategies the second, and cognitive strategies the third. With regard to individual strategies, the effective listeners used planning, monitoring, self-evaluation, practicing, bottom-up, top-down, note-taking, grouping, summarization, and the social strategy significantly more frequently than the less effective listeners. As expected, the less effective listeners used translation more frequently than the effective listeners.
In a nutshell, in this section, I have reviewed five related studies. In terms of test effects, the findings of Cheng (2000) and Chen & Luo (2001) were inconsistent. From the statistical analysis, it appeared that formats were not a significant factor affecting listening comprehension. However, from the results of the questionnaires, their subjects expressed different opinions about test formats. In terms of text effects, Wu (2000) found that text types had significant effects on listening comprehension and dialogues were easier than statements. As for listening strategy use, Teng (1996) and Wang (2002) both found that the effective listeners used listening strategies significantly more frequently than the less effective listeners. However, they differed in their classification system of listening strategies and in the ranking of categories of listening strategies. Despite the findings, these studies differed from the present study
in some ways. None of them explored the relationship between test/text effects and listening strategies. Besides, except Wang (2002), the other researchers did not adopt standardized tools. Moreover, since there is still no classification system of listening strategies agreed upon by researchers, more explorations are needed to better understand students’ strategy use.
2.4 Summary of Chapter Two
In this chapter, I have reviewed the issues in listening texts and listening strategies and related studies. It was found that various factors and strategies had effects on listening comprehension. However, there is still no agreement upon test effects, text effects, and listening strategy use. Among these studies, few or even no studies have been conducted on test effects, text effects or listening strategy use of senior high school students in Taiwan. Most important of all, none of the studies reviewed in the present study examined the relationship between test formats and listening strategy use, and the relationship between text types and listening strategy use. Therefore, more work on these issues is still desired.
