Listening self-efficacy beliefs, L2 listening proficiency, and listening strategy training: An experimental study

________________


INTRODUCTION
gain success by following them, they get more motivated. Observing successes that are gained by sustained effort boost learners' beliefs in themselves. The third source of self-efficacy is available when learners are persuaded that they are capable and can complete a task successfully. In fact, teachers or parents can strengthen students' beliefs, which would lead them to put more effort on learning tasks. Graham (2007) claims that we need to make learners notice that their success or lack of success are due to their strategy uses, not their being talented or not. The last source of self-efficacy comes from affective and physical states. Depression, for instance, decreases an individual's confidence in their capabilities, and they may show poor performances.
Self-efficacy has been the focal point of numerous studies in the field of second language learning. A group of researchers have attempted to describe the relationship between self-efficacy and overall language learning (Gahunga, 2009;Magogwe, & Oliver, 2007;Mahyuddin, Elias, Muhamad, Noordin and Abdullah 2006;Mills, Pajares, & Herron, 2007;Raoofi, Tan & Chan, 2012;Woodrow, 2011). These studies have revealed that there is a significant positive relationship between self-efficacy and overall language learning success. For example, Raoofi et al. (2012) shows that learners' selfefficacy profiles can be a strong predictor for achieving language skills. Mills et al. (2007) investigated the L2 French students' academic performance in relation with their self-efficacy and found out that those who perceived themselves as proficient in using effective metacognitive strategies were more successful in French. On the contrary, Anyadubalu, (2010) studied the relationship among anxiety, self-efficacy, and students' performances in the English language and revealed that there was no significant relationship between self-efficacy and English language performance. Although this study shows no relationship, many studies that highlight the positive relationship between self-efficacy and using learning strategies can be an indicator of importance of self-efficacy and students' beliefs about themselves (Diseth, 2011;Kim et al., 2015;Pajares, 2002;Schunk, 1991;Anam & Stracke, 2016;Wilfong, 2006;Wong, 2005;Yang, 1992;Yang, 1999;Yang & Wang, 2015). The findings of these studies reveal a significant positive relationship between self-efficacy and the using learning strategies by students. For instance, Anam & Stracke (2016) studied on Indonesian students' strategy use in learning English and its relationship with their self-efficacy beliefs and found out that students with high self-efficacy use different strategies compared with those who have low self-efficacy. In another study, Yang and Wang (2015) demonstrated how explicit strategy training increased students' selfefficacy and strategy use. Bonyadi, Nikou, and Shahbaz, (2012) explored the relationship between Iranian students' self-efficacy and strategy use; however, the results showed no significant relationship between them. Nosratnia et al. (2014) also found out that metacognitive awareness is a better predictor for successful strategy use than self-efficacy beliefs.
Another group of studies have attempted to explore the correlation between students' selfefficacy and their anxiety level (Çubukçu, 2008;Ghonsooly & Elahi, 2010;Huerta et al., 2017;Kirmizi & Kirmizi, 2015;Martinez, Kock & Cass, 2011;Mills, Pajares & Herron, 2006;Torres & Turner, 2016;Woodrow, 2011). Significant negative correlation has been found between students' anxiety and their level of self-efficacy in Torres and Turner (2016). Some studies explored the relationship between anxiety and self-efficacy regarded with students' writing skills and they all reported that writing selfefficacy is a strong predictor for writing anxiety (Huerta et al., 2017;Kirmizi & Kirmizi, 2015;Martinez et al., 2011;Woodrow, 2011). Mills et al. (2006) conducted a similar research on students reading and listening skills which had similar results. Çubukçu (2008) revealed that there was no significant correlation between self-efficacy and anxiety; she argues that students' anxiety could be due to educational settings, not their self-efficacy. Overall majority of these studies have found out that self-efficacy has positive effects on individuals as they claim that those with high levels of selfefficacy could deal better with different tasks, unfamiliar contexts, and interpersonal relationships; students with high self-efficacy tend to endure more in difficulties and challenges. They argue that selfefficacy causes higher level of motivation, since it makes students conscious of their progress in a skill. Most of the studies support the strong positive relationships between self-efficacy and learning strategies and academic achievement among students, as suggested by Zimmerman and Cleary (2006). Students with high self-efficacy are more active in the learning process, in other words, they try harder, persist longer, and have fewer negative emotional reactions in facing difficulties. Finally, it can be claimed that students with higher self-efficacy have more control over their cognitive processes in general.

Self-efficacy and Listening Comprehension
Listening skills are considered to be crucial for success in language learning. According to Dunkel (1986), the main path to reach proficient speaking is through listening mastery. Rubin (1994) lists a few factors contributing to listening comprehension one of which is the listener's emotions that the listener undergoes in the learning context. In a recent inquiry, Graham (2006) studied a group of intermediate learners of French in England and conducted a questionnaire to find out which language skill they had been more successful. The results revealed that very few students had seen listening as a fulfilled skill; following the interviews from the same group, it is noted that there were statements of demotivation regarding listening (Graham, 2006). Such findings have led a group of scholars to study the effect of self-efficacy and anxiety on learners' performance of listening comprehension. For example, Chen (2007) compared scores of listening self-efficacy of Taiwanese students with their L2 listening anxiety and found out that second language listening self-efficacy could predict the listening scores better than second language listening anxiety. Mills et al. (2006) found out that self-efficacy in listening was positively correlated with listening proficiency and negatively correlated with listening anxiety, accordingly, it can be expected that with higher level of self-efficacy, students will experience less anxiety.
Apart from lowering anxiety, Arnold (2000) believes that self-concept can affect overall listening comprehension, the change of which as MacIntyre and Gardner (1991) argue, can be challenging; however, it is worth it. Boosting self-efficacy and listening performance accordingly can be achieved through teaching listening strategies explicitly, which aims at boosting learners' sense of control and expectations of success (Chamot, 1987;Nunan, 1999;Yeldham, 2009). Researchers argue that low levels of self-efficacy regarding listening may be caused by the techniques used by teachers to teach listening in the classroom (Chambers, 2007;Field, 2008;Graham, 2006). They note that listening is considered as an activity to be 'delivered' while it should be considered as a skill that needs to be developed.
In most L2 teaching contexts, teachers emphasize more on quantity of practice than quality and they rely on repetition of the listening texts and never implement pedagogic principles (Graham & Santos, 2013). The importance of listening comprehension, learners' self-perception and necessity of teaching strategies have been the focus of several studies. Graham (2011) notes that both selfefficacy and performance of listening can be improved by explicit teaching of listening strategies which can happen through increasing learners' sense of control and making them aware of the relationship between strategies used and the outcome achieved. Graham and Macaro (2008) showed that learners who received listening strategy instruction scored better on a listening test, and their self-efficacy for listening also had improved. The same result was reported by Goh and Taib (2006) who claim that strategy instruction and discussing strategy uses with students improved their listening strategy knowledge and their confidence. Strategies like verbalization or modelling by a teacher are also believed to improve self-efficacy among the learners; Schunk and Rice (1983) reported that modelling listening strategies leads to higher self-efficacy and better listening performance.
In addition to cognitive strategies, metacognitive strategies are also believed to increase L2 learners' performance (Flavell, 1979;Schoenfeld, 1987). Metacognitive knowledge is about the selfregulation of language learning such as planning, monitoring, and evaluating the learning process (O'Malley & Chamot, 1990;Wenden, 1998). Since listening is a receptive language skill that does not push learners to produce language, researchers believe that the use of metacognitive strategies in listening comprehension can activate learners' language knowledge while listening (Goh & Taib, 2006;Graham & Macaro, 2008;Vandergrift & Tafaghodtari, 2010). It is believed that proficient listeners use more metacognitive strategies, which lead to better listening performances. Students need to know what they need to achieve, and which strategies can be the most effective.
Mastery of listening comprehension as well as having high self-efficacy play key roles in overall language proficiency. There is also a codependency between high achievements in listening comprehension tasks and learners' high self-efficacy. While Çubukçu (2008) claims that the reason for anxiety among Turkish students is not due to their low self-efficacy but the overall educational setting in Turkey, it is assumed that explicit strategy training, modelling, and guiding students in selfregulating themselves will improve learners' listening comprehension performance, which will result in higher self-efficacy and lower anxiety levels. Therefore, this study aims to find out if teaching strategies and improving metacognitive abilities will alleviate learners' anxiety along with increasing their listening comprehension and self-efficacy or not.

Reseach Design
This study is a quasi-experimental design with non-equivalent pre-test and post-test control groups. In quasi-experimental research design, as opposed to the true experimental designs, the subjects in the experiment are assigned to groups in a non-random fashion (Creswell, 2012). Therefore, a quasiexperimental design was utilized in the present study as there were two experimental groups and one control group, the participants of which were not chosen randomly; all three groups took pre-test and post-test, the experimental groups received treatment whereas the control group did not.

Participants
For the purpose of the current study, convenience sampling was used. Ninety students from elementary and pre-intermediate levels of a language school located in Eskisehir, Turkey participated in the experiment. These levels were preferred because higher-level students might have already been familiar with some metacognitive strategies. Before the implementation, a placement test (Macmillian Straightforward Placement Test) was held to check the homogeneity of the learners. Participants were between the ages of 14 and 16 and the gender of the learners were almost similar in number: 43 males and 47 females. Thirty students from B1 level were assigned as the control group and the remaining 60 were placed in two experimental groups. All of the participants in the control group were in B1 level, and the experimental group participants were in two levels of A2 and B1 according to CEFR that were in different classes.

Trainers
Three trainers applied the treatment sessions in the study: the trainer of the B1 experimental group was a native speaker teacher from the USA. There were an Iranian teacher for the A2 experimental group, and a Turkish teacher for the B1 control group. The trainers of the experimental groups were already assigned to the related groups by the school administration, which means they were not selected by the researcher. Fortunately, the trainers of the experimental groups were also IELTS instructors, therefore, they were already familiar with the listening strategies they were going to teach. Before the pre-test, a meeting was held with the trainers, and the aim and approach of the study was explained to them. Before each treatment session, the listening tasks were given to them, and the trainers of the experimental groups were given a list of the related strategies they were going to teach to the groups. They taught metacognitive strategies such as planning, and self-monitoring along with listening cognitive strategies like guessing, note taking, inferencing, catching the gist of the text as well as listening for details. The trainers modeled all of these strategies before asking students to do them on their own.

Instruments
Three instruments were used to collect data for this study: • English Listening Self-efficacy Questionnaire

English Listening Self-efficacy Questionnaire
Learners' listening self-efficacy beliefs were measured by the Turkish version of an English listening self-efficacy questionnaire (See Appendix), which was developed by Rahimi & Abedini (2009). This questionnaire measures learners' confidence and self-perception during listening comprehension tasks and in communicating with native speakers and teachers. The original questionnaire was adapted for Turkish students in Turkish by using back-translation method. The questionnaire was first translated to Turkish and the translated version was retranslated into English to find out the flaws. The last version, which was in Turkish, was then edited and confirmed by an expert in the field. In order to prevent any misinterpretation of the questions, the questionnaire was piloted on seven learners who were not part of the study. The final questionnaire consisted of 18 Likert-scale items and the participants needed to read and decide if they: (1) strongly disagree (2) disagree (3) have no idea (4) agree (5) strongly agree.

Preliminary English Test (PET)
PET is developed by Cambridge ESOL called the Cambridge Main Suite, which measures the participants' proficiency in B1 level and is designed to be used for teenagers. Considering the age and level of the participants of this study, PET was used to measure the participants' listening comprehension in B1 level. This test has three main sections: reading/writing, listening, and speaking. The listening part consists of four parts ranging from short exchanges to longer dialogues and monologues. It measures learners' understanding dialogues and monologues on daily topics. The listening section has 25 items and takes 35 minutes for the test takers to complete.

Key English Test (KET)
KET is developed by Cambridge ESOL called the Cambridge Main Suite, which measures the participants' proficiency in A2 level and is designed to be used for teenagers. Considering the age and level of the participants of this study, KET was used to measure the participants' listening comprehension in A2 level. KET was used to measure the participants' listening comprehension in A2 level. KET has three main sections of reading/writing, listening, and speaking. The listening part consists of five parts ranging from short exchanges to longer dialogues and monologues. The listening section has 25 items and takes 30 minutes.

Listening Activities from British Council Website
Teachers who participated in this study used listening activities from British Council Website for the treatment sessions to teach and model metacognitive strategies used during listening activities. This website was chosen since the activities were designed for PET and KET exams, which could prepare the participants for the post-test exam.

Procedure
At the onset of the study, a placement test was held to confirm the homogeneity of the participants in both control and experimental groups. Later, the questionnaire of listening self-efficacy was given to the participants, which took 10-15 minutes. Having finished the survey, the listening comprehension tests were given to the participants. The participants in the control group and B1 learners of the experimental groups took PET and the rest of the experimental group took KET. Both of the listening tests had 25 questions, which took 35 minutes to complete. During the following weeks, the participants in the experimental group were trained for one month, by giving them listening activities, which were taken from British Council website. The activities were given in every session and the trainers were informed about how to train the learners. During the training sessions, the metacognitive strategies were taught in the classes. Some other strategies such as listening for gist and the details were also practiced. They were also given hints on how to answer different types of listening questions (e.g. true-false, fill-in-the-blanks, multiple choices). Other strategies were also used by the trainers to increase participants' self-efficacy. The strategies used for this purpose were adapted from Dörnyei (1994): giving affiliative motives, modelling the task first, and giving motivational feedback. Meanwhile, the control group was only receiving the activities and there were no instructions and no feedback given. To find out the efficacy of the treatments, after one month, the participants of both control and experimental groups were given a post-test. For the post-implementation session, students responded to the same questionnaire and took the listening test used for the pre-test.

Data Analysis
Quantitative data analyses were used to answer the three research questions posed for the purpose of the present experiment. To answer the first researh question, the learners' pre-test and post-test scores from PET and KET were compared via the paired samples t-tests. To answer the second research question, similar to research question 1, paired samples t-test analysis was used. To learn if there was any correlation between learners' post-test listening proficiency and listening self-efficacy scores of each group, as an answer to the third research question, Pearson's correlation analysis was calculated.

FINDINGS AND DISCUSSION
To have an overview of each group's performance on pre-and post-tests, the mean scores are listed in Table 1.  Table 1 demonstrates the number of the participants, minimum and maximum scores, mean scores, and standard deviations of the control group and the experimental groups of A2 and B1, both before and after the training period. Considering the control group, we can notice a slight improvement in their listening comprehension scores; there is also a noticeable amount of improvement in their self-efficacy scores. In the experimental groups, the participants of A2 level are seen to have improved their listening comprehension scores and their self-efficacy as well; however, their self-efficacy improvement is not considerable. The participants of the experimental B1 group seems to have a noticeable rise in their listening comprehension scores while their self-efficacy scores have slightly declined.

Strategy training and listening proficiency scores (RQ1)
The first research question was to find out if teaching Turkish EFL learners metacognitive and listening strategies improves their listening proficiency in a significant way. As Table 1 shows, the mean scores of the experimental and control groups all improved in the post-training listening proficiency test. Paired Sample t-tests were also conducted to find out whether these improvements were significant. As Table 2 shows, there is no significant difference between the pre and post-test of control group (t (29)= -1.992, p=.056). As the analysis indicated, the control group had no progress during the one-month time span. However, results show that there is a significant difference between pre-(M=17.30, SD= 3.109) and post-test (M=20.37, SD= 2.297) scores of A2 level experimental group (t (29) = -6.860, p< .001). There is also a significant difference between listening pre-(M= 17.30, SD=3.271) and post-test (M=21.27, SD= 2.212) of B1 level experimental group (t (29) = -8.650, p<001). This means that teaching listening strategies helped Turkish EFL learners to make a significant progress in their listening proficiency tests. This finding confirms the statement that teaching cognitive and metacognitive listening strategies explicitly can improve the learners' listening comprehension performance (Flavell, 1979;Goh & Taib, 2006;Graham & Macaro, 2008;Graham, 2011;Macaro, 2008;Schoenfled, 1987;Schunk & Rice, 1983). It can also be understood that listening comprehension needs explicit training and development, and learners need to know that they can control their performance by implementing appropriate strategies (Graham, 2011). Among the various strategies used was modelling of the instructor, which can be notified as an effective way to guide students (Goh & Taib, 2006;Schunk & Rice, 1983). Metacognitive strategies, which were also included in the training sessions, had noticeable roles played in the post-tense results since their engage learners during the listening tasks and help them to process the task deeper (Goh & Taib, 2006;Graham & Macaro, 2008;Vandergrift & Tafaghodtari, 2010).

Strategy training and self-efficacy (RQ2)
The second research question was about finding out if teaching listening strategies to Turkish EFL learners improved their listening self-efficacy in a significant way. Considering the pre-and posttraining mean scores of the three groups, a slight improvement was noticed (Table 3). To find out if the self-efficacy improvements were significant or not, paired samples t-tests were conducted for the groups. As Table 3 shows, there is no significant difference between the pre and post-training scores of control group (t (29) = -1.979, p= .057). Considering the A2 level experimental group, there is no significant difference between their pre-and post-training scores either (t (29) = -.340, p=.736). The same result can be seen for the B1 level group, and there is no significant difference between their pre-and post-training scores (t (29) = .389, p=.707). These results reveal that teaching listening strategies had no effect on experimental groups' listening self-efficacy. According to Chambers (2007), Field (2008, and Graham (2006), teaching listening strategies can increase the learners' control over the listening tasks, which leads to higher self-efficacy in them. This claim has been approved by Graham (2011), Macaro (2008), Macaro (2008), and Schunk and Rice (1983); however, the finding of this study contradicts with these findings. Turkish EFL learners had no significant improvement over their self-efficacy scores. According to Bandura (1995), one of the most important sources of self-efficacy is mastery, or in other words, the amount of successful experiences that learners have had during their educational period. It can be inferred that learners need more success to be able to have higher self-efficacy. Another source of self-efficacy according to Bandura (1995) is learners' physiological status, which includes their affective filters. Sparks and Ganschow (1991) claim that affective filters consist various emotions like stress, anxiety and lack of self-confidence. Similarly, Çubukcu (2008) has recently reported general anxiety among Turkish learners, which she relates to the educational system of Turkey. It can be inferred that anxiety affects learners' overall self-efficacy beliefs.

Listening proficiency scores and self-efficacy (RQ3)
The third research question was to find out if there was any significant correlation between listening proficiency scores and listening self-efficacy scores. Table 4 presents the results of the correlation analysis.  Table 4 indicates that there is no statistical correlation between A2 level group's listening proficiency post-test and their post-training listening self-efficacy scores (r=.085); the same result can be seen for the B1 group (r= -.041). Considering the control group, though, we can see a significant moderate positive relationship between their listening proficiency post-test and their post-training listening self-efficacy scores (r= .588, p<05). This finding contradicts with the studies reporting a significant positive relationship between self-efficacy and language performance (Gahunga, 2009;Magogwe, & Oliver, 2007;Mahyuddin, et al. 2006;Mills, et al., 2007;Raoofi, Tan, & Chan, 2012;Woodrow, 2011). Nevertheless, it is in parallel with Anyadubalu (2010) and Bonyadi et al. (2012) who had found no significant correlation between performance and self-efficacy. Anyadubalu (2010), found out that the learners who had lower self-efficacy did not necessarily have a poor performance in learning English; however self-efficacy was negatively correlated with anxiety. In other words, selfefficacy does not affect language performance on its own; but higher self-efficacy can lower the anxiety, which leads to a better performance.

CONCLUSION
This study attempted to find out if explicit teaching of listening strategies would increase learners' listening comprehesion scores. The results revealed that the experimental groups had a significant improvement in their scores. This implies that the outcome of the listening comprehension tasks would be noticeably better if teachers train their learners explicitly in what strategies can be more efficient for a particular task and how to implement those strategies. Learners should also be able to monitor their actions and evaluate themselves to find out whether they are on the right path or not. As students might be confused at the onset of the lessons and considering the fact that metacognitive strategies might be intangible for the learners, teachers' modelling the strategies can be essential during the training sessions. Teachers also need to observe their learners carefully to provide efficient feedback to them while they are implementing the strategies. To sum up, language teachers need to raise their students' consciousness on the nature of the listening skills and the needed strategies to perform the tasks. They should model those strategies, introduce and practice new strategies frequently, and remind students to plan and set goals and conduct self-evaluation (Rubin, Chamot, Harris, & Anderson, 2007). Another finding of the study was that explicit teaching of strategies had no significant effect on learners' self-efficacy, which led to the finding that there was no correlation between listening scores and self-efficacy scores of the experimental groups. Referring to Çubukçu (2008), general anxiety that is present in Tukish educational setting might lower learners' self-efficacy, which calls for further research on the elements that cause the anxiety and the ways that the system can be revised to lower it. The anxiety factor is one of the factors that this study should have considered and controlled. A limitation of the study was the duration of intervention sessions: the training period of the experimental groups took only one month, in a longer time span for training, the participants might increase their sense of mastery and accordingly their their self-efficacy scores.