EFL Learners’ Autonomous Listening Practice Outside of the Class

This study utilizes the design of an informal, semi-structured self-directed English listening program beyond the classroom in a university in Taiwan. The purpose was to examine Mandarin-speaking EFL learners’ autonomous learning processes and their perceptions toward the program. Twenty-two participants voluntarily signed up for the ten-week program, in which TED videos were used as the listening materials. Except for the orientation during the first week, all learning activities took place outside of the classroom at each learner’s pace. Data included 446 listening logs, end-of-program questionnaires, and interviews. The findings reveal that during the autonomous learning process the participants adopted various strategies at the stages of setting goals, selecting materials, developing learning methods, and evaluating learning outcomes. The strategies and self-regulation skills were shaped by both personal and contextual factors. Active metacognitive processes were also observed. Although learners’ participation decreased over time, they valued the opportunity to exercise control over their learning by having greater flexibility than that of taking a course. This study extends our understanding of learner autonomy in a specific context, which could shed light on the future designs of the self-directed language program.


2001)
, exploring the nature of autonomous out-of-class language learning in this context, provides important theoretical and pedagogical implications.
This study utilizes the design of an informal, semi-structured self-directed listening program outside the classroom, and examines EFL learners' learning processes and their perceptions toward the program. With the aid of technology, the forms of language learning in the twenty-first century will inevitably undergo processes of change and transformation.
On the one hand, taking courses and fulfilling course requirements may not be the major option for students who wish to learn a foreign language. On the other hand, however, students may not necessarily know how to self-regulate their language learning in informal, unstructured environments. Through an understanding of the autonomous learning process, this study intends to shed light on the design principles of out-of-class language learning arrangements, which may play important roles in supporting learners to transition from the stages of teacher-directed learning to self-directed learning, or simply in offering more diverse out-of-class language learning modules.

Literature Review
The concept of autonomy overlaps and interacts with several concepts in education theories such as self-directed learning and self-regulation. Lai (2017) argues that autonomous learning and self-directed learning share similar dimensions and goals. Accordingly, this study uses the two terms interchangeably. Self-regulated learning and performance, is defined as "the processes whereby learners personally activate and sustain cognitions, affects, and behaviors that are systematically oriented toward the attainment of personal goals" (Zimmerman & Schunk, 2011, p. 1). Effective use of self-regulatory strategies has been found to greatly influence language learners' autonomous behaviors (Kormos & Csizér, 2014).
Studies on autonomous language learning have been conducted in formal, institutional contexts such as blended learning which combines face-to-face classroom instruction and out-of-class activities (Hafner & Miller, 2011;Kessler, 2009;Snodin, 2013) and online language courses (Lee, 2016). For example, Snodin (2013) integrated a course management system (CMS) into a face-to-face classroom. The study suggests that learner autonomy was advanced through carefully designed activities in a supportive environment, which grants learners opportunities to exercise control over their own learning. In another study, Lee (2016) examined autonomous language learning in a full online Spanish course. Self-regulation was found to play a key role in the autonomous learning process.
It is not until recent years that researchers have advanced the field of autonomous language learning to a context beyond formal curriculum. Studies have explored how learners engaged in out-of-class learning activities without the need to comply with institutional requirements. Through learner diaries, Sockett and Toffoli (2012) reported that learners first listened to songs for a gist, and then used online written lyrics to resolve comprehension difficulties. Rosell-Aguilar (2013) surveyed the listening habits of iTunes U language learners, and found that most language learners rarely or never took notes during the listening process. In addition, they listened to podcasts while engaging in other activities, which suggests that learning in this context is casual. These findings contradict those of previous research which was conducted in formal settings and showed that listening to podcasts was perceived as a formal academic activity. The contrast implies that autonomous language learning is a context-dependent construct (Benson, 2001). One of the aims of this study is to extend our understanding of the construct in an informal, semi-structured context. Figure 1 illustrates the conceptualization of autonomous learning in different contexts, in part based on Lai's (2017) definition. The program designed for this study involved neither grading nor credits (i.e., informal), but was implemented under an instructor's guidance by offering consultation upon learners' requests and providing regular weekly progress reports to learners (i.e., semi-structured). Few studies have explored language learner autonomy in a similar context. One exception is a study by Bailly (2011). An out-of-class language learning program was developed for teenagers in France, and provided various learning resources such as advisory sessions, online materials designed for the program, and conversation sessions with native speakers. Students voluntarily participated in the program, learned at their own pace, and chose their own learning methods and resources. The findings show that students did not seem to know how to adopt learning strategies for lighter activities (e.g., watching TV and playing online games) without teacher guidance. Students also reported various difficulties, for example, a lack of obligation to learn, a lack of systematic monitoring of the learning processes, and an inability to set realistic individual learning objectives. Bailly's study again suggests that autonomous language learning beyond formal curricula constitutes unique learning processes and difficulties.

Autonomous Learning in Different Contexts
This study aims to build on our understanding of this less explored context. It is through the understanding of these learning processes and difficulties that effective out-ofclass language learning programs can be developed and tailored for different learners' needs.
Language learning and teaching therefore can take various forms with an ultimate goal of fostering lifelong language learners. Two research questions were addressed. (1) How do EFL learners practice English listening using TED talks in an out-of-class self-directed program? (2) What are EFL learners' perceptions toward the out-of-class self-directed program?

Participants
Twenty-two participants (M age = 23; 11 males and 11 females) were recruited in a university in Taiwan, including 12 graduate students, nine undergraduate students, and one adult EFL learner. They voluntarily signed up for the TED listening program. More than half of them reported to have taken at least one formal English listening course. Their levels of English listening skills, as measured by a TOEFL listening practice test, ranged from intermediate to high.

The TED Listening Program
The duration of the TED listening program was 10 weeks (Table 1). TED videos (https://www.ted.com/) were chosen as the listening materials for two reasons. The videos provide authentic listening materials covering a variety of topics, from which learners can choose the ones that interest them for the self-study purpose. Also, most of the videos come with subtitles and interactive transcripts, which can serve as learning resources for individual listening practices. This TED listening program was a non-credit self-directed learning program. Each participant was required to watch at least four TED talks per week for a period of eight weeks. Participants could freely decide on when, where, and how they watched each TED talk. The deadline to finish watching four TED talks was every Sunday at 11pm.

Table 1
The Procedures of the TED Listening Program and Data Collection

Weeks
The TED listening program Data collection Week

Listening Log
Participants had to submit a listening log for every TED talk that they watched. Each log entry required participants to document their listening process, report listening difficulties, and reflect on the content of the talk. The listening log consisted of 17 questions.
Fifteen of them were structured in the form of multiple choice or multiple response questions to encourage participants' response rate. One of them was an optional, open-ended question, which allows participants to reflect on the listening process. The last question asked participants to write a 50-word reflection on, or summary of the talk. Listening logs were collected using Google Forms, which allows access to the instant update on participants' learning processes.

Facebook Community
A secret Facebook group page was created to provide an online social platform for participants. Participation in the Facebook group was voluntary and access to the group was restricted to the participants of the study. Specific ethical principles about using the Facebook group were also explained to the participants in the consent form and during the orientation.
They were encouraged to post at least one piece of reflection (at least 50 words; written in English) on a TED talk they had watched and reply to at least one post per week. They were also encouraged to share their language learning strategies or ask questions on the Facebook page.

Weekly Progress Report
The researcher read through all listening logs, Facebook posts and responses submitted for the previous week every Monday morning, and sent a weekly progress report to each participant via email. The weekly progress report included (1) a summary of the total number of TED talks the student had watched and the total number of Facebook posts and responses the students wrote in the previous week, (2) a response to students' questions or learning difficulties recorded in the listening log, if any, and (3) personal messages to encourage their accomplishment of the weekly goal.

Data Collection and Procedures
This mixed-methods study adopts a sequential, explanatory design in which quantitative data were collected first, and qualitative data were collected at a later stage to provide explanation and support narratives for the quantitative data (Creswell et al., 2008).

Data Analysis
The data analysis involved three major stages. The first stage was the quantification of the data collected from listening logs, Facebook posts and responses, and the end-of-program questionnaire to understand the overall accomplished rate and participation pattern over the eight weeks. The results obtained from this stage were used to select participants for later interviews and to form interview questions. The second stage was to further quantify the data from listening logs and the end-of-program questionnaire to reveal the participants' autonomous learning processes in the program (Research Question 1) and their perceptions toward the program (Research Question 2). To answer Research Question 1, the listening logs and the end-of-program questionnaires were categorized relative to four aspects (i.e., setting goals, selecting learning materials, developing learning methods, and evaluating learning outcomes) based on a review of previous studies on language learner autonomy (Little, 1991). To address Research Question 2, the end-of-program questionnaires were analyzed to understand the participants' perceptions toward each major component of the program. During the third stage, the interview data were transcribed verbatim and coded inductively. The data were initially coded and analyzed with the two research questions in mind: the autonomous learning process and participants' perceptions. Sub-themes were later developed based on multiple readings of the data and quantitative findings. The interview data were used to explain and offer narrative support of the quantitative data and were examined for several rounds to identify additional themes to supplement the quantitative findings.

Results
Overall Performance Figure 2 illustrates the total number of listening logs, Facebook posts, and Facebook responses completed by participants by week. took less than 15 minutes; 25% took more than 30 minutes.

Figure 2
Total Numbers of Required Exercises Submitted by Week

Required excercises submitted by week
Listening logs

Facebook posts
Facebook responses

Setting Goals
Participants were asked to set a goal for each round of listening practice. Data analysis of the 446 listening logs showed that the top three goals selected by participants were related to understanding main ideas: (i) understanding main ideas and purposefully learn English vocabulary or expressions (31%), followed by (ii)  for himself-writing a summary. To be able to write a summary, he had to get the main ideas of the talk, which was already a challenging task for him.

Selecting Learning Materials
Of the 446 TED talks watched by the participants, the most common length of the video was 6 to 15 minutes (58%), followed by less than 5 minutes (29%), and 16 to 25 minutes (13%). The interviews indicate that participants used a variety of strategies and website functions to select the TED videos. For example, they avoided the talks delivered by speakers with unfamiliar accents or forced themselves to watch talks beyond personal interests. Also, these strategies were not consistent throughout the eight weeks and were adjusted according to individual needs.
In the beginning, I chose the ones that interested me, but then I realized that I had to learn things from other disciplines. I started to expand my search focus (Participant 3,

Interview).
A word cloud (Figure 3) was generated to illustrate the common themes of the content that the participants listened to. Based on the topics and the links provided by the participants in listening logs, the researcher retrieved the related tags for each talk on the TED website.
Related tags (N = 2377) were available for 83% of the TED talks that the students watched. As can be seen from the figure, the most common theme of the topics selected by the participants was related to health, followed by culture, science, business, and technology.

Developing Learning Methods
Listening logs show that 39% of the TED talks were watched from the beginning to the end once, 38% of them were watched partially, only 18% of the talks were watched from the beginning to the end twice, and 5% of them were watched from the beginning to the end three times.
The qualitative data further revealed that participants developed a variety of individualized listening practice cycles, which can be divided into two types. The most common practice type was to listen once and repeat the parts where comprehension difficulties arose. For example, I did not watch a video for more than once. If I didn't understand a sentence, I rewound immediately and listened to that sentence again. If the entire talk was really difficult, I listened to the talk and read the transcript at the same time…I also paused the video to take notes or look up words in the dictionary (Participant 21, Interview).
The other type was to listen more than once. The first time was usually listening without subtitles, and the subsequent rounds with subtitles. For example, A closer examination of participants' listening logs reveals that most of them did not follow only one practice cycle. Participants adjusted the practice cycle based on their evaluation of whether they reached the desired level of comprehension. In other words, they were constantly monitoring and evaluating their own learning.
As shown in the previous quotes, learners did not just listen; they used resources (e.g., subtitles, dictionaries, and Google) to enhance their learning. Listening logs show that only 5.61% of the listening practices did not include the use of any learning resources. The five most frequently used resources were English subtitles (314 out of 446 listening logs; 70%), English interactive transcript (42%), Chinese-English dictionary (36%), Chinese subtitles (34%), and Google (28%).
Although subtitles and interactive transcripts are easily accessible on the TED website, learners did not always turn them on when listening. As shown in Table 3, when English subtitles were used, they were mostly used for 1-60% of the practice time; reported use of subtitles or transcripts for 100% of the practice time was rare; more than half of the listening logs indicated that Chinese subtitles were not turned on at all. This finding suggests that learners selectively used subtitles or transcripts to aid listening comprehension during listening practices. They seemed to strategically incorporate subtitles or transcripts into the listening practices. For example, as illustrated in previous quotes, English subtitles or transcripts were used to resolve comprehension difficulties or to confirm the gist of the talk.
Chinese subtitles, on the other hand, appeared to be used as the last resort: I used Chinese subtitles for four times throughout the eight weeks. Those were the times that I understood every word in the English subtitle, but still couldn't understand what the speaker meant…When the language went beyond the literal meaning, I turned on the Chinese subtitle to comprehend (Participant 12, Interview).  19%). During the interviews, participants also expressed the need to have some form of evaluation to help them monitor their listening outcomes, for example, comprehension questions.

Self-regulation in an Out-of-Class Program
The aforementioned autonomous listening processes were driven by another set of learning processes at the macro level-self-regulation. The most salient difficulties in such a program reported by the participants were not being able to require oneself to complete the program requirements (M = 3.40; SD = 1.23) and too busy to complete the program requirements (M = 3.95; SD = 1.05), as measured by the end-of-program survey on a 5-point Likert scale. Both difficulties were related to learners' capacity for self-regulation. The interviews revealed that participants employed both internal and external forces to regulate their learning. The internal forces included self-reminder, self-reward, self-requirement, and a sense of responsibility, as reported by most participants who constantly engaged in the program throughout eight weeks. For example, Self-reminder: I set a phone reminder every weekend…but after 4 or 5 weeks, this had become a habit (Participant 12, Interview).
Self-reward: I gave myself small rewards after completing the task, for example, taking a shower, going to the gym, having some food…Soon I got used to it and it had become part of my life rather than a required task (Participant 15, Interview).
Self-requirement: If I did not accomplish the goal this week, I would ask myself to make them up in the following week (Participant 16, Interview).
Self-imposed obligation: Since I signed up for this program, I had to be responsible (Participant 2, Interview).
The external force came from the pedagogical mechanisms of the program. According to the end-of-program questionnaire, the weekly progress report sent out by the teacher was considered as the most commonly used approach to push and motivate participants to accomplish weekly requirements (M = 3.95; SD = 1.19). Several participants used the weekly progress report to regulate their learning: I don't think I managed my time well. I sometimes put it off until the day before the weekly progress report was sent. The fact that a report about my accomplishment would be sent tomorrow motivated me to listen more (Participant 3, Interview).

Learners' Perceptions Toward the Out-of-Class Self-Directed Program
The end-of-program survey revealed participants' perceptions toward the program.
Overall, the program was considered interesting (M = 4.25; SD = 0.77). In the survey, participants also rated their perceived usefulness toward seven major activities in the program. The finding showed that they held positive attitudes toward activities that required or were related to personal efforts such as watching four TED talks per week (M = 3.90; SD = 1.24), filling out a listening log for each talk (M = 4.15; SD = 1.14), posting reflection on Facebook (M = 3.25; SD = 1.37), and receiving weekly progress reports from the teacher (M = 3.75; SD = 0.97). The activities that involved interaction on Facebook with peers such as reading and responding others' reflection were perceived less helpful with means less than 3.00 out of a 5-point Likert scale. According to the interviews, one major reason was that most students did not know each other in person; they were not motivated to interact with strangers on an online platform.
When I realized that I actually knew two participants in person, I became eager to read what they wrote on Facebook compared to those posted by the people I didn't know (Participant 2, Interview).

Discussion and Conclusion
The most notable strategies employed by participants were metacognitive listening strategies, including "planning for, monitoring, and evaluating comprehension" (Vandergrift & Goh, 2012, p. 90). The metacognitive strategy use is based on the metacognitive knowledge that learners possess, which can be divided into person, task, and strategy knowledge (Vandergrift & Goh, 2012). To illustrate, Participant 12, an advanced listener, began each round of listening practice by planning specific goals, one of which was to comprehend the talk to a level that would allow him to write a piece of critical reflection.
This goal was based on his person knowledge that he was a language learner who was capable of critically reflecting on the content of the talk and that he was likely to develop a sense of critical thinking. While listening to the talk, Participant 12 constantly monitored his comprehension, identified comprehension difficulties, and sought resources to address the difficulties, for example, using Chinese subtitles to solve comprehension difficulties arising from figurative language uses. This strategy use builds on his task knowledge (i.e., In this study, the autonomous learning processes, which were characterized by individual and dynamic strategy uses, were shaped by both personal and contextual factors.
For example, participants' goal setting for each round of listening practice was influenced by personal factors such as interest in the topic of the talk or self-perceived listening competence and contextual factors such as the required 50-word writing task. At the macro-level, the eight-week autonomous learning processes were also driven by personal and contextual factors such as a variety of self-regulatory strategies and the weekly program requirement.
The interplay between the two forces can be conceptualized from the realist perspective of agency and structure (Gao, 2010). Realists maintain that learners constantly reflect on the existing contextual elements and reconstruct the meanings of the context for the "realization of their own interests" (Sealey & Carter, 2004, p. 11). As can be seen from the findings of this study, it is through this continuous exercise of agency that autonomous learning takes place.
The overall engagement in the program decreased throughout the duration of study, a finding often reported in the literature of informal language learning beyond the classroom (e.g., Lin et al., 2016). Different from previous studies, however, the informal out-of-class language learning was semi-structured and implemented under teacher guidance. Qualitative findings show that under this structure, the participants adopted strategies to sustain their learning motivation and behaviors. For example, they used the contextual elements such as watching four TED talks per week and receiving a weekly progress report from the instructor, to break down the seemingly intangible, long-term goal into the manageable, short-term ones.
Several participants started to watch TED talks one day before the weekly progress reports were sent; some made up work that had fallen behind in the following week. Although these learning behaviors show signs of procrastination, they could also be interpreted as the approaches to support persistence. This finding suggests that other than having access to learning resources, language learners need other forms of support such as teacher guidance and semi-structured or structured activities to resist attrition during the autonomous learning process (Lin et al., 2016).
Compared to other forms of learning activities, the participants in this study engaged less in and perceived less positively toward activities that involved online social interaction.
One major reason was that they did not know each other in person since the program did not offer face-to-face interactive opportunities. This suggests that simply setting up an online social platform does not guarantee social interactions among learners, especially those who are not acquainted with one another in person. Specific steps need to be taken to facilitate peer interactions in the context of out-of-class language learning, for example, offering faceto-face discussion sessions, as suggested by the participants of this study. The sharing of a strong sense of community in a self-directed learning group would encourage collaborative learning and in turn may reduce the learner attrition rate (Lai, 2017). More importantly, the apprenticeship formed in the community of practice through observing and interacting with peers (Oxford, 2011) could drive learners to constantly reflect on their own learning, which plays an important role in the development of autonomy (Lai, 2017).
The findings suggest that autonomous language learning in this out-of-class context was mostly intentional, different from the incidental learning behaviors reported in previous research situated in informal, unstructured contexts (e.g., Rosell-Aguilar, 2013). As shown in the interviews, the intentional learning behaviors were in part prompted by the contextual factors (e.g., writing a 50-word reflection or summary for each talk and posting it on Facebook). This finding can be informative for instructors who wish to develop students' autonomous language learning competence beyond the classroom. Most out-of-class language learning experiences were reported to be casual and incidental and mainly involved receptive activities (Lai, 2017). A balanced and diverse learning environment has been found to positively correlate with effective cognitive and affective learning outcomes, for example, an environment that encourages learners to engage in both receptive and productive activities or learning experience that integrates diverse use of resources (Lai et al., 2015). Given that the finding of the present study showed that the contextual factors shaped students' learning behaviors, carefully designed guidelines or activities for out-of-class learning are critical to learners' construction of effective out-of-class language learning environments.
The present study has some limitations. Multiple semi-structured interviews conducted during the program would have allowed more systematic observations of the changes and reasons behind the fluctuation in the autonomous learning process. Furthermore, the participants of this study voluntarily signed up for this zero-credit program. It is likely that they demonstrated higher levels of motivation compared to other EFL learners.
Generalization of the findings needs to be cautious. Also, interviews with all participants including those who reported none or very few learning activities would enrich our understanding of factors that contribute to attrition.
The findings of this study extend our understanding of learner autonomy in a specific context, which could shed light on the future designs of the self-directed language program.
More specifically, as Lai (2017) states, "Autonomy is not only teachable, it also needs to be consciously supported, fostered and maintained" (p. 10). In addition to incorporating autonomy into the curriculum, informal and semi-structured programs beyond the classroom could be designed to constantly support the development of learner autonomy.

Notes on the Contributor
Fang-Ying Yang is an assistant professor in the Institute of TESOL at National Chiao Tung University, Taiwan. Her research interests include learner autonomy, self-directed language learning beyond the classroom, and L2 listening pedagogy.