Learner autonomy in language learning: the development of a rigorous measuring scale

Abstract

Learner autonomy has long been considered a requirement for university students. Attempts have been made to develop scales for measuring learner autonomy, but those built are either not psychometrically sound, too long not suitable for implementing in class, or based on a different conceptualised definition. This study, therefore, aims at developing a brief and sound scale to measure level of autonomy of language learners (called LLAS). Questionnaire development procedure started with adapting items from three existing questionnaires reviewed in the literature. The first questionnaire version (26 items) was piloted on a group of English major students (n = 220). Principle components analysis produced a 23-item scale with six subscales. Cronbach’s alphas and Principle component analysis were then rerun for the new 23-item scale. Results of both reliability and validity of this 23-item scale were found to be satisfactory, suggesting that the scale is reliable and valid.

BACKGROUND

Learner autonomy has been widely discussed for the last four decades and continues to be of great interest to scholars and educators. It is believed to be one of the prerequisites for life-long learning 1 ; it has received increasing attention when there is a gradual shift of educational focus, from teacher-centered to learner-centered 2 . With the advancement of technology, autonomous learners can access learning resources from every corner of the globe, and as such learning is not constrained inside the four classroom walls. Autonomous learning contributes to learners’ comprehension and strongly supports their language learning process 3 . To be successful, learners, especially those at the tertiary level, are expected to be proactive, take initiatives and be more independent in their studies. If learner autonomy plays such an important role in learners’ success and one of the objectives of higher education is to support the development of learner autonomy, instruments for measuring learner autonomy are needed. Though some attempts have been made to build a sound measure of autonomous learning (e.g., Self-directed Learning Readiness Scale 4 , Autonomous Learning Scale 5 , and Self-Efficacy Questionnaire of Language Learning Strategies 6 ), there seems to be a lack of a relatively short and comprehensive measure. This paper, therefore, aims at exploring how language learner autonomy can be measured, and on that basis, it proposes a questionnaire that can be used to measure learner autonomy of language learners.

THE CONCEPTUALIZATION OF LEARNER AUTONOMY

Learner autonomy is often referred to as a significant requirement to be successful in higher education; however, there seems to be no consensus on what it exactly means. In the early literature it is often referred to as self-directed learning 7 . In particular, Holec [ 8 , p. 3] defines it as “the ability to take charge of one’s own learning”. In Holec’s sense, it is a potential capacity to act in a particular learning situation, not learners’ actual behaviour in that situation. This ability is not what learners are born with but could be acquired through a purposeful learning process. Autonomous learners are able to identify their own learning objectives and to select resources and learning activities. Autonomy in Holec’s sense also means the ability to control and make decisions in the learning process, including planning what and how to learn, monitoring the acquisition procedure and evaluating what is acquired. Little 9 calls this aspect of meaning self-regulation instead of self-direction. Of the same view with Holec 8 , Little [ 9 , p. 3] considers learner autonomy as “the willingness, proactive and reflective involvement in one’s own learning”. In Little’s view, learner autonomy depends on the initiative of the learner a lot more than it does on the input given by a teacher or a textbook. The initiative is shown through efforts to seek help and cooperation with others since, as Little 9 argues, “autonomous learners do things for themselves, but they may or may not do things on their own” (p.223). Instead of considering this aspect of meaning as part of self-direction, Little 9 and some other scholars (e.g., 6 , 10 ) call it self-initiation. In the present study, learner autonomy is operationalized as a concept comprising two elements: self-initiation (learners’ motivation, positive attitudes and efforts to learn) and self-regulation (the ability to identify learning objectives, to select resources and to plan and monitor learning activities).

HOW TO MEASURE LEARNER AUTONOMY

Learner autonomy is believed to be problematic to measure in a traditional sense due to the complexity of the construct 3 , 6 , 11 . Degree of autonomous learning depends on the cultural context, the particular learning situation, the learning stage and individuals’ experience 10 , 12 . However, it is possible to identify the strength of autonomous learning if the concept can be broken into quantifiable components 3 , 6 . A number of studies have been conducted to investigate the strengths of learner autonomy. To measure level of learner autonomy, different approaches have been proposed such as teachers’ observation and first person narrative 11 , interviews and students’ learning journals 13 , students’ self-assessment 14 , 15 and peer assessment questionnaires 16 . Among various approaches, learners’ self-assessment seems to be the most prominent one since it is difficult to assess learners’ autonomy from an external perspective 3 , 12 . Assessment from an external source can only identify autonomy behaviour, not the capacity to behave autonomously 5 , 6 .

Several models to measure learner autonomy have been proposed. The most widely used measure is Guglielmino’s [ 4 , cited in 5 ] Self-directed Learning Readiness Scale. This measuring scale, however, has been reported to be problematic with its construct validity and therefore was recommended not to use 17 . Macaskill and Taylor 5 later built the Autonomous Learning Scale of 12 items based on that Self-directed Learning Readiness Scale. The questionnaire consists of two groups of question items, independence of learning and study habits . Question items on independence learning explore students’ responsibility for learning, their openness to experience, and intrinsic motivation while items on study habits monitor students’ study practices, time management and attitudes to lone working. This questionnaire has been built for the purpose of monitoring students at higher education in general, not focusing on language learners. It fails to elicit detailed information about strategies that students can use to manage their learning 6 . Neither does it contain question items to explore learners’ ability of goal setting and social interactive aspects.

Another scale is the one built by Nguyen 10 based on two components, self-initiation and self-regulation. The questionnaire was built following three principles: having the concept defined based on quantifiable components, employing both qualitative and quantitative data collection methods, and ensuring that the tool is piloted and validated. It contains 31 items on self-initiation and 22 items on self-regulation. Self-initiation refers to learners’ willingness to learn which is broken into reasons for learning and making efforts to learn whereas self-regulation involves learners’ cognitive skills of planning, monitoring, and evaluating. Though being built through a rigorous process, Nguyen’s 10 questionnaire was developed with a specific group of students in mind, students studying writing skill. It is, therefore, not ideal for the purpose of evaluating language learners’ autonomy at different stages.

To the best of our knowledge, Self-Efficacy Questionnaire of Language Learning Strategies (SeQueLLS) built by Ruelens 6 is the most recent scale. It was constructed by blending the construct of self-efficacy beliefs and learner autonomy with the argument that students with a high sense of self efficacy are more likely to be more responsible for their own learning. The questionnaire aims to explore students’ self-efficacy beliefs about the use of cognitive, metacognitive and social learning strategies to manage learning. The basis for the questionnaire includes (1) identifying learning needs and goal setting, (2) selecting learning approaches, (3) seeking social assistance, (4) organizing the learning environment, (5) monitoring the learning, (6) evaluating the learning process and outcomes, (7) transferring acquired skills and information to other contexts [ 6 , p. 377]. Though rigorous and involving both learner-task and learner-peer interaction, the questionnaire fails to explore learners’ motivation and attitudes towards learning, which is an important indicator of autonomous learning. Apart from that, two aspects in Ruelens’ 6 questionnaire, (4) organizing the learning environment and (7) transferring acquired skills and information to other contexts, are not considered as indicators of learner autonomy from the operationalised definition of the present study. From the review of the concept and the existing questionnaires, there appears to be a lack of a sound and comprehensive questionnaire for measuring learner autonomy.

QUESTIONNAIRE DEVELOPMENT PROCESS

Given the above discussion about what learner autonomy is and how to measure it rigorously, this paper attempts to construct a questionnaire exploring learner autonomy of English major students, which was named Language Learner Autonomy Scale (LLAS). The questionnaire has been built through three steps: (1) adapting the existing questionnaire, (2) piloting the questionnaire, and (3) revising the questionnaire.

Adapting existing questionnaires

Based on the two elements of learner autonomy of the operationalised definition (self-initiation and self-regulation), we built a questionnaire by adapting the questionnaires of Nguyen 10 , Macaskill and Taylor 5 and Ruelens 6 . The first element, self-initiation, was broken into two sub-elements, motivation and attitudes and making efforts to learn . This first element aims at exploring learners’ willingness to learn, positive attitudes towards learning and their efforts to learn through seeking assistance and working cooperatively with peers. Self-regulation was also divided into two sub-elements comprising of the ability to identify the needs and learning goals and the ability to select learning resources and planning learning activities. Table 1 presents themes, sub-themes and the number of questions in each theme.

Table 1 Summary of themes and number of questions in each theme

Macaskill and Taylor’s 5 questionnaire asks participants to rate using a 5-point Likert scale with Very like me at one end and Not at all like me at the other end of the scale. Both Nguyen’s 10 and Ruelens’s 6 questionnaires, in contrast, ask participants to rate each statement on a 5-point and 7-point Likert scale of agreement, respectively. Both rating scales are appropriate for measuring learners’ capacity, and for the reason of familiarity to the participants, we chose agreement scale (see Appendix A for the full questionnaire).

Piloting the questionnaire

The questionnaire was designed in Google Forms and distributed to students of Year 1 and Year 2 in a program of the English Faculty of a university in the South of Vietnam. The age range of the students was from 18 to 22. To collect the data, we visited each class, explaining the purpose of the study and the questionnaire to the students, and asking them to complete the questionnaire on a voluntary basis. The students were also encouraged to note down and report to us items that were not clear. This was an effort to collect learners’ reflection on the clarity of items in the questionnaire for revision. All the students present in the classes agreed to participate in the study and completed the questionnaire in about 15 minutes on average. The total number of questionnaires completed and valid was 220. No reports or suggestions on items that should be reworded were received. After the data were collected, the responses from the Google Forms were extracted in an Excel file, which was then cleaned and imported into the IBM SPSS Statistics 26 Program for analysis. The Likert-scale items were coded with 1 for Strongly disagree, 2 for Disagree, 3 for Neither agree nor disagree, 4 for Agree, and 5 for Strongly agree.

Reliability of the questionnaire

To ensure the reliability of the Likert-scale items in the original questionnaire, we checked the Cronbach’s alpha coefficients () for all the subscales and the corrected item total correlation for each item. The results are presented in Table 2 .

Table 2 Reliability statistics of the original Likert-scale items

As can be seen from Table 2 , the Cronbach’s alpha for the first subscale (SIM) will improve if item SIM4 is deleted. Similarly, the Cronbach’s alphas for the second and fourth subscales (SIE and SRP) will improve if items SIE4 and SRP4, respectively, are deleted. Items SIE4 and SRP4 also have low corrected item-total correlation. Therefore, these three items should be deleted from the questionnaire.

Validity of the questionnaire

To validate the construct of the questionnaire, we conducted an exploratory factor analysis (EFA) of all the Likert-scale items, using the Principal Component Analysis as the extraction method with Varimax rotation and coefficients with absolute values less than .50 being suppressed. As shown in Table 3 , the Kaiser-Meyer-Olkin (KMO) value is .851, which is greater than .500. The significance level (Sig.) is .000 (less than .050). It can thus be concluded that an EFA is appropriate for this study.

Table 3 Reliability statistics of the original Likert-scale items

As shown in Table 4 , the Rotated Component Matrix yielded from the EFA suggests seven factors.

Table 4 Rotated Component Matrix of the original Likert-scale items

The results of PCA also showed that the two items (SRP4 and SIM4) should be removed from the questionnaire. Item SIE4 was the only item left; it is therefore also removed from the questionnaire. The final questionnaire thus includes only 23 items. The PCA was rerun for the new set with 23 items. The Rotated Component Matrix shows that the PCA suggests six factors as shown in Table 5 .

Table 5 Rotated Component Matrix of the revised Likert-scale items

As can be seen from Table 5 , items in each of the two subscales of self-regulation (coded SRN and SRP) are correlated highly with each other within their group. The two subscales of self-initiation (coded SIM and SIE) are suggested to be split into four smaller subscales. Therefore, we decided to name the smaller subscales appropriately; in this way, it would be easy for researchers using this scale to refer to them when analysing results. Then, the reliability of the new set (with the six subscales) was checked. The Cronbach’s alpha coefficients for the six subscales are presented in Table 6 .

Table 6 Reliability statistics of the revised Likert-scale items

The Cronbach’s alphas of all the six subscales are above the required threshold of .700. The revised Likert-scale items for Learner Autonomy can thus be considered reliable.

DISCUSSION AND CONCLUSION

From this study, it can be said that the strength or level of learner autonomy could be explored and measured rigorously. The questionnaire developed in this study, based on the operationalised definition comprising two elements, self-initiation and self-regulation, was shown to be reliable and valid. This 23-item scale is not necessarily the best replacement for other existing scales but could be a preferable choice for teachers and educators who look for a brief measure that is easily administered and can generate results that are simple to interpret and monitor. Different from Nguyen’s 10 questionnaire, which was designed to be context-specific (i.e., in learning writing only), this questionnaire aims at measuring learner autonomy of language learners in general, not just one language skill; it is thus expected to be widely applicable. Future researchers who are interested in measuring language learner autonomy can use the questionnaire developed in this study as a research tool which is neither too narrow (about one language skill) nor too broad (about learning in general) as in the existing literature.

Although self-assessment is considered as the most prominent method of measuring learners’ capacity to behave autonomously, it is not completely certain that learners are actually self-initiated and self-regulated in learning as they self-report in the questionnaire. Where possible, teachers’ observation could be exploited as an additional data collection method to triangulate learners’ self-report data. This data set could play a significant role in interpreting and reinforcing findings from the self-report questionnaire. In summary, once the concept is defined as quantifiable components and steps of developing a questionnaire (designing, piloting, and revising) are carefully followed, it is possible to develop a rigorous measure.

ACKNOWLEDGEMENTS

This article is part of a research project, coding TC2024-08, which was funded by University of Social Sciences and Humanities, VNUHCM.

CONFLICT OF INTEREST AND DATA AVAILABILITY STATEMENT

The authors state no conflict of interest and there are no data associated with this article.

AUTHOR’S CONTRIBUTIONS

Cao Thi Phuong Dzung: in charge of collecting data, analysing data, and writing the introduction, literature review, discussion and conclusion of the article

Pho Phuong Dzung: in charge of collecting data, analysing data, and writing the methodology and result section of the article

APPENDIX A: QUESTIONNAIRE

Table 7 Apendix A

APPENDIX B: REVISED QUESTIONNAIRE

Table 8 Apendix B