Strategy-based instruction: A learner-focused approach to developing learner autonomy

1 Approaches to fostering LA

Benson (2001) classifies six approaches to developing LA. These approaches are resource-based, technology-based, curriculum-based, teacher-based, classroom-based, and learner-based. They focus on different aspects of control in the learning process.

Resource-based approaches place emphasis on the provision of opportunities for learners to direct their own learning in self-study, self-access, and distance learning. Materials and counselling have become the main instruments for the operation of resource-based approaches. Researchers have explored the engagement of learners in more active and creative roles rather than responsive and reproductive ones (Littlejohn, 1997; Sheerin, 1997; Sturtridge, 1997), and in using authentic materials (Lee, 1996; McGarry, 1995). Considerable efforts have been invested in discovering counselling frameworks (Dingle & McKenzie, 2001; Kelly, 1996; Pemberton, Toogood, Ho, & Lam, 2001) and specific counselling methods (Carter, 2001; Cotterall & Crabbe, 2008; Voller, Martyn, & Pickard, 1999).

Technology-based approaches emphasize learning opportunities made available by diverse forms of technology. Some forms of technology can provide opportunities for collaboration which resource-based approaches often fail to achieve. The most popular form of technology-based approaches is computer-assisted language learning (CALL), which makes use of CD-ROMs and the internet for language learning by incorporating an interactive video programme in self-access centres (Gardner & Garcia, 1996), or using analytic techniques of linguistic texts to plan, generate ideas, write, and revise written work (Milton, 1997). E-tandem – learning a language through a partnership between a native speaker and a non-native speaker – is another way of making use of technology for enhancing LA (Brammerts, 2003; Brammerts & Calvert, 2003).

Curriculum-based approaches focus on the negotiation between teachers and learners. The negotiation is intended to enhance learners’ participation in making decisions on learning content, activities, and tasks as well as to evaluate learning. Curriculum-based approaches take two forms: a weak and a strong version of the process syllabus (Benson, 2001). The weak version involves learners’ project work in which determinations on content and methods are made by themselves (Cunningham & Carlton, 2003; Nix, 2003; Stephenson & Kohyama, 2003). In the strong version, the syllabus is not predefined. Rather, it is selected, organized, negotiated and renegotiated by teachers and learners as the learning goes on (Cotterall, 2008; Dam, 1995).

The focus of teacher-based approaches is on teacher professional development and teacher education. These approaches have been developed on the assumption that changing teachers’ beliefs about autonomy, building their commitment to autonomy, and encouraging practices that support LA will result in classroom changes in favour of LA. Research has mainly focused on understanding the concept of teacher autonomy and on working out principles for fostering it (Benson, 2000; Little, 1995; McGrath, 2000; Thavenius, 1999; Vieira, 1999) and for developing the teacher’s role (Aoki, 2002; Yang, 1998).

Classroom-based approaches highlight changing relationships and practices inside the classroom. The changes enable teachers to transfer responsibility and control over learning goals, the learning process, and the assessment of learning outcomes to learners (Crabbe, 1993; Shao & Wu, 2007; Smith, 2001, 2003). The most popular forms of these approaches include cooperative learning (Mizuki, 2003; Tagaki, 2003), portfolios (Kohonen; 2000, 2001; Nunes, 2004; Rao; 2005; Shimo, 2003), self-assessment (Nachi, 2003; Thomson, 1996), peer-assessment (Miller & Ng, 1996; Natri, 2007), and out-of-class learning (Hyland, 2004; Pearson, 2004; Pickard, 1995, 1996; Yap, 1998).

Despite a large body of research advocating different approaches to fostering LA, the effect of a particular approach has yet to be identified. A few remarks about LA research could be drawn on the above approaches. First, most LA studies so far are descriptive and exploratory. While these studies have an important function of generating insights into learners’ autonomous behaviours, they have not provided sufficient empirical evidence for the effectiveness of any approach. Second, researchers have so far focused on changes in learning behaviours that learners have made as a result of engaging in classes where particular approaches to developing LA were applied. There seems to be an assumption that before learners participated in autonomy-based classes they all had a low level of LA even though no initial assessment was made. Third, most studies take LA for granted, and explore various approaches to fostering it. Very few studies have examined the effect of LA on language learning results.

A few studies do show a link between LA and language learning outcomes (e.g. Champagne et al., 2001; Dam & Legenhausen, 1996; Vickers & Ene, 2006). Champagne et al. (2001) documented an exploratory action research project involving a group of pre-Master’s students at a language centre in Thailand. As such, their aim was to describe their approach to language education and how learner autonomy was incorporated into their teaching. Their evidence of language improvement at the end of the programme was mainly ‘alternative’ in nature, comprising ‘portfolios of work and participants’ self-perceptions of progress together with teacher observations’ (p. 49).

Dam and Legenhausen (1996) showed how a class of twenty-one 12-year-old students at a Danish school who negotiated an English language syllabus of what they needed to learn compared with another Danish class who followed a traditional textbook-based syllabus and a text-book based English language class in a German grammar school. At the beginning of the programme, the ‘autonomous class’ learners were asked to (1) bring to class the English they see around them, (2) tell the teacher what they wanted to learn, (3) find five words they wanted to learn in a picture dictionary, (4) listen to nursery rhymes, songs and fairy tales, and (5) create word cards with their first language (L1) or pictures on one side and the English equivalent on the other side. After four weeks, 400 words appeared in the learner-created list shared among the whole class. This list was shown to contain more than the textbooks covered within the same time frame. A vocabulary recall task at the end of 7.5 weeks suggested that the autonomous group was either as good as or better than their textbook counterparts in the other two comparison groups. A receptive vocabulary test at the end of 15 weeks suggested that the autonomous group performed better than their German counterparts in auditory recognition, but not in written recognition and spelling. Despite the incompatibility of the groups, these results sound encouraging to us in that deep learner involvement in the learning process is not only possible, but also useful.

Vickers and Ene (2006) focused on a self-correction task among a group of advanced English as a Second Lanuage (ESL) composition students at a university in the USA. Thirteen learners were selected based on a test (pre-test) of 10 picture-based production items involving the past hypothetical conditional. The 13 learners scored below 90% and above 45% on the pre-test. The learners were first asked to write a paragraph on a hypothetical situation with prompts such as ‘If I, If the teacher, If the committee …’. On another day, the learners read a rule describing the past hypothetical conditional followed by a 400-word text written by a native speaker with uses of the past hypothetical conditional bolded. The learners were asked to compare their own version with this native speaker version and make the necessary corrections. Within 10 days of the first composition, the same learners wrote another paragraph on another hypothetical situation. One week later, a post-test similar to the pre-test showed that the average score became 93.05%. A delayed-test five weeks after the post-test showed an average of 92.46%. Again, the study is among the few that show concrete language gains as a result of self-directed, focus-on-form tasks. Strictly speaking, however, the study does not involve much learner autonomy, because it basically shows that learners can pick up a linguistic form from indirect form-focused instruction. The lack of a control group also compromises the strength of the argument.

2 Learner-based approaches and strategy-based instruction

While resource-, technology-, curriculum-, and classroom-based approaches to LA development concentrate on the provision of opportunities for learners to actively engage in and exercise control over their learning, learner-based approaches seek to equip learners with specific skills and strategies which enable them to take up the learning opportunities.

So far, the most convincing evidence that LA promotes learning comes from learner-based approaches of strategy training. Some of these studies focus on improving students’ metacognition and self-regulation skills; some direct students’ attention to task analysis; and others involve comprehensive training packages that attempt to improve both metacognitive management of learning and the analysis of tasks. For example, Goh and Taib (2006) explored metacognition instruction in listening for young ESL learners in Singapore. Two types of metacognitive knowledge comprising task knowledge and strategy knowledge were examined. They found a positive relationship between their metacognition instruction and the achievement of greater language learning results. Butler’s (1997) four studies applied a model of tutoring where the teacher assisted students in selecting, adapting, and initiating learning strategies based on task requirements. Butler found evidence of students’ improvement in task performance, use of strategies, self-monitoring, perceptions of self-efficacy, and patterns of attribution. Similarly, Rubin and McCoy’s (2008) experiment identifying the role of task analysis and its effects on language performance revealed gains in task knowledge and exam performance of the experiment groups. Evidence of further use of task analysis by learners in the learning process was also shown.

Strategy-based instruction (SBI) is an approach that focuses on the training of strategic learning by incorporating the training of strategies into the regular language curriculum (Rubin, Chamot, Harris, & Anderson, 2007). Rubin et al. (2007, p. 142) identified four steps as the core features of an SBI model: (1) awareness raising, (2) presentation and modelling, (3) providing multiple practice opportunities, and (4) evaluating the effectiveness of strategies and transferring them to new tasks. A clear and comprehensive SBI framework that has all the above four features built in is Chamot, Barnhardt, El-Dinary, and Robbins’s (1999) Cognitive Academic Language Learning Approach (CALLA) model. A prominent feature of this framework is the increased responsibility on the part of learners when they move from one stage to another. Teachers remove the scaffolding step by step to let learners take on their responsibilities as learners. Flexibility is another strength of this framework. Depending on learners’ knowledge and their experience in the use of a certain strategy, time allocation and teacher support for each step might be adjusted accordingly. Since SBI is integrated into the curriculum, the time allocated for each step can also be extended or shortened to fit in the content of instruction. An experimental study that explicitly used Chamot et al.’s (1999) CALLA framework is that of Gu (2007), where strategy training was integrated into the primary five writing curriculum in Singapore. The experimental groups significantly outperformed the control groups in their writing scores on both the post-test and the delayed test. In addition, Gu also reported encouraging comments from the teachers and heads of department of the schools where the SBI training took place.

In this study, we operationalize LA as a learner’s self-initiation and self-regulation at the tertiary level in the Vietnamese EFL context. Self-initiation, a learner’s volition and willingness to learn, is in turn operationalized as ‘reasons for learning’ and ‘making effort to learn’. Self-regulation, on the other hand, focuses more on the learner’s strategies and skills of metacognitive self-management, such as planning, monitoring, and evaluating. The study reported here is a follow-up of Nguyen’s (2008) survey study in which LA was found to be linked to EFL proficiency. This study will, however, go beyond the Nguyen (2008) study and see if SBI as a means to promote LA will be effective in improving EFL learning results and in improving LA. Specifically, the following research questions will be addressed:

1 Participants

This study involved both teachers and learners from a university in Vietnam. The teacher participants were chosen on a voluntary basis. All the four teachers who were teaching academic writing to third-year English-major students were invited to take part in the study. Two of them expressed their interest in becoming involved. Both of them had been teaching English at university level for 13 years. Both earned their Master’s degrees in applied linguistics.

There were four classes of approximately 130 third-year English-major students at the university where the data were collected. Forty students were training to become teachers of English and the others were training to become interpreters. The research involved all 91 students who were training to become interpreters. The other students were not invited to participate in the study partly due to teacher allocation, and partly due to the fear of introducing an intervening variable. The resulting number of male and female participants was eight and 83 respectively; all aged between 20 and 22. By the time they took part in this study, they had received approximately 1000 periods (45 minutes each) of English lessons. These included listening comprehension, speaking, reading, writing and grammar. Out of the 1000 periods of English lessons taken, there were 150 periods of writing lessons.

The 91 participants were kept in three intact groups as assigned by the Department of English where the data were collected. There were two control groups and one experimental group. The decision for the selection of experimental and control groups was made on the basis of the learners’ latest writing examination results, because the focus of the SBI training was placed on the writing skills and not on other language skills. Given the practical constraints of intact classes, it was important to ensure group compatibility as much as possible. Control group 1 (n = 26) and the experimental group (n = 37) both had a mean writing score of 6.3 on their latest writing test. Control group 2 (n = 28) scored a mean of 6.8 on the same test. These differences were not statistically significant.

2 Materials

An SBI intervention package was developed to integrate the training of metacognitive regulation into a regular 36-hour EFL writing course. LA was elicited by a general LA questionnaire and a questionnaire that targeted the planning, monitoring, and evaluating of a writing task. The questionnaires were comprised mainly of statements measured on a Likert scale of 1–5 (1 = never; 5 = always) (for more details of the development and validation process of these questionnaires, see Nguyen, 2008, 2012). Potential changes in LA were elicited by administering the same set of questionnaires at the beginning and the end of the programme. Possible improvements in writing were tested through pre- and post-tests of writing. The end-of-term writing examination, set by the university six weeks after the SBI programme, served as a delayed test. Interviews with teacher and learner participants were also conducted to obtain some qualitative insights about the SBI programme.

3 Procedures

As part of their writing course, the 91 students spent 60 minutes in class writing a comparison and contrast essay of about 150 to 200 words, which served as pre- and post-tests of their writing. They then completed the LA questionnaires the following day.

The 36-hour academic writing course with the experimental group was co-taught by one of the two teacher participants and the first author. The teacher taught students the contents of the course, which took three-quarters of the course time (27 hours). The first author spent the other quarter of the time (nine hours) teaching the metacognitive skills of planning, monitoring, and evaluating a writing task. In the two control classes where the students did not receive SBI training, the teachers conducted their 36-hour writing lessons as normal, following the same topics in the curriculum for the academic writing course and the same textbook.

The first author observed three writing lessons in each of the three classes at the beginning, middle, and end of the writing course. All students in the experimental group were encouraged to write diaries in which they reflected on the SBI training sessions and their use of the skills taught. After the completion of the programme, the two teachers were interviewed and five student volunteers from the experimental group took part in a group interview. Finally, 11 learners from the three classes volunteered to participate in a follow-up interview which was conducted by email.

For each SBI session, the CALLA framework developed by Chamot et al. (1999) was adapted. Instead of the five basic stages of preparation, presentation, practice, evaluation, and expansion, this study included seven steps: review or warm-up, preparation, explanation, demonstration, practice, evaluation, and summary (see Table 1). A Mountain Story adapted from Chamot et al. (1999, pp. 90–92) was used as teaching material across the nine SBI sessions. Each session started with the researcher summarizing the main points of the previous session or asking students questions about current events happening in their city, or activities they undertook over the weekend, as well as future events or their short-term plans. The preparation stage focused on exploring the students’ background knowledge of the topic and of the target strategy. In this stage, they were asked to read the Mountain Story in order to find the target strategy and to reflect on the experiences they had with the target strategy. The teacher then presented the strategy in focus by various techniques such as explanation, demonstration, and thinking aloud. Students were then given a block of time to work in groups on tasks that practised the strategy being taught. They were provided with worksheets, tasks, or sample essays to practise the target strategy. The class would then evaluate in various forms and reflect together about the choice, use, and effectiveness of the writing strategy, as well as on how well they had learned it and what other similar tasks the strategy could be used for. The expansion stage was realized as homework where the students were asked to use the strategy on tasks similar to those that had been practised in class. They were also encouraged to make use of their newly learned strategy for the whole writing course. As in Chamot et al.’s (1999) SBI framework, the strategy training was incorporated into the regular curriculum, and the responsibility for the choice and use of strategies was gradually transferred from the trainer to the students.

OPEN IN VIEWER

Table 1.

Structure of an SBI session.

Procedures	Activities	Responsibility	Time
Warm-up	Review of previously learned strategy/ Talking about daily events and activities	Teacher and students	3 minutes
Preparation	Finding the target strategy in the Mountain Story and eliciting students’ experiences with the use of the target strategy	Teacher and students	7 minutes
Explanation	Explaining the target strategy	Teacher	10 minutes
Demonstration	Modelling the use of the target strategy	Teacher	10 minutes
Practice	In-class actual use of the target strategy in writing tasks	Students	20 minutes
Evaluation	Evaluation of the use of the target strategy in in-class writing tasks	Students	5 minutes
Summary	Review of the content of the session with emphasis on the target strategy	Students	5 minutes

1 Does SBI training improve English writing?

Descriptive statistics and analysis of variance (ANOVA) comparing the mean scores of the experimental and control classes for pre-, post-, and delayed writing tests are presented in Table 2. Before the SBI took place, the writing mean scores of the three groups varied slightly. Control 2 had the highest score (5.860), followed by the experimental group (5.833) and Control 1 (5.120). The F value for the pre-test (F(2, 80) = 2.323, n.s.) indicates that the differences in the mean writing scores of the three groups were not significant at this stage. Immediately after the implementation of SBI, all three classes demonstrated gains in their writing scores. The experimental group, however, achieved the greatest mean score of 7.348, obtaining an increase of 1.515 marks compared with the mean score of the pre-test. The scores of Control 1 and Control 2 were 5.7760 and 6.500 respectively. These two groups both achieved a growth of 0.640 marks. ANOVA results (F(2, 80) = 8.134, p < .05) showed significant differences in the mean scores for the post writing test of the three groups. Post-hoc comparisons (Table 3) suggest that the experimental group significantly outperformed both Control 1 (p < .05) and Control 2 (p < .05).

OPEN IN VIEWER

Table 2.

Pre-, post-, and delayed-tests across three groups.

	Group	Mean	SD	N	F	Significance
Pre-test	Experimental	5.833	1.4878	33	2.323	.105
	Control 1	5.120	1.5294	25
	Control 2	5.860	1.1413	25
	Total	5.627	1.4289	83
Post-test	Experimental	7.348	1.5983	33	8.134	.001
	Control 1	5.760	1.5487	25
	Control 2	6.500	1.2829	25
	Total	6.614	1.6199	83
Delayed test	Experimental	6.2727	1.54662	33	6.930	.002
	Control 1	4.8800	1.83303	25
	Control 2	6.2400	1.20000	25
	Total	5.8434	1.65634	83

OPEN IN VIEWER

Table 3.

Post-hoc comparisons of writing scores across three groups.

Tests	(I) Group	(J) Group	Mean difference (I–J)	Standard error	Significance
Pre-test	Experimental	Control 1	.5140	.3748	.174
		Control 2	.0594	.3668	.872
	Control 1	Experimental	−.5140	.3748	.174
		Control 2	−.4547	.3989	.257
	Control 2	Experimental	−.0594	.3668	.872
		Control 1	.4547	.3989	.257
Post-test	Experimental	Control 1	1.2791 *	.4158	.003
		Control 2	.8890 *	.4069	.032
	Control 1	Experimental	−1.2791 *	.4158	.003
		Control 2	−.3901	.4425	.380
	Control 2	Experimental	−.8890 *	.4069	.032
		Control 1	.3901	.4425	.380
Delayed test	Experimental	Control 1	1.3927 *	.4105	.001
		Control 2	.0327	.4105	.937
	Control 1	Experimental	−1.3927 *	.4105	.001
		Control 2	−1.3600 *	.4379	.003
	Control 2	Experimental	−.0327	.4105	.937
		Control 1	1.3600 *	.4379	.003

Note: * p < .05.

In the delayed test, scores dropped across the three groups. Nevertheless, the experimental group managed to keep the highest score among them, with a mean score of 6.2727. The figures for Control 1 and Control 2 were 4.88 and 6.24 respectively. Table 2 shows statistically significant differences in the delayed test among the three groups in general (F(2, 80) = 6.930, p < .05). Post-hoc comparisons (Table 3) suggest that the experimental group performed significantly better than Control 1 (p < .05) but not Control 2.

Tables 4 and 5 reveal significant differences among the three tests and among the three groups. These results indicate to us that the experimental group not only outperformed the control groups in writing improvements after SBI training, they also managed to maintain their edge over the other groups in the delayed test six weeks after the training stopped. Interestingly, the delayed test was the end-of-term writing examination set by the university, not by the researchers. Unlike the pre- and post-tests which required students to write comparison and contrast essays, the delayed test required students to compose an argumentative essay. This appears to show that the participants were able to extend their strategy use beyond the comparison and contrast type of essay used for SBI training.

OPEN IN VIEWER

Table 4.

Within-participants effects for three tests across three groups.

Source	Sum of squares	df	Mean square	F	Significance
Tests	39.461	2	19.731	27.847	.000
Tests * Group	10.878	4	2.719	3.838	.005
Error (Tests)	113.367	160	.709

OPEN IN VIEWER

Table 5.

Between-participants effects for three tests across three groups.

Source	Sum of squares	df	Mean square	F	Significance
Intercept	8751.664	1	8751.664	1685.387	.000
Group	67.889	2	33.945	6.537	.002

2 SBI training and LA

Before SBI, the differences among the means on most aspects of LA did not appear to be great. The gap in self-initiation among the three groups was negligible. Reasons for learning English for all three groups fluctuated from 3.846 to 3.944. Students in Control 2 were slightly more motivated to learn English and made a little more effort to learn English than their counterparts in the other two groups. The other element of LA, self-regulation, however, varied among the three groups. The experimental group practised self-regulation skills of planning, monitoring, and evaluating less than the two control groups. Control 1 had the highest planning scores, reaching around 3.83 for both general and task-specific planning, followed by Control 2 with the mean scores of roughly 3.70 for the same elements. Both groups spent the same amount of time on monitoring their work, though not as much as on planning. However, Control 1 evaluated their written work (3.620) slightly more than Control 1 (3.437). ANOVA results show that only one of the six variables, namely Pre-SBI general planning, revealed a significant difference among the experimental group and the other groups. Post-hoc comparisons also showed that Control 1 and Control 2 performed significantly better than the experimental group (p < .05) in general planning. Control 2 also outperformed the experimental group (p < .05) in evaluating. In general, it could be said that while the experimental group showed slightly lower mean scores on self-initiation, the differences among the three groups were insignificant. However, the experimental group did not demonstrate the same level of self-regulation as the other two groups, especially on general planning and evaluating (Table 6).

OPEN IN VIEWER

Table 6.

Descriptive statistics for LA across three groups.

			N	Pre-SBI		Post-SBI		Gain (post–pre)
				Mean	SD	Mean	SD	Mean	SD
Self-initiation	Reasons for learning English	Experimental	37	3.8889	.66046	3.8408	.69019	−.0480	.43235
		Control 1	26	3.8462	.56313	3.8718	.53858	.0256	.37622
		Control 2	28	3.9444	.49367	3.8095	.60470	−.1349	.48795
		Total	91	3.8938	.58065	3.8400	.61758	−.0537	.43520
	Making efforts to learn English	Experimental	37	2.7629	.47783	2.8600	.50484	.0971	.39250
		Control 1	26	2.8129	.55243	2.8986	.51943	.0857	.39340
		Control 2	28	2.9318	.46577	3.0211	.50548	.0893	.26228
		Total	91	2.8292	.49633	2.9206	.50827	.0914	.35394
Self-regulation	General planning	Experimental	37	3.4162	.55253	3.8270	.59285	.4108	.55965
		Control 1	26	3.8308	.61108	3.8538	.64883	.0231	.65563
		Control 2	28	3.7500	.61010	3.8429	.54258	.0929	.48529
		Total	91	3.6374	.61022	3.8396	.58819	.2022	.58840
	Task-specific planning	Experimental	37	3.6036	.59622	3.9550	.59790	.3514	.54118
		Control 1	26	3.8397	.56859	3.7885	.64903	−.0513	.52232
		Control 2	28	3.7024	.55806	3.7144	.45540	.0120	.39526
		Total	91	3.7015	.57869	3.8334	.57731	.1319	.52323
	Monitoring	Experimental	37	3.3563	.54939	3.6585	.61074	.3022	.59151
		Control 1	26	3.4056	.54741	3.3182	.64089	−.0874	.55416
		Control 2	28	3.4318	.57048	3.5292	.66208	.0974	.44462
		Total	91	3.3936	.55016	3.5215	.64380	.1279	.55744
	Evaluating	Experimental	37	3.2826	.57367	3.7248	.58857	.4423	.58922
		Control 1	26	3.4371	.61712	3.4755	.68045	.0385	.37542
		Control 2	28	3.6201	.46949	3.5584	.48969	−.0617	.41109
		Total	91	3.4306	.56899	3.6024	.59206	.1718	.52999

Post-SBI mean scores (Table 6) seem to indicate that not much was changed in the two self-initiation variables; and despite a general pattern in favour of the experimental group in self-regulation variables, the differences in mean scores do not seem to be great. However, taking into consideration the experimental group’s low Pre-SBI scores, it is not surprising to find the experimental group outperforming the two control groups in terms of gain scores in self-regulation. Indeed, when we performed ANOVA analysis on the gain scores (Table 7), all the self-regulation differences between the experimental group and the other two groups were significant below the .05 level. Post-hoc comparisons (Table 8) show that none of the differences on self-initiation variables between the three groups were significant, but the majority pairs of comparisons reveal a significant advantage of the experimental group over the two control groups on all self-regulation variables but general planning and monitoring, where the advantage of the experimental class over Control 2 was not significant.

OPEN IN VIEWER

Table 7.

ANOVA for gain scores of LA across three groups.

			Sum of squares	df	Mean square	F	Significance
Self-initiation	Reasons for learning English	Between groups	.350	2	.175	.921	.402
		Within groups	16.696	88	.190
		Total	17.046	90
	Making efforts to learn English	Between groups	.002	2	.001	.008	.992
		Within groups	11.273	88	.128
		Total	11.275	90
Self-regulation	General planning	Between groups	2.779	2	1.390	4.309	.016
		Within groups	28.380	88	.323
		Total	31.160	90
	Task-specific planning	Between groups	3.057	2	1.528	6.232	.003
		Within groups	21.582	88	.245
		Total	24.639	90
	Monitoring	Between groups	2.356	2	1.178	4.047	.021
		Within groups	25.611	88	.291
		Total	27.967	90
	Evaluating	Between groups	4.695	2	2.348	10.036	.000
		Within groups	20.585	88	.234
		Total	25.280	90

OPEN IN VIEWER

Table 8.

Post-hoc comparisons of LA gains across three groups.

		(I) Group	(J) Group	Mean difference (I–J)	Standard error	Significance
Self-initiation	Reasons for learning English	Experimental	Control 1	−.07369	.11147	.787
			Control 2	.08687	.10911	.706
		Control 1	Experimental	.07369	.11147	.787
			Control 2	.16056	.11863	.370
		Control 2	Experimental	−.08687	.10911	.706
			Control 1	−.16056	.11863	.370
	Making efforts to learn English	Experimental	Control 1	.01139	.09159	.992
			Control 2	.00777	.08965	.996
		Control 1	Experimental	−.01139	.09159	.992
			Control 2	−.00362	.09748	.999
		Control 2	Experimental	−.00777	.08965	.996
			Control 1	.00362	.09748	.999
Self-regulation	General planning	Experimental	Control 1	.38773 *	.14533	.024
			Control 2	.31795	.14225	.071
		Control 1	Experimental	−.38773 *	.14533	.024
			Control 2	−.06978	.15467	.894
		Control 2	Experimental	−.31795	.14225	.071
			Control 1	.06978	.15467	.894
	Task-specific planning	Experimental	Control 1	.40263 *	.12673	.006
			Control 2	.33933 *	.12405	.020
		Control 1	Experimental	−.40263 *	.12673	.006
			Control 2	−.06331	.13488	.886
		Control 2	Experimental	−.33933 *	.12405	.020
			Control 1	.06331	.13488	.886
	Monitoring	Experimental	Control 1	.38962 *	.13806	.016
			Control 2	.20481	.13513	.289
		Control 1	Experimental	−.38962 *	.13806	.016
			Control 2	−.18482	.14693	.423
		Control 2	Experimental	−.20481	.13513	.289
			Control 1	.18482	.14693	.423
	Evaluating	Experimental	Control 1	.40380 *	.12377	.004
			Control 2	.50395 *	.12115	.000
		Control 1	Experimental	−.40380 *	.12377	.004
			Control 2	.10015	.13172	.728
		Control 2	Experimental	−.50395 *	.12115	.000
			Control 1	−.10015	.13172	.728

Note: *p < .05.

These findings were corroborated by the group interviews conducted with five students from the experimental group immediately after the SBI training. All five of them said that the way they approached a writing task had changed since they embarked on the SBI training. All stated that they did a lot more thinking before they wrote. One said that she paid more attention to the content of the essay and learned to search for information on a topic. Another pointed out that she learned how to organize her essay:

I do not put down so many ‘small’ ideas in my essay now. I choose only two ‘big’ ideas and write about them in one paragraph. I used to be very ambitious because I always wanted to write as much as I could. (Student 4: turn 1, Question 1, Group interview)

It is interesting to note that monitoring was the least used among the four components of self-regulation even after the training. Some students merged monitoring with evaluating because they did not have time for both. That said, students in the experimental group began to realize the importance of monitoring in spite of the perceived lack of time. As one student noted:

While I am writing, if I get stuck with one idea, I will just move on. I will come back to it later. It saves me time now because before I kept forcing myself to stick to it and by all means to finish what I wanted to write no matter how much time it took. I ended up not having enough time for the rest of the essay. (Student 3: turn 2, Question 1, Group interview)

As for evaluation, most students associated it with checking spelling and grammatical mistakes. One student, however, said that she checked not only spelling and grammatical mistakes, but also the development of her ideas:

I also check if I have provided enough examples or facts to illustrate my points. Sometimes I add connective words to make my essay better. (Student 1: turn 2, Question 1, Group interview)

The group interview not only reinforced the results of the post-SBI training questionnaire, but also provided insights into the aspects of each metacognitive skill that students exercised. For example, students did a lot of planning of content, organization, and language, but they failed to properly plan time for their essay. The interview suggested that shortage of time was a major reason for the lack of monitoring during writing. As far as evaluating was concerned, students focused more on the local aspects of evaluation such as spelling and grammar, rather than on global ones such as organization or idea development.

1 Enhanced self-regulation

SBI resulted in the development in self-regulation. While some learners who had not often exercised self-regulation skills previously started to plan, monitor, and evaluate a writing task, other learners who were familiar with metacognitive skills managed to widen the repertoire of each skill. For instance, in addition to content planning that learners had already undertaken before the SBI training, after the SBI they considered vocabulary they would use in their writing (Student 5, Question 4, Group interview). Others evaluated the provision of examples and facts, and cohesive devices in their essays, as well as, in most cases, checking spelling and grammar errors (Student 3, Question 5, Group interview).

Learners engaged in self-regulation considerably more after the SBI training. However, the findings also suggest that monitoring was the least exercised aspect of self-regulation, even after training. In fact, similar findings have been reported elsewhere. For example, Sert (2006) found that Turkish learners failed to monitor their own learning. Similarly, White (1995) examined the use of metacognitive strategies by classroom and distance learners and found that learners in both instructional contexts did not monitor as much as they planned and evaluated. The Vietnamese learners in this study also employed more planning and evaluating than monitoring strategies. This could partly be attributed to the possibility that many of them equated monitoring with evaluating. They tended to focus more on monitoring language problems, rather than on other aspects such as monitoring task progress, task concentration, task performance, or their own emotional reactions. Since monitoring was the least exercised among the three metacognitive skills, it is suggested that more monitoring-focused training could help learners master and execute the monitoring skill in their learning.

2 Enhanced writing performance

The experimental group’s accomplishment in writing in both the post-test and the delayed test is in tune with the results of studies on SBI, such as those by Butler (1997), Goh and Taib (2006), Gu (2007), and Rubin and McCoy (2008). These studies indicated the usefulness of strategy training to learners’ improved learning results among different age groups in a variety of contexts. The flexibility of SBI is probably a strength that makes it work effectively irrespective of the setting and the level of learners. The success of the SBI training also suggests to us that the self-regulation element of LA is teachable. This is congruent with Holec’s (1981, p. 3) view that LA is most often acquired through a systematic and intentional process.

3 SBI implementation

In addition to the training of planning, monitoring, and evaluating, several practical strategies during the SBI programme could have also explained the positive results achieved by the experimental group. These include transfer of responsibility from the teacher to the learner, explicit statement of lesson objectives, and self- or peer-correction of errors.

The success of the SBI training appeared to be closely linked to the gradual transfer of responsibility from the teacher to the learners. The responsibility was shared with learners by the teachers who engaged them in the stages of which the Vietnamese teacher is normally in charge, such as giving new input or theoretical background for an item. Acceptance of responsibility, which facilitates the management of learning and the right to make decisions in learning, is fundamental in the exercise of LA (Little, 1999; Scharle & Szabo, 2000). There has been ample evidence that a well-balanced lesson where the responsibility is shared and is gradually shifted from the teacher to learners would be beneficial for language learning (Cunningham & Carlton, 2003; Dam, 1995; Nix, 2003; Stephenson & Kohyama, 2003). It is vital that there be a proper transfer of responsibility from teachers to learners (Chan, Sprat, & Humphreys, 2002).

Having the objectives of the lessons explicitly articulated seemed to have made a contribution to enhanced LA in the experimental group. Gagne (1974) maintains that statements of learning objectives are essential not only for the teacher, but also for the learner. He argues that ‘for the students, appropriate communication of learning objectives may be an important element in the establishment of motivation and the feedback from completed learning’ (p. 74). This suggests that explicit articulation of objectives is crucial for learners. However, it is commonly observed that in the Vietnamese classroom, especially at the tertiary level, teachers normally do not tell their students about the objectives of each lesson. They might either perceive this to be a point for students to think about, or be influenced by the implicitness of the Vietnamese culture where most people prefer their interlocutors to arrive at conclusions or draw implications by themselves from what is said or taught. For these students, making both language and strategy objectives explicit in this study may well have oriented the students to the main points of the lessons and ‘pushed’ them towards the targets ahead.

Giving learners opportunities for self- and peer-correction seemed to be another SBI implementation technique that could have motivated our learners to practise using metacognitive skills effectively. Previous studies have endorsed practical values of peer- and self-evaluation (Miller & Ng, 1996; Nachi, 2003; Natri, 2007; Thomson, 1996). Empowering learners with the right to correct themselves and others is, in essence, giving them a chance to evaluate themselves and others. Teachers, by doing so, help create opportunities for their students to practise and to improve their evaluation skills, an essential aspect of metacognition.

4 LA in Vietnam

The psychological model of LA with its emphasis on the internal modification within each learner appears to be well-suited to the Vietnamese educational context, where the ‘what’ of learning is predetermined by the school curriculum and the teachers. Despite the lack of a voice in curriculum design, learners can be empowered to make decisions on how to learn. Although this is a constraint, it is at the same time a facilitating factor that reinforces the inner transformation inside learners. There may be limits to what students can do in the classroom, but this encourages them to exploit what they are currently entitled to. In essence, when learners were receiving SBI training, they were not asked to make their own decisions as to what they wanted to learn, but ended up getting more involved in their own learning processes by creating learning environments for themselves outside the classroom, interacting with peers and teachers, and actively using planning, monitoring, and evaluating skills to complete their writing tasks. It was evident that the students became more responsible for their learning by employing the metacognitive skills they had been taught. In this regard, autonomy could be seen as ‘something which is internal to the learner and which is not necessarily tied to any particular learning circumstances’ (White, 1995, p. 209). Irrespective of the limited freedom of the educational context of Vietnam, learners managed to make changes in the way they approached writing tasks to attain better writing results. In any language learning context where the decision on what students learn in class still rests with the school and the teachers, this study has shown that providing students with strategy training would be useful to enhance LA and to improve language learning results.