Keywords

Introduction

Investigation of second and foreign language (henceforth L2) learning strategies started with studies on “good language learners” in the 1970s, where the primary focus was on elucidating the characteristics that caused some learners to excel in L2 attainment (Naiman et al. 1978; Rubin 1975). Results of these studies consistently indicated that certain types and ranges of learner techniques and behaviors (i.e., learning strategies) were shared by successful L2 learners. They also implied that active participation in the learning process through effective use of strategies made some learners more successful than others in L2 learning. These appealing outcomes sparked further research on L2 learning strategies (e.g., O’Malley and Chamot 1990; Oxford 1990), and by the beginning of the 1990s, strategy research had established itself as one of the main areas of L2 research (Ellis 1994).

One of the reasons for such interest in L2 strategy research is the opportunity it offers for delving into the “black box” of complex L2 learning mechanism, that is, what is going on inside the brain during L2 learning (Dörnyei and Ryan 2015). Its early research thus focused mainly on the cognitive aspects of L2 learning. However, L2 research has since expanded itself into the metacognitive and the social dimensions. The affective component has also been emphasized because at the heart of successful L2 learning lies the agency of the learner that serves as the initial driving force as well as the sustaining power of strategy use. Present-day L2 strategy research, therefore, encompasses all aspects of the L2 learning: its metacognitive, cognitive, affective (agentive), and sociocultural dimensions.

In this chapter, first, an outline of the field is provided with emphasis on taxonomies of strategies, variations in strategy use, strategy clusters, and strategy instruction, all of which have been the main concerns of L2 strategy researchers and practitioners. The importance of the person-task-context configuration is emphasized herein. Then, major controversies in L2 strategy research – that is, the fuzziness of definitions, weak theoretical underpinnings, and a paucity of methodological rigor – are elaborated on. Subsequently, the possibility of replacing the concept of strategies with that of self-regulation is discussed, and its merits and problems are compared. Based on the discussion, this chapter argues that discarding the concept of learning strategies altogether is in fact not advisable and that strategies and self-regulation, in fact, complement one another. Consequently, to understand a larger picture of L2 learning processes, a theoretical framework encompassing both strategies (employed in a specific setting) and self-regulation is called for. Last, future directions of L2 strategy research and its applications in the L2 classroom are discussed.

Major Topics Relating to L2 Learning Strategies

Classification of L2 Learning Strategies

As described above, investigation in this field started with the identification of learning strategies often used by successful language learners (e.g., Rubin 1975). Efforts to identify these strategies led to attempts to classify them in the early phase of research. For example, Rubin (1981) identified two types of learning strategies: direct and indirect. The former category deals with strategies directly related to L2 learning, such as clarification/verification, monitoring, memorization, guessing/inductive inferencing, deductive reasoning, and practice. In contrast, the latter category covers strategies not directly related to L2 learning, such as creating opportunities for practice and production tricks.

Another attempt to categorize learning strategies, which is based on J. R. Anderson’s cognitive theory in psychology, was made by O’Malley and Chamot (1990). The researchers subdivided a total of 26 identified learning strategies into three categories: metacognitive, cognitive, and social/affective. The category of cognitive strategies contains strategies for more effective learning, such as mental manipulation or transformation of materials to enhance comprehension or retention, while the component of metacognitive strategies includes those used for coordinating learning processes, such as problem identification, planning for learning, increasing learning opportunities, monitoring performance, and evaluating how well one has learned. The category of social/affective strategies deals with strategies used to learn with others and control the social and affective aspects of learning. The first two categories correspond approximately to Rubin’s direct and indirect strategies. The addition of a third category, although containing such diverse behaviors as questioning for clarification, cooperation, self-talk, and self-reinforcement, was considered a crucial step in the direction of acknowledging the importance of interactional/communication strategies in L2 learning, which learners use to overcome difficulties in conveying their intended meaning.

Oxford (1990) took the categorization effort a step further and proposed a comprehensive classification system. Under the direct/indirect dichotomy, six categories were included in her system (see Fig.1), in which equal importance was placed on the metacognitive, cognitive, and social/affective dimensions of L2 learning. The very extensive nature of this classification scheme, however, produced the problems at the same time. The most prominent problems were the removal of the theoretical underpinnings of the strategies and possibility of overlapping among the subcategories. The former problem led to the criticisms from several researchers, such as Dörnyei (2005), Ellis (1994), and Skehan (1989), which will be summarized later in this chapter.

Fig. 1
figure 1

Oxford’s taxonomy (Based on Oxford 1990)

Full size image

The latter issue called attention to the fluidity of roles or functions that learning strategies may play (Cohen and Wang 2018). For example, the metacognitive strategy of planning can be considered a cognitive strategy since planning requires reasoning. Moreover, practicing patterns, a cognitive strategy, could entail the metacognitive moment of planning/evaluation or possibly an affective moment when frustration accompanies the monotonous pattern practice. Thus, a strategy may assume one function at one time, while the same strategy can play another function at another time. This view consequently resulted in the recent idea of paying special heed to examining how strategies actually work rather than to classifying them under a rigid taxonomy system (Cohen 2018; Oxford 2017; Oxford and Amerstorfer 2018; see, however, Véliz 2012, for the relevancy of a conventional taxonomy system in a highly contextualized study).

Variation in L2 Strategy Use

From the 1990s onward, in accordance with the research trend of focusing on the individual learner rather than on the learner in general, increased attention has been paid to variation in L2 strategy use. Influences on strategy use of variables such as age, gender, L2 proficiency, motivation, personality, ethnicity, culture, task types, and learning situation have been empirically examined, and it has been found that these variables generally affect L2 strategy use. For example, several researchers (e.g., Dreyer and Oxford 1996; Peacock and Ho 2003) have shown that the learner’s gender often exerts a significant influence on strategy use, with females reporting higher strategy use than males in many cultures. However, some exceptions have also been reported, with no difference or with males surpassing females in some types of strategy use (e.g., Griffiths 2003; Wharton 2000). Motivation is another variable found to significantly affect strategy use, although the causality involved (i.e., whether motivation spurs strategy use or, conversely, strategy use leads to better language performance, which in turn increases motivation) remains unclear (see Takeuchi et al. 2007, for a review of the influences of various variables).

Another example is the effect of task type. Studies have shown that the type and demands of tasks that L2 learners are engaged in can affect their strategy use (e.g., Ikeda and Takeuchi 2000; Oxford et al. 2004). Still another instance is the influence of the learning situation, that is, English as a second language (ESL) versus English as a foreign language (EFL). Takeuchi (2003), for example, revealed that relatively scarce opportunities for input and output in the Japanese EFL context have resulted in a unique pattern of learner strategy use.

The results of these studies have contributed to the notion that the use of L2 learning strategies is fully situated within a given learning context, and its use and effectiveness depend on the learner him/herself (e.g., age, gender, L2 proficiency, motivation, personality, ethnicity), the learning task at hand (e.g., types, complexity, and difficulty), and the learning context (e.g., culture, SL vs. FL differences). This person-task-context configuration (Gu 2012) of specific learning situations suggests that no learning strategy is inherently good or bad, but has the potential to be used appropriately and effectively in specific learning contexts (Hsiao and Oxford 2002).

Strategy Clusters

In addition to taxonomy and variation, researchers have increasingly pointed out that successful L2 learners employ more than one strategy at a time in a task (e.g., Macaro 2006; Oxford 2011; Vandergrift 2003), although it is true that strategies sometimes function individually (Cohen 2011). Macaro (2006, p. 328) argued that “for a strategy to be effective in promoting learning or improved performance, it must be combined with other strategies either simultaneously or in sequence, thus forming strategy clusters.” In strategy clusters, a metacognitive strategy or a series of metacognitive strategies plays an important role because it helps learners plan how they will approach a task and evaluate which strategies work best for them (Cohen 2011).

An example of a strategy cluster can be found in L2 listening comprehension. In this process, the following strategies may be deployed to attain the goal of comprehension, thereby forming a strategy cluster: using linguistic clues, paying attention to paralinguistic features, guessing unknown vocabulary, employing background knowledge, selecting relevant information, and checking interpretation with the context of the message. This cluster might be combined with another cluster of strategies regarding, for example, responding to what learners comprehend in the case of an L2 interactional task. The combination of these strategies requires the orchestration of clusters of strategies, that is, higher levels of metacognition such as choosing and evaluating strategies from a range of strategies or strategy clusters for attaining a specific goal (Macaro 2006, p. 328). Thus, investigating the orchestration in a specific learning task, or the orchestrating of “a cycle of cognitive and metacognitive strategies” (Vandergrift 2003, p. 490) to achieve a certain learning goal, has been identified as a critical topic in L2 strategy research.

Strategy Instruction: Approaches and Procedures

Identifying behaviors and techniques of successful L2 learners and then having novice learners emulate them may be a major reason for the onset of research on L2 learning strategies. Accordingly, a variety of approaches and methods of strategy instruction have been proposed. Among these, the Cognitive Academic Language Learning Approach (CALLA) (Chamot and O’Malley 1994) and Styles- and Strategies-Based Instruction (SSBI) (Cohen 2011) are prominent instructional models that have been implemented in school settings around the world. Both models focus mainly on the cognitive aspect of learning.

The CALLA model was originally developed in response to the needs of ESL children in the USA, for whom language problems often resulted in poor academic performance in school. It is an instructional model based on cognitive theory in psychology. The model attempts to integrate instruction in content areas such as mathematics, science, and social studies with the development of the language skills needed for school learning and also with explicit instruction in the use of learning strategies for academic tasks. These skills are taught in a five-stage sequence of preparation, presentation, practice, self-evaluation, and transfer. In the sequence, the responsibility of learning is gradually shifted to the students, thereby helping develop independent and self-regulated learners. The model has been used extensively in L2 classrooms (including ESL, EFL, and foreign languages other than English) and bilingual classrooms. The similarities of this model with the Content and Language Integrated Learning (CLIL) model have been pointed out by some researchers (Grenfell and Harris 2017; Oxford and Amerstorfer 2018).

The SSBI model is based on learner-focused language teaching that explicitly combines learning styles and strategy training activities with classroom language instruction. In this approach, students are given the opportunity to understand not only what they can learn in the L2 classroom but also how they can learn an L2 more effectively and efficiently. The model emphasizes both explicit and implicit integration of language learning strategies and language use strategies (Cohen 2011) in the everyday L2 classroom. Instruction in SSBI is based on the following five steps: preparation, awareness-raising, training, practice, and personalization of strategies.

As Rubin et al. (2007) summarized, most of the approaches for L2 strategy instruction, including the two models described above, have observed the following four common steps in their instruction: (1) assessing and then raising awareness of the strategies that learners are already using, (2) teacher presentation and modeling of the strategies that learners are going to learn, (3) multiple practice opportunities and the incorporation of scaffolds that are gradually removed, and (4) self-evaluation of the effectiveness of the strategies taught and transfer of them to new learning tasks and contexts. This four-stage procedure, as Gu (2019) pointed out, may be easily adopted by many L2 teachers, because it closely resembles the traditional presentation-practice-production (PPP) language teaching model. In addition to the common four-stage procedure described above, researchers seem to agree that explicit training , that is, directly teaching the purpose and use of an individual strategy, is likely to be more effective than implicit training , and that integrated (curriculum-embedded) and context-specific instruction is more effective than separate and general instruction (Grenfell and Harris 2017).

More recently, a new approach of strategy instruction has been proposed, in which the bottom-up, co-constructive, and recursive natures of instruction are emphasized (Gu 2019). In the Strategic Content Learning (SCL) model , originally developed in educational psychology, pre-selected sets of strategies are not used for instruction. Instead, the teacher and learners collaboratively engage in problem-solving to identify task demands and appropriate strategies to address them. In other words, the strategies to be taught emerge directly from the demands of each task. This also warrants the personalization or differentiation of strategies. It is true that the attempts to validate this approach are fairly limited in existing L2 research, but this can be expanded into a good alternative or a supplement for the traditional approaches of strategy-based instruction because the SCL approach is well grounded in constructivism and sociocultural theories of learning, rather than cognitive roots in the traditional approaches. Furthermore, Grenfell and Harris (2017) contend that many features of SCL already underlie much of the four common steps of strategy instruction described above, thereby proposing the integration of traditional strategy-based instruction with SCL. The resulting approach is called integrated strategy-based instruction (ISBI) .

Strategy Instruction: Assessing Its Effects

To assess the effects of these strategy-based interventions, many empirical studies have been conducted. These studies have attempted to ascertain whether L2 learning strategies are teachable to various types of learners in different skill domains and task-settings. Some of these studies have confirmed that learning strategies can be taught to various types of L2 learners and that their L2 capacities and/or learning attitude can be enhanced or improved thanks to strategy instruction (e.g., Aghaie and Zhang 2012; Goh and Taib 2006; Mizumoto and Takeuchi 2009; Nguyen and Gu 2013). Other studies, however, have reported negative or mixed results (e.g., Ikeda and Takeuchi 2003; Rossiter 2003; Vandergrift and Tafaghodtari 2010), producing uncertainty regarding the effect of strategy instruction. Amid this confusion, skeptics (e.g., McDonough 1999; Rossiter 2003) have cautioned L2 practitioners against spending too much time and energy on strategy instruction. Furthermore, doubts on the effectiveness of strategy instruction have been cast due to the methodological limitations of previous research, such as small sample sizes, nonrandom group assignment, the exclusion of contrast groups, the uncertainty of long-term effects, and a lack of valid/reliable instruments for outcome measurement (Plonsky 2011, p. 994). In addition, as explained above, successful strategy use is context- and learner-dependent. Thus, the effect of strategy instruction may also depend on (1) the context in which strategies are taught and used and (2) the characteristics of the learners involved in it. Accordingly, contextual and learner variables, such as “SL vs. FL” differences, “classroom vs. laboratory” settings, task type, and the age, personality, and proficiency of the participants, are considered influential in determining the effect of strategy instruction (Plonsky 2011).

To partially resolve this confusion, Plonsky (2011) conducted a meta-analysis regarding the effects of strategy instruction in L2 research, in which 16 individual studies and 1,099 learners were involved. Meta-analysis is a statistical procedure that focuses on contrasting and combining results from different empirical studies with the aim of finding patterns among the results or other important relationships that may come to light in the context of multiple studies. His findings indicated a small to medium overall effect (weighted d = 0.49) of strategy instruction. Skill-wise, medium to large instructional effects were obtained for treatment groups “in reading, speaking, vocabulary, pronunciation, and strategy use; more modest effects were found for writing and attitude toward language learning; negligible effects resulted from studies with listening, grammar, and general language ability as dependent variables” (Plonsky 2011, p. 1010). The variables found to moderate the effectiveness of instruction included the type of strategies taught, the learning environment, and the length of the intervention.

Plonsky’s study reported another interesting finding. Many researchers claim that metacognitive strategy use, which orchestrates the use of cognitive and affective strategies, is an essential component for promoting L2 learning through strategy instruction (e.g., Cohen 2011; Goh and Taib 2006; Nguyen and Gu 2013). However, Plonsky’s meta-analysis revealed that although instruction in both cognitive and metacognitive strategies is helpful, it is generally found that strategy instruction is the most effective when the target strategies to be taught are narrowed down.

Plonsky (2019) recently conducted another meta-analysis on the effects of strategy instruction, in which 77 individual studies with a total number of 7,890 learners were included. The overall weighted mean effect size had a d value of 0.66, which is considered a medium effect size (see Adrasheva et al. 2017, for the meta-analysis with a much larger effect size). Compared to his previous study, the 2019 meta-analysis reported a larger effect size. One reason for this result, according to Plonsky, is an increase in our understanding of how to design and carry out strategy instruction. Skill-wise, experimental groups outperformed control groups when the target skill involved speaking (d = 1.00), reading (d = 0.82), vocabulary (d = 0.63), and writing (d = 0.59). Regarding listening and general proficiency, the effect sizes were small (d = 0.06 and d = 0.05, respectively). The weak outcome of listening strategy instruction in this study corroborated the results of Grenfell and Harris (2017)’s study and Hassan et al. (2005)’s systematic review. Plonsky (2019)’s updated study also reported that (1) strategy instruction involving metacognitive strategies demonstrated much larger effects, (2) longer interventions produced larger effects, (3) teaching a single strategy led to larger effects than teaching multiple strategies at once, and (4) the effects of strategy instruction were much stronger with intermediate and advanced learners than with novice learners.

It is true that some researchers have claimed that time in the classroom is better spent learning the L2 itself rather than learning how to learn the L2. The studies summarized above, however, have shown otherwise, namely, that instruction in L2 learning strategies can be beneficial, at least, in some populations of L2 learners and in some domains of L2 language use. Consequently, many researchers in the field confidently continue to work on L2 strategy instruction, thereby contributing to its further expansion and refinements (e.g., Chamot and Harris 2019; Grenfell and Harris 2017; Oxford 2017; Oxford and Amerstorfer 2018).

Controversies in L2 Strategy Research

Definition of Learning Strategies

Around the turn of the century, a new perspective, a sociocultural one, was introduced into L2 strategy research (Gao 2007; Norton and Toohey 2001). Studies based on this perspective emphasize the importance of the situated learner. The use of learning strategies therefore can be uniquely dependent on the sociocultural contexts in which the learning is situated. The most prominent change in climate at this time, however, was that conventional L2 learning strategy research began to be criticized from various perspectives (e.g., Dörnyei 2005; Ellis 1994; McDonough 1999; Skehan 1989; Tseng et al. 2006). Due to these criticisms, interest in L2 learning strategy research was “at an all-time low” (Gu 2012, p. 330) despite the strenuous efforts of several prominent experts in the field (e.g., Cohen 2011; Griffiths and Oxford 2014; Oxford 2011), some of whom have maintained that “interest in language learning strategies remained vibrant” (Griffiths and Oxford 2014, p. 1).

One of the criticisms concerning L2 strategy research was related to the diversified conceptualizations of learning strategies or the fuzziness in the definition of this term. Ellis (1994), for example, contended that “definitions of learning strategies have tended to be ad hoc and atheoretical” (p. 533). In another instance, Tseng et al. (2006) lamented this lack of shared definition by asserting that, “there was no coherent agreement on exactly what the defining criteria for language learning strategies are and regrettably the situation remains the same today” (p. 80). One notable debate here is whether L2 learning strategies should be regarded as either observable behaviors or internal cognitive operations or both. Tseng et al. insisted that “a phenomenon is highly unlikely to be both behavioural and cognitive in nature” (p. 80), although Gu (2012) rebutted this contention by arguing that it is not a matter of categorical distinction but of a “graded degree of membership” (p. 335).

Another noteworthy debate was what distinguishes strategic learning from regular learning. Dörnyei (2005) pointed out that the distinction depends on how particular learners assume a certain behavior or technique as “appropriate” for the improvement of their L2 abilities. For example, reading aloud L2 passages many times, which is reported to be a popular L2 learning behavior in East Asian countries (Takeuchi 2003), can be strategic if the learner thinks this strategy is appropriate for attaining a specific purpose and uses it intentionally. Learning strategies conceptualized in this manner can then only be defined in terms of a learner’s intentions and his/her efforts, which has resulted in, according to Dörnyei, the unstable definitions of L2 learning strategies.

Several rejoinders to the criticisms described above have been made thus far. For example, Grenfell and Macaro (2007, p. 26) responded to such criticisms by pointing out several problems in Dörnyei’s claims. Furthermore, Oxford (2017, p. 48), in a rigorous survey that produced 33 definitions of learning strategies, offered one single but encompassing definition that might solve some of the problems pointed out by the critics. Dörnyei and Ryan (2015), in an update on Dörnyei (2005), however, still remain uncomfortable about the absence of consensus in the definition of L2 learning strategies.

Theoretical Underpinnings

The weak theoretical underpinnings of L2 strategy research have also been pointed out by several researchers (e.g., Dörnyei 2005; Skehan 1989). In fact, according to Dörnyei, an ambiguous or blurred taxonomy of L2 learning strategies, which has caused confusion among researchers and practitioners, can be attributed mainly to this issue of under-theorization. The most widely recognized taxonomy of L2 strategies (Oxford 1990), for example, includes the category of compensation strategies (i.e., strategies used to compensate for limitations in L2 use). These strategies, however, are related to language use rather than learning (Cohen 2011). Dörnyei (2005, p. 168) thus argued, “the two processes are so different both in terms of their function and their psycholinguistic representation that they are best kept separate.” Note, however, that even in language use, L2 learning opportunities exist.

Another confusion (Dörnyei 2005) derived from the weak theoretical underpinnings in L2 strategy research is the inclusion of the category “social/affective strategies” (i.e., strategies used to learn with others and to control the social and affective aspects of learning) in O’Malley and Chamot’s widely used taxonomy (O’Malley and Chamot 1990), which is based on J. R. Anderson’s cognitive theory in psychology. The “social/affective” category encompasses such miscellaneous behaviors as “cooperation,” “questioning for clarification,” and “self-talk,” none of which are directly related to the cognitive basis of this taxonomy. In fact, Hsiao and Oxford (2002) confirmed that the explanatory power of the taxonomy was increased if this category was divided in two separate categories, which is, according to Dörnyei (2005), one indication of under-theorization in L2 strategy research. However, as O’Malley and Chamot themselves admitted, the category of social/affective strategies was not their main concern from the beginning, and thus it is not surprising that this specific category received less theorization in their classification system.

The criticisms described above have led to the contention that there is no consensus on strategy categorization (Dörnyei and Ryan 2015). Oxford (2017), however, countered this line of contention by arguing that L2 learning strategy research needs to have a somewhat eclectic theoretical foundation because it may include behaviorist, cognitive, metacognitive, communicative, and sociocultural dimensions, thereby producing a “web of interlocking theories” (p. 60). This eclectic nature of strategy research inevitably results in the difficulty of straightforward categorization based on one single theoretical perspective. In a different vein, Macaro (2006) also responded to the lack of consensus regarding strategy categorization by abandoning the attempt to achieve an “all-encompassing definition of a strategy” (p. 320) and instead choosing to list a series of features essential for describing a strategy: location, size, abstractness, relationship to other strategies, explicitness of goal orientation, and transferability.

Methodological Rigor

Strategy research has also been under fire regarding the ways in which studies have been conducted. Tseng et al. (2006) criticized the research instruments used to assess strategy use. Among these instruments, the most prominent is the Strategy Inventory for Language Learning (SILL) (Oxford 1990). The SILL has been the mainstay of L2 strategy research (Dörnyei 2005; Oxford 2011). It is also an easy-to-use inventory for L2 teachers to assess learners’ strategy use, and, accordingly, it has been used widely. The inventory is made up of six subscales based on Oxford’s taxonomy of learning strategies and includes 50 items (ESL/EFL version) in total. Scale scores are obtained by calculating the mean of the item scores within each subscale. Its scale descriptors contain 5-point Likert scales, ranging from “never or almost never true of me” to “always or almost always true of me.”

Tseng et al. (2006) criticized the instrument by asserting that “the scales in the SILL are not cumulative in nature and computing mean scale scores is psychometrically not justifiable” (p. 83). They argued that the items in the SILL focus on specific strategic behaviors, not on general trends and inclinations corresponding to a certain learner trait behind a scale. Thus, “one can be a generally good memory strategy user while scoring low on some of the items in the memory scale” (Dörnyei 2005, p. 182). In other words, the instrument does not ensure that each item in the scale has a linear relationship with the corresponding underlying trait of learners. In addition, the SILL adopts scale descriptors indicating frequencies of strategy use. Consequently, a high score on the SILL can be achieved by using as many different strategies, as frequently as possible. A high score therefore does not guarantee that the learner is an adept strategy user, but rather that the learner is a frequent user of many different L2 learning strategies. In fact, research has suggested that unsuccessful learners frequently use a variety of strategies, but in an unconnected or random way (e.g., Vann and Abraham 1990). With L2 strategy use, what matters most is the quality, not quantity, of the strategies employed, and “the more, the better” is not always the case (Yamamori et al. 2003). Thus, the casual use of the SILL or SILL-type inventories for research can be problematic regarding psychometric properties. Instead, Dörnyei and Ryan (2015) recommended the use of the Motivated Strategies for Learning Questionnaire (MSLQ) , which they believe is more psychometrically robust and theoretically valid (see Teng and Zhang 2016, for the problems concerning the MSLQ).

Yet another problem related to strategy inventories is the misuse of statistical procedures for Likert-type scale data (Griffiths and Oxford 2014). Likert-type scales, which have been utilized frequently in strategy inventories, produce ordinal data rather than interval or ratio data. For the former type of data, non-parametric statistical tests (such as Wilcoxon signed rank tests), rather than parametric ones (such as t-tests), are considered appropriate for use in terms of statistical robustness. Many studies in strategy research, however, have assumed that Likert-type categories constitute interval measurement and accordingly have applied little caution when using parametric statistical procedures, which are optimized for interval or ratio data. Such a misuse of statistical procedures might have caused problems in data analysis in existing strategy research, although Griffiths and Oxford (2014) insist that statisticians have emphasized “the minimal difference caused by using parametric tests rather than non-parametric tests when analysing results of scales with non-equal intervals” (p. 4). In this connection, the use of the Rasch model (Bond and Fox 2015), which converts ordinal scales into interval ones, might be a viable solution to avoid the misuse of statistical procedures and increase the robustness of the analyses involved in the use of strategy inventories.

Strategy inventories have also been criticized for their lack of context. They tend to ask learners to generalize their actions across task-settings or contexts (or even across past experiences). The results consequently reflect learners’ perceived strategy use rather than their actual use of strategies in specific tasks. As many researchers have pointed out (Oxford 2017; Takeuchi et al. 2007), however, the influence of task-setting or context on L2 strategy use is enormous. Thus, avoiding the decontextualized use of a single pre-existing strategy inventory has been called for in recent research (e.g., Oxford 2017; Woodrow 2005).

As described above, the conventional use of inventories such as the SILL might fail to capture the reality of L2 learners’ strategy use. To grasp its real and dynamic use in context, Macaro (2006) proposed a theoretical framework in which he defines learning strategies in terms of a goal, situation, and mental action. He then argued that the effectiveness or non-effectiveness of strategies derives from the way they are used in tasks and learning processes. This framework introduced a rationale for studies using a strategy inventory in a specific task-setting or context (Hsiao and Oxford 2002; Ikeda and Takeuchi 2000). The use of a scenario-based questionnaire, which embeds the responders in a specific kind of task-setting by providing them with L2 learning-related scenarios, has also been proposed recently regarding the use of affective and meta-affective strategies (Gkonou and Oxford 2016). In addition, some attempts to adapt the SILL to a specific population in a specific context have been made. For example, in the context of primary and secondary ESL learners in the USA, Ardasheva and Tretter (2013) showed how the SILL could be adapted based on recent conceptual controversies to yield a more robust, population-specific measure.

The studies mentioned above indicate that the criticisms of the lack of context in the use of strategy inventories can be solved through the creative efforts of professionals in the field. In this connection, researchers (e.g., Gao 2007; Grenfell and Harris 2017; Oxford 2011; Woodrow 2005) also have argued for an increase in the use of qualitative research methodologies such as analyzing think-aloud protocols, interviews, conversations, and narratives, all situated in specific task-settings, thereby providing contexts for strategy use to which researchers, practitioners, and learners can refer. Likewise, a sociocultural approach to L2 strategy research is also called for to paint a holistic picture of L2 learners’ strategy use in context (Gao 2010; Oxford 2017).

L2 Learning Strategies and Self-Regulation

Replacing Learning Strategies with Self-Regulation

Reflecting on the variety of debates outlined above, Dörnyei and his colleagues have proposed abandoning the concept of L2 learning strategy and replacing it with that of self-regulation (Dörnyei 2005; Tseng et al. 2006). This proposed change entails shifting the research focus from concrete techniques or behaviors (i.e., learning strategies) to the underlying capacity or the trait behind the use of strategies, be it self-regulation or another. This shift might coincide with Chamot and Rubin’s (1994) contention that the successful learners cannot be described by their use of strategies alone (behaviors) but rather by their ability (trait) to understand and develop a personal set of effective strategies. Macaro (2001) also agreed with the existence of some underlying trait of successful learners by contending that L2 learners with proactive tendencies appear to learn best.

To promote this line of inquiry, therefore, Tseng et al. (2006) developed a survey instrument measuring self-regulatory capacity using action control strategies, based on volitional research in educational psychology. In their study, these strategies were situated in one skill domain, that is, L2 vocabulary learning. The newly developed instrument, the Self-Regulating Capacity in Vocabulary Learning (SRCvoc) , is made up of five components: commitment, metacognitive, satiation, emotion, and environmental controls. The SRCvoc attempts to measure the intentions and abilities to learn L2 vocabulary rather than the learning techniques or behaviors involved in doing so.

Complementing Learning Strategies with Self-Regulation

One thing that needs to be pointed out here is that although Tseng et al. (2006)’s work offered the field a new way of research, most of the studies relating to the SRCvoc by then were either validation or adaptation studies (e.g., Mizumoto and Takeuchi 2012), indicating that “self-regulation has not greatly expanded in its scope since this time” (Rose et al. 2018, p. 158). Furthermore, several researchers have taken exception to giving up on the concept of learning strategies altogether and replacing it with that of self-regulation. Gao (2007), for example, argued that Tseng et al. (2006)’s model of self-regulation and the models of strategy use are in fact complementary, as they measure the beginning and end product of the same event, respectively. Stated differently, a model of self-regulation is “about looking at the initial driving forces,” while those of strategy use examine “the outcome of these forces” (Rose 2012, p. 95), both of which constitute strategic learning . As Dörnyei (2005) himself admitted, the SRCvoc does not measure strategy use (i.e., the end product) but rather the learner’s underlying trait (i.e., the beginning) that will eventually result in strategy use. Thus, considering the still-outstanding necessity of investigating the end product, especially for the purpose of learning/teaching, research on strategy use in a specific task-setting is still worth conducting, even after the introduction of self-regulation. As Oxford (2017) and Rose et al. (2018) contend, learning strategy use undoubtedly remains an essential component, even in the face of self-regulation (see also Thomas and Rose 2018, for recent developments).

Correspondingly, evidence indicating that Tseng et al. (2006)’s model of self-regulation is not immune to criticisms and that it needs a complementary scheme for more effective explanation of L2 learning processes has also been mounting. Rose (2012), for example, showed that the taxonomy introduced in Tseng et al. (2006)’s model suffers from the same “fuzziness” in definitions for which strategy research has been criticized. He argued, based on his empirical inquiry, that “environmental control may not be a separate category of control in itself, but a self-regulatory mechanism or strategy to control other forms of motivation” (p. 95). He nevertheless posited that self-regulation is a useful concept in exploring strategic L2 learning. Furthermore, Ranalli (2012), by comparing alternative concepts of self-regulation, argued, “depending on the model one adopts, self-regulation is not only compatible with the study of specific strategies but useful for shedding new light on strategy research and integrating it with research in other related areas” (p. 357).

Based on these arguments, Oxford (2017), Ranalli (2012), and Rose et al. (2018), among others, have maintained that hastily discarding L2 learning strategy research as a field is not advisable, nor is it beneficial for L2 learners, either. Instead, as Rose (2012) insisted, research frameworks encompassing both self-regulation and strategy use in specific task-settings need to be introduced to understand the whole picture of L2 learning processes. One example of such frameworks is the Strategic Self-Regulation (S2R) model proposed by Oxford (2017). The model attempts to explain L2 learning through the concept of self-regulation; the use of tactics, strategies, and meta-strategies; and several types of knowledge underlying the use of meta-strategies. The model includes strategies for three key dimensions of L2 learning that learners can regulate: cognitive, affective, and sociocultural-interactive. Covering all three dimensions are the mental processes, that is, meta-strategies, that help learners manage their strategy use. Another feature of this model is the inclusion of tactics in addition to strategies. Tactics are the specific manifestations of a strategy by a learner in a certain setting for a specific purpose. In other words, tactics are highly specific uses of strategies. Strategies, on the other hand, are more abstract, broad, and general and can give rise to many kinds of tactics depending on the learning situation and the purposes of an L2 learner. The S2R model can be an appropriate theoretical foundation encompassing both self-regulation and strategy use in specific task-settings or skill domains, even though this model has not yet been adopted widely (Rose et al. 2018) and that some concerns could be raised for the usefulness of distinguishing strategies from tactics in its application in the classroom.

Recently, empirical attempts to connect the constructs of strategies and self-regulation have also been emerging. Among them, one example worth mentioning is Teng and Zhang (2016), in which the writing strategies of 790 undergraduate students learning English in China were investigated by using a newly developed questionnaire. The questionnaire, Writing Strategies for Self-Regulated Learning Questionnaire (WSSLQ) , is based on the concepts of both self-regulation (the driving forces) and L2 strategy research (the outcome of the forces). The study successfully validated the questionnaire by showing self-regulation as a higher-order construct, which explained the correlations with lower-order writing strategies, thereby demonstrating the compatibility of the two concepts. It also verified the relationship between strategy use and EFL writing proficiency.

Another example is Ziegler (2015)’s large-sampled (N = 572) study of grades 4–9 EFL students in Germany, in which the relationships of the SRCvoc construct with the motivational characteristics of self-regulated learners (motivated learning strategies) measured by the MSLQ motivation scale were investigated using regression analyses. The results showed that the SRCvoc predicted each of the MSLQ characteristics significantly, thereby confirming the SRCvoc as a reliable and valid measure for the behaviors exhibited in the performance phase of the self-regulated learning cycle. At the same time, Ziegler (2015)’s study established an empirical link between the constructs of strategies and self-regulation, consequently providing the evidence to support the complementarity of the two constructs.

Yet another example was a recent study on the use of writing strategies by 37 Japanese EFL university students (Sasaki et al. 2018). In the study, the self-regulatory assumption of Oxford (2017)’s definition of learning strategies was adopted and the use of three writing strategies interacting with other relevant cognitive and environmental variables was investigated longitudinally using a mixed methods approach. The results revealed that the self-regulatory perspective was useful in modeling the systematicity and individuality observed in strategy development, thus lending support to the complementarity of the concepts of strategies and self-regulation.

Conclusion: Future Directions of L2 Strategy Research and Its Applications

As shown above, evidence that the concepts of learning strategies and self-regulation are compatible and may even be complementary has been mounting. Dörnyei and Ryan (2015), in an update of Dörnyei (2005), also agreed with this movement toward complementarity. Research on L2 learning strategies, as well as its applications in the classroom, is thus unquestionably worth conducting even after the introduction of self-regulation into the field. This chapter accordingly concludes with some descriptions of possible future directions of L2 strategy research and its applications in and outside of the classroom.

First, future research should focus on a specific population in a specific task-setting and context, since strategy use depends largely on learners, tasks, and contexts. In this connection, the sociocultural perspective (including activity theory and collaborative learning) as well as that of complex dynamic systems theory (Larsen-Freeman 2016) might provide a better landscape for strategy research as they place special emphasis on L2 learning in a specific context. Future research should also pay heed to such theoretical frameworks as self-regulation and learner autonomy to better investigate how strategy use and other higher L2 learning processes are related to each other. By examining the use of learning strategies situated in a solid theoretical framework, a better general understanding of L2 learning processes can readily be obtained.

Second, strategy research should observe the trend in the direction of qualitative data collection methodologies including narratives, interviews, diaries, journals, portfolios, and think-aloud protocols. In particular, longitudinal data collection using these methods would be welcome. Furthermore, qualitative data analysis using such methods as the grounded theory approach might become prevalent (Oxford 2017). However, conventional quantitative research using strategy inventories will undoubtedly continue to be conducted, focusing on a specific population in a specific domain, task-setting, and context. Similarly, to assess learners’ strategy use, strategy inventories will continue to provide valuable information to L2 practitioners, especially at the beginning and evaluative stages of strategy instruction.

Use of advanced statistical procedures such as decision tree analysis (e.g., Mizumoto and Takeuchi 2018), structural equation modeling (SEM) (e.g., Ardasheva 2016), and the Rasch model and the use of effect sizes might also gain momentum. Decision tree analysis is a form of predictive modeling in which learner behavior is analyzed, and future courses of action are decided. SEM is a type of multivariate statistical procedure used to analyze structural relationships between measured variables and latent constructs. The Rasch model introduces a method of converting ordinal scales into interval ones, while the effect size is a quantitative measure of the strength of a phenomenon, which is presumed to be free from the effect of sample size. Statistical procedures such as these could yield the ways to circumvent various problems pertaining to the traditional use of statistics in the field. In addition to the improvement of quantitative procedures, mixed methods approach, which combines quantitative and qualitative methods, will also be used widely.

Third, assessment issues in strategy instruction, along with concrete instructional techniques, could become prominent topics in the field. Improvement of L2 abilities through strategy instruction is the original motive behind L2 strategy research. However, assessment of the improvements has not drawn due attention thus far. For the successful implementation of strategy instruction, the advancement of assessment methods is indispensable. Thus, more accurate measurements of the self-regulatory capacities, L2 abilities, learning attitudes, emotions, and acquisition/transfer of strategies need to be further explored in future studies. Concrete techniques of strategy instruction (including needs analysis, strategy selection, sequencing, combination, presentation, and evaluation) as well as materials development and the use of technology (Ranalli 2018) also need to be further investigated. The investigation should, of course, be situated in a specific skill domain, task-setting, and learning context, so that more appropriate implementation of strategy instruction would become possible.

Cross-References