Investigating language learning strategies: Prospects, pitfalls and challenges

The overarching goals of the study of strategies are likely to remain similar, focusing on their identification, the influences mediating their use, the effectiveness of different strategic devices with respect to L2 learning, and the contributions of pedagogical interventions targeting LLS. Still, such empirical investigations will have to become much more sharply focused, better take account of the contribution of the individual and the context, and get extended to cover new contexts and situations (see Pawlak & Oxford, 2018).

One way in which the field can be taken forward is by shifting the focus from examining generalized LLS use to probing how strategic devices are employed in specific domains, such as TL skills and subsystems, translation, culture, affect or communication. While this is by no means tantamount to taking the position that overall patterns of strategy use should no longer be explored, not least because language ultimately has to be used as an integrated entity rather than a set of unrelated components, it is clear that each TL domain has its own specificity and different strategies may be needed to aid the process of mastering it. For example, learning and using grammar is bound to require heavier reliance on a unique set of cognitive strategies (Pawlak, 2018c) than participating in spontaneous interactions (Nakatani & Goh, 2007; Pawlak, 2018b) or getting to know L2 culture (Oxford & Gkonou, 2018). If such differences are glossed over and researchers merely opt for slight adjustments to existing classifications and data collection tools, they might be missing out on the distinctive strategies and constellations thereof that aid the mastery of each domain, with the effect that it may become impossible to devise successful strategic intervention. It should also be kept in mind that while there is copious research on LLS in some domains (e.g. vocabulary, reading), there are areas that have garnered limited attention from specialists. These are strategies for learning pronunciation (e.g. Pawlak & Szyszka, 2018), grammar (e.g. Pawlak, 2018c), pragmatics (e.g. Sykes & Cohen, 2018), translation (e.g. Cohen, 2001) or speaking (not just communication strategies) (Pawlak, 2018c), and it is reasonable to assume that these areas should be researchers’ first priority. On a cautionary note, there are clearly overlaps between certain domains, because, for instance, the investigation of speaking strategies will inevitably entail taking into account elements of grammar, lexis, pronunciation or pragmatics.

Whether LLS are explored in a more general manner or in regard to specific domains, large-scale questionnaire-based studies need to be complemented with the examination of how strategies are deployed in various learning tasks. Empirical investigations adopting this situated approach are still few and far between, which is disconcerting because strategy use is dependent on context and general tendencies, whether they are established for a single learner or a group of learners, may not reflect LLS employed in a specific situation (see Ehrman et al., 2003; McDonough, 1999; Oxford, 2017). When strategies for learning L2 grammar are examined, for example, a learner might indicate high frequency of monitoring oral production in response to a Likert-scale item, but then fail to report it at all when doing a grammar exercise or using a structure in spontaneous interaction as other concerns have to be prioritized in view of limited attentional resources (e.g. recalling relevant rules, searching for the right words). Importantly, such a contextualized approach does not only apply to strategic devices used for learning and using TL skills and subsystems. It is equally feasible when investigating strategies for managing affective states, since, while reliance on narratives or hypothetical scenarios (e.g. Gkonou, 2018) may be revealing, much more interesting and ecologically valid insights can perhaps be gleaned by having learners report on strategies they actually apply to grapple with genuine situations that transpire in classrooms (e.g. giving a speech in front of a class, doing a listening comprehension exercise or completing a communicative task in a small group).

The issues that also surely warrant further investigation are the wide array of variables that mediate LLS use as well as the link between strategic learning and attainment. This is because although the available empirical evidence is abundant, the research findings are far from conclusive and many matters are in need of further illumination. First, the role of some ID factors has barely been examined even though they may constitute crucial influences on how learners choose and employ LLS. Relevant examples could be enjoyment and other positive emotions, willingness to communicate, working memory, but also concepts that are just beginning to be seen as viable lines of inquiry in ID research, such as boredom (Kruk & Zawodniak, 2018), curiosity (Mahmoodzadeh & Khajavy, 2018) or grit (Yamashita, 2018). When we consider the last of these, understood as perseverance and passion for achievement of long-term goals (Duckworth, Peterson, Mathews, & Kelly, 2007), there are grounds to assume that its levels impinge upon the employment of different types of strategies. Second, some constructs that have been investigated as mediating variables in LLS use have been subject to reconceptualization focusing on their multicomponential nature, with the effect that the existing findings may have to be verified. This is perhaps best evident in the case of motivation, which is currently mainly investigated in terms of the different elements of the L2 motivational self system (Dörnyei, 2009). Third, if the spotlight in LLS research is to be redirected at specific TL skills and subsystems, the impact of the mediating variables would need to be examined with respect to those domains rather than in a general manner. After all, it is one thing to show that LLS use is related to beliefs and another to demonstrate that this also applies to the link between strategies for learning grammar and beliefs about how this subsystem should be taught and learned. Fourth, it would make sense to illuminate the relationship between strategy use, attainment, and clusters of different variables, such as motivation, personality or emotions, as research conducted to date has largely been confined to exploring the impact of single variables. Fifth, also in this area it seems indispensable to complement questionnaire studies with a more situated perspective where diverse influences on strategy use are examined with reference to the performance of particular tasks by particular individuals. For instance, it would be insightful to reveal what individual and contextual factors impact the employment of LLS in different stages of a communicative task and subsequently relate these factors to the quality of the interaction and attainment, both in terms of communicative and linguistic goals. Sixth, more research is needed that would appraise the effectiveness of the use of LLS but also go beyond establishing correlations between reported frequencies and measures of proficiency (e.g. course grades, self-assessment), not least because it is exceedingly difficult to make claims about causality. Moreover, a more nuanced operationalization of achievement is needed that might focus on particular TL skills and subsystems and in the latter case tap into the contribution of LLS to the development of explicit knowledge (i.e. such that is conscious and requires adequate time to be accessed) and implicit or highly automatized knowledge (i.e. such that is activated quickly and can thus be drawn on in spontaneous communication) (DeKeyser, 2010, 2017; Ellis, 2009). Finally, it would be instructive to uncover how the use of LLS is related to and fosters agency, autonomy and self-regulation, a link that is assumed (Oxford, 2017; Thomas & Rose, 2019) but should also be empirically verified, irrespective of whether this happens by means of carefully designed inventories or in specific situations.

Yet another intriguing question is whether LLS use remains relatively stable over time or it is subject to changes with respect to its intensity, the palette of strategies that learners opt for, or the functions that strategic devices perform. Empirical investigations could shed light on such dynamism, both quantitatively and qualitatively, over longer periods of time, looking, for example, into how patterns of strategy use change over the course of an academic year or the duration of a program. Alternatively, also in this case a more situated stance could be adopted, just as Cohen and Wang (2018) did when examining changes in the use of vocabulary learning strategies. A middle-of-the-road position is also viable, in which case fluctuations in LLS use are investigated as a function of the focus of a given lesson or its phase, a sequence of several classes, group dynamics, the requirements of a specific homework assignment and so on. It should be emphasized that such research may but by no means has to be conducted within the framework of CDST, a point that will be expanded upon below. This is the case with motivational dynamics which are often studied with recourse to this paradigm (Dörnyei et al., 2015) but can also be approached from other theoretical perspectives (e.g. Pawlak, 2012).

There are also numerous contexts, however broadly defined, in which strategies should be subjected to more intensive investigation. On the most general level, research on LLS would surely benefit from achieving a balance in the settings in which it is undertaken, whether they are defined by access to TL (e.g. foreign vs. second), educational level (e.g. secondary vs. tertiary), program type (e.g. students majoring in an L2 vs. those pursuing other majors), course type (e.g. a general course in L2 vs. one focusing on L2 for specific purposes), age (e.g. children vs. adults), language learnt (e.g. L2 vs. L3) or socioeconomic status (e.g. refugee vs. immigrant populations). More specifically, one area that deserves closer scrutiny is LLS use in different additional languages that learners are trying to master, whether in general or with respect to specific domains. The work that has been conducted thus far is extremely scarce (e.g. Haukås, 2015; Pawlak & Kiermasz, 2018) and even though such attempts have been made in studies of multilingualism, their focus has been confined to demonstrating that multilinguals use strategies more frequently than bilinguals (e.g. Kemp, 2007). There is also a paucity of research targeting the employment of strategies in content-based language instruction, whatever form it may take, be it English-medium instruction (EMI) or content and language integrated learning (CLIL) (e.g. Ruiz de Zarobe & Zenotz, 2018). The situation is similar when it comes to research into LLS use in study abroad, with the empirical evidence being tenuous (e.g. Briggs, 2015), even though strategic learning in such settings might boost self-regulation on different levels (e.g. academic, linguistic, social). Yet another context that needs to be singled out as worthy of more robust empirical investigation is the employment of LLS by third-age learners. Also here relevant studies are few and far between (e.g. Pawlak, Derenowski, & Mystkowska-Wiertelak, 2018; Piechurska-Kuciel & Szyszka, 2017), which is regrettable because L2 learning in this age group is gaining popularity and adept use of strategies could make this process truly more effective, self-directed and enjoyable (Oxford, 1989, p. 235). Finally, there is a pressing need to give more attention to the application of strategies in technology-enhanced L2 learning. This is because more and more learners are availing themselves of such opportunities by, for example, engaging in synchronous and asynchronous computer-mediated communication, regularly using educational software, drawing upon Internet-based resources, or being constantly immersed in social media. While the body of research has been on the increase (see Zhou & Wei, 2018), more studies are indispensable as this area holds a lot of promise. After all, patterns of LLS use are likely to deviate from those predominant in traditional modes of learning, other mediating variables may come into play, and new technologies offer more precise ways of recording strategy use and relating this use to learning gains.

The last research direction whose significance needs to be underscored is connected with empirical investigations of the effectiveness of pedagogical interventions composed of different forms of SBI. This line of inquiry has unique value as it is the most relevant to everyday concerns of practitioners. In fact, offering teachers concrete evidence that strategy training is effective and providing them with realistic guidelines on how to most beneficially implement it can reasonably be regarded as the ultimate goal of LLS research in the first place. As Pawlak and Oxford (2018, p. 529) emphasize, ‘the results of these empirical investigations should culminate in well-designed programs for strategic intervention in different areas, such that would be feasible and implementable, rather than merely reflecting the wishful thinking of their creators, in most cases researchers.’ In his meta-analysis of relevant studies, Plonsky (2011) found that SBI is moderately efficacious but that its utility is mediated by a host of variables. These include, among others, the choice of LLS that are introduced and practiced, the duration and intensity of the intervention, the manner in which it is carried out, or the type of program in which it occurs. It is such issues that future research should address, focusing to a greater extent on specific TL domains and ensuring that not only gains in strategy use are measured but also the effects on actual attainment, which, as mentioned earlier, should in itself be evaluated in a much finer-grained way (e.g. with reference to different types of L2 knowledge). The role of such studies cannot be overestimated as it is incumbent on researchers to produce convincing proof of the effectiveness of LLS if teachers are expected to place a higher premium on SBI.

While methodological considerations have inevitably come up on several occasions in this article, it is warranted to pull the different threads together and offer general pointers on how future research on LLS should proceed. First and foremost, it is clear that successful research on ID factors, including LLS, should seek to combine a macro-perspective and a micro-perspective. In other words, if we wish to uncover more general patterns of strategy use and factors affecting those patterns, we still need large-scale studies involving respectable samples, well-designed inventories and advanced statistical procedures. At the same time, if we want to probe deeper and shed light on the use of strategies in specific contexts, such as the completion of various learning tasks, we need research projects that would tap strategic learning in a more situated manner, thus allowing a window on the dynamics of LLS as well as the contextual and individual forces underlying it. This shows that the macro- and micro-perspective are not mutually exclusive but can complement, enrich and inform each other, whether this happens within a single study or becomes possible once adequate empirical evidence has been accumulated.

Whatever approach to examining LLS we adopt, the best option is to rely whenever possible on mixed-methods research with an eye to getting a multifaceted view of the issues that we set out to investigate (Pawlak & Oxford, 2018). Thus, quantitative data obtained through questionnaires can be better understood when complemented with qualitative information from interviews with selected participants, whereas the strategies deployed in task performance can be tapped through several instruments, such as self-assessment grids, audio recording, think-alouds or immediate reports. Naturally, there is also a need to design valid and reliable tools for assessing LLS both in general and in different domains, and then amend and update them as research findings become available. This applies in equal measure to carefully designed inventories, such as those targeting grammar (Pawlak, 2018a) or vocabulary learning strategies (Gu, 2018), and tools for tapping into LLS use in context, which could perhaps draw on latest advances in computer technology. In fact, there seems to be little justification why well-tried questionnaires should be discarded even if they suffer from obvious flaws, all the more so that all instruments used in SLA research possess their own share of shortcomings. A prime example is the SILL, which, as Amerstorfer (2018) has so convincingly shown, can still be of value to research, particularly in combination with other tools, and it can certainly be capitalized on by teachers wishing to identify or teach strategies. Obviously, we should not overlook the innovative approaches and techniques that can be harnessed in the service of LLS research. The most promising include: (1) retrodictive qualitative modeling (Dörnyei, 2014), where current patterns of strategy use can be accounted for in terms of prior experiences (see Oxford, 2017), (2) the idiodynamic method (MacIntyre & Legatto, 2011), where LLS use could be explored on a second-by-second basis in specific tasks with the help of dedicated software, (3) narratives (e.g. Gkonou, 2018), which could come from diary entries or in-depth interviews, (4) vignette methodology (Gkonou & Oxford, 2016), where learners are asked to respond to hypothetical scenarios, (5) the use of vision to reveal strategies (e.g. Oxford et al., 2018), or (6) decision-tree methods (e.g. Mizumoto & Takeuchi, 2018), which allow modeling prototypical use of strategies in context. It remains to be seen to what extent research carried out with the help of these approaches can enhance our understanding of the use of LLS.

Since some of the techniques mentioned above are anchored in CDST (Larsen-Freeman & Cameron, 2008), the contribution of this theoretical perspective to the study of LLS should be given some consideration. While all efforts to help the field reinvent itself should be appreciated and there is much potential in retrodictive qualitative modeling or idiodynamic methodology, it would certainly be a mistake to view this paradigm as a magic bullet that offers a sole solution to all the woes of LLS research. One reason for this is that adherence to this stance would necessarily bring with it the abandonment of the macro-perspective and run the risk of leaving us with numerous, thought-provoking pieces of a puzzle which, however, could be totally unrelated and prevent us from discerning the complete picture. It could be argued that the entire field of ID research faces such a danger if, in response to what may turn out to constitute little more than a passing fad, it is forced to follow this direction. It is difficult to see how interpreting LLS use with the assistance of the terminological apparatus of CDST (e.g. attractor states, signature dynamics, perturbations) can get us any closer to understanding how strategies are applied, how they contribute to effective learning or how they can successfully be taught. While the value of this theoretical stance can hardly be denied, it appears to be no more useful than, say, theories of automatization. In fact, lack of dogmatism should continue to characterize LLS research since it is this property that has always been its hallmark and helped the field withstand the barrage of criticisms in the last decade.

To quote from Pawlak and Oxford (2018, p. 532) once again, ‘whatever aspect of LLS we choose to investigate and whichever data collection tools we employ, no matter how innovative they might seem, the yardstick for appraising the empirical evidence we generate will be the degree to which it translates into everyday L2 instruction.’ A warranted question then is how researchers in the field of LLS have been faring on this front. Sadly, the answer cannot be very positive, although this reflects the gap between theory, research and practice visible in all other strands of SLA (see Nassaji, 2012). On the one hand, elements of SBI have made their way into coursebooks and the need for strategic interventions even gets mentioned in national curricula. On the other hand, the focus on strategies is typically marginal, educational materials usually fail to provide SBI in a principled manner, and teachers are reluctant to devote time to strategies with so many other priorities on their agendas (see Chamot, 2018; Chamot & Harris, 2019). In the view of the present author, the blame for this unfortunate situation must be laid at the feet of specialists, who, despite all the talk about the need to bring strategies into the classroom, have failed to demonstrate how LLS contribute to proficiency, have not done enough to convey research findings to practitioners in an approachable manner, and have often formulated unrealistic expectations about SBI. While existing guidelines are certainly helpful (e.g. Gu, 2019), it is wishful thinking to believe that entire programs in any school system can be dedicated to strategic L2 learning. Thus, modest but also feasible proposals might be more successful, with greater chances that they find reflection in curricula, coursebooks and educational materials.

However, what is perhaps the most disconcerting is that while the latest developments may have given the field a momentum and opened up new lines of inquiry, it is hard to see how they can make LLS research more relevant to practitioners. Let us focus on three examples at this point. First, with respect to the application of CDST to the study of strategies, one cannot help wondering what L2 pedagogy stands to gain from relevant research and the answer is far from clear-cut. It might not really matter to teachers that much that the functions of strategies keep fluctuating in the performance of an activity, nor do they have to be persuaded that L2 learning is complex and dynamic if the apparently straightforward act of correction involves making a number of decisions in a split second. Second, when we examine the encompassing definition recently proposed by Oxford (2017), we cannot avoid the question of who it is intended for. While it may certainly serve as a point of reference for researchers and raises our awareness of how LLS operate, it might be close to impossible to design studies which would reflect all of the definitional features at the same time. However, having been a secondary school teacher for many years, the present author is at a loss to think of ways in which practitioners could benefit from it if they care to read it in its entirety. It is unlikely that such complex, convoluted and elaborate characterizations of LLS are likely to make the concept appealing to materials developers or convince teachers that orchestrating SBI is a worthwhile endeavor. Truth be told, it is the succinct definition by Griffiths (2018) that may at the end of the day make practitioners more sympathetic to the LLS cause. Third, recent years have seen a growing, one could even say excessive, focus on strategies that learners fall back on to regulate their emotions and some studies even seek to evaluate pedagogic interventions in this area (e.g. Bielak & Mystkowska-Wiertelak, 2018). While such attempts are laudable, it is doubtful that the findings will make their way into the classroom. One reason is that teachers do not have necessary qualifications to tamper with students’ affective states and some may be reluctant to do so. Moreover, many learners might shy away from having their affectivity meddled with by people they barely know. Most importantly, though, if there is limited willingness to focus on strategies that could help learners make headway with TL skills and subsystems, it is unrealistic to expect practitioners to allot time to instruction in affective LLS. The intention here is by no means to discredit any of these research avenues but simply to emphasize that they may do little to bridge the precipice between research and practice. For this to happen, we need more consistent findings that would be disseminated among teachers in language they comprehend, tangible benefits of SBI would need to be shown, and pedagogical recommendations would need to be realistic and sensitive to the realities of specific educational contexts.