Recently, an unusually large systematic review of language learning and communication strategies was published, written by Javad Belali and Reza Khany. The authors analysed more than 400 studies, spanning roughly 50 years of research on second and foreign language acquisition. Their central question was straightforward: which strategies genuinely help learners improve their proficiency, and under which conditions?

The conclusion appears clear at first glance. Metacognitive strategies, such as goal setting, planning, monitoring one’s own learning and reflecting afterwards, show the strongest effects. In some controlled studies, the reported effects are even substantial. Cognitive strategies, such as summarising, repetition and structuring information, also work, but particularly in richer language environments where learners receive abundant authentic input. Affective and social strategies, including emotion regulation, collaboration and motivational support, show smaller effects. Not necessarily because they are less important, but because they are rarely studied systematically and rigorously. The review also highlights how skewed the field is: most studies focus on individual cognitive processes, while sociocultural context, multilingualism, language transfer and policy receive minimal attention. Strikingly, around four out of five studies do not even report an explicit theoretical framework guiding the research.

That is the summary. And it is precisely here that the interesting part begins.

A quick reading might suggest a simple takeaway: metacognition works, so let us invest more in metacognitive training. That is not wrong, but it is incomplete. The review not only tells us what works, but also what doesn’t. It also tells us something slightly uncomfortable about how we conduct research in education.

What appears to work most strongly often coincides with what we can measure most comfortably. Metacognitive strategies fit neatly into questionnaires, pre- and post-tests, and effect sizes. They align well with experimental designs and produce clean numbers. Affective processes, social dynamics and institutional culture are messier. They unfold slowly, depend heavily on context and resist reduction to a single scale. The consequence is predictable: they become peripheral in the evidence base.

This does not mean that emotion, motivation or collaboration matter less. It means that our research instruments are less well-equipped to capture them. The apparently modest effects of affective strategies say as much about methodological limitations as they do about pedagogy. What we are seeing is not simply a hierarchy of what learners need, but also a hierarchy of what researchers can measure reliably.

A second important message in the review concerns context. Strategies do not operate in isolation. Their effectiveness is shaped by the learning environment. The same cognitive strategy yields larger effects in immersive second-language settings than in classroom-based foreign-language contexts. That sounds obvious, yet it challenges the tendency to treat strategies as transferable tools: this works, therefore apply it everywhere. The review suggests that effectiveness is configurational. It depends on the interactions among learner characteristics, instructional design, exposure conditions, and social context.

There is a broader lesson here that reaches beyond language learning. In education, we actively search for universal recipes. We often treat evidence-informed practice as a hunt for the strongest average effect. But this review forces us to pause. Effect sizes do not float freely. They are produced within specific methodological choices and particular contexts. What we label as working depends on how we define variables, how we measure them and what we choose to isolate.

It is also telling what remains underexplored. Multilingual contexts, third-language acquisition, cross-linguistic transfer, and institutional factors are underrepresented in fifty years of research. That is not a trivial oversight. It reflects a persistent monolingual and individualistic lens, even though many classrooms today are multilingual and socially complex. We know quite a lot about how an individual learner plans and monitors their learning. We know far less about how learners co-construct meaning within dynamic, culturally situated environments. This made me think about my own work on collective student efficacy.

For me, that is the real value of this review. Not simply that it confirms the importance of metacognition, but that it reveals how selective our knowledge base is. The strongest effects tend to emerge where our measurement tools feel most at home. That is not a flaw. It is a reminder.

For practice, this leads to a balanced conclusion. Yes, it makes sense to explicitly teach learners how to plan, monitor and reflect. Metacognitive routines deserve a central place in good teaching. But that should not come at the expense of attention to motivation, safety, relationships and context, simply because those elements are harder to quantify. If we focus only on what produces the cleanest numbers, we risk overlooking the conditions that make learning possible in the first place.

Evidence is indispensable. But it is never neutral. What we know about what works is always shaped by how we choose to look. And that may be the most important lesson this review offers.