Those among us who you follow debates about languages education are likely aware that Modern Foreign Languages are facing a crisis. With the exception of English, the global lingua franca, fewer young people study languages, participation rates are stubbornly low (unless language courses are mandatory), and exam outcomes reinforce the sense of decline. This is the broad picture across national contexts, and it makes for gloomy reading.
But is this really the whole story? Or might the familiar narrative obscure important variation between schools, and with it the hope of change? I recently came across a new article1 by Zhu Hua, Elin Arfon and Ann-Marie Hunter (2025) in The Language Learning Journal, which invites us to think differently. In the article, the authors show how local contexts, school cultures, and histories interact in ways that shape very different experiences of language learning, and —by using Complexity Theory as a theoretical lens— they explain how these interactions lead to quite different outcomes. Their findings challenge some familiar assumptions and, perhaps, offer a more hopeful way of thinking about the future of languages education.
This is a rich and thought-provoking piece, I believe, which rewards a closer look, so in what follows, I will try to unpack its content; and also add some thoughts about what the authors’ findings and insights imply for language education policy. If this sounds like something you’d like to think along with me, let’s read on.
A complexity perspective on language education
Key to the authors’ perspective is the conceptualisation of schools as Complex Adaptive Systems.2 This reflects the fact that schools bring together into one entity people and policies, rules, routines and resources, traditions and aspirations, ideologies and lived experiences; and that they all interact to produce learning outcomes. This is a perspective I have developed in more detail in my own work (e.g., A language school as a complex system3 and The intentional dynamics of TESOL4), and as I have argued (sometimes with colleagues), it is an intuitive way to make sense of phenomena we see in language education.
What is Complexity?
Complexity theory, or Complex Dynamic Systems Theory (CDST),5 as seems to be the preferred usage these days, takes as its starting point that systems (like schools) are comprise “multiple interacting elements [which] adapt and evolve over time” (Hua et al., 2025, p. 3). The dense web of relations between these elements, and the complex ways with which these elements interact with each other and the environment, produce stochastic activity. This means that the outcomes that are never fully predictable but they are not random either.
Features of schools as complex systems
In their article, Hua et al. (2025) point out that schools, viewed as complex systems, exhibit the following properties.
Openness
Complex systems, like schools are open, in the sense that they constantly interact with the wider environment in which they are embedded. For instance, the operation of a school will be, directly or indirectly, influenced by economic conditions, governmental priorities etc. In fact, it is perhaps better to think of systems having interfaces rather than boundaries. No matter what you call them, the boundaries of a system are are ‘fuzzy’ and hard to trace with precision.
Non-linearity
Non-linearity, a core property of complex systems, means that outcomes are not proportionate to effort. Language teachers know that sometimes considerable investment in time and effort can have little obvious effect. This happens because systems have the ability to constantly adapt to change (see below). But equally, seemingly trivial events (aha! moments) can trigger impressive results, if they push a system past its tipping point.
Emergence
Complex systems are sites for emergence. That is to say, because of complexity, the whole system is “more than the sum of its parts”. Properties are generated which are not easy to explain as the aggregate of the system components. Sometimes, for instance, classes have a ‘life of their own’, as Juup Stelma and I describe in Chapter 7 of the Intentional Dynamics of TESOL.
Feedback
Complex systems contain feedback loops. This means that activity within a system amplify or dampen out activity elsewhere. Motivated behaviour, for example, might produce success in learning, and this might bring about additional motivation in a ‘virtuous circle’.
Attractors and tipping points
In A language school as a complex system, I use the term ‘affordance landscape’ to describe the range of possibilities that a system has. This typically consists of attractors and tipping points. Attractors are patterns of behaviour that resist change. For example, think of persistent mistakes or fossilised language are attractors in learning. Tipping points are moments of when the system undergoes radical restructuring. My favourite example is the communicative turn of language education in the early 1980s.
Path dependence
Path dependence describes describes how the present state of a system influences its future states. This is, broadly, what Juup Stelma and I describe as the ‘historicity‘ of a system. In our work we argue that historicity (or ‘path dependence’) is what produces normative activity, i.e., doing things in schools because of the force of tradition and past knowledge. One example of this would be the resilience of grammar instruction even in contexts where teachers and management nominally espouse communicative language teaching (Kostoulas, 2014).
Adaptiveness
Implicit in the name Complex Adaptive Systems is the idea that systems constantly change. In fact, even when they appear tranquil, systems are in what we call ‘dynamic stability’. My Year 1 classes, for example are dynamically stable: every June my students leave, and every September a similar number of 18-year-olds enroll.6 Changes in the environment can trigger small-scale change in individual parts of the system, which we call adaptation. But also, changes that push a system past its tipping point bring about system-wide restructuring (‘catastrophic’ change).
Heterogeneity
In saying that complex systems are heterogeneous, we mean that their constituents are not just diverse but also belong to different categories: A school comprises students and teachers (people), as well as rules and routines (activity) and so on.
Self-organisation
Complex systems resist top-down control, or –in the authors’ words, they are “difficult to manipulate, control or manage”. This also means that causality works differently in complex systems: it’s less about cause and effect, and more about ‘steering’.
The Iceberg Model
Building on this conceptualisation, Hua, Arfon and Hunter put forward a striking metaphor: the “languages education iceberg“. The iceberg consists of two parts, above and below the waterline. Above the waterline we see the visible data points that dominate public debate: things like participation rates, exam outcomes, policy decisions. But the real complexity lies beneath the surface, where we have the interacting mesh of factors like leadership commitment, parental attitudes, teacher recruitment, school ethos, pupil motivation, even the school’s historical trajectory. These submerged elements interact continuously, shaping what becomes visible above the waterline. Look only at the tip, and you miss the mass beneath.
The way Hua, Arfon and Hunter connect these ideas to real schools is particularly illuminating. If you’re curious to see how these abstract principles play out in practice, you like what follows. The findings of the article make the complexity of language education vividly concrete.
Findings from the study
In their study, Hua et al. (2025) bring together survey and case study data from the National Consortium for Languages Education (NCLE) programme in the United Kingdom, which ran between 2023 and 2025.
The survey study, which draws on evidence from 108 schools, suggests that the national averages reported in the official UK statistics conceal very different lived realities across schools. For example, participation in language courses ranged from 26% to 80% depending on how school policies. Even among schools where language courses were compulsory, attainment ranged from less than 30% to almost 100%. In other words, compulsory entry delivered strong results in some contexts but in others it led large numbers of learners to fail in securing a pass.
To better understand the variation, the authors compare case studies of two different schools. School 1 is a selective grammar school7 with a strong tradition of languages and a longstanding International Baccalaureate (IB) ethos. In this school, it appears that parents and school leadership strongly support language learning. Student motivation is sustained by international trips, extracurricular culture clubs, and a multilingual intake. That is not to say that there are no challenges, such as large classes or heavy workloads. However, but these are outweighed by a culture that normalises and values languages.
The picture in school 2, a non-selective academy8 seems somewhat more troubled. Although this school also had an IB tradition, parents appeared to question the value of compulsory languages, pupils lacked motivation, and teachers reportedly faced relentless pressure. In this school committed leadership and cultural investments, such as international visits, seemed to have little impact. On the whole, the school struggled to maintain its compulsory language policy, and was experimenting with alternatives.
Access the article
Click on the button to access the article, Complexity in secondary languages: How the local context shapes languages education in England (Hua et al., 2025).
What does this study tell us?
These findings complicate any assumption that a single policy will “fix” languages education. Accountability frameworks interact differently across schools, depending on intake and ethos. Some pupils thrive under compulsory study policies; others are set up for frustration.
The authors suggest a more context-sensitive approach. Rather than imposing uniform solutions, policymakers should support schools in developing their own pathways, drawing on their unique histories and communities. They also raise the possibility of alternative qualifications for pupils for whom GCSEs are inaccessible, an idea that challenges the dominance of a single route.
This approach seems to be in line with the perspective I often take in my writing, namely that language education needs to build on local particularity rather than aspire to uniformity. What works in one setting may not translate neatly to another, causing ’tissue rejection’. This happens because each school —indeed, each classroom— operates within its own ecology of learners, teachers, and community influences. Policies designed around averages risk overlooking this local knowledge and, in doing so, miss opportunities for meaningful improvement.
In my own work, I have often argued that sustainable change in language education grows from within systems, not through externally imposed reform. Complexity theory helps articulate why: it reminds us that education is not a machine to be fine-tuned, but a living network of relationships, histories, and aspirations. Recognising this complexity does not make reform impossible. It simply reframes it as a process of nurturing conditions under which positive change can emerge.
Seen this way, Hua, Arfon and Hunter’s call for context-sensitive policy is both pragmatic and hopeful. It acknowledges the diversity of schools, teachers, and learners—not as a problem to be managed, but as a source of strength and creativity for the future of languages education.
Closing Remarks
The familiar narrative of decline in language learning is not wrong, but it is incomplete. If we limit ourselves to the “tip of the iceberg”(i.e., participation figures and exam outcomes) we risk misinterpreting what is really happening in schools.
A complexity perspective invites us to look deeper: to see schools as evolving systems shaped by multiple, interacting forces. For policymakers, this means resisting one-size-fits-all solutions. For teachers and school leaders, it means reflecting on the unique configurations that sustain (or hinder!) language learning in their own settings.
In short, the story of language education is not just one of decline. If you look hard enough, it is also one of variation, adaptation, and complexity. And that, perhaps, is a more hopeful way to begin thinking about the future.
I’d be interested to hear how these ideas resonate in your own context. Do you see similar patterns of variation or complexity in the schools and classrooms you know? Feel free to share your reflections in the comments section below.
Related posts
Summary
- National averages hide wide variation in how languages are taught and learned across schools.
- Schools function as Complex Adaptive Systems, i.e., dynamic networks where people, policies, and contexts interact in unpredictable but non-random ways.
- We can use the iceberg model to visualise how visible outcomes (participation, exams, policy) rest on deeper factors such as leadership, ethos, motivation, and history.
- Even with the same policy, schools differ greatly. Real improvement depends on context-sensitive, locally grounded approaches, not one-size-fits-all reforms.
Key questions & answers
What is Complexity Theory in education?
It’s a way of understanding schools as complex systems, i.e., networks of people, policies, and practices that interact dynamically. Instead of linear cause-and-effect, complexity thinking highlights interdependence, emergence, and adaptation.
Why does this perspective matter for languages education?
Because it explains why the same policy can succeed in one school and fail in another. It shifts attention from “what works” in general to “what works where, for whom, and under what conditions.”
What does the “languages education iceberg” illustrate?
It shows that visible outcomes like exam results rest on deeper, often invisible factors: leadership, ethos, resources, motivation, and school history. To understand the visible “crisis,” we must also look beneath the surface.
Does this mean policy doesn’t matter?
Policy matters, but its effects depend on context. Complexity thinking calls for context-sensitive policy: one that enables schools to build on their own histories, communities, and capacities rather than enforcing uniform models.
How can teachers and school leaders use this insight?
By reflecting on their own systems: identifying what supports or constrains languages learning locally, and nurturing conditions that allow positive change to emerge. Small shifts that can have large effects over time.
Footnotes
- Hua, Z., Arfon, E., & Hunter, A. M. (2025). Complexity in secondary languages: how the local context shapes languages education in England. The Language Learning Journal, 1–21. https://doi.org/10.1080/09571736.2025.2557456 ↩︎
- The complexity literature has a rich nomenclature, with multiple terms describing different nuances of the same construct. While many people use the term Complex Dynamic(al) Systems to indicate that systems are in constant flux, the term adaptive can be helpful in showing that this change is the process of constant interaction with the system’s environment. ↩︎
- Kostoulas, A. (2018). A language school as a complex system: Complex systems theory in English language teaching. Peter Lang. ↩︎
- Stelma, J. & Kostoulas, A. (2021). The intentional dynamics of TESOL. De Gruyter. ↩︎
- In my own writing, I sometimes reserve the term ‘complexity theory/thinking’ for a ‘softer’ use of CDST, which uses concepts developed in the sciences as metaphors for language education. Here I will use both terms interchangeably. ↩︎
- And yet, year after year, I age. ↩︎
- In the UK system, grammar schools are state-funded selective secondaries, which admit pupils based on academic ability. This is typically determined by entrance exams taken when one is 11 years old. ↩︎
- Academies are independently run schools in England, which operate with greater autonomy than traditional state schools. ↩︎
About me
Achilleas Kostoulas is an applied linguist and language teacher educator at the University of Thessaly in Greece. He holds a PhD and an MA in Teaching English to Speakers of Other Languages from the University of Manchester (UK) and a BA in English Studies from the University of Athens in Greece.
A lot of his research in language education has adopted a Complex Dynamic Systems Theory (CDST) perspective, starting with the pioneering Manchester Roundtable on Complexity and ELT in 2015. More recent examples of such work include the monographs A Language School as a Complex System (2018) and The Intentional Dynamics of TESOL (2021, with Juup Stelma), as well as multiple articles and encyclopedia entries on CDST in language education.
In addition to his work on complexity, he has researched teacher research literacy, critical perspectives on language education, and the role of AI in language teaching and learning. He works closely with colleagues on several international projects, including the ECML initiative Artificial Intelligence in Language Education and the Erasmus+ project Research Literacy of Teachers (ReaLiTea).
What drives his teaching and research is a strong belief in research-informed practice and critical reflection, and he welcomes opportunities to exchange ideas, collaborate, and speak on complexity in language education, teacher research literacy, critical applied linguistics, and AI in language education.
About this post
I wrote this post in October 2025, prompted by the publication of Hua et al. (2025). The reflections and interpretations presented here are entirely my own and do not represent the views of the article’s authors, the University of Thessaly, or any other collaborators or affiliates.
The featured image is by Oleksii (Adobe Stock) and is used under licence. The butterfly motif was chosen to evoke the “butterfly effect,” a metaphor often used in discussions of Complexity Theory to illustrate how small changes can have far-reaching consequences.
All images include descriptive alt text for accessibility. Artificial intelligence was used to assist with copy-editing and search optimisation, but all ideas, interpretations, and final editorial choices are my own.
Leave a Reply