Complex Dynamic Systems Theory in Language Education

I am happy to announce that the International Encyclopedia of Language and Linguistics (Hilary Nesi & Petar Milin, eds.) will be carrying an article by Juup Stelma and me. The article that we contributed focuses on how Complex Dynamic Systems Theory (CDST) is used in language education. In this post, I will provide a brief overview of the perspective we put forward in the article, and if you scroll all the way to the bottom of this page, you will be able to download a post-print version.

---

What is Complex Dynamic Systems Theory in language education

Names and a tentative definition

Complex Dynamic Systems Theory has been called by many names: complexity theory, chaos theory, complex dynamic(al) systems theory, complex adaptive systems theory, and more. What all these names index, however, is a shared way of thinking about things: a perspective that focuses on holistic descriptions of structure, activity and outcomes.

In the context of language education, Complex Dynamics Systems Theory (CDST) is a metatheory that can be used to provide a holistic explanation of language learning structure, activity, and outcomes. […] The emergent structures, activity, and outcomes that are subsumed under ‘language education’ are in part stochastic and in part purpose-driven; CDST, then, represents an attempt to theoretically account for this hybridity.

Kostoulas & Stelma (2024)

From the sciences to language education

The origins of the theory can be traced back to the sciences: biology and mathematics, for example. However, in the last 30 years or so, there has been an increasing drive to use insights from complexity thinking in the social sciences (e.g., Byrne & Callaghan, 2013)¹ in general, and in applied linguistics more specifically (Larsen-Freeman & Cameron, 2008a). Applied linguistics and language education have been drifting apart for some time now, and this is important to keep in mind as you read this post. While bearing in mind that CDST has been used in diverse ways across applied linguistics, our focus will be mostly on how the relevance of the theory in language teaching and learning.

In retrospect, it seems difficult to miss the parallels between CDST and language education. For example, a lot of language teaching and learning shows dynamic stability. This means that our teaching adapts to minor change (e.g., the arrival of a new student, the introduction of a new coursebook) while preserving its core characteristics. And yet, when a lot of changes push us past a tipping point, change can be catastrophic.² We know, often from frustrated experience, that the effort we invest in our teaching is not always matched by proportionate outcomes. But equally, that a single moments of inspiration can produce extraordinary success. And we also know that two lessons are rarely the same, even if we use the same lesson plan and have similar classes, because minute differences can completely transform the experience.

Some history of CDST in language education

As I mentioned above, the origins of Complex Dynamic Systems Theory lie in the sciences. It was Diane Larsen-Freeman who first argued that many aspects of language and language teaching (e.g., grammar) shared properties with complex systems. Her early work, in the 1990s, was immensely influential in popularising CDST.

The twenty or so years after that saw the publication of several robust theoretical and empirical contributions. Some of the most important ones were:

• Larsen-Freeman and Cameron’s seminal (2008a) volume, Complex Systems and Applied Linguistics;
• The Five Graces ‘Position paper’ (2009), which argued that “Language is a complex adaptive system”;
• An edited volume by Verspoor, de Bot, and Lowie (2011) that brought together thinking by the ‘Groningen group’ of CDST scholars;
• Zoltán Dörnyei’s Second Language Motivational Self System.

Although scholarship in this period was fruitful, it also seems somewhat disjointed, with many different ‘silos’ of thought, and more than a few different names to describe similar things.

An important development in recent years has been that CDST has evolved into a metatheory, or ‘theory of theories’, which provides scholars with a shared vocabulary and common methodological tools with which to conduct their research. This third period of complexity research is, in some ways, the point when CDST in Language education attained theoretical maturity.

Describing language education as a Complex Dynamic System

In our article, we suggest that there are three ways to conceptualise a Complex Dynamic System such as language education.

Aspect	Highlights	Includes
Structural	Entities in the system	Inventories of system components and subsystems, boundaries
Relational	Interconnections in the system	Affordances, constraints, activity
Intentional	‘aboutness’ of the system (what the system does/is for)	Outcomes

The structure of language education

To begin with, we could focus on the various components that make it up, such as students and teachers, coursebooks and syllabuses, policies and practices. What such a list suggests is that Complex Dynamic Systems are heterogeneous. This means that they comprise things that belong to very different category classes (e.g., people, things, ideas, and even other systems nested inside them). They also have ‘fuzzy’ boundaries, because everything in a system connects to entities outside it (‘should foreign policy be part of the language education system?’ ‘how about the salience of foreign music genres like K-pop, which provide opportunities for incidental learning of vocabulary?’).

A structurally oriented description will include: (a) the constituent entities of the system, (b) the boundaries of the system, (c) the timescales in which it operates and (d) the things that interact within this time-space. It will attempt to explain how these elements connect to produce the system of language education, and how this system interconnects with others, such as mainstream education and language policy.

The activity of language education

The second way to describe language education as a CDS would be to explain how the various structural elements interact. It is important to remember that a system is not just an set of adjacent components. Rather, it is the relations between them that matters, and these relations come into being as the components interact. I, Achilleas Kostoulas, am many things to many people, but I am a teacher (with specific, defined roles) in relation to my students and my employers. It is the set of relations that defines the system, and the system is this web of relations.

A relation-oriented description of a CDS should do two things. First, it can explain how these relations construct the system (this is different from listing the entities, above). It should also explain how activity emerges from the structure of the system and how the system structure is reshaped through activity (e.g., how a lesson pattern can become routinised). It involves identifying constraints in the system (e.g., ‘I can’t use the target language to define a word with this beginner class’) and affordances for action (e.g., ‘what can I do with AI, which was not possible a couple of years ago?’)

The intentionality of language education

The last way to talk about language education as a complex dynamic system is to focus on its intentionality: what it ‘is for’. This is perhaps the most challenging part to describe, but here goes: Systems do not exist in the physical world, in the same sense that you and I, a language textbook and a whiteboard exist. A system is a mental construct, produced by our mind’s ability to forge meaningful connections. All these components that make up a system come together in our mind, i.e. they become as a system when we try to understand how something (an outcome) happens.

Systems do not exist in the physical world.
They are products of our minds’ capacity to make meaning.

In philosophy, the connection between an internal construct and something else³ is called ‘intentionality’. Intentionality is the property that connects the feeling of love with what one loves, my thinking about a lesson with the lesson itself, an opinion about someone with that person in the real world. In CDST, it is the intentionality of a system that meaningfully connects all the components and their relations. Different intentionalities will produce descriptions of different systems. For example, I might try to connect various elements into a CDS of language learning, whereas someone else might connect a similar (but not necessarily identical) set of elements into a CDS of linguistic imperialism. The two systems will overlap and interpenetrate, some (or most) constituents will belong to both (in the same way that I can be a teacher and a father, depending on which system your perspective pigeon-holes me), and what differs is their ‘aboutness’ (intentionality). Adding the element of intentionality in the description of a complex dynamic system is what distinguishes social understandings of CDST from uses of the theory in the sciences.

This CDST view of language education, while informed by the scientific origins of the theory, is distinctly social and native to what makes language education meaningful to its participants.

Kostoulas & Stelma (2024)

Using Complex Dynamic Systems Theory in language education

Complex dynamic systems theory has informed applied linguistics and language education in three main ways.

Complexity as an object theory

One way to use CDST is as an ‘object theory’. This means using it as a way to make sense of phenomena such as grammar acquisition, learner progress etc. Experienced language teachers know that these are ‘messy’ phenomena and that it’s often hard to pinpoint what cause produces every outcome. We know, to return to a previous example, that two lessons are rarely the same even if we have similar learners, and the same lesson plan – there are always minute differences, sometimes even imperceptible ones, which produce huge differences. A large strand of research has turned to CDST and attempted to use sophisticated mathematical tools to disentangle all this complexity. This strand of research, which views complexity as a challenge one needs to overcome mathematically, is what Edgar Morin called “restricted complexity”.⁴

There has been much useful research carried out in this paradigm. For instance, people have used tools such as non-linear equations and time-series analysis to describe how things like second language acquisition or motivation develop over time, and what patterns of change or activity we might expect on various timescales. A lot of work by the Groningen group has been informed by this perspective, and there are strands of applied linguistics where it has been both popular and productive.

Complexity as a meta-theory

An alternative perspective (what Morin called “generalised complexity”) involves using CDST as a broad conceptual metaphor and to map onto it all the phenomena that interest us in language teaching and learning. To return to the example I used before, one might want to see a language class as a CDS. Just like the complex systems that the natural sciences describe, a class is nested into broader systems (e.g., a school, an educational system) and contains smaller systems nested inside it (e.g., students). It also overlaps with other systems and has ‘fuzzy’ boundaries (e.g., students and teachers move in and out of the class).

Our capacity to perceive systems in the social world corresponds to the way we make sense of the physical world.

This is not to say that activity in a class must have a one-to-one correspondence with activity in, say, a biological ecology. What it does mean is that our capacity to perceive systems in the social world corresponds to the way we make sense of the physical world. And once we accept this, it follows that this shared ability can provide us with a way to talk across disciplines, and across teacher-researcher divides.

Work in this tradition has tended to be more useful in language education than in the strands of applied linguistics that have diverged from it. It has also been relatively more theoretical and more synthetic than work in the empirical tradition described above. Some examples (not always explicitly self-described as CDST), include Al-Hoorie et al. (2023), Dörnyei et al. (2014), Tudor (2001), Sampson (2016), and Stelma and Kostoulas (2021). The aim of such work is, ultimately, to help us understand “what is meaningful and meaningless, acceptable and unacceptable, central and peripheral, as theory” (Larsen-Freeman, 2017, p. 21).

Complexity as a way of understanding language education

A final strand of complexity-informed scholarship has a methodological focus. Work in this strand aims to develop methodological procedures and instruments that lend themselves to generating complexity informed descriptions. Some salient examples include Larsen-Freeman and Cameron’s (2008b) article, as well as more recent contributions by Phil Hiver and Ali Al-Hoorie (2016, 2019).

Why is any of this important?

I have written about CDST and language education often in this blog, and I doubt that this will be the last time. The reason is that, for me, a CDST perspective is more useful to teachers than many other research perspectives are.

I mentioned, earlier in this post, that applied linguistics is broadening and drifting away from language teaching and learning. This is not an unwelcome development, but it comes at a cost. It means that the connections among the informing disciplines are lost and – at the same time – the research-practice gap keeps widening. What CDST, viewed as a metatheory, provides is a shared discourse space, where the people who teach languages and the people who study them can work together to explore meaningful connections between ideas.

Despite its name, CDST it is actually an intuitively accessible way of making sense of our language classrooms, and many of the conceptual tools it uses are useful metaphors that we can use to talk about our teaching and experiences. Having access to the same ‘language’ about our profession, is – I would argue – one important step towards redressing inequalities in knowledge production, and giving teachers more power.

---

Notes

1. Just so you know, if you buy a book from Amazon after following one of these links, I will get a small commission, at no additional cost to you. You may also want to consider supporting an independent bookstore with your purchase. ↩︎
2. I use the term ‘catastrophic’ in a technical sense here, to describe a sudden, major change, rather than a calamitous outcome. ↩︎
3. Whether this ‘something else’ is, necessarily, an external (non-mental) ‘thing’, or whether internal constructs can also qualify is a matter of debate, which need not concern us here. ↩︎
4. I couldn’t find an online link to this article, but you can find the full reference in the postprint. ↩︎

---