Introduction: Applying Complex Dynamic Systems Principles to Empirical Research on L2 Motivation

Zoltán Dörnyei, Peter D. MacIntyre and Alastair Henry

When nonlinear system dynamics was introduced into second language acquisition (SLA) research – under various rubrics such as chaos theory (Larsen-Freeman, 1997), emergentism (Ellis & Larsen-Freeman, 2006), dynamic systems theory (de Bot et al., 2007) and complexity theory (Larsen-Freeman & Cameron, 2008) – the new approach, which may be seen as the 'dynamic turn' in SLA, resonated with many scholars because nonlinear system dynamics appeared to nicely describe several puzzling language learning phenomena. To offer but one illustration, the so-called 'butterfly effect' explained why language teaching input sometimes had considerable impact on the learners' progress, whereas at other times it led only to minimal, if any, uptake. The dynamic principles introduced also made intuitive sense research-wise. We have long known that the manifold issues and factors affecting SLA are interrelated, and the new paradigm represented a holistic approach that took into account the combined and interactive operation of a number of different elements/conditions relevant to specific situations, rather than following the more traditional practice of examining the relationship between well-defined variables in relative isolation.

Thus, proposals for a dynamic paradigm shift in the research community during the first decade of the new millennium were generally well received. However, by the end of the 2010s it had become noticeable that while there was a growing body of literature on complex dynamic systems within SLA contexts, very little of this work was empirical in nature. In other words, scholars spent much more time talking about research in a dynamic systems vein than actually doing it. Furthermore, even when dynamic principles were referred to in data-based studies, this was often to explain away difficult-to-interpret results, stating in effect that such results occurred because of the unpredictable or 'emergentist' nature of the system. At the same time, in informal conversations at conferences, it was not at all uncommon to hear scholars privately express the sense of being at a loss as to how exactly to go about researching dynamic systems.

The Challenge of the New Paradigm

This growing uncertainty was to some extent understandable since – as Dörnyei (2009) summarised – at least three aspects of such an approach inevitably pushed researchers into unchartered territories (for a detailed overview, see Verspoor et al., 2011).

• Modelling nonlinear change (especially quantitatively); this has been succinctly summed up by de Bot and Larsen-Freeman (2011: 18) as follows: 'If the process is nonlinear, how is it possible to make any predictions that are likely to hold up?'

• Observing the operation of the whole system and the interaction of the parts, rather than focusing on specific units (e.g. variables) within it. In de Bot and Larsen-Freeman's (2011: 18) words: 'if everything is interconnected, how is it possible to study anything apart from everything else?'

• Finding alternatives to conventional quantitative research methodologies that, by and large, relied on statistical procedures to examine linear rather than dynamic relationships.

The combination of these three issues seriously questioned the feasibility of investigating cause–effect relations, the traditional basis of generalisable theories in the spirit of the 'scientific method' (see Dörnyei, 2007). As Byrne and Callaghan (2014: 173) put it:

we cannot decompose the system into its elements and use control over discrete elements whilst varying just one of them, either directly or through the use of treatment and control groups, in order to establish causality in terms of the properties of those elements.

We should note here that the challenge that applied linguists and language psychologists have been facing is not merely having to master new research skills in order to find their bearings in a novel paradigm, but is related also to a much broader issue: the difficulty of transferring the nonlinear systems approach from the natural sciences – where dynamic systems theory has been flourishing in several areas (such as thermodynamics) – to the social sciences. In the natural sciences, where the main units of analysis are molecules or objects, it is possible to reconstruct...