Dynamic Systems Theory

Dynamic Systems Theory (DST), also known as Complex Dynamic Systems Theory (CDST) or Complexity Theory, views language as a complex adaptive system — nonlinear, emergent, self-organising, and sensitive to initial conditions. Diane Larsen-Freeman (1997) was the first to apply DST to SLA, arguing that language development shares the properties of complex systems studied in mathematics, physics, and biology. Kees de Bot, Wander Lowie, and Marjolijn Verspoor further developed the framework, with their 2007 paper "A Dynamic Systems Theory approach to second language acquisition" and Verspoor, de Bot & Lowie's (2011) edited volume A Dynamic Approach to Second Language Development establishing CDST as a major paradigm in SLA research.

Core Properties

Language as a dynamic system exhibits:

Property	In language development
Complexity	Multiple interacting subsystems (phonology, lexis, syntax, pragmatics, motivation, social context) influence each other simultaneously
Nonlinearity	Small changes in one variable can produce large, disproportionate effects. Progress is not smooth or incremental.
Self-organisation	Patterns emerge from the interaction of components without a central controller or innate blueprint
Sensitivity to initial conditions	Two learners with slightly different starting states (L1, motivation, aptitude) may follow radically different developmental paths
Variability	Variability is not noise — it is a signal of the system reorganising. Increased variability often precedes developmental leaps.
Interconnectedness	Subsystems compete for resources. Growth in one area (e.g., vocabulary) may temporarily cause regression in another (e.g., fluency).

Key Departures from Traditional SLA

DST challenges several assumptions in mainstream SLA:

• No fixed stages — unlike Interlanguage theory or Processability Theory, DST does not posit discrete stages. Development is continuous, variable, and path-dependent.
• No single causal variable — there is no single factor (input frequency, UG, motivation) that determines acquisition. The outcome emerges from the interaction of all variables.
• Individual variation is the norm — group averages obscure the actual developmental trajectories of individuals. DST emphasises single-case studies and dense longitudinal data.
• Attractor states — the system tends toward certain stable configurations (attractors). Fossilization can be reconceptualised as a strong attractor state that the system has settled into, rather than a permanent endpoint.

Research Methods

DST has prompted methodological innovation in SLA:

• Longitudinal case studies with dense data collection (multiple measures at frequent intervals)
• Min-max graphs and moving correlation analysis to detect variability patterns
• Retrodictive qualitative modelling — explaining individual developmental trajectories after the fact, rather than predicting group outcomes
• Rejection of cross-sectional designs and group-mean comparisons as inadequate for capturing dynamic processes

Relationship to Other Theories

DST shares common ground with Emergentism and Usage-Based Theory in rejecting innate grammar and emphasising that structure emerges from use. However, DST is not a theory of what is learned (as usage-based accounts specify) but rather a theory of how development unfolds — the dynamics of change itself. It is a metatheory that can potentially accommodate findings from multiple SLA traditions.

Criticisms

• Descriptive, not explanatory — critics argue that DST describes the properties of development (nonlinear, variable, complex) without explaining why specific structures are acquired or what drives development
• Difficulty with prediction — by definition, complex systems are sensitive to initial conditions and difficult to predict. This limits the theory's practical utility for teachers and syllabus designers.
• Methodological demands — the research methods DST requires (dense longitudinal data, single-case designs) are time-consuming and difficult to generalise from

Teaching Implications

• Variability in learner performance is normal and expected, not a sign of failure
• Rigid linear syllabuses may not align with how language actually develops
• Different learners may need different input and support at the same nominal "level"
• Regression in one area during growth in another is a natural feature of the system reorganising