Instructed second language acquisition from a complex dynamic systems perspective Zoltán Dörnyei (University of Nottingham) The behaviour of a complex system is not completely random, but neither is it wholly predictable. (Larsen-Freeman & Cameron, 2008a, p. 75) What is a ‘complex dynamic system’? A system can be considered dynamic if it has: (a) two or more elements that are (b) interlinked with each other, and which (c) also change in time. These simple conditions can result in highly complex system behaviour. Simplest example: the ‘double pendulum’ The system’s behaviour is: Complex to the extent of being unpredictable. Nonlinear – no simple linear, cause-effect relationships. The system’s behavioural outcome depends on the overall constellation of the system components – how all the relevant factors work together. Discussed by four interrelated theories: complexity theory, dynamic systems theory, chaos theory and emergentism. Difficulty of researching complex dynamic systems Most common research paradigms in the social sciences tend to examine variables in relative isolation. Most established statistical procedures (e.g. correlation analysis or structural equation modelling) are based on linear relationships. Quantitative research methodology in general is problematic, because it is based on group averages, eliminating idiosyncratic details. Three potential research strategies Focus on identifying strong attractor-governed phenomena Focus on identifying typical conglomerates Focus on identifying typical dynamic outcome patterns ‘Retrodictive qualitative modelling’ In any domain: limited range of system outcome patterns (e.g. typical types of behaviours/learners/ achievement). This is the essence of self-organisation. By identifying the main emerging system prototypes we can trace back the reasons why certain components of the system ended up with one outcome option and not another. Thus, we do retro-diction rather than pre-diction. Illustration of RQM Dynamic system: language classroom System outcome options: learner prototypes Research objective: to understand what kind of a conglomeration of learner factors and classroom processes “pushed” a learner into the particular prototype he/she embodies. Through in-depth interviewing we aim to assemble a qualitative model of the main system components and development patterns. Three-step research template Step 1: Identifying salient student types in the classroom Step 2: Identifying students who are typical of the prototypes and conducting interviews with them Step 3: Identifying the most salient system components and the signature dynamic of each system Identifying salient student types in the classroom Possible sources of information: classroom observation interviews with teachers and students focus group discussions with teachers and students questionnaires (e.g. cluster analysis of the data) Motivation Cognition Emotion Behaviour Ask lots of questions Clumsy, inflexible Rigid, active need clear guidelines Quiet, obedient, rigid, responsible, fossilized in their learning strategies, Alex (M) Motivated for general subject & Less able also in English Low English proficiency Cheerful Mary (F) Hardworking Motivated Self learning, Will learn autonomously Learn only when pushed Does not do his homework seriously Slow learner Empathy, crying after receiving a test paper Mediocre in his studies Neutral in emotional, gentle, lucid Saki (F) Has intrinsic interests in learning English Serious in learning Has good memory Emotionally stable, Has acquired various learning Confident strategies Chris/Kris (M/F) Motivated High language ability Has a lot of expectations for teachers and themselves Worry a lot, not cheerful in general Negative in their way of thinking Loves comparing with others, Likes competition Helen (F) Not hardworking Not motivated Low in language ability especially when compared with students in a good class Reserved Not happy Not confident in English or any other subjects Proud of being in an elite class Has interiority complex Problematic in teachers’ eyes Her homework is messy Danny (M) Not hardworking Not motivated withdrawn Low in language ability even in a regular English class Reserved Not happy Not confident in English or any other subjects Proud of being in an elite class Has interiority complex Problematic in teachers’ eyes His homework is messy Rex (M/F) Obedient, attention seeking, would try to make some jokes in class, Funny Detail-minded, organized, independent in everything, capable of handling everything, helpful, Well-behaved Identifying students who are typical of the prototypes Interviewing prototypical students Examples of factors addressed in the interviews: attitudes towards L2 learning; L2 learning habits and styles; self-appraisal of language aptitude L2 learning goals and desires; vision of being future L2 speakers external influences such as family and friends; career considerations experience of learning the L2 at school; various situation-specific ‘pushes’ and ‘pulls’; the impact of the L2 teacher(s) Identifying salient system components and signature dynamics The interviews allow us to identify: The most salient factors affecting the students’ learning behaviour – the main components of the qualitative system model. The trajectory of each learner’s development that culminated in their specific system outcome – the system’s ‘signature dynamic’ that explains why a particular student ended up in a particular attractor state (i.e. learner type) and not in another. In sum... A retrodictive qualitative model portrays how the salient system components interact to create a unique development path (or ‘signature dynamic’) that leads the learner to a specific system outcome as opposed to other possible outcomes. Interpreting the findings of RQM In conventional research, once we arrive at an explanation of a phenomenon, we use this to make predictions in the form of testable hypotheses. BUT: In dynamic systems approaches expectations that are based on prior experiences have only limited predictive power. In dynamic systems what has happened might not happen again because of the changes in the context and in other system parameters. Interpreting the findings of RQM BUT: The essence of RQM is that while we cannot generalise any signature dynamics from one situation to another, the identified patterns are fundamental enough to be useful in understanding the dynamics of a range of other situations. This is the quintessence of qualitative research logic. Conclusion Retrodictive qualitative modelling offers a research template for deriving essential dynamic moves from idiosyncratic situations. The process aims at generating abstractions that help to describe how social systems work without reducing those systems to simplistic representations. Thus, retrodictive qualitative modelling is an attempt to essentialise rather than simplify.