CONFERENCE ON COMPLEX SYSTEMS 2019
October 3, 2019
Fourth CCS Satellite Symposium on Complex Systems and Education: Applications in Language Pedagogy, Assessment, and Complexity Literacy
He Sun (Nanyang Technological University, Singapore; email@example.com)
Vahid Aryadoust (Nanyang Technological University, Singapore; firstname.lastname@example.org)
Hiroki Sayama (Binghamton University, USA; email@example.com)
Matthijs Koopmans (Mercy College, USA; firstname.lastname@example.org)
Dimitrios Stamovlasis (Aristotle University of Thessaloniki, Greece)
Since at least 20 years ago, complex dynamical systems (CDS) theory has motivated a shift in attention from outcomes to process in education and language learning. Despite discernable progress, one of the challenges of the CDS approach that remains to be addressed is the effective integration of quantitative research methods into the rapidly expanding theoretical discussions (Lowie, 2017). Many educators have presented CDS as a new approach by relying on metaphors and analogies to describe the relevant underlying concepts such as iteration, emergence, and attractors, while paying comparatively little attention to operationalization and measurement. There is a consensus over this lack in both language learning and education fields. For example, Koopmans and Stamovlasis (2016) have argued, “while this work is valuable in its own right, it does not have the level of conceptual and methodological specificity
that is required to capture the dynamical processes hypothesized in the dynamical literature…, nor does it speak to the specific gaps in our knowledge that result from the relative absence of dynamical perspectives in empirical research in education” (p.2). Similarly, van Geert (2007) refers to the extant CDS approach in language learning as the “informal” model, calling for formulating “formal” methodologies which would embrace mathematical and quantitative modeling. There is an important related question: How rigorously can CDS concepts, methodologies and techniques be learned or taught to those outside of the CDS research domain, such that the perspective gains wider acceptance? The proposed satellite symposium, therefore, aims to address this gap and discuss the formal models’ affordances, challenges, and promises for the fields.
This satellite is a full-day session with 8 invited and contributed talks, plus overarching introductory and concluding remarks.
Lynch School of Education & Human Development Boston College , USA
Patrick McQuillan, an Associate Professor in the Lynch School of Education at Boston College, has a Ph. D. in cultural anthropology from Brown University. He teaches courses on curriculum theory and history education as well as seminars on qualitative research. His current research focuses on school reform, with an emphasis on the role of the prinicpal in transforming urban schools as understood through the lens of complexity theory. He has organized complexity-related sessions for the American Educational Research Association in 2014 and 2016 and recently presented at the Complex Systems Institute at MIT. His other professional writing includes Reform and Resistance in Schools and Classrooms: An Ethnographic View of the Coalition of Essential Schools (Yale University Press, 1996; co-authored with Donna Muncey) and Educational Opportunity in an Urban American High School: A Cultural Analysis (SUNY Press, 1998).
Quantifying the Metaphor:
Looking at Educational Change Through the Lens of the Complex Adaptive System
Efforts at educational reform are too often ineffective. Commonly, this occurs because change is seen as a linear matter of technical precision. But systems change is more complex. To create a more holistic conception of educational change, this paper draws upon complexity theory, in particular, the analytic construct known as a complex adaptive system (CAS; Stacey, 1996; Lewin, 1999), an heuristic attentive to the iterative, recursive, interdependent, and non-linear nature of educational transformation. To address this matter, I examine work of the Lynch Leadership Academy (LLA) at Boston College, a principal leadership program.
Specifically, I focus on three aspects of the CAS framework: (1) Whether principals generated a sense of disequilibrium, perturbing their school systems toward change; (2) Whether in response to disequilibrium principals distributed control throughout schools to enhance networking capabilities; (3) And the centrality of maintaining balance among system elements.
The following questions constitute elements of a rubric designed to illuminate each CAS dimension:
Was a context created which disrupted normal routines?
Did new opportunities emerge?
Could the system revert from reverting to the status quo?
Did people have autonomy to act in response to disequilibrium they experienced?
Did new networks emerge?
Were multiple persons and structures integrated into the change endeavor?
The “edge of chaos”
Were key points of tension balanced—not too much, not too little?
Challenge & support: Were people challenged to innovate but supported in this work?
Authority & autonomy: Was decision-making broadly balanced?
Moreover, quantifying the understandings generated by the educational transformation rubric offers the following potential benefits:
Creating a basis for assessing various drivers and impediments to change;
Quantifiably evaluating a reform over time or before or after specific changes are implemented;
Generating dialogue based on a common language and related understandings.
Lewin, R. (1999). Complexity: Life at the edge of chaos. Chicago: University of Chicago Press.
Stacey, R. D. (1996). Complexity and creativity in organizations. San Francisco: Berrett-Koehler Publishers.
National Institute of Education, Nanyang Technological University, Singapore
Dr. Sun He is a research scientist at National Institute of Education, Nanyang Technological University, Singapore. Her major interests are 1) effective mother tongue language and English development with eBook/paper reading, 2) parental tips provided with Apps for scaffolding children’s early bilingual development, 3) individual differences in early bilingualism from the perspective of cognitive and environmental factors, and 4) harmonious bilingual education and children’s social-emotional wellbeing. Her work uses corpus (e.g., video observations), experiment (e.g., eye-tracking, skin conductance, test), and questionnaire to explore the impact of home and school environment on the variability and stages of early bilingual teaching and acquisition. Dr. Sun He has published her work in distinguished journals, such as Bilingualism: Language and Cognition, the International Journal of Bilingual Education and Bilingualism, Applied Psycholinguistics, and Frontiers in Psychology.
To Explore the Dynamics of Teachers’ Questions and Comments and Bilingual Preschooler’s Chinese Learning With Mixed Effects Model and Dynamic Motifs Detection
Preschool teacher’s comments and questions during Shared Book Reading (SBR) have been found to facilitate children’s early language and literacy development. To date, much more is known about the impact of these language strategies on children’s learning outcome than on their learning process. The current study focused on both aspects with two goals (a) to investigate the relation between teachers’ high/medium/low cognitive loaded questions and comments, and children’s receptive vocabulary outcome, and (b) to explore the association between different types of questions and comments and children’s immediate responses interpreted using complex dynamic system theory. We studied Mandarin teachers (n = 31) and English-Mandarin bilingual children (n = 505; ages 4-5) in Singapore. Teachers’ and children’s conversations during the reading were transcribed, and teachers’ comments and questions were coded as high, medium or low cognitive demand levels. Children’s receptive vocabulary was tested in the first and second academic years at kindergarten. Mixed effects models and motif detection analysis were used to address the two research questions respectively. The results demonstrated that teachers’ low level commenting and questioning strategies were significantly correlated with children’s vocabulary growth, and only certain types of medium and low level questions and comments elicited verbal response from children. Our findings suggest that low cognitive demand questions and comments might match preschoolers’ Mandarin development at the age of four to five in the Singaporean bilingual context. Questions and comments, such as asking children to describe an object in the picture book, could significantly result in more responses from children and maintain them in the conversation. The underlying dynamical process and implications for teachers are discussed.
National Institute of Education, Nanyang Technological University, Singapore
A Novel Scientometric Investigation of the Evolution of Language Assessment Specialty: identifying Within- and Between-Domain Research Trends
The present study is the first scientometric investigation of language assessment to identify the most influential trends that have shaped its scholarly focus since 1984. The database used for the analysis comprised papers and cited publications in Language Assessment Quarterly (LAQ, 2009 – 2018) and Language Testing (LT, 1984 – 2018), the two flagship research outlets in language assessment. Multiple visualization and text-mining methods were employed to detect the latent patterns among publications. First, the two journals were situated on a dual-map overlay, which revealed significant connections between LAQ and LT with “psychology, education, and social sciences” as well as “economics and economic and political sciences” fields on the Web of Science, revealing previously unrecognized interdisciplinary research trends. Next, a co-citation cluster analysis revealed 17 major clusters, including cognitive diagnostic assessment, assessment literacy, language proficiency, interactional competence, and issues surrounding performance assessments such as validation of rating scales, rater effects, and fairness. Co-citation cluster analysis further detected a number of publications, countries, institutions, and cited journals (including LAQ and LT) that have significantly shaped research directions in the field. Unlike the previous studies where narrative reviews have been mainly used to compartmentalize the history of language into multiple clear-cut phases, it was found that the detected trends did not linearly progress; rather, multiple research trends with different foci emerged, overlapped, and then phased out. “Punctuated equilibrium”, a term from evolutionary biology, is used as a metaphor to interpret trends in the emergence, continuation, and end of clusters. Lastly, the implications of this study for the specialty and future research directions are discussed.
Mercy College, USA
Dr. Matthijs Koopmans joined the faculty at Mercy College in 2011, and he serves as Assessment Coordinator to the programs at the College that prepare educators. Previously, he worked for several research and consulting organizations, and conducted research on an independent contract basis to evaluate the effectiveness of educational interventions, such as math, science and technology initiatives, second language programs, and full day vs. half day kindergarten, focusing mostly on urban schools in the United States. He has taught at several colleges and universities in the Northeastern region of the U. S., and served as president of the Society for Chaos Theory in Psychology and Life Sciences from 2005 -- 2007. He co-edited books on the application of complexity theory to education (published by Springer) and psychology (Cambridge University Press). He has published papers about time series with fractal patterns in numerous peer reviewed journals, and wrote a book on the subject for The Sage University Series on Quantitative Applications in the Social Sciences, which is due out in the spring of 2020. He earned his Doctorate in 1988 at the Harvard Graduate School of Education.
Problem Formation and Problem Resolution in American Schools
In spite of the potential of formal education to contribute to the emancipation from poverty, we are faced in the United States with a persistent gap between rich and poor, and our public schools are widely criticized for not being able to prepare sufficient numbers of students for college and for the demands of the modern workplace. One set of responses to this state of affairs has been the structural reform of the educational system, which has included such initiatives as longer school days, block scheduling, magnet schools, teacher pay incentives and charter schools. Over the past three decades, reforms of this nature have been undertaken in many places and a case study literature has proliferated to evaluate the effectiveness of these efforts. Another set of reforms is of a financial nature, often mandated by the courts in decisions that note a failure by states and municipalities to enable schools to fulfill their constitutional obligation of providing equal educational opportunities, due to underlying inequalities in the money flow.
This presentation approaches these issues from a complexity theory standpoint, which makes a distinction between first order change (‘the more it changes, the more it stays the same…’, or ‘when the solution becomes a problem.’) and second order change, which describes the transformation of a system beyond its initial settings (Watzlawick et al., 1974), and includes qualitative and catastrophic transformations, self-organized criticality and the edge of chaos. This latter kind of change may be a prerequisite for true reform. Many of the current attempts can be seen as first order change (problem formation). We identify the growing points in the system that can generate second order change (and hopefully, problem resolution).
Watzlawick, P., Weakland, J., & Fish, R. (1974). Change: Principles of problem formation and problem resolution. New York: Norton.
National Institute of Education, Nanyang Technological University, Singapore
Student population dynamic model to simulate real world university scenario and examine the cause and effects of policies
We intend to develop a dynamic model that simulates a real world university environment, and captures the dynamical forces at play between different characters. We define the characters as students, teachers, professors and heads of departments, each character equipped with a different set of attributes and functions. The characters interact with each other and execute functions, and the effect gets propagated in the entire population. For example, a teacher choosing to execute absolute grading of exams will have a different effect on student population than if the teacher chooses to execute relative grading of exams. The values of parameters at the end of the two different sets of actions will be different for students. Similarly, executions of policies by heads of departments will have a ripple down effect in the entire population. In the intended model, we will allow for differences among students with regards to their academic standing. Similarly, teachers and professors will be stratified according to their qualifications and experience. The model aims to represent a simulated environment where the cause and effects of policies and decisions can be observed experimentally, and their consequences can be seen in detail.
Ismo T. Koponen
Department of Physics, University of Helsinki, FINLAND
Learning science lexicons: Complex Adaptive Systems Perspective
Learning lexicon of terms is important part of learning science, because the terms provide the basis of the language of science. In communication between the students and teachers, the lexicons have also a central role because communication requires certain similarity of the lexicons; lexicons need to be shared.
A long-standing question in learning and acquisition of knowledge is how structure of communication and shared/unshared knowledge are related and affect the emergence of consensus knowledge. Several studies have attempted to find out beneficial knowledge sharing patterns which could signal better acquisition of knowledge. From viewpoint of acquiring new knowledge, sharing knowledge with dissimilar peers may be advantageous. However, if group members are prone to seek consensus, the interaction with dissimilar members may not always lead to increase or improvement of knowledge.
In this talk I will discuss an agent-based model, where lexicon acquisition is seen from a perspective of complex adaptive systems. The terms to be learned is modelled as network of terms, with a structure that resembles the empirically revealed structure of scientific lexicon. Each agent posses a fraction of that network, and through interaction and comparison of their lexicons, attempt to reach consensus. It is shown that in presence of consensus bias, small changes in bias to share knowledge with similar peers may have
strong effects how similarity of shared knowledge between peers evolves. This, on the other hand, affect the communication patterns of individual agents and the group dynamics. Because of such reciprocal interactions, new and stable shared lexical networks are formed; new cliques of shared lexicons emerge as kinds of attractor states. Such finding resembles interestingly the group learning dynamics, where different groups may have different emphasis on key terms they found important.
Martin F. Gardiner
Brown University, Providence, R.I., U.S.A.
Evidence for a bifurcation in the development of verbal language skill: insight from music skill training
Data from longitudinal studies in Cambridge, Massachusetts of students progressing in verbal language skill in Elementary school between 2003 and 2009 shows evidence of a bifurcation in verbal language skill development, as measured by standardized test which begins at 3rd grade. At Peabody school, percentages of students who reached grade level or higher on basic reading skills was the same as in the rest of the district until end of 2nd grade. But by end of 3rd grade, percent scoring at grade level or higher on verbal language at Peabody jumped upward, while percent at grade level or higher in the rest of the district actually declined, superiority of performance of Peabody students compared to rest of the district now becoming highly significant statistically. This difference of Peabody from the rest of the district remained significant in 4th grade as well. The superiority from rest of district at verbal language beginning at 3rd grade was especially notable in students who qualified for free or reduced lunch and thus came from impoverished families.
The Peabody students received the same verbal language curriculum as the rest of the district, and were actually located in a district with a high percentage of impoverished families. Analysis controlled for differences of Peabody from the rest of the District in percentage of students from poor families or with learning disabilities. During these years, however, Peabody was the only school in the District that integrated a specific form of music skill training into its curriculum beginning by 1st grade. That it was students that received this training that jumped away from the others in verbal skill can illuminate the existence of the bifurcation discussed here, and can also help to account for the jump in verbal language performance at Peabody compared to the rest of the district.
Martin F. Gardiner has a Ph.D. in Biophysics (UCLA) specializing in brain research, an M.S. in Electrical Engineering (Stanford) and an M.S. in Physics (Columbia), and extensive training in music. He has held research positions at Brown University and been on the teaching faculty at New England Conservatory of Music for almost two decades. Beginning with a paper with colleagues in Nature in 1996 he uses music skill as a window to examine more general components of human cognitive and emotional skill learning
Aristotle University of Thessaloniki
Bifurcations in educational research: Methods and Interpretations
Complex Dynamical Systems (CDS) perspective as a meta-theoretical framework, has already embrace educational research and the implementation of the proper nonlinear tools has presented a considerable number of investigations, and thus it is legitimate to talk about a paradigm shift in this field. An intriguing issue, within the new perspective, is the emergence of a nonlinear behavior with two stable stages or two equilibrium positions. Whenever this behavior becomes empirically ascertained by means of catastrophe theory modelling, the traditional epistemology is questioned. Merely CDS framework can explain such unexpected behaviors, which might lead to abrupt changes, called discontinuities. To this end, understanding the crucial role of variables known as bifurcation factors is the central goal of many research endeavors. Fields and areas such as decision-making, motivational and achievement goal theories, intentionality, cognitive and psychometric variables in learning and performance, fatigue and overload phenomena, illicit behaviors, are some examples where the application of nonlinear methodology has opened new avenues for theory development. A lucid review of the relevant literature, is part of this presentation, showing how CDS and nonlinear dynamics have influenced contemporary scientific inquiry, along with methodological and statistical issues. It is interesting and a surprising exception that, while CDS educational research, in general, has remained behind comparing to other scientific disciplines, the area of applying stochastic catastrophe theory, recently has presented a lot of work. This is because, given the available methodological assets for analyzing empirical data, there are fundamental philosophical reasons for reconsidering the nature of the educational processes. Education is by definition “change” and changes might not be predictable as we traditionally think.
*Short Workshop: Time Series and the Analysis of Fractal Patterns Using R: Applications to Education and Other Social Science Data
In spite of the longitudinal nature of many educational phenomena such as teaching and learning, reducing the achievement gap and school reform, educational research lacks a tradition of conducting time series analysis to capture these processes. A similar situation can be found in many other social sciences. Moreover, to the extent that time series analyses are conducted in those fields, the approach is typically not extended to include the estimation of fractal or irregular patterns and other complex phenomena. Fractal patterns in time series data are indicative of adaptive behavior at the edge of chaos and self-organized criticality in systems, and therefore should be of interest to complex systems researchers.
The purpose of this workshop is to demonstrate how fractal patterns can be estimated using time series analysis. Two distinct but related approaches will be discussed: fractional differencing, also known as autoregressive fractional integrated moving average (ARFIMA) models, and power spectral density analysis. Both approaches will be illustrated using R, the open source statistical programming environment. This workshop will take participants step by step through the estimation process and the model diagnostics, and we will go over the R programming statements and the output they produce. We will also discuss what indicators of fractality are telling us about the systems of interest, and touch upon the limitations of a fractal interpretation of statistical data.
The workshop consists of six parts:
A brief conceptual overview of fractional differencing and power spectral density analysis;
Introduction to the arfima, fractal and fracdiff packages in R used to conduct the analyses;
Discussion of a few typical data scenarios that fractal researchers need to be able to recognize;
A step by step run through the diagnostic and analytical process, the R programming statements plus interpretation of the output;
Discussion of the parameter estimates (specifically the differencing parameter d and the Hurst exponent H) in terms of complexity theory;
Generation of power spectra in R and their use to argue for fractal patterns.
The following illustrative data sets are used for this demonstration:
Daily attendance rates in one New York City high school from 2010 to 2014 (N = 735);
Monthly unemployment figures in the United States from 1948 through 2017 (N = 837);
Daily recordings of the number of births to teens in the state of Texas from January 1, 1964 through March 10, 1966 (N = 800).
Intended Audience: Those with an interest in the use of complexity theory in educational research and other social sciences, and those interested in the extension of conventional inferential statistics to the detection and analysis of fractal patterns and other complex phenomena.
Facilitator: Matthijs Koopmans – Mercy College, USA
Matthijs Koopmans has published extensively about time series with fractal patterns in journals such as Nonlinear Dynamics, Psychology and Life Sciences, Complicity: An International Journal of Complexity in Education, Journal of Experimental Education and Fluctuation and Noise Letters. He also wrote a book on the subject for The Sage University Series on Quantitative Applications in the Social Sciences, which is due out in the spring of 2020.