Language Learning ISSN 0023-8333
Variability and Variation in Second
Language Acquisition Orders: A Dynamic
Wander Lowiea,b and Marjolijn Verspoora,b aUniversity of Groningen and bUniversity of the Free State, South Africa
The traditional morpheme order studies in second language acquisition have tried to demonstrate the existence of a fixed order of acquisition of English morphemes, regardless of the second language learner’s background. Such orders have been taken as evidence of the preprogrammed nature of language acquisition. This article argues for a process-based, dynamic explanation of development, in which each developmental step is based on the dynamic interaction of all processes involved. Due to the complexity of these interactions, the developmental process cannot be predetermined and fixed.
Although stages of development like the acquisition order of morphemes are commonly observed as a grand sweep effect at the group level, these stages may be meaningless at the level of the individual language learner. This paradox shows we can only make the observations that our method allows us. If we are interested in grand sweep effects that may be generalizable to large populations of learners, we will have to carry out group studies with representative samples that can be analyzed using Gaussian statistics based on the normal distribution. But if we are interested in how an individual learner progresses over time as a result of changing variables in a changing context, we will have to conduct longitudinal studies and use nonlinear methods of analysis.
Keywords variability; morpheme orders; dynamic systems; second language acquisition
Second language acquisition (SLA) research in the late 1970s and early 1980s studied the sequence of acquisition of morphosyntactic characteristics of
English by second language (L2) learners with different first language (L1)
Correspondence concerning this article should be addressed to: Wander Lowie, University of
Groningen, Department of English, P.O. Box 716, Groningen, 9700 AS, Netherlands. E-mail: email@example.com
Language Learning 65:1, March 2015, pp. 63–88 63
C© 2015 Language Learning Research Club, University of Michigan
Lowie and Verspoor Variability in Acquisition Orders: DST backgrounds. The purpose of these studies was to see if the L2 would develop in a fixed order, irrespective of the learner’s L1. If this were the case, it could be argued that L2 learning is a universal and therefore predetermined process, which would point to both the innateness and the systematicity of syntax and grammar. The studies showed that approximately the same order of acquisition of English morphosyntax was found for learners with very different mother tongues. This caused a great deal of discussion about the interpretation of these findings, and even though our thinking about language acquisition has changed over the past 40 years, the morpheme order studies are still under debate. Given the symposia and colloquia at major recent conferences in the field, it could even be argued that there is an upsurge of interest in these studies. The discussions have centered around two questions. The first question is whether a fixed order of acquisition has convincingly and consistently been demonstrated. Not surprisingly, there is still disagreement about this point. Second, if we can speak of a fixed order, the question is what causes the order. Is a fixed order indeed evidence of the universal nature of acquisition that is the same for all language learners, all types of language learning, and irrespective of the learner’s L1?
Or are there other explanations for the observed similarities?
In this article we address both of these questions by discussing the possible existence of acquisition orders and their implications about the underlying learning process from a Dynamic Systems Theory (DST) point of view. We explain how DST can provide new insights into these questions by studying the type and amount of variability in language use. The objective of DST applications to SLA is to acknowledge the fact that language development should be seen as an individually owned process rather than a product and that this process is shaped by the nonlinear relationships of changing components over time. We argue that the process-based view and its consequences for the type of analyses we perform have important implications for the way we interpret language learning data. In this article we first explore the relevance of variability for the explanation of development, alongside a dynamic reinterpretation of the morpheme order studies. Subsequently, we discuss the theoretical and practical implications of these observations in view of both theory formation and empirical studies in the field of SLA. Finally, we try to point out the interconnections with the other contributions in this Special Issue.
Toward a Process-Based Explanation of Development
The prevailing approach to SLA up to the beginning of this century has been to employ the product of language use as the measure of evaluation, usually
Language Learning 65:1, March 2015, pp. 63–88 64
Lowie and Verspoor Variability in Acquisition Orders: DST in combination with one or more independent variables, such as motivation, aptitude, anxiety, et cetera. The question then is in what way and to what extent the dependent measure (the product of language learning) can be accounted for by one or more of the independent variables. Using analyses of variance, regression analyses and, more recently, mixed-model analyses of means comparisons of group results, conclusions are drawn about the nature of the learning products and the factors that have played a role in shaping these products.When representative samples are assumed to have been drawn, the results are then generalized toward bigger populations of learners, which are in the end supposed to inform the process of language learning in general. This approach is well attested and generally applied in most fields of research and as long as basic assumptions regarding normality of the distribution, homogeneity of variance, representativeness of the sample, and careful operationalization are certified, the validity of the outcomes is assumed to be warranted (Lowie &