Declarative and procedural memory systems and linguistic difficulty in instructed second language acquisition of English

Abstract

In contrast to first language (L1) acquisition and child second language (L2) learning which almost always lead to native-level proficiency in the language, L2 acquisition is highly variable in the rate of learning and the ultimate level of achievement. Very few learners achieve nativelike proficiency in L2 comprehension or production. To explain the modulating factors in L2 acquisition and the nature of L2 knowledge systems, current approaches that view language as inextricably intertwined with cognition have called for the investigation of the role of the two most important long-term and domaingeneral memory systems, the declarative and procedural memory systems, and of how these two memory systems are influenced by external factors such as exposure condition and the complexity of learning stimuli (e.g., DeKeyser, 2016; Hamrick, Lum, & Ullman, 2018; Housen & Simoens, 2016; M. Paradis, 2009; A. S. Reber, Walkenfeld, & Hernstadt, 1991). A great deal of L2 acquisition research has examined the role of learning conditions (e.g., Norris & Ortega, 2000; Spada and Tomita, 2010) and language aptitude (e.g., Carpenter, 2008; Dornyei, 2005; 2006; Grey, Williams, & Rebuschat, 2015; 2017; 2019; Li, 2018; Yalçın & Spada, 2016), largely ignoring the three-way interaction of the memory systems, learning conditions and linguistic complexity, which offers one promising way to characterise L2 acquisition. Adopting an ex post facto design, this study investigated the role of the declarative and procedural memory systems in the acquisition of the knowledge of 14 grammatical structures of L2 English in a classroom environment in Malawi where grammar rules are explicitly taught. The goal was to present the overview of the educational effects on L2 English in Malawi. Quantitative data were collected from 103 L2 English learners at primary school, secondary school and university levels. In the absence of clear ways of distinguishing between simple and complex grammatical structures, a criterion based on Pienemann’s processability theory was proposed to evaluate the linguistic complexity of grammatical structures targeted in the present study. Declarative memory measures were the Continuous Visual Memory Test, the Llama-B test, the DAT Verbal Reasoning Test and the Three-Term Contingency Learning Task. Procedural memory was assessed using the Serial Reaction Times and Llama-D tests. The Timed Grammaticality Judgment Test and the Elicited Imitation Test assessed learners’ procedural language knowledge whereas the Untimed Grammaticality Judgment Test assessed learners’ declarative language knowledge. All the tests were administered using a computer except for the Untimed Grammaticality Judgment Test and the DAT Verbal Reasoning Test which were paper-and-pencil-delivered. The evidence of test validity was explored using correlational, exploratory and confirmatory factor analyses. To simultaneously investigate various interrelated dependence relationships among variables, data were analysed using a series of structural equation modelling analyses and one-way between-group analyses of variance. The principal findings of the study were that (1) procedural memory played no role in the learning of both simple and complex grammatical structures; (2) linguistic complexity and exposure type appeared to modulate only the declarative learning processes; and (3) no group differences were found in the procedural memory system. These results suggest that the declarative/procedural learning and declarative/procedural memory distinction may indeed be a one-to-one relationship, without the conscious, declarative learning processes interfacing with the unconscious, procedural processes. They also suggest that procedural processes display tighter distributions in a population when compared with declarative systems. Methodological implications and practical suggestions for pedagogy and future research are identified as well.