Developing Concepts in Second Language (L2) Science from ‘Bits and Pieces’ of Prior Knowledge: A Multiple Case Study of Spanish-Speaking 4th and 5th Graders

Earth science concepts sit at the intersection of the compounding difficulties English language learners’ (ELLs) face in content-integrated instruction. Emerging first language (L1) skills, emerging L2 language, and the difficulty of science terms interact with second language (L2) content learning. Moreover, L2 science concepts rarely develop incidentally through L2 socialization and instruction in the same naturalistic way as they do in the L1. Leveraging L1 resources may help ELLs in this process, yet there is a dearth of studies that consistently do so. This dissertation explores how, and the extent to which, leveraging ELLs’ pre-existing knowledge may support science and L2 language in a mutually supportive way (Kecskés et al., 2020). To this end, the study piloted a tutoring program—the Concept Learning Procedure (CLP)— to ‘grow’ L2 science connections and complexity by pulling together ‘bits and pieces’ of pre-existing knowledge through socially-mediated tasks. The double goal of leveraging resources and fostering development translated into a teaching implementation emphasizing explicit, linguistically-motivated, conceptually-driven, language-focused strategies. The CLP tried to promote a reduced set of eight, fourth- and fifth-grade science words and their central conceptual relationships, embedded in two cohesive, cognitively-challenging earth science units, with gradual decontextualization and support from the L1. This multiple-case study explored how sixteen upper-elementary, Spanish-speaking participants developed those eight concepts in comparison to six Controls in regular instruction in a dual program (RIDP). The initial iterative, qualitative coding of observations collected with a four-mode assessment protocol tentatively revealed a large volume of modifications and incorporations, and a relatively common emergent-target learning trajectory, both of which preliminarily underscore the potential benefits of the CLP. Analysis of students’ pre- and post-test conceptual knowledge using holistic rating found that those ELLs with emerging knowledge at pre-test could show larger quantitative gains even if more emerging. In contrast, advanced students may experience marginal quantitative gains, with greater qualitative restructuring and conceptual connections. In addition, English proficiency (EP) and length of stay (LS) impacted students’ pre-test knowledge levels, reinforcing the previously observed exponential, reverse trends. Yet, neither seemed to limit ELLs’ access to science content in CLP instruction, given its translanguaging mode. The study also showed that the CLP tentatively reconceptualized students’ pre-existing knowledge —even lack of— as a function of concept complexity, not task type. Aligned with the literature, synonymous concepts had higher reconceptualization potential than polysemic ones, likely due to differences in their nature and cognitive effort. Yet, dual-meaning concepts emerged as unique in complexity and cognitive demands. Finally, science concepts introduced in the CLP preliminarily outperformed Control counterparts, potentially due to the overall CLP method, and, in particular, to three strategies —referred to as MEM-SFA: leveraging morpho-etymology (ME), creating and extending metaphors (M), and examining core conceptual features (SFA). Case studies of three students showed the potential of these strategies, although emerging learners tended to use decontextualized definitions, and connections across concepts less frequently. Finally, the study could not corroborate the pilot’s parallel theory of quantitative and qualitative learning gains between prior knowledge, EP and LS. An alternative two-dimensional model was proposed to account for the data in the study and the apparent reconciliation problems. Future research should refine CLP strategies, increase instructional intensity, and isolate the impact of each task. Overall, this study highlights the tentative benefits of a tutoring program leveraging prior knowledge and employing a translanguaging mode. While results are not generalizable, this study might provide insights to support emerging ELLs in content-integrated science instruction.