Research on differential effects of cognitive, social and instructional factors often confounds hierarchical relationships within 'container variables'. The Proximity Hypothesis posits that effects of distal factors are mediated and explained by proximal factors interacting directly with learners. This is investigated longitudinally in regular and bilingual primary schools using multilevel modeling. (Preprint on Researchgate)
Studies investigating factors influencing instructed early FLL show that learners' individual differences in language background/s, cognition, social factors, as well as L2 input quality and quantity present strong predictors for L2A. These factors are intrinsically intertwined; research on their combined and differential effects is scarce. Major difficulties lie in the hierarchical level of variables, the assumed nature of their effects and their statistical representation: An often neglected problem is that socioeconomic status,migration background or program represent 'container variables' comprising factors on various conceptual levels (e.g., type of linguistic input or interactions in the family). A second aspect is the differentiation between distal (program) and proximal (linguistic input) factors regarding the assumed point of contact with the learner. The Proximity Hypothesis posits that distal variables exert only indirect effects mediated by proximal factors: Effects are best explained using factors that interact directly with the child. In this vein, this paper investigates the impact of different variables on FLA by unpacking container variables, disentangling conceptual and causal hierarchies, accounting for mediating and moderating relationships and analyzing their differential contributions. To do so, data from a longitudinal project (N=262) in German regular and bilingual primary schools (L2 English) are investigated using multilevel and structural equation modeling to account for the nested data structure. Cognitive and linguistic variables were assessed using standardized tests. Social variables (parental education, cultural capital, family interactions) were elicited using a parental questionnaire. Instructional variables include L2 contact, intensity, and input as operationalized with the Teacher Input Observation Scheme. Preliminary results suggest that L2 proficiency is mainly predicted by proximal input and cognitive variables. Linguistic and social backgrounds show differential effects which are stronger in regular than in bilingual programs. The effect of container variables is mediated by proximal variables. Results are discussed with reference to the Proximity Hypothesis. (A preprint to the paper on this talk can be found on Researchgate.)