This dissertation aims to understand the maturational and non-maturational aspects of early bilingualism and language attrition in heritage speakers who have acquired their L1 incompletely in childhood. The study highlights the influential role of age and input dynamics in early L1 development, where the timing of reduction in L1 input and the quality of L1 input largely determine ultimate L1 attainment.

     Ninety heritage speakers of Korean in the United States who took part in this study showed great variability in their knowledge of L1, depending on the age at which L1 exposure was reduced and on the types of L1 knowledge tested. A multivariate regression analysis explored to what extent such L1 variability can be explained by maturational and non-maturational factors, in order to provide rational explanations for the divergent L1 outcomes in heritage language acquisition. It also examined how different types of L1 knowledge are selectively affected by each factor. The results showed that the maturational factor accounted for the largest variance in their ultimate L1 outcomes, but that non-maturational factors aided in further explaining the L1 variance. The multivariate model was able to predict not only the degree of incomplete L1 knowledge among heritage speakers, but also the types of L1 knowledge that they are likely to acquire incompletely. 

     The current study suggests that variability in ultimate L1 outcomes among heritage speakers can largely be understood as a function of the age at which their exposure to the L1 was reduced and the nature of L1 input they received in childhood. It is argued that lack of exposure to the L1 in childhood has a long-lasting effect on heritage speakers' L1 grammar and that, in this input-constrained context, non-maturational factors come into play, where language aptitude compensates for reduced L1 input, and language attitude allows for prolonged L1 input.