This book defines and reviews five central problems in the study of second language acquisition (SLA): (subconscious) transfer, staged development, cross-learner systematicity, incompleteness, and variability. The central thesis of the text is that linguistic knowledge is an autonomous component of SLA, distinct from the socially determined use of the second language (L2) and distinct from the psychological capacities for understanding, storing, and producing utterances in the L2. In chapters 2-4 attempts to explain these phenomena via early linguistic, sociolinguistic, and cognitive approaches are examined. It is argued that they all fail because they attach insufficient importance to the nature of language. In chapters 5-9 the central problems are approached from the perspective of Universal Grammar and parametric variation: it is considered that this approach provides greater insights into transfer, staged development, cross-learner systematicity and into some aspects of completeness, but that it has difficulty accounting for variability. Variability, it is then argued in chapters 10-13, is more attributable to factors related to language use and language processing. The most important of these are: the learner's need to develop hypotheses from data where Universal Grammar may not be accessible or applicable; the learner's need to transform linguistic knowledge into the productions required for language processing in real-time; and the learner's need to communicate effectively with an incomplete linguistic system. The variability observed in second language learners who began learning after the age of seven is attributed to the use of multiple knowledge sources and the different kinds of productions which may underlie second language use. The strands making up this argument are then brought together in chapter 14 in a single model and indications of further directions for research are provided. (Contains numerous references.) (Author/NAV)