Fluency in a second language is considered important by both learners and teachers but is not well understood. This paper describes what is known about second language fluency and describes a number of psychological learning mechanisms that might explain how fluency develops. These include the mechanisms underlying the contrast between automatic and controlled processing, the learning mechanisms postulated within Anderson's ACT* theory of cognition, Bialystok's conception of the control dimension of language development, the notion of restructuring, recent proposals for the redefinition of automaticity as retrieval from memory (both instance and strength versions), and chunking theories. The paper concludes with some suggestions for research into the development of second language fluency itself that can fill gaps in existing knowledge and reduce our dependence on other fields for explanatory principles, while contributing simultaneously to discussion of the mechanisms responsible for skill development in general.

Tacit linguistic knowledge derives from experience of the input data or from domain-specific principles without the interaction of experience. Recognition of constraint violations is generally deemed to call upon the second kind of knowledge source. This paper proposes that this epistemological distinction should be reflected in native speaker and nonnative speaker performance. More precisely, performance should be more uniform across the species where it engages knowledge without the interaction of experience. A review of six studies investigating nonnative speaker acceptability judgments supports the predicted asymmetry. The paper concludes that the observed asymmetry supports the modular position on nonnative linguistic knowledge.

In this paper a theoretical model of dyadic native-nonnative speaker (NS-NNS) discourse is proposed in which discourse is described in terms of three features: interactional contingency, the goal orientation of participants, and dominance. The model is then used to study the discourse of 30 dyadic oral interviews of the Cambridge First Certificate in English examination. Results demonstrate the effectiveness of the model in abstracting the structure of oral interview discourse. They show that the discourse of oral proficiency interviews is characterized by greater reactiveness by NNS candidates and greater orientation toward goals by NS examiners. Variation in the structure of the discourse is also investigated in this study. This is shown to be related to the examiner, the theme of the interview, the task in which the participants are engaged, and the gender of examiner and candidate.

This paper reviews work in the philosophy of science pertinent to theory formats and relates it to recent second language acquisition (SLA) theories. Recent developments in philosophy of science and science studies have shown that the theory format advocated by philosophy of science for much of this century is unhelpful; developments in theory format in cognitive science, particularly psychology, artificial intelligence, and linguistics, have proceeded on lines independent of the older tradition in theory format. The naturalization of philosophy of science has resulted in improved understandings of what is necessary in a theory if it is to be adequately explanatory. SLA theory development has largely taken place in ignorance of such recent developments, and initial critiques of SLA theory from within the field reflected the earlier conception of theory. However, SLA research has reached the stage where a meta-understanding of theory formats, in terms of the components of a theory and the language in which a theory is couched (its formalism), is badly needed to facilitate theory development.

It goes without saying that longitudinal second language (L2) data have contributed—and will continue to contribute—much to L2 theory construction, yet one of the more difficult problems involved with this and similarly large-scale corpora is their very manageability. In years past, a common solution to this problem involved creating card catalogs for what appeared to be crucial parts of the data; more recently, researchers have resorted to computational resources, generally electronic databases. Nonetheless, these solutions remained strictly local: Access to such analytical means was generally restricted to the researchers involved. This situation is beginning to change, however, because, as this report will show, more child as well as adult L2 data are finding their way into the Child Language Data Exchange System (CHILDES) archive, thereby permitting widespread access to these data.