V arious theories have been proposed to account for second language production and the systematic variation characteristic of such production . Most of these theories , however , constitute only partial explanations because they often fail to handle mixed empirical findings about factors held to affect second language output and underlie systematic variation . This paper proposes a new theoretical framework for posing and addressing research questions about second language production in general and systematic variation in particular . Building on influential cognitive models of second language learning and use , the theoretical framework comprises three cognitive dimensions , viz . knowledge representation , attentional focus , and processing automaticity , which interact with each other . The theoretical framework is justified by drawing on the cognitive literature and cumulative findings from second language acquisition (SLA) research . The paper also outlines possible applications of the framework to issues of interest to SLA researchers working on a number of different fronts .