The effects of task features on lexis and grammar in L2 oral performance
This thesis is based upon an experimental study designed to establish the effects of task features on lexis and grammar in L2 (English as a second or foreign language) learners' oral performance by investigating the complex interactions between a wide range of interrelating variables. Specifically, task conditions (the presence or absence of planning time with descriptive vs. narrative tasks) are analysed with respect to lexical measures (word range, lexis-to-grammar ratio, lexical density, lexical choice, syllabic range, lexical strategy use and evidence for lexical stretching) and to grammatical measures (complexity and accuracy), as well as to measures of fluency. The thesis aims to look into the relationship between lexis and grammar within the context of spoken discourse where task features like planning time and task type interact. It is particularly focussed on lexis (which has so far been underplayed in L2 acquisition research). The concept of lexical stretching is proposed, as a parallel to the already familiar notion of grammaticallinterlanguage stretching, and evidence for lexical stretching is provided by drawing parallels between the quantitative, statistical analysis of oral performances and qualitative analysis of protocols held with learners on the completion of tasks. The research study addresses such questions as the effect of the provision of planning time on lexical vs. grammatical stretching: is there a trade-off between them, and is this further influenced by task type (operationalised here as descriptive vs. narrativebased tasks)? It also examines the ways in which contextual and interpersonal factors influence interlanguage use, particularly the use of lexis and grammar. Based on the analyses (both quantitative and qualitative), it is concluded that not only are there interdependencies between lexis and grammar, and most strikingly within lexis, but also there are contextual and interpersonal constraints on L2 learners' output, suggesting an interaction of task features and contextual factors.