Abstract/Summary

Discourse competence is suggested as an integral part of L2 ability. A review of literature draws attention to the role of discourse competence in real-life communication, as it can promote effective language use in a given context, where L2 users know how to follow appropriate social conventions (Bachman & Palmer, 1996; de Jong, Steinel, Florijn, Schoonen & Hulstijn, 2012; Kuiken, Vedder, & Gilabert, 2010). Yet, little is known about the extent of the contribution discourse features of spoken performance make to assessment of L2 ability. This thesis presents the findings of two interrelated studies, both of which aimed at investigating L2 learners’ use of discourse features across different proficiency levels. Moreover, the thesis aimed at examining L2 ability through two different methods of assessment (i.e., a standardized multiple-choice test and raters’ holistic judgements) in order to find out which approach exhibits a clear relationship with use of discourse-related features of competence. An additional aim of the thesis was to explore the impact of task design on use of discourse features in L2 learners’ oral performance. More specifically, Study One aimed to investigate the use of Epistemic Markers (EMs) and Lexical Fillers (LFs) across different levels of proficiency in terms of their frequency and the effect of task complexity on use of these features. Data was collected from 63 EFL learners through a standardized test (Oxford Placement Test, OPT) that placed them in various proficiency levels and two oral tasks of different complexity (i.e., operationalised in terms of different degrees of reasoning and justification demands). The speech samples were recorded and assessed by experienced raters for overall L2 ability. Then, a frequency analysis (Fung & Carter, 2007) was conducted to examine the use of EMs and LFs. The descriptive analysis showed that participants at higher proficiency levels made more frequent use of EMs and LFs than less proficient participants. The findings from the correlation analyses revealed that use of discourse features is positively associated with L2 ability in both assessment methods. Further analyses showed that certain levels of proficiency are distinguished through use of EMs and LFs. In addition, the t-test results demonstrated no effect of task complexity on learner performance. Study Two was motivated by the findings of Study One and was aimed at expanding the design in several directions. In specific, it aimed to examine the use of Discourse Markers (DMs) across more proficiency levels and the effect of task type on the use of these discourse features. For the analysis of DMs, a new measurement of DM complexity was developed. A total of 73 EFL learners completed the OPT that determined their level of proficiency and performed three oral tasks of different type (i.e., a personal information task, a decision-making task and a narrative based on pictures) in a counterbalanced design. Learners’ speech performances were transcribed and assessed for both overall L2 ability and discourse structure by experienced raters. Then, the dataset was analysed for DM sophistication, diversity and frequency. The descriptive data analyses indicated similar use of DMs in terms of their sophistication across different proficiency levels. On the contrary, the analysis showed a more diverse use of short DMs as proficiency increased, whereas mixed results were produced for the diversity of longer DMs. The analysis also revealed that the most frequent use of DMs was made by learners at the lowest level of proficiency, while the use of DMs was very similar between the highest proficiency levels (i.e., B2 and C1). Inferential statistics confirmed that different forms of assessment led to different results. In particular, the findings from the correlation analyses showed that assessment through raters’ judgements reflects a clear relationship with use of DMs for certain levels of proficiency, whereas no such association was observed through the OPT. Moreover, the ANOVA analysis demonstrated that different types of tasks can affect frequency of DMs. The significant implications of the findings for L2 research, language testing and teaching will be discussed.