Roles of Different Dimensions of Vocabulary Knowledge in L2 Reading Comprehension: Readers, Texts, and Tasks

Abstract

Abstract Informed by the Lexical Quality Hypothesis (Perfetti, 2007) and the Reading Systems Framework (Perfetti & Stafura, 2014), this thesis aims to further current understandings about roles of vocabulary knowledge in second language (L2) reading comprehension. It sets out to test 1) how different dimensions of vocabulary knowledge (i.e., size vs. depth), particularly different aspects of vocabulary depth knowledge, collectively and relatively predict reading comprehension for different types of texts and tasks; and 2) how the contribution of different dimensions and aspects of vocabulary knowledge to reading comprehension may be mediated by readers’ inference making skills. Two distinct but related studies were conducted that involved separate groups of participants of the same background, that is, Chinese-speaking university students in the UK. While the two studies followed the same design and answered the same set of research questions, they purposefully differed on the type of texts for reading comprehension. Study 1 (N = 123) focused on the comprehension of narrative texts, whereas Study 2 (N = 121) focused on that of informational texts. For each study, a battery of established and researcher-developed tests was administered to participants on a group basis or individually to test their vocabulary size, vocabulary depth (semantic network knowledge and polysemous word knowledge), inferencing skills (bridging and elaborative), working memory (digital span and operation span), and passage comprehension (long vs. short passages). In both studies, participants’ different aspects of vocabulary knowledge and reading comprehension were measured through a set of paper-based tasks, while their inference making skills and working memory were measured with computer-based decision tasks. Various correlation-based methods, including hierarchical regression analysis and structural equation modelling (SEM), were conducted to answer the research questions in each study. Hierarchical regression analyses revealed that in both studies, the three vocabulary knowledge measures collectively explained a substantial amount of variance in reading comprehension; and each was a significant unique predictor. The relative effects of different types of vocabulary knowledge on reading comprehension, however, varied between the two studies. In Study 1, when narrative reading comprehension was the outcome measure, vocabulary size had the greatest unique effect, followed by semantic network knowledge and polysemous knowledge. In contrast, in Study 2, where the focus was on comprehension of informational texts, semantic network knowledge was found to be the strongest predictor, and the two depth aspects demonstrated a stronger effect than did vocabulary size. Further, in both studies, differential patterns were revealed when the three vocabulary measures were regressed on different reading tasks (long vs. short passages). In both studies, subsequent SEM analysis showed that, controlling for working memory and inference making, a latent variable of vocabulary knowledge represented by the three measures directly and significantly predicted reading comprehension; its indirect effect on reading comprehension, through the mediation of inference making, was also significant. Thus, disregarding text type, the effects of vocabulary knowledge on reading comprehension were partially mediated by inference making skills. The mediating effect in Study 2 for informational texts appeared to be smaller than that in Study 1 for narrative texts. Further SEM analysis tested, in both studies, how different vocabulary knowledge, particularly the two depth aspects, may differentially contribute to reading comprehension through the mediation of bridging and elaborative inferences. Among the various mediated routes in the two studies, only the indirect effect of word associations (an aspect of vocabulary depth) on reading comprehension through the mediation of bridging inference was found to be significant in Study 1 for narrative comprehension. Taken together, the findings of the two studies in this thesis have filled several important gaps in the literature on L2 vocabulary knowledge and reading comprehension. In particular, they have expanded our understanding of how various types of vocabulary knowledge, inference making skills, and working memory interrelate and work in tandem in L2 reading development. By attending to both reader and non-reader factors, this research has generated new insights into the lexical basis of L2 reading. It highlights the complexity of lexical involvement in L2 text reading and comprehension in relation to dimensions/aspects of vocabulary knowledge, mediating functions of inference making, types of texts and comprehension tasks that may involve different reading purposes and activities. Pedagogically, the findings across the two studies suggest that, to enable successful reading comprehension, disregarding text type, readers should have a deep understanding of words or a high-quality vocabulary; and vocabulary depth deserves special instructional attention even among advanced L2 learners.