BUILDING A CORPUS-BASED INSTRUCTIONAL VOCABULARY MODEL: INTERDISCIPLINARY ACADEMIC WORDS IN UNIVERSITY READING SUPPORT COURSES

Abstract

While research demonstrates the importance of vocabulary knowledge for reading comprehension (Anderson & Freebody, 1981; Guo, Roehrig, & Williams, 2011), many studies suggest wide reading exposure for vocabulary acquisition (Nagy, Herman, & Anderson, 1985; Stahl & Nagy, 2006). However, there is compelling evidence for the benefits of direct vocabulary instruction (Elleman, Lindo, Morphy, & Compton, 2009; Laufer & Rozovski-Roitblat, 2015). Due to the volume of vocabulary acquisition that is needed, time considerations and subjectivity can make instructional vocabulary selection seem futile. To overcome these issues, the use of corpora has enhanced the possibilities of identifying high-frequency lexical items in a systematic and more objective manner. Drawing on insights from multiple disciplines, this dissertation adopts a corpus-based approach for the selection and teaching of high-frequency interdisciplinary academic vocabulary in a model that accommodates the needs of students in university reading support courses. The adopted corpus-based model identifies instructional vocabulary with the support of the Word and Phrase Tool (Davies, 2013), a Web interface that allows users to scan texts in order to gain instructional information about their lexical profiles. Considering academic vocabulary to be an interrelated feature of academic language, the dissertation shows how a well-designed empirical selection process can advance the capabilities of instructors in ways that address some valid concerns of earlier word list approaches. The instructional portion of the model is organized around a researcher-developed method called item discussions that uses a graphic organizer for students to record vocabulary explorations that are attentive to known aspects of vocabulary development such as definition, morphology, and synonymy and, additionally, to generate sentences using each term in a context similar to that in which it was encountered. To further investigate the effectiveness of this corpus-based instructional model, a preliminary small-scale experimental study was conducted with students enrolled in two reading support classes at a large university (N = 18). Moderate to strong effect sizes were found for similar-to-instructional context encounters (g = .73) and the full instructional model (g = .95), suggesting that the model offers a promising approach. The discussion includes pedagogical implications and further directions for research and implementation.