English as an Additional Language (EAL) learners have been shown to struggle academically in English only learning environments (e.g. Coxhead & Boutorwick, 2016; Murphy & Unthiah, 2015). Various assessments tools have therefore been adapted to explore the academic performance of these learners. These tools include assessments for: vocabulary levels (e.g., VLT; McLean, Kramer & Beglar, 2015; Nation, 1998), word decoding (e.g. the Single Word Reading Test (SWRT); Foster, 2007), reading comprehension (e.g. the York Assessment of Reading Comprehension (YARC); Snowling et al., 2009), and global academic proficiency (e.g. the C-Test, Dörnyei & Katona, 1992). Given that none of these tools were specifically designed for the EAL learner, it remains difficult to determine if the results they provide equate to valid assessments of EAL learners' academic skills.
Accordingly, the current study examines the validity of four different assessment tools, chosen based on testing adopted in recent EAL studies, by exploring their constructive validity and predictive power. We assessed 18 EAL and 14 L1 (English) learners (n=32) at an international school in Japan. We measured: vocabulary knowledge (VLT); word decoding skills (SWRT); reading comprehension (YARC); and, overall academic proficiency (C-test). Results from these measures were compared with results obtained from an assessment of the learners' academic proficiency done using the EAL Assessment Framework for Schools, a newly created assessment framework designed to assess EAL learners' academic and linguist proficiencies (Bell Foundation, 2016).
A Rasch analysis and Cronbach's alpha were used to examine the construct validity of each assessment and person and item reliability. Both analyses showed the four assessments to be highly reliable. A multiple regression analysis was then used to determine the extent to which the assessments predicted variances in academic proficiency. This analysis showed that for all learners both vocabulary (r=0.001) and reading fluency (r=0.004) were statistically significant predictors of academic proficiency. However, when the data were analyzed by group (L1 compared to EAL learners) only vocabulary remained a statistically significant predictor of academic proficiency (r=0.018), and only for the EAL learners. These results highlight the importance of vocabulary for EAL learners and the potential of using vocabulary assessment in both the classroom and for research purposes. The current paper forms the basis of a wider study exploring the development of EAL specific word lists and vocabulary assessment tools. We discuss our findings in this light and in terms of the importance of contextually specific assessment tools and test validity.