This study examines the validity of score interpretations of a classroom-based oral reading fluency (ORF) assessment for 60 Korean high school EFL learners using an argument-based validation framework. Focusing on the evaluation and explanation inferences, the research integrated quantitative measures with qualitative analysis of rater verbal protocols. Quantitative analyses yielded excellent inter-rater reliability, while qualitative evidence confirmed that rater reasoning was predominantly rubric-referenced, demonstrating that raters prioritized multidimensional cues—specifically phrasing and pace—over impressionistic judgments. Regarding the explanation inference, prosody correlated more strongly with post-oral-reading comprehension than reading rate did, positioning expressive phrasing as a superior indicator of meaning construction in EFL reading. Furthermore, jagged profiles offered a nuanced way to diagnose reading difficulties such as the gap between decoding automaticity and prosodic performance. These findings demonstrate that incorporating prosody into ORF assessment provides valuable diagnostic insights, enabling a shift from product-oriented testing to processoriented formative assessment in EFL reading.

In many high-stakes testing situations, test-takers are not allowed to draw on external writing resources while writing, a practice observed more frequently in classroom settings. This may pose problems with the representativeness of test tasks and score interpretations. This study investigates the domain definition of one particular test known as the English Placement Writing Test within an argument-based validation framework. Focusing on the domain definition inference, the following rebuttal was evaluated: Certain essential contextual factors in the academic writing domain are not modeled in the test tasks. To do so, lower- and intermediate-level ESL students (n=92) who previously took the test were surveyed and interviewed regarding their uses of computerbased and face-to-face human-assisted writing tools. Results showed that students at both levels were statistically similar in their attitudes toward and uptakes of such tools while writing. The difference in availability of external writing tools between the target and test domain may point to issues with task authenticity of the test.