Towards enhanced second language reading comprehension assessment: Computerized versus manual scoring of written recall protocols

Abstract

Second language (L2) reading comprehension assessment has long relied upon classical quantitative, product-oriented measurement techniques (i.e., multiple-choice and cloze) in both research and classroom assessment. As Bernhardt (1991) clearly demonstrated, these traditionally employed assessment methods are unable to capture the complex processes that take place between learner and text. The present paper has as its central purpose to enhance and extend the efficiency, consistency, and validity of an alternative measure, the immediate free recall protocol. Unlike the multiple choice or cloze tests, the recall protocol is a truly integrative authentic-task measure, firmly grounded on a constructivist model of reading comprehension. Given an understanding of the model, the literature base on memorial representation, and the formulation of a "weak-rule" scoring system, the present study demonstrates a computerized recall protocol scoring system that has high correlation with traditional manual scoring methods. The results further demonstrate that the computerized procedure provides efficiency in delivery and scoring, enhances consistency, is practical for large-scale assessment, and can lead to improved diagnostic and placement testing. Using this system as part of a multiple-measures approach, valid and reliable quantitative score information is readily available and directly linked to a qualitative database ripe for additional examination to advance L2 reading comprehension research and model development.