Web-based Diagnostic Reading Assessment

This study attempts to diagnose college students' grammatical problems in reading tests (mainly vocabulary in Bachman and Palmer's model). Based on the literature of process-oriented reading researches (Grabe, 1991, 1999), reading assessment researches (Alderson, 2000), and the model of language ability proposed by Bachman and Palmer (1996), a framework of diagnostic reading assessment is established. Then, a self-designed diagnostic reading assessment system (Diaras) is used to collect the data. The character of the system is interactive, which is achieved through two teacher clients and one student client. The teacher clients handle the generating, administering, and grading of diagnostic tests while the student client enables the taking of diagnostic tests.Before doing the online tests, a questionnaire based on the framework was conducted to find out the factors that the students think most affect their reading performance. In addition to the data obtained from the questionnaire, the data mainly consist of four types: (1) difficult language points, (2) confidence about an answer, and (3) scores in the reading tests, and (4) reading time and question-answering time.The data were collected through the following techniques: (1) four passages of reading comprehension from the CET-4 of June 1999, (2) written protocols of difficult language points, and (3) confidence about an answer.The questionnaire reveals that the factors that language learners think most affect their performance are vocabulary knowledge, affective" schemata and prior knowledge, etc. The experiment shows that vocabulary knowledge is the most important factor that affects their reading performance. The study also indicates that one important factor is confidence. The more confidence that the students have, the more answers that they choose are right. Time is another important factor. On average, highly successful learners have more adequate time than others in answering the questions.