This article is a literature review which seeks to answer four questions: 1) What is washback? 2) How does washback work? 3) How can we promote positive washback? 4) How can we investigate washback? Building on the 'Washback Hypothesis' proposed by Alderson and Wall, and suggestions from Hughes, the article proposes a model that identifies participants, processes and products which may influence, or be influenced by, washback. Strategies for investigating wash back are also discussed.

In 1996, we published an article that investigated the characteristics of basic language testing courses in terms of the instructors, course characteristics, and students. The present study is designed to describe the same characteristics of basic language testing courses in 2007 and to examine how such courses have changed since 1996. To those ends, we used the same questionnaire that was used in 1996 with 20 new items added. The main topics of the questionnaire included the instructors' background, the topics covered in class, and the instructors' perceptions of students' general attitudes toward language testing at the beginning and end of the course. While most items were of the Likert-scale type, some were open-ended. In the first three months of 2007, email messages were sent out to the ltest-l and ilta-mem lists asking recipients to access a Survey Monkey website that contained our questionnaire and respond to it. A second emailing was sent out several weeks later to involve those who had not responded the first time. A total of 97 participants cooperated with usable data. The results describe the instructors, course characteristics, and students, as well as differences and similarities between the 1996 and 2007 results in quantitative and qualitative terms.

This paper first reviews recent research on evaluating second language learners’writing skills. It then discusses research on a categorical instrument for evaluating compositions written by upper intermediate university ESL students. The form of the instrument used in this study included five equally weighted criteria for scoring: (1) Organization, (2) Logical Development of Ideas, (3) Grammar, (4) Mechanics, and (5) Style. An experiment was conducted under controlled conditions in which ten raters scored fifty randomly selected compositions. Regression analysis and generalizability theory were used for investigating the reliability of the instrument. In addition, information was obtained from the raters as to their reactions to the instrument. The results indicate that the scoring instrument is moderately reliable. More useful, perhaps, is the demonstration provided of the effects on reliability of changing the number of raters and/or criteria. The paper concludes with a discussion of the issues involved (both with this particular instrument and with analytic scoring systems in general) and of possible future research on the evaluation of nonnative speakers’writing.

No abstract

No abstract