A current view about grammar is that it has to be explicitly instructed. However, teachers of English as a foreign language (EFL) are usually not native speakers. When the class size is big and students come with various needs, it can be very taxing for a single teacher to try to provide timely and individualized feedback. Peer cooperation can be an alternative but it has to be examined whether students are competent or how they evaluate the arguments about a language they are not fully proficient with. Relevant factors should also be identified to see how they affect students' acceptance of peer feedback. This alternative can be especially challenging in Asian cultures where teachers are traditionally regarded the only authority in the classroom. This study is carried out to examine the baseline case of peer-assessment in a least favorable natural class environment. Three classes participated in the study. Grammatical questions are generated by the students and were randomly assigned to each of them to answer. Each student then rated responses to a minimum of 5 questions including the one he or she authored. They were also required to justify their ratings in writing. The FACET program were used to analyze the data. A survey accompanied with detailed rating information was sent later to each student for cross-validation. Results show obvious rater leniency/severity difference. However, students do have the ability to evaluate and learn from each other. Grade is the biggest concern in their willingness to accept peer ratings and teacher's authority is confirmed. The conclusion is that higher-order thinking skills of adult students can be utilized to facilitate learning and assessing. Fair scores can be calculated to improve grading accuracy. Other guidelines on course design and classroom assessment are discussed.

Keywords: EFL, peer-assessment, classroom assessment, validity