THE EFFECTS OF QUESTION GENERATION TRAINING ON READING COMPREHENSION AND QUESTION PRODUCTION (STUDENT, ANSWER, RELATIONSHIPS).

Abstract

Investigates the effects of question generation training on reading comprehension with a focus on the number and types of questions generated. Additionally, this study investigates effects of reading achievement and text difficulty on students' comprehension and question generation abilities. 295 sixth grade students, stratified according to reading achievement, were randomly assigned to one of three treatment groups: question generation training, question generation without training, or control assignment. Within treatments, students were assigned to text versions according to reading scores. Stanines 1-3 were assigned to read the original or adapted version. Stanines 4-9 read the original version. Two phases were involved: (1) the question generation training group participated in one forty minute training session with the researcher that involved a discussion about questioning; a model passage and demonstration of the procedure; group participation in question generation; and individual practice in question generation. Other groups continued regular classroom work during this period. (2) All three groups participated in one testing session involving three fifteen minute activities. Appropriate directions were given and students read the original or adapted version; generated questions or completed a control assignment; and, then completed a thirty-item multiple-choice comprehension measure. An ANOVA indicated a significant treatment effect on reading comprehension in favor of the control assignment over the question generation without training group. Significant differences also resulted in terms of reading achievement: high achievement students scored higher than low achievement students. However, low achievement students who read the adapted version scored significantly higher than those who read the original story. Four ANOVAs indicated significant treatment effects on question production in favor of the question generation training group. Those trained generated significantly more textually explicit and implicit questions and significantly fewer passage nonspecific questions than the untrained group. Significance also resulted in terms of achievement level and text difficulty. Finally, significant differences were noted between low ability students reading original versus adapted versions. Those students who read the original generated significantly more textually explicit and passage nonspecific questions, yet significantly fewer textually implicit questions than those who read the adapted version.