2018 年 29 巻 p. 305-320
This article reports an action research on the effects of process-writing instructions with datadriven learning and the order of meanings approaches. A specific theme of this action research was how to verify the degree of success in teaching because the score gains from pretests to posttests are usually found in practical research. Low-proficient university students were taught English writing for a semester in regular classes to develop accuracy in recognizing phrase chunks, using verbs, and ordering words and phrases in sentence composition. In accordance with the steps of action research, the author examined the expected effects of each instruction approach from meta-analysis results. By developing the meta-analytical criteria that determine how effectively the approaches would work according to the author’s educational settings (e.g., proficiency, instruction length, teaching approaches, and targeted skills), the results of a first cycle indicated that the data-driven learning did not develop participants’ writing skills as much as previous practices. Based on the reflections from the first cycle, the order of meanings approach was introduced in a second cycle, resulting in successful improvement of the author’s practice. This study discussed the (in)effectiveness of the building of a meta-analytical criterion for the progression of our profession.