Morishita, M. 1 , Chang, F. 2 & Harada, Y. 3
1 Kobe Gakuin University
2 University of Liverpool
3 Waseda University
232 Japanese EFL learners performed computer-based sentence production tasks. Proficiency level was manipulated by using two universities that differ in their respective average Oxford Quick Placement Test scores, i.e., higher-proficiency students with CEFR B1 (n = 83) and lower-proficiency students with CEFR A2 (n = 149). Dative primes were PD sentences (e.g., the father found a car for his son) versus DO sentences and question primes were wh-questions versus Yes/No-questions. Structure and lexical overlaps were manipulated within subjects. Prime sentences were presented phrase-by-phrase in the center of a circle on the screen. Students had to select the matching phrases on the circle for primes and then produced target sentences on their own from a set of phrases on a circle which constrained the type of structure that was produced. The students showed dative and question structural priming, but proficiency modulated this effect. For datives, higher-proficiency created a stronger priming effect. For question priming, Yes/No production was higher after Yes/No than after wh-question primes and this priming effect was larger for higher-proficiency students. This shows that the ability to use the input prime sentence for learning was different depending on previous experience as expressed by proficiency.