L2 acquisition of Japanese null arguments

Abstract Adopting the anti-agreement hypothesis that argument ellipsis is allowed in languages lacking φ-feature agreement, the present study investigates whether second language (L2) learners of φ-feature agreement languages (L1-English L2 learners of Japanese; ELJs) and L2 learners of anti-agreement languages (L1-Mandarin Chinese L2 learners of Japanese; CLJs) can access sloppy interpretations of Japanese null arguments. The results concerning null subjects indicate that both L2 groups behaved similarly to Japanese native speakers (JNSs) by allowing the sloppy reading of null subjects. However, further analysis by L2 proficiency revealed distinct patterns. Intermediate ELJs exhibited L1-transfer of φ-feature agreement, assigning null subjects the strict reading, while the advanced ELJs were able to discard the φ-feature, allowing the sloppy reading. In contrast, CLJs accepted the sloppy reading from the intermediate level, indicating that the absence of the φ-feature functioned as positive transfer. Regarding the results with null objects, both L2 groups performed like JNSs, allowing the sloppy interpretation. Assuming that deleting a feature present in L1 is more difficult than adding a feature absent in L1 to L2, we will explain the results in terms of L1 transfer.