Second language acquisition of interface properties: Case of Japanese wa and ga

Event details

PhD Seminar

Date & time

Monday 14 October 2019


Seminar Room 1.13, Coombs Extension Building (8), Australian National University


Haruka Woods


Haruka Woods

This study investigates the second language (L2) acquisition of interface properties, with special focus on the Japanese particles wa and ga. The choice between wa, the topic marker, and ga, the nominative case marker, is under influence of semantic (e.g. Noda 1996, Masuoka 1987) and discourse (e.g. Kuno 1975) information; in other words, it is a multiple-interface property (Montrul 2011). Based on the vulnerability in L2 acquisition of interface properties reported in previous studies (e.g. Sorace 2005), it is predicted that the difficulty in the L2 acquisition of wa and ga also stems from its characteristics as an interface property. It has also been argued that interface properties can cause residual difficulties at the near-native level and might not be fully acquirable by L2 speakers, i.e. Interface Hypothesis (Sorace and Filiaci 2006).

However, in previous studies on L2 acquisition of wa and ga, the affecting factors on the choice between these particles have been mainly examined individually and not enough attention has been paid to the potential interactions between these factors. Therefore, how multiple-level factors affect Japanese L2 learners’ choice between wa and ga remains a mystery. In order to uncover this mystery, the current study investigated the effect of various factors and their interactions on Japanese L2 learners’ use of wa and ga. 134 L1 English / L2 Japanese speakers with beginner, intermediate, advanced and upper-advanced proficiency levels and 36 Japanese native speakers participated in an online cloze test. These participants were instructed to choose either wa or ga to mark the nominative NP in sentences that were under various semantic and discourse conditions. The collected data was analysed statistically using binomial logistic regression models.

The results indicate that there are significant interactions between a discourse factor on one hand and semantic factors on the other, which means that the effect of discourse varies depending on the semantic characteristics of the sentence. The results of the follow-up interviews, which were conducted with 10 participants with various proficiency levels, also support these statistical results. The results also indicate that the difference between native and non-native speakers is mainly caused by non-native speakers’ lower ability to reflect discourse information onto the choice of the particles, which results in higher susceptibility to semantic factors. Finally, the results also indicate that the gap between native speakers and upper-advanced speakers in the ability to appropriately process discourse information remains, providing supporting evidence to the Interface Hypothesis.

Updated:  7 July 2017/Responsible Officer:  Director, Culture, History & Language/Page Contact:  CHL webmaster