[PS-3.22] Do word-based expectations influence structural priming?

Arai, M. 1 , Chang, F. 2 & Ishikawa, M. 3

1 Seijo University
2 The University of Liverpool
3 The University of Tokyo

The difference between expectations and the input has been argued to support the change in structures in syntactic priming (error-based learning, Chang et al., 2006). Priming is sensitive to expectation manipulations that involve the same structures that are used in the primes (Kashack et al., 2006), but few studies have examined whether word-based expectations can influence priming of structures. The current study examined passive structural priming in Japanese by using fillers with the dative case marker ni. This case marker is ambiguous, because it is also used to mark the agent in the Japanese passive. Increasing dative expectations with fillers should increase prediction error with ni in passive primes. We varied whether the fillers all had dative ni (All-Dative-Filler), no dative ni (No-Dative-Filler), or half dative ni (Half-Dative-Filler). Crucially, experimental items were identical across all lists. Our results (N=43) showed that passive structural priming was strongest in All-Dative-Filler list (13.9%) and weakest in No-Dative-Filler list (4.0%). The effect in Half-Dative-Filler list did not differ from All-Dative-Filler list (13.6%). Our results provide the first evidence that passive structural priming can be modulated by expectations for a case marker in dative structures, which supports word error-based learning accounts of priming.