Testing Implicit Learning with Case Marking Variation

Abstract

INTRODUCTION Structural priming is known to be insensitive to the identity of closed class elements such as prepositions. For example, Bock (1989) found that primes containing a different preposition than targets (e.g., A cheerleader saved a seat for her friend) were equally successful in priming target constructions (e.g., The girl handed the paintbrush to the man) as were primes containing the same preposition (e.g., A cheerleader saved a seat to her friend) (see also Ferreira, 2003; Pickering & Branigan, 1998; Scheepers, 2003). Implicit learning models (Change et al., 2006; Reitter et al., 2011) suggest that closed class elements do not have long-lasting and cumulative effects on structural priming. This claim, however, has not been tested with case-markers, which are a key closed class element in many languages such as Korean. Here we test whether morphological variation in case marking leads to long-term structural priming in Korean. Similar to English ditransitive constructions, Korean allows two alternate ditransitive constructions such as ‘Mary-NOM John-DAT book-ACC gave’ (DAT-ACC) and ‘Mary-NOM John-ACC book-ACC gave’ (ACCACC). The two structures are commonly suggested to have the same underlying structure but result from morphological variation in case marking (e.g., Oh, 2006; Sohn, 2001; Yook, 2013, but see Jung & Miyagawa, 2004). We investigate whether morphological variation in ditransitive constructions can lead to long-term structural priming in comparison to transitive constructions (actives/passives) that have different underlying structures. EXPERIMENT We recruited 22 native Korean speakers. We examined long-term structural priming using a sentence completion methodology (e.g., Pickering & Branigan, 1998; Kaschak et al., 2006). The experiment consisted of three phases: pretest phase to measure baseline production of ditransitive and transitive constructions (12 items each), priming phrase where participants were exposed to only one dative and transitive construction (24 items each, e.g., Mary-NOM __-ACC __-ACC gave/ Mary-NOM __-by was pushed), and posttest phrase to measure the long-term structural priming effect (12 items each). Participants were asked to describe a picture depicting a ditransitive or transitive event using given sentence stems during the priming phase but without stems during the pretest and posttest. RESULTS The analysis of transitive constructions revealed a significant interaction effect between priming and test sessions (p<.05). That is, Korean speakers exposed to passive constructions during the priming phrase were significantly more likely to produce passive constructions during the posttest (16% more passives) than those who were exposed to active constructions (2% more passives). This suggests that experience with a particular transitive construction can lead to the long-term adaptation within the sentence production system in Korean. The exposure to a dative construction, however, did not result in any longterm adaptation. Whether participants produced a DAT-ACC or ACC-ACC construction during the priming phase, they produced a DAT-ACC structure 100% of the time during the posttest phase. This suggests that ditransitive constructions that involve morphological variation with the same underlying structure do not cause long-term structural priming as suggested by implicit learning accounts of structural priming. Note, however, that Shin and Christianson (2009) showed that ditransitive constructions resulted in short-term structural priming between Korean and English. This may suggest that morphological variation may cause short-term structural priming without causing long-term adaptation. More broadly, this indicates that different learning mechanisms may be involved with short-term and long-term effects of syntactic experience and that syntactic experience leading to short-term priming does not necessarily lead to long-term adaptation.