Slide Slam J11 Sandbox Series
Masked translation priming with or without translation equivalents: ERP evidence from Korean-English bilinguals
Hyoung Sun Kim1, Say Young Kim1; 1NeuroCognition of Language Laboratory, Hanyang University, Seoul, Korea
Numerous behavioral and neurophysiological studies have consistently found robust masked translation priming in the forward direction (e.g., Gollan et al., 1997; Hoshino et al., 2010; Midgley et al., 2009; Schoonbaert et al., 2009), in turn demonstrating how bilingual word recognition and underlying lexical organizations are distinct from that of monolinguals. Crucially, however, most studies have used translation equivalents as prime-target pairs, which makes it difficult to pinpoint the locus of the observed priming effects. Nearly full semantic overlap between prime-target pairs may ensure priming even in the absence of mandatory translation; however, whether robust masked translation priming is possible without semantic mediation remains unclear. To examine whether L1-to-L2 translation is an automatic process in bilinguals, the present study investigates the presence of masked priming that is obtainable only if translation takes place automatically and rapidly. Thus, we manipulated the relationship between prime-target pairs as follows. Relative to an identical L1 prime (e.g., 얼굴 /elkwul/ “face”), L2 targets were its translation equivalent (ID; face) or phonologically overlapped with its translation to varying degrees: moderate (MOD; fake), minimal (MIN; fool) or unrelated (CON; lime). Highly proficient late Korean-English bilinguals performed a masked priming lexical decision task (prime duration: 48ms) on 524 critical targets (131 per condition). Provided subliminal presentation of an L1 prime is subject to fast and automatic translation robust enough to produce priming without mediation via the semantic level, varying degrees of phonological overlap between prime-target pairs may differentially affect behavioral and ERP responses to target words. In lexical decision latencies, while facilitation should be observed for ID in comparison to CON, inhibition for phonologically overlapped targets (MOD and MIN) is expected in line with the Interactive Activation framework, analogous to competition between within-language form distractors and related targets (Guo et al., 2012). Although a preliminary behavioral study with eight participants did not reveal any statistically significant difference between CON and the critical conditions, a trend that is consistent with the predictions was observed with a 24ms facilitation for ID, and a 20ms and 4ms inhibition for MOD and MIN, respectively. To assess the nature of this behavioral trend, ERP responses will be simultaneously obtained. Lateral inhibition at the lexical level will translate to greater N400 negativity for MOD compared to MIN targets, which in turn should be more negative than CON, while the negativity for ID targets should be significantly reduced. If lexical activation spreads top-down to the sublexical level, differential modulation of the N250 component reflecting sublexical-to-lexical mapping (Grainger et al., 2006) is expected across conditions. Critically, N250 negativity should be smaller for MOD compared to MIN targets, reflecting a greater ease in sublexical-to-lexical mapping of targets due to preactivation of overlapping sublexical representations constituting the prime’s translation. Whether the behavioral results translate to modulations of ERP responses reflecting sublexical and lexical processing of L2 targets will be discussed as evidence that contributes to our understanding of processes involved in bilingual word recognition.