Task effects in masked cross-script translation and phonological priming

This study examined cross-language masked priming with Korean-English unbalanced bilinguals. The aim was to determine if the type of prime and target relationship and the type of response task would modulate across-language priming effects. The nature of the relationship between prime (Korean: L1) and target (English: L2) pairs was manipulated so that they shared semantics and phonology (cognate translations), semantics only (noncognate translations), phonology only (homophones), or neither phonology nor semantics (baseline). These prime types were tested in three different response tasks, i.e., lexical decision, naming, and semantic categorization. In the lexical decision task (Experiment 1), significant priming from cognate and noncognate translation primes was observed. Homophone primes did not produce a significant priming effect.]it the naming task (Experiment 2), both cognate and homophone primes produced significant priming effects but the noncognate translation primes did not. Experiment 3 replicated the homophone priming effect and showed that it was unlikely to be due to the shared initial phonemes of primes and targets. Finally, the semantic categorization task (Experiment 4) showed cognate and noncognate translation priming but not homophone priming. The results indicated that priming was affected both by prime-target relationship and by task type; this outcome was discussed in terms of the regulation of lexical information by a task-decision system. (C) 2003 Elsevier Inc. All rights reserved.