Do Visible Semantic Primes Preactivate Lexical Representations?

Considerable research effort has been devoted to investigating semantic priming effects, particularly, the locus of those effects. Semantically related primes might activate their target's lexical representation (through automatic spreading activation at short stimulus onset asynchronies (SOAs), or through generation of words expected to follow the prime at longer SOAs). Alternately, semantically related primes might aid responding after target identification (i.e., postlexically). In contrast, masked orthographic priming effects appear to be lexical and automatic. Lexical processing of targets is facilitated by orthographically similar nonword primes and often inhibited by orthographically similar word primes (Davis & Lupker, 2006). Using the lexical-decision task (LDT), we found additivity between the facilitative effects of visible semantic primes and the facilitative effects of masked orthographically similar nonword primes at long and short SOAs, consistent with a postlexical locus of the semantic priming effects. Also consistent with this conclusion, semantic primes affected the skew of the distribution (larger effects on longer latency trials), whereas masked orthographic primes did not. In a final experiment, visible primes that were semantically related to the masked orthographic word primes did not make those primes more effective lexical inhibitors of orthographically similar targets (independent of SOA). Taken together, our findings suggest that the impact of a semantic prime is not to increase the lexical activation of related concepts. Rather, they suggest that the locus of semantic priming effects in LDTs is postlexical, in that discovering the existence of a relationship between the prime and target biases participants to make a "word" response.