Single word context effects : facilitation or inhibition?
Page, Duane J.
The aim of this research was to examine how single word context affects word recognition in naming and lexical decision tasks. Overall, word targets are recognized faster and/or more accurately when they are preceded by related word primes compared with neutral primes and, in some circumstances, are identified more slowly and/or less accurately when they follow unrelated primes compared to neutral primes. A primary concern is that there is little empirical and theoretical justification for choosing one among a variety of neutral primes. Four current models of semantic priming were used to confine the search for an appropriate neutral prime. The results of a tone detection task in Experiment 1, and a dual naming/probe task in Experiment 2, showed that nonword primes produce the same alerting effects and consume as much attentional capacity as word primes. Experiment 3 demonstrated that nonwords do not introduce any perceptual or response biases. Thus, nonword primes appear to satisfy the criteria required ofa neutral context. Measures of facilitation and inhibition calculated from a nonword baseline at long SOAs revealed that semantic priming was facilitatory in a primed naming task in Experiment 2 and inhibitory in a lexical decision task (LDT) in Experiment 3. At short SOAs, there were facilitatory priming effects only in the naming and LDT. In a lexical matching task (LMT) in Experiment 4, priming effects were restricted to short SOAs. Evidence of inhibitory priming effects in the LDT was inconsistent with those models of priming that propose that context effects are the product of strategic processes. The use of nonword primes in Experiments 3 and 4 at short SOAs revealed that nonwords can facilitate the recognition of word targets when nonword primes are created from related word primes. The effects of nonword priming are discussed in terms of two models that purport that semantic priming is the result of automatic processes. The overall findings of these four experiments are inconsistent with current models of semantic priming. The informational constraint model is introduced as an alternative form of explanation for semantic priming. In this model, various types of lexical information are represented in a distributed fashion coupled with a mechanism of sensitization and habituation by which participants gain access to the specific modules of information. The model emphasizes the constraints that learning and attention provide in order to account for the effects that relatedness proportion, nonword ratio, stimulus sets, and different neutral primes have on word identification.