Effects of semantic and associative relatedness on automatic priming

Models of automatic priming of word identification can be divided into those based on associative relations (e.g., spreading activation) and others based on semantic similarity (e.g., distributed models). In three experiments, associative relatedness was manipulated by presenting asymmetrically associated word pairs in both their forward and backward directions. Priming was comparable in both directions for semantically related pairs. Furthermore, priming was not obtained in either direction when pairs were associated but not semantically similar. The absence of inhibition, practice, and nonword ratio effects suggested that priming was not the result of nonsemantic, controlled processes. These results indicate that semantic similarity, and not associative relatedness, is both necessary and sufficient to produce automatic priming. (C) 1998 Academic Press.