Priming patterns are different in masked word identification and word fragment completion

Indirect tests of remembering have revealed two different patterns of priming following generation and reading tasks: (1) read words produce more priming than generated words, which produce little or no priming relative to new words, and (2) both read and generated words show reliable and equivalent priming. In a series of six experiments using both mixed and blocked presentation of encoding tasks, we confirmed that the word fragment completion task reliably produced the first pattern of results whereas we found that the masked word identification task almost always produced the second pattern of results. Only when three different tasks were presented in a blocked design during encoding did the identification task lead to less priming for generated than for read words. We conclude that the brief presentation of a whole word in the masked word identification task makes contact with an initial interpretive encoding that includes records of conceptual as well as perceptual operations performed during encoding. (C) 1997 Academic Press.