ABSTRACT
Priming effects due to similarity of form rather than meaning (i.e. form-priming) are readily demonstrated in a lexical decision task with masked primes that are different by one letter from the target (substitution primes). This effect can be interpreted as an activation process, where activation is induced in the detectors of all words that resemble the stimulus. An alternative account is that form-priming is really a case of identity-priming: X primes Y only if X is taken to be an instance of Y. The strongest version of this account predicts that priming should be restricted to the best match for the prime. Two experimental tests of this hypothesis are reported. Both predictions are falsified. First, it is shown that substitution priming is independent of the lexical status of the prime, i.e., attitude and antitude prime the target APTITUDE equally well. Priming should not occur in the first instance since APTITUDE could never be the best match for the prime attitude. Second, it is shown that the best match hypothesis fails to explain why substitution priming is not obtained for words with a large number of neighbours, since it predicts that priming should be obtained when the number of prime neighbours is kept to a minimum, which was not the case. In a third experiment, it is shown that form-priming does have repetition-like properties, since it persists across an intervening word, which would not be expected in an activation model.
