RECOGNITION MEMORY ERRORS PRODUCED BY IMPLICIT ACTIVATION OF WORD CANDIDATES DURING THE PROCESSING OF SPOKEN WORDS

Abstract Prior research has shown a high incidence of false positives on a recognition memory test for new test words assumed to have been implicitly activated during on-line processing of spoken sound stimuli (nonwords). Error rates were higher for test words disqualified late in nonword study items, compared to those disqualified early. In the present experiments only words were used to minimize post-lexical implicit activation of test words during study. Test words with disqualification points occurring late in study words had more false positives than those disqualified early (Experiments 1 and 2). Experiments 3 and 4 reduced available on-line processing time by using shorter study words. In these experiments recognition-memory errors were higher for both categories of experimental words (with late or early disqualification points) compared to control words, and there was some evidence of more errors to experimental test words disqualified late in their respective study words compared to those disqualified early. The implications of these results for cohort theory and pre-lexical processing are discussed.