Abstract: Our reasearch includes two experiments. The first one includes 3 le-vels of processing and 3 implicit memory tests, we found no levels ofprocessing effect in primed word fragment completion. Whether there werelevels of processing effect in word stern completion or not had no relationto the possibility of word stems which could be completed to form manydefferent words or just to form one word. The second experiment, using aquestionnaire technique, asked the subjects to answer whether he(she) usesdeclarative retrieval strategies (DRS) during the completion. We found thaksubjects who used DRS completed a higher proportion of semantically en-coded words than structurally encoded words, whereas the performance ofthe subjects who had not used DRS was not affected by semantic versusstructural encoding. So there were no levels of proessing effect in wordslem completion.

Key words: NULL