PREFIX UNITS WITHIN THE MENTAL LEXICON

International audienceThree masked priming experiments associated with the lexical decision task were carried out in order to examine the cognitive processing of prefixed words in French. To this end we systematically compared the effects of the prior presentation of prefixed words (e.g., prénom), prefixed nonwords (e.g., dénom) or orthographic nonwords (e.g., danom) on the recognition latencies of their root (e.g., nom) or of another related prefixed word (e.g., surnom). When compared to unrelated primes, both prefixed words and nonwords facilitated target recognition (Experiments 1 & 2) and this was not an effect arising from the frequency ratio between roots and prefixed derivations. However, when morphological priming effects were measured against orthographic nonword controls, that where combinations of existing roots with nonexisting prefixes, morphological effects did not differ significantly from orthographic effects (Experiment 3). This finding suggests that morphological priming effects do not totally depend on the decomposition of the prime in two distinct morphemes, as suggested by Rastle & Davis (2008) but tend to be sensitive to formal factors (more precisely overlapping roots), even though they cannot be reduced to simple orthographic priming. Taken together, the present data moderate the full decomposition approach of morphological processing. A new model is proposed, integrating both sublexical units corresponding to "morphomes" (Aronoff, 1994) and supralexical units assimilated to "base-lexemes"