The Identification-Production Hypothesis predicts that the effect of divided attention (DA) at encoding should be larger when priming tasks involve divergent search processes through many different competitors, because they are supposed to place heavier attentional demands on frontal lobe functions (Gabrieli, Vaidya, Stone et al., 1999). This hypothesis was tested in two experiments using the Word Fragment Completion (WFC) task with unique solutions (which relies on convergent lexical search towards single appropriate representations: the identification version of the WFC) or multiple solutions (which implies a competition between equally plausible responses: the production version of the WFC). In Experiment 1 the two versions of the completion task were found to be equally unaffected by the imposition of a short-term memory load at encoding. In Experiment 2 the reduction of target words' study time (from 2.5 to 1 s), and the use of a blocked (rather than mixed) design, significantly diminished the level of priming in the DA condition: However, contrary to the predictions, the degree of impairment was not greater for multiple-solution fragments. Results are in line with recent failures to support the identification-production hypothesis in normal adults (Geraci, 2006; Prull, 2004)