In two experiments this study tested the "Graded Salience Hypothesis" and the "Defaultness Hypothesis." It weighs the effects of linguistic versus pictorial contexts in terms of activation (or suppression) of default, salient meanings when context invites nondefault, less-salient alternatives. Using a naming task, Experiments 1 and 2 looked into the processing speed of ambiguous probe words, following a linguistic or pictorial prime, supportive of the less-salient, nondefault meaning. Prime presentation was either limited to 350 ms (Experiment 1) or self-paced (Experiment 2). Findings support the "Graded Salience Hypothesis" and the "Defaultness Hypothesis," showing that, as predicted, default meanings were activated initially even when context, whether linguistic or pictorial, was strongly biased toward the alternative, nondefault meaning. These novel findings suggest that linguistic and pictorial contexts affect linguistic disambiguation to the same extent.