A greater tendency to complete single-completion word stems (e.g. "BEY") to form previously read whole words (e.g. "BEYOND") was found when test stems were presented in the same letter case as their previously encoded words, compared with the different letter case, but only when stems were presented directly to the right hemisphere (i.e. in the left visual field) and not when they were presented directly to the left hemisphere (i.e. in the right visual field). This finding with single-completion stems was robust (i.e. observed for both lowercase and uppercase stems) when the initial encoding task was perceptually demanding, but it was test-case dependent (i.e. observed for uppercase but not lowercase stems) when the initial encoding task was not perceptually demanding. Results and theory help to explain why letter-case-specific priming in right-hemisphere test presentations is typically test-case dependent when priming is measured using perceptual identification at test, but is consistently robust when priming is measured using multiple-completion word stems (e.g. "BEA") at test. Demands from both the stimuli and tasks affect the relative contributions of abstract and specific subsystems to the processing of visual forms.