Recent studies have shown that a subjective experience of difficulty associated with cognitive operations can paradoxically improve cognitive performance. A theoretical account is that a text written in a hard-to-read font can improve retention because disfluency encourages people to engage in deeper processing of the information. Our study critically examined this view by an experiment. Forty-seven participants were asked to learn 40 words written in an easy-to-read font or a hard-to-read font. As a result, disfluency raised the percentage of correct answers; however, this increase was only for participants who have a lower working memory capacity (WMC) or it was for all participants regardless of their WMC depending on the measuring method of WMC. We proposed an alternative view of the effect of disfluency to account for the results assuming multiple routes for information inputted. According to this view, people with low WMC tend to process peripheral information (i.e., surface features of letters, a part of which is relevant to disfluency) resulting in cuing them to retrieve stimuli. People with high WMC, in contrast, are able to focus exclusively on central information (e.g.,meaning of letters) and they may not benefit from the effect of disfluency. This view is consistent with the results of previous studies that suggested the effect of disfluency is not necessarily robust.