Judgments of learning (JOL) are often used to assess memory monitoring at encoding. Participants study a cue-target word pair (e.g., mouse-cheese) and are asked to rate the probability of correctly recalling the target in the presence of the cue at test (e.g., mouse -?). Prior research has shown that JOL accuracy is sensitive to perceptual cues. These cues can produce metamemory illusions in which JOLs overestimate memory, such as the "font-size effect" (Rhodes & Castel, 2008), which occurs when participants inflate JOLs for pairs presented in large versus small fonts without a concomitant increase to recall. The present study further tests the font-size effect and examines whether other perceptual manipulations can affect the correspondence between JOLs and recall. Experiments 1A and 1B were designed to replicate the font-size effect and test whether the effect extended to highlighted pairs that were related or unrelated in the same study list. Experiment 2A and 2B examined font size and highlighting effects on JOLs using only unrelated pairs. Finally, Experiment 3 tested whether Sans Forgetica--a perceptually disfluent font designed to improve memory--would result in inflated JOLs and/or recall. Large fonts similarly increased both JOLs and recall relative to small fonts, highlights had no effect on JOLs or recall, and Sans Forgetica font yielded a memory cost (though no effect on JOLs). Collectively, perceptually fluent and disfluent study pairs do not appear to inflate JOLs relative to subsequent recall.