The Impact of Disfluent Fonts on Natural Paragraph Reading: An Eye-Tracking Study

Fonts play a crucial role in shaping the reading experience by influencing both readability and cognitive load. Fluent fonts are generally easier to read due to their clear and familiar design, while disfluent fonts disrupt the reading flow with irregular strokes, decorative features, and inconsistent spacing. Prior research has suggested that disfluent fonts may introduce “desirable difficulties,” slowing reading but potentially enhancing comprehension and retention through increased cognitive effort. However, the extent to which disfluent fonts actually promote learning and recall, particularly in more natural reading contexts, remains contested. Most existing studies have focused on isolated words, short sentences, and page or multi-page length texts, leaving the effects of font disfluency during paragraph-level reading largely unexplored. This study investigates the impact of font disfluency on reading comprehension and immediate recall in an ecologically valid, paragraph-level reading task. Ten English-speaking participants (aged 18-25) silently read college-level English passages presented in three fonts with varying styles and fluency levels: Brush Script MT (disfluent, handwritten), TH SarabunPSK (fluent, sans-serif), and Angsana New (fluent, serif). Eye movements were recorded using a high-resolution eye-tracking system, and participants answered comprehension questions immediately after each passage. Results indicated that Brush Script MT, the most disfluent font, significantly increased reading time and was rated as the most difficult to read. However, no statistically significant effects of font style were found on comprehension accuracy or the time spent answering comprehension questions. Eye-tracking analysis at the word level revealed that Brush Script MT elicited the longest fixation durations and displayed a more leftward preferred viewing location (PVL) compared to Angsana New and TH SarabunPSK, suggesting that disfluent fonts disrupt visual processing at the word level. Despite this visual disruption, paragraph-level reading comprehension remained unaffected by font disfluency. Together, the findings suggest that while disfluent fonts can increase cognitive load and perceived difficulty, they may not enhance or impair comprehension in natural reading contexts. These results advance the understanding of how font disfluency affects reading processes in more realistic settings, as well as offer novel insights into the impact of widely-used but under-researched Thai fonts on reading behaviour.