A new eye-tracking study suggests that positive emojis affect reading by nudging how we process text even before we fixate on the symbol itself. The research, published in the Quarterly Journal of Experimental Psychology, found that a forthcoming cheerful face emoji is linked with readers pausing slightly longer on the preceding word, while finishing the overall sentence more quickly. The findings point to parallel processing, where emotional symbols and neighbouring words are taken in together rather than one item at a time.
Emojis—tiny pictorial marks representing faces, gestures and objects—have become a staple in digital conversation, often guiding how we interpret tone and emotion. Previous work has shown that people recognise emotions from emojis faster than from photographs, with happy expressions identified most swiftly.
Beyond overall impression, psychologists use high-speed eye-tracking to explore how emojis influence the split-second mechanics of reading. The study examines how the parafovea, the near-peripheral vision just to the right of what is being read, previews upcoming characters while the fovea focuses on the current word.
Led by Demian Stoianov, then a doctoral researcher at Macquarie University and the University of Groningen, the project tested whether facial expressions on mid-sentence emojis alter the way readers process surrounding words. Stoianov is now a postdoctoral researcher in the Forkosh Lab at the Hebrew University of Jerusalem.
In setting up the experiment, the researchers created 100 emotionally neutral sentences split into two clauses by a dash and selected 64 of the most neutral dash-separated sentences after a prescreening exercise conducted online by 34 Australian university students. Each sentence paired with one positive and one neutral Apple face emoji yielded 128 versions, divided into two lists so each reader saw 64 sentences—half with a positive emoji, half with a neutral one.
A total of 50 Macquarie University undergraduates (36 women, 14 men), aged 18 to 34, participated after calibration issues excluded one candidate. The study used a head-stabilised setup with an infrared eye tracker recording 1,000 samples per second; data from the right eye were analysed. After finishing each sentence, participants rated the sentiment of either the sentence or the emoji on a 5-point scale.
The researchers tracked both early eye movements and later measures. Early metrics included first-fixation duration, single-fixation duration and gaze duration, while late measures covered total reading time, the likelihood of backward glances and overall dwell time for each word and sentence.
When the emoji itself was fixated, positive faces showed the expected positivity advantage: shorter total reading times (about 298 milliseconds) than neutral faces (about 362 milliseconds), fewer backward glances, and shorter initial fixations on the emoji (about 208 ms vs 226 ms).
But on the word immediately preceding the emoji, the opposite pattern emerged. Across all early measures, readers paused slightly longer on the preceding word when the upcoming emoji conveyed a positive emotion rather than a neutral one; for example, single-fixation durations averaged 196 ms before a positive face compared with 192 ms before a neutral face. The late measures on that preceding word did not reach statistical significance.
Adjusting for emoji familiarity, arousal, visual complexity and ambiguity was crucial. In models that did not account for these factors, the valence effects on the preceding word disappeared, underscoring the modesty of the observed differences and prompting the authors to describe the interpretation as tentative.
Stoianov explained a potential mechanism: “Our eyes sometimes land just short of where we aim them. We wondered whether readers who appeared to be looking at the end of a word had actually been aiming for the emoji beside it.” Even after excluding glances that landed on the word’s final letter or the space before the emoji, the difference between positive and neutral faces persisted, lending some support to parallel-processing models of reading.
Looking at the sentence as a whole, readers completed trials more quickly when a positive emoji appeared, averaging 2,523 milliseconds compared with 2,585 milliseconds for neutral faces. Higher emoji arousal also correlated with faster sentence reading, a pattern the authors describe as statistically reliable and consistent, albeit modest in raw terms.
Exploratory sentiment ratings aligned with these reading shifts: participants judged positive emojis more positively (averaging 3.1 on the 5-point scale) than neutral ones (2.3), and sentences containing positive faces were rated more positively than the same sentences with neutral faces. By contrast, a separate pretest comparing dash-separated sentences (without emojis) found those plain versions scored slightly higher on positivity than the positive-emoji versions, a nuance the researchers note could reflect how emojis influence perceived sincerity.
Commenting on the implications, Stoianov emphasised the preliminary nature of the findings: “We’d like to know whether readers pick up on a frown as early as a smile.” He noted that a future study could vary emoji placement within a sentence to see how location and clause boundaries affect word processing, and could explore negative or high-arousal expressions to test whether they draw peripheral attention as readily as positive cues.
Several data-handling details are worth noting. The researchers performed quality control by merging very brief nearby glances and removing trials affected by blinks or excessively long fixations, ultimately excluding about a quarter of trials. They also conducted a non-preregistered follow-up check to address potential mislocated fixations, which reduced data loss to roughly 6 per cent and supported the notion that boundary effects around a small symbol can artificially inflate skipping rates when glances land slightly off-target.
Across all trials, valence effects were most evident at a middle distance of four to ten characters before the emoji, with the observed pause in the preceding word persisting even when final letters and spaces near the emoji were excluded from the analysis. There were no statistically significant differences on the word immediately after the emoji, suggesting mid-sentence integration and ongoing meaning construction may mute any immediate emotional priming.
The study, titled The Effect of Positive Emojis on Reading: An Eye-Movement Study, is credited to Demian Stoianov, Nenagh Kemp, Signy Wegener and Elisabeth Beyersmann.
