The concept of reading fluency combines accuracy and speed of reading with the ability to comprehend the content being read. Some definitions of reading fluency focus more on letter recognition and reading speed (Meyer and Felton, 1999), while others include content comprehension (Pikulski and Chard, 2005).

Many factors affect reading fluency. Reading fluency is affected by the shape or legibility of the typeface (Ali et al., 2013), type size (Mueller et al., 2014; Su et al., 2018), and the overall typographic design of the text (Koch, 2012). When reading on a screen, reading fluency is also affected by the screen resolution, as with a higher resolution, letters and their features can be displayed better (Bessemans, 2016a; Bigelow, 2019).

Several studies showed that the shape of letters and the text can influence the comprehension of the read content (Choi et al., 2018) and the actual memorisation of the read content (Lewis and Walker, 1989; Gasser et al., 2005). Poorer fluency results in poorer information processing and, consequently, poorer comprehension and memorisation of the text (Novemsky et al., 2007; Oppenheimer and Frank, 2008; Meyer et al., 2015; Bjork and Yue, 2016; Pieger et al., 2016; Rummer et al., 2016; Sanchez and Nayor, 2018; Dressler, 2019; Wu et al., 2019).

Studies examining how using a perceptually difficult-to-process typeface with an increased desirable difficulty designed specifically to reduce legibility, such as Sans Forgetica, found either no processing or memory benefit of such typefaces or even yielded a memory cost (Geller et al., 2020; Taylor et al., 2020; Wetzler et al., 2021; Cushing and Bodner, 2022; Maxwell et al., 2022). However, there is also a whole series of studies which showed that poorer fluency of the text or desired difficulty in the fluency of the text resulted in better processing of the text and consequently in better memorisation of the read content (Diemand-Yauman et al., 2011; Macdonald and Lavic, 2011; Bjork et al., 2013; Halin, 2016; Pieger et al., 2016).

Numerous studies demonstrated that reading fluency affects the learning process, more specifically short-term and long-term memory (Weissgerber and Reinhard, 2017), as well as metacognition (Yue et al., 2013; Ilic and Akbulut, 2019). Based on the shape of letters and the text, readers can predict how long it will take them to read the text and remember the content of the text (Beier and Larson, 2013; Price et al., 2016). Higher reading fluency should promote a positive attitude towards the text, consequently the feeling of better memorability of the text, and it should allow for better memorisation and comprehension of the text (Song and Schwarz, 2008; Mueller et al., 2013; Labro and Pocheptsova, 2016; Pieger et al., 2016; Mead and Hardesty, 2018). In contrast, poorer fluency should promote poorer attitudes toward the text and readers should assume that they will spend more time reading and memorising the text.

Emotions play a specific role in reading. The typographic design or the shape of the typeface has a great impact on the reader’s mood, more specifically on their emotional response or feeling of pleasantness that the reader experiences when reading certain letterforms (Larson and Picard, 2005; Larson et al., 2006; Koch, 2012; Petit et al., 2015). The shape of letters and the text can suggest the nature and content of the text to the reader (Lewis and Walker, 1989; Ehsen and Lupton, 1998; Celhay et al., 2015; Bigelow, 2019; Davis, 2019; Raden and Qeis, 2019).

Several studies have shown that the perception of shapes, tastes and sounds evokes various feelings in humans, including the feeling of pleasantness (Childers and Jass, 2002; Brumberger, 2003; Mackiewicz, 2005; Shaikh et al., 2006; Bar and Neta, 2007; Tsonos and Kouroupetroglou, 2011; Amare and Manning, 2012; Crisinel et al., 2012; Ngo et al., 2013; Velasco et al., 2014, 2015a,b, 2016, 2018a; Salgado-Montejo et al., 2015; Jordan, 2017; Davis, 2019; Haenschen and Tamul, 2019). Round and rounded shapes, as well as symmetric shapes evoke more pleasant feelings than angular or pointed and asymmetric shapes (Bar and Neta, 2007; Ngo et al., 2013; Turoman et al., 2018; Velasco et al., 2018b).

We have not found a study that would examine how these features of human perception can be effectively used in the typographic design of educational materials, but based on the previous studies we can assume that round typefaces would evoke more pleasant feelings than angular ones.

The core of typographic design are typefaces, which can be grouped based on the shape of the main strokes, and the transitions between the strokes and the stroke ends (terminals, serifs). One group of typefaces contains round/rounded typefaces and the other group contains angular/pointed typefaces. A typical example of typefaces that could be classified in the round/rounded group based on their design features are typefaces that belong to the group of Venetian, Garalde and Transitional typefaces (McLean, 1997; Možina, 2003). Typefaces that could be classified in the angular/pointed group based on their design characteristics are typefaces that belong to the Didone, Slab-Serif and Sans Serif group (McLean, 1997; Možina, 2003).

Previous studies found that rounded, organic shapes of strokes and softer transitions between the strokes and stroke ends are perceived as more pleasing whereas the letters with more geometric stroke shapes, sharp transitions between strokes and finial stroke are found to be less pleasant (Spence and Deroy, 2012; Hyndman, 2016). The feeling of pleasure we experience when reading different typefaces influences motivation and concentration (Mano, 1997; Koch, 2012), memorisation and comprehension of a text (Mano, 1997). However, research addressing how the reader’s emotional response to the shape of letters affects reading fluency is scarce.

It has been shown that perception in reading also depends on the age of the reader. Children and adult readers differ in the level of development of cognitive and physiological abilities until the age of four, after which the ability to recognise letters should be the same in children and adults (Woods et al., 2005). However, studies reported that children from 4 to 11 years old react to different stimuli, e.g., colour, shape, taste, smell, differently from adult students (Gollely and Guichard, 2011). It has been discovered that reactions to the same stimuli are different also in younger adults (under 35 years old) and older adults (over 60 years old) (Piqueras-Fiszman et al., 2011). In younger readers (aged 7 to 9 years), typefaces with serifs and a difference in stroke width were found to lead to more fluent reading, whereas sans serif typefaces that have no or minor difference in stroke width result in fewer reading errors (Wilkins et al., 2009). It is also claimed that a larger type size allows faster decoding of information and better memory, but only in children (age 9 to 12 years old), not in adult students (Abukaber and Lu, 2012). Children (from 7 to 12 years old) read letters that are heterogeneous in shape more easily (Wilkins et al., 2009; Abukaber and Lu, 2012); especially the heterogeneity in the shape of letters seems to greatly aid visually impaired children (age 5 to 10 years old) in reading (Bessemans, 2016b). In the study conducted by Katzir et al. (2013), the increased desirable difficulty of the typeface affected reading fluency, demonstrating positive effects in older children (11 years old), but negative effects in younger children (8 years old).

Our study had two aims. The first aim was to determine how the shape of the typeface (round/rounded vs. angular/pointed) affects reading fluency, subjective reading experience, and reading performance. The second aim of our study was to investigate whether the effect of typeface shape is the same for younger, less experienced readers and for adult, more experienced readers.

We used an eye-movement tracking device as it provides objective measures of reading fluency (Piqueras-Fiszman et al., 2013; Franken et al., 2015). We monitored the reading speed and regressive saccades as measures of reading fluency. We also used this device to observe changes in pupil size, which should be indicative of the reader’s emotional response (Hess and Polt, 1960; Margareth et al., 2008; Wang et al., 2018). Objective measures of emotional response to different shapes of typefaces were complemented with subjective ratings of feelings of pleasantness. Text comprehension and memorisation of what was read were also observed as indicators of reading performance.

To track eye movements, we used a Tobii X120 eye-tracking device and Tobii Studio 3.4.8 software (Tobii AB, Sweden). The device tracks eye movements by tracking the reflection of the image from the cornea. The corneal reflection is generated by infrared emitters on the front of the device that create IR light patterns that are then reflected off the cornea. The device contains a camera that is sensitive to IR light and monitors each movement and fixation of the eye based on the reflection of IR light from the cornea (Tobii Pro, 2017).

Before the measurements, each participant had 5 min to adapt to the lighting conditions in the test room and to perform a nine-point screen-based calibration of the device. We used an LCD screen with a resolution of 2,400 × 1900 pixels (pixel size 0.27 mm) and a refresh rate of 60 Hz.

Prior to the main study in which we investigated how the shape of different typefaces affects the pleasantness ratings and the reading speed, memorisation and understanding of a text, we conducted two preliminary studies. The purpose of the first preliminary study was to select eight texts comparable in cognitive load and the purpose of the second preliminary study was to select eight typefaces.

The measurements were done in a quiet room with walls painted with grey matte paint in accordance with the ISO 3664 standard (ISO 3664, 2009). The letters of the texts that the participants read on the screen were dark on a light background (text colour: #000, background colour: #eee) according to the ISO 12646 standard (ISO 12646, 2015). The participants were located at a distance of 60 cm +/− 1 cm from the screen, in line with the recommendations of the ISO 9241-303 standard (ISO 9241-303, 2011). Their movements were not restricted, but they were asked to remain at a fixed position.

First, we examined the relationship between the subjective and objective indicators of the emotional response to reading. We found that ratings of pleasantness predicted pupil size (b = 0.01, β = 0.17, SE_b = 0.004, t(111.6) = 3.32), with strong support for the alternative hypothesis that the two variables are correlated (BF = 51.75). This suggests that readers responded emotionally to less or more pleasant typefaces. Pupil size was larger when reading typefaces were rated as more pleasant than when reading typefaces were rated as less pleasant (see also Figure 2).

Next, we examined the effects of typeface shape, age, and their interaction on pleasantness ratings and pupil size. Table 1 shows the results of linear mixed modelling and the Bayesian factors for each effect. BF values greater than 1 indicate evidence for the tested model, i.e., the model with both fixed and random effects, and values less than 1 indicate evidence for the reference model without fixed effects. In Table 1, we see that our data show very strong evidence for the effect of typeface shape on the pleasantness ratings. The extremely high BF value for the effect of typeface shape indicates that the model with fixed and random effects was preferred to the model with only random effects.

Figure 3A shows the pleasantness ratings and Figure 3B shows the pupil size under different experimental conditions. Typeface shape, as already mentioned, affected the ratings of pleasantness. In general, readers rated round/rounded typefaces as more pleasant than the angular/pointed ones. No such effect of typeface shape was observed in the pupil size data. Pupil size was only slightly larger for round/rounded typefaces than for angular/pointed typefaces (Figure 3B). Pleasantness ratings and pupil sizes were relatively similar in children and adults.

Our data showed no evidence of a fixed effect of typeface shape on the number of regressive saccades; however, there was moderate evidence that typeface shape affected reading time per character (see Table 1; Figure 4A). The round/rounded typefaces had lower reading time per character than the angular/pointed typefaces. Thus, we can confirm that round/rounded typefaces allow for more fluent reading than angular/pointed typefaces.

An interesting discovery was that there was no fixed effect of age on the number of regressive saccades (see Table 1; Figure 4B). However, there was strong evidence for the effect of age on reading time per character (see Table 1; Figure 4A). Children read more slowly than adults.

Figures 5A,B show comprehension and memorisation scores under different experimental conditions, respectively. The analysis revealed no fixed effects of typeface shape, age, or their interaction on text comprehension or memorisation beyond the random effects. The BF values were in favour of the models with only random effects.

The pleasantness of the typeface was tested with a hedonic scale in which readers rated how pleasant they found the typeface. Both children and adults found round/rounded typefaces more pleasing than angular/pointed typefaces (see Table 1; Figure 3A).

The effect of different typeface shape on subjective experience was also tested by measuring pupil size while reading different typefaces. The measured pupil size was slightly larger when reading round/rounded typefaces (this can also be seen in Figure 3B), which was also perceived as more pleasant by the readers. The rated typeface pleasantness correlated with pupil size (see Figure 2), supporting the assumption that the shape of the typeface influences the reader’s emotional experience (Hess and Polt, 1960; Margareth et al., 2008; Wang et al., 2018). However, only the fixed effect of typeface shape on pleasantness ratings was convincing, whereas the effect of the typeface shape on pupil size was less remarkable. The analysis indicated that small differences in pupil size when reading round/rounded and angular/pointed typefaces could be a consequence of interindividual differences and could be attributed to random effects, i.e., to individual differences in pupil size and interindividual variability in the effect of typeface shape on pupil size. The fact that the effect of typeface shape on pupil size was smaller than effect of typeface shape on pleasantness ratings might indicate that factors other than the reader’s emotional response, e.g., surprise (Preuschoff et al., 2011) or cognitive load (Gavas et al., 2018), influenced pupil size, although we tried to control for cognitive load by selecting texts with homogeneous difficulty.

We found that the shape of the typeface had an effect on one of the measures of reading fluency, i.e., reading speed. Readers read round/rounded typefaces faster than angular/pointed typefaces (see Table 1; Figure 4B). Typeface shape did not show notable effects on other measures of reading fluency and reading performance measures. It is possible that our comprehension and memory tests were not discriminative enough to detect differences between the two typeface shapes. Future studies should use psychometrically validated measures of memorisation (and comprehension) for the texts used in the study.

Based on the results of our study, we can conclude that the shape of the typeface can influence reading speed and feelings of pleasantness while reading. Round/rounded typeface shapes may be perceived as more pleasant than angular/pointed shapes. Round/rounded typefaces also support reading fluency and allow readers to read faster.

Reading time varied by reader age – as expected, children read more slowly than adults, who tend to be more experienced readers (see Table 1; Figure 4B).

A somewhat surprising result was that the number of regressive saccades during the reading was not affected by age; that is, children did not have, on average, a higher number of regressive saccades than adults, as would be expected given their reading experience (see Table 1; Figure 4A). There were also no major differences between children and adults in text comprehension and memorisation. This can probably be explained by the fact that the texts used were not complex; they were easy to read and could be processed easily by both age groups. Future studies should examine how different reading parameters change with increasing text difficulty and whether age interacts with text difficulty in predicting reading performance and emotional and physiological responses during reading.

Our study had several limitations. Even though we used texts of comparable difficulty, factors other than typeface shape may have influenced the results.

First, different participants might have responded differently to different texts. Their emotional response might depend on their specific interests (e.g., adults might respond differently to descriptions of animals than children). This could have increased the between-subject variability of the data.

Second, the typefaces we used differed in some characteristics that could affect reading parameters, such as typographic tonal density and overall character size: for example, we controlled for the x-height, but the different typefaces had different sizes of ascenders and descenders. As a result, the whiteness in the ascenders and descenders of the different typefaces was different, resulting in different line spacing, even though the leading was set to the same size (e.g., to 140%). Because of the different whiteness in ascenders and descenders, and because of the different counter shapes of the letters of different typefaces, the typographic tonal density value of texts in different typefaces will always be different, even if we unified the size of the x-height. Previous studies (Franken et al., 2015; Pušnik et al., 2016) have shown that factors such as these can affect reading speed and letter recognition. Future studies should investigate how manipulating a single feature of the typeface (e.g., only the shape of the strokes, while controlling for all other features, if possible) affects reading.

Third, the COVID-19 pandemic made it difficult to include larger samples, and the power of our complex statistical tests was low. Future studies should include larger samples.

Nevertheless, we believe that our results, although they should be considered preliminary, are quite informative because different measures of text processing were used, and although the fixed effects studied did not appear to be salient, all results pointed in the same direction – reading was more pleasant and fluent, and reading performance was minimally better with round/rounded typefaces compared to angular/pointed ones. Further studies will need to be conducted to provide more evidence, but our results suggest that it is important to consider typeface shape when examining reading or comparing findings from different studies.

Based on the results of our study, the use of round/rounded typefaces is recommended for the design of educational materials because readers or learners experience more pleasant feelings when reading than with angular/pointed typefaces. Using round/rounded typefaces also allows learners to read faster, which can have a positive impact on the learning process. The effect of typeface shape was similar in primary school pupils and university students, showing that the effect of typeface shape can be generalised across ages for simple texts. The typefaces with round/rounded shapes could be recommended for the design of educational materials used on the screen of a digital device for less experienced and more experienced readers. Such typefaces could make the learning process easier and more enjoyable.