A Comparative Analysis of Bionic and Traditional Reading
Meghann Rupert, Deniz Tinc, and Izaz Zubayer
Department of Cognitive Science, Simon Fraser University
The Cognitive Science of Reading
December 4, 2023
Introduction
Many people struggle with maintaining focus during reading, and while there are specialized fonts designed to address the issue, a new method has surfaced that uses a unique approach. This new method called bionic reading was developed by a Swiss typographic designer named Renato Casutt. Bionic reading strategically bolds the most concise parts of words that serve as artificial fixation points, guiding the readers' eyes and enhancing reading efficiency (Sener, 2023). The theory posits that readers focus on the bolded part of the word and their brain completes the rest, enhancing the overall process. These bold parts are called "fixations", enabling readers' eyes to transition easily from one fixation to the next, letting their brains fill in the rest of the words and promoting a swifter and more effective reading experience (Sener, 2023). Since the brain can process information faster than the eyes, reducing the number of letters the eyes need to focus on enables quicker reading while in theory retaining the full contextual understanding.
The previous literature in this field consists of diverse studies on reading comprehension and speed, studying various factors that play a role in information processing. Vitu (2005) investigated how eyes extracted information in reading and the role of regressive saccades during this process. Soleimani & Mohammadi (2012) explored the effect of typographical factors on reading speed, comprehension, and recall. Rayner et al. (2016) examined the relationship between speed reading and comprehension. Owens et al. (2019) investigated gist comprehension using Rapid Serial Visual Presentation (RSVP), analyzing the impact of presentation speed and style on readers' ability to extract essential meaning. Powell & Trice (2019) demonstrated the impact of Dyslexia, a font specialized for people with reading disabilities, mostly ADHD.
However, this paper distinguishes itself by aiming to find a broader understanding of the effects of bionic reading on reading speed and comprehension. Rather than focusing on a specific target group, our paper strives to investigate the comprehensive dynamics between bionic reading technology and reading speed and comprehension. The main focus is the difference between bionic reading and traditional reading methods. It is hypothesized that bionic reading will improve reading efficiency by reducing reading time while preserving reading comprehension.
Methods
Participants
The sample included seven students at Simon Fraser University, comprising 3 female and 4 male participants. The mean age was 21.43 with a standard deviation of 0.904. All participants volunteered to take part in the study and had an English proficiency level of intermediate (B1) or greater. Three participants reported English as their first language, while the other four native languages were Bangla, Spanish, French, and Cantonese. There were no reported diagnosed reading or learning disabilities.
Materials
The independent variable in this study is the type of text, bionic or traditional. As previously discussed, the bionic text utilizes boldness to create a visual guide for fixations when reading (see Figure 1 for example). This variable was manipulated by randomly assigning either bionic or traditional reading texts to the participants. All texts were generated by ChatGPT (see Appendix A). The overarching dependent variable is reading efficiency, which is measured in several ways: reading duration, fixation duration, saccade duration, reading comprehension, and subjective experience. Reading duration, fixation duration, and saccade duration were all measured and analyzed using the EyeLink 1000 Plus arm-mounted eye-tracking device and software. Reading comprehension was measured by a comprehension question after each text. The study also included two questionnaires for the participants to complete. One questionnaire was given before the experiment, to establish the participants' reading abilities and other demographic information (see Appendix B). A second questionnaire was given after the experiment to get responses from the participants on how bionic reading affected their subjective reading experience (see Appendix C).
Procedures
After consenting to the study, the participants were asked to complete the pre-study questionnaire. The participants were then instructed on how to progress through the experiment. One by one, each participant had a target sticker placed on their forehead to allow the camera to track the participant's head position during blinks or other sudden movements. Then, the participants underwent the calibration phase which is conducted to provide the system with fixation samples from known target points on the monitor using the left eye of each participant. This data is then validated to ensure accuracy. The participants then began the reading task where they were presented with a series of text passages in a random order, and randomly assigned either regular or bionic text, which were counterbalanced to ensure equal distribution. Each passage was followed by a comprehension question, which the participant had to respond to by pressing the 'a' key for true and the 'l' key for false. After the reading tasks, the participants were asked to complete the post-study questionnaire.
Results
Average Reading Duration
The average reading duration for bionic text passages among all subjects was approximately 19690 ms, while the average for regular text passages was approximately 19952 ms. The average difference in duration between bionic text and regular text was -262.14 ms (SD = 1876.22 ms, ɑ = 0.05, p-value = 0.74). See Table 1 for more details.
Performance Data
| Subject | Bionic Text (ms) | Regular Text (ms) | Difference (ms) |
|---|---|---|---|
| 1 | 22114 | 20980 | 1134 |
| 2 | 17616 | 18439 | -823 |
| 3 | 14254 | 15139 | -885 |
| 4 | 16392 | 16846 | -454 |
| 5 | 18746 | 16414 | 2332 |
| 6 | 33192 | 37191 | -3999 |
| 7 | 15513 | 14653 | 860 |
Average Eye Movement Duration
The mean duration for the current fixation across all subjects for bionic text passages was 220.66 ms, while the average next fixation duration was 218.98 ms (see Table 2). For regular text passages, the average duration for the current fixation across all subjects was 222.78 ms, while the mean of all next fixation durations was 223.45 ms. The average saccade duration for bionic texts for all subjects was 49.65 ms and 51.01 ms for regular texts.
Fixation and Saccade Durations
| Text Type | Current Fix. Duration (ms) | Next Fix. Duration (ms) | Next Saccade Duration (ms) |
|---|---|---|---|
| Bionic (avg of 7 subjects) | 220.66 | 218.98 | 49.65 |
| Regular (avg of 7 subjects) | 222.78 | 223.45 | 51.01 |
| Subject | Comprehension Scores (Bionic) | Comprehension Scores (Regular) |
|---|---|---|
| 1 | 9/9 | 8/10 |
| 2 | 8/9 | 7/10 |
| 3 | 9/10 | 8/10 |
| 4 | 10/10 | 10/10 |
| 5 | 10/10 | 9/10 |
| 6 | 10/10 | 8/10 |
| 7 | 9/9 | 10/10 |
Reading Comprehension Scores
The average score for comprehension questions following a bionic text passage for all subjects was 97%, with five out of seven subjects scoring 100%. Meanwhile, the average score for questions following traditional text passages for all participants was 85%, with only two scoring 100%. See Table 3 for the raw scores for each subject.
Table 2
Subjective Experience
Five out of seven participants reported having no prior exposure to bionic reading, with one participant expressing some familiarity and another reporting a high level of familiarity. All participants conveyed a sense of comfort while reading traditional texts, with 4 indicating a high level of comfort and 3 reporting a moderate level of comfort. A majority of participants, constituting 57% (4 out of 7), perceived bionic text as more easily readable than traditional text, with 2 participants noting a significant improvement and 2 expressing a slight preference. Two participants discerned no discernible difference, while one found bionic text marginally more challenging. Notably, 71% of participants (5 out of 7) expressed an inclination toward considering bionic reading in their future reading endeavours.
Figure 1: Bionic text vs Traditional text
Discussion
Using bionic text resulted in a reading duration that was half a second shorter than conventional texts. Additionally, participants exhibited a reduced average fixation duration for bionic reading, yet achieved a higher accuracy in answering questions. These findings suggest that bionic reading facilitates lexical access, enabling individuals to identify words more effortlessly than with traditional texts. Consequently, this enhanced lexical access likely contributed to a faster comprehension of word meanings. The results also showed that fixation duration with bionic reading was significantly reduced on average compared to regular text. In eye tracking, fixation time is the duration eyes are focusing on a specific point such as a word. It can be inferred that individuals can process words quicker and more effectively with bionic reading, as indicated by shorter durations of fixations during the process. Participants P1, P2, P3, and P6, who were non-native speakers, exhibited a noteworthy pattern in their comprehension performance as observed in Table 3 of the results. It became apparent that these individuals made more mistakes in comprehension when presented with the regular text format. Gilakjani & Ahmadi's (2011) study on reading comprehension of L1 and L2 readers also supports these results, as familiarity with a language helps readers' comprehension by utilizing their background knowledge, called background schemata. L1 readers showed a more successful activation of their background schema compared to L2 readers, as the text was more familiar to those with L1 proficiency. Reading in a second language is found to involve extra cognitive skills which might have affected their comprehension performance. This observation underscores the potential influence of text format on the comprehension outcomes of non-native speakers, suggesting a need for further investigation into the impact of text presentation on language learners.
Building on the insights from Dyson & Haselgrove’s (2001) study, which highlighted the advantages of a medium line length of 55 characters per line, our approach to crafting paragraphs has been guided by a commitment to optimizing reading experiences. We recognize that this specific line length supports effective reading at both normal and fast speeds, leading to heightened comprehension levels and faster reading compared to shorter lines. Acknowledging the observed speed-accuracy trade-off in comprehension, we have intentionally structured our paragraphs to align with the optimal conditions identified in the study. By doing so, we aimed to mitigate potential biases and prevent reader exhaustion that may be associated with longer character lengths.
Expanding on Jackson & McClelland’s (1979) findings, which propose a connection between individual differences in reading speed and central cognitive processes, our interpretation considers the influence of language comprehension and verbal aptitude on reading speed. Notably, their research revealed that faster readers often exhibited lower accuracy criteria, indicating a nuanced interplay between speed and accuracy. Given these insights, we might speculate that participants 1 and 5, who demonstrated a potential correlation with lower reading speed and perhaps made more errors, could be associated with lower verbal aptitude. This hypothesis aligns with the idea that verbal proficiency plays a critical role in influencing reading speed. However, it is crucial to approach such assumptions with caution, as various factors can contribute to individual differences in reading performance, and additional research or data would be needed to validate this interpretation.
Nation (2009) emphasized the role of reading under pressure in reading dynamics. Although the study found that increasing reading speed positively correlated with comprehension, reading under pressure could affect enjoyment and potentially hinder reading comprehension. These effects could account for participant 2’s comprehension scores being slightly lower compared to the others
Implications & Limitations
The implications of this study extend significantly to the realm of education. Given the proven enhancement of comprehension through bionic reading, educational institutions should contemplate integrating this method into their reading materials. Moreover, standardized tests like the SAT could adopt such reading methods in their reading sections to evaluate students based on their comprehension skills, as it may significantly speed up both their reading and comprehension skills. Beyond education, the professional arena offers another promising avenue for the application of bionic reading technology. In fields that demand swift information processing, such as financial institutions, emergency services, healthcare, customer support, research and development, and legal practices, the adoption of this technology could prove invaluable. Professionals in these sectors could leverage bionic reading to stay ahead of the constant influx of new literature, research findings, and emerging trends, thereby facilitating continuous learning and proficiency maintenance.
Given the novelty of this research area, certain limitations warrant consideration for future investigations. The restricted sample size hinders the generalizability and replicability of the study's findings. One finding was that the overall reading duration was slightly shorter on average when reading bionic texts compared to regular texts. Future experiments should explore the impact of using longer text passages to assess whether this method significantly affects reading time on a larger scale. The present study was conducted with the researchers' presence in the room, which may have introduced potential biases. Future studies could increase measures to minimize subject expectancy bias. Additionally, the silent reading method employed by participants could have influenced their reading pace, possibly due to phonological encoding.
It is important to consider that much of the current research posits that individual differences are a significant factor as the results showed significant individual differences but only minor distinctions overall. Additional research on this topic is necessary to develop a more comprehensive understanding of the effects of bionic reading.
References
Dyson, M. C., & Haselgrove, M. (2001). The influence of reading speed and line length on the effectiveness of reading from screen. International Journal of Human-Computer Studies, 54(4), 585–612. https://doi.org/10.1006/ijhc.2001.0458
Gilakjani, A. P., & Ahmadi, S. M. (2011). The relationship between L2 reading comprehension and schema theory: A matter of text familiarity. International Journal of Information and Education Technology, 1(2), 142–149. https://doi.org/10.7763/ijiet.2011.v1.24
Jackson, M. D., & McClelland, J. L. (1979). Processing determinants of reading speed. Journal of Experimental Psychology: General, 108(2), 151–181. https://doi.org/10.1037/0096-3445.108.2.151
Nation, P. (2009). Reading faster. International Journal of English Studies, 9(2). Retrieved from https://revistas.um.es/ijes/article/view/90791
Owens, J. W., Chaparro, B. S., & Palmer, E. M. (2019). Exploring website Gist through Rapid Serial Visual Presentation. Cognitive Research: Principles and Implications, 4(1). https://doi.org/10.1186/s41235-019-0192-1
Powell, S. L., & Trice, A. D. (2019). The Impact of a Specialized Font on the Reading Performance of Elementary Children with Reading Disability. Contemporary School Psychology. https://doi.org/10.1007/s40688-019-00225-4
Rayner, K., Schotter, E. R., Masson, M. E. J., Potter, M. C., & Treiman, R. (2016). So Much to Read, So Little Time: How Do We Read, and Can Speed Reading Help? Psychological Science in the Public Interest, 17(1), 4–34. https://doi.org/10.1177/1529100615623267
Sener, E. (2023). A Comparison of Memory Performances for Expository Scientific Prose and Diagram in Flat vs. Spatially Distributed Layouts in Virtual Reality.
Soleimani, H., & Mohammadi, E. (2012). The effect of text typographical features on legibility, comprehension, and retrieval of EFL learners. English Language Teaching, 5(8). https://doi.org/10.5539/elt.v5n8p207
Vitu, F. (2005). Visual extraction processes and regressive saccades in reading. Cognitive Processes in Eye Guidance, 1–32 https://doi.org/10.1093/acprof:oso/9780198566816.003.0001