Harder-to-read fonts boost student learning

Making learning materials more difficult to read can significantly improve student performance. Yes, you read that correctly. Connor Diemand-Yauman and his colleagues think the effect occurs because fonts that are more awkward to read encourage deeper processing of the to-be-learned material.

Diemand-Yauman first tested this principal in the lab with 28 participants (aged 18 to 40) who spent 90 seconds learning the seven features associated with three alien species. Half the students learned from materials written in clear 16-point Arial font, whereas the other half learned from materials written either in 12-point Comic Sans or 12-point Bodoni. As the researchers explained, these last two fonts are obviously more difficult to read when considered side-by-side with the Arial font, but viewed on their own few people would notice anything amiss. Fifteen minutes later the participants were tested and the key finding was that those who learned from the harder-to-read fonts answered 86.5 per cent of questions correctly, compared with the 72.8 per cent success rate achieved by the participants who learned from the clearer font.

For a follow-up study the researchers collaborated with a high school in Ohio. Teachers sent in their work-sheets and power-point slides and the researchers made them more difficult to read. They did this either by switching the fonts to Comic Sans Italicised, Haettenschweiler or Monotype Corsiva, or, if the materials were hand-written, simply by shaking them about in a photo-copier to make them blurry. The history, English and science teachers used the manipulated materials for one of their classes but not the other, which acted as a control. You guessed it, of the 220 participating pupils, those who learned from the harder-to-read materials subsequently performed better in the relevant class assessments than did the pupils who learned from the unadulterated materials (for more statistically minded readers, the effect size was d=.45).

When people find something easy to read, they take that as a sign that they’ve mastered it. Conversely, the researchers believe harder-to-read fonts provoke a feeling of lack of mastery and encourage deeper processing. However, there’s obviously a balance to be struck. If material becomes too difficult to read, some students may simply give up. Another possible mechanism is that the less legible fonts are somehow more distinctive, rendering them more memorable. Diemand-Yauman’s team doubt this explanation because distinctiveness should wear off over time, and anyway they didn’t use any fonts that pupils wouldn’t have seen before.

The researchers think their finding could be the tip of the ice-berg as regards using cognitive findings to boost educational practice. ‘If a simple change of font can significantly increase student performance, one can only imagine the number of beneficial cognitive interventions waiting to be discovered,’ they said. ‘Fluency demonstrates how small interventions have the potential to make big improvements in the performance of our students and education system as a whole.’

ResearchBlogging.orgDiemand-Yauman, C., Oppenheimer, D., and Vaughan, E. (2011). Fortune favors the Bold (and the Italicized): Effects of disfluency on educational outcomes. Cognition, 118 (1), 111-115 DOI: 10.1016/j.cognition.2010.09.012

Post written by Christian Jarrett (@psych_writer) for the BPS Research Digest.

30 thoughts on “Harder-to-read fonts boost student learning”

  1. You need someone who knows fonts, to review this.

    The first three examples (clear 16-point Ariel font, 12-point Comic Sans or 12-point Bodoni) have a ringer: the last is a serif typefaces designed for running text; the first and second are non-serif typefaces designed for titles or constant-width needs.

    in your second list, Monotype Corsiva is a serif font for running text.

    I would expect a short story written in 12-point Bondoni (or Monotype Corsiva) to be far easier to read than Ariel (at any size).

    1. Indeed, you would have to isolate for variables such as serif and sans serif. Personally, I’m much more engaged when reading Comic Sans and Bodoni than I am reading Ariel (which is a font I really dislike). Both Comic Sans and Bodoni are quite ‘readable’ fonts; perhaps, less scannable, but certainly readable. In fact, if you’re isolating for readability you first need to reference data on readability measures of each typeface, then measure a sans serif font against another sans serif font – one with different readability measures. From this synopsis, there are way too many variables that weren’t isolated. Drawing a conclusion that readability is the independent variable is unsound.

    2. And FYI: 16 pt Ariel is more difficult to read, by readability measures than 12 pt Ariel on printed material. Readability changes as one moves from print to digital (larger sizes are easier to read online). Moreover, 12 pt fonts encourage faster reading, as people are able to process more words (we read in chunks of sentences) than larger sizes.

  2. It's an interesting idea, makes sense. Sometimes when I'm reading something, I'll get through it nice and fast and then realize I haven't really taken any of it in – although this doesn't happen often nowadays, but when I was a student it used to be a big problem.

    I started making quick notes summarizing every paper I read, not because I planned to consult the notes (I never did) but because it forced me to understand them.

    It really helped me.

  3. A fascinating idea!! As a 5th grade teacher, I am constantly looking for ways to increase comprehension and to heighten the level of focus for my students. In today's high tech climate, we sometimes need to look for simplistic ways (and altering fonts is one of them) to try and reach all kinds of learners and attend to multiple levels of intelligences, and multiple levels of learning styles. This is worth a try–Brilliant, Mr. Diemand-Yauman!!

  4. I bet at least part of this is because the reader has to slow down and that gives more time for language processing unrelated to the visual part to occur.

  5. Hey cool! I remembered a post in either this blog or one of the other cog sci blogs I read where I saved a bookmark to the source material on how difficult fonts affected people's judgement of the difficulty of a task, or how long it would take, or their perception of reliability or beauty.

    So I wonder if people in a class would procrastinate doing the homework because of perceiving how difficult the material is to read? 🙂

    here is the other article.


    “If it’s easy to read, it’s easy to do, pretty, good, and true

    Hyunjin Song and Norbert Schwarz describe some fascinating findings on how fluency affects judgement, choice and processing style”

  6. Could you get them to experiment on people doing programming tasks by playing with fonts in their editors? fixed length or not, syntax highlighting or not, etc.

    It is fun to think of all the different things one could try.

  7. Sorry to follow up so many times, but this topic is fun and I keep having more things to wonder about.

    I wonder if the teaching material were presented in a harder to read font, but assignments were presented in an easy to read font, that students would have more confidence in their abilities to do the assignments.

    btw, I wish I had an rss feed to the comments in this blog so that I could follow subsequent discussions.

  8. WRT the question of if the harder fonts affect reading speed – my own research (undergraduate senior thesis in Cog Sci, a very long time ago) showed that degraded fonts don't significantly slow down readers as long as the fonts are legible. They do complain that the text is harder to read, however, and they perceive their speed as slower.

  9. I do not like the Ariel font or Helvetica. They are not good fonts, cause your I and l look the same, among other odd issues… comic sans is a bit better though.

  10. ….you should have posted this in a more difficult to read font.

    Im just sayin…

  11. Any chance the groups that scored higher were actually just made up of people who memorize things better or who were smarter? Why not test the same group twice with different sets of info so that you know the intelligence and ability to memorize is the same?

  12. If this is serious research on fonts, don't you think it would make sense to spell the name of the font correctly? It's ARIAL. No e.

  13. This is total rubbish (given the fonts that were used in this test) and unfortunately, from the comments, has led several people off to implement changes that will be damaging to there craft of teaching. Like all bad science, trying to undo the harm that has been done will never be accomplished. Shame, Shame, Shame!!!

  14. Your post doesn't mention that Connor was an undergraduate at Princeton when he conducted this research. That's impressive in and of itself… and if you take the time to read about him, he's a pretty impressive guy in many other ways as well, from reality TV star to student body president.

  15. Apple, pears and poor research! Newspapers may be on the wane in terms of their media clout but there is over 100 years of international research (e.g. at sources like http://www.printanddigitalresearchforum.com) that show the differences in readability of both typefaces and individual fonts. Fundamentally seriffed fonts like Bondoni (Times New Roman etc) can be read with greater comprehension than san serif fonts like Arial and Comic Sans.

    This study hardly qualifies as 'research', as the sample is too small (28) to draw and statistically significant conclusions and there was an even more fundamental error; using two different font sizes (16 and 12 point).

    Perhaps all this work shows is that a basic misunderstanding of typography and design is very widespread. The wonder of technology is it allows fast experimental design, but it also allows lots of design that while visually sophisticated, actually gets in the way of communicating. For example, comprehension is also far lower where there text is reversed out, or where layout of text does not scan top to bottom left to right. Will this change, probably, as people grow up reading on screen, but what this piece really shows, is that we have forgotten some of the basics of typography and design to the detriment of readers.

  16. Richard – to clarify, this study isn't an investigation into which types of font are easier to read. It's about the effects of readability on memory and levels of processing. You say that the sample is too small to reach statistically significant conclusions – this is incorrect, since statistically significant differences were observed between the groups.

  17. Interesting that the statistical difference was noted, but I too wonder whether the reason for the learning difference is actually what the researchers concluded. Serifed fonts are easier to read (at small sizes) than sans serif fonts, and I wonder if the effect might be explainable by the fact that a 12-point sentence is more readily scanned by the eye than a 16-pt one that may require more eye movement. In the second test, the Haettenschweiler is definitely harder to read because it's a very narrow font–but does that again make it more quickly scanned? I wonder how the handwritten and blurred elements did in comparison to the clear versions. (Teachers still use handwritten docs?)

  18. Here's your problem: “Half the students learned from materials written in clear 16-point Arial font, whereas the other half learned from materials written either in 12-point Comic Sans or 12-point Bodoni. As the researchers explained, these last two fonts are obviously more difficult to read when considered side-by-side with the Arial font, but viewed on their own few people would notice anything amiss.”

    “Obviously more difficult to read”? Given that this assertion is the basis of the research Diemand-Yauman might have checked it. He would have tripped over and perhaps been swallowed by existing research on what is readable, different contexts which affect readability, comparing typefaces for readability/legibility, etc.

    This says nothing about the hypothesis, which is perfectly reasonable and worth investigating. Diemand-Yauman hasn't investigated it, however.

  19. In respect of the last comment, I think the clue is in the information; 'when considered side by side'. The students were offered Arial and Comic Sans or Bodoni separately, in which case, as the researcher indicates, 'few people would notice anything amiss'. I would suggest, therefore, that it also might escape their notice that either font is 'more difficult to read'. They simply aren't. Annoying, maybe, but more difficult? It's an interesting hypothesis but one which hardly seems adequately tested by this piece of research.

  20. I am a second year Undergraduate Psychology student and want to test this for my practical report. I was going to basically replicate it but change the materials used, so instead use a factual extract of text rather than made-up alien names. Having read all of the comments on here I was wondering if anyone could suggest a better font instead of Bodoni or Monotype Corsiva which would be suitable to test alongside Arial and Comic Sans? I don't know much about fonts or what serif typefaces/non-serif typefaces means, I just thought it was an interesting thing to investigate.
    I very much appreciate any suggestions.

  21. I looked at fonts and intelligibility some years ago.

    Almost all the research I read concluded that if the material is to be read and retained, a font with serifs is better than one without. If what’s wanted is for content to be acted on immediately and not retained, a sans-serif font is better.

    I suspected (it was the eighties) that fonts with serifs are processed more carefully, so more meaning falls into short term memory and from there can, with attention, proceed to long term. Simple sans-serif fonts are more easily digested and needn’t register in more than an eidetic way.

    Complex sans serif fonts like Comic Sans Italicised still need more processing so stand a better chance of making it into real memeory.

    Courier, without proportional spacing, came with manual typewriters and is now rare except in line printers and computer interfaces like this one (though the text is transformed for display). Courier also works well with optical character recognition – a place Arial does less well with no serifs and fewer distinguishing features. Not all serifs appear equal!

    Personally, I like the look of text with plenty of white space and serifs, but Arial (almost Helvetica by another name, avoiding royalty dues) is a pervasive thing, but a bit sterile.

    Maybe I should have been a middle aged monk, dedicated to illuminated script on deer hide…

Comments are closed.