I’m delighted to have heard back from Stephen Banham at RMIT who says that this research is in its infancy and that research papers will follow over the next six months or so. 

Tweeting about Sans Forgetica brought some precedents to light. In particular, Dominique Joseph alerted me to previous discussions in the plain language community and cited this paper, which itself contains quite a number of earlier references.

Diemand-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.

As the typography in the title shows (well done, Cognition, for letting its hair down on this occasion although you could have aligned the x-heights), this study also used less legible type to affect learning. Here’s a good summary and discussion of it: https://digest.bps.org.uk/2010/12/17/harder-to-read-fonts-boost-student-learning/

And Gianni Ribeiro sent me a link to a critique of this team’s work by Meyer et al entitled ‘Disfluent Fonts Don’t Help People Solve Math Problems‘. 

My reading of the Sans Forgetica team’s press release is that they see the font change as (presumably) noticeable by the reader, but that the difficulty it causes affects them in a way that they are not actually aware of (that is, deeper cognitive processing takes place while they are struggling to read the text surface).

However, Diemand-Yauman and his colleagues seem to suggest that readers see the font change as an explicit signal that they need to interpret and potentially act upon

‘Importantly, disfluency can function as a cue that one may not have mastery over material (for a review, see Alter and Oppenheimer (2009)). For example, studies have shown that fluency is highly related to people’s confidence in their ability to later remember new information (e.g. Castel, McCabe, & Roediger, 2007). To the extent that a person is less confident in how well they have learned the material, they are likely to engage in more effortful and elaborative processing styles (Alter et al., 2007).’

So, assuming the RMIT finding is valid, we seem to have a debate about how the illegible font is acting on the reader – whether in a deep unarticulated way, or at the surface level of deliberate reader choices.

Meyer et al characterise it like this:

‘Many distinguish intuitive thoughts, released merely by exposure to stimuli, from reflective thoughts, occurring after deliberate deployment of additional operations (Shweder, 1977).’

They do an extensive review of the Alter et al (2007) study, which they report has been cited numerous times, including popular books by Malcolm Gladwell and Daniel Kahneman. Their paper is a detailed meta-review of a number of studies which try to replicate the effect, and a detailed critique of methodologies employed. Their conclusion is clear from the paper’s abstract:

Prior research suggests that reducing font clarity can cause people to consider printed information more carefully. The most famous demonstration showed that participants were more likely to solve counterintuitive math problems when they were printed in hard-to-read font. However, after pooling data from that experiment with 16 attempts to replicate it, we find no effect on solution rates. We examine potential moderating variables, including cognitive ability, presentation format, and experimental setting, but we find no evidence of a disfluent font benefit under any conditions. More generally, though disfluent fonts slightly increase response times, we find little evidence that they activate analytic reasoning.

In my view this is the level of review and critical thinking required before research results are released into the wild.

Ignoring for now this rather devastating demolition job, and going back to the Sans Forgetica and Diemand-Yauman findings… if disfluency is a signal to process the content differently, then it seems no different from other kinds of cues which do not involve illegibility or disfluency – for example, highlighting something in colour, or a teacher saying: ‘make sure you get this, because it’ll come up in the test’.

In fact, I can’t see how the font variation can go un-hypothesised by the reader, who is bound to ask themselves ‘why are these words in bold/italic/a crazy font’.

And, by the way, I italicised the word ‘can’ in the last sentence to emphasise it, not so you would remember just that word. Font changes already exist in our writing system, and have generally agreed functions.

In the world of instructional research, typography has occasional moments in the sun. This one reminds me of 1980s work on ‘typographic cueing’ (for example, Glynn 1978) which also used highlighting to signal important concepts. A problem with many such studies is that text is already visual as well as verbal. Typography and layout already exist and are used by readers to navigate documents and ideas, and to read actively and strategically.

And underlying all of this is an often unacknowledged debate about whether learners are passive sponges, soaking in knowledge (squeezed out later in a test, as proof of learning) or whether they are active participants in education.

Instructional designers have been debating this for many years. When I first worked alongside them at the Open University in the 1970s, educational psychology was in transition from behaviourism to cognitive theories. Behaviourists essentially saw humans as a sophisticated form of rat, whose responses to stimuli (rewards and punishments) could be studied and manipulated. They looked at observable behaviours, rather than speculate too much about invisible cognitive processes. In the educational context, this led to ‘programmed learning’ in which content was learned in tiny steps, with success rewarded along the way. With the exception of certain types of industrial training this led nowhere.

An influential researcher at the time was Ernst Rothkopf,* whose theory of ‘mathemagenic behaviours’ suggested that there are behaviours that give rise to learning, which can be induced or encouraged, even though not observed. Mathemagenic behaviours could be encouraged through, among other things, frequent inserted questions in text, which appeared to influence learning not only of the topics thus highlighted, but of other parts of the text also. Many, many studies were published on the topic.

I only mention it in order to introduce a famous review of the theory by Ronald Carver (1972) – a very readable and quite excoriating critique that’s a great introduction to instructional research at that time. I like his conclusion that ‘It appears it would be a questionable use of the practical decision-maker’s time for him to wade through this recent research since it is mainly irrelevant to most applied situations.’ Still applies…

It’s relevant to this current discussion for three reasons.

Firstly, Carver is very convinced that the speed at which students read, and the time they spend on text affects learning – so this supports the Sans Forgetica team’s view that slowing people down is a good thing. He cites a very early paper by Green (1931) who first pointed this out, and criticises numerous researchers for failing to control for this and report on it.

Secondly, he brings the learner’s own strategy into the foreground. He talks of ‘self-directed reading as a problem-solving process’ and of the reader’s ‘plan’ or ‘program’ being of primary interest, rather than speculation about observed behaviours.

Lastly, he points out that a statistical difference in a lab experiment is not in itself of value unless it relates to a theory (and I would extend this to: unless it survives in a practical environment). Without the theory, which enables generalisation, the result doesn’t matter.


Alter, A. L., & Oppenheimer, D. M. (2009). Uniting the tribes of fluency to form a metacognitive nation. Personality and Social Psychology Review, 13: 219–235.

Alter, A. L., Oppenheimer, D. M., Epley, N., & Eyre, R. (2007). Overcoming intuition: Metacognitive difficulty activates analytic reasoning. Journal of Experimental Psychology, 136: 569–576.

Carver, R.P. (1972) A critical review of mathemagenic behaviors and effect of questions upon the retention of prose materials. Journal of Reading Behavior, 4: 93-119.

Castel, A. D., McCabe, D. P., & Roediger, H. L. III, (2007). Illusions of competence and overestimation of associative memory for identical items: Evidence from judgments of learning. Psychonomic Bulletin and Review, 14: 107–111.

Diemand-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.

Glynn, S.M., (1978) Capturing readers’ attention by means of typographical cuing strategies. Educational Technology, 18 (11): 7-12.

Green, E.B. (1931) Effectiveness of various rates of silent reading of college students. Journal of Applied Psychology, 15: 214-227.
Meyer A., Frederick S., Burnham T.C., Guevara Pinto J.D., Boyer T.W., Ball L.J., Pennycook G., Ackerman R., Thompson V.A., Schuldt J.P. (2015). Disfluent fonts don’t help people solve math problems. Journal of Experimental Psychology: General, 144(2): 16-30

Rothkopf, E.Z. (1970) The concept of mathemagenic activities. Review of Educational. Research, 40: 325-336.

Shweder, R. A. (1977). Likeness and likelihood in everyday thought: Magical thinking in judgments about personality. Current Anthropology,
18, 637– 658.

*In case I’m misunderstood, my memories of Ernst (who I met at several conferences) were of an exceptionally warm, generous and approachable man who thought deeply about teaching and learning. I often quote one of his observations in support of the importance of layout – that people often remember things from the position on the page where they read them.

Rothkopf, E.Z. (1971) Incidental memory for the location of information in text, Journal of Verbal Learning & Verbal Behavior, 10, 608–613