Pretty ugly

The paper by Diemand-Yauman, Oppenheimer, and Vaughan, ‘Fortune favors the bold (and the italicized)’, made it to the Today programme on Friday, courtesy of Jonah Lehrer of Wired. This paper reports two linked studies investigating the claim that a degree of disfluency in documents (put simply, making them harder to read) leads to improved memory performance.

Its findings have been reduced to ‘ugly fonts help you learn’, ‘typographic history is wrong’, even ‘Kindles are bad for you’, but I’m sure these leaps are not justified by the paper itself. As someone who thinks that good typography consists of good writing that is well articulated, well set, and properly printed (or presented onscreen), I find the concentration on the ‘magic bullet’ effect of a typeface change worrying. Setting your text in Caslon won’t make you write like G. B. Shaw, despite the Shavian insistence that his works be set exclusively in that typeface.

First, let’s look at the typefaces that the investigators considered disfluent: Comic Sans, Comic Sans Italicized, Bodoni MT, Monotype Corsiva, Haettenschweiler (similar to Compacta Bold). The first experiment compared Arial with Comic Sans Italicized and Bodoni MT. It isn’t clear from the paper whether the italic or roman version of Bodoni MT was used – one of the problems is that we are only shown two example stimuli from the first experiment, and none from the second experiment. For an investigation in the effect of visual presentation of text, this is a pretty big omission.

With the exception of Haettenschweiler, all these fonts are near normal in weight and proportion. Comic Sans may be the butt of jokes and typographic snobbery, but its letterforms are clearly within the norms of the western type canon. Monotype Corsiva might appear ‘arty’ to some but again it isn’t really a leftfield font. At first sight Haettenschweiler is – at least it has normal letterforms emboldened and condensed to a very abnormal degree for a text face – but on reflection it’s a fairly conventional font for newspaper headlines. So while a longish text in Word might be really difficult, a short bullet-pointed list, or a straightforward PowerPoint slide might not be so problematic. But there’s a special problem with Haettenschweiler, which we’ll come to.

For experiment 1, we know that simple, comparable texts were set in 16pt Arial, 16pt Comic Sans Italicized (60% greyscale), and 16pt Bodoni MT (60% greyscale). I’ve recreated one of the experiment 1 texts in all the typefaces tested in both experiments (I’ve sloped the Comic Sans by 12 degrees).

Different (from Arial) the variant fonts may be, but I would contend that they are not ‘ugly’. Screen resolution may improve or degrade the effect of tinting the type, but these variants are well within the norms of screen display and print production. (Remember Haettenschweiler and Monotype Corsiva weren’t used in experiment 1.)

Experiment 2 set school learning materials in the ‘disfluent’ typefaces. The process isn’t described very fully, because we get no idea of what the documents (Word and PowerPoint) looked like to begin with, or how much text they contained. We are told that one teacher objected to the use of Haettenschweiler because it was ‘difficult to read’ – Comic Sans Italicized was immediately substituted. We are also told that ‘the font size of the supplementary material was not changed unless the size coupled with the disfluent font made the text illegible as reported by the teachers or the experimenters, in which case the font size was adjusted to allow legibility’ (my italics).

The tables in the paper demonstrate the statistical basis of the claim that the disfluent presentations were more memorable to students. But they don’t break down the results by document type and font. When questioned on their feelings about the material they read, the students reported no discernibly different responses to the materials they were presented with, and that this ‘lack of observed liking/motivational differences between fluency conditions is unlikely to be due to insensitivity measures.’ (This strikes me as very odd, because the work done by Sue Walker demonstrates that schoolchildren can articulate their responses to typographic presentation very clearly.)

So what might we conclude from this paper? Its conclusion is that ‘small interventions have the potential to make big improvements in the performance of our students and education system as a whole.’ What they have found is that changing a document from a typographic default, a norm, makes for an improvement. That somehow doesn’t come as a surprise to me: Arial was designed to look like Helvetica, not to be the ideal typeface for children’s educational materials. Increasing inter-liner spacing often improves a text no end. All the variant fonts, Haettenschweiler excepted, have what conventional typographers might consider either better inter-linear spacing or a more even rhythm of character spacing. Comic Sans (and perhaps even Monotype Corsiva) resemble handwriting. Haettenschweiler looks like the bold, brassy headlines in tabloid newspapers.

What if the findings actually indicate the opposite: that Arial is the ‘ugly’ font, and moving away from a crowded, over-tightly spaced font improves things? Far from working because they are ‘ugly’, these other typefaces might work simply because they are ‘different’.

¶ See also Alison Black’s blog.