Using difficult typefaces--to increase comprehension

samuel_b's picture

Has anyone at Typophile ever designed something to be harder to read on purpose--to make sure the reader slows down, and thus better understands the text? In particular, has anyone ever done this with a textbook or other educational materials? Or has anyone made another design decision with a similar effect, such as printing the text in light gray?

Note: there is some empirical research suggesting that using more difficult typefaces can increase comprehension. Dan Kahneman discusses some of it in Thinking Fast, Thinking Slow, and a very recent example is the new study described here, which finds increased gains for dyslexic readers:

http://www.telegraph.co.uk/education/educationnews/10089102/Hard-to-read...

Chris Dean's picture

See also:

Diemand-Yauman, C., Oppenheimer, D. M. & Vaughan, E. B. (2010). Fortune favours the bold (and the italicized): Effects of disfluency on educational outcomes. Cognition, 118(1), 111–115.

(discussed in an older Typophile thread — Disfluency disrupts the confirmation bias)

Yue, C. L., Castel, A. A. & Bjork, R. A. (2013). When disfluency is—and is not—a desirable difficulty: The influence of typeface clarity on metacognitive judgments and memory. Memory & Cognition, 41(2), 229–241.

hrant's picture

The only point of making people slow down is to help them catch typos. Sloth is not a virtue.

hhp

JamesM's picture

Not sure that techniques for dyslexic readers would also work for folks without that problem.

I haven't read the studies, but generally speaking designers make something harder to read if their clients want to discourage folks from reading it. For example, in annual reports bad news is sometimes put in dense paragraphs of all caps, with numbers spelled out, making it harder to read.

Nick Shinn's picture

I don’t equate slowing down reading with making it more difficult to read.

Jens Kutilek's picture

That Arial is easy and Monotype Corsiva is hard to read is just an assumption the teachers did make. Maybe Corsiva is more readable because it's an italic, maybe because the letter forms are more differentiated than in Arial. The same results could be argued to just prove the opposite, that Corsiva is more readable and thus the pupils' comprehension was better.

This study doesn't really hold up to scientific standards.

samuel_b's picture

Nick, could you give any specific examples of the difference? How would you encourage readers to slow down without making the text more difficult to read?

Jens, I haven't read the study cited in the news article and can't vouch for its research design. But it's not the only one--I mentioned Kahneman's secondary work, and Chris listed other studies. I don't care about the details of this study (or the dyslexia vs. general population point). I'm interested in the general principle.

Nick Shinn's picture

How would you encourage readers to slow down without making the text more difficult to read?

Speed/slowness and ease/difficulty are not the parameters I use for typeface design or typographic design. I prefer not play the game according to those rules.

The following example isn’t “text type” per se, which is what some say should be optimally decoded at rapid speed prior to comprehension, but poetry, where the idea is to savour the words. That principle applies to prose, too, although to a lesser extent.


Verlaine: «À la Princesse Roukine», Parallèlement. Paris, Ambroise Vollard, 1900

Thomas Phinney's picture

Even if the general principle applies to experimental conditions where the subjects must study the material, that may not translate well to the real world, where in 99% of cases people have the option of just not frickin' reading it, because they lost interest in doing the extra work of reading something that was harder to read.

John Hudson's picture

Poetry is an interesting case of text that generally calls to be read slowly, preferably out loud. When we talk about the high contrast romantic types of Didot and Bodoni being relatively more difficult to read than lower contrast, oblique axis types, we often overlook just how much poetry the Didots and Bodoni were publishing.

Chris Dean's picture

A quick review for those who don’t have database access (or don’t want to pay $40.USD to download it).

For starters, the authour’s and those who administered the tests are middle-school teachers, not research scientists.

Subjects (Ss)
275 students in Years 9–11 (age range= 13–16 years).
Mean and standard deviation not reported.
Ss were previously grouped into four “bands” by Middle Years Information System tests.
Some were diagnosed with Dyslexya, but there was no agreed upon definition, nor were they diagnosed by the same person with the same methods. Uncontrolled.

Experimental design
Between Ss. Each group of Ss only received one condition (font—Arial or Monotype Corsiva).

Methods
Double blind study. Neither the experimenter (a teacher) nor Ss knew what condition they were about to administer/receive. During a 40 minute lesson of students copying material from the blackboard and answering questions from a textbook Ss were stopped, and shown a powerpoint slide. It contained the following text:

The white dwarf star Amethyst is in the constellation Cassiopeia. It is half the mass of the Sun but has a diameter one hundred times less than the Sun. On its surface it has a temperature of twelve thousand Kelvin. It is mostly made up of carbon. Currently, we have found three planets orbiting
around it.

The line-breaks can not be discerned from the figure presented in the paper.

Ss were given 90 seconds to read the text in silence. The lesson continued. Approximately 35 minutes later, they were administered the following questions:

1) What is the name of the star?
Amethyst/Amey/Myethyst/Alpha-Centauri

2) In which constellation is it?
Cassiopeia/Cancer/Capricorn/Pavo

3) What type of star is it?
White Dwarf/Red Dwarf/Red Giant/White Giant

4) How many times the mass of the Sun is it?
Half/Double/Ten Times/Five Times

5) What is the temperature on its surface?
Twelve Kelvin/Twelve Thousand Kelvin/Twelve Thousand Centigrade/Three Thousand Kelvin

6) What is it mostly made up of?
Carbon/Helium/Iron/Cadmium

7) Have we found planets orbiting it?
Yes/No

They were on paper. Their typographic design is not reported.
121 Ss saw Arial, 154 saw Monotype Corsiva.
Ss were grouped into two groups by random balanced assignment—random assignment followed by re-grouping Ss to ensure that each group had a similar percentage of Ss from all bands, including readers with Dyslexia.

Results
Across bands, and including dyslexic readers, Ss who saw Monotype Corsiva scored 12.8% higher. This was statistically significant.
Within bands, excluding dyslexic readers, the mean difference between Arial and Monotype Corsiva was 11.5%. This was statistically significant.

The interesting part of the results
Across bands, the difference between Arial and Monotype Corsiva was 12.8%. Within bands, the mean difference was 11.5%. However, when looking at only readers with dyslexia, the difference between Arial and Monotype Corsiva was 19%. While arithmetically different from the overall difference, it was not statiscically significant.

In one sentence
When students with Dyslexia saw a slide with a passage of text set in Monotype Corsiva compared to Arial , they performed significantly better on a short comprehension test, but the difference between their scores was not significantly greater than the difference between the scores of students without Dyslexia.

In conclusion
A nice short and easy to read article. however, it is full of confounds and impossible to reproduce from the paper. In the concluding remarks of Diemand-Yauman et al. (2010, pp 4), they state “The potential for improving educational practices through cognitive interventions is immense. 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. Fluency demonstrates how small interventions have the potential to make big improvements in the performance of our students and education system as a whole.”

This is an great example of non-scientists testing disfluency theory, and a testament at how easy it may be to implement these techniques in real-world settings, for which they should be applauded.

Frida VI's picture

That's very interesting research. I am also working on my thesis, trying to find the link between letterform and learning ability. Thanks for posting!

Chris Dean's picture

@Frida VI: Where do you study?

Jens Kutilek's picture

Thank you Chris for posting your summary.

«a simple change of font can significantly increase student performance» – I think no typographer would have claimed anything else ;)

Nick Shinn's picture

One slide in a 40-minute lecture.
What typeface(s) were the other slides set in?

Chris Dean's picture

Good catch Nick, you spotted a typo. When I wrote “40 minute lecture,” I meant ”40 minute lesson.” I have corrected.

At some point, during a normal class of students taking notes from a blackboard and answering questions from a textbook, Ss were stopped, and were shown a single slide. One group saw Monotype Corsiva, the other, Arial. There were no other slides.

Nick Shinn's picture

And what did the slides look like?
Was the setting skewed towards an outcome that supported disfluency theory?—as was the case in a previous experimented that “compared typefaces”:

Chris Dean's picture

There are no accurate visual representations or descriptions of the slides in the paper regarding size, dimensions, leading, visual angle, &c.

What do you mean by “Was the setting skewed towards an outcome that supported disfluency theory?”

As reported, it was a double blind study, and Ss were assigned to two different groups via random balanced assignment in an effort to ensure that the populations of the experimental and control groups weren’t unbalanced in any direction according to ability.

From my perspective, confounds aside, there was no deliberate attempt to skew the results via text setting.

JamesM's picture

Am I correct that the test consisted of just 1 slide?

I can see how slowing down might increase comprehension when dealing with a few sentences. But would the same result happen when reading a whole book, or would people (like Thomas mentioned) simply give up?

A real-world example: If someone sends me a handwritten "thank you" note, I'll take the time to decipher the writing. But a few years ago someone showed me a handwritten 1800s diary of our church history. Even though the subject interested me, I gave up after a couple of pages because reading the handwriting was just too slow and difficult.

Chris Dean's picture

Yes. There was just one slide (per condition).

Nick Shinn's picture

What do you mean by “Was the setting skewed towards an outcome that supported disfluency theory?”

In the sample I posted, one typeface was set large, bold and solid, to represent “typical”, and the other small, italic and screened, to represent “hard to read”. So, it was not typefaces that were being compared, but settings, according to the prejudices of the researchers, who rigged the experiment to set one typeface so that it was “harder to read” than the other.

Without seeing the slides involved here, I can nonetheless state that the Arial setting will look really boring, just the kind of thing that would make students’ eyes glaze over and their minds wander, while the Corsiva setting will have a degree of novelty that warrants more attention.

As I pointed out in a previous thread, what is being compared is not, as the researchers believe, difficulty of typeface, but appropriateness of type choice and type specs, about which they are somewhat ignorant, because that ain’t their bailiwick.

So congratulations, they have discovered typography!

The researchers here are victims of the Beer Test phenomenon: a small amount of something unusual will have immediate highlight appeal, but is not the kind of thing for prolonged consumption. Display Beer, as Hrant termed it.

Syndicate content Syndicate content