Mary Dyson

The devil’s in the g‑tails


Wong, K., Wadee, F., Ellenblum, G., & McCloskey, M. (2018). The devil’s in the g-tails: Deficient letter-shape knowledge and awareness despite massive visual experience. Journal of Experimental Psychology-Human Perception and Performance, 44(9), 1324–1335.


I am always excited when I come across a psychology paper which describes a study of a typographic variable. One of the first things I do is look at the list of references to see whether it includes any sources from typography. There rarely are any and I must remind myself that it is unreasonable to expect researchers in another discipline to be aware of the research literature on typography. It may not matter, and this is probably the case with these studies. However, I’m a fan of interdisciplinary research.

The paper reports on three experiments which explore skilled readers’ knowledge of two letter shapes, ‘g’ and ‘a’, considering that such knowledge is central to reading. A more informal account of the work is given at

The first experiment used a qualitative approach and asked 38 undergraduate students (probably from psychology) whether they could think of any letters that had two distinct forms in lowercase print, and to draw these forms. The researchers explain the distinction between clearly different shapes and ‘minor font variations’ such as serif vs sans serif. If the letter ‘g’ or ‘a’ was not named, two further stages of this experiment were used to try to tease out knowledge of the two forms of ‘g’ and ‘a’. Only two people named the letter ‘g’ as having two forms and they were unable to draw the looptail ‘g’ (also known as double-storey ‘g’). Only one participant succeeded in drawing a form that was considered sufficiently accurate, according to their specified criteria.

Their second experiment attempted to focus readers’ attention on the looptail ‘g’ by asking 16 different participants to read a passage and read aloud any word with a ‘g’ in it. They were then asked to write the letter ‘g’ they had seen. One reproduced a looptail ‘g’; eight drew an opentail ‘g’ (single-storey); and the others drew incorrect shapes which looked something like the looptail ‘g’.

I was not surprised by this inability to reproduce a familiar letter shape as I believe we cannot necessarily reproduce (in this case draw) what we can perceive. I was therefore incredibly pleased to find that they had done the study that would test this. The third experiment showed four versions of the looptail ‘g’ to a further group of participants, one being the correct version and the other three distractors; the same procedure was carried out with the double-storey ‘a’. They were asked to choose the correct letter shape. All those who were shown the letter ‘a’ chose the correct version whereas only seven out of 25 chose the correct ‘g’ which is chance level, equivalent to guessing. The majority (14) chose the version in Figure 1, perhaps because of the greater similarity to the single-storey ‘g’.

One of the versions of the letter ‘g’ used in the experiments

Figure 1: One of the versions of the letter ‘g’ used in the experiments.

I had expected readers would be able to discriminate among alternatives and choose the correct version. But those of us who are not type designers might find that the longer you stare at this incorrect form, the more plausible it becomes. The reason for perfect performance with the letter ‘a’ may be the implausibility of the alternative shapes and possibly greater or more perceptually salient deviations from the correct shape. Consequently, no comparison should be made between knowledge of the letter ‘a’ and the letter ‘g’. It would be worth repeating the letter ‘g’ part of this third experiment with more participants and different distractors.

The researchers’ discussion of the results is very thorough. There is an interesting discussion of the role of writing in learning letter shapes, but I find the exploration of the apparent lack of accurate letter-shape knowledge the most compelling. I favour the explanation that our knowledge of letter shapes does not need to be a complete representation when we recognise letters as part of the reading process. We need only learn visual features that distinguish one letter from another. This makes sense when we consider the variety of shapes across different fonts. It is possible that the difference between the same letter in different fonts is as great as the difference between the two forms of the letter ‘g’. A similar point is made by Walker and Reynolds (2003) and their study with children between 5 and 7 years old provides an interesting comparison. Unlike undergraduate students, most children in their study were aware of the different forms of both ‘a’ and ‘g’. The things we forget when we grow up!

Enjoyed the article?

Sign up for our newsletter and get notified when we publish the next one.


Walker, S., & Reynolds, L. (2003). Serifs, sans serifs and infant characters in children’s reading books. Information Design Journal, 11(3), 106–122.

More reading

See all articles