Studying eye tracking helps us present content for learners
Understanding how people read and absorb information is pretty important to anybody who wants to get a message across using text. Fortunately, research using eye-tracking equipment has given us useful information about this, particularly when it concerns reading from a screen. What follows is a very brief summary of some of the areas the eye-trackers have been concentrating on in recent years.
The idea of observing the readers’ eyes as they read is not new. In 1908, Edmund Huey was already experimenting by attaching a thin aluminium pointer to the cornea (it’s OK, subjects were given cocaine to dull the pain!) which scratched a pattern on a revolving drum when the eye moved.
Modern experimental setups are much more sophisticated. An eye-reader usually consists of one or more cameras attached to the screen. Eye movement is calibrated using software in a few seconds, and much of the processing is automated from then on.
Thanks to numerous studies, we know that our eyes take a surprisingly irregular path across any substantial text. This has implications for anybody writing and presenting information.
Readers skim text
As we read a text, our eyes don’t follow letter after letter, but do a series of leaps along the line, called saccades. Each landing is called a fixation, and we skip around a third of the words in a text. The idea that readers skim text like this is clear, and numerous commentators (particularly Jakob Nielson) have repeatedly emphasised what this means for content creators.
On average, only 2% of nine-letter words are skipped. But very short words like “the” and “of” are skipped 50% of the time. Curiously, an experiment with German compound nouns split into their shorter component parts helped readers speed up their reading despite the unfamiliarity of the fragments. (Compound words combine several shorter words and fragments – like überschallgeschwindigkeit, meaning supersonic).
What happens during a saccade isn’t always clear, but we do know that the parafovea – the large portion of our vision that isn’t the subject of our gaze – still feeds us plenty of information. This makes perfect sense: we easily notice if somebody comes into the room even when we’re engrossed in a book, so it’s likely that we’re doing some processing of words we only see out of the corner of our eyes.
The process of deciding how far to leap along a line isn’t yet fully understood. We quite often make mistakes about where to land, and our eyes sometimes have to sweep back to the left to get more information (between 10% and 15% of saccades consist of these regressions). How the brain makes these decisions is now enjoying a great deal of research, and the only safe conclusion for now is “it depends”.
Unusual letter combinations stop us in our tracks
Not surprisingly, spelling mistakes attract our attention. They only grab us for a few hundred milliseconds, but that’s enough to interrupt the flow. In the sentence “The doctor told Fred that his drinking would damage his liver very quickly”, the word liver tends to be skipped because it’s predictable. Accidentally use “livor”, and the eye will home in on it like the proverbial sore thumb.
As readers, we’re also interrupted by unusual letter combinations. Studies have shown that rare combinations like “pne” used in words like pneumatic and pneumonia attract the eye until they can be processed.
Word order matters
Processing is slowed down if something unexpected happens in the sentence. Many sentences prime us to expect certain words or ideas, and if they don’t appear we instinctively start checking more closely. So “John admired Mary because she was a very smart woman” is a smoother ride than “John admired Mary because he liked very smart women”.
Serif or sans serif? Choose either!
Perhaps surprisingly to the modernists who favour sans serif faces, numerous studies have found little difference when comparing them with serif fonts. IBM and Google researchers found no significant difference in reading speed when they asked dozens of readers to plough through a text in Helvetica (sans serif) and dozens more to read Georgia (serif). The same study also found that varying the size through 10 point, 12 point and 14 point resulted in little variation.