What makes an ‘A’ an ‘a’?

How are we able to identify what an object is, just by looking at it? If you were to describe to a young child what the letter ‘A’ looks like, you might talk about it having three lines like a triangle, two of them leaning toward each other, etc. But do you think you could come up with a single description that explains why all of the shapes in the figure above are an uppercase ‘A’? Surely not—yet nonetheless, we can seemingly tell without effort that we are looking at a chart of the uppercase letter ‘A’, and not some other letter or shape. How do we do this?

Arguably the most straight-forward way would be for us to simply memorize each instance of the letter ‘A’ that we have seen, and “label” in our memory, so to speak, as examples of that letter. But consider all of the different ‘A’s we have seen—different fonts, different sizes, styles, different people’s handwriting—we would have to memorize thousands upon thousands of shapes for each letter! This burden on our memory would be inefficient to the point of absurdity, and this theory of letter recognition (called “template matching”) has largely become discounted.

Roman letter features
Fig. 2 taken from Fiset, D., Blais, C., Ethier-Majcher, C., Arguin, M., Bub, D. N., & Gosselin, F. (2008). Features for uppercase and lowercase letter identification. Psychol. Sci.

The best research on this question generally agrees with a theory called “feature matching”, which proposes that instead we memorize a (much smaller) set of features that describe the letter’s shape. This means we identify a letter by determining what its visual features are: lines, curves, intersections, etc., and comparing those features to lists in our memory. For example, when you read this letter ‘A’, visual processes in your brain determine that it is composed of two diagonal lines that meet in an L-intersection, a horizontal line that intersects them in two T-intersections, etc. The idea that our brains respond to such basic features is iwell-supported by scientific research (and arguably goes back to the cat experiments of Hubel & Weisel— check that out below!).

The upshot of feature matching is that a lot of the variability we see in fonts, size, etc., don’t matter—think of all of the ‘A’s in that figure that have in common those two diagonal lines and a horizontal line. Of course, some shapes that we consider to be the letter ‘A’ look quite different, and so it must be the case that we memorize a separate set of features to identify those ones—most obviously, the lowercase ‘a’ has very different features than the uppercase ‘A’ (two different shapes that share the same letter identity are called “allographs”; look out for Part 2 of this blog entry for more on that topic!).

The downside of feature matching is that it is not obvious what the features we actually use are—and so researchers remained challenged with determining this. One thing we do know is that expert readers, like you as you read this blog, pay attention to different features than do people who don’t have expertise (e.g., children, or second-language learners; see references). The most likely reason for this might be that experts have a better sense of what is or isn’t important—and when it comes to reading, what is most important is distinguishing between different letters. This means we become better are ignoring irrelevant features that don’t change the letter’s identity (think about some of the ‘A’s in the figure that have a lot of extra flourishes—you know that they’re not important, but someone just learning to read might not!).

Suggested reading: Hofstadter, D. (1995). On seeing A’s and As. Stanford Humanities Review, 4(2), 109–121. https://web.stanford.edu/group/SHR/4-2/text/hofstadter.html

References:
Wiley, R. W., Wilson, C., & Rapp, B. (2016). The Effects of Alphabet and Expertise on Letter Perception. Journal of Experimental Psychology: Human Perception and Performance, 42(8), 1186–1203. https://doi.org/10.1037/xhp0000213

Palmer, S. E. (1999). Vision science: Photons to phenomenology (Vol. 1). Cambridge, MA: MIT Press.

Gibson, E. J. (1969). Principles of perceptual learning and development. East Norwalk, CT: Appleton-Century-Crofts.

Fiset, D., Blais, C., Ethier-Majcher, C., Arguin, M., Bub, D. N., & Gosselin, F. (2008). Features for uppercase and lowercase letter identification. Psychol. Sci.

Courrieu, P., Farioli, F., & Grainger, J. (2004). Inverse discrimination time as a perceptual distance for alphabetic characters. Visual Cognition, 11(7), 901–919.

Is it true that we don’t look at every word when we’re reading? What are our eyes doing when we read?

Aoccdrnig to a rscheearch at Cmabrigde Uinervtisy, it deosn’t mttaer in waht oredr the ltteers in a wrod are, the olny iprmoetnt tihng is taht the frist and lsat ltteer be at the rghit pclae. The rset can be a toatl mses and you can sitll raed it wouthit porbelm. Tihs is bcuseae the huamn mnid deos not raed ervey lteter by istlef, but the wrod as a wlohe.”

 

You’ve probably encountered that paragraph before (or, if you’re like me, a bunch of times—it seems to go viral every other year!). How true is this claim that “the human mind does not read every letter by itself, but the word as a whole”? The first part has some truth to it—we do not focus on each letter in a word as we read—but the second part is very misleading! It is a pernicious myth that we learn to read by memorizing words as a whole shape. And it’s easy enough to come up with examples where jumbling letters in this way is a real problem—calm becomes clam, blow becomes bowl, etc. So, what do we actually need to look at, when we’re reading?

 

As you’re reading this sentence, you might feel that your eyes are moving smoothly across it. In fact, when we read text, whether its on a printed page or a computer screen, our eyes more in a series of short jumps, called saccades. These saccades are very fast, around 20-35 milliseconds, and in between them our eyes fixate on the text. These fixations can be brief (150 milliseconds), or relatively long, say one half of one second.

 

So, what is it that we look at during these periods of fixation? It is true that we do not focus on every single word when we’re reading—this is more or less for two reasons. First, we’re able to perceive several letters within the fovea (the center of our gaze): in languages like English, which are written from left to right, we can see a few letters to the left of our fixation and maybe 12-15 to the right (in languages written from right to left, like Arabic and Hebrew, readers can perceive more letters toward the left of fixation that the right!). This means that during each fixation, we take in a few words at a time, unless there are very long words. When we saccade to our next fixation, we are able to skip over some words because we actually have already seen them. This means, of course, that one of the challenges of reading is remembering the words and letters you have recently seen (in working memory) and integrating them with new information, as you continue to saccade through the sentence.

 
visible_light_eye-tracking_algorithm

The second reason we do not need to fixate on every word is because we are often able to predict what words are going to follow—and we can use this ability to predict to speed our reading. This is often true of function words (words like “to”, “the”, and “do”), but also in sentences where the context leads to a very high probability for a certain word. Imagine that in one fixation you read “They sang Happy…”—you can guess that almost definitely the next word is “Birthday” (in fact, when we read sentences where we expect one word and it ends up being another, this surprise has consequences—it will cause us to slow down dramatically in our reading speed and often to double back and re-read!).

 

How do we know these things about reading? Mostly through the use of a machine called an eye tracker, which allows us to know (with very high temporal precision) where someone is looking. There are many videos online where you can see demonstrations of an eye tracker at work. This one in particular “How we read shown through eye tracking”) shows how we move our eyes from one line of text to the next—and how this is affected by the way that the lines are (or are not) justified!

 

Links:

“Eye movements in skilled readers”

“What eye movements during reading reveal about processing speed”

“How we read shown through eye tracking”