Shapes of Letters

Is there any pattern to the letters across written languages, asks Stanislas Dehaene in Reading in the Brain. At first sight, it seems like an idiotic question: the Roman, Arabic, Devanagari and Chinese scripts hardly resemble each other in any way. But look at things a little differently, and a pattern emerges.

 

Perhaps, one hypothesis goes, certain letter-like shapes are omnipresent in what we see around ourselves. For example:

The frequency of these shapes is even higher if rotated shapes are counted as the same shape:

If you felt that rotating the shapes and still thinking of them as the same letter is cheating, think again: our brain adjusts for orientations – how else would it recognize the same object again, when it is viewed from a different angle?

 

Next, consider these drawings, where the same shape is drawn with varying degrees of completeness:

 

Notice how the middle column is almost interpretable whereas the first column isn’t? Why is that? The difference between the two is that the second column includes the junctions, the meeting point of two surfaces.

 

Now look back at the first picture at the top. The F, T and Y shapes are junction shapes – perhaps this is why our brains evolved to recognize those letter-like shapes, says the hypothesis.

 

But, you counter, the F, T and Y shapes are Roman characters. We don’t find them in the other scripts. Doesn’t that prove this hypothesis can’t be a universal reason for character shapes?

 

Marc Changizi answered that question. Two lines can intersect to form only 3 letter shapes: L, T and X. Three lines can intersect in far more ways: F, K, Y, Δ etc. He then looked for the frequency of these shapes in various scripts… while ignoring their orientation, i.e., a tilted (or mirror-image) F still counted as an F. Guess what he found? Across all scripts, the L and T shapes occurred the most, X and F were next, and so on. These patterns were the same across scripts.

 

In fact, the frequency of occurrences of these shapes in the scripts matches the frequency of occurrence of these shapes in natural objects. This then is why many believe the following:

“Our primate brain only accepts a limited set of written shapes.”

And:

“We did not invent most of our letter shapes: they lay dormant in our brains for millions of years, and were rediscovered when our species invented writing the alphabet.”

 

It sounds like an interesting hypothesis.