Fascinating... I’m looking forward to seeing that article. Of course, I imagine there’s probably a strong correlation between those shapes that are easiest to perceive and those that are easiest to form. I’m really curious how they went about collecting their data...

—Jim K.

The paper is available on Prof. Changizi’s website.
http://www.changizi.com/

They take an interesting tact and analyze letters as orientation invarient topologies. This means that there is no difference between L, v, and 7. They find that the letters in different orthographies tend to use the pattern of topologies. They go on to find that these same topologies correlate well with those seen in natural scenes, but do not correlate well with the number of hand movements needed to create the topology.

For this to be useful for type design, the work will need to be extended past topology to looking at geometric differences within each of their topologies. They do note this in the paper as a possible future direction.

Cheers, Kevin

They take an interesting tact and analyze letters as orientation invarient topologies. This means that there is no difference between L, v, and 7.

Or, say, between an A and an upsidedown A. I’m not sure whether that makes me or Hrant the orientation invarient topologist. :)

these same topologies correlate well with those seen in natural scenes

Nick,
Instead of Russia, That photo looks like it could have been taken in Pittsburgh in the 60s.

ChrisL

“Learning to read must involve a pre-emption of part of the existing visual system and it’s conversion, by minimal modification, to process the shapes of letters and words.” [Stanislas Dehaene, Laurent Cohen, Mariano Sigman and Fabian Vinckier: “The neural code for written words: a proposal” (Trends in Cognitive Sciences, Vol 9 No 7, July 2005)

So the Changizi hypothesis is not far fetched.

Dehaene, et. al., add: “Indeed, in most expert readers, a localized region of the left occipito-temporal sulcus, just lateral to the fusiform gyrus, systematically takes on the function of identifying visual letter strings. The ’visual word form system’ is thought to play a pivotal role in informing other temporal, parietal and frontal areas of the identity of the letter string, for the purpose of both semantic access and phonological retrieval. Thus it must provide a compact ’neural code’ for visual words, a cell assembly unique to each word, yet invariant under changes in location, size, case* and font.”

In “Neuroimaging Studies of Reading Development and Reading Disability” Kenneth R. Pugh, et. al. (W. Einar Mencl, Annette R. Jenner, Jun Ren Lee, Leonard Katz, Stephen J. Frost, SSally E. Shaywita, Bennett A. Shaywitz), posit “a word-recognition system in which processing is fast because linguistic knowledge is encapsulated in highly efficient pattern-recognizing templates.” (Learning Disabilities Research & Practice, 16(4), 240-249.

I’ve tried to argue for neural encoding of whole words at the ’pattern of features’ level. It seems to me that, in addition to direct encapsulation of ’linguistic knowledge’ in these ’templates’ or ’neural codes’, there are—yoked to these complexes—’chirographic scripts’, which form the basis of our knowledge of orthgraphic shape.

In other words: while letter topologies do not perhaps correlate well with the number of hand movements needed to create the topology, our sensory-motor familiarity with the concrete movements we make in writing senquences of letters underlies our abstract knowledge of orthgraphic codes (and how they are most effectively learned). When we get stuck in spelling we imagine writing the word.

* debatable

>When we get stuck in spelling we imagine writing the word.

It is striking to a non-Chinese that when Chinese meet and give their names, and want to know the character—it gives a meaning, and there may be many for each sound—they draw it in the air with one finger against the palm of other hand. Then the person observing will instantly recognize the character, though it is invisible.

I just tried it with roman characters and it works for me also.

Interesting! (I should have said ’whole-word chirographic scripts’)

> our sensory-motor familiarity with the concrete
> movements we make in writing senquences of letters ...

Just one problem with that theory: we don’t write any more.

> When we get stuck in spelling we imagine writing the word.

You don’t know that to be generally true. I can assure you that I for one don’t do that. I visualize the letterforms. Or type them out! This, even though I did write a lot as a child. And one can only hope that the kids of the future will avoid such pointless intermediate steps entirely:
http://news.bbc.co.uk/2/hi/uk_news/education/3103421.stm

hhp