Is there a name for this aspect of written language?

Allographs?

That’s a related idea, @submariner. Thanks for that word.

I guess another way of putting my question would be, what is the name for what allographs have in common?—for example, the essential G-ness of a G or fiveness of a 5?

Haha, that reminds me of the time a mathematically gifted friend of mine was trying to figure out a puzzle that asked what B, C, D, G, J, O, P, Q, R, S and U had in common. He was converting the letters into numbers and trying various permutations and was stumped. I took one look at it and said, “They all have curved strokes.”

I guess you could define a letter in terms of the graphic features of the strokes used to write it, like the way that phonemes are described in terms of phonetic features, but I don’t know any technical terms for those features.

I’ll bet Steven Pinker would have a word to two to say about this. What a great question. Is Gestalt Psychology involved here?

Not for the specific features, @submariner, but for the essential form, generically: for the itness of it that you would find in Plato’s ideal realm. Let’s say the term is gazfap. We’d use it in a sentence like this:

The gazfap of the letter U is a curved enclosure that is open on top; of the letter C, an enclosure that is open on the right. (Note that the curved bottom of the U is necessary to distinguish it from a V, but a C can actually be recognized even if it’s made with square corners, as long as it’s open on the right—and doesn’t have a little tail that makes it a G.)

@jazzticity, is it? I don’t know. Do you think so? You might be right that Steven Pinker would have a word to say, but so, I expect, would typography designers.

Part of the answer can be covered by topology. Two things are topologically equivalent if one can be twisted to look like the other. C and G are topologically the same but are topologically distinct from A. This still does not explain the distinctness, for example, not only C and G, but also S, Z, M, U and N, which are all topologically equivalent.

@submariner: Ha! That was on my extra credit homework for my gifted school when I was a kid. I still like to pull that brain teaser out on my friends. :)

@Jeruba: I passed this along to a friend, she studied English Linguistics – maybe she can help. I’ll let you know!

This was such a great question I had to look it up. I believe it’s called ‘Geometric Moments’ and you can see it used in this paper, published in the National Institute of Health just this past September. They studied not only English letters, but also Chinese letters to better establish a universal standard of symbol recognition and visual acuity in literate adults.

Fascinating reading.

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2935333/

This is, indeed, a very curious question that has me thinking. This will likely be the Fluther question I carry with me through today!

I am not sure I can help, but I keep thinking about the term “glyph.” You couldn’t use it in the sentence above as “The glyphness of the letter U . . . .” loses something.

Phonemes and graphemes are related concepts and glyphs are the elements that comprise graphemes. Graphemes correspond to single phonemes, so the concept of graphemes has as much to do with spoken language as phonemes in most cases. At the level of the glyph (I am thinking of heiroglyphics) the glyph may not represent a phoneme but rather the written element itself (e.g., pictogram). This seems to get closer to the concept you are describing (an E can have many horizontal strokes, while an F may have only two to allow recognition).

Although not a graphology term per se, the word “construction” does substitute for “gazfap” in your sentence. I also like the word “typology” as mentioned above. I liken it, in my head, to topology and, as an avid reader, the notion of “the topology of the very letters of her correspondence drew me in and I traced her glyphs with my finger to assure myself of the ink on the page and of her affection.”

These are great comments. Thank you! I’m interested in anything more you care to add.

I must also note that I make no claims for my gazfaps of U and C above; I just tossed them out as examples. There’s probably a Boolean aspect to the definitions, and there certainly would have to be for computer recognition (which may be where the answer is to be found). For example, a U is still a U if it is square on the bottom, but not if it has any sort of downward-facing point.

For now I think I will let formal essence = gazfap in my mind, while still longing for a better term.

I think it’s interesting to observe that for some letters, the formal essence is all there is. You can embellish it, but there’s nothing you can take away. An I is a vertical stroke. An O is a complete or virtually complete enclosure. An H must have a cross stroke between two verticals. By contrast, as this logo strikingly illustrates, an A can be an A without the cross stroke.

This whole idea is for me a very old one, but I’ve never tried to articulate it before.

To me it’s part of the ever-fascinating question of pattern recognition, which I see as crucial to survival. It’s also related to things that I know have been studied, such as human recognition of words even when partially concealed and of context-dependent interpretation of graphic elements that might otherwise remain inscrutable—such as when piecing together ancient manuscripts (or just reading my brother’s handwriting).

But again, my question is not “What makes an N an N?” but “What do you call whatever it is that makes an N an N and not an M?”

“I think what this person is describing is a very specific PART of allography. Without its characteristics, a letter or word can’t be an allograph. There’s no word to describe it that I know of because the “essential form without which a letter can’t be a letter” tends to fall under the bigger umbrella of allography.”

—Says the girl with the degree in English Linguistics.

On a side note, the ability to recognize things outside of their normal appearance is fascinating to me. There’s been a lot of effort in computers/robotics to ‘teach’ this ability to machines. For example, we know these are turkeys but a computer viewing them wouldn’t be able to identify them all under that general heading without loads specific programming. Same with the KIA ‘A’.

There are two different cases to consider. One would be recognition of a letter in any orientation and the other would be specifically along a horizontal line. For example, upside down V used in KIA that you show works in the second case but not in the first.

I found this paper which seems to be germane.

Less helpful, but interesting is page 8 of this academic paper.

Neither source labels the science of alphabet recognition with a formal name, but I hope they are of some assistance or relevance.