Module +Similarities


Hello everybody,
As non Latin type designer I was wondering where to start to Design Latin type face. When I found
Hrant Papazian Module + similarities it was a lot of time saver . I have made an extension to Hrant’s Module + Similarities called  Aleme’s tree.

I think it will be helpful to have this kind of tree for all scripts .
John Hudson whom I respect a lot who Designed (Nyala Ethiopic ) among many type faces ,has this to say in one of his presentation .  
“The first things are those that all type designers must know: the way in which shapes relate to one another; how their relative proportions and spacing contribute to the visual rhythm of text; how their weight, height and stroke modulation need to vary so that, optically, they appear balanced and aligned. “

What do you think ? What can be added or subtracted?
Thanks in advance, Aleme



  • Nina Stössinger
    Nina Stössinger Posts: 151
    edited May 2016
    So an “h” does not belong in the “arch” category?? I would expect that things would have to be able to belong to at *least* two categories at once.

    As an aside: It is the rule on TypeDrawers to have one’s username be constituted of one’s complete first *and* last name; thanks.
  • Aleme Tadesse
    This is not complete list . My idea is After the everyone chips in we will have a complete tree .
    Aleme Tadesse
  • Nina Stössinger
    Not sure, I would see this less as a case for (hierarchical) trees and more as a more complex network or something, where letters are not categorized in total but according to their (multiple) parts. But I’m happy to see where it goes.

    (And thanks for the name: If you could edit your username accordingly that would be great. Thanks!)
  • John Hudson
    John Hudson Posts: 3,013
    edited May 2016
    I doubt if there is a better way to visualise sharing of features than simply list of letters containing those features. Any diagrammatic approach is likely to end up being more visually complicated than what it is intended to display.

    The other problem with such schemas is that as soon as you have declared that a specific set of letters all share a particular feature or characteristic, I can come up with a design in which that is not the case.

    PS. Aleme, really nice to see you here.
  • Aleme Tadesse
    Hi John, I am very glad I am back . Will e-mail you  later .

    I agree additional diagram may complicate approch to start designing new or foreign writing system. I belive at the same time it some how save time . 

    for example if you need to design Georgian type face wouldn't be nice if there is handy  
    suggestions how the forms are related to each each other?  

    BTW I don't see Hrant here. Do you know where he is ?

  • Simon Cozens
    Simon Cozens Posts: 726
    I wonder if another problem with these taxonomies is that they encourage a cut-and-paste approach to design. "Extend the n to make a h." If it were that easy we could have automated most of the process by now. Just draw the letters.
  • Ray Larabie
    Ray Larabie Posts: 1,385
    When I'm dealing with an unfamiliar script, I'm looking for contentious relationships. Those relationships have practical implications which limit forms. For example: gy. You can't make a y without considering how it fits with g and vice versa. To someone unfamiliar with Latin alphabets, the never-used Qg relationship might seem important. While learning Greek, there were some letter combinations that looked like a train wreck, but never appear in use. And other combinations that weren't obviously a problem but occur frequently in Greek text. A chart that showed me these problematic relationships would have been very helpful.
  • Christian Thalmann
    Latin does need Qy and to support High Valyrian, though.  :grimace:
  • Denis Moyogo Jacquerye
    @Ray Larabie the “never-used” Qg may be used in Naro language where qg is a digraph, for example in “qgóé” when the initial letter is capitalized. Depending on the scope of your design you may have to consider those cases.
  • Hrant Հրանդ Փափազեան Papazian
    Double blast from the past! Aleme (great to "see" you again) and my diagram...

    Humans being visual creatures, diagrams generally help more than lists, but only if the intent warrants it. Unlike your tree, my original Modules+Similars diagram isn't about a progression; it tries to statically map two kinds of relationships (neither one of which can be effectively represented by a list) with the intent of helping people balance belonging versus divergence. In that way it's more in line with what Nina is thinking.

    That said a progression/tree is also worth pursuing, since it can be useful as well, just in a different way: helping people implement an efficient order of making their glyphs, especially in a script they don't have a native grasp of. In a relatively straight-forward writing system like Latin that might only save a little bit of time however; although I guess it can still bestow a certain confidence.

    But there is a danger, that Simon alluded to:
    Formulaïc design (for example leveraging the conventional broad-nib pen ductal "logic") gets to a marketable product faster, but makes you blind to the unexplored. Basically, blind to a big chunk of culture. Now, some people still manage to go exploring, but many get stuck in a comfortable rut and never recover.

    A related point: no diagram is gospel. Just because you could –fruitfully– deviate from it doesn't make it useless.

    Ray's point concerning linguistic frequencies is also important.


    BTW this is my first post to TypeDrawers. So Aleme, it's your fault.  :-P

  • PabloImpallari
    PabloImpallari Posts: 783
    edited June 2016
    Grouping letters by 1 single criteria is never enough, since there are many interconnected stuff going on. Alphabets are interconnected systems (Latin or any other script).

    For example:
    In the Devanagari version of the Testing Page, under the "Groups" tab, you can see some of the "groups by criteria":
    1) By Graphical Similarities
    2) By Position of the vertical bar
    3) By Root and derived letter
    4) By Knots and Loops

    Similar criteria groups can be made for Latin, Cyrillic, Greek, etc..