Interdiacritical Keyboard Layout

Grzegorz Rolek's picture

With multiple languages at work, input of diacritical letters by keyboard happens to be quite problematic. Dozens of quirky key combinations of your native layout, still giving you not even a fraction of all the diacritical variety, or switching from language to language for specific input, each defined without single, predictable manner.

Trying to deal with this mess, I once made a keyboard with a single and intuitive method for possibly all diacritical letters. The method is simply the Option key + the base letter, followed by typing diacritic's name. Therefore pressing, say, Opt + e, followed by typing acute spits an é. There's also no need to type the entire name, since the keyboard estimates it and suggests early available output. You can get the layout and view more info at

Typographers, linguists and editors — all the people who are dealing with diacritics in their daily work — would, I think, find the keyboard useful. Especially touch-typists used to type with uninterrupted flow, what the method inherently allows. But I'd rather hear what do you guys think.

twardoch's picture

Interesting idea! Since you haven't mentioned this, I'll add that this is for Mac OS X.

Kristians Sics's picture

Really good idea. It should help a lot. Thank you!

Birdseeding's picture

The Swedish standard keyboard has five combinatory diacritics - acute and grave accents, circumflex, tilde and umlaut - mapped out on two buttons of the keyboard. You press the button (or the button and shift/button and alt gr, yes mine is a pc) and then the letter, getting you a good portion of the needed glyphs. Wouldn't a system like that but with a greatly extended range of diacritics be more efficient?

Grzegorz Rolek's picture

But system you're describing, Johan — and, in fact, system of all those standard keyboards — is exactly what I was trying to avoid, as it forces you to be aware of a bunch of things:
• What modifier keys are required? Shift? Alt/Opt? Both?
• What keys trigger what diacritics? Apostrophe gives acute or grave? Colon as umlaut? What about others, less intuitive ones? And, hey, it's still punctuation, not diacritics!
• What is the actual combination? In other words, what modifiers in what combination with what other keys give what diacritics? You see the problem.

Early on, I've tried a system based simply on Option key plus a key of diacritic's initial letter, followed by the base letter. But it just couldn't work with caron, cedilla, circumflex and so on. I'd have to make some hacks, therefore no solution either.

What makes the Interdiacritical method effective is its ridiculous simplicity, intuitiveness and extensibility: with a single Opt key you can enter basically all imaginable diacritics above all Latin letters, knowing only the diacritic's name. And with multiple languages to deal with efficiency is not a priority — that's what language-specific keyboards are for.

dberlow's picture

This is a great idea! Out of curiosity, are the names of diacritics the same in all languages, and are there any diacritics with accents in their names?

Theunis de Jong's picture

David, were you thinking of háček? The name itself contains a haček, so trying to enter its name leads to an infinite loop :-)

Unless you settle for its alternative, caron.

Synthview's picture

I feel comfortable with standard Mac method, using the US-international keyboard layout (it covers the western latin languages). I feel it's the fastest method: (cmd+key) + your letter. Of course you have to remember the key for each accent, but they're not so many.
However I have to point some incoherences as, for instance, "å" letter is the direct result of cmd+a, while "à" is cmd+` + a

To access the complete latin accents, and have a more consistent generation system [(cmd+key) + key] you can switch to US-Extended.
for custom tuning, you can modify your keyboard layout with the Ukulele App.

Queneau's picture

I work mostly in three different languages (dutch, german english) on a dutch keyboard, which works fine most of the time. The only difficulty I have is when I write long german texts, as the accents and characters commonly used (ä,ü,ö,ß) are not, as on a german keyboard, directly accessible, but only through the pattern alt+u+a/o/u and alt+s. This becomes even more of a hassle when using it on capitalized letters, which are common in german because every noun is capitalized. So typing Ä requires the combination alt+u, shift+a. This can become a little annoying over time...

And I guess languages that use more diacritics, and more complicated ones (like stacked diacritics) can be even more difficult to write. So I guess there definately is a need to simplify this, but I'm not sure this will work over longer texts as it still requires quite a lot of extra typing. So when using multiple languages at the same time, I would prefer programming new key combinations with only single keys (like the ones mentioned above). When only using the diacritic in a few words I probably would go through the glyphs panel.

Igor Freiberger's picture

I use Windows and expanded my keyboard adding AltGr + key combinations for most letters with diacritics. This could be done with a free keyboard editor from Microsoft.

One can add characters to AltGr + A-9 and Shift + AltGr + A-9. Thus, the possibilities are quite wide (49 x 2 in a Portuguese keyboard). The only problem are conflicts with keyboard shortcuts in some programs, as InDesign or Photoshop, so it's up to the user to select keys for one and other functions.

I don't know if this is possible in Mac. As there is no AltGr key, Grzegorz solution may be the better choice in this platform.

What about to use stylistic sets to achieve this? Say, in a ssXX feature one can add the same substitutions set on ccmp feature, but then using combinations like letter + diacritic abbreviation + trigger. Something like a/ac= subs by aacute (this is just a sample, not the correct syntax). Is this a very unorthodox idea?

Grzegorz Rolek's picture

David, Theunis, that was a good one. In fact, just to save you a hassle, I went into various names, as there's no limit for that. You just type the name you're familiar with and the diacritic pops out. For now they're mostly Unicode names and some of their widely recognizable equivalents. Still flattened to plain Latin, though.

Andreas Stötzner's picture

For the Windows party also the Europa-Keyboard may be worth considering. There is obviously not yet an English (or other) version of the page, but (the developer Karl Pentzlin is a colleague of mine) there might be ways of approaching him.

We are still sticking with a keyboard invented about 150 years ago and lacking the slightest innovation ever since. I wonder why. It seems to be the only component of our machines the developers deny any prospect of progress for. I still don’t understand.
As well as I don’t understand why the entirety of the academic world is still so lazy in forwarding suggestions to get rid of that old-fashioned dinosaur-keyboard. Myriads of scientists let themselve get tortured by the limitations of the average *typewriter* for decades now …

I do hope, in the age of Unicode, multiscriptive typefaces and heavily increased worldwide text interchange they will wake up soon and re-think the whole thing, right from the ground.

Sorry for the radicalism.

Queneau's picture

You are right on the money, Andreas. Making new keyboards might actually make working and researching quite a lot more effective!

dberlow's picture

>... invented about 150 years ago and lacking the slightest innovation ever since.

That is a bit extreme. It's just the keys, not the board.

>You just type the name you're familiar with

that is nice, I can go back to the names I first used, "pointy hat thing", "eyes", ":)"

uwe's picture

I've experimented with various input methods some years ago. For longer texts anything beyond two keystrokes per accented letter gets tedious. It's not only the number of keystrokes that grows but also the typing time per keystroke. It seems that one can do at most two keystrokes unconsciously — if the keystroke sequence is longer, the brain starts task switching: one stops thinking about the text, one starts thinking about the required key sequence, one finishes the key sequence, and finally continues thinking about the text.

Unfortunately, accessing every accented letter with two keystrokes does not work (unless one uses a keyboard with thirty dead keys). But one can access the most frequent ones with two keystrokes, and that's what counts in practice. Here's the recipe: Take one key as a reserved <Accent> key (I prefer <Tab>, but that's a matter of choice). Typing <Accent> replaces the letter before the cursor by another one: a becomes ä, A becomes Ä, e becomes é, s becomes ß. To get other letters with the same base letter, type <Accent> repeatedly. Every press of the <Accent> key yields one more replacement: ä becomes à, à becomes á, á becomes å, and so on.

As a result, I can get the frequent letters (in a given language) with keystroke sequences that are both mnemonic and short (the average length of the keystroke sequences is very close to 2). For rare letters occurring in some foreign proper name, I may have to type <Accent> ten times, but for letters that I need only once per month, that doesn't matter. The only thing I have to remember is the rule "type the base letter and then type <Accent> sufficiently often until the required letter appears". The method also has the advantage that one can type in the natural (handwriting) order: first the base letter, then the accent. And in contrast to dead keys, one needs only one reserved <Accent> key. (Conveniently located keys are a scarce good!)

I've checked the orthographies of 28 European languages. For 21 of them, everything works smoothly, for the remaining 7, it works with some tweaking. (Here's the full story.) Still there are some limitations:

The method clearly does not work if the number of accented letters per base letter is too large (such as the 17 variants of o in Vietnamese or the 23 variant of alpha in polytonic Greek). Even five derived letters (e.g. in Yoruba) may be too much.
If one switches between different languages, one has to switch the replacement scheme. Typing French text with a replacement scheme for German is a sure way to give you RSI.
Unlike dead key or AltGr techniques, the method requires support by the application. It's an input method, not a keyboard. (In fact, I have only implemented it as an editor macro.)

hrant's picture

I will delve into everything in this thread more deeply, but in the meantime I will point out that a central thing here is portability: since this can't be about standardizing a particular novel keyboard input method*, having a way of taking it with you -without imposing on a host system- is critical. How to do that? Maybe the best solution is cloud-based.

* If Dvorak didn't work...


JamesT's picture

What would be great would be the addition of diacritic keys. For example, if I wanted to type å or ư, for example, I would only need to press "a" and the "ring" key, or "u" and the "horn" key. These could perhaps replace the dual availability of the numerals on the standard keyboard.

Grzegorz Rolek's picture

First, I must say that Uwe just nailed the subject with his research and all this makes my little method rather weak, basically. Now portability is another problem entirely. I’m not really a fan of syncing everything everywhere mantra, nevertheless this would imply a system-level support, as any third party solution would have to be imposed on the host system anyway.

James, isn’t this is how OS X Lion somehow deals with it? Hold-pressing a base letter gives you a little pop-up with a row of few related diacritics, all with associated numeral as a shortcut. By the way, this row is also tab-enumerable, what quite closely resembles Uwe’s very own idea.

Syndicate content Syndicate content