About

Each keyboard layout claims it is the best because they each use a different metric. Colemak gained a boost in fame when it was declared to be the best in the Carpalx study. Workman claimed to improve on Colemak because it modified the key cap scores used in the Carpalx study by adding a subjective gauge of lateral movement and finger curling preferences to the cap scores. David Piepgrass likened the Asset layout to Colemak, but by declaring Asset to be closer to QWERTY, said it would be better than Colemak. It's also interesting to note that Colemak also advertises similarity to QWERTY.

Norman vs. "the others", the history

Every layout has its strengths. The author of Workman is confident in his key cap scoring model, which if applied to U.S. English yields lower effort scores than any of its competition, however the full optimization of Workman compromises the locations for copy/paste, putting them in less efficient locations for frequent copy operations. Accepting Workman as "fully optimized" also means agreeing with the scores used in the Workman design.

Colemak and Asset are similar layouts, but in the design of Asset, the same-finger assignment from QWERTY is raised as a higher concern in the design. Though Colemak is touted for its supposed ease of switching from QWERTY, Asset keeps 20/26 letters on the same finger, whereas Colemak only keeps 15/26. As the Carpalx study shows, the theoretical loss of efficiency by keeping the QWERTY same-finger design is fractional, perhaps inconsequential.

Blue arrows are keys that stayed with the same QWERTY finger in Norman. Green are keys on the same hand, but different fingers, and yellow keys changed to a different hand.

The drastic movement of keys in the Dvorak layout is the main reason Colemak's FAQ cites for Dvorak adoption failure. Efforts to achieve "full optimization" by moving keys all around the keyboard like in QGMLWY, away from their originating QWERTY hand and finger, are also destined to increase the likelihood of adoption failure.

When switching from QWERTY, as a user who types enough to be concerned enough to switch, the loss of productivity in typing speed and in gained frustration is cause to minimize not only the general movement of key caps, but the finite movement to a different hand or finger on the same hand. Maintaining common shortcuts is essential to adoption. The better job you can do at minimizing frustration and productivity loss, the more likely a conversion away from QWERTY will be successful.

What’s remarkable is that with only 12 keys you get most of the improvement of Dvorak at nowhere near the expense of its learning curve. This shows that you don’t need to throw out QWERTY, you just need to fix it.

The importance of common shortcuts and historic link to QWERTY is evident when comparing the overall success of adoption between Arensito, introduced 2001, and Colemak, introduced 2006. Arensito has essentially no resemblance to QWERTY. Colemak has been adopted in the main keyboard options during operating system installs for Apple, Linux, Chromium OS, NetBSD, and FreeBSD. It also has add-on options for just about any other operating system. Arensito is available as an add-on for Windows only.

When trying to improve on any of the available layout alternatives to QWERTY, all the authors of Asset, Colemak, Workman, Arensito, Dvorak, Capewell-Dvorak, Klausler, and Carpalx variants should be considered as they each have slightly different goals and research findings.

I tried for existing layouts after researching them as possible candidates. Here are some of my findings from studying, and trying some of the aforementioned:

Asset

  • Asset is so great that six years later, its creator doesn't even use it.
  • I think its emphasis on QWERTY similarity has validity based on my experience trying to train to a key shifted to a different finger on the opposite hand.
  • When I started designing my own keyboard layout, I independently started with the same 8 keys on the home row as are used with Asset.

Colemak

  • Colemak gained popularity to a large degree by advertising similarity to QWERTY.
  • The community around Colemak has been very successful in getting it added in feature positions for popular operating systems.
  • As OJ Bucao observed when designing Workman, the Colemak layout has a high degree of lateral finger movement to the center column for D and H. I believe the center column should be de-prioritized and not considered part of the "home row."
  • Included in freedesktop.org's xkeyboard-config.

Workman

  • OJ Bucao is a big fan of the TypeMatrix keyboard, a non-standard, grid style keyboard. When launching www.workmanlayout.com, he even created the feature layout image using the grid instead of a standard staggered design. This likely impacted the design of the staggered scores.
  • Workman's model is designed around a score for each key cap.
  • The scoring model treats the left hand as equal to the right.
  • The keyboard shortcuts for copy/paste are shifted to the right unnecessarily.
  • Included in freedesktop.org's xkeyboard-config.

Dvorak

  • Dvorak is a huge break away from QWERTY. There is a high degree of frustration and a long period of productivity loss in converting to it.
  • Common shortcut keys are shifted around the keyboard and therefore should disqualify it from consideration by anyone if for no other reason, Colemak's wide availability.
  • Included in freedesktop.org's xkeyboard-config.

Parameters for Norman

I am right handed. My right hand generally does everything better and faster than my left. My fingers are doing the work, not the key caps. Fingers should be considered, scored by ability and assigned a load, not the keyboard.

My pinkies are slow. On a keyboard, I don't press with my pinky finger independently so much as I hold it stiff and press my entire hand in the direction of my pinky. I use my right ring finger for the delete key.

When I have tried learning alternative keyboard layouts, I find myself using the same finger on the wrong hand (i.e. typing E with my left middle finger in the Workman layout). I think when users are considering alternative keyboard layouts, they've already become proficient with QWERTY. There are deep brain connections after several years of QWERTY typing. Learning a new keyboard layout that shifts keys to different fingers or to the opposite hand as QWERTY compounds the learning curve and increases the chances for conversion failure.

Undo/copy/cut/paste, save, select all, and quit are common shortcuts, which belong on the left hand and are already in optimal locations. They should remain together and on the left hand so copy/paste operations can be performed while mousing with the right hand.

Software developers use common shortcuts, which are universal across applications and operating systems. Undo, copy, cut, paste, select all, and quit should be one of the first things a person uses when using a computer. In fact, they're so important, that just about sums up the only shortcuts to have their own implementations in mobile operating systems like iOS and Android.

Layout design exclusions

Changing the keyboard layout is not a complete solution to preventing repetitive strain injury and carpal tunnel syndrome. There are additions you can make to your injury prevention plan, however I posit the following, which I list as exclusions for consideration in the design of a keyboard layout.

  • Grid layouts like TypeMatrix are not likely to be adopted widely by computer manufacturers. TypeMatrix is as expensive as some fully-functional tablet computers. I used to own their 2020 model but sold it before they came out with the 2030. I think their 2030 model does fix some of their earlier design flaws and would be excited for them to implement Norman in some way. I simply didn't consider them mainstream enough to weight in the design of the Norman layout.
  • Contoured keyboards like the Kinesis are even less likely to be adopted widely.
  • Keyboards with cords are dinosaurs.
  • Wireless keyboards like the FrogPad with multiple parts still require significant effort to learn new typing formats.

Goals for Norman

When setting the goals for the Norman layout, there was really only one primary goal:

Find a balance between typing optimization and conversion difficulty.

Super typing speed is not a goal of Norman, but it is a side-effect. Reduced effort is a much higher goal, both in effort to convert from QWERTY and in finger movement. In fact, applying some thought and modern analysis tools to the QWERTY layout can yield speed improvements by simply moving the most common 6 characters to the home row.

It just so happens that the Norman layout almost beats out Workman's efficiency scores in US English using Workman's own scoring model. That said, I do not agree with the Workman scoring model because it rates key caps instead of fingers. I do not think the left hand should be rated equally to the right, however the design of Norman doesn't depend on key cap scores.

When designing Norman, the first task was to find the most frequently used characters in US English that I am likely to type. I am not likely to type any large volume of Alice in Wonderland, as is so commonly used as sample input to analyses. I took a sample of my own writing, 5,755 words using 33,319 characters to decide which characters I type commonly. The Asset layout did a similar study and there are websites like letterfrequency.org that list the outcomes of more generic frequency studies. My frequency outcome was similar to other frequency studies:

etoainsr ldhu cmpgwfybvkxjqz

The spacing was added for emphasis. The Norman layout places each of the most common 8 letters somewhere in the proper home row. Their exact location in the home row is guided by their QWERTY finger. Only the letters T and R were in contention for shifting to an alternate finger. The lesser frequency of R guided it to the weak pinky finger and results in the first of 4 letters to move to a different finger than would have been used in QWERTY. This meant adopting the model of several other layouts like Asset, Colemak, and Workman, to shift semicolon out of the home row and to the even weaker pinky position.

Given my right hand preference, and the guidance of QWERTY fingers, L, U, and H were all able to take prime positions on my right hand in the top row, on strong fingers, and maintain their QWERTY finger. Letter D was able to maintain its QWERTY finger and keep a strong finger on my left hand.

From this point forward, the remaining list of characters is as follows in order of frequency:

mpgwfybkj

Note, the common shortcut keys are not considered in the list. They are fixed in the layout. Regardless of their frequency, their added efficiency by saving typing with their operations, or by not introducing additional finger stretching disqualifies them from moving.

Considering the next letter M, its QWERTY position was already in a location for favoring my right hand on a strong finger. Since P had to shift to make room for the semicolon, I also assigned P to my strong right hand finger.

The process for the rest of the letters follows the same process. Evaluate the letter's QWERTY finger and decide the optimal location for it to be pressed, avoiding the center column as much as possible.

Comments

So after a few hours (in succession) from researching ergonomic accessories such as keyboards to layouts, what would you suggest for someone who isn't even that great with QWERTY and doesn't know any other alternatives. My WPM is about 32 from "rate my typing" and I'm a non-touch typer. One layout you didn't mention was the Maltron layout which focus more on the shape of the keyboard.
I guess the ultimate question would be: "What is the best English language keyboard starting from when a child first uses a computer?"

I think you need to start with a keyboard layout that's built-in on most operating systems and only if you know how to enable it on a computer when you move to a different one. That leaves the choices at QWERTY, Dvorak, and Colemak, I think. My kids are going to start with QWERTY, then I will make them aware of Colemak and Norman. If they want to diverge from those, then we can have a debate about changing Norman again or do their own thing.

I designed Norman with practicality in mind. Using custom-molded keyboards is a specialty and an extreme I'm not attempting to consider. There are compromises of practicality that need to be made when diverging from QWERTY, which is why ZXCVASQ were not keys I considered to move around in the Norman layout. Taking my Macbook to a park to work and hauling a Kinesis keyboard a mile down a trail with me isn't a reasonable thing to do. I even think it's a legitimate argument to say you'd only consider QWERTY, Dvorak, and Colemak simply because that's all that mobile devices like iOS and Swiftkey support.

Thanks for the reply!
At the moment only QWERTY and Dvorak are standardized, I had to download the program for Colemak but there's a portable version for USB which is nice. I guess your kids got no choice because the school is using QWERTY?

Fair point about the custom molded keyboards and it's impracticality. It is true that Maltron keyboards are very expensive. I'd imagine to buy one would mean conversion for life due to the investment haha

I'm not really an enthusiast but I tend to over-research things even if it's the tiniest interest and all this is very interesting to me :)

Hey!
I just wanted to add to those who want to try dvorak or colemak, cause this website helped me a lot - http://www.typingstudy.com/en-us_dvorak-3/ http://www.typingstudy.com/en-us_colemak-3/
Maybe you should write the developers to think of developing a typing tutor for Norman Layout as well? :)

I can type in QWERTY pretty fast. My speed is around 90-100 wpm. I was looking for an alternative layout to reduce fatigue. I like Workman's design philosophy and Norman's even more. So I decided to use my own slight variation of Norman to better adaptation to my hand anatomy, that is I grow nails on my right hand fingers because I play the classical guitar. I can't curl my right hand fingers to rest on the home row as comfortably as the left hand ones due to the nails. My home position for the right hand fingers are JIO; on QWERTY, so I decided to swap out the O and I keys on the Norman layout for R and L. This way I can rest on the N I O and H keys, while maintain the familiarity of the O, I and L key positions compared to QWERTY. I also use the right middle finger to type the U key. The left hand key arrangement is still the same as Norman. I jokingly call it "Nailman". The keyboard analyzer website says the layout is slightly better than the original Norman considering my unorthodox home positions. So far so good.

It took me only under 1 hour to remember the position of all the keys. I feel great comfort, so much better than QWERTY. After 2 days of intensive training now I can type at 55 wpm.

And here come the problems. The layout makes typing the frequent letters/common words so easy, but also makes the hard words harder. The difference between the efforts needed to type easy and hard words is larger, so it's like hitting the bumps on the road when driving. I want to talk specifically about the keys K and J, which are the hardest ones to reach. The B key also has 5 points like the J key, but actually reaching for the B key using lateral movement of the index finger and the wrist is much easier and I never experienced any problem with it. The K key requires a hard upward reach by the left index finger if you don't move the whole hand and arm up. It makes frequent combinations like "ke" and "ck" really awkward, considering many common words contain them: make, take, worked, talked, back, pack, suck, etc. The combinations of J with any vowels other than A and E, which are located on the left side, are hard to type: en-joy, joi-n, ju-ne, ju-ly, ad-jou-rn, etc. There aren't many words with the letter J, but whenever you encounter them it's a pain to type compared to other words, especially when learning. It's worth to mention that typing these words is already easier on my variation, because the J and those vowel keys are on the same row. I never noticed the problem with the QWERTY layout because words are somewhat more equal in terms of effort needed to type them.

There is another problem, that is I mistype a lot the D/E pair because they switch places compared to QWERTY. I don't consider it's an inherent problem of the layout though. I suggest moving the J and K keys somewhere else.

Thanks for taking input seriously. Ive used Dvorak primarily for 20+ years and can touchetype QWERTY as well. Having both is handy. I did not find it tedious to switch as I was suffering from RSI with QWERTY.

To learn Dvorak all I did was touch-type my ABC's over and over for 2 weeks. I don't believe Dvorak made me faster but it certainly relieved my suffering.

Good luck with the layout. I encourage any effort to get input right.

change of subject: This took a full 5 minutes for me to write on my phone (Note 2). I could have typed it in less than 45 seconds on Dvorak. I have tried NUMEROUS keyboards on Android. still going slow. Maybe you can fix that? :)

If your only goal is to win speed typing competitions, then favoring the already dominant hand makes sense. But I'd rather use both hands equally, to train my (non-dominant) left hand to be more useful and since my right hand is already at greater risk for RSI since it gets used more for most functions besides typing. This makes the deliberate right-handedness of Norman a deal-breaker for me--enough of a reason to stick to Workman. Any plans to create a variant of Norman that keeps the other features of the design philosophy but moves some frequently-used letters from right to left?

I actually felt like leaving E and D on the left made it left-heavy.

If you were to move common letters from the right hand side to the left, then you've broken much of the theory behind using Norman in the first place. If you were re-training from QWERTY and moved U to the left, then you've lost the mental assignment for the U to be on the right hand, but it's double bad because you'd have to move a left-hand-assigned key to the right.

Trying to train your non-dominant hand to be equivalent to your dominant hand is silly. As long as you're using both hands to type, your dominant hand is also gaining "experience." Your non-dominant will always be behind on the experience curve, in my un-scientific opinion. I think you'd be better-off switching your mousing to your left hand.

The only change I'm considering right now is putting the E, D, and I back in their QWERTY positions, but then I'm basically making a new layout by doing so.

Add new comment

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
By submitting this form, you accept the Mollom privacy policy.
dumb