Ergonomic layout of a standard keyboard "NEO"

Optimized for the German linguistic area (and western/european area/unicode)

usable for english and with minor adjustmens of special characters for other languages


 Last update: 29.6.2005
  1. Ergonomic layout of a standard keyboard "NEO"
    1. Optimized for the German linguistic area (and western/european area/unicode)
    2. usable for english and with minor adjustmens of special characters for other languages
    3. Short history
    4. What is the NEO layout?
      1. The paradigms of this layout
    5. Statistic comparison
      1. Statistic distribution
    6. Various keyboard layouts
    7. Change of the keyboard
    8. Key simulator
    9. Files
    10. Links

Short history

Most of you have a so-called QWERTY keyboard, and you may believe, this keyboard is well designed and it was developed on the basis of ergonomic criteria. Wrong.

Already in 1932 invented Dr. Augusts Dvorak a keyboard layout with improved usability, with clearly less effort and risk on repetitive strain injury, as well as easy to learn. But his invention remained almost unknown until today, although many computers systems already support this keyboard layout, if you only would know and turn it on (Mac, Windows, Linux...).

The learning of a layout costs between 25 hours and 100 hours time, there are empiric reports on the net, even switching between layouts is only one minute of "remapping" in mind. So there are no rational reasons against learning a new, more efficient layout. The old layout slows down your typing rate between 50% and 75%, fatigues and leads often to repetitive strain injury. The whole world is talking about software ergonomic, invests in faster computers, while they persist on a keyboard layout, which was designed around 1870, for to avoid fast typing on mechanical typewriters that lead to jamming. So the most used types are put on the keyboard with the maximum distance, so that the types don't jam and as a good argument to sell these old scrappy typing machines, the upper row was designed to type "TYPEWRITER" on it. For such a important interface between machine and human this is unbelievable, no matter how long it has "worked". (more on that you'll find down below in the "link"-section)

What is the NEO layout?

Here and now I would like to present a layout, which I call NEO and developed in 2004. There are ergonomic layouts as from Dvorak (1932), Meier (1964) and other one, but these layouts were not fitting in their paradigma or conceptions. Sometimes they are outdatet, relay on a language that has changed too much, tried to fit all languages or were simply not designed with using statistics and therefor failed optimizing. You can find a good site for ergonomic layouts from Hartmut Goebel, some of the scientific and ergonomic basis on this site I based on at beginning development.

The paradigms of this layout

  1. the fastest and most comfortable letter takes place on the baseline, therefore all frequent indications should be on this baseline
  2. the pairs of letters, which follow most frequently one another, should be distributed on both hands, this is also positiv for the typing rate
  3. if certain letter sequences cannot be distributed conceptionally on both hands, then they should follow drum patterns from the inside outward, optimal way with a finger between them, speak of index finger after ringfinger or small finger, of the middle finger after small finger and if necessary, always strictly arranged according to frequency distributions in reverse. 
  4. the strongest load should be appropriate on index and middle finger, the upper row is better than the lower, ring and small fingers should be used few in the upper row and avoided on the lower. 
  5. most people are right-handed, so this should count in, the standard keyboard is very much left-handed, because the left hand has to do most of the work
  6. the german language has priority, since it is my native (it's not that great difference to the english statistics anyway), after that english, programming and shell instructions (Unix). 
  7. if the place for a key is statistically ambivalent on it's old place, it should remain there for better learning effect
  8. if the place for a key is statistically ambivalent, it should remain arranged in logical blocks, since thats easier to learn and speeds up by use of mnemobocks. This makes it easier for not "one finger on sight typer". 
  9. unavoidable "worst case" scenarios, as moving from upper to lower row should remain statistically irrelevant
  10. conventional keyboards should work, ergonomic re-organization of the keys are not planned (although some ideas for this exist)

By cryptographic statistics of the frequency distribution of the letters in the german language, as well as the frequency of certain letter sequences and efficiency of certain palpation sequences, the resulting arrangement of the keys are more or less inevitably. Since existing layouts mostly ignore or break several of my paradigms, it came to the development of a new layout - NEO.

The NEO layout deduces itself particularly from my own needs to pass the limiting borders of qwert, with fewer fatigue symptoms, because I would call myself a busy writer.

Statistic comparison

A little statistics will show the strength of  NEO, compared to other layouts, the possible improvements and the limits.

For the developement I used statistics and ergonomic scientific investigations from Hartmut Goebels site, other sources like word statistics and some small statistic programs I developed for this task. It came to some interesting results. Limited to the basic row, a list of the most frequent german words was matched on the QWERTZ keyboard and scores only 75 words, on Dvorak 1400, on de_ergo (more eMeier) approximately 1600 and on the NEO design over 3600. This resulted rather coincidentally from the optimization of the statistic distribution and shows, this is a good way in optimizing a keyboard layout. For the english words it was not as clear, but still significant better than all other layouts (including dvorak!).

On the Wikiweb you can find a list of the most frequent 207 german words, with which a quite convincing statistics can be set up. These 207 words result together in approximately 54% of the on the average written german text. The following statistics shows, how much the weighted words of it can be written with the fastest keys. The 50,5% to the priority at the end of the "NEO" column 189 of the 207 words, which means 91% of the words altogether, however the priority of the words arise at the beginning of the list more frequently than at the end. I measured  even distributed within the groups and would actually ignored the Zipf law, the consideration the projection/lead of NEO here, since the missing words of the first group are in the last third of the list. Still better performance on the baseline is to be expected after the german spelling reform, because many of the words in this list are still according to old kind with "ß" instead of "ss" written and the "ß", after the reform, loses statistic priority in the text flow.

finger\weighted % vocabulary (words) NEO
de_ergo
de-more eMeier
Dvorak
QWERTZ
basic position
11,8 (23)
11,0 (21)
11,0 (21) 4,7 (12)
2,4 (4)
+indexfinger spread
21,4 (51)
13,8 (30)
13,8 (30) 18,7 (43)
2,4 (4)
+indexfinger up
30,4 (83)
26,2 (75)
26,2 (75) 19,0 (46)
4,1 (10)
+middlefinger up
32,8 (98)
30,7 (105)
30,7 (105) 25,3 (58)
12,7 (31)
+indexfinger down
38,9 (124)
43,6 (147)
36,2 (125) 31,1 (82)
16,2 (42)
+indexfinger spread up
43,6 (157) 45,2 (154)
45,2 (154) 32,1 (83)
20,6 (55)
+indexfinger spread down
47,7 (181) 47,0 (169)
47,0 (169) 33,3 (91)
35,8 (126)
+ringfinger up
50,5 (189)
48,8 (184)
48,8 (184)
35,7 (107)
43,3 (167)

You'll recognize that the frequency of the punctuation marks "," and "." and therefore the Dvorak allocation starting from the middle finger remains unconsidered in an somewhat unfair way and performce more badly in the comparison, particularly to the QWERTZ. Also poor performance of the QWERTZ layout is typical in this most important range.

I think this small statistics is a clear argument for the NEO design, but everyone should decide for himself. The layout could probable be optimized still on a special word list, however at the price of the large whole, because at the end again the total statistics of the german language will count in and particularly on this global statistic NEO is optimized for the total mass of text.

The frequent emerging of the word "paragraph" in the list is most suspect, I think, this is a word, which I personally just used just now the first time in my life. If you go through the first three wordgroups of the list, it determines that the most frequent words can be typed extremely simply on the NEO keyboard, to a large extent on the basic LINE and in a continuous direction of finger movement, which brings naturally clear speed advantages. That is however likewise not because of an optimization on this word list, but that the layout gives generally the most frequent letter sequences in bi and trigrams and also tries to optimize the finger running in one direction like Rohmert measured is the fastest way and which produces high typing speeds.

The speed of highly developed ergonomic layouts develop to a certain maximum, which permits not much more increases any more. So after NEO, I presume is no relevant gain. Even if a seldom used key could be highly optimized (which I do not think), the overall gain will be too small after all.


Statistic distribution

20%
Q
V
L
C
W
K
H
G
F
J
ss
70%
U
I
A
E
O
S
N
R
T
D
Y
10%
Oe
Ue
Ae
P
Z
B
M
,
.
X


4%
8%
9%
26%


27%
13%
8%
5%

47% 53%

Additionally to this allocation the NEO layout was supplied with a large number of special character on the AltGR and Shift AltGr level, which supports all scientific working and also most western unicode characters, like french, danish, and so on. Starting with euro-sign and cent, many greek letters, scandinavian or other special characters and a pronounced DEAD key holding back, thus, with which letters can be modified. There are akut's, cedilla, ogonek, tilde or cirkumflex and many more modifiers.

Various keyboard layouts

You can find the common keyboard layouts in the overview briefly arranged on this page.

Change of the keyboard

The NEO layout aims at the 10-finger-touched-type, why letters on the keys are actually irrelevant, but many people like the correct letters on the keys. In order to reach this simply and favorably, there are the following possibilities, for whose successful execution I can naturally take no responsibility. I personally did without renaming the keys of my keyboard, since I write blind anyway.
  1. Pry off the keys with a simple screwdriver carefully and put them to the new place. That can be done very simply and should anyone should succeed. Unfortunately many modern keyboards are build in such a way that the rear row keys have another height than the front, so that it will be rough like copplestone. Before you seizes this simplest possibility, you should examine  whether the keyboard permits this at all. Otherwise the second method remains.
  2. This method is somewhat more complex, however a bit more elegantly than the first. With a glaseraser or similar equipment you rub out the letters on the keys. Glaseraser consist of a small bundle glass fibers, which scrub the letters off the plastic keys with their sharp ends very precisely and carefully.
    Afterwards you rub new rubbing letters on the keys, you get such foil letters in each well sorted stationer's shop. This method works usually very well, however the keys should be absolutely dust, dirt and grease-free. The letters in factory are put on mostly with the same method. Careful work provides here good appearance.
  3. There are small caps for keyboards, so-called keypreserver. Put these caps simply in the desired order on the right keys. Since I know this method only of hearing, I cannot say anything further, also not where one gets such caps.

Key simulator

The key simulator i planned to work on, which packs the scientific realizations over the ergonomics of keyboards into a small Java program, which calculates, how quickly a certain text with a certain layout could be typed on the average under optimal conditions, is on ice now, because of lacking time and usefulness. The results should be predictable anyway, because NEO fits perfect in the paradigma. The only thing you'll gett would be exact measurements how much speed you gain, not if the NEO works fine.

Since the designers of the other layouts did a damn good word anyway, I can't say I did this alone, because these people backed me up with their knowledge and withoud NEO won't exist in the way it is now.

Files

 With improved Linux version and Windows XP driver

For printing layout I prepared a small pdf file, which you can download by clicking the on the link. Because I am still in development any changes can happen. You find as well the layout de_ergo layout of Hartmut Goebel for xkb/Qt which is an improvement of the Meier layout and you may use them in comparison.

For learning the new layout, I wrote a learning lesson for KTouch the KDE typing program from the kdeedu package. This typewriter learning lesson was made by transfering the spirit of the DIN typewriter lesson and is now as close you can come to that standard. The lesson covers 100 short chapters, together with exercises and such things and ends, where the lesson would be exactly become the same for QWERTZ layout, with the numbers for example. I recommend to continue the course with exercises and sample letters from the text book for typewriter, where this course ends. On older Linux versions the special characters of the training file on ISO8859-15, coded in UTF-8, would have to be back-coded, approximately with recode or iconv.

A readme file describes the installation in the archive for installation on Linux and Windows XP.

So far no drivers for Mac, Win98 and Linux/Console are available, if necessary I can write the Linux/Console drivers subsequently. For the Mac I can't help with lack of Macs here, the Win98 I don't yet know how to.

This software is under the GNU Public Licence publishes, all rights reserved, no warranty.

You download the files here:
(including Linux Driver, Windows XP driver, readme, sourcefiles, ktouch learning lesson)
tgz: NEO layout for Linux/KDE and Windows XP
zip: NEO layout for Linux/KDE and Windows XP

Hanno Behrens

History:
28.01.2005: english version of NEO-Homepage
27.01.2005: Version 1.00 - Again! Now also drivers for Windows XP

Thanks to
Hartmut Goebel, Johannes trunk, Markus Hanauska, Denis Knauf

Back

Links

Hartmut Goebel DE-ERGO

Wikipedia - keyboard
Wikipedia - alphabet