Can anyone weight in on personal experience and if you felt wrist strain? As a pianist, I think the emphasis on finger-movement chording is misguided. Yes you can play chords on the piano but the entire point is to minimize finger movement - send as much movement as possible to the hand, wrist, forearm, arm, and body so as to minimize local movement. The reason is because if you try to do everything with your fingers, you increase your odds of getting tendonitis. Using a device that is designed to localize movement just to the fingers - through chording no less - strikes me as completely backwards ergonomically.
I see the benefit for those that already have movement limitation, but not really otherwise.
Using a device that is designed to localize movement just to the fingers - through chording no less - strikes me as completely backwards ergonomically.
That's a good point on chording mechanics, and useful to compare with Douglas Engelbart's piano-like chording keyboard[1]. Also, if the current design permits minimal movement precision-grip use (vs. crossing into power grip), it may not be too bad.
The tricky design challenge of the Twiddler is that it's intended for portable, handheld use. E.g. I met Steve Mann using an earlier Twiddler rev. some years back as part of his then-current cyborg rig.
Perhaps it's best to look at this device as another design compromise for a portable keyboard, instead of a desktop keyboard replacement. I wouldn't necessarily want to write a novel on my phone, for similar suitability-to-purpose reasons.
I spent some time working with an early model Twiddler many years ago for a prototype we were building for an industrial application. I loved the idea and was able to get to a pretty decent speed with it, but it was really hard on my wrist. I would not recommend it for constant use. I've since developed some tendinitis and occasional trigger finger (all unrelated to Twiddler use), and I can say it would be a terrible idea for me to try to use one now as gripping makes it worse.
Around ~6 years ago I was intrigued by the idea of using a regular midi keyboard as an input device. I was having some RSI issues, and was looking for alternatives. I'm not a professional pianist, but I play both pianos and keyboards. Any regular midi keyboard doesn't have key shadowing issues, so it's very easy to implement chording mechanics. I remember I was only vaguely aware of stenotyping, and certainly didn't know about plover back then, so I just tried some random approaches.
The first issue with chording keyboards is the tradeoff you make between the amount of keys in a chord required to perform all permutations readily available in a typing keyboard. You can, optimally, pack all permutations only using ~10 keys, but you might require to perform either 6 key chords (3 keys per hand simultaneously), introduce sequences, or introduce trigger delays.
Now, while doable, doing a 6-key chord was quite hard for me. I couldn't remember all the required permutations, and it also required to setup some quantization in order to detect the chords correctly. Quantization in turns imposes an upper-bound on CPM you can attain.
Using more space with smaller chords (a larger set of 2 octaves with ~3-4 keys per chord) was easier for me, but the regularity imposed by the chords (jamming the required keys at the same time) was still unnatural.
I tried to setup some different mechanics, using a sort of tree dictionary where the keys enters the branches. An approach that worked well was to map the frequent character tuples in adjacent keys. Performing CDE, CDF, DEF, EDC combos would result the vowels, which were mapped repeatedly at other locations using the same musical intervals. Typing with this system felt much more "music style": infrequent letters were mapped sparsely on the keyboard, with vowels available in every octave. You could move left-to-right to type. It's also much easier to press keys in _fast_ sequence than trying to perform a perfect chord timing, or at least that was my impression. I didn't need any quantization this way.
Still, for anything but text (ie, programming), a regular keyboard felt so much faster. I remember attaining ~30 wpm or so before giving up entirely.
I wish I had investigated some chord/phonetic based system, but I'm not sure how much that could've helped with regular programming.
As I mentioned elsewhere here, I had a similar experiment (around that same time, I think :-P). What I found worked pretty well was what I called "partially chorded" - where chords in the lower octave determined a mapping in the upper. This eliminates worry about simultaneity and allows one-character-per-keypress responsiveness while still exposing tremendous representation (8 choose 3 times 8 is 448). I found it surprisingly usable in my test application, grouping keys by a fairly cursory analysis of what was likely to go together - minimizing changes of map seems desirable. I never got around to implementing a keyboard driver for it, so it was only available in my test application. I don't think I have the code available anymore, unfortunately - it's something I've always been meaning to revisit.
I developed a software demo (for single hand chording on the keyboard) that updates the character tree representation, possible outputs are shown as the chord sequence is formed. And the adjacent keys idea you have which I call "bigram rollover" works really well in this type of system, because you can tell what neighboring characters are.
The "bigram rollover" is key to reducing the number of key events, and the added bonus is that it should help ramp up the steep learning curve with getting the right sequence.
I am also investigating some sort of phonetic/bigram mapping system (maybe similar to the plover/stenotype system), seems it's the way to go, as shorthand seems to solve the main issue with chording which is high number of key events for a single output.
I am actually looking for help designing the mapping as explained above, it seems we have very similar ideas, perhaps we can collaborate?
I tried many keyboards due to RSI but in the end I'm currently very happy with a kinesis advantage, which made me lose a lot of interest in chording keyboards.
In the end, there's nothing in stenotyping that you couldn't do with a regular keyboard. The advantage in stenotyping is not chording per-se (which is used to reduce the number of keys more than anything else), but the input/composition method, which is highly specialized for text (and english text at that).
There's no convenient way to compose arbitrary symbols in a stenotyping machine. In a regular keyboard there are just more keys and thus the requirement for key composition is almost zero (you still "chord" with shift/control/alt if you think about it). More keys = more immediate symbols. On my S90, which is a full-length keyboard, I only have 88 keys with a key travel length of 1.2m! By comparison, I have (more than) 104 keys in 40x20cm right now.
I agree, stenotype's advantage is not chording and I think our insight is similar; it's the phonetic shorthand which is breaking a word down into syllables which is always shorter than the number of characters (for English anyways).
I think of it as a readily available syllable->word database, which any chording mechanism can take advantage of.
You're right, arbitrary symbols is not stenotype's strength, but chording is somewhat, at least when number of key switches is limited and the whole keyboard size is really small.
I suppose the question is whether chording in a confined space is any better than a full size keyboard, because there is zero or less wrist movement and fingers do not need to reach out (no positional issues).
The answer to that question is yes, chording is better in a setting like mobile phones, which by the way is where we do most of our typing these days, so you shouldn't lose interest.
I don't have any personal experience with the Twiddler. But I have read a lot about computer-based repetitive strain injury and how to avoid it. Dr. Pascarelli's book is a really good resource here.
Normally the advice to "minimize finger movement" is really intended to protect your wrists, not your fingers. If you need to hit a key, but you avoid moving your hand and instead move your fingers, you will be stretching your wrist. If you consistently bend your wrist to the left or right (radial deviation or ulnar deviation) to reach notes, you may get carpal tunnel syndrome. This is basically an inflammation of the soft tissue surrounding the nerve that passes through the "carpal tunnel," a circle of bone in your wrist. The same thing happens when using a flat computer keyboard, and for the same reasons. This is why the correct keyboard posture is the V-shaped one, and the correct way to reach keys is to move your hand, not stretch your wrist.
There are keyboard designs that can (almost) completely avoid ulnar and radial deviation. For example, the SafeType, a vertical keyboard, is a lot easier on the wrists. It's shaped roughly like an accordion, with the keys on both sides. This eliminates the temptation to bend your wrists to reach a key on the other side of the keyboard. The vertical "handshake" position is also a lot less stressful on the wrist than the flat keyboard one.
There are finger issues that can come up if you strike the keys with too much force. These are especially a problem on the ultra-shallow (low travel) keyboards that Apple and other manufacturers are pushing now. In general, though, I would expect a design that only required finger movement to be much easier on the body than a traditional design, PROVIDED that the keys used were good quality, high travel keys. (This is a very important "provided" since a lot of keyboards nowadays are absolute garbage!)
I'm curious if any research has been done on the ergonomics of chording keyboards. If not, perhaps research on RSI issues with trumpets would be a good proxy? In general, trumpet players don't have to move their hands, but only their fingers while playing. On the other hand, trumpets have a lot of vibration, which tends to exacerbate RSI issues.
I have been looking around to see if there is high travel key switch which I believe is key to a good chording experience, something in between a piano key and a keyboard key.
The trumpet should be a good model for the perfect chording experience.
I have played the trumpet years before (in high school) and I recall that there wasn't much strain to the waist or finger. Is there any professional trumpet players here whom can share some experience in regard to the strain at high playing speeds? The location of the strain? Where the limitations are?
I'd be interested in hearing more about the finger issues, since I've had some trouble with my finger joints while typing. I got around it by switching to Cherry Brown and Red keyboards and learning to type on them without bottoming out (which also boosted my typing speed), but it still flares up if I use a low-end keyboard for a few days. I've been assuming it's a bit of osteoarthritis but haven't seen a doctor about it.
One thing I played with is what I called a "partially chorded" keyboard, where chords in one section determine a mapping for the other. I used this to let me type on a midi piano keyboard (where the two sections were neighboring octaves), and it was surprisingly useable for the minimal experience I had with it. Of course, ease of use at the outset is only a poor proxy for capabilities in the long run...
I would note that it doesn't suffer nearly as much from the problem you describe, as changes of chorded position happen more rarely. It might well still have grown painful as I got faster, but it was interesting (and fun to be able to type on a piano keyboard - I've been wanting to rig up my organ to accept input that way).
Chording keyboards [are] a poor idea that will never fly.
How then do you explain the success of stenotype keyboards, which use chording heavily? According to http://en.wikipedia.org/wiki/Stenotype,
"a trained court reporter or closed captioner must write speeds of approximately 180, 200, and 225 words per minute (wpm) at very high accuracy." I don't believe any non-chording-keyboard users can sustain a rate of 225 wpm.
It would be interesting to know if these high speeds be achieved with a non-chording keyboard using sequential keypresses, if the keypresses were post-processed in a manner similar to the stenotype.
Is that 200 WPM counting expanded words or the raw steno notation?
From the same Wikipedia page:
"Some court reporters use scopists to translate and edit their work. A scopist is a person who is trained in the phonetic writing system, English punctuation, and usually in legal formatting."
I.e. there is a post-processing pass whereby phonetic gibberish is translated into legible English with good spelling, grammar and punctuation.
Is the WPM calculation done on the raw input to the scopist, who expands it to proper English, or is it calculated over their output?
See, in ordinary English-language typing tests, you type full sentences. Mistakes in spelling and punctuation subtract from your WPM score.
Is that 200 WPM counting expanded words or the raw steno notation?
It counts expanded words. In steno, the typist's chord (known as a "stroke") is a phonetic abbreviation that is automatically expanded into a word or phrase by software with a lookup table. Some steno software does slightly more complex expansion, but the vast majority of it is pure dictionary lookup.
in ordinary English-language typing tests, you type full sentences. Mistakes in spelling and punctuation subtract from your WPM score.
I see what you mean. It looks like WPM in steno is measured at sub-100% accuracy rates, whereas on a QWERTY keyboard it's measured at 100% accuracy. I'm not sure how to precisely compare steno speeds to QWERTY, but no QWERTY typist has ever come close to the speed of the fastest stenographers. For example, according to http://www.guinnessworldrecords.com/world-records/3000/faste..., "Mark Kislingbury of Houston, Texas, USA is the National Court Reporters Association speed and real-time champion, achieving 360 words per minute with 97.23% accuracy, at the NCRA 2004 summer convention on 30 July 2004."
i think the point is to mash the word like a piano pushing CABD all at the same time, so instead of typing 30 letters to make a sentence you mash word1 word2 word3 word4 word5, similar to 5 keystrokes instead of 30... im looking forward to trying one myself
I spent a few months learning to type with various layouts on the Twiddler 2, but never approached my keyboard speed. Also, the joystick was inaccurate and slow. A clickable trackball in the same position would be so much better.
I think one-handed chording keyboards could be very useful in some niches, but they're a waste of time for most of us.
I'm hoping the Twiddler will provide some variation from regular typing to help keep my hands/arms/wrists more comfortable.
I'm also very interested in experimenting with wearable computing... I've built my own chording keyboards in the past for projects, but was trying to do it cheap/quick so the enclosure and buttons were not easy to use.
20 comments
[ 1.5 ms ] story [ 57.1 ms ] threadI see the benefit for those that already have movement limitation, but not really otherwise.
That's a good point on chording mechanics, and useful to compare with Douglas Engelbart's piano-like chording keyboard[1]. Also, if the current design permits minimal movement precision-grip use (vs. crossing into power grip), it may not be too bad.
The tricky design challenge of the Twiddler is that it's intended for portable, handheld use. E.g. I met Steve Mann using an earlier Twiddler rev. some years back as part of his then-current cyborg rig.
Perhaps it's best to look at this device as another design compromise for a portable keyboard, instead of a desktop keyboard replacement. I wouldn't necessarily want to write a novel on my phone, for similar suitability-to-purpose reasons.
[1] http://www.loper-os.org/wp-content/mw/motd_ui.jpg
[2] Image in [1] via http://www.loper-os.org/?p=861
The first issue with chording keyboards is the tradeoff you make between the amount of keys in a chord required to perform all permutations readily available in a typing keyboard. You can, optimally, pack all permutations only using ~10 keys, but you might require to perform either 6 key chords (3 keys per hand simultaneously), introduce sequences, or introduce trigger delays.
Now, while doable, doing a 6-key chord was quite hard for me. I couldn't remember all the required permutations, and it also required to setup some quantization in order to detect the chords correctly. Quantization in turns imposes an upper-bound on CPM you can attain.
Using more space with smaller chords (a larger set of 2 octaves with ~3-4 keys per chord) was easier for me, but the regularity imposed by the chords (jamming the required keys at the same time) was still unnatural.
I tried to setup some different mechanics, using a sort of tree dictionary where the keys enters the branches. An approach that worked well was to map the frequent character tuples in adjacent keys. Performing CDE, CDF, DEF, EDC combos would result the vowels, which were mapped repeatedly at other locations using the same musical intervals. Typing with this system felt much more "music style": infrequent letters were mapped sparsely on the keyboard, with vowels available in every octave. You could move left-to-right to type. It's also much easier to press keys in _fast_ sequence than trying to perform a perfect chord timing, or at least that was my impression. I didn't need any quantization this way.
Still, for anything but text (ie, programming), a regular keyboard felt so much faster. I remember attaining ~30 wpm or so before giving up entirely.
I wish I had investigated some chord/phonetic based system, but I'm not sure how much that could've helped with regular programming.
The "bigram rollover" is key to reducing the number of key events, and the added bonus is that it should help ramp up the steep learning curve with getting the right sequence.
I am also investigating some sort of phonetic/bigram mapping system (maybe similar to the plover/stenotype system), seems it's the way to go, as shorthand seems to solve the main issue with chording which is high number of key events for a single output.
I am actually looking for help designing the mapping as explained above, it seems we have very similar ideas, perhaps we can collaborate?
In the end, there's nothing in stenotyping that you couldn't do with a regular keyboard. The advantage in stenotyping is not chording per-se (which is used to reduce the number of keys more than anything else), but the input/composition method, which is highly specialized for text (and english text at that).
There's no convenient way to compose arbitrary symbols in a stenotyping machine. In a regular keyboard there are just more keys and thus the requirement for key composition is almost zero (you still "chord" with shift/control/alt if you think about it). More keys = more immediate symbols. On my S90, which is a full-length keyboard, I only have 88 keys with a key travel length of 1.2m! By comparison, I have (more than) 104 keys in 40x20cm right now.
I think of it as a readily available syllable->word database, which any chording mechanism can take advantage of.
You're right, arbitrary symbols is not stenotype's strength, but chording is somewhat, at least when number of key switches is limited and the whole keyboard size is really small.
I suppose the question is whether chording in a confined space is any better than a full size keyboard, because there is zero or less wrist movement and fingers do not need to reach out (no positional issues).
The answer to that question is yes, chording is better in a setting like mobile phones, which by the way is where we do most of our typing these days, so you shouldn't lose interest.
Normally the advice to "minimize finger movement" is really intended to protect your wrists, not your fingers. If you need to hit a key, but you avoid moving your hand and instead move your fingers, you will be stretching your wrist. If you consistently bend your wrist to the left or right (radial deviation or ulnar deviation) to reach notes, you may get carpal tunnel syndrome. This is basically an inflammation of the soft tissue surrounding the nerve that passes through the "carpal tunnel," a circle of bone in your wrist. The same thing happens when using a flat computer keyboard, and for the same reasons. This is why the correct keyboard posture is the V-shaped one, and the correct way to reach keys is to move your hand, not stretch your wrist.
There are keyboard designs that can (almost) completely avoid ulnar and radial deviation. For example, the SafeType, a vertical keyboard, is a lot easier on the wrists. It's shaped roughly like an accordion, with the keys on both sides. This eliminates the temptation to bend your wrists to reach a key on the other side of the keyboard. The vertical "handshake" position is also a lot less stressful on the wrist than the flat keyboard one.
There are finger issues that can come up if you strike the keys with too much force. These are especially a problem on the ultra-shallow (low travel) keyboards that Apple and other manufacturers are pushing now. In general, though, I would expect a design that only required finger movement to be much easier on the body than a traditional design, PROVIDED that the keys used were good quality, high travel keys. (This is a very important "provided" since a lot of keyboards nowadays are absolute garbage!)
I'm curious if any research has been done on the ergonomics of chording keyboards. If not, perhaps research on RSI issues with trumpets would be a good proxy? In general, trumpet players don't have to move their hands, but only their fingers while playing. On the other hand, trumpets have a lot of vibration, which tends to exacerbate RSI issues.
The trumpet should be a good model for the perfect chording experience.
I have played the trumpet years before (in high school) and I recall that there wasn't much strain to the waist or finger. Is there any professional trumpet players here whom can share some experience in regard to the strain at high playing speeds? The location of the strain? Where the limitations are?
I would note that it doesn't suffer nearly as much from the problem you describe, as changes of chorded position happen more rarely. It might well still have grown painful as I got faster, but it was interesting (and fun to be able to type on a piano keyboard - I've been wanting to rig up my organ to accept input that way).
How then do you explain the success of stenotype keyboards, which use chording heavily? According to http://en.wikipedia.org/wiki/Stenotype, "a trained court reporter or closed captioner must write speeds of approximately 180, 200, and 225 words per minute (wpm) at very high accuracy." I don't believe any non-chording-keyboard users can sustain a rate of 225 wpm.
Related: https://news.ycombinator.com/item?id=8510409 ("Plover: Text to thought at 240 WPM")
It would be interesting to know if these high speeds be achieved with a non-chording keyboard using sequential keypresses, if the keypresses were post-processed in a manner similar to the stenotype.
"Some court reporters use scopists to translate and edit their work. A scopist is a person who is trained in the phonetic writing system, English punctuation, and usually in legal formatting."
I.e. there is a post-processing pass whereby phonetic gibberish is translated into legible English with good spelling, grammar and punctuation.
Is the WPM calculation done on the raw input to the scopist, who expands it to proper English, or is it calculated over their output?
See, in ordinary English-language typing tests, you type full sentences. Mistakes in spelling and punctuation subtract from your WPM score.
It counts expanded words. In steno, the typist's chord (known as a "stroke") is a phonetic abbreviation that is automatically expanded into a word or phrase by software with a lookup table. Some steno software does slightly more complex expansion, but the vast majority of it is pure dictionary lookup.
in ordinary English-language typing tests, you type full sentences. Mistakes in spelling and punctuation subtract from your WPM score.
I see what you mean. It looks like WPM in steno is measured at sub-100% accuracy rates, whereas on a QWERTY keyboard it's measured at 100% accuracy. I'm not sure how to precisely compare steno speeds to QWERTY, but no QWERTY typist has ever come close to the speed of the fastest stenographers. For example, according to http://www.guinnessworldrecords.com/world-records/3000/faste..., "Mark Kislingbury of Houston, Texas, USA is the National Court Reporters Association speed and real-time champion, achieving 360 words per minute with 97.23% accuracy, at the NCRA 2004 summer convention on 30 July 2004."
I spent a few months learning to type with various layouts on the Twiddler 2, but never approached my keyboard speed. Also, the joystick was inaccurate and slow. A clickable trackball in the same position would be so much better.
I think one-handed chording keyboards could be very useful in some niches, but they're a waste of time for most of us.
I'm also very interested in experimenting with wearable computing... I've built my own chording keyboards in the past for projects, but was trying to do it cheap/quick so the enclosure and buttons were not easy to use.