What actually went into the design of the 22-key English Stenotype?
259 views
Skip to first unread message
Steven Tammen
Jan 21, 2016, 8:10:45 PM1/21/16
to Plover
Having spent some time thinking about what is actually ideal, it seems to me that chording the ending consonants -D and -Z from home position would be more optimal than the pinky reach currently in place. Keeping everything on "home position" gives us exactly 20 keys to work with: a top key and bottom key for each normal digit, and two vowel keys horizontal to each other for the thumb. Because the easiest motions for fingers involve simple flexion and contraction (think high-five and claw, respectively), having a top key and bottom key is better than any lateral motion for these fingers (so anything other than these two keys is sub-optimal, even small lateral movements like those of the Datahand). Thumbs tend towards a different type of motion that is roughly 45 degrees to the plane, so lateral keys for them make the most sense (though it would be better if their natural range of motion were incorporated more into physical stenotype design, in my opinion).
All this somewhat philosophical. Having come from the ranks of keyboard layout optimizers (and the particular breed unhappy with Colemak, Dvorak, and Workman for not being "good enough", read: custom designed algorithmically generated layouts), it strikes me that steno shares with QWERTY the dubious honor of having been invented before the advent of computing and large-scale data analysis. Now, from what research I have done, chording considerations matter much less than letter placement for touch typing. That is to say, while you could try to make steno inefficient by making T require the left hand chord currently in place for Z, it still wouldn't actually matter a great deal. Steno has effectively zero travel distance for chords in home position, and thus it is almost impossible to be inefficient with it, once you get things into muscle memory (chording four keys may be less good than chording one, but the different is negligible after practice). Thus, the impulse to go off and redesign the current layout with algorithms to squeeze out a bit more efficiency is tempered by practical realism: while 22-key steno isn't perfect, the marginal benefit of statistical improvement is low, while the time involved is astronomical. This is true of orthographic methods as well (e.g., Velotype).
So, with that long preamble, my question is pretty simple: why exactly do we have STKPWHRAO*EUFRPBLGTSDZ? Especially, why are the letters placed as they are, and why do we have -D and -Z requiring pinky extension? Are these things historical artifacts, or are there specific quantitative reasons for pinky extension (e.g., you need these keys to account for all ending phonemes, there is no elegant way to design chords for -D and -Z while maintaining steno order and common strokes, etc.)? I fear I am at a disadvantage because I am still waiting for my HRUF to come so I can start practicing: all I have is abstract "head knowledge". Perhaps these things are obvious once you get into steno, but Google failed me in turning up any papers on the design principles for the layout, why specific chords were chosen for letters (e.g., Z, J), and how it all fits together. In an ideal case I would work out some system in Plover to chord -D and -Z and then never leave home position for the rest of my life. Is this possible? If so, is it a bad idea? Why?
All this somewhat philosophical. Having come from the ranks of keyboard layout optimizers (and the particular breed unhappy with Colemak, Dvorak, and Workman for not being "good enough", read: custom designed algorithmically generated layouts), it strikes me that steno shares with QWERTY the dubious honor of having been invented before the advent of computing and large-scale data analysis. Now, from what research I have done, chording considerations matter much less than letter placement for touch typing. That is to say, while you could try to make steno inefficient by making T require the left hand chord currently in place for Z, it still wouldn't actually matter a great deal. Steno has effectively zero travel distance for chords in home position, and thus it is almost impossible to be inefficient with it, once you get things into muscle memory (chording four keys may be less good than chording one, but the different is negligible after practice). Thus, the impulse to go off and redesign the current layout with algorithms to squeeze out a bit more efficiency is tempered by practical realism: while 22-key steno isn't perfect, the marginal benefit of statistical improvement is low, while the time involved is astronomical. This is true of orthographic methods as well (e.g., Velotype).
So, with that long preamble, my question is pretty simple: why exactly do we have STKPWHRAO*EUFRPBLGTSDZ? Especially, why are the letters placed as they are, and why do we have -D and -Z requiring pinky extension? Are these things historical artifacts, or are there specific quantitative reasons for pinky extension (e.g., you need these keys to account for all ending phonemes, there is no elegant way to design chords for -D and -Z while maintaining steno order and common strokes, etc.)? I fear I am at a disadvantage because I am still waiting for my HRUF to come so I can start practicing: all I have is abstract "head knowledge". Perhaps these things are obvious once you get into steno, but Google failed me in turning up any papers on the design principles for the layout, why specific chords were chosen for letters (e.g., Z, J), and how it all fits together. In an ideal case I would work out some system in Plover to chord -D and -Z and then never leave home position for the rest of my life. Is this possible? If so, is it a bad idea? Why?
Mirabai Knight
Jan 21, 2016, 8:30:43 PM1/21/16
to ploversteno
You know how in Scrabble the S tile is one of the most valuable tiles you can have in your rack, even though it has a low point value? Because you can use it to pluralize almost every noun and turn a ton of verbs from imperative to indicative. It's an incredibly useful and versatile letter. Think if there was a tile that let you add "ed" to any verb to make it past tense. That would be pretty dang useful too, right?
That's why the D and Z keys have their own column at the far right of the keyboard: So you can write almost any syllable in English in a single stroke, and then have the option of nearly doubling the number of strokes you can write, by adding on that S or -ED inflection by just sliding your pinky over to one or both of those keys without lifting up the other ones. The asymmetry of the keyboard instinctively bothers a lot of newbies to steno, until they realize it's English's asymmetry -- having most of its noun and verb inflections conveyed by an S or -ED at the end of a word -- and not a flaw in steno itself.
--
You received this message because you are subscribed to the Google Groups "Plover" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ploversteno...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Steven Tammen
Jan 21, 2016, 9:26:09 PM1/21/16
to Plover
This makes a good amount of sense. Do you know off the top of your head if -s and -ed are actually the most common affixes? I'm sure there are other things like -ing up there too, but I'm guessing you are right.
I already knew this asking the question, but I was interested in the possibility of chording say, -T and -S for -D and then using the S already there, eliminating the extra pinky movement. As I said, much of the curiosity is likely me not knowing all the stroke conflicts doing such would create, but I think it is an interesting idea nonetheless (at least philosophically). Could you go over some words that wouldn't work with the above? Are there actually things where you would need to stroke -SD or -TD?
What about other strokes like TP for F and TPH for N? I'm assuming they aren't entirely arbitrary, so is there anywhere where the choices are described logically?
Thanks!
I already knew this asking the question, but I was interested in the possibility of chording say, -T and -S for -D and then using the S already there, eliminating the extra pinky movement. As I said, much of the curiosity is likely me not knowing all the stroke conflicts doing such would create, but I think it is an interesting idea nonetheless (at least philosophically). Could you go over some words that wouldn't work with the above? Are there actually things where you would need to stroke -SD or -TD?
What about other strokes like TP for F and TPH for N? I'm assuming they aren't entirely arbitrary, so is there anywhere where the choices are described logically?
Thanks!
Mirabai Knight
Jan 21, 2016, 9:29:01 PM1/21/16
to ploversteno
Use TS for -D and you've just added an extra stroke to the thousands and thousands of words ending in TS.
http://www.wordfind.com/ends-with/ts/
--
Steven Tammen
Jan 21, 2016, 9:37:43 PM1/21/16
to Plover
Ah... I think I better learn bit more before asking further silly questions. See: I knew there were reasons, I am just too new to see them!
I'm still interested in learning more about the layout, and as I said, there is disappointingly little on the internet about it, even though it has been around for a while. If anyone has links or just general knowledge, It'd be greatly appreciated!
Thanks,
S.
I'm still interested in learning more about the layout, and as I said, there is disappointingly little on the internet about it, even though it has been around for a while. If anyone has links or just general knowledge, It'd be greatly appreciated!
Thanks,
S.
Gabriel Holmes
Jan 21, 2016, 10:24:26 PM1/21/16
to Plover
There are a number of ways of looking at this. I tend to start with math. A "normal" steno machine has 23 keys. Clearly not all 8,388,607 subsets can be realized as strokes.
There are, some modern machines for which:
a) The left S key is split
b) The number bar is split into as many as twelve discrete pieces
c) The asterisk is split into four distinct pieces
d)There are keys to the left of the left S key
Putting aside issues of comfort and risk of injury, one could, in theory, produce as many as 457,275,455 distinct strokes.
We can also look at this from the standpoint of phonetics. Steno is based on the idea of keeping up with human speech, which is also subject to certain constraints. For instance, within a given syllable, sounds can only occur in a certain order. The more sonorous, or "vowel like" occur in the middle of a syllable, and become less vowel like as you move outward. That's why "R" and "L" are towards the middle, whatever type of machine stenography you're dealing with (velotype, stenotype, and palantype). It's hard to come across estimates of exactly how many humanly possible syllables there are -- but even some of the larger estimates are less than the number of practical strokes even on a "classic" steno machine. For polysyllabic words, you can invoke the "Maximum Onset Principle" and focus on syllables with very simple (if any) codas. This would tend to discourage you from torturing your right pinky. Basically, the layout of any flavor of steno machine reflects universal phonotactic constraints (to the extent that such constraints exist). Incidentally, from a practical standpoint, there are only a million or so English words (let alone syllables) so in some sense all this fretting over syllables is irrelevant. Now, most languages don't have as many words as English -- but assume for the sake of argument that they do (also, there only a couple hundred thousand that are actually in common use). The largest number of languages spoken by any one person is 58 -- so even if that person knew every single word in every one of those languages and each of those languages had as rich a vocabulary as English, it would still be less than the 457,275,455 figure.
As for your comment about stenography having been developed before the advent of computing and large scale data analysis -- a lot of things were, but we don't begrudge them that because they basically work: vaccines, suspension bridges, airplanes, and beer to name a few. While computational tools have helped since then, they haven't changed radically since then. And I'd suggest that in the case of stenography, it's sufficiently customizable by its nature that the basic design is basically irrelevant -- that's what "real time theories" are for.
There are, some modern machines for which:
a) The left S key is split
b) The number bar is split into as many as twelve discrete pieces
c) The asterisk is split into four distinct pieces
d)There are keys to the left of the left S key
Putting aside issues of comfort and risk of injury, one could, in theory, produce as many as 457,275,455 distinct strokes.
We can also look at this from the standpoint of phonetics. Steno is based on the idea of keeping up with human speech, which is also subject to certain constraints. For instance, within a given syllable, sounds can only occur in a certain order. The more sonorous, or "vowel like" occur in the middle of a syllable, and become less vowel like as you move outward. That's why "R" and "L" are towards the middle, whatever type of machine stenography you're dealing with (velotype, stenotype, and palantype). It's hard to come across estimates of exactly how many humanly possible syllables there are -- but even some of the larger estimates are less than the number of practical strokes even on a "classic" steno machine. For polysyllabic words, you can invoke the "Maximum Onset Principle" and focus on syllables with very simple (if any) codas. This would tend to discourage you from torturing your right pinky. Basically, the layout of any flavor of steno machine reflects universal phonotactic constraints (to the extent that such constraints exist). Incidentally, from a practical standpoint, there are only a million or so English words (let alone syllables) so in some sense all this fretting over syllables is irrelevant. Now, most languages don't have as many words as English -- but assume for the sake of argument that they do (also, there only a couple hundred thousand that are actually in common use). The largest number of languages spoken by any one person is 58 -- so even if that person knew every single word in every one of those languages and each of those languages had as rich a vocabulary as English, it would still be less than the 457,275,455 figure.
As for your comment about stenography having been developed before the advent of computing and large scale data analysis -- a lot of things were, but we don't begrudge them that because they basically work: vaccines, suspension bridges, airplanes, and beer to name a few. While computational tools have helped since then, they haven't changed radically since then. And I'd suggest that in the case of stenography, it's sufficiently customizable by its nature that the basic design is basically irrelevant -- that's what "real time theories" are for.
Gabriel Holmes
Jan 21, 2016, 10:29:10 PM1/21/16
to Plover
Another thing, related to Mirabai's point -- I think linguistically there's a lot to be said for regarding syllables for the purpose of stenography as being strictly subordinate to morphemes -- thus "cats" should really be thought of as "cat" + "s". Sure, maybe sometimes you might get lucky and be able to do it all in one stroke, but you shouldn't plan on that any more than you should plan your budget by assuming you're going to win the lottery or inherit a lot of money from a rich uncle.
On Thursday, January 21, 2016 at 8:10:45 PM UTC-5, Steven Tammen wrote:
Steven Tammen
Jan 21, 2016, 11:43:31 PM1/21/16
to Plover
Thanks for the insightful comments.
I have seen many of the new steno machines with the split S key, another column to the left of that, multiple asterisk keys, etc., but for the purpose of pure stenography it seems that it is better to chord more and move less, since chording requires almost no movement and is thus faster/more ergonomic. If you presuppose such a belief, you have each finger spanning two possible keys, and end up with 22 normal keys (with my new understanding of the -D and -Z keys) plus the asterisk for disambiguation. Any additional keys I'm planning to use for "leader characters" to signal the start of command strokes, symbols, and such (or perhaps "leader strokes", depending on how easy that is to implement). I haven't fully developed it yet, but I'm planning on writing almost exclusively in Emacs except making it modal like Vim for some things (numbers, which you almost always enter in sequences, and arrows + pgup/pgdn keys which often have multiple presses in short time-periods). Extensibility here seems like the key, which is harder in Vim than in Emacs since Vim doesn't run a lisp interpreter or equivalent internally. Anyhow, getting a bit off topic...
The mathematical treatment of these issues is fascinating. I think the main thing to take away is that spending tons of time on this becomes impractical because there is simply so much to sift through. I had to do some reading to understand your points (for posterity: phonotactic constraints, coda), and I couldn't agree more now than I've thought about it. If you avoid having difficult codas in intermediate syllables, use of -D and -Z can be (mostly) reserved for pluralizing/past-tensing morphemes and your right pinky can remain happy and RSI free. Also, thinking about -D and -Z as "things you add on the end of ___" makes it easier to understand why they exist. For cats, I'm guessing you would probably stroke it KA-TS. But for words without the easy pluralization, that's where the -Z comes in: you just tack it on the end, and save yourself another stroke. Speed >> minor comfort improvement.
Finally regarding your last point, you are right of course. I fear I (and the rest of my generation with me) forgets how new computers really are. My parents have told me stories of terminals in college taking up entire rooms, and boxes with megabytes (!) of RAM. I guess my overzealousness comes from our society's stubborn insistence on mediocrity for so many other things that can be improved with computers but aren't because of historical factors and change friction. As the saying goes, if all you have is a hammer, then everything looks like a nail.
Please do correct me if I am thinking about things in the wrong way!
S.
I have seen many of the new steno machines with the split S key, another column to the left of that, multiple asterisk keys, etc., but for the purpose of pure stenography it seems that it is better to chord more and move less, since chording requires almost no movement and is thus faster/more ergonomic. If you presuppose such a belief, you have each finger spanning two possible keys, and end up with 22 normal keys (with my new understanding of the -D and -Z keys) plus the asterisk for disambiguation. Any additional keys I'm planning to use for "leader characters" to signal the start of command strokes, symbols, and such (or perhaps "leader strokes", depending on how easy that is to implement). I haven't fully developed it yet, but I'm planning on writing almost exclusively in Emacs except making it modal like Vim for some things (numbers, which you almost always enter in sequences, and arrows + pgup/pgdn keys which often have multiple presses in short time-periods). Extensibility here seems like the key, which is harder in Vim than in Emacs since Vim doesn't run a lisp interpreter or equivalent internally. Anyhow, getting a bit off topic...
The mathematical treatment of these issues is fascinating. I think the main thing to take away is that spending tons of time on this becomes impractical because there is simply so much to sift through. I had to do some reading to understand your points (for posterity: phonotactic constraints, coda), and I couldn't agree more now than I've thought about it. If you avoid having difficult codas in intermediate syllables, use of -D and -Z can be (mostly) reserved for pluralizing/past-tensing morphemes and your right pinky can remain happy and RSI free. Also, thinking about -D and -Z as "things you add on the end of ___" makes it easier to understand why they exist. For cats, I'm guessing you would probably stroke it KA-TS. But for words without the easy pluralization, that's where the -Z comes in: you just tack it on the end, and save yourself another stroke. Speed >> minor comfort improvement.
Finally regarding your last point, you are right of course. I fear I (and the rest of my generation with me) forgets how new computers really are. My parents have told me stories of terminals in college taking up entire rooms, and boxes with megabytes (!) of RAM. I guess my overzealousness comes from our society's stubborn insistence on mediocrity for so many other things that can be improved with computers but aren't because of historical factors and change friction. As the saying goes, if all you have is a hammer, then everything looks like a nail.
Please do correct me if I am thinking about things in the wrong way!
S.
Message has been deleted
Hanchul Park
Jan 22, 2016, 9:45:49 AM1/22/16
to Plover
I think the current design of 22-key stenotype is due to old custom just like QWERTY, even though it is very smartly designed from the beginning. I don't get the imbalance of 7:11 ratio of left and right finger keys. Many input devices contain buttons assigned for two or more fingers like the star key of the stenotype or the space bar of the QWERTY keyboard, but it makes almost no sense. It is far better to divide it into two or more keys and give potentially different functions. If someone says 7:11 imbalance is because English has complex codas compared to onsets, it is OK but if I have additional keys for the left hand, then I could use it for word boundary control, initial vowels, or anything I can imagine.
It is theoritically possible that the minimal key layout of 20 keys is the best if it can boost speed of strokes. But we have two things to think about. The first thing is additional keys would cause finger stretches, but they can reduce the number of strokes like some people pointed out. The second is once a layout is used widely, then reducing the number of keys respecting former users of the layout is very hard or impossible.
2016년 1월 22일 금요일 오후 1시 43분 31초 UTC+9, Steven Tammen 님의 말:
It is theoritically possible that the minimal key layout of 20 keys is the best if it can boost speed of strokes. But we have two things to think about. The first thing is additional keys would cause finger stretches, but they can reduce the number of strokes like some people pointed out. The second is once a layout is used widely, then reducing the number of keys respecting former users of the layout is very hard or impossible.
2016년 1월 22일 금요일 오후 1시 43분 31초 UTC+9, Steven Tammen 님의 말:
Steven Tammen
Jan 22, 2016, 3:17:13 PM1/22/16
to Plover
On "sharing" keys with fingers, I agree with you. The huge spacebar is one of the things that annoys me most about normal keyboards: thumbs are among our strongest most dextrous fingers, and yet they are relegated to sitting unused.
Your proposition of initial vowels is interesting, but there are many sounds, and the maximum onset principle implies that vowels will not be the first thing in the onset excepting for the beginning of words. Additionally, there are may words that can be stroked in one go even so ("able", for example). Do you have a formulated plan with this concept in mind? How would you account for all the possible sounds?
Your proposition of initial vowels is interesting, but there are many sounds, and the maximum onset principle implies that vowels will not be the first thing in the onset excepting for the beginning of words. Additionally, there are may words that can be stroked in one go even so ("able", for example). Do you have a formulated plan with this concept in mind? How would you account for all the possible sounds?
Message has been deleted
Mirabai Knight
Jan 22, 2016, 8:57:10 PM1/22/16
to ploversteno
Always worthwhile to mention Palantype layout (the other English chord-based layout that is used for professional captioning) in these discussions.
https://silenceteller.files.wordpress.com/2010/08/apalankeyboarddiagram.jpg
On Jan 22, 2016 8:54 PM, "Hanchul Park" <lhu...@gmail.com> wrote:
The idea of initial vowels can be found at https://groups.google.com/forum/#!searchin/ploversteno/orthography/ploversteno/Td8WcHyx0nw/UnGQQZdGCwAJ. It is very true initial vowels are less frequent than final -s or -ed, whenever it is available it can reduce the number of strokes. Alive, asleep, assign, abandon for example. It also can be used for many briefs not currently possible.
It is just an example of possibility of additional keys. Once every steno machine has more keys, I am certain many stenographers will find their use.
2016년 1월 23일 토요일 오전 5시 17분 13초 UTC+9, Steven Tammen 님의 말:
Hanchul Park
Jan 22, 2016, 9:09:55 PM1/22/16
to Plover
* I deleted my post to edit it but Mirabai replied it before that. Sorry for that.
The idea of initial vowels can be found at https://groups.google.com/forum/#!searchin/ploversteno/orthography/ploversteno/Td8WcHyx0nw/UnGQQZdGCwAJ. It is very true initial vowels are less frequent than final -s or -ed, but whenever it is available it can reduce the number of strokes. Alive, asleep, assign, abandon for example. It also can be used for many briefs not currently possible. But its best use would be the article "a" before many nouns.
The idea of initial vowels can be found at https://groups.google.com/forum/#!searchin/ploversteno/orthography/ploversteno/Td8WcHyx0nw/UnGQQZdGCwAJ. It is very true initial vowels are less frequent than final -s or -ed, but whenever it is available it can reduce the number of strokes. Alive, asleep, assign, abandon for example. It also can be used for many briefs not currently possible. But its best use would be the article "a" before many nouns.
It is just an example of possibility of additional keys. Once every steno machine has more keys, I am certain many stenographers will find their use.
2016년 1월 23일 토요일 오전 5시 17분 13초 UTC+9, Steven Tammen 님의 말:
Steven Tammen
Jan 22, 2016, 9:22:53 PM1/22/16
to Plover
For writers that support the split S, do you think it would be better to have the A on the bottom key, say, of that? Assuming one more full column, that would give us three keys to work with initial vowels, and we could expand it even to more than just A. Working with this concept we could have the three keys, A+S, the chord of the other two, S+the top one, A+bottom one, and all four if that would be feasible to push down with your pinky. That gives 8 different combinations for initial vowel sounds!
This sort of thing is what I love doing. Although I think it will be hard to come up with frequencies for which vowel sounds come up in the onset, such an analysis and layout design could open the door to a whole bunch of new natural one-stroke words, as well as the opportunity for many new briefs. Does the sound of this interest you?
This sort of thing is what I love doing. Although I think it will be hard to come up with frequencies for which vowel sounds come up in the onset, such an analysis and layout design could open the door to a whole bunch of new natural one-stroke words, as well as the opportunity for many new briefs. Does the sound of this interest you?
Gabriel Holmes
Jan 22, 2016, 11:12:24 PM1/22/16
to Plover
I'm telling ya, man, someday there are going to be PhD theses in linguistics about this stuff.
Reply all
Reply to author
Forward
0 new messages