In the future… a new Gregg!

I was reading the uptake of shorthand thread about how we need to modernize the system and I came across an idea– a vision dare I say!

The pda and tablet market are prime for a unique text entry system. When I’m rich and influential in 10 years (as we all are going to be) I want to hire a software engineer and a linguist and we’ll collaborate to faithfully modify Gregg to be more suitable for computer recognition. We will write a book and sell the software for pda and computer. The system will retain its fluidity and hopefully will be just as suitable for normal writing as it is for writing on a touch screen.

This is a grand vision, but its an idea that I cannot imagine will ever escape me.

I’d like to know from people here their thoughts of Gregg for computer recognition. Specifically, how one achieves to influence a society. To take an idea to reality, and to spread that reality as courageously as John R Gregg spread Gregg in the beginning. It would take some serious determination to push something like this onto the world, but I believe that its a worthwhile endeavor!


(by michael_lisitsa for


31 comments Add yours
  1. My first thought is we don't need yet another system of shorthand. My input in the other thread was aimed at making the teaching materials for the present system more appropriate for what I perceive to be the current market.

    You've certainly put your finger on one aspect of why shorthand in particular and Gregg in general has faded. Gregg himself was a tireless promoter, and the decline of shorthand correlated with the decline of Gregg himself (yeah, yeah, there were lots of other factors–and a younger Gregg might well have adapted to the new realities). And any attempt to influence a society, to take an idea to reality, is going to take enormous determination. So you're taking on an enormous project.

    I haven't entirely caught the gist of your vision. I'd like to hear more. What I'm getting is that you want a system which can be readily input to a PDA or tablet pc. In other words, a _much_ faster tablet data input. People would use a shorthand system to write on their pda/table, the system would recognized and put the text on the screen. It would be similar to the current graffiti systems. Am I getting the idea, and if not, tell us more!

    My next thought, as a mathematician and software engineer (for longer than I want to think about) is that Pitman would be better suited for computer recognition. Computer recognition of shorthand suffers from all the difficulties of computer recognition of any sort of longhand. It's a challenge to do optical character recognition on printed/typed material. But with that material at least every character is uniform. Once your system recognizes the characters reliably, you're home free. Handwriting is much more difficult because the characters are never uniform, joins between characters are highly variable, people get sloppy with writing, and so on. Gregg suffers from the same problems. You can't just recognize characters. Again, it's a first impression, but the best bet for a design would recognize on an outline by outline basis. My impression of Pitman is that the characters are more consistently joined, but all the problems of idiosyncratic writing remain.

    For the PDA/tablet input, users do seem to accept the limitations of the graffiti, and might — or might not — be willing to learn a system of shorthand that would make the task easier when taking notes in the field. Your system would need to recognize all the abbreviations and brief forms and convert them (from context) into readable text. But I really don't see it as being an incredibly difficult problem. It would take a lot of coding — not going to be a weekend project by any means.

    As we talk about generating new teaching and reading material in hard copy, the computer could really aid in that. There's the same problem that you just can't type Gregg. The joins between the characters are too variable. Maybe you could design a rules based program which would adjust the joins between characters appropriately. But I'd still start with the premise of writing by outline of the whole word. In this day and age memory is simply not a problem. So we undertake the task of scanning the entire Gregg dictionary (pick an initial system), ocr the text, extract images of each outline, identify the line of writing, and store it in an indexed array. From there it's a fairly straightforward task to read from a text file, match each word against the dictionary, and insert the associated image(adjusted for line position) into the output file. Words that didn't have a match would just be inserted as text — sorry about that — later on someone would have to add the missing words to the dictionary.

    And again, not conceptually difficult, but not a weekend project.

  2. There is different ways of trying to get Gregg popular again. I believe that new material and marketing it to school systems etc. could work.

    Another way, which is what I was thinking, would be to introduce it through the PDA tablet market because thats a market where there is a serious shortage of a fast writing alternative. The problem of marketing it in the same way as it has always been done, is that people will still come up with the same excuses (tape recorders, no application in the job market), to not learn it. You might go to all the trouble of rewriting a book only to be knocked back with the same reasons that made the system fail during Series 90 and afterwards.

    I agree, it is not a weekend project. In fact I was thinking that it would be a serious project which would need several professionals and several years. You say that its not incredibly difficult, but one might think that Handwriting recognition is not super complex but look how many years and how many professionals have dedicated themselves to making it work and Windows Vista even doesn't do the best job.

    With my limited knowledge, I was thinking that because shorthand is one continuous line with (usually) clearly defined changes at the boundary, then it is easier for the software to follow along.

    For example: You start with a 't' stroke. The computer recognizes a positive sloping line with a reasonable constant gradient. What could happen next:

    Small circle – It would look for a 360 change in gradient and an intersection of the line.
    Big Circle – ditto but it would look for a larger average radius of the circle.
    'n' – It would look for an obtuse angle and a new line with a constant 0 gradient
    'o' hook – etc etc

    Its definitely not as simple as above, and perhaps Pitman might be easier to recognize, but I think it is possible and John you seem to agree.

    The problem with Pitman is that it has the inherent disadvantage of being UGLY. It is true! It just doesn't look so nice, so I doubt people will want to go back to a system that they had already shunned 100 years ago.

  3. The newer tablet PC's claim to have handwriting recognition, here's a Microsoft article   A lot has been done.  Microsoft, of course, will view all its work as proprietary.  I need to research what might be available open source.  The tools and techniques which work for cursive would provide a guide to what might be done with a shorthand system.   I was doing some work the other day with error correction codes, and it occurred to me that there are parallels with shorthand and cursive.  For those not into digital communication, when your computer sends data somewhere it adds bits to the data which essentially provide extra data that the receiving end can use to determine if an error occurred and if so make an attempt at correcting it.  Whole textbooks have been written on error correcting codes.  But the essential concept is that if you send only the bits necessary for the data, if some of those bits get corrupted the data is lost.  But if you add additional bits, they can be used to correct errors.   Similarly with handwriting.  Normal cursive has lots of extra information which the reader (human or machine) can use to figure out the message.  All those extra loops and lines and curls provide lots of extra clues.  By definition, any shorthand system whether alphabetic or symbolic eliminates  those extra clues because they take time to write.  So a shorthand recognition system is going to have to rely on the user to be careful about forming characters.  I think that's stating the obvious, but it at least gave me a context of why.   A major challenge to my idea of using Pitman is that tablet PC's have no way of recognizing stroke thickness — well, the user could write with an italic stylus I suppose.  But I think that simply brings us back to Gregg.   The article in the URL speaks of a neural net approach.  I haven't worked very much with the concept, but I suspect it's going to be the way to go.   This is a way cool project!     I'm afraid that in the cool, business oriented light of day it's going to be hard to sell to a major manufacturer.  But it could be a really fun open source project.      

  4. Thinking about it, this endeavor has two parts: (1) recognizing the strokes into letters, and (2) transcription of the letters into words or phrases. I believe that the difficulty here is not (1), but (2). Handwriting recognition software can recognize cursive letters, so I believe that recognizing Gregg strokes won't be difficult to program. One could also "train" the computer to recognize the strokes, for example, in training mode, you could write and m vs an n, a t vs a d, etc., or write simple words (like day, may, knee, etc.), so that the program has an idea of proportion. Part (2) would be the difficult aspect, as the program would have to select or show you different options for the transcription of strokes, or build in some intelligence as to which is the better match. I believe it is doable, but it would probably look like some of the translation engines that we have available right now on the web (Babelfish, etc.), which sometimes give translations that are a little off. (Though I would be happy if the program were to do something like that!)

  5. "The problem with Pitman is that it has the inherent disadvantage of being UGLY. It is true! It just doesn't look so nice, so I doubt people will want to go back to a system that they had already shunned 100 years ago."

    Actually, Michael, Pitman is alive and well–and taught in courses–in the Commonwealth.

    I respect your opinion about Pitman's beauty (or lack of it). When I was growing up in the. 70s, Gregg was the method taught in schools; since I had written a zillion term papers in cursive, I thought Gregg, looking too much like cursive, looked boring. When I saw Pitman, I thought, "Wow! This looks groovy!" (It was the 70s). Pitman looked like some strange, esoteric language from another planet; I was determined to learn it.

    While I have never finished my Pitman studies, I still think it looks other-worldly. Surely taste in shorthand beauty is subjective…

  6. George, good to have you actively contributing again. Since I'm about to retire I'm holding off on Pitman studies until after that fateful day dawns, But your comment about the appearance of Pitman made me wonder: Would a DNA test on Sir Isaac's remains reveal alien genes?
    Good-naturedly yours!

  7. Well I guess it might appeal to the block letter students these days. It irritates me when some girls gets a compliment on her handwriting– handwriting which is written really small, entirely by the fingers, disjointed and with no sense of flow whatsoever. You've seen it before, its that uniform writing which is nearly as easy as text to read, but that has the same personality as a block of text.   Of course Pitman appeals to some, but dare I cause more insult, abstract art appeals to some as well. At our university a new work of abstract art (a sink stuck into a wall) was praised for its aesthetic beauty. To me it looks like a wash basin, glued to a wall, not art!   At the same time, I at one time or another will consider learning Teeline because I would like to see how a system based on alphabet adapts to fast writing. It would surely make it easier for the computer to work with a choice of letters than a choice of sounds — bringing Chucks argument that 'recognizing' strokes is not the major problem, but converting this information into words is!  

  8. I agree that Gregg would be much harder than Pitman. Gregg varies a lot. A simple "right hook" can be at several angles, depending what letter it came off. I've seen "N" in some books at a 15 degree angle.

    Then we have the penmanship issues. Circles are rarely true circles. The circle in TA is not a full 360 — it stops at right angles to the T (if you write as the text says — otherwise all bets are off).

    Pitman texts look like a robot made the marks. The circles are full circles. The lines are on specific angles and exactly one of three lengths. The curves are parts of a clock. Pitman can probably be done by looking at outlines as a whole.

    Gregg, on the other hand, has to be traced. When I was learning to read Gregg, DJS, it took a while for me to realize I should trace the outlines to distinguish between D and J. The book didn't talk about left/right motion for O/U; I'm not sure what terms they used, but I had to figure it out myself when they talked about turning it on it's side for "noon".

    I have a PDA, and find Graffiti frustrating. It keeps reading the wrong letter. I replaced the screen, which helped somewhat, but when I do a lot of typing I use the pop-up keyboard. My current one, admittedly 2001, doesn't like fast writing.

    John's outline-level database would work for text-to-shorthand. Writing the program would be a weekend or two. The database? Ugh. Also add something for phrases. If the word is in this subset, look at next few words.

    Tablet PCs sometimes have weight detection. I have some artist friends who paid a few hundred for pads that have enough weight discrimination to keep them happy, and there are much more expensive ones out there. So the technology exists and will come down in price.

    Good point about Teeline. That may be the one to go with. It's actively taught. One thickness. Clear and consistent angles and lengths, although that varies with writer. One downside is that writers are encouraged to customize it almost from day one. Faster to learn at first, but it maxes at about 140.

    I was just telling Husband about $500 for a book in another thread. He said things about chicken stratchings. He's slowly learning. Teeline looks like chicken scratchings. Pitman looks like chicken peckings. Gregg looks like chickens ice dancing.


  9. "Of course Pitman appeals to some, but dare I cause more insult, abstract art appeals to some as well. At our university a new work of abstract art (a sink stuck into a wall) was praised for its aesthetic beauty. To me it looks like a wash basin, glued to a wall, not art!"

    Of course, you are quite right, Michaell–aesthetics is completely subjective.

    Michael, after learning the superior Gregg, I think you would be disappointed with Teeline. From what I've heard, it's just far too subjective a system; year-old texts are difficult to read even to the person who wrote it.

    My understanding is that you're learning Anniversary Gregg, which I believe is the apogee of the system. You can fly to the moon with that one; I think you have what it takes to get there. 🙂 I'd love to be able to hear of a Gregg writer who can write over 200 wpm.

    If I might digress for a moment, I'd like to ask you a question. According to your profile, you're from Australia–a Pitman bastion. How did you get exposure to Gregg? (The only place in the world Gregg completely dominated was the United States.)

  10. "Well I guess it might appeal to the block letter students these days"

    Michael, I think Gregg would appeal more to today's students, because they're not so used to seeing cursive! (It would look as esoteric to these students as Pitman looked to me in my day.)

    (Spoken conspiratorially) After writing a zillion words in cursive in the 70s, it's hard for me to believe the fact that some students nowadays aren't even TAUGHT cursive–but it's the truth, at least here in the States. Good God, what is this world coming to? 😉

  11. Cricket: "I agree that Gregg would be much harder than Pitman."

    Gosh, I don't know about that one, Cricket; Pitman theory–at least, Pitman New Era–is quite difficult for the average Joe. Besides there being a billion rules, there are a billion short forms to memorize. (It's fair to say, however, that written Pitman is considerably easier to read than written Gregg.)

    Of course, I could be wrong. If there is a forum-poster who has learnt both systems, I'd be much obliged to hear from you.

  12. Well the internet is the biggest source of information. I wasn't actually in Australia anyway, I was in Israel surfing the net and randomly thought that it would be cool to be able to write shorthand. I had experimented with alphabetic systems when I was like 16.

    The first system I came across was Teeline. I spent 2 days learning it, but then when I saw the Lords prayer with how much information had been left out, I thought that the system must be too subjective and I moved to Handywrite. Handywrite I spent 5 days learning and finished the whole online instructional manual that the guy had loaded on there. i was really excited about Handywrite — all its promises of cursive writing and its beautiful lines. I succumbed to common sense and decided to learn the founding system that Handywrite was based on — Gregg!

    The whole process was done between me and the internet!

  13. I'll restrict the discussion to text-book writing.

    "It's fair to say, however, that written Pitman is considerably easier to read than written Gregg."

    That's what I meant. Pitman shapes are more consistant than Gregg. A medium horizontal light line on the line is always the same length, depth and angle, no matter what's around it.

    On the other hand, that variable vowel bit (s-t-r) would be a challenge.

    Every new Gregg writer has complained about line length and slope changing depending on what's around it.

    Teeline was invented by a Pitman teacher whose students found all the rules difficult to learn. Things like circle direction and the precision bother new students. I think he threw the baby out with the bathwater.

    Teeline students are encouraged to customize it very early on. I told a Teeline teacher that Gregg students are encouraged to read others' writing, and she was amazed. "Wouldn't that discourage them from customizing it?" and "What if their own version is different?"

    I think that's why Teeline is hard to read. If your personal system changes (or is forgotten), then you don't have a reverse-dictionary to fall back on when reading old notes.

    While advanced Gregg writers do leave out information, we're also given lots of guidance, told, "If in doubt, write it out," and not to create brief forms for rarely-used words.

    It's easier to identify Teeline letters than Gregg letters, at least while learning, but converting those to proper English might be a problem. Also, Teeline has something like 60 — almost every pairing gets a symbol.

    I think a full machine-readable, handwritten shorthand would need to be designed with the goal in mind. We need shapes a machine can identify, and more specific interpretation rules.

  14. CB wrote: Every new Gregg writer has complained about line length and slope changing depending on what's around it.

    This was not true when Gregg was taught in a classroom. None of my classmates complained about the nature of outline construction. From the first week of writing (we were using the Simplified Functional Manual) we traded steno books after writing connected matter and were expected to be able to read our peer's notes. Of course in 1958 there were probably no more than 20 or 21 students in a shorthand class so we received a lot of individual personalized instruction.

    The problem appears to stem from self-study writers who lack and experienced teacher to guide them through the introductory texts. These same learners seem at the commencement of their study to believe shorthand can be absorbed effortlessly. Not so!

    If any of my readers learned cursive writing in elementary school, I'm sure they recall the hours of intensive practice required to acquire handwriting the teacher deemed satisfactory let alone good. Gregg shorthand is of the same nature. To be able to read and write good Gregg does require many hours of study and application.

    It would be great to see Gregg restored with new texts to the public education system, but would today's "instant gratification" students have the patience to actually commit to the hours it would take them to master the system?

  15. " 'It's fair to say, however, that written Pitman is considerably easier to read than written Gregg.' "

    That's what I meant. Pitman shapes are more consistant than Gregg. A medium horizontal light line on the line is always the same length, depth and angle, no matter what's around it"

    I see, Cricket. The first time I missed the point. Thank you for the clarification.

  16. "George, good to have you actively contributing again. Since I'm about to retire I'm holding off on Pitman studies until after that fateful day dawns, But your comment about the appearance of Pitman made me wonder: Would a DNA test on Sir Isaac's remains reveal alien genes?
    Good-naturedly yours!"

    Thank you for giving me a chuckle; thank you again for your kind words.

  17. Well George 200wpm really is reaching for the moon. The Gregg Writer used to advertise when people got the Diamond Medallion for 200wpm and these people were usually employed with shorthand and had attended several years of shorthand tuition.

    I've actually been kinda stuck at 100 – 120wpm for quite a while now. I have been improving here and there but progress is really quite effortsome at this point. I reckon I will make a real dash for the line at my 1 year anniversary (September 15th) since starting shorthand and try to get a bit more speed.

    Recently I have been sent "Gregg Speed Building for Colleges" which is a big book that was used as a textbook for speed building courses in college (obviously). Has anyone on this forum had success with the Speed Building books, I'm only up to page 19!

  18. Thanks. I was just dawdling through it writing everything out not under dictation so your response will change my approach. I need one of those tape recorders that someone suggested a few days ago, or as a temporary measure I'll use my phone!

  19. Yes. I like those books because (1) they reinforce theory (as they will present new examples of word building), (2) they stress phrase building throughout the whole book (phrasing will become natural and automatic), (3) they provide penmanship exercises, and (4) their speed drills are numerous. Here are some recommendations on how to study with the book (from the teacher's manual):

    1. Phrase drills: Each phrase drill should be dictated at a minimum of one phrase per second. Read back, and redictate at a much faster rate.

    2. Phrase letters: These must be read at a high speed before attempting to take them from dictation. The manual recommends writing 3 to 5 copies of the plates, so that the phrases are cemented in the brain. For example, on page 14, letters 8 and 9 should be read in 2 min total. Then, the letters should be dictated at a rate 20 words higher than the student's customary rate on business letters of average difficulty (this should take about 3 min). Read back, and redictate at a much faster rate. The student should write these letters at a rate of 120 wpm with very little repetition. Read back the last portion of the letter.

    3. Building sustained speed (Assignments 4 & 5): Read the material first. Write the preview and master the words. The exercise on derivatives of common words should be written 5 times. For Assignment 4, as a warm up for dictation, letters 10-12 should be dictated as a single take (3 min). Read back portions of the letters (1 min). Redictate, pushing the speed to 120 wpm (2 min). Read back the take (1 min). Read the preview on page 16 rapidly without consulting the key. Dictate the preview three consecutive times, increasing the speed at each time. After this, you will be ready for a 5 min take on page 17-18. Stop at 5 min, whether you have completed the letters or not. Redictate at higher speed, 20 words faster, in half- or one-minute takes.
    Repeat the same procedure for Assignment 5.

    There is a separate section in the manual on how to do the penmanship drills.

    As you can see, it is really an intense book when used as directed. It will definitely push your speed.

  20. One *phrase* per second? I struggle with one *word* per minute! Ah, well, used to be struggling with one sound per minute, so progress is being made.

    Michael, I'm pretty sure Windows comes with a basic TTS system. Not as good as Cepstral, but at your speeds good enough, especially if you preview the text.


  21. As Stewie would respond on "Family Guy":

    Good grief, woman, how can you expect to attain speed and knowledge if you persist in the sleight-of-hand to keep asking fruitless no-single-correct-response questions rather than baring your pen and commencing your much-needed practice?


  22. The units of the Speed Studies are designed to supplement the study of the manual (whether the regular one, or the functional method). They provide additional reading and writing material, introducing new vocabulary and reinforcing common words. Since you are at a learning stage, your primary purpose should be to read the passages fluently, and copy them with the best shorthand possible, reading aloud the passage as you copy. Dictation is only attempted at this stage with material that has been drilled; that is, read and copied, and it should start slowly, with few sentences at the beginning, with the book open so that you don't get discouraged. These initial dictation periods should be very brief (half, one, two minute tops). As you grow more comfortable, you can increase the length of dictation to say 5 or 10 min, at an easy speed for you, and your dependence on the open book will eventually diminish. Eventually, you will be able to take new material. Your ultimate goal for a beginning shorthand course should be to write approximately one word each second (60 wpm) comfortably and perfectly, with no hesitation whatsoever.

  23. Chuck I think you've made the same oversight as I did by confusing "Speed Studies" with "Speed Building". These are two different ranges, Speed Studies to be used while you're learning the system, but Speed Building was I assume intended for a second or third year of shorthand where the theory has been learned and you need to get more confident.

    As someone told me on the forum before: there was a few versions of Speed building. I have "Speed Building for colleges" which is as you can guess for college use when you've already learned shorthand at school. The book is much thicker than the Speed Studies manuals with a lot of dictation material and difficult word drills.

    In response to critical, I can't stand computer voices. I once tried listening to a Project Gutenburg e-book (David Copperfield) and it was torturous. Project Gutenberg has mostly live readers but some books it has computer spoken and it is not worth listening to. No emotion, no intonations.

    Anyway I've ordered a Sony digital notetaker online from America. The ones in store cost too much and you can get smaller memory ones (64MB) very cheap on ebay. Hopefully it is good quality with a nice speaker!

  24. On the topic about inputting Gregg into the computer… I still think it would be an excellent idea to use MSN's draw feature to send Gregg messages back and forth. Of course, you *need* a tablet for this unless you want your outlines to be ungodly ugly 🙂

    I'm considering getting a cheapo tablet this Christmas to help protect myself against tendinitis (especially since my future career depends on my hands!) and would love to chitchat with someone Gregg-style 🙂 It's certainly a heck of a lot faster than snail-mailing letters back and forth!

  25. For resisting any RSI, I can't emphasize enough an ergonomic keyboard and properly fit desk. If you get even a twinge, see a physiotherapist or occupational therapist. Trust me. For three years I had an annoying knot in my neck. Then I damaged my rotator cuff. The muscle problem had blocked a nerve which caused my shoulder muscles to be weak. A nerve all down my arm had seized up– white fire when she twisted it just right. The knot really was a knot, surrounded by lazy muscles, and the other side was way to tight — no way I could stretch just the right muscles by myself. Took under a month to fix it all, another two months to confirm the fix held, and the little twinges I thought were carpal tunnel cleared up as well.

    LOL about "just do it". You're right — questions are a way of avoiding the real work.

    Thanks for the advice on Speed Studies. Sounds like I'm doing things pretty much the best way.

    Finally, the "modern shorthand". I've tried several alternate onscreen keyboards for my PDA. All of them involve "strokes, where you start on one square and move to another, usually four cardinals and four corners. EasyWrite is the most extreme — most letters involve a stroke. Most systems use strokes to modify the letter you start with. E.g., move up to make capital, move right to add space. Start at T and move left for The.


Leave a Reply