garye@hp-ptp.HP.COM (Gary_Ericson) (06/10/89)
I have been thinking through the differences in creating text between typing on
a keyboard and writing letters by hand (e.g., pencil and paper). I know that
typing is considered generally faster, but I think it depends on the setup.
I'm wondering what studies might be available that could fill in the matrix
below. I'm concerned with execution times for the *average* person in each
category, not the maximum (expert) or minimum (novice) times.
My theory is that typing speed in a bad environment (when you can't touch-type
efficiently) approaches the speed of handwriting, especially when you get into
handheld sizes. I would say that the notion that typing is always faster than
handwriting applies only to when both are done in optimal environments. In bad
environments, I'd wonder if handwriting would win or tie.
The keyboard matrix consists of 4 types of layout vs. 3 types of keys vs.
3 sizes. In each combination (36 total), the speed would be marked for 2 kinds
of typists. Here's the beginning of the matrix:
LAYOUT KEYS SIZE touch-typist non-typist
----------- ------------ -------- ------------ ----------
QWERTY standard full ? ?
QWERTY standard half ? ?
QWERTY standard handheld ? ?
QWERTY membrane full ? ?
QWERTY membrane half ? ?
QWERTY membrane handheld ? ?
QWERTY touchscreen full ? ?
QWERTY touchscreen half ? ?
QWERTY touchscreen handheld ? ?
alphabetic etc. etc. etc. etc.
Dvorak etc. etc. etc. etc.
MicroWriter etc. etc. etc. etc.
Definitions:
MicroWriter layout = a grouping of 5 keys played in "chords"
standard keys = physical keys that move when depressed
membrane keys = physically outlined, but are flat
touchscreen keys = flat and have no physical outline
full-size = the standard computer keyboard
half-size = something other than standard, i.e., the keys are a
little smaller and pushed closer together (partially
defeating touch-typing)
handheld = very small, e.g., calculator, Casio, Sharp Wizard
The handwriting matrix would be smaller:
HANDWRITING desktop surface handheld surface
----------- --------------- ----------------
printing ? ?
cursive ? ?
shorthand ? ?
Does anyone have any pointers to information that would begin to fill in these
numbers? Anyone have an opinion about my theory?
Gary Ericson - System Engineering, Workstation Technology Division
phone: (408)746-5098 mailstop: 101N email: gary@hpdsla9.hp.comgln@arizona.edu (Gary L. Newell) (06/12/89)
I think that your theory is incorrect. Average writing speeds for English are 1.5-2.5 characters per second (with peak rates approaching 5-10 characters a second for things like a sequence of 1's). The only time that handwriting interfaces would improve upon keyboard entry is for large alphabet languages like Chinese or for applications using a small set of gestures or math. symbols. Of course, if you intend a user population which is unfamiliar with or simply not very good using keyboards - then I guess that it might be usefull - say for inventory updates on the warehouse floor or the factory floor (something hand held...). IBM has done work recently on Gestural interfaces and has a variety of papers on the subject. One in particular 'ON-Line Handwriting Recognition - A Survey" could probably give you a good feel for the current trends in this area. I do think that certain applications lend themselves to handwritten interfaces assuming that many of the hardware problems are overcome. The use of a VERY thin transparent tablet over a flat display comes to mind for example..... It also seems a reasonable optional entry mode for almost any interface. Single gestures can indicate target, scope, and action for many commands in applications like editting - much better than most keyboard designs. Oh well, I could go on with this topic for a while but it gets boring I guess..... gln
lambert@cwi.nl (Lambert Meertens) (06/13/89)
In article <1440001@hp-ptp.HP.COM> garye@hp-ptp.HP.COM (Gary_Ericson) writes: ) My theory is that typing speed in a bad environment (when you can't touch-type ) efficiently) approaches the speed of handwriting, especially when you get into ) handheld sizes. I would say that the notion that typing is always faster than ) handwriting applies only to when both are done in optimal environments. In ) bad environments, I'd wonder if handwriting would win or tie. In article <11581@megaron.arizona.edu> gln@arizona.edu (Gary L. Newell) reacts: )) I think that your theory is incorrect. Average writing speeds for English )) are 1.5-2.5 characters per second (with peak rates approaching 5-10 )) characters a second for things like a sequence of 1's). How does it follow that Gary_E's theory is incorrect? I would think the major objection is that it is unfalsifiable in the Popperian sense, since if typing speed far exceeds writing speed in a certain environment, defenders of the theory could easily maintain that that particular environment is not bad. I am not a touch-typist; in spite of concerted efforts to become one I would keep typing things like "exrtaordianrily". Compounded with "shift over" errors, counting the time needed for making the corrections, my speed was at the best (after more than a month of only touch-typing) less than ten words per minute. With two fingers (one of which is mainly moving between the shift key and ESC), I get 2.3 characters per second FOR PLAIN TEXT. My hand-writing speed is... 2.3 characters per second. So I am my own portable "bad environment". Much of the stuff I do is anyway riddled with non-alphabetic characters (like in *(c->buf.pos++)), which are differently placed on each of the three keyboards I regularly use and half of which are shifted -- touch-typing is no help there! I have noticed, by the way, that I am not really doing much worse on this kind of text (where every typo must be corrected!) than my touch-typing colleagues. Gary L. N. also states: )) The only time that handwriting interfaces would improve upon keyboard entry )) is for large alphabet languages like Chinese or for applications using a )) small set of gestures or math. symbols. I would approach this differently: Keyboard entry condemns us to stay bound to a small set of symbols, which may be OK for some applications, but which blocks development in many areas: engineering, mathematics, chemistry. If a good handwriting interface were available, I would use it immediately, quite possibly exclusively. -- --Lambert Meertens, CWI, Amsterdam; lambert@cwi.nl
norman@cogsci.ucsd.EDU (Donald A Norman-UCSD Cog Sci Dept) (06/13/89)
Typing versus handwriting: why the "versus," why not "along with"
Students in the class I teach called, appropriately enough "cognitive
engineering," have performed a number of relevant analyses, that
although not of publishable quality, I suspect are valid.
Over the years, several groups in my class have proposed hand-held
portable computers or note-taking devices. Each has proposed using
both handwriting and typed input (with attached or optional
keyboards). Their studies of handwriting speeds lead me to the
following conclusions:
1. There is a very sizable proportion of the population* that PREFERS
handwriting. Some proportion prefers keyboards. NOTE: the
preferences are often very strong and are NOT always related to
performance data. That is, take some who strongly prefer input method
A and demonstrate that they are actually faster with method B, they
will not change their preference. Speed is not the only variable of
concern, and often is a secondary variable.
..................................................................
* The population studied was mostly college students, but included
some professional engineers and management.
..................................................................
2. If the device is hand held, so that one hand is occupied supporting
the device and only one hand is available for input, for a large class
of people, handwriting is faster than typing.
Note that good typists can still type more quickly using only one hand
than they can write: I have heard of a typist who only has
one hand who can go extremely fast (I can't recall the exact figure,
but 80 wpm comes to mind). This is rare, but it shows that training
can compensate, and that one-handed typing is not necessarily slow.
Nonetheless, many people can handwrite FASTER than they can type if
restricted to one hand.
3. A number of years ago I compared novice** typists on a variety of
keyboards: qwerty, alphabetic, dvorak, random. Performance was about
the same on all of them. Alphabetic arangement did not help and
random arrangement did not hurt. And any knowledge of qwerty helped
qwerty, so that arrangement was superior.
...................................................................
** We tried to get people with no experience at typing, but in our
environment, this seemed impossible. So we settled for those who
claimed they could not type, and whose initial performance was very
very minimal. (Norman & Fisher, Human Factors Journal. A long time
ago.)
...................................................................
We did not measure handwriting speed, but the novices typed around 10
wpm, which is a little faster than claimed for handwriting.
4. There are specialized input devices that are faster than qwerty
(or even Dvorak). Chord keyboards, such as court stenographers use,
are better. Note that part of the speed from a court stenographer
results from the fact that they type a shorthand -- the keys produce
sylables, not letters, and often only the typist can read their own
output (each stenographer has idiosyncratic abbreviations). Modern
systems feed the keyboard to a computer that, in principle, can
produce printed text on-line.
The learning time on these systems is huge: months and years of
practice are required. Not for everyday folks.
I have also used other chord systems, some one-handed. They may work
well for special cases, for those who are willing to take the time to
learn them, and then take the time to stay in training. But I do not
believe these will ever be popular among the general public.
---
What do I conclude? That each input device has its place. With
regard to the original question of keyboard versus handwriting, it
would be a mistake to market a note-taking device with only tablet
(handwriting) or only keyboard input. Sizable proportions of the
population prefer each. Moreover, I suspect that any individual would
switch back and forth, depending upon circumstances and the nature of
the input material.
BUT, if you are going to include a keyboard:
1. Make it full size (it can fold up to a small size, but it should
unfold to full size)
2. Make it qwerty. Alphabetically arranged keybaords are an
impediment to most (e.g., the otherwise virtuous Sharp Wizard).
(Please don't restart the Dvorak wars. Dvorak keyboards are better.
But it is too late.)
don norman
Don Norman INTERNET: dnorman@ucsd.edu
Department of Cognitive Science C-015 BITNET: dnorman@ucsd
University of California, San Diego AppleLink: D.NORMAN
La Jolla, California 92093 USA
[e-mail paths often fail: please give postal address and full e-mail path.]chuck@melmac.harris-atd.com (Chuck Musciano) (06/14/89)
This discussion is quite interesting, although I feel that I am off the
"norm" presented by previous posters. Personally, I can type much faster
than I can write, and I am not a touch typist. I am a four-finger/two-thumb
typist, and sometimes achieve great bursts of speed, as long as I can look
at the keyboard while I type. This makes me very fast when I am typing
thoughts, and horribly slow when I am retyping something else.
I would like to see a computer which could read my handwriting, since a
lot of the time, *I* can't read my writing. My handwriting was never good,
and since I find myself typing almost everything I need to write, my hand-
writing has atrophied to the point where I have a hard time remembering certain
little used cursive letters, like capital Q. If I want to be clear, I print,
which is horribly slow and tedious. When I write out greeting cards and the
like, I usually have hand cramps after about 15 lines of text. Now that I
can reach my parents via e-mail, I don't even need to write letters.
My use of computers has evolved along similar lines. When I started out,
I would write out programs in longhand, and then type them in (on an ASR-33
teletype at 110 baud!). After a while, I would just note down psuedo-code,
and convert to real code as I typed. Finally, I just started composing code
on the fly. I haven't put a line of code on paper in about six years.
I compose all my code using top-down decomposition, usually after thinking
about the problem until it just "becomes obvious".
Frankly, anything with only a written interface would be completely
unacceptable to me. I can think of no other single thing which would
hamper my productivity more than having to write, rather than type.
This is on a tangent, but my discussion about my programming techniques
started me thinking (always dangerous :-). When I am faced with a problem
to be solved (requiring some programmed solution) I sort of think about it
initially, and then don't think about it. Sometime later, maybe days, maybe
weeks, the full solution just pops into my head, and I sit down and type all
the code in. If I try to force the issue, and type in the code before I
am ready, the results are usually disastrous. If I am patient, I usually
wind up with an elegant solution.
This often works in a heirarchical manner, too. Often, the initial
solution is the top level structure of the program, particularly in the
case of complex problems. After I get that stubbed out, the lower level
details start filling in, in a similar manner. Even big things I have written,
with many thousands of lines of code, all formed this way.
I suppose my question is, is this a common problem solving method? Do
others have this approach? I don't know if it is the traditional "Aha!"
method, since I can "queue" jobs up in my mind and just have them get solved
in some order at a later date. Any thoughts on this? Am I totally weird?
What sort of interface design criteria might be applicable to take advantage
of this sort of mindset?
Chuck Musciano ARPA : chuck@trantor.harris-atd.com
Harris Corporation Usenet: ...!uunet!x102a!trantor!chuck
PO Box 37, MS 3A/1912 AT&T : (407) 727-6131
Melbourne, FL 32902 FAX : (407) 727-{5118,5227,4004}
Oh yeah, laugh now! But when the millions start pouring in, I'll be the one
at Burger King, sucking down Whoppers at my own private table! --Al Bundythom@dewey.soe.berkeley.edu (Thom Gillespie) (06/14/89)
In article <751@cogsci.ucsd.EDU> norman@cogsci.UUCP (Donald A Norman-UCSD Cog Sci Dept) writes: >Typing versus handwriting: why the "versus," why not "along with" > The one thing which 'writing' allows which 'typing' doesn't is the ability to express ideas without words - with pictures, jottings , diagrams. You can get a lot of information down with just a few well placed arrows and some text. The problem is that the systems we have now don't let us express our information as best we'd like. It is an either or situation but the real either/or is not in terms of just text and words. It should be in terms of text and graphics. I'd like to be able to doddle on this posting but i can't. Text I'd rather 'type', ideas I'd rather 'express'. I need at the least a keyboard and a mouse - and some better soft/hardware. For lack of a better phrase I suggest that there is a learning disablity which I call 'visual dyslexia', the inability ot understand ,manipulate and express ideas with visuals. A lot of people suffer from this disability and it will become more apparent as we move into the area of Hypermedia, text is not as important as we think. > >What do I conclude? That each input device has its place. With >regard to the original question of keyboard versus handwriting, it >would be a mistake to market a note-taking device with only tablet >(handwriting) or only keyboard input. Sizable proportions of the >population prefer each. Moreover, I suspect that any individual would >switch back and forth, depending upon circumstances and the nature of >the input material. > It would be interesting to see how users used this 'writing/keyboard'. I'd guess that you'd see a few pictures. Apple's HyperCard really gives you this sort of capability but not in a convenient form for class note taking. Maybe when the DynaBook mates with HyperCard we'll see the DynaCard and our problems will be solved. --Thom Gillespie
gln@arizona.edu (Gary L. Newell) (06/15/89)
In article <8181@boring.cwi.nl>, lambert@cwi.nl (Lambert Meertens) writes: > If a good handwriting interface were available, I would use it immediately, > quite possibly exclusively. Using a handwriting interface for text-entry would be foolish at best. I don't care whether you are a touch typist or not - it is simply too slow and requires to much screen space and presents a variety of other problems. Unless the interface improves upon keyboards (for example, editting of existing text or manuscripts - clearly a stylus/transp. tablet interface would be superior - again, chinese text entry is another example) for a particular application, it would seem too slow to be useful. > --Lambert Meertens, CWI, Amsterdam; lambert@cwi.nl By the way, perhaps I misunderstood you - but 10 words a minute??? What? Do you type with your toes...???...
gln@arizona.edu (Gary L. Newell) (06/15/89)
In article <2203@trantor.harris-atd.com>, chuck@melmac.harris-atd.com (Chuck Musciano) writes: > I would like to see a computer which could read my handwriting, since a > lot of the time, *I* can't read my writing. My handwriting was never good, > and since I find myself typing almost everything I need to write, my hand- >writing has atrophied to the point where I have a hard time remembering certain > little used cursive letters, like capital Q. If I want to be clear, I print, A few interesting points - first, it isn't clear how recognition is effected by legibility - most recognition methods use stroke counts and direction changes or stroke sequences etc. to recognize characters - it may be that a system could actually recognize 'what you meant' to draw better than you or I. Secodn, cursive recognizers are pretty bad (in my opinion), even most recognizers of printing require that the user print in boxes, and many require that the letters all be uppercase etc.... > which is horribly slow and tedious. When I write out greeting cards and the > like, I usually have hand cramps after about 15 lines of text. Now that I > can reach my parents via e-mail, I don't even need to write letters. This is what I tried to point out in my last posting - a stylus/tablet interface is simply a very poor idea for applications with a large percentage of text entry. Editting, spreadsheets, some formula writing, are better applications for these systems (assuming that human factors problems along with hardware problems are overcome and a reasonable system appears). > My use of computers has evolved along similar lines. When I started out, > I would write out programs in longhand, and then type them in (on an ASR-33 > teletype at 110 baud!). After a while, I would just note down psuedo-code, > and convert to real code as I typed. Finally, I just started composing code > on the fly. I haven't put a line of code on paper in about six years. This seems to be more common than I thought - I used this 'creative method' myself and it seems that during a design phase or initial creation, it is common for people to use pen/paper due to a variety of properties - it seems that gestural interfaces may also be effective in this area. gln