notebook

Up-and-down more awkward for right-hand-only

So, I switched cold turkey to right-handed Dvorak yesterday, and within an hour discovered that it really is easier to stroke M and F as in the Internet fingerings (see my last posting).

Paint me green and call me Gumby.

The search for one-handed Dvorak fingering

I've been typing Dvorak for two years. I highly recommend it. Perhaps I'll advocate it more in a later posting.

This posting is about fingering for the one-handed Dvorak layouts. I recently decided I wanted to learn one-handed so I can type quickly when I have just one hand free (such as while manipulating books, eating, and walking). I explore what I know of the fingering, and determine to practice the right-handed fingering which makes most sense to me.

While the one-handed Dvorak layouts are easily enabled in Windows, information on fingering is hard to come by and not very authoritative. Authoritativeness matters here not because there is necessarily any one right way to finger, but because Dr. Dvorak went to great lengths to optimise typing, and using his layouts approximately the way he designed them should avail the advantages he designed into them. He was positively obsessed with bettering typing, and it is unlikely anyone will revisit the manifold issues he considered any time soon, so for now the most practical way to optimise typing with standard keyboard equipment is to try to get into his flow.

Randy Cassingham mentions in his 1986 book The Dvorak Keyboard (pages 71-72) that

Dr. Dvorak had written a special one-hand typing manual for Col. Allen which was widely copied and distributed to other one-handers, but the manual has long been out of print. ... In fact, it is very difficult to find any information at all about these special designs.

If anybody ever turns up a copy of that manual, I would surely like to read and publicise it. Absent Dr. Dvorak's manual, we're a bit on our own.

In a couple of places on the Internet, I have found graphic or text descriptions which individuals are promulgating as the best fingerings, typified by Bob Harrell's description.

Left-handed fingering as elsewhere on the Internet
Right-handed fingering as elsewhere on the Internet

I tried these fingerings for a couple of days before suspending my practice because they felt wrong. Several finger reaches seem both difficult and unlike two-handed reaches. In two-handed typing, each finger is used to type the keys NNW and SSE. I call this up-and-down, though it shears a bit left of north.

Two-handed Dvorak fingering (up-and-down)

I felt the reaches should stay the same on one-handed, for two reasons: up-and-down is most comfortable to me, and up-and-down maximally leverages my muscle memory from two-handed typing.

Up-and-down left-handed fingering
Up-and-down right-handed fingering

I decided to look up Dr. Dvorak's 1936 Typing Behavior to see if I could divine his later one-handed fingerings from the data he considered in laying out the his two-handed keyboard. My look is first-approximation: I consider only single-letter frequency, neglecting digraphs and error rates and other matters I won't allocate time to study.

Here are the letter frequencies Dr. Dvorak gives on page 347 of his book:

                                                                                                                                                                 

letter	count
E	43051	12.67%
T	35096	10.33%
O	29623	8.72%
A	27558	8.11%
H	23434	6.90%
N	22080	6.50%
I	21896	6.45%
R	20010	5.89%
S	16919	4.98%
L	13104	3.86%
D	12075	3.55%
U	9968	2.93%
W	9479	2.79%
F	9246	2.72%
Y	9040	2.66%
M	8289	2.44%
C	7334	2.16%
B	5408	1.59%
G	4607	1.36%
P	4519	1.33%
V	3441	1.01%
K	2440	0.72%
J	437	0.13%
X	420	0.12%
Q	116	0.03%
Z	81	0.02%

Dr. Dvorak labels the fingers as L-1 for the left index, etc. I'll use the same labels.

The first question I had regarding the one-handed layouts is which the home keys for each hand are. They must be on the same row as in two-handed typing — reaching two rows down would make about as little sense as reaching three rows up. The top row of each one-handed layout contains 10% of strokes (letters); the next row 27; the home row 44; and the bottom row 18.

My next question was which key to strike with the index finger. One guesses E, but just to verify, I summed up the finger loads for L-1 and R-1 on each A, E, and H, assuming up-and-down stroking. For the left hand, they are 17%/28/22/33, 45/22/19/15, and 67/19/9/6. Obviously only L-1 on E provides balanced loads (taking into consideration the strength and skill of the index finger relative to the others). For the right hand, they are 8%/28/23/41, 36/23/21/20, and 60/21/10/10. Again, only R-1 on E provides balanced loads.

Then I compared the loads of each up-and-down fingering and the fingerings I found on the Internet, charted below:

For the left hand, the above chart shows the Internet fingering a bit unbalanced, with L-2 doing less work than its neighbours while it is a stronger finger. For the right hand, it could be a toss-up as the differences between loads are minor. I would however consider the left- and right-hand fingerings in concert, my thinking being that logic behind good fingering should apply just as well to both hands, so the inversion in performance of the left-handed Internet fingering does not inspire confidence in me that the right-handed Internet fingering would be most sensible.

I do want to move forward with learning one-handed typing. The right-handed Internet and up-and-down fingerings differ only in the fingers used for M and F, which together account for 5% of strokes. The left-handed Internet and up-and-down fingerings differ in the fingers used for F, M, L, J, U, R and S, or 23% of strokes. For now I am going to practice right-hand typing using up-and-down fingering. It seems the safer bet and I am right-handed, anyway. If I later try the Internet fingering, I will only have to relearn M and F. With some more experience maybe I will be able to offer a more qualified opinion on left-handed fingering, especially after feeling how mirroring versus reaching works out.

If anyone has materials or insight to share, please comment or trackback.