Thursday, November 20, 2008
An excerpt from an email to an friend:
My curiosity got the best of me and thinking about keyboard layouts, and since I hadn't written any programs in a while, I figured I'd look into the real differences between Qwerty and Dvorak, without giving any judgment about the results.
A few definitions:
Home row words: All the letters are on row 2, that is, where the home keys are.
One row words: All the letters are the same row.
Stretch: A stretch of 2 key rows between two adjacent letters in a word, e.g. "m" and "o" on a Qwerty keyboard.
Stack: When two adjacent letters in a word are keyed by the same finger, e.g. "e" and "d" on a Qwerty keyboard. I do take into account that the index fingers do double duty with columns 5 and 6. I do not account for a little hand shift that might unstack the letters as for "h" and "u" on a Qwerty keyboard.
I ran the analysis on the first 10 chapters of Joyce's Ulysses (taken from this site: http://www.online-literature.com/james_joyce/ulysses/) all pasted together in a single file. Non-letter characters were changed to white space, all letters were converted to lower case.
And the counts came out like this:
$ cat ulysses.txt perl keyanal.pl
Dvorak results:
Total words (12984) and chars (388403)
One row words (627) and chars (76495)
Home row words (564) and chars (75442)
Stretch words (1907) and instances (2109)
Stacked words (2202) and instances (2413)
Qwerty results:
Total words (12984) and chars (388403)
One row words (287) and chars (25529)
Home row words (60) and chars (7406)
Stretch words (8095) and instances (14875)
Stacked words (4975) and instances (6333)
The differences in the numbers do seem reflected in the way the typing feels, something subjective for sure.
My curiosity got the best of me and thinking about keyboard layouts, and since I hadn't written any programs in a while, I figured I'd look into the real differences between Qwerty and Dvorak, without giving any judgment about the results.
A few definitions:
Home row words: All the letters are on row 2, that is, where the home keys are.
One row words: All the letters are the same row.
Stretch: A stretch of 2 key rows between two adjacent letters in a word, e.g. "m" and "o" on a Qwerty keyboard.
Stack: When two adjacent letters in a word are keyed by the same finger, e.g. "e" and "d" on a Qwerty keyboard. I do take into account that the index fingers do double duty with columns 5 and 6. I do not account for a little hand shift that might unstack the letters as for "h" and "u" on a Qwerty keyboard.
I ran the analysis on the first 10 chapters of Joyce's Ulysses (taken from this site: http://www.online-literature.com/james_joyce/ulysses/) all pasted together in a single file. Non-letter characters were changed to white space, all letters were converted to lower case.
And the counts came out like this:
$ cat ulysses.txt perl keyanal.pl
Dvorak results:
Total words (12984) and chars (388403)
One row words (627) and chars (76495)
Home row words (564) and chars (75442)
Stretch words (1907) and instances (2109)
Stacked words (2202) and instances (2413)
Qwerty results:
Total words (12984) and chars (388403)
One row words (287) and chars (25529)
Home row words (60) and chars (7406)
Stretch words (8095) and instances (14875)
Stacked words (4975) and instances (6333)
The differences in the numbers do seem reflected in the way the typing feels, something subjective for sure.
# posted by goldpython @ 8:07 AM 
