An interesting read for those of you looking at designing your own custom layouts, or if you just like big data analysis:
English Letter Frequency Counts: Mayzner Revisited by Peter Norvig (a Director of Research at Google Inc.), published in January 2013
--Chris
Topics: bigrams, n-grams, letter frequency, word counts/lengths
Indeed, that reference is both thorough and well presented.
*** Learn Colemak in 2–5 steps with Tarmak! ***
*** Check out my Big Bag of Keyboard Tricks for Win/Linux/TMK... ***
This refutes a common criticism of Colemak that H deserves to be in the home position because 'the' is the most common word in English or because of "ETAOIN SHRDLU"
Oooh, the n-gram table would make for a good workout.
--
Physicians deafen our ears with the Honorificabilitudinitatibus of their heavenly Panacaea, their sovereign Guiacum.
This refutes a common criticism of Colemak that H deserves to be in the home position because 'the' is the most common word in English or because of "ETAOIN SHRDLU"
Not such good news for Workman then!
The order I have generated, from the same list out-of-copyright books the carpalx guy uses, is this (including symbols):
E T A O I N H S R D L U M W C F Y G , P B . V K ' ; ? J X Q : Z
Peter Norvig's frequency table looks interesting, seemingly being based on such a wide selection of material, I would be interested in trying it out, but the problem is it doesn't include common non-alphabetic symbols. I would at least want to simulate the main punctuation keys eg comma, dot, semicolon/colon.
Using Colemak-DH with Seniply.
The home position alone is no guarantee for a good THE experience anyway. The TH and HE bigrams are important of course, but this can be solved in various ways. I really like these bi- and trigrams in Colemak (especially with the DH-mod).
*** Learn Colemak in 2–5 steps with Tarmak! ***
*** Check out my Big Bag of Keyboard Tricks for Win/Linux/TMK... ***
Indeed, regardless of whether R or H is the slightly more frequent, they are pretty close and only one of them can get a place in the top 8. I instinctively feel Colemak's choice of having R on the home row is correct. But the main thing is, whichever one loses out needs to get a next-best slot, which of course is the point of Mod DH.
I started looking at the Peter Norvig bigrams data in a bit more detail, comparing it to the data I have been using previously (which I'll call the carpalx data). Interestingly, the order of same-finger bigrams occurrence is also noticably different, although the overall totals are pretty much the same. I have excluded punctuation bigrams because the Norvig data doesn't have them.
With Norvig bigram frequencies, I find the most frequent Colemak same-finger bigrams are:
SC 0.1547%
UE 0.1475%
PT 0.1058%
NL 0.0638%
NK 0.0516%
KN 0.0514%
EU 0.0312%
DG 0.0310%
WR 0.0308%
YI 0.0288%
Whereas the carpalx data gives:
KN 0.1124%
UE 0.1110%
SC 0.1025%
NK 0.0941%
NL 0.0840%
PT 0.0727%
LK 0.0397%
YI 0.0393%
LM 0.0366%
WR 0.0329%
The hand balance is also a little different (letter keys only):
Norvig: L: 48.34% R: 51.66%
carpalx: L: 46.96% R: 53.04%
Not sure there is any major conclusion to be drawn, but thought I'd post the info anyway.
Using Colemak-DH with Seniply.
