Progress report on Digid (keymap based on 5-bit visual chording diagrams rather than a “home row” like the default mapping or single key mappings).
Two things stand out:
o All MCC characters and “,”, “’”, “?”, , , , and are keyed into a single column is critical to my speed, and
o I have mastered, and get a lot of extra speed from, almost all the 50 bigrams and trigrams I mapped, and I am surprised at how consistently I remember to use them.
The way my mind works, I visualize my finger placement as I memorize, therefore my cognitive load is LOWER rather than higher for more complex chords:
o I definitely feel more confident remembering multi-stroke strings like “ck” (in my head I see “SCC then 2CC on rows 2-3 of left column,” which is an easily visualized pattern for oLoo->oLLo).
o This is the same phenomenon as when we read; we do not read individual words, but instead remember long and distinctive phrases we encounter often.
o My most common errors and the ones I have the hardest time eradicating are short, symmetrical SCCs, especially confusing “d” and “n” (both are single characters centered in the right column, ooRo and oRoo) and “a” for “e” (ooMo and oMoo).
o Short common 2CCs are also easy to reverse: “s” and “r” (2CCs on rows 3-4, and ooMM). In an article I typed on writing I spelled it “wsiting” or “wsitesr” practically all the way through.
If I use a bigram or trigram MCC that is not derived directly from the characters (i.e., “ir” is like “I+,” and “en” is like “e+f” and “com” is like “c-f-a”) I get disoriented if I immediately repeat the final character “r”, “n” or “m”, which I have to remember has suddenly “moved” back to its familiar location.
- I started by using a descriptive graphic, the digigram, to determine how to map the characters. I lucked out, since this allowed me to map almost all the common bigrams and trigrams:
o Because of the way odd-even character numberings work with my graphic, the most common consonants were on the right column and all the vowels on the middle column;
o Simpler graphic symbols for the more common letters led to simpler keyings -- only “z,” “v,” etc., led to “trumpet” fingerings;
o The left column, which I had to use for punctuation and directional key mappings, is where “cgkqw” ended up;
o The saddest casualties were the surprisingly common combinations with “c” and “L” (oRRo, which cannot combine with “A” ooMo or “e” oMoo) but we can’t have everything.
o Folks, ignore all the research that says “H” is a common character. Eliminate the common bigrams and trigrams and you seldom run into it.
o “J,” RRRR, is just painful for me to key with my left hand. But my graphic logic said that is where this seldom-used letter had to go. There IS a god.
My physical dexterity is apparently low compared to others on this forum, especially in the right column (which requires me to curl my left fingers most tightly):
o Stutter, hitting a character twice, is my second most common error.
o By having mapped most of the possible MCCs, I almost always encounter some output when I hit the wrong keys; the penalty for my low dexterity and 50 bi/trigrams is a lot of backspacing.
o I “remember” errors (i.e. I notice them as soon as I hit the key); if I start out keying slowly then I can correct errors before entering, and it turns out that with experience the errors are only dexterity problems and not cognitive.
o I was lucky that my digigraphs did not lead to a lot of hard-to-strike multicolumn chords like LRLM.
o My third most common error is missing strokes, which are the only errors I do not “remember.”
o Good placement is critical (3CC on right column oRRR).
Miscellaneous
o I exploit striding (see Lance Gatlin’s Typemax) to some degree, and it speeds me up; e.g, by holding the “t” with my pinky I can quickly get “th-at” LooR->ooMR)
o I was lucky with punctuation, especially by concentrating it on the left column or using “r” (**MM) or (***M) as a typical set-key combination.
o For memorizing numbers, it has been easier for me using 1-9 character which set with the letters A through I (ooLM->oRoM->oLoM->ooRM->oMoM->RooM-> LoLM->RoRM->LooM), rather than using more obvious physical numeral layouts (e.g. phone or canonical order)
o Directional MCCs (whose angles feel like they “point” left or right) work well for muscle memory, including arrow controls and bracket/brace/para character pairs.
o I found I got no benefit from including in the bi/trigrams.
o Generally the bigrams either start or end in “a” or “e,” but few duplicate consonants, so it is not hard to decide which multigram combination to use. “Has” as “ha-s” is easier than “h-as,” but this sort of ambiguity seldom comes up.