Over 40 years ago I wrote a GWBASIC program that sent random 5 letter groups of letters and/or numbers. Like the Farnsworth system (although we had never heard of it before) it asked for a starting and a target speed and adjusted the spacing between letters so that the characters were sent at the target speed but the thinking time between characters was adjusted to give the starting speed. If you then typed the characters as they were sent, the spaces between characters would gradually decrease if you got it right, but if you got it wrong the letter that was incorrect was given a higher weighting so that it occurred more often. After a period of correct input, the weighting would decrease and the speed would start to ramp up again. This either gets your morse up to your typing speed, or in my case, my typing up to my morse speed. Groups of letters could be enabled individually so that someone just starting could begin with EISHTMO before adding additional letters as proficiency increased.
Unfortunately GWBASIC died with the advent of Windows3 and other versions of BASIC did not have the BEEP command that was used, but recently I have rewritten the program in Python and it runs on a Raspberry Pi2 at least up to 50 WPM. Danny, G3XVR -----Original Message----- From: Elecraft [mailto:elecraft-boun...@mailman.qth.net] On Behalf Of Robert G Strickland Sent: 05 December 2015 03:22 To: Elecraft Subject: [Elecraft] [OT] increasing CW copy speed: practice slow -v- practice fast The following comments/questions focus on increasing CW copying speed, not the task of initially learning the code. That said, there may be an overlap between the two tasks. W1AW starts its CW practice speeds fast and then slows down. Presumably, as the speed get slower the mental demand lessens and copying becomes easier; then, ease of copying starts occurring at higher speeds over time/trials. >From my days studying animal learning, I remember significant research to the effect that starting a new task in the easiest form [slow CW speed] lessened/prevented errors and, by the end, resulted in quicker and more accurate learning. I tend to practice 3-letter groups at 35-40 wpm, 5-letter groups at 30-35 wpm, and 7-letter groups at 25-30 wpm [for better or worse]. This is somewhat geared to DX contesting since call signs are not "words." All that said, I'm starting to wonder if the animal study folks may have a point. How about the reverse of the above approach. For example, start with 3-letter groups at an error free speed, slowing increasing speed as long as the error rate stays under some value [5%, say]. Keep working at a given speed until the error rate is reached, then increment. Proceed in this fashion until a goal speed is reached. Then, repeat in the same fashion for longer letter groups. The same approach could be used with numbers, complete call signs, and sweepstakes type exchanges. The general idea is to minimize the error rate so that only correct neural networks are formed in the brain. These can be slowly stretched, perhaps like increasing strength in weight training and increasing range of motion after orthopedic surgery, all the time working at the edge to slowly increase capacity. This might also be applicable to increasing the speed of characters as in the Farnsworth method. I'm interested in what folks think. ...rober -- Robert G Strickland, PhD ABPH - KE2WY rc...@verizon.net.usa Syracuse, New York, USA ______________________________________________________________ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:Elecraft@mailman.qth.net This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html Message delivered to danny.higg...@keme.co.uk ______________________________________________________________ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:Elecraft@mailman.qth.net This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html Message delivered to arch...@mail-archive.com