Subj : Yet another morse proggie
To   : Deuce
From : Angus McLeod
Date : Tue Aug 23 2005 08:49 pm

  Re: Yet another morse proggie
  By: Deuce to Angus McLeod on Tue Aug 23 2005 15:42:00

 > So bascially, you're suggesting something along the lines of:
 > i) 2 minutes of drilling farnsworth style using similar group

Or perhaps until the student sustains a particular success rate for a 
certain span of characters.

 > ii) 5 minutes of drilling using words, pause after each group rather than ea
 > character, real words in somewhat sensical order

You can send each word but you send the character at (say) 20 WPM, and set 
the inter-character gap at (say) 5 WPM, so they have lots of time to 
decipher each character.  And again, the period might be determined by the 
success.  Say you keep it up until you get 10 words correct in a row.

 > iii) 5 minutes of drilling as with ii, but using all groups drilled to date.

Uh, yes... OK.  Again, maybe the period determined by the rate of success.

 > Then, the user decides when to move to another group.  During the first grou
 > of course, iii is skipped.

Exactly.  

 > So bascially, a 12 minutes drill session (With, of course, breaks in between

The .rc file would record what group the student was currently exercising 
on, and the student could repeat the exercise as often as they liked, 
until they decided it was time to advance to the next group.  Perhaps the 
program could refuse to allow the step forward -- or conversely, recommend 
it -- depending on how well the student has done.

I've never given machine-taught CW too much thought before, but it strikes 
me that the machine should be able to maintain a table indicating which 
characters are most difficult for the student.  A SCORE, so to speak.  
This should allow the program to actually *tune* the selection of "random" 
characters.  You could insert a slightly higher percentage of those 
characters that the student is having difficulty with, to help them learn 
those characters.  Or you could increase the frequency of those characters 
the student knows well, so as to ease up on the student for a while and 
bolster their self-confidence.  Perhaps at different times, different 
tunings would be appropriate?

With this in mind, the program might (in step iii) gradually start 
introducing characters from earlier groups.  Perhaps only throwing in an 
occasional one, and a well-known one at that.  But slowly increasing the 
number of "older" characters, until the student was being exercised with 
the full complement of characters learned so far.  In essence, steps ii 
and iii might be a single, mutating phase, wherein at first only those
characters from the latest group are exercised, but as the student hits 
certain success targets, more and more older characters are added to the 
exercise pool, until it contains all characters learned so far.

Perhaps the number of exposures to a particular character should be taken 
into account when computing it's score.  Obviously, the first time you are 
tested with a character and you get it RIGHT, you don't want to give it a 
score of 100%!  So maybe the score should be based on the last X exposures 
to that character, keeping characters you have not yet been exposed to 
very often with a low score?

I am just brainstorming with myself, now.


---
 þ Synchronet þ CQ DX!  The ANJO BBS calling on 56K dial-up...

.