By listening to the morse code audio files with QSZ 3, I noticed I almost always miss the same particular words independent of the speed of the code. It does not matter whether it is at 25wpm, 30wpm, or 35wpm.
The longer the word, the more difficult it becomes to get the word. And if I miss at the first time, listening to it two more times does not help much in most cases.
Some examples are:
10 absolutely 10 acceptable 10 additional ... 11 advertising 11 alternative 11 application ... 12 championship 12 construction 12 contribution ... 13 automatically 13 communication 13 comprehensive ... 14 administration 14 recommendation 14 transportation
Maybe I should concentrate on words, say length more than 10, to save my time
% gawk '{print length($1)}' English2000.txt > English2000-length.txt
% R
> data=read.table("English2000-length.txt")
> hist(data$V1,xlim=c(0,14),main="Histogram of Word Length",xlab="Word Length",col="green")