Call Search
     

New to Ham Radio?
My Profile

Community
Articles
Forums
News
Reviews
Friends Remembered
Strays
Survey Question

Operating
Contesting
DX Cluster Spots
Propagation

Resources
Calendar
Classifieds
Ham Exams
Ham Links
List Archives
News Articles
Product Reviews
QSL Managers

Site Info
eHam Help (FAQ)
Support the site
The eHam Team
Advertising Info
Vision Statement
About eHam.net

   Home   Help Search  
Pages: Prev 1 [2]   Go Down
  Print  
Author Topic: Iambic vs non Iambic: evaluating the myth (new ed. of the book ZART available)  (Read 9934 times)
STAYVERTICAL
Member

Posts: 875




Ignore
« Reply #15 on: August 23, 2010, 05:03:45 AM »

Hi erm,
your message, even if OT, opens a very interesting front of discussion: is there any "real" CW still alive ? However, in this topic we are discussing about effectiveness of iambic vs non-iambic and posts on this line are the welcome to tell about our own experience in iambic keying.
If you open a new topic I will be glad to discuss it, there are a lot of things to talk about ...
73 Carlo

Hi Carlo, I must confess, I am not very good at iambic operation, although I am very impressed by those who are.
I keep trying, but the brain pathways for single paddle keying seem to be laid down with concrete not asphalt in my case.
As to whether there is any real CW left - as long as people read it by ear I think it is alive and well.
CW is one of those skills which elicits both wonder and bemusement in those who don't practice it.
Like admiring a fine painting when a photograph would be quicker to produce, CW expresses the humanity behind the key.

John, no offence was taken - forums should be a venue for the genuine expression of ideas, emotions included, and I did stray off topic.

73s.
Logged
W9OY
Member

Posts: 1346


WWW

Ignore
« Reply #16 on: August 28, 2010, 08:04:00 AM »

I think a statistical analysis that does not statistically take into account the error rate does not address the issue.

So what if your sending method has a 12% more efficient wrist action, if you make 12% more mistakes what does that do to efficient flow of information?   Who says 12% more wrist action is not trivial?  What is the increased wrist action due to mistakes and how does that affect efficiency?

As to a strait key being the most pure form of CW  Until I went to a keyer CW was a chore not a joy. 

73 W9OY
Logged
IK0YGJ
Member

Posts: 43


WWW

Ignore
« Reply #17 on: August 28, 2010, 10:33:16 AM »

We have to understand that the evaluation of iambic keying is a twofolded problem.
At one side  is the efficiency of the keying method, that can be evaluated using Information Theory (not statistics): iambic is undoubtely the most efficient method. Purpose of my article is to answer to this face of the problem.

The other side is how humans interact with the various manipulation technique. To address such matter we should evaluate the average error rate (using a representative sample of CW operators) and how it increases with speed.

I have no statistical data on this subject, but my personal experience of QRQ operator tells me that the error rate is quite the same with iambic and non iambic up to 40 WPM. From 40 WPM up, the error rate increases more with the iambic key than with the single lever.

However, the single lever is sensitively more fatiguing so, there is no "one size fits all": it depends on  the QSO style. When I QSO "armchair" style, iambic is MY preferred method. When I want to QRQ I (but not always) put up the HST and have *a lot* of fun with it.

73 Carlo
Logged
W0XI
Member

Posts: 67




Ignore
« Reply #18 on: August 28, 2010, 08:04:12 PM »

Hello Carlo, fellow information theory and statistics Man!

Thanks for posting your chapter. I'll read through it! Info measures and stats approaches are essential for any detailed inquiry. Good for you!  Smiley

I'm sure you've read William Peirpont's book - also on line - "The Art and Skill of Radio Telegraphy." I found the frequency use of English letters interesting. The letters E,T,I,N,A, while the shortest are not the set of the most frequent. He reports E, T, N, O, R, and A as most frequency. His source information was text taken from a standard source of words for encryption analysis used at the time of his writing, about 1960. He also lists the most frequent 100 words, used by a number of CW learning programs, such as JustLearnMorseCode.

You might find the following interesting reading - also on the web - "Speaking In Shorthand - A Syllable-Centric Perspective For Understanding  Pronunciation Variation," by Steven Greenberg, a linguist. Rather than dealing with marks and spaces, letters, and words, he deals with syllables and words. His source material is from conversations rather than text! Hence, his top 100 word list differs from that of Peirpont's, found in most ham-driven CW practice programs. I use his "conversational" list for code practice. He found that the top 100 words account for 66% if those spoken! The top ten words account for 26%! What was even more interesting was that 81% of the words used were mono-syllabic. 95% of the words used were one or two syllables! Regarding the duration of syllables - kind of like the duration of code letter length - the mean for English speaking was 200 ms and the SD was 100 ms. Japanese mean was 166 ms. A graph of current syllable duration versus previous syllable duration clusters at 200, 200 ms. What this says is that it is very likely a mono-syllable word will follow a long word but not vice versa. Since the durations of Morse letter codes are approximately proportional to their frequency of use, the same clustering of code pairs likely occurs. Fascinating.

Why bother with the stats from linguistics for the study of CW? They reflect what we hear in a conversation, spoken or coded. 73s. Phil, W0XI.



 
Logged
WX7G
Member

Posts: 6212




Ignore
« Reply #19 on: September 02, 2010, 07:02:40 PM »

Pietro at Begali keys says that some of the High Speed Telegraphy competitors use single lever keys. No iambic here.

An ARRL article on the US team at the HST says that the single lever paddles are required to hit the highest speeds when sending figures (numbers). No iambic needed for numbers.
« Last Edit: September 04, 2010, 01:35:27 PM by DAVE CUTHBERT » Logged
N3QE
Member

Posts: 2359




Ignore
« Reply #20 on: September 03, 2010, 05:57:30 AM »

You might find the following interesting reading - also on the web - "Speaking In Shorthand - A Syllable-Centric Perspective For Understanding  Pronunciation Variation," by Steven Greenberg, a linguist. Rather than dealing with marks and spaces, letters, and words, he deals with syllables and words. His source material is from conversations rather than text! Hence, his top 100 word list differs from that of Peirpont's, found in most ham-driven CW practice programs. I use his "conversational" list for code practice. He found that the top 100 words account for 66% if those spoken! The top ten words account for 26%! What was even more interesting was that 81% of the words used were mono-syllabic. 95% of the words used were one or two syllables! Regarding the duration of syllables - kind of like the duration of code letter length - the mean for English speaking was 200 ms and the SD was 100 ms. Japanese mean was 166 ms. A graph of current syllable duration versus previous syllable duration clusters at 200, 200 ms. What this says is that it is very likely a mono-syllable word will follow a long word but not vice versa. Since the durations of Morse letter codes are approximately proportional to their frequency of use, the same clustering of code pairs likely occurs. Fascinating.
I think the above is a very important observation about both spoken language and CW intelligibility. That intelligibility is not solely "information efficiency" in either. That a preferred usage alternates between long and short syllables or between dots and dashes.

For example, if you apply the most simplistic "information efficiency" measure like "shortest callsign in CW" to the contest callsign "EE5E" you might come to the conclusion that this is the best callsign from some simple measures like number of finger movements or most information in the least time.

But if you look at the spots that circulate around during contests, it is very clear that the most commonly busted (miscopied) callsign of all is "EE5E". Common busts are E5E, EI5E, I5E, etc.

What's the actual informational purpose of having the shortest callsign if it's busted so often?

So there are layers of efficiencies that live above the very most simple measures like "highest information density" or "fewest finger movements", and I think that the study that W0XI talks about goes beyond those to looking at pairs of elements or pairs of letters or pairs of syllables which are very important.

Tim.
Logged
IK0YGJ
Member

Posts: 43


WWW

Ignore
« Reply #21 on: September 12, 2010, 01:21:44 AM »

Hi Phil and Tim,
excuse me for the late replay, I don't know why I did not receive the notification via e-mail.
I think that an approach to learning CW from a linguistic perspective can be of tremendous efficience.
My opinion is that the learning approach by syllabes and words reflects the fact that we learn by "acknowledging patterns", shortest patterns exactly.
An interesting analysis to perform would be the one of mutual occurrence, where couples of letter groups (2 or 3 elements) are learnt by sound, starting with the most frequent ones.
Learning QRQ CW can be achieved with significant less effort in this way.
73 and tnx for the interesting comments.
Carlo
Logged
W0XI
Member

Posts: 67




Ignore
« Reply #22 on: September 12, 2010, 08:33:59 AM »

Hi Carlo, Tim

Yes, "clumping" - a loose linguistic term for combing a few sounds - is a known learning strategy. I am only a recent reader of linguistics, a novice with no formal training there; however, I am finding it fascinating/fun. For example clumping a phone number, e.g. 7851112222 into 785-111-2222 makes it easier to copy and understand. So rufzXP study of XX and YYY letter combinations helps in learning to copy call signs, e.g. KA 8 XYZ. I've tried that his last week and can report that my speed and accuracy have both increased - and my typing, hi.

Somehow, this result occurs due to the nature of our short term memory (STM). Some authors think STM naturally decays, like voltage in an RC circuit. Others think that is too is long term memory but is replaced quickly by new distractions: more data, noise, or wandering mind. (Including mine !). By clumping, we are able to pack in a bit more data. The range of STM is about 4 to 7 "elements" of information.

The mechanisms (of the mind and body) used for following behind are yet a mystery to me. Clearly stored memory by sender and receiver contributes to faster code. I can see that analogy clearly when comparing it to the strategies used for multi-tone modems and ARQ protocols such as Pactor. With ARQ, the data is repeated until the S/N - sometimes - can correct the error. 73s. Phil, W0XI
Logged
N3QE
Member

Posts: 2359




Ignore
« Reply #23 on: September 13, 2010, 07:14:08 AM »

Yes, "clumping" - a loose linguistic term for combing a few sounds - is a known learning strategy.
I think it's more than just a learning strategy. Alternation of symbols really does enhance legibility and communications, be it vocal or CW.

In colloquial terms the alternation of short and long symbols produces distinctive rhythms that are important in how we think about the word as a whole. And above a certain speed in CW the right unit of sound is definitely the "word", and not the "symbol" or "letter".

In Information Theoretical terms, the partial-response-maximum-likelihood decoding process seems to work substantially better when symbols don't repeat over and over again, but alternate. It's like the variety of symbols (dot, dash, interlement space, intercharacter space) all work together to frame the word against fades or noise or QRM.

Lose the variety or alternation of symbols, and you're stuck with e.g. EE5E being the most busted call.
Logged
Pages: Prev 1 [2]   Go Up
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.11 | SMF © 2006-2009, Simple Machines LLC Valid XHTML 1.0! Valid CSS!