How Closed Captioning Works (lifted from a GD-P Thread, but this isn't about the primary)

THUNDER HANDS

(1000+ posts) Send PM | Profile | Ignore

Thu Feb-21-08 11:08 AM
Original message

How Closed Captioning Works (lifted from a GD-P Thread, but this isn't about the primary)

Edited on Thu Feb-21-08 11:59 AM by Magic Rat

There were two threads in GD-P about Al Queda being substituted for Hillary Clinton on a closed captioning feed of CNN's broadcast Tuesday night. Most people thought this was a conspiracy on the part of CNN.

http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=132&topic_id=4681189&mesg_id=4681189
http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=132&topic_id=4674045&mesg_id=4674045

I tried in vain, for more than a dozen posts, to prove that it wasn't, but most people didn't get what I was saying. Maybe I can make it more clear by starting this and trying to educate some people as to how the closed captioning process works.

This is a stenograph machine:

This is what the keyboard looks like with letters on it:

As you can see, there are only 17 different letters on it. That's because you have to hit those letters in combination with other letters to make completely different letters.

For example, the TPH letters on the left-side, when hit together, form the letter N.

The TP letters on the left side, when hit together, form the letter F.
The PH letters on th eleft side, when hit together, form the letter M.

IF you're wondering about the vowels, and where the I is, that's the E and U buttons hit together.

It's very complicated and takes about a year to learn that in school.

Now, as a basic court reporter, you have to learn this letter pattern, which teaches you how to form the different letters, as well as how to make words and sentences in one stroke (ie, hitting specific keys at the same time) as well as how to do it really really quickly.

Most people type on their home computer at about 30 to 75 wpm.

To graduate from court reporting school, you have to be able to type 225 wpm.

To do real-time, live tv captioning, you have to type up to 350 wpm.

THREE HUNDRED AND FIFTY WORDS A MINUTE. That's almost six words a second.

And you have to do it without thinking. Your hands have to be an extention of your ears. You heard the words and type them without looking at the keyboard and without processing any of the information you're hearing. They could be talking about war, puppies, sex, garbage, insurance, accidents, Anna Nicole Smith, Sesamie Street, ect. It is just the transition from spoken word to typed-form that matters.

Now, that brings me to why I started this post. The CNN Al Queda issue.

People have claimed, incorrectly, that CNN is responisble for this. CNN does not hire their captioner. They are not sitting next to the table of reporters taking down what they say like in a court room. They are someone like you or me, sitting at home, watching it on tv and transcribing it. When the tv program takes a commercial break, the captioner gets up and goes to the bathroom, or goes to check on their kids, or goes to get something to eat - but they have to be back in time to pick up when the show starts again.

During election night coverage, which frequently goes on for hours without a break, that captioner will be sitting at their chair, immobile, for hours on end typing away while their back, their arms, their hands and their mind, gets really really tired.

It's very likely that after an hour or two of solid non-stop transcribing, that the captioner jsut slipped up and substituted Al Queda for Hillary Clinton.

It would be very easy to do, and I'll show you how that could happen.

I'll refer back to the keyboard for a moment....

Now, do you see the BG on the right hand side? Press those together and that's a K. Press the H on the left and the BG on the right, you can make H-K, a shortform for Hillary Clinton.

Now, do you see the HR on the right hand side, below the 4? Press those together and it's an L. Press that on the left and the BG on the right, and you can make L-K, a shortform for Al Queda.

Now, the captioner would have to pre-program these shortforms into their machine beforehand, or else it would take a fraction fo a second longer for them to type out the words Al Queda (because it would look like this A-L / KAEU / DA*. Those / marks indicate different strokes.

So you can do L-K in one stroke, or A-L / KAEU / DA* in three strokes. When you have to type 350 wpm, you do everything in as few strokes as possible.

Hillary Clinton, if it had to be written out, would be H*EUL / RAOE / KHREUPB / TO*PB. Four strokes, as opposed to H-K, which would be one.

Again, it's just to save time. Because in real-time captioning, you can't afford to miss what people are saying. It's better to get a word wrong, than to get big chunks of text missing.

So, I hope that explained things a little better, and people don't try to go ballistic and see consiracies everywhere when there's logical explanations for things. I know we've been conditioned for the last 7 years to look under every rock for a hidden agenda, but in this instance, it was just a case of a tired captioner meeting a slipped finger and two words that, even though aren't spelled the same, can be quite similiar to the person typing them out.

Thanks for reading, I hope I haven't bored you too much.