WEER NOT BAD SPELERZBY PATTY WHITE, CSR, RPR, RMR, CRR Whether you are a new caption viewer or someone who has been using captions for many years, as you watch captioning, the thought may cross your mind that the captioner either (a) is from a foreign country and has little facility in spelling or (b) is a lazy typist who doesn't want to check their spelling some of the time. Nothing could be further from the truth. Almost all captioners who undertake a captioning career are some of the best, most motivated, ambitious and highly skilled from the court reporting profession. Caption Colorado has a standard of a 98.5% accuracy rate, which all captioners meet and most exceed. What that means, in practical terms, if a speaker is talking at about 250 words per minute and there is one error every 100 words, you will see an average of 2.5 errors or so every minute, and sometimes that seems like a lot. Some of the captionists with our company are the best in the world and write above 99.5% accuracy, and that will produce about one mistake every minute. Over the course of a 30-minute newscast, with the best captioner available, there will be approximately 10-20 errors in the captioning you see. It is an imperfect technology. Captioning is a translation; much as Spanish is translated into English, a captioner translates steno into English. Steno machines are what captioners use to write on and capture the spoken word, called stenography, writing by sounds, and often writing several words in one stroke, which is what gives us the speed to keep up with most speakers. Most captioners can write at speeds in excess of 260 words per minute. Attached to our steno machine is a cable that runs to a computer. Loaded into the computer is software that captures that steno stream, reads it, then throws it up against a steno -> English dictionary to try to find a match for that steno. As it finds a match, it translates it into English, which is what you see on your screen. Sounds like a little hocus-pocus and magic, and in many ways, it is! But like all things, magic doesn't always work, or it works with unexpected and, often, unwanted results. A captioner builds their dictionary over many years, receives extensive training and continuing education on differentiating words that sound alike as they write them, like hear/here, sea/see, their/there/they're, and thousands of others like them. None of those words can be written in steno the same way, and each of them must have an entry in our steno -> English dictionary for a match to be made and the proper word to appear on your screen. Captioners also train to hook word parts together. So if we have to write the word "conundrum," we have all the word parts defined, con, un, and drum, so they will hook together to form the complete word. The down side of that is sometimes word parts hooking together will form a misspelled word that sometimes closely, and other times not so closely, approximates the word that was said. As you might imagine, with the hundreds of thousands of words and names in the English language, it's almost impossible for a captioner to have every word entered in that dictionary, much less every possible derivative of each word. When you see something roll across the screen that looks like complete gibberish, that is what is known to us as an untranslate. Our dictionary just couldn't find a match and translated it just as we write it on our machine. So something that appeared as "DWIS" should be the word "device." If the computer couldn't find the correct match and spelling in the dictionary, it translates the keys that we depress.....D, W, I, S. However, on a steno machine, in reality, the keys that we depress in one stroke to write that word are TKWEUS. That's not very helpful to someone reading captions when you see DWIS go across the screen, but, again, it's a limitation of the technology. Something else you will see from time to time are words that make no sense in the context of the sentence, and you may be wondering whether your captioner's IQ gets out of single digits or that they just don't understand the English language and are making poor choices for a word substitute. As an example, if someone said the word "trellis," and I wrote it and it wasn't in my dictionary, it would translate "tremendously is." That's because I would write it in two strokes, TREL/IS. TREL is a brief form for the word "tremendously." Once I put those two strokes together in my dictionary, it will forever after translate as "trellis," provided, of course, it is written the same way every time and another key on the steno machine isn't hit accidentally or a key is left out or the speaking speed is too fast and one hand doesn't come up off the keyboard from the first stroke fast enough and both of those strokes wind up in one stroke, or... well, there's just a multitude of ways to mess this up. Word boundary problems are something else you'll encounter while watching captioning. To use an example of "target," and "plantar," if someone were to use the phrase, "They plan targets all over the nation." in my dictionary resides an entry for the word "plantar," and the two strokes for that word, not surprisingly, are PLAN/TAR. So the computer, doing it's normal gyrations, finds a match for "plantar" in the above-referenced sentence, so what would come across your screen would be "They plantar gets all over the nation." Looks silly, doesn't it? Let's be honest, it looks ridiculous. The good news is..... once I put those three strokes in my dictionary PLAN/TAR/GET as "plan target," you won't have to see that particular bit of ridiculousness from me again. The bad news is... if someone starts talking about "The definition of plantar gets into the medical..." it will translate as "The definition of plan targets into the medical..." Another part of that is all the words that can either be one word, two words or hyphenated. For instance, the word "kickback" can either be one word or two words, so we cannot define it in our dictionary as one word and must write it a different way. There are many other examples of that, like sometime/some time, touchdown/touch down, pickup/pick-up/pick up, etc. The faster the news is going, the more difficult it is to get context and make those differentiations. One other common thing you may see happens with proper names. If Mr. Russo were referred to, the captioner may opt to shorten his name, try and stroke it and hope for the best, or finger spell it. Finger spelling is where the reporter strokes each letter in the name, much like is done on a regular typewriter, so the name appears intact. Difficult to do with rapid speakers. If that name were written out in steno, it would most likely translate as "reduce so" instead of "Russo," the reason being that it would be stroked on the steno machine as RUS/SO, and RUS is a common brief form in steno for "reduce." So if that name comes up many times, it may be shortened or a pronoun "He" substituted rather than confuse you with the words "reduce so" appearing over and over again. Once I get the name Russo in my dictionary as a definition of the strokes RUS/SO, then that name will forever after translate. The bad news is, if someone says, "They will try to reduce so much of the budget" will very likely translate as "They will try to Russo much of the budget." Another example of a name is "Gonzalez." If the captioner did not have that name in their dictionary, it may translate as "gon is alles," or "gon is a less," or "gons alles." Why? Because that name can be divided in many places, and it is very, very hard to remember from word to word precisely where you would break the word when writing in phonetics, and if you break it in a different place than what is in your dictionary, it won't translate properly. Other times, what you will see that is unreadable are plain old-fashioned mistakes. Captioners often have to write in excess of 260 words per minute, and that makes it almost impossible to write perfectly. We do try, but often fall short. Think of when you type on the typewriter as fast as possible; the faster you go, the more mistakes you make. Sometimes we just get our fingers on the wrong keys or drag in an extra key, which causes the stroke not to be read by the computer properly. There are times when we just don't understand the person speaking clearly, sometimes mishear a word they say, sometimes our fingers are puffy and sluggish... in other words, we're human. There are some differences in captioning. What I have explained above is real-time, live captioning. Most television programming and movies have the captioning done in post-production, and the text and words are error-free because there is the time to check spellings and edit the text. Some programming and much of the national news is done in such a way where much of the text of the show is written ahead of time. Most of the work of captioning a local newscast is done "on the fly." As it is being said, we are writing it in steno and it is being sent to the television station at that instant. There is no time to back up and correct our mistakes; we have a split second to write it and to write it correctly. In addition to the words that have to be written, captioners must also insert punctuation at high speeds, correctly format numbers, dollars, hyphens, musical notes, trying to rely on the natural pauses in speech for comma insertion, which is often faulty. There are also some technical problems that can occur in captioning transmission. We send captioning to a TV station via modem and phone lines, and at the station it is then placed on line 21 of the video picture. Line 21 is very fragile. If the reception on your TV is not crystal clear, you may get garbled characters in your captions, missing letters or words. We love this technology and the work we do and hope you find it useful. Any comments are more than welcome, along with your questions. As with all technology that is developing and evolving, we're improving our skills, the software is improving, and we hope to one day be able to do real-time captioning almost error-free. Until that day comes, we thank you for your patience, your interest, and your kindness. CAPTION COLORADO |
  |
