<NCAM Logo

International Captioning


North America

Here is a table showing the broadcast standard, number of TV households and amount of cable and VCR penetration in North America:



TV households


















Amount of Captioning

In North America (primarily in the United States), captioning has come a long way since its inception in 1972, when only about a half-hour per week of programming was accessible to deaf and hard-of-hearing viewers. Today, across American networks (ABC, CBS, NBC, FOX and PBS) over 200 hours per week of captioned programming is available, plus an additional yearly total of 3,000 commercial spots. Over 150 hours per week of basic cable programming are captioned, and nearly all movies shown on pay-per-view cable services are captioned. Additionally, over 2,000 home movies are captioned each year. The U.S. has over 70 captioning agencies, all but a handful of which are for-profit organizations. All real-time captioning at the national level is handled by The Caption Center, the National Captioning Institute, VITAC, Real-Time Captions, Inc. and Media Captioning Services. National-level off-line captioning is handled by a small group of agencies as well, due to the strenuous demands of network production schedules.

In Canada, there are approximately 25 captioning agencies. Approximately 32% of all Canadian broadcasts are captioned, including the entire prime-time schedule. Because Canadian broadcasters air large amounts of programming from the United States, much of this captioning has already been done by U.S. agencies. Nonetheless, Canadian agencies provide captioning for about 30% of the total. It is interesting to note that three organizations in Canada have developed the capability to provide real-time captions in French (currently, not even France is broadcasting real-time captions in French). These organizations hope to begin broadcasting French real-time captions for French-language broadcasts on the French CBC service in 1995. If a translation from, say, French to English is necessary, this is currently accomplished by having a translator translate the spoken French into spoken English; the stenocaptioner would subsequently write the translator's spoken output as real-time captions. The delay for this method can be as high as five to ten seconds.

Canada has been greatly affected by the passage of the Television Decoder Circuitry Act in the U.S. Because manufacturers ship the same TV sets to both the U.S. and Canada, virtually every new 13" and larger television in Canada contains built-in caption decoding circuitry. The same holds true for other FCC regulations, such as field 2 captioning (see below), which will also affect Canada because of the "spill" of programs and consumer hardware from the United States. Additionally, on March 24, 1995, the Canadian Radio & Television Commission (CRTC) issued landmark rulings which will greatly affect captioning in that country. The rules state that, by September 1, 1998:

  • large stations (over $10 million annually in advertising revenue and network payments) must caption all their local newscasts using real- time captioning or another high-quality method. Ninety percent of all programming is required to be captioned during the broadcast day. These requirements are mandatory and are subject to court action;
  • medium stations ($5-10 million in advertising revenue and network payments) are expected to captioning their local newscasts using real- time captioning or another high-quality method. Ninety percent of all programming is expected to be captioned during the broadcast day;
  • small stations (under $5 million in advertising revenue and network payments) are encouraged to caption their local newscasts using real- time captioning or another high-quality method. The CRTC encourages such broadcasters to caption 90% of their programming.

In Mexico, the story is different. Because of the lower per capita income and lack of infrastructure development in Mexico, the influx of American- produced television and consumer electronics products developed for the U.S. market is small. However, with NAFTA and the increasing economic development of Mexico, the export of American culture and American-style products into Mexico is expected to increase. (Additionally, it should be noted that one Texas-based company is developing Spanish real-time captioning software. While it intends to provide Spanish real-time captions for the United States, this company eventually hopes to offer the service to Mexico.)

Field 2 Data Transmission

To date, the digital data for closed captioning have been carried on line 21, field 1 of the vertical blanking interval. Recently, the FCC set aside field 2 of line 21 for carrying additional captioning information, as well as Extended Data Services (XDS; see below). Field 2 decoding capability is optional for television receiver manufacturers. However, it is expected that most will support this option.

Line 21, field 1, carries four data channels, known as CC1 and CC2 (caption channels) and T1 and T2 (text channels). CC1 is the channel for synchronous captions- the type of captioning available since 1980. It is possible to simultaneously transmit caption data on CC1 and CC2, but doing so effectively cuts the data rate from approximately 60 characters per second to 30. This slow transmission rate results in captions which often lag far behind the audio. Now, with field 2 as a parallel thoroughfare, four more channels are available- CC3 and CC4 for captions and T3 and T4 for text. Data are transmitted simultaneously yet separately, thus permitting full use of each field's data capacity.

The primary use of line 21, field 2 will be for the transmission of a second stream of captions, which will allow the viewer to choose between regular captions and edited captions, or to choose between English captions and those in another language. As of this writing, one manufacturer is already selling field 2-capable set-top decoders in the United States, and many new TV receivers are now being equipped with field 2 decoders. It is expected that most manufacturers will follow suit in the near future. (See Appendix C for a partial list of television receivers and set-top decoders capable of decoding line 21, field 2.)

Extended Data Services

One extra channel, Extended Data Services (XDS), is carried in the excess capacity of line 21, field 2. XDS opens a wide array of new uses for this additional VBI real estate. Its immediate use will most likely be to send data which interacts with consumer hardware or other special devices built into TVs and VCRs. For example, an XDS data packet may set a VCR's clock and calendar, create an index of contents on a videocassette or tell a VCR when it's time to tape a program. XDS can also be used to transmit network or station call letters and channel numbers, program titles, a brief program description or timing information about a program in progress. Many innovative uses of XDS are currently being examined and may be in use by late 1995.

Multi-lingual Captioning

The idea of transmitting captions in more than one language or reading level is not new. It was explored in the United States in the mid-1970s when the The Caption Center published a report on "multi-level" captioning. This report examined ways to edit captions for different reading levels, helping to pave the way for the use of captions as an educational tool. Until recently, however, it was difficult and not always practical to send more than one stream of useable text within the confines of the line-21 system. This difficulty was not with the technology itself, but with the broadcast system (see above). Until recently, only line-21, field 1 was assigned to carry caption data.

With the opening of line 21, field 2 for data transmission, however, the production of multi-lingual captions is now practical. It is anticipated that in the United States, broadcasts featuring caption data on both field 1 and field 2 will appear soon. Additionally, the introduction of new hardware now makes it possible in non-broadcast situations to use lines 10-21 of the VBI for additional caption encoding. Using both fields and all four channels available on each line, this would theoretically make available a total of 44 channels of captioning (in addition to the four already available on line 21). However, in broadcast situations many of these lines are already being used for other services (ghost cancelling, Nielsen AMOL (Automated Measurement of Line-Up, a ratings measurement technology), StarSight, etc.); thus, it is not likely that broadcasting caption data on these lines will become a reality in the U.S.

In general, the captioning industry around the world is optimistic about multi- field, multi-language captioning. As encoders and decoders with multi-field capability become available, programs with more than one data stream can undoubtedly become the rule, rather than the exception. It will be up to the program providers and electronics manufacturers, however, to determine how these separate yet simultaneous data streams can best be utilized. In the U.S., some ideas already being discussed include using field 1 for "regular" captions, or for verbatim captions, and field 2 for so-called "easy-reader" captions- captions which have been edited to a slower reading speed and/or easier reading level, or for captions in a second language, such as Spanish. In Canada, field 2 is being eyed for the French language. Dual-language broadcasts will undoubtedly appear in North America once the installed base of televisions with field-2 capability grows.

Data Exchange

The sharing of caption files between the United States and Canada has been going on for several years. (In the technical sense, this is nothing special since both countries use the NTSC format and the same encoding/decoding technologies.) The Caption Center, for example, regularly shares some of its caption files with broadcasters in Canada. First, they undergo a slight modification to indicate to the Canadian broadcaster where the first program video appears for each act of the show, where the program breaks are and where the end credits begin. Once this information has been incorporated in the file, the data are transferred via modem to Canada for encoding prior to broadcast. These changes are necessary because the programs are slightly edited for Canadian re-broadcast.

Data exchange also occurs between North America and other parts of the world. The Caption Center has cooperated with the Independent TV Facilities Centre, IntelFax and the Australian Caption Centre (the latter as recently as July, 1994) in an effort to make the sharing of caption files more feasible. The caption text itself is simply converted to an ASCII format and transferred to the foreign agency. However, because the United States uses the NTSC format (which operates at 30 frames per second) and Australia uses PAL (operating at 25 frames per second), information regarding the way the original captions were timed has to be reformatted or discarded. Stylistic changes are sometimes made at the receiving end (American vs. European spelling conventions, for example), and the larger issues of editing are resolved by the caption agency in accordance with their audience's preferences. However, a significant amount of time is saved by not having to transcribe and format the text into captions.

Additionally, in 1994 The Caption Center began using its own software to convert some of the BBC's EBU-standard subtitle files into caption files useable by The Caption Center's captioning software, CC Writer©. The software automatically converts the timing information in the original files from 25 frames per second to 30 frames per second, and also retains placement information. However, because the BBC captioning facility and The Caption Center use different stylistic and editing conventions, caption writers at The Caption Center must still spend some time working with the text. Time is still saved over the conventional captioning process.

Finally, there is some discussion between captioning agencies about retaining original spelling and editing conventions when caption files are shared internationally. For example, The Australian Caption Centre (ACC) does not change stylistic conventions or spelling of captions in files received for American programming. Since the program is an American creation, the ACC believes that it is perfectly acceptable for the captions to conform to American style.

International Caption Exchange Protocol

In the United States, the issue of overcoming data-format incompatibility was initially investigated in the early 1980s, when the National Captioning Institute hired Andrew Lamborn, of Synapse Systems in the United Kingdom, to develop a point-to-point method of exchanging data. The proposed solution, known as the International Caption Exchange Protocol (ICEP) would allow caption files created at NCI's facilities to be converted from NTSC to PAL format and then sent to the BBC for re-encoding. In addition to the relatively simple problem of straight data conversion, NCI also wanted to work out problems with editorial and style differences- the most simple, for example, being the spelling of "color" vs "colour", and the more complex being the use of color captions vs italics to indicate emphasis.

In addition to NCI, the project involved a number of other captioning agencies in North America, including The Caption Center and the Canadian Captioning Development Agency (Toronto, Ontario). Unfortunately, NCI made the decision to shelve the ICEP project in the mid 1980s when competitive interests took precedence. Thus, the solutions proposed in ICEP were never put into practice.

The work on ICEP, however, was not totally wasted. In the late 1980s, Andrew Lamborn was again hired by NCI to write software which would transcode real- time caption files created by NCI to the new European standard, known as the European Broadcast Union standard (EBU standard; see below). While the transcoding scheme was a success, it is not now widely used because the cost is somewhat prohibitive, and the demand for instant translation of American captions for use in Britain is somewhat limited at this time, since American news and sports (the programs most often done in real-time) are not widely broadcast in the U.K. and Australia. (Lamborn has since written software to transcode EBU- standard files to NTSC format.) It is also not widely used because it is a proprietary scheme and unavailable to the captioners of CBS News, NBC News, CNN and the MacNeil/Lehrer Newshour.

Teletext in North America

Teletext was used in the United States in the mid-1980s with relatively little success. Known as the North American Broadcast Teletext Specification, or NABTS, it was a different standard from World System Teletext (used in most other parts of the world). Its design was based on the French Antiope and Canadian Telidon systems.

Its biggest proponent was CBS, which marketed it under the name ExtravisionTM. Extravision, like its counterparts in Europe, was designed not only to deliver national and local current-events information, advertising, games and other services, but also to provide teletext closed captioning as an alternative to- and, it was hoped, a replacement for- line-21 closed captioning. For a period in the mid- to late-1980s, captioning agencies (such as The Caption Center) worked with the network to provide caption data in both the line 21 and NABTS teletext formats. Thus, at least in theory, viewers who owned line-21 and NABTS teletext decoders could choose between caption streams. (There was no such thing as a decoder which could accommodate both technologies, however. Separate decoders had to be purchased.)

Unfortunately, this service never really gained public acceptance in North America. There were several reasons for its relative failure; high on the list was the cost to viewers: consumer-level decoders cost approximately $2,000 and required a special television with component inputs, something the average viewer did not have. Additionally, broadcasters were reluctant to commit large amounts of space in their vertical blanking intervals to a service which was limited in its one-way capabilities, especially at a time when two-way communication was being touted as the wave of the future.

Despite these drawbacks, NABTS teletext in North America never completely vanished. As a matter of fact, some insiders in the television industry predict a revival of the service. One venture, currently underway in the United States, uses the VBI to feed data (such as stock-market quotes) to a computer rather than to a television set. In addition to paying a subscription fee, users must purchase an expansion card/TV tuner for a PC to receive the teletext signal and download the data to the computer. The technology used for this service was derived from the BBC's Ceefax system.

Another teletext-like service available in North America is StarSight Telecast. Launched in early 1994, the StarSight service consists of a program guide and automatic VCR programming information. The data are transmitted in the PBS vertical blanking interval on a single line at a rate similar to that for line-21 closed captioning. Local PBS stations bridge the information into their on-air feeds. A StarSight-equipped TV can tune into the VBI, retrieve the program schedule and other information and, at the viewer's command, display the information on the television screen. The StarSight decoder may be a set-top unit, or may be built into the television receiver or cable converter box. Users must pay a monthly fee for the service.

In addition, Nielsen embeds AMOL data into the VBI. In Canada, the Toronto Globe and Mail and other Canada-wide papers are embedded in the VBI of the Sports Channel and are used to deliver the paper to the computers of blind persons.


NCAM | ncam@wgbh.org | WGBH Educational Foundation | main.wgbh.org