Joe Clark: Media access

You are here: joeclark.orgCaptioning and media access
Localization and accessibility

Updated 2002.02.03

Localization and accessibility

SUMMARY – People will put up with captioning, subtitling, dubbing, and audio description that are not rendered in their particular dialects, but you risk alienating, annoying, or distracting your audience unless you localize your access provisions.

What’s the problem?

A lot of people hate to admit it, but the oldschool language techniques of subtitling and dubbing are just as much a form of accessibility as captioning and audio description. (The people who hate to admit it are chiefly subtitlers, who view captioning as mere transcription, an artless postproduction process that is beneath their station in life.) All those techniques are part of the same family because they accommodate features the audience cannot change, or cannot change easily. (You can’t stop being deaf or blind, and you cannot suddenly start to understand a foreign language.)

It’s obvious that language varieties are relevant to subtitling and dubbing because those forms explicitly deal with translation. The same fact is less obvious for captioning and description given that those forms are rendered in the same language as the original (with rare exceptions). However, just as subtitling and dubbing really are related to captioning and description, localization is relevant to all four.

About this document

This document is ruminative and anecdotal and attempts to begin a discussion about localization issues in accessibility based on actual experience and a smattering of research and theory. Many readers will find it too meandering, but them’s the breaks. There is an emphasis on English-language variants.

Localization

First, we need a couple of definitions, in this case adapted from Mozilla:

Localization
Modifying or adapting a production to fit the requirements of a particular locale. The cutesy abbreviation used is L10n (referring to the number of letters between L and n)
Internationalization
Designing and developing a production to function in multiple locales. The cutesy abbreviation used is i18n

Americans rarely encounter anything that’s been localized, though in fact a great many products actually have been: Products manufactured in Europe were modified to work on 120-volt power using American-style two- and three-prong power plugs, for example. TV sets use the NTSC standard, not PAL, and come equipped with caption decoders and V-chips. (It is said that there are no U.S. production facilities for TV sets at all. In effect, every American television is a localized product.)

Actors localize themselves all the time. Russell Crowe, Guy Pearce, and Nicole Kidman appear to speak fluent American despite actually being Australian. Robert Downey Jr. made a valiant effort to sound Australian in Natural Born Killers, but American actors are notoriously bad at putting on any kind of accent. (Something to do with the training, presumably.) Yet the cliché joke about Meryl Streep amounts to a joke about localization: All she is is an accent.

Literature localizes itself: Not only can you read a French or a Japanese translation of a book, you can read a localized English-language variant of a book. British, American, Canadian, and Australian editions of, say, a Margaret Atwood novel will all look slightly different. Some examples:

British
U.K. spellings (neighbour, organise, tyre, pluralized adjectives in locutions like plastics industry), single quotation marks
American
U.S. spellings (neighbor, organize, tire, singular adjectives in locutions like plastic industry), double quotation marks
Canadian
Canadian spellings (neighbour, organize, tire, singular adjectives in locutions like plastic industry) or U.S. ones, double quotation marks
Australian
U.K. spellings, double quotation marks; in a pinch, can use a Canadian edition

And actually, if you put these last examples together – focusing on the issues of (spoken) accents and (written) orthography – you pretty much describe the problem of localizing accessibility in the written medium. We’ll come back to this shortly.

Ideal vs. tolerated

Now let’s draw a distinction between what’s ideal for an audience and what is merely tolerated.

If your DVD of a German- or Spanish- language movie (Run Lola Run, Before Night Falls) contains a dubbing track using American English accents, a British audience will be able to understand the dialogue and will tolerate the American accent if there is no other option. (Whether an American audience would put up with British accents is rather less certain. A Canadian or Australian audience probably would.)

Here “tolerate” must be understood to mean “put up with” or “be annoyed by.” A significant number of viewers will be aware of the dubbing accents, an awareness that may be in the forefront of audience consciousness through every minute of the film. To deploy a term much overused in the accessibility business (chiefly by opponents), these viewers will be distracted by the accents. Toleration equals distraction.

However, it would be considered ideal to use British dubbing voices for the U.K. market (actually, the Scots and Irish might possibly object) and American voices for the U.S. market. In this ideal case, the degree of distraction is dramatically reduced even for the very sensitive, and a much larger number of viewers will not even notice the dubbing after the first moments. This ideal case involves much less putting up with, annoyance, and distraction. (It can never be eliminated, at least not in every production a viewer sees over a long period of time; nearly every nondisabled viewer of a dubbed film has some awareness of the dubbing process at the outset.)

We can divide the topic into two parts.

Subtitling/dubbing

The entire purpose of dubbing and subtitling is to translate. The fact that a production’s source and target languages differ is self-evident, and there aren’t all that many localization issues involved beyond the fact that the translation is itself a form of localization. The complications reside in the distinctions within the same target language.

The dubbing industry’s trade dispute between Quebec and France is such a perennial story in Canada it’s become the only thing a lot of people know about dubbing at all. (See, for example, “In Quebec, Studios Face a French Knot,” whatever a French knot is.)

Three claims are made on this topic:

  1. Quebec actors need the dubbing work. It’s obviously true.
  2. Europeans cannot understand Quebec accents. That’s an old chestnut rendered absurd and outdated by the vast numbers of African and Arab immigrants regularly heard on French television. (I see them all the time on TV5.) Their accents differ much more strongly from national French than Quebec accents do, yet such voices remain understandable. This argument is so stereotypical of French snobbery that you’d think the French would no longer cling to it (until it occurs to you that the French might actively cultivate the image of being snobs). Moreover, Quebec French comprises a range of accents, and, while mainstream television newsreader accents are still noticeably different (you would never mistake a Quebec newscast for a French one), both are perfectly understandable.
  3. Quebec dialect differences do a better job of adapting Hollywood films for Quebec viewers. It’s indisputably true, but this claim quickly becomes a case study of ideal vs. tolerated. With the exception of colloquial words with distinct meanings in France and Quebec, Quebec French-speakers can still understand national French dialects.

Now, it is actually possible to find a production subtitled in the source language. Just talking about the whole issue gets confusing, because if the subtitles and the dialogue are the same language, aren’t we talking about captioning? Not in this case, because there is never any effort to position captions to indicate speakers, explicitly identify speakers, or notate “non-speech information” like sound effects. Two examples can be found in any good video store:

  1. Trainspotting, in which Begbie’s unending stream of invective in the original Scottish was subtitled in Queen’s English (admittedly, the practice could be explained by the noisy disco sequence in which it took place).
  2. The Acid House, also adapted from a book by Irvine Welsh, where nearly every Scot speaker was subtitled into English.

Especially nowadays, television newscasts are apt to subtitle the English-language comments from “foreign” sources if their accents or if the audio quality make it unlikely that the assumed audience could understand the speech otherwise.

Within-language subtitle-captioning is also a common practice in Oriental films. It’s possible to watch Chinese-language films on TV every Saturday night here in Toronto with two parallel strands of subtitles – one in Chinese, the other in English. (I’ve seen up to four levels, actually – Chinese, English, Thai, and Vietnamese all at once. I’ve also seen movies spoken in Chinese and subtitled solely in Chinese.) In this case, the issue is the mutual incomprehensibility of Cantonese and Mandarin, which nonetheless use nearly the same writing system save for a limited repertoire of characters specific to the various dialects.

On no more than four occasions, I have witnessed TV closed captions that translate the original language into English, a sin of considerable gravity. (The cases are a documentary on an elephant trainer whose British-accented Hindi commands to the elephant were translated; The Suburbanators, in which Arabic was captioned in English; and an episode of Blue Murder, in which Spanish was translated. I seem to recall there was one other example a very long time ago.)

Subtitlers translate, captioners render, which is why subtitled productions must be captioned to make them accessible to deaf viewers.

Captioning/description

As Begbie, in one of his calmer moods, would put it, the shit hits the fan when it comes to intra-language distinctions in captioning and audio description.

International captioning transfers

Unbeknownst to Americans, captioning viewers in the U.K. and Australia are commonly forced to read American captions. One can fairly say “forced” because, to this day, four of the Top Five U.S. captioners caption in all-capitals by default (NCI, Vitac, Captions, Inc., and CaptionMax); only the Caption Center captions in “mixed case,” and even then not all the time. But captions in the U.K. and Oz have always been mixed-case. (Their caption fonts are less illegible than ours used to be. The claim that current caption fonts built into TVs are as hard to read as the old TeleCaption fonts is actually false; the Caption Center is right to have switched to mixed case.)

Conversion and cost

American programs very often are captioned by importing and converting the original American captions. It’s cheaper than recaptioning, though that also happens – one informant tells me he watched Sex and the City in London with what were obviously made-in-the-U.K. captions.

In extremely rare cases (Da Vinci’s Inquest is the only one I know of), Canadian captioned shows appear overseas with imported uppercase Canadian captions. Overseas captioners leave the uppercase captions alone. It is thus immediately apparent to viewers when captions derive from another country, since ALL THE CHARACTERS APPEAR TO BE SHOUTING.

But there is almost no traffic in captions from the U.K. and Oz to North America. On very rare occasions, it has in fact come to pass that a World System Teletext file has been converted for Line 21 captioning, but the preference here is to simply recaption from scratch. It is sometimes unavoidable anyway: British miniseries appearing on, say, Masterpiece Theatre are surrounded by intros and extros that change the program, so the whole thing must be captioned anyway, and why not be consistent all the way through?

Also:

  • Canadian programs (canonically, the DeGrassi cycle, but also “industrially Canadian” programs like CSI and Due South) are also captioned in the U.S.; very rarely indeed (The Jewel in the Crown miniseries, a couple of episodes of Nero Wolfe, Queer as Folk), Canadian-captioned shows air with Canadian captions in the U.S.
  • Also uncommonly, British programs are separately captioned in Canada and the U.S.

(Increasingly, I find that shows captioned well in the U.S. are dumbly miscaptioned from scratch by Canadians – Prime Suspect is an example. Canadians will even recaption shows previously captioned in Canada – the examples again are the DeGrassi cycle and Da Vinci’s Inquest, though this phenomenon is not a localization issue.)

In all these respective cases, regardless of the English dialect used, the local variant used in the captioner’s country is the one you end up reading in captions no matter where you are. U.S.-captioned shows use American English, Canadian-captioned shows Canadian English. Presumably, Australian programs air in the U.K. (which uses the same World System Teletext apparatus) with Australian-made captions in Australian English, which may differ chiefly in the use of double rather than single quotation marks.

Captioning orthography

Moreover, American captions feature American spellings. Perhaps this is not so terrible, really, since the programs themselves are American; hearing people already have to tolerate American accents, so it is arguably symmetrical and consistent for captioning viewers to tolerate U.S. spellings.

In fact, the Australian captioning guidelines, available only as a Microsoft Word document, say:

§3.7  American spelling

American spellings should only be used for specific American words. In an American program, characters may use certain distinctly American variants of words. Where an alternative spelling exists that reflects American pronunciation, this should be used. This rule only applies to specific words; Australian spelling should still be used for the program as a whole.

Examples: airplanemomassasshole

American proper names should not be changed to Australian spelling. The name of an American place or organization may contain a common word spelled according to the American standard. This is the correct spelling for that name, so the word should not be changed to Australian spelling.

Examples: Pearl HarborLincoln CenterDepartment of Defense

The number of American English words with possible divergent American and Australian spellings is rather small, though I note that the Australian advice quoted above ignores inflectional endings like -ize. Thus the sentence “Don’t you be criticizing my mom, asshole, or I’ll kick your ass right out of this airplane” could be rendered in an Australian caption as “Don’t you be criticising my mom, asshole, or I’ll kick your ass right out of this airplane” and still comply with the guideline.

There is a certain flexibility in North America that compares well with the Australian guideline: The old Caption Center training manual specifically authorizes retaining proper names with non-American spellings (e.g., Department of Labour), while Americans and Canadians are perfectly happy to write words like arse, arsehole, and mum (but not aeroplane, a word that has always been ethnically cleansed to airplane every single time I’ve heard it).

Also, it is possible to live-caption a show in the U.S. or Canada and have the captions automatically converted to World System Teletext format overseas, a process that effectively captions a program live in multiple countries using two different systems. It is quite possible to caption in mixed case. In North American in the English language, only Valerie Waite at Waite & Associates ever bothers to use mixed case, and only occasionally; essentially all real-time captions here are set in uppercase save for speaker IDs and miscellany, and overseas viewers will read and tolerate such captions. (French- and Spanish-language real-time captions appear in mixed case.)

I know of one captioner that routinely captions Hollywood movies separately in U.S. and British orthography.

How important is this?

Just how important it is to render text in a certain national orthography is something of a “soft” issue: Since the resulting words are still understandable to any English-speaker, it is difficult to make the case that, say, American spellings are simply wrong in the U.K. They would certainly be wrong if originally written or produced there, but as imports, it’s hard to work oneself to high dudgeon.

But if the issue is not important per se, it is nonetheless not unimportant; it is somewhat important, if only for the following reasons.

Canadians, Australians, and the British may legitimately tire of American media dominating their television and movie screens. It’s bad enough to be stuck with American voices; must they be stuck with American spellings, too? Americans, who generally speak exactly one language (American!), react conversely when confronted with another orthography: Can’t these people learn to spell?

All those groups will nevertheless be able to understand the English words. They may simply resent having to understand words that seem subtly misspelled. These are, to reiterate, soft complaints, since any English variant is fundamentally comprehensible.

Many other languages do not display the variation in spelling that English does. French is an example: Apart from the different words used in different nations, French is written almost uniformly across the globe. (A strong case can be made that slang, informal, or “uneducated” French looks very different when written down. You definitely can spot Quebec French a mile away in that case. But Irvine Welsh’s novels, written in what amounts to broad phonetic transcription of the original Scottish, don’t look anything like standard English either.)

In other languages that use one of two writing systems – Latin and Cyrillic characters in Serbo-Croat; Urdu and Hindi (where vocabulary differences are noticeable); and the many efforts to render aboriginal languages like Navajo and Inuktitut in Latin script – the distinctions are not trivial and it is quite possible to alienate a portion of the audience that could nonetheless actually understand the target language were it rendered in the alphabet or syllabary they are used to.

This has actually come up in captioning, and it happened at the very dawn of closed-captioning. The NBC television miniseries Shogun (1980) contained lengthy passages in spoken Japanese, something that U.S. television would never put up with now. NCI captioned the Japanese in romanized English (romaji), and in fact asked staff from the Japanese embassy to proofread the romaji. (I know this because I interviewed an NCI manager over the phone during the series’ run. When I say I go back 20 years in this business, I really mean it.) The manager claimed to have received phone calls from hearing people wondering if captioning viewers were given an undue advantage – were they seeing a translation? The answer is no.

Character encodings

In closed captioning, decades of incompetence on the part of engineers have saddled us with decoder fonts that do not have enough characters to typeset even “simple” languages like French and Spanish properly. (I am setting aside for the moment the issue of typographically accurate hyphens and dashes, quotation marks, and pi symbols.)

After two upgrades (in other words, on the third try), the Line 21 system still lacks a wide swath of characters used in French, including uppercase accents (they’re not optional no matter who has told you otherwise: CHIRAC TUE and CHIRAC TUÉ have rather different meanings) and letters with diereresis/umlaut/tréma. Those same failings make it impossible to caption Spanish properly, either.

Nor are angled quotation marks or guillemets present (« and »). While neutral English quotation marks are widely used and are tolerated, the absence of proper quotation marks, coupled with the abject stupidity and typographic ignorance of certain captioners, forces viewers to deal with << and >> (two less-than or greater-than characters) as a pretended substitute. (To draw an English parallel: You can’t type two apostrophes and pretend they make a double quotation mark.)

World System Teletext fonts also lack enough characters to caption European languages, though that tends to vary by decoder generation and model. The Australian Caption Centre guidelines state: “Ignore all diacritical marks, because the teletext font doesn’t allow for them (don’t use e to represent an umlaut).”

It’s an entirely avoidable predicament that remains outrageous, forcing captioners working in “foreign” languages to misspell words – all because the unilingual geeks designing the systems didn’t do their homework.

What happens with subtitling?

Does anything like this happen with subtitling?

Yes, occasionally. It is quite possible for the British or Australians (or anyone else anywhere in the world, actually) to watch productions subtitled in the U.S., which pretty much always use U.S. spellings. French-language Canadian films all seem to be subtitled by Robert Gray of Kinograph, who reports that he chooses American spellings except for certain National Film Board productions, which use Canadian orthography.

It has certainly been my experience that subtitlers residing in countries where English is not a national language subtitle in U.K. English.

Accordingly, it is quite possible to watch a production subtitled into your own language even though it does not quite use the same spellings you do.

Description voices

This is the toughest case: Should the voices of audio describers match the production or the audience? The easy answer is “the audience,” but there are many provisos.

Similarities with dubbing

What few researchers there are in the field of subtitling will disagree with my terminology in this section, but this document is written for laypeople, after all. In dubbing and subtitling, you can retain cultural references from the original (deutschemarks remain deutschemarks, kilograms remain kilograms) or adapt them to the culture of the target language (deutschemarks become dollars or pounds, kilograms stay kilograms or are turned into Imperial measure).

Right there, you can already see that translating a production for the U.S. differs from the same task intended for a U.K. audience. (What do Canadians prefer? We like to consider ourselves more international than the Americans, so maybe we’d prefer deutschemarks to stay the same; we use the metric system, but absolutely everyone understands Imperial measures. How about the Australians, who also have “dollars” but use metric measures?)

If you’re dubbing a production, do you want it to sound seamless and natural to an American or a British audience? (Or neither?) Are you willing to produce two dubbed versions, in U.S. and in U.K. English? Canadians will be able to tell it’s an American voice, while Scottish and Irish listeners will know that the English dialect is not the same as their own. Neither version sounds like Australian English, to say nothing of, say, South African English.

The issue, then, is the disparity between the audience’s vernacular and the dubbing vernacular. But as always in accessibility, this issue revolves around ideal vs. tolerated: If the only dubbed version available is in U.K. English, you can live with that.

It must be pointed out that it is very rare indeed to find a movie released in North America with more than one English-language dialogue track (excluding audio description). The only one I know of is the recent DVD of Mad Max, which specifically markets itself as having English and Australian English (!) dialogue (“Original Australian Language... English Dubbed”).

Description disparity

If a disparity between dubbed voices and the audience is undesirable in dubbing, an analogous disparity is required in audio description. The listener must be able to distinguish the description narrator’s voice from every other voice in the production, which can itself include narrators, actors, interview subjects, and celebrities.

With so much to define yourself against, doesn’t it make sense for the description voice to be as different as possible from every other voice? When describing an American TV series, why not use a British voice?

Audio description is unlike the other accessibility techniques in that it must be distinguishable and unobtrusive at the same time. (Captions and subtitles are distinguishable and obtrusive by definition. Dubbing is obtrusive whenever viewers notice the mouth movements do not match, itself a fact that everyone notices at least momentarily, or when obvious cultural, national, or ethnic disparities are at work, like a screenful of black American actors speaking fluent Chinese.) The goal is to simply hear and understand the description as part of the audio flow of the program. The awareness of who’s the description voice compared to every other voice must be subconscious most of the time.

(There are notable exceptions, as when description must call attention to itself – by stating “Titles appear” or “End credits” before such sections, or when the description provider identifies itself. Or, when descriptions must be given over dialogue, the fact that main dialogue is now missing brings the description track to the forefront.)

In choosing description voices, then, the describer must be just different enough to be noticeable, but not so different as to be too noticeable. It is not a simple case of choosing a female narrator for a production whose voices are nearly all male, or vice-versa; in fact, that can be insensitive and ridiculous. (T2 and Black Hawk Down could only have been described by guys. The converse is not entirely true: Chick flicks like Waiting to Exhale and When Harry Met Sally do not necessarily require a female narrator. A documentary on female workers in maquiladoras or on breast cancer or pregnancy would almost demand a female describer.)

Audience voice vs. production voice

In description, we match the describer’s voice to the production so that the audience won’t be constantly aware of the disparity between describer voice and other voices. We are putting the needs of the audience first.

Accordingly, the case can be made that productions must be described in the vernacular of the audience. In fact, this case has already been made, and is a standard technique of audio description.

Examples:

  • Harry Potter was described by DVS for U.S. and Canadian theatrical release using an American narrator. It was also described by ITFC for U.K. release using a British narrator. The film itself is British and every voice in the movie speaks in some kind of British accent, but it’s the audience that counts.
  • Moulin Rouge is an Australian film (made with American money) with a cast whose accents are a kind of overstated characterization device and range from British to American to a rather coarse Spanish. IADA described the film for DVD using a Scottish narrator. The film’s male lead, Ewan McGregor, is a Scot; his mother, Carol McGregor, runs IADA, and that presumably is why a Scottish narrator was used, but the fact remains that only a U.K. audience would consider the describer’s accent unobtrusive, and even that might not be true given that Scottish accents are “marked” or nonstandard in the U.K.
  • Bob & Margaret is a Canadian television series set at first in London and later in Toronto. Bob and Margaret themselves are British. The program is described for Canadian TV using Canadian accents.
  • A range of British productions have aired on Masterpiece Theatre (including Henry V with extended passages in French, or the closest that British actors can approximate it), all described by DVS with American accents.
  • Movies with extensive subtitling, like The Joy Luck Club, require voicing the subtitles. (Opponents of audio description constantly freak out at the idea of describing subtitled films. It’s happened repeatedly already, and it’s no big deal.) In that example, a male and a female narrator were used to embody the multiple characters; that male narrator was also the main description voice throughout the film.
  • I had previously written, in covering U.K. “guidelines” on audio description:

    On the topic of “regional accents,” here’s what the Guidelines say: “Occasionally a slight regional accent may fit the bill, but each programme has to be assessed separately. [While] a regional accent for a regional programme might seem logical, in practice it is quite difficult to find the accent which suits everybody. For example, there are many Lancashire variations, whereas standard English is at least understood by most people.” The issue is complex in the U.K., with its hundreds of regional dialects and the dominance of Received Pronunciation – the type of English the Royal Family or prestige newsreaders speak. By “standard English,” the Guidelines clearly mean “newsreader accent,” or Received Pronunciation.

    Canadian and American descriptions of British programming will always exhibit a contrast between the actors’ many accents and the describer’s. (Same with Australia, one would think.) But we live with this: As the Guidelines tell us, local vernacular must always win out.

    Bilingual viewers in Scandinavia were asked if they would prefer an audio description in English or in their own local language when it accompanied an English-language import. The English audio description would have the advantage of matching the programme language and might be available to purchase with the programme but it would not reflect the culture of the target audience. A description in the local language could help to clarify any difficult English programme dialogue, and would avoid any miscomprehension of an English description which might arise. Opinion was unanimously in favour of a local language description.

  • There are already several cases of DVDs described once in the U.S. and once again in the U.K. There is no technical reason for the duplication. Rather, it centres more around the local-audience issue (Her Majesty Mrs. Brown was described for American TV as part of the Masterpiece Theatre series, which is always described in American voices) and the fact that the U.K. description provider wanted a kind of brand exclusivity. Even now, with so little description available, describers who also sell home videos are completely unwilling to sell each other’s stock, preferring to reinvent the wheel.

Separate described versions

Does this mean an English-language production must be described anew in each country with a definable English variant? In many cases, yes.

  • The accents used by Canadian and American description narrators are similar enough to be interchangeable. (Any competent linguist can list a range of differences, but whatever they may be, they are unobtrusive.) The U.S. and Canada can carry on a kind of free trade in audio description.
  • If the only available source of description uses an accent that is not your own, don’t let that stop you from buying, exhibiting, broadcasting, or watching the production.
  • If you have the time, budget, and rights clearances to make it possible, redescribe in the local vernacular.

There is a clear exception to be made here, and we draw our attention back to the issue of recaptioning. U.S.-captioned programs (some of them British) have been and are recaptioned in Canada. In every single case, the original U.S. captions were of good or excellent quality while the Canadian captions were appalling. It is quite easy to find home videos captioned well or excellently in the U.S. and recaptioned poorly in Canada. On rare occasion, Canadian broadcasters defy governent regulations and simulcast programs well-captioned in the U.S. with low-quality Canadian captions.

So, if your choice is one of a recognized, high-quality describer and a pack of amateurs in your home country, pick the high-quality describers no matter what the difference in accent and vernacular may be. A good description in the “wrong” accent beats the pants off a lousy description in the “proper” one.

Conclusions

How important is it that captioning, subtitling, dubbing, and audio description be provided in the accent, orthography, or vernacular of the specific national or ethnic audience?

Captioning
Somewhat important
Subtitling
Not very important
Dubbing
Important
Audio description
Very important

But in no case is it absolutely essential.