3 screen translation - tolk- och översättarinstitutet tÖi/screen_translation.pdf · 3 screen...

23
1 3 Screen Translation In order to show how Scandinavia fits into the European media translation scene, and to place the texts under investigation in a wider context, , this chapter is dedicated to a short survey of audiovisual translation and related subjects. As Scandinavia has been called “a bastion of subtitling” (Ivarsson and Carroll 1998: 5), the focus is on interlingual subtitling and related topics. Most of this chapter is based on the research of others, notably Gottlieb (chiefly 1994, 1997, and 2001) and Ivarsson and Carroll (1998), but the ef- forts of many other researchers and practitioners are also included. Many of the terms used in the present study are introduced and explained here, and most of these are commonplace in the discipline and the profession of subti- tling. There are a few new angles on some aspects, however, notably ‘central cueing’ (3.3.2.1) and the ‘contract of illusion’ (3.3.4). The chapter contains a short synchronic survey of language transfer methods, but I have refrained from doing any historical review, as the focus of this study is not really dia- chronic. This survey culminates in a more detailed description of subtitling as a phenomenon and as a process, and of various other phenomena pertinent to subtitling, such as condensation and exposure times. 3.1 Language transfer on screen The issue of what constitutes translation is a sticky one, and there is dis- agreement between scholars and practitioners in the discipline as to what should be included in the term. Some, like Cattrysse (2002), take a broad view and include the adaptation of books and plays for the cinema, whereas others (e.g. Wildblood 2002) take the more purist view and do not even in- clude interlingual subtitling. In the present study, a middle road is taken, the core idea of which is that whenever a message is transferred from one lan- guage to another, translation takes place. However, there are many forms of language transfer that share common traits with translation, and these are dealt with in this section. Screen translation, which itself is a form of lan- guage transfer on screen, will nevertheless be discussed in the next section of this chapter. The scope of this survey is further limited by the audiovisual focus of the present study; forms of language transfer that do not include a screen, such as literary translation and most forms of interpreting are not included in the

Upload: phamkhuong

Post on 25-Mar-2018

254 views

Category:

Documents


1 download

TRANSCRIPT

1

3 Screen Translation

In order to show how Scandinavia fits into the European media translation scene, and to place the texts under investigation in a wider context, , this chapter is dedicated to a short survey of audiovisual translation and related subjects. As Scandinavia has been called “a bastion of subtitling” (Ivarsson and Carroll 1998: 5), the focus is on interlingual subtitling and related topics. Most of this chapter is based on the research of others, notably Gottlieb (chiefly 1994, 1997, and 2001) and Ivarsson and Carroll (1998), but the ef-forts of many other researchers and practitioners are also included. Many of the terms used in the present study are introduced and explained here, and most of these are commonplace in the discipline and the profession of subti-tling. There are a few new angles on some aspects, however, notably ‘central cueing’ (3.3.2.1) and the ‘contract of illusion’ (3.3.4). The chapter contains a short synchronic survey of language transfer methods, but I have refrained from doing any historical review, as the focus of this study is not really dia-chronic. This survey culminates in a more detailed description of subtitling as a phenomenon and as a process, and of various other phenomena pertinent to subtitling, such as condensation and exposure times.

3.1 Language transfer on screen The issue of what constitutes translation is a sticky one, and there is dis-agreement between scholars and practitioners in the discipline as to what should be included in the term. Some, like Cattrysse (2002), take a broad view and include the adaptation of books and plays for the cinema, whereas others (e.g. Wildblood 2002) take the more purist view and do not even in-clude interlingual subtitling. In the present study, a middle road is taken, the core idea of which is that whenever a message is transferred from one lan-guage to another, translation takes place. However, there are many forms of language transfer that share common traits with translation, and these are dealt with in this section. Screen translation, which itself is a form of lan-guage transfer on screen, will nevertheless be discussed in the next section of this chapter.

The scope of this survey is further limited by the audiovisual focus of the present study; forms of language transfer that do not include a screen, such as literary translation and most forms of interpreting are not included in the

2

survey. There are many excellent other sources for these subjects, such as Nida 1964, Toury 1995, Venuti 1995 or Landers 2001 for literary translation and the works of Wadensjö 1998, Gile 1998, 1999 or Pöchhacker 2004 for interpreting.

3.1.1 Intralingual subtitling

Intralingual subtitling, or Same Language Subtitling (SLS) as it is sometimes called (e.g. by Kothari et al. 2002), is a form of language transfer on screen that does not involve translation. Instead, there is only a shift from the spo-ken mode of the dialogue in a film or TV programme to the written mode of the subtitles. This is sometimes done for pedagogical reasons, e.g. for the benefit of language learners, but its main raison d’être is to be an accessibil-ity tool for the deaf and hard of hearing. As such, it exists mainly as a teletext option on television, which can be turned on or off at the viewer’s discretion (and also as an optional feature on DVDs). In this form, intralin-gual subtitling is often better known as ‘closed captions’, as opposed to ‘open captions’, which are subtitles for television that the viewers cannot turn off. It should be mentioned, however, that intralingual subtitles are not synonymous with closed captions. In Scandinavia, some programmes are subtitled intralingually using open captions, which are often removed for reruns of the programme.

In the last few years there has been an increase in both the production of and the academic interest in intralingual subtitles. In the United Kingdom, for instance, the Communications Act of 2003 requires that 60% of all pro-gramming in that country be subtitled for the deaf and hard of hearing, and the BBC has announced a goal of 80% by 2008 (Hearing Concern) and ulti-mately 100% soon after that. As a good example of the recent academic in-terest in intralingual subtitles, Sahlin (2001) deserves to be mentioned for her thorough methodology and substantial corpus.

In Scandinavia, the situation for the deaf and hard of hearing is approach-ing that of the U.K. In both Denmark and Sweden, the public service com-panies are aiming for 100% subtitling for all pre-recorded domestic pro-gramming in the near future, according to Gitte Heide Olsen of DR (personal communication (henceforth p.c.), October 2006) and Magnus Rönnlid of SVT (p.c., October 2006). Certain genres are excluded from this goal, how-ever, such as young children’s programming, which tends to be dubbed or voiced-over because of its being aimed at a pre-literate audience, and music programmes. To this should be added that all foreign-language programming is subtitled, so all in all, most of what is aired on Scandinavian television is subtitled, either intra- or interlingually. Comparatively little intralingual sub-titling in Scandinavia contains renderings of non-verbal audio material (e.g. a caption saying: “the phone rings”), however. Instead, most of what is pro-duced focuses on the verbal audio channel, rendering dialogue only.

3

Many of the concerns and media-specific constraints (cf. 3.3.3) of intra-lingual subtitling are shared by its translational counterpart. Both are subject to norms of exposure times, reading speed constraints, subtitle density and condensation issues. Pioneering efforts have also been made by Neves (2005) to bridge the gap between the two categories of subtitling, by produc-ing interlingual subtitles for the deaf and hard of hearing. These are interlin-gual subtitles (i.e. translations from other languages) that use all the extra annotation common to intralingual subtitles that give the deaf and hard of hearing access to the non-verbal audio features of a film or TV programme. In fact, this has been even further developed by Neves by experimental in-clusion of special symbols, such as emoticons (2005: 226ff.).

Even though intralingual subtitling shares many of the media-specific constraints of interlingual subtitling, it also shares a media-specific con-straint with dubbing (cf. section 3.2.2, below). As many people with hearing impairments use lip-reading as an additional source of information, the mes-sage in the subtitles should not be too far removed in lexis or syntax from the original, which means that a certain degree of synchronization is called for. There is thus a limit to how much paraphrasing is used in intralingual subtit-ling. This in turn calls for fairly verbatim subtitles, which might cause prob-lems of a different kind. The hearing-impaired are in no way a homogeneous group, as the degree of loss of hearing varies, and equally importantly the time when the hearing was lost. The congenitally deaf are known to be slow readers (cf. e.g. Sahlin 2001: 99 ff. or de Linde and Kay 1999: 11f) and would not be helped by verbatim subtitles, since these would require a very high reading speed.

3.1.1.1 Live subtitling

One form of subtitling that appears in both inter- and intralingual form is live subtitling. It is most commonly used for intralingual subtitling, so it will be discussed here.1 Not much research has been carried out on the topic of live subtitling so far, probably because it is a fairly rare phenomenon, but that is changing. For instance, a seminar was held by the University of Bologna’s department of translation, language and culture at Forlì about the topic in November 2006 (www.respeaking.net). Also, live subtitling is often brought up at conferences on multidimensional translation. The information in this section stems from such conference presentations, notably Müller (2005) and Carroll et al (2006).

Live subtitling, as the name suggests, is done in real-time when a pro-gramme is aired. It is used exclusively for live broadcasts (e.g. of sporting events) and television news. To aid the subtitlers, stenography or special ‘Velotype’ keyboards with short commands have been developed to increase

1 For a more detailed classification of the different kinds of subtitling that exist today, the reader is referred to Bartoll 2004.

4

their typing speed. However, it is nowadays common for live subtitlers to use respeaking. This is done with the aid of voice recognition software. The subtitler repeats an edited version of what is said on-screen and may also proofread and edit the subtitle in real-time. It is of course impossible to achieve the same synchrony in live subtitling as in pre-produced subtitling, but the lag is no more than a few seconds. This form of language transfer puts enormous stress on the subtitler and for this reason, it is common for more than one subtitler to be involved. Typically, two subtitlers take short turns to subtitle a TV programme when typing is the mode of production, and a great deal of preparation goes into the process, in the form of research-ing background data and issues that might come up during a broadcast. Of-ten, a simultaneous interpreter is also included in the team, as the skills of simultaneous interpreting (such as split-second decision making, and stress tolerance) are as important as those of subtitling, when it comes to live subti-tling (Carroll, p.c., May 2007).

In Denmark, DR started up a service of subtitling of live interviews and the like based on respeaking on October 1, 2006. This was done with the aid of two government ministries and a Danish software company, and initially in corporation with TV2, according to the head of subtitling at DR, Heide Olsen (p.c., February, 2007). Live subtitling for national news and some other live broadcasts has been the practice on SVT in Sweden since 1988, according to the head of subtitling at SVT, Rönnlid (p.c., October, 2006). This has been done using typing, but SVT is now moving into respeaking as well. On SVT, 60% of all live national programming was subtitled live or semi-live (using pre-typed subtitles that are cued live) in 2006. This is an impressive figure, but not as impressive as the BBC which subtitles 95% of its live programming live, meaning 20,000 hours of live subtitling per an-num, according to David Padmore of Red Bee Media (formerly BBC Broad-cast, p.c., October, 2006).

3.1.2 Audio description

Just as the hearing-impaired can get increased access to audiovisual material through intralingual subtitling, the vision-impaired can get increased access through audio description (AD). Just as interlingual subtitling makes the aural visible, audio description makes the visual aural. The idea is fairly simple: a narrator describes what can be seen on-screen in as great a detail as possible and/or necessary: “the action, body language, facial expressions, scenery and costumes” (Benecke 2004: 78). Even though the general idea might be simple, the execution of it is often very tricky business indeed. The narration should not block or clash with pre-existing dialogue, sound effects or relevant music. This makes AD very complicated in films that already contain a great deal of aural information, e.g. rapid dialogue, and such films are generally not audio described, according to Benecke (2004: 79).

5

Audio description is not nearly as widespread a phenomenon as intralin-gual subtitling for the deaf and hard of hearing, but it is on the rise. Accord-ing to Orero, the history of audio description in countries such as the U.K., the U.S., Canada and Spain shares a common development from amateur practice to professional endeavour:

Beginning with live and open AD for special play performances, moving to recorded AD for films to be sent internally to members of associations, to the hopefully comprehensive standard, public and professional AD which will be present in Digital television (DTV) and commercial DVDs (2005: 8).

According to Benecke (2004: 78), AD began in the early 1980s in the U.S. and developed somewhat independently in the U.S. and in various major European countries. In the U.S., the top four networks have to offer 200 hours of audio described programming each, and in the U.K. 4% of all pro-gramming has to be audio described, according to Benecke (2004: 78). In Scandinavia, audio description is very scarce indeed, even if Clark (2001) states that Sweden was the first country to offer audio description in a movie theatre. At the moment, AD, or syntolkning/synstolkning (“sight interpreta-tion”) as it is called in Sweden and Denmark is available on neither Swedish nor Danish television (Heide Olsen and Eva Åkerberg, p.c., October 2006). There are, however, moves towards this accessibility tool in Scandinavia. A few DVD films have an AD option, and there is pressure from lobby groups such as Dansk Blindesamfund ([The Danish Society for the Blind]) to intro-duce it on television as well (cf. Dansk Blindesamfund).

As indicated by Orero’s quote above, AD can be produced live, just like live subtitling. This is described by Matamala (2005) and Matamala and Orero (2007), who add that up to three different audio describers may be involved to describe different aspects of an opera performance in the U.S. Presumably, however, the intensity involved in live audio description may not be as heavy as it is for live subtitling. In Sweden, the European Athletics Championships 2006 were audio described live (Dansk Blidesamfund).

In Sweden, there is also a service called ‘talande textremsa’ (spoken sub-title), which is also known as audio subtitling (cf. Matamala and Orero 2007). This is a service for the partially sighted, and it is produced electroni-cally, as voice synthesizing software ‘reads’ the subtitles out loud. However, accessing this service requires additional hardware for the users. This service does not exist in Denmark at the present time.

3.2 Audiovisual translation ‘Screen translation’ and ‘audiovisual translation’ are two nearly synonymous terms, as indeed are ‘media translation’ and ‘multimedia translation’. The

6

last term is theoretically the broadest, but it is arguably most often used for the translation of multimedia applications, such as computer games, web sites and computer software (cf. e.g. Hernández Bartolomé and Mendiluce Cabrera 2005). This in turn is closely connected with issues of localization, which is beyond the scope of the present survey. Instead, the reader is re-ferred to O’Hagan (1996), O’Hagan and Ashworth (2002) and Pym (2004, 2006). According to Google (www.google.com), ‘media translation’ is the most common term, but ‘screen translation’ and ‘audiovisual translation’ (AVT) are the most common terms in academic circles, even though ‘media translation’ is also used here2. ‘Audiovisual translation’ would include all forms of translation that appears in an audio and/or visual context. ‘Screen translation’, on the other hand, would include any translation that appears on a screen, be it the silver screen, a computer screen or TV screen. This would imply that surtitling for the opera and theatre is included in AVT, but not necessarily in screen translation. In these contexts, AVT would thus be a superordinate term to screen translation. On the other hand, surtitled texts are displayed on screens, even if the source text is not, so the terms can also be used interchangeably. In the present study, ‘Screen Translation’ and ‘AVT’ will therefore be used as synonyms.

The three main modes of AVT are subtitling, dubbing and voice-over. The first of these is discussed at length in section 3.3, as it is the form of AVT that the present study investigates. The first two of these will be briefly discussed presently. First of all, however, it may be warranted to give a few observations on a form of language transfer that is very interesting as it could be said to be based on quasi-translation: versioning.

3.2.1 Versioning

Unlike the three prototypical forms of AVT (dubbing, subtitling and voice-over), which are forms of post-production translation, versioning could be described as pre-production translation. Instead of producing a film or TV programme and then translating it into various target languages, in the case of versioning multiple-language versions of the same film or TV programme are produced.

Versioning is one of the oldest forms of AVT. According to Gottlieb, it began already in the first few years of sound film. In its infancy, the same director shot the same film twice (or as many times as required) with the actors speaking different languages (1997: 52). Chiaro (forthcoming) de-scribes how this mode of AVT became so popular for Laurel and Hardy films that when dubbing later was used, the Italian dubbing actors had to mimic Laurel and Hardy’s American accents in their Italian. This mode of

2 For instance “(Multi) Media Translation” was used in the title of Gambier & Gottlieb (2001).

7

transfer was of course very costly, and such ‘simultaneous’ versions are hard to find today.3 Instead today, we have what Gottlieb calls “adaptations” and “formatted versions” (1997: 53).

In adaptations4, the rights to produce a TL version of a SL programme or film are bought by a TL producer. The TL producer would then stage the film or TV programme again, in the TL with TL actors. The ST script is bought along with the production rights, so it is still a case of translation (Gottlieb 1997: 53). However, it could be described as a covert (cf. House 1997: 69) form of translation, as the fact that the viewers are watching a translation is often toned down. More often than not, the ST is only men-tioned in the end credits, and the title of the product is usually changed. Also, a great deal of cultural adaptation and other strategies are used in the translation, which makes the TT appear as a TL original. Adaptations are also known for being freer than other forms of translation, so much so that whole scenes are edited out or added. There is thus a good reason for talking about a ‘version’ (in House’s (1997: 73f) terms) rather than a ‘translation’.

A few Swedish examples of adaptations of anglophone TV series are Rena rama Rolf, TV4 (the Honeymooners, CBS), En fyra för tre, TV4 (Three’s Company, ABC) and Albert och Herbert, SVT (Steptoe and Son, BBC). This final example is often seen as a typical Swedish comedy, and people appear very incredulous when they are told that it is an adaptation. Very successful adaptations, such as this one, may take on lives of their own, with new episodes or whole series being written, which make use of the original characters and setting, but which have no SL screenplay counterpart, in which case we can no longer talk about a translation. Occasionally, an original TV series, such as Only When I Laugh (BBC) has been shown in Sweden, only to be adapted and rerecorded in Swedish a decade or two later, in this case as Sjukan.

Adaptation is also used for films. In Scandinavia we have the example of the very popular Danish Olsenbanden films (1969–1981), which were adapted for Norway, using the same title during a longer period, 1969–1998, and later for Sweden as Jönssonligan (1981–1995). As regards feature films, this is a common form of AVT on the American market. The general Ameri-can public is known to be reluctant towards subtitles (cf. e.g. Zoglin (1992), The Economist (1997)), and there are many examples of how films in other languages have been versioned in Hollywood, such as the Danish film Nattevagten (1994), which starred Ewan McGregor in its 1997 Hollywood version Nightwatch. In fact, there is even a further degree of cultural domes-tication for the American market. The British mockumentary sitcom The

3 One such example is the German version of Monty Python’s Flying Circus made by the original cast in 1971 (www.IMDb.com “Monty Python’s fliegender Zirkus”) 4 Dries (1995: 40–41) further divides adaptations into “remakes” and “double versions”, the main difference being the time lag between the versions. Her term for “formatted versions” is simply “formats”.

8

Office, BBC, from 2001 was made by NBC in an American version of the same name in 2005.

If adaptations are just borderline translations, it is even harder to find ar-guments for calling formatted versions translations. In the case of formatted versions, a programme format is exported, rather than a programme script. The TL producers buy the format of the show, often including rights to lo-gos, a set or even a locale (e.g. in the French format Fort Boyard, France2, which has both Swedish and Danish formatted versions) and then fill it with TL content. Perhaps the most common genre for formatted versions is the game show. Examples of these could be Who Wants to be a Millionaire, ITV and Jeopardy!, NBC, both of which exist in Swedish and Danish formatted versions produced locally.

3.2.2 Dubbing

When a film is dubbed, the original SL soundtrack is removed and replaced by a TL soundtrack. This means that dubbing creates “in theory, the perfect illusion” (Gottlieb 1997: 54) of not being translated at all. In actual fact, there can be a huge credibility gap between the linguistic ‘reality’ produced by the soundtrack, and the cultural reality of the images, and the content of the dialogue. However, in the dubbing countries, the audience is used to that and is not bothered by the fact that John Wayne apparently speaks German or that six very American friends drinking coffee in Manhattan discuss an outing to Central Park in French. For people in subtitling countries, such anomalies are hard to come to terms with, because of their background, and they often find it hard to “suspend their linguistic disbelief”, as Romero Fresco (2006) puts it, the way that people in dubbing countries do. In a later section (3.3.4), however, I will show that people in the subtitling countries are indulging in a similar kind of suspension of disbelief.

As Chaume (2004b: 40) points out (a point also made by Gottlieb 1997: 54) dubbing can be both intra- and interlingual. Interlingual dubbing is the most obvious form, as the replacing soundtrack is in another language, but intralingual dubbing is probably more common. In this case, the replacing dialogue is identical to the replaced dialogue, with the difference of being produced after the filming of the scenes. The reason for intralingual dubbing is to ensure sound quality, particularly on outdoor sets. For the purposes of this survey, however, we will only concern ourselves with interlingual dub-bing, as this is the kind that involves translation.

Just as subtitling has its media-specific constraints, which will be de-scribed later (3.3.3), dubbing has constraints which can be very complex indeed. Most of these are to do with various forms of synchronization, and these have been described thoroughly by e.g. Chaume (2004b). The most obvious form of synchronization is lip-synch, the matching of the words to the characters’ lip movements. In extreme close-ups, this needs to be quite

9

rigorous to meet standards of quality, especially when it comes to open vow-els and labial consonants (Chaume 2004b: 41). Chaume lists two more forms of synchrony, namely kinetic synchrony, which matches the voices with the characters’ body movements, and isochrony, which matches the length of the dubbed utterance to the length of the original utterance (2004b: 41).

From the audience’s perspective, the use of dubbing has the comfortable advantage of hearing everything in their own language. From a production perspective, it has the economic disadvantage of being about 15 times as expensive as subtitling, according to Luyken et al. (1991: 105). Dries (1995: 14) advises that Luyken et al’s figures should not be over-emphasized, as they are based on a European average, and the cost of dubbing varies a great deal between countries. Also, even if Luyken et al’s study might be the most comprehensive of its kind, it is somewhat dated by now. Dries’s own study indicates a more current difference, with dubbing being ten times as expen-sive as subtitling (1995: 30). The reason for the difference in cost is that, apart from the translating and editing of the script, dubbing requires a whole new cast of dubbing actors. From a language political perspective, dubbing has been described as nationalistic or protectionist. For instance, in Italy, the Fascist government forbade the use of any language other than Italian on the silver screen in the 1930’s (Guardini 1998:91). In their historical overview of AVT, Ivarsson and Carroll write:

Dubbing was favoured in countries with very strong nationalistic currents where vehement defence of the national language was common. These voices had been heard for centuries in France and Spain and were even more force-ful in the recently united countries of Germany and Italy. With the rise of Fascism in Italy and Spain and once the Nazis took power in Germany, legis-lation was introduced in these countries sanctioning dubbing and forbidding or limiting subtitling. In France of the 1930’s only ten of the country’s 4000 cinemas were permitted to show subtitled films. […] These countries have stayed strongholds of dubbing until today. (1998: 10–11)

The historical coincidence of the advent of the talkies and the instalment of Fascist or Nazi governments in three of the four main dubbing countries in Europe is thus used to explain why these countries use dubbing as their main form of AVT today. The weakness of the argument is that France was never Fascist, even if French linguistic protectionism is well known. The economic argument coincides with the historical argument, at least for Western Europe. As the cost of dubbing is so great, it only makes economic sense to dub if the language community is large enough. This would explain why Spain is a dubbing country, while its neighbour Portugal is a subtitling coun-try. There are fierce arguments both on the side of subtitling and on the side of dubbing (cf. Gottlieb 1997: 55ff, Diaz Cintas 1999, and Kilborn 1989, 1993). In the end, the choice probably boils down to what the viewers are used to, and it is as hard to make a French person read in the cinema, as it is

10

to make a Dane suspend her/his linguistic disbelief. Lately, however, it has been easier to find cinemas showing subtitled films in Germany.

In Scandinavia, dubbing is almost exclusively found in films and TV pro-grammes for pre-school children and cartoons and computer-animated fea-ture films like Shrek (2001) or Toy Story (1995). Lately, however, animated films from Disney, Dreamworks and similar studios are now released in both dubbed and subtitled versions in Scandinavia, when in 1990, only the dubbed versions were available.

3.2.3 Voice-over

As the name suggests, voice-over is a translation superimposed on the origi-nal dialogue. The original dialogue is toned done (but still audible) and one or two speakers read a translation of the original dialogue. Normally, this translation method is slightly asynchronous, giving the audience a few sylla-bles of the original dialogue at the beginning and the end of each utterance unimpeded, without any TL voice-over, presumably to give the audience a chance to hear what the original sounds like and who the speaker is.

Voice-over could be described as a budget form of dubbing, as it does not involve a full cast of dubbing actors; one or two speakers can do the job. According to Luyken et al’s somewhat dated figures, the cost of voice-over is about 1/7 of dubbing, but still twice as expensive as subtitling (1991: 105). On the other hand, voice-over – unlike dubbing, but like subtitling – is an overt form of translation, as the original is audible in the background.

Today, voice-over is mainly used for TV translation in Eastern Europe, e.g. in Poland and Russia, where it is almost exclusively used for foreign language programming on TV. It can occasionally be found in Western Europe as well. In documentaries and news reporting, e.g. on CNN, when a person speaking some ‘exotic’ language is seen on screen, voice-over is of-ten the preferred mode of AVT. This is particularly true of translation into English. In Scandinavia, these situations normally call for subtitling. In Denmark, and sometimes in Sweden, what is called third-person voice over (i.e. the narrator reads the translation in third person, rather than ‘pretending’ to be the person seen on screen; Gottlieb 2001: 78) is used for children’s programming from other Scandinavian countries. A case in point is the popular Swedish Astrid Lindgren stories, which are very often voiced-over when aired in Denmark. Apart from Pippi Longstocking et al, voice-over is rare in Scandinavia today, even if it was part of the origin of Scandinavian television. Gottlieb and Grigaravičiūt÷ (1999/2001: 76) describes the follow-ing situation from TV’s infancy in Denmark:

Before introducing subtitles in 1955 […] DR, Danish national TV had its (few) foreign films translated by a film professor who read all the lines off-screen to a small, but enthusiastic Danish audience. However, this practice

11

was cancelled after the (male) professor had done his best to render three French girls in an agitated discussion […]. Then the director general of DR interfered, and since that day no feature films have been voiced-over on Dan-ish TV.

3.3 Subtitling Even though both dubbing and, to a lesser extent, voice-over, occur in Scan-dinavia, subtitling is by far the AVT of choice in the Nordic countries and the rest of the present study will, as its name indicates, deal with Scandina-vian subtitles. Furthermore, apart from what was said about intralingual sub-titling in 3.1.1, it will deal with interlingual subtitling, i.e. subtitling that includes translation. Thus, ‘subtitling’ will henceforth be used as an abbre-viation of ‘interlingual subtitling’, unless otherwise specified. A distinction will also be made between process and product. The term ‘subtitling’ will be used about the activity of producing subtitles, and also about the field in general. The term ‘subtitles’ will be used about the end product of the proc-ess of subtitling, i.e. the subtitles themselves.

I will refrain from writing a historical description of the development of subtitling in Scandinavia or elsewhere. Instead, the reader is referred to Ivarsson (2002), Ivarsson and Carroll (1998: 9–32), Gottlieb (1994: 13–28, 1997: 49–68, 2001: 29–34) and Tveit (2004: 23–27). However, in chapter 5, a short diachronic investigation of the development of Scandinavian subti-tling norms of the last twenty-odd years will be made. Also, in chapter 8 there will be a brief diachronic discussion, as this is of importance to the main investigation in the present study. Furthermore, I will not make any survey of the different media for which subtitling is used in Scandinavia today, even if there are interesting differences in the techniques and norms which govern subtitling for the cinema, DVD and TV. For an exposé on subtitling for these media, the reader is referred to Ivarsson and Carroll (1998: 19–32, 63–71) or Bartoll (2004). The focus in this section, as indeed in the rest of the present study, will be on television subtitles.

3.3.1 The nature of subtitles

If you ask the man on the street, or rather on the TV couch, what subtitles are, you are likely to get an answer that sounds something like this: “Subtit-les are lines at the bottom of the screen that display a translation of the dia-logue.” Even though this definition is easily accessible, for the present pur-poses, it is clearly inadequate, and it can also be shown to be erroneous. First, a subtitle could be one single line, or it could be any number of lines. In theory, you could cover the whole screen with subtitles, even if that would be rather pointless. Normally a subtitle consists of one or two lines (referred

12

to as a one-liner or a two-liner, respectively), but it could occasionally con-sist of three lines. Second, the lines need not be at the bottom of the screen. In Japan, for instance, subtitles are sometimes placed vertically on the side of the screen (Gottlieb 2001: 15). That never happens in Scandinavia, but occasionally, subtitles may be placed in other positions than at the bottom of the screen, for instance if there is a risk of their blocking something impor-tant in the picture there. For instance, subtitles (or supertitles?) displaying the name and occupation of the speaker could be placed in the top left-hand corner of the screen. Third, as explained in section 3.1.1, the texts in intra-lingual subtitles are not translations. Finally, the subtitles do not only display the dialogue; instead they reproduce any verbal material in the ST. Not in-frequently, subtitles show written material from the ST, e.g. captions, street signs, headlines or letters.

Gottlieb has a much more accurate semiotic definition. According to him, subtitling is:

A. Prepared communication B. using written language C. acting as an additive D. and synchronous semiotic channel, E. and as part of a transient F. and polysemiotic text.5

(2001: 15, emphasis in the original)

Each of the emphasized words in this definition serves to differentiate subti-tling from other forms of translation. A contrasts subtitling with simultane-ous interpretation, but does not work very well for live subtitling (cf. 3.1.1.1 above). B contrasts subtitling with dubbing and voice-over. C contrasts sub-titling with dubbing or literary translation, as subtitling adds a semiotic channel of information, whereas dubbing replaces an existing channel. D refers to the notion that subtitles should come on screen at the exact moment when an utterance starts, and ideally end at the end of an utterance. Very often though, subtitles tend to linger somewhat after the end of an utterance, unless the constraints of the dialogue prohibit it. Subtitling is thus less syn-chronous than dubbing, which ideally should match the original utterance word for word, and sometimes even phoneme for phoneme. Like D, E con-trasts subtitling with literary translation and refers to the fact that the subtit-les follow the image and disappear after reading. Unlike the case of literary translation, the TV viewer cannot go back and reread a part of the text that s/he has missed or misunderstood. This explains why individual subtitles should as far as possible be syntactically and semantically self-contained (cf.

5 This definition does not say anything about translation; it is open to both intra- and interlin-gual subtitling.

13

points 4–7 of the Code of Good Subtitling Practice, Ivarsson and Carroll 1998: 157). F also contrasts subtitling with literary translation, but says more about the medium in which it occurs.

The concept of the polysemiotic text has been developed by Gottlieb on many occasions. According to Gottlieb, “[t]he term ‘polysemiotic’ refers to the presence of two or more parallel channels of discourse constituting the text in question” (2004: 227). Gottlieb divides a polysemiotic text, such as a feature film or television programme, into four distinguishable semiotic channels. These are:

1. Verbal audio: the dialogue and its paraverbal elements 2. Non-verbal audio: (background) music and sound effects 3. Verbal video: displays and captions 4. Non-verbal video: composition and montage

(Gottlieb 1997: 143, my translation)

Subtitles primarily interact with the two verbal channels, editing and trans-lating them (or editing and transcribing them, in the case of intralingual sub-titling). However, intralingual subtitling may also transcribe the non-verbal audio channel and all subtitling should be in harmony with the flow of the non-verbal video channel.

Interlingual subtitling is unique in that the message is not only transferred from one language to another, but also from one mode to another: from the spoken mode (usually) to the written mode. Gottlieb has called this semiotic jaywalking (2001: 16), in that it does not go straight from spoken SL to spo-ken TL, as in interpreting, nor from written SL to written TL, as in literary translation, nor straight from the spoken to the written code of the SL or the TL as would be the case in transcription. Instead, interlingual subtitling “jaywalks” or crosses over from SL spoken mode to TL written mode. So, whereas literary translation or interpreting are horizontal in nature, interlin-gual subtitling is diagonal. Intralingual subtitling, on the other hand, could be described as vertical, as it is a form of (edited) transcription.

The move from the spoken to the written format brings with it a necessity for editing: “[…] the deletion or condensation of redundant, oral features is a necessity when crossing over from speech to writing” (Gottlieb 2001: 20). This is particularly true when it comes to spontaneous, non-scripted speech. In fact, when many people see a carefully made verbatim transcript of their own spontaneous utterances, they see it as near gibberish, so a certain reduc-tion due to the semiotic transfer from one medium to another is probably a good thing. Otherwise, “the audience would be taken aback by reading the oddities of spoken discourse” (Gottlieb 1997: 113). This may not be true to such a high degree when dealing with the scripted speech of fiction. After all, we are dealing with a ‘polysemiotic U-turn’ there, from written script to oral performance and back to written subtitles.

14

3.3.2 The subtitling process

The preceding section discussed the semiotic nature of the medium of subtit-ling. This section will discuss the practical reality of subtitling. Bearing in mind that there is a great deal of variation in the working processes and con-ditions of individual subtitlers, what I call the ‘subtitling situation’, I will here try to outline the process of subtitling. This will be done very briefly, as the object is not to instruct anyone in how to become a subtitler (there are more suitable texts for that, e.g. Ivarsson and Carroll 1998, Tveit 2004 and any number of in-house guidelines), but rather to explain some of the terms and notions that will surface in later parts of the present study. In this sec-tion, as in the rest of the present study, I will primarily concern myself with subtitling for television, even if the subtitling of DVDs, cinema releases, corporate videos, video games etc. is fairly similar (cf. Ivarsson and Carroll 1998: 63–71 for some of the more important differences).6 The outline of the process in this section is based on my own experiences as a subtitler, pre-scriptive guidelines, and the accounts of other subtitlers, notably Wildblood (2002), Pollard (2002) and Ivarsson (2002).

The process starts with a subtitling company procuring a commission from a broadcaster to subtitle a film or a TV programme. Traditionally, when the originator of the process is a public service broadcaster, this first step has been left out, as they have had their own in-house subtitlers, but that is now changing. For the commercial sphere, the commission is normally awarded to the lowest bidder. The subtitling company subcontracts the commission to a freelance subtitler, who receives either a digital file or an analogue VHS tape of the film to use as the basis of her/his translation. The subtitler also often, though not always, receives a script of the dialogue as an auxiliary source of information. More often than not, these scripts are not the original screenplays, but post-production transcripts, which record any changes to the original screenplay and thus give a rendering of the final ver-sion of the film. However, these transcripts are often unreliable, and sub-titlers are told to give priority to the film itself. Superimposed on the film is an electronic time code, which could be said to be an electronic clock, that starts at the beginning of the film, and then counts hours, minutes, seconds and frames. There are 25 picture frames to a second in European television (Ivarsson and Carroll 1998: 141). A frame could thus be said to be used as a measure of time in subtitling, representing 1/25 of a second. The reason for having frames, rather than fractions of seconds as the smallest unit of time is perhaps obviously of a practical nature. Even though film appears to be mov-ing pictures (i.e. ‘movies’), it is not continuous, but a rapid succession of still pictures, technically known as frames.

6 These differences are mainly of a technical character, such as different reading speeds and more use of one-liners and a different airing process in the cinema, as the subtitles are etched into the film itself; cf. Ivarsson 2002.

15

The next step in the process is the subtitling itself. The subtitler has a multitude of tasks to juggle (cf. e.g. Nørgaard 1989 or Gottlieb 2001: 41–52). The most important ones are to segment the ST into coherent subtitles, to translate the subtitles into the TL and to cue the subtitles onto the elec-tronic time code, so that the subtitles appear and disappear on the screen at the proper time. This last task is known by a number of terms, e.g. spotting, cueing and time-coding. ‘Cueing’ will be the term used henceforth. Now-adays, most subtitlers work from home, using one of a number of PC-based software programmes for subtitling. The film can either be integrated with the programme, in which case the cueing can be done as you translate, or the subtitler can work with twin screens, segmenting and translating first, and then cueing the subtitles onto the time code later. The process can also be done the other way around, so the segmenting and cueing is done first and the translation is done afterwards. The software programmes have perks that assist the subtitlers in their task, by indicating whenever a line becomes too long, or ‘nudging’ the original cueing so it coincides with scene changes. Modern software also has voice detection functions, which help the subtitler align the subtitles exactly to the soundtrack. These improvements in techno-logy, in combination with shorter deadlines and heavier workloads have shortened the preparation time for one hour of (intralingual) subtitled pro-gramming from approximately 40 hours in 1980 to less than ten hours today, according to Lambourne (2006). He also points to another related and equally important trend in the subtitling process, namely that prices of (in-tralingual) subtitling have been cut by two thirds since 1990.

The result of the subtitling activity is an electronic subtitle file, with the in and out times of the subtitles marked. Converted into a normal text file, the result can look like the extracted example below:

00:08:25:16 00:08:32:00

Hun vil hellere steges levende end gå om bord i din stinkende jolle igen! 00:08:32:04 00:08:37:00 Vær du glad for, jeg ikke giver dig en endefuld!

The example shows how the first subtitle begins at 8 minutes, 25 seconds and 16 frames (i.e. 0.64 seconds) after the beginning of the film (in this case The Piano) and stays on the air for slightly more than 6 seconds, until 8 minutes and 32 seconds after the beginning of the film. There is then a gap of four frames (i.e. 0.16 seconds) before the next subtitle comes on. There is always a very small gap like this between subtitles; otherwise the viewers might not notice that another subtitle has appeared. The next subtitle then

16

comes on at 8 minutes, 32 seconds and 4 frames, and stays on until 8 min-utes and 37 seconds.7

The subtitler has other tasks, apart from segmenting, cueing and translat-ing. The most important one could be called editing, which means that the subtitler needs to adapt the text in the subtitles, so that it meets the spatial and temporal constraints of the medium. I will return to this issue in the next section, and again in chapter 5, but as a rule, this means reducing or con-densing the verbal content in the original polysemiotic text.

Another task of the subtitler is to research the ST. This includes checking the spelling of any names or other verbal material that are to be retained in the TT. It also means making sure that any references, cultural or otherwise, in the ST are represented in the subtitles in a coherent way, and making sure that the subtitler understands all aspects of the ST in order to render such references in an accurate way in the subtitles. In highly topical program-ming, such as current affairs programmes and topical talk shows, researching takes up a considerable amount of the subtitler’s time.

The subtitler then proofreads and re-edits the subtitling file (often several times) before handing it back to the subtitling company. In many cases, the subtitling company has an editor who revises the subtitling file, looking for errors and discrepancies. This person can also occasionally be called upon to do the cueing of the segmented and translated text. As of November 2006, revising (and also reviewing, if so required) by someone other than the sub-titler is a mandatory step in the subtitling process for any company certified to meet European standards of translation quality (BS EN ISO 9000: 2005 and BS EN 15038: 2006). However, there were no such companies in Scan-dinavia when the present survey was carried out.

The final step of the process is then that the subtitling file is turned over to the original commissioner, who broadcasts the subtitle file (or files, as international channels, such as the Discovery channel, may air in up to eight languages simultaneously (cf. Pollard 2002: 26)) along with the ST, in the case of TV subtitles. Synchrony between the ST and the subtitles is secured through the use of the electronic time code. It has been known to happen that the time code is broken and/or is erroneous, in which case the match will be incomplete. In that case, the subtitles will not fit the utterances on the screen, which is possibly the most annoying subtitling error of all. However, this can normally be sorted out during the airing process (cf. Pollard 2002: 26).

3.3.2.1 Central cueing

The process described above is much simpler than it was as recently as twenty years ago (cf. e.g. Ivarsson 2002), but as each step in the process adds to the cost of the final product, there is an incentive among commercial

7 It is a mere coincidence that both subtitles disappearance coincides with whole seconds and zero frames. The out-time of a subtitle can be at any number of frames.

17

subtitling companies to simplify it even further. This has been done through a process that I, for want of a better term, call ‘central cueing’, which basi-cally means that the cueing is done centrally, rather than by individual sub-titlers. Central cueing is highly relevant to the present study, as its effects on TTs and on national norms of subtitling is substantial, and this will be fur-ther discussed in chapter 5 and 8.

When subtitling DVDs, subtitlers are often provided with a so-called ‘Genesis’ file (cf. e.g. Gardner 2005: 30), which is basically an intralingual subtitling file of the ST. The advantage of this is that the segmentation and cueing have already been done, allowing the subtitler to focus on translating, thus minimizing the cost for each translation, by removing the burden of two of the subtitlers’ main tasks. This is also done when subtitling for television, but here, a Genesis file is rarely used. Instead a first-generation translation is used as ‘master template file’. This means that a first-generation subtitler carries out all the tasks described above. The resulting subtitling file is then given to second-generation subtitlers, who can focus on producing transla-tions into other TLs without having to bother about segmentation or cueing. In this way, a great deal of time and money is saved. I came into contact with the process as a subtitler and the process was further explained to me by Johan Norberg of SDI Media (p.c., September 2004).

In the early days of central cueing, there were some positive reactions to it (cf. e.g. Ivarsson 1989: 107), because of the savings in time and money it afforded. However, there have been many negative reactions as well. One effect of central cueing is that apart from the time code, the original transla-tion is normally also sent to the second-generation subtitlers, who may use it as a pivot translation8 or choose to ignore it and work straight from the ST. As the Scandinavian languages are broadly mutually intelligible (at least in their written forms), it is very common for subtitlers to use the first-generation translation as pivot translations, as it saves them much of the labour of editing. The second-generation translation is then just written into the same subtitles as the first-generation translation, replacing it. In these cases, the first-generation translation could in fact be considered the ST (or at least a co-ST) of the second-generation translation. There are a number of drawbacks associated with pivot translation, as it makes the TT one step further removed from the ST (cf. e.g. Gottlieb and Grigaravičiūt÷ 1999 or Zilberdik 2004). However, most of these appear only when the translator’s knowledge does not include the SL, but only the language of the pivot trans-lation, and this is rarely the case in central cueing. On the contrary, as master template files used in e.g. Croatia may be produced in Sweden, or vice versa, and most commercially successful films today tend to be in English, quite the opposite is true. However, in Scandinavia most master template files are

8 A pivot translation is a translation from which other translations are made, rather than mak-ing these translations directly from the ST; cf. section 2.1.

18

in another Scandinavian language, and as a result of poor proofreading, you sometimes find traces of the first-generation translation in the second-generation subtitles. In the main corpus of the present study, there are a few lines of Norwegian in the Swedish subtitles of The Simple Life (8: 8.25).

One beneficial effect of central cueing is that you get an extra second reader (the second-generation subtitler) who scrutinizes the first-generation subtitles, and who may spot and report any errors in them, unless it has al-ready been aired. This might have been the case with a centrally cued film in the main corpus of this study (What Women Want), where the Swedish first-generation translation had four rather serious translation errors, which were all gone in the Danish second-generation translation. One less beneficial effect is that the pivot translation may itself be a source of translation errors, as in an oft-cited example (from “bøfsiden” www.titlevision.com). Here, ‘breakfast sausage’ turned into a Danish term corresponding to ‘lunch bas-ket’ through interference from a Swedish pivot translation, as the Swedish word for ‘breakfast sausage’ (frukostkorv) is very similar to the Danish word for ‘lunch basket’ (frokostkurv). The work of the second-generation sub-titlers is perhaps less interesting than first-generation subtitling, which may cause some of these peculiarities. With the segmenting, cueing and editing (and pragmatically speaking also the research) removed from the task, what is left is less challenging. The second-generation subtitlers may feel that they are not producing any original work, hence the tendency of not crediting second-generation subtitlers.9

Whether or not the first-generation translation is used as a pivot transla-tion, the centrally cued time code must be used by the second-generation subtitlers, who may, but rarely do, alter it, unless a serious error is discov-ered in it. That means that the cueing of the time code is not changed to match any national norms or individual subtitlers’ preferences. This ten-dency has caused some concern in the subtitling world (cf. e.g. Imhauser 2002: 23). The effect of central cueing on national norms will be investi-gated empirically in chapter 5.

The logical next step after employing central cueing is of course to use translation memories and even machine translation for going from the first-generation to the second-generation translations. There have been experi-ments with machine translation into other Nordic languages from a first-generation translation, which I was involved in some years ago, but the re-sults were discouraging. Still, machine translation keeps developing, and new techniques using statistical analysis and translation memory are ex-plored for these purposes, which are showing greater success rates (cf. Volk and Harder 2007 or Armstrong et al 2007).

9 Subtitlers (and subtitling companies) are normally credited, most often by having their name appear in a separate subtitle when the end credits start or when the name of the scriptwriters of the ST appears.

19

3.3.3 Constraints of subtitling

The constraints of subtitling are so marked and so common as to seem con-stitutive, resulting in subtitling being called “constrained translation” (Tit-ford 1982). It has later been shown that other forms have their constraints as well (e.g. by Chaume (2004b) for dubbing), and if you take such paratextual considerations as deadlines into account, it is probably safe to say that all forms of translation are constrained. However, the constraints of subtitling are what Gottlieb calls “infamous” (2004: 219), and many translation deci-sions that are made in subtitling would be incomprehensible without knowl-edge of these, so they will be surveyed briefly in this section. One constraint has already been discussed (in section 3.3.1 above), namely the issue of the semiotic switch from spoken to written language, which brings with it thor-ough editing of (spontaneous) speech. Another issue is the spatial and tem-poral constraints and the condensation that these bring with them.

3.3.3.1 Spatial and temporal constraints

The spatial constraints of subtitling are based on the simple fact that you can only fit a certain number of characters into a line, and that you can rarely use more than two lines in each subtitle. At this point, it should perhaps be pointed out that the character and not the word is considered the basic unit in subtitling, as a word can be any number of characters.10 It should also be pointed out that a ‘character’ could be defined as any visible result produced by pressing a keyboard key. This means that in subtitling e.g. commas, full stops, dashes, exclamation marks and blank spaces are also characters.

The number of characters that can be fitted into a single line varies, de-pending on a few factors. In subtitling, italics need more space than un-marked text. Capital letters are very bulky, and ‘m’ and ‘w’ take up more space than ‘i’ and ‘l’ (cf. Ivarsson and Carroll 1998: 44). In academic texts on subtitling, there is quite a range as to how many characters you can fit into a line; Schröter (2005: 27) states a range of 30–40, Tveit gives 38 as a Scandinavian maximum (2004: 107) whereas Luyken et al (1991: 43) gives 28–38 as being the most common figures. It should be noted that the low figure of 28 stated by Luyken et al may be caused by its being the oldest of these sources and that for technological reasons, subtitled lines used to be shorter before the process was completely computerized. Current Scandina-vian guidelines give a maximum of 35 characters per line (fewer if italics are used), but in actual fact, you regularly come across lines that are longer. In the main corpus of the present study, there are examples of lines with as many as 42 characters. In view of this, and to have a round number for cal-culations, a full line of contemporary subtitling in the present study is con-sidered to have 36 characters, and a full two-liner 72 characters, even if that 10 However, in a background study (Pedersen 2003a: 8) I found that the average length of a word in soaps and sitcoms, in both SL (English) and TL (Swedish), is circa five characters.

20

is in the upper half of the scales given above. For older subtitles, a shorter line of 28–32 characters is considered normal. It should, however, be borne in mind that most lines are not full.

The temporal constraints are closely linked to the spatial ones, in that the message contained in the lines needs to be displayed for a certain amount of time (the so called display or exposure time) in order for the viewer to be able to read it. Generally speaking, most writers in the field consider some three seconds for a full one-liner (de Linde and Kay 1999: 7) and six seconds for a two-liner (Ivarsson and Carroll 1998: 64f) for TV subtitles something of an average. Gottlieb notes a limit of 12 characters per second, something which has been called the 12 cps rule. That means that a full two-liner, i.e. 72 characters should stay on the screen for 6 seconds (72/12 = 6) in order for 90% of the viewers to be able to read it (Gottlieb 2001: 20). Reality is more complicated than this, however, as you normally are supposed to take the TT audience into account. Children’s programming is often given longer expo-sure times, because of the reduced reading speed of the TT audience. Also, the nature of the TT affects reading speeds: complex lexis and syntax require longer exposure time. The polysemiotic nature of the medium also affects reading speed. If there is much information coming through the non-verbal video channel (as is the case in action-packed scenes), the viewers’ reading speed is lowered by their attention being focussed on the picture. Con-versely, if the verbal audio channel coincides with the subtitles to a high extent (as it would if e.g. long names and titles are retained in the subtitle), the intersemiotic redundancy offers a possibility to raise reading speeds. However, I know of no research into whether these factors are actually taken into consideration in real life.

3.3.3.2 Condensation

In Gottlieb’s defining features of subtitling (cf. 3.3.1 above), there is no mention of condensation, which some might find surprising. Yet condensa-tion (or the slightly more negative term ‘reduction’) is not a necessary prop-erty of subtitling; it is just very common. Because of the constraints of subti-tling, all the verbal content in the ST cannot always be represented in the subtitles. Something usually has to be left out, or rather: the remaining TT message has to be edited so as to (ideally) say the same thing, in fewer words (or at least in fewer characters), as the ST, and that is what is meant by condensation.

Even if condensation is extremely common, there are a few cases when no condensation is needed. This is often the case when captions are subtitled, or when the dialogue is very slow. In the last instance, there are also exam-ples of expansion of the original utterance in order to maintain a consistent pace in the subtitles (cf. Wildblood 2002: 41). This is, however, rare, and even slow speech is often condensed slightly as a result of the switch from spoken to written language.

21

The condensation rate varies, depending on the pace and complexity of the dialogue. According to Gottlieb, “[e]specially with up-tempo speech, the subtitler may choose to sacrifice close to 50 % of the dialogue – measured in quantitative terms [i.e. a simple word count of ST and TT] – in order not to exceed the normal television ‘speed limit’” (2001: 20). I have found exam-ples of extremely rapid and nonsensical speech being reduced by 75% of the original (in the sitcom Little Britain, BBC). However, these are extremes, and the average condensation rate is much lower. A Norwegian study of subtitling of feature films (Lomheim 1995, quoted by Gottlieb 2001: 57) shows a quantitative reduction of 22–35%. De Linde and Kay (1999: 51) found quite different results in their investigation of British subtitles for the deaf and hard of hearing. There, the mean condensation rate was as high as 43%. In a background study for the present study, I found that the average condensation rate was 31%, which is well in line with the Norwegian fig-ures, and also consistent with more sweeping generalisations of “a third” found in the literature on subtitling (cf. e.g. Gottlieb 1997: 73). It is less con-sistent with the British findings, but this is probably due to the fact that they studied subtitling for the deaf and hard of hearing. Intralingual subtitles often come at a slower pace, because “in general deaf people do not read written text as quickly as hearing people” (de Linde and Kay 1999: 19; cf. section 3.4.1. above), and this results in more condensation. Also, subtitling strate-gies in Britain differ from the ones used by Scandinavians. For instance, in the UK subtitles can appear before the beginning of the utterance they trans-late, and this hardly ever happens in Scandinavian subtitling practice.

These figures suggest that the TT audience misses out on much of what is said in the ST, but the situation is not as bad as it seems. The quantitative condensation rate may average a third, but Gottlieb has shown that there is not a qualitative loss of information of the same amount. Instead, what is condensed is spoken language features, such as repetitions and false starts, so-called “intrasemiotic redundancy” (1997: 101). De Linde (1995) agrees and claims that the reductions in subtitling are not random in this respect, but systematic. Also, fortunately, the viewers can pick up much of what is lost in the condensation of dialogue through other channels, mainly through picto-rial information. Because of this so-called “intersemiotic redundancy” (Gottlieb 1997: 101), the viewers are compensated through other channels, so the total loss of information is not as dire as the quantitative figures sug-gest.

The condensation rate does not vary only according to the speed and complexity of dialogue, or as a result of the smaller intrasemiotic redun-dancy of scripted speech, but also according to national norms, as could be seen from the discussion of de Linde and Kay’s figures.

22

3.3.4 A contract of illusion

In Pedersen (forthcoming b), I introduce the idea of a tacit contract of illu-sion, which I will expand on here, as it is highly pertinent to both the nature of subtitling and to the present study. It is based on the nature of watching subtitled television. Let me first quote Samuel Taylor Coleridge. Whenever we encounter an object of art, like a statue, a painting or a poem, we make use of “the willing suspension of disbelief for the moment that constitutes poetic faith” (1817/1985: 314). Serious spectators tend to enjoy the beauty of Venus de Milo, and not to make the complaint that her armless state makes her unrealistic. The same is true when we watch fiction on television, or go to the cinema. For instance, “[i]n any fiction film the spectator knows very well that the actor and character are not the same, yet at the same time must believe that they are for the film to work” (De Cordova 1986: 132). The same goes for the story, the setting etc. The viewer may well know that M*A*S*H was filmed in the Rocky Mountains, but disregards that knowl-edge, and pretends that what s/he is seeing took place in Korea. Or the viewer can easily realize that Forrest Gump could not possibly have run across America an absurd number of times without even changing his shoes, but disregards this realization to get more pleasure from the film. In other words, the viewer has to suspend his/her disbelief in order to enjoy the film.

I propose that the viewers extend this suspension of disbelief to the subtit-les. There is a tacit agreement, a ‘contract of illusion’ if you will, between the subtitler and the viewers to the effect that the subtitles are the dialogue, that what you read is actually what people say.11 In reality, of course, it is not. For one thing, the verbal material is in another mode, writing instead of speech, and this means that the code is also changed from spoken to written. In practice this means that only a few oral forms are retained in the subtitles. The language is cleaned up; more often than not, hesitations, false starts etc. are excluded (cf. section 3.3.1 above). Furthermore, there is the condensation issue. What you read is on average about a third shorter than what people say (cf. 3.3.3 above). And even more strikingly, in interlingual subtitling, what is written is in another language from what is said. Taking all this into account means that it requires quite a good deal of willing suspension of disbelief for the viewers to sign the contract of illusion. In Scandinavia though, people grow up with subtitling, and the contract of illusion is something they have internalized.

So, the viewers are willing to suspend their disbelief, and as part of their commitment to the contract of illusion, the subtitlers have traditionally tried to make their work as unobtrusive as possible. From this practice comes the focus on a “transparent” (Gottlieb 2001: 51) translation, or as Lindberg puts it at the end of his guidelines for Danish subtitlers: “good subtitles are those 11 This is similar to the “suspension of linguistic disbelief” found in dubbing by Romero Fresco (2006).

23

you never notice” (my translation, italics in the original). Swedish guidelines also encourage this contract of illusion by giving advice such as “[T]hink […]: How would a Swede express herself in the same situation?” (Lan-guageLand 23, emphasis removed). Domesticating advice like this results in a fluent TT of the kind so scorned by Venuti:

Fluency assumes a theory of language as communication that, in practice, manifests itself as stress on immediate intelligibility and an avoidance of polysemy, or indeed any play of the signifier that erodes the coherence of the signified. (1995: 60)

I am not saying that fluency is always the best translation strategy, but the transient (cf. 3.3.1 above) nature of subtitling makes “immediate intelligibil-ity” a necessity. Television viewers do not have the option of going back to make sense of a resistant TT.

Fluency does not necessarily mean domestication, though. The guidelines of the Swedish public service broadcaster, SVT, actively discourage exces-sive domestication, as this might actually threaten the contract of illusion.