Abstract

The purpose of this paper is to consider the theoretical issues relating to the viewership of subtitling. I begin with some considerations regarding subtitling audiences, then, following Delabastita's model for film semiotics, I discuss the cognitive demands on viewers of subtitled material. In this regard, I compare and contrast reading of translated literary texts and the viewing of subtitled film. I also discuss the status of image in the subtitling process.

Finally, building on Halliday's notion of situation and the theory of film developed by Balazs, I suggest a viewer-oriented, image-bound approach to subtitling. This includes a critical review of current approaches, especially as regards emphasis on reading speed and the concept of loss in the reading of subtitles.

ubtitling should not be considered a second-rate, debased form of translation, as is increasingly recognized. Universities have begun to include it as a subject for study in their curricula, and every year it is the focus of several academic conferences as this relatively new but very rich field of study attracts more and more researchers.

Subtitlers need to be visually literate and fully familiar with the language of cinema.

From the industry's point of view, subtitling has become very much a part of cinema, with the development of readily-available, cost-effective services to reach international audiences and markets. This broad reach is one reason for the considerable interest in the field, in addition to its special place for translation studies, since subtitling involves cultural adaptation and language transfers for a single film to reach people in several different countries representing a variety of norms.

In this paper, I will be considering the often neglected issues of viewership in subtitling, particularly for film. I begin with a review of the extent of subtitling audiences around the world, moving on to consider the cognitive efforts involved in the viewing of subtitled material. This includes a critical review of current theoretical analysis.

Subtitling viewership—varied and able

Subtitled material attracts more and more viewers, especially where cinema is concerned, in many different countries and language groups. There are a variety of explanations for this enthusiasm.

Film lovers want to watch their favorite films in the original. This may be matter of hearing the real voices of their favorite actors, which they can do with subtitles. Another important incentive is the desire to learn a foreign language, as is illustrated by viewing patterns in my home country, Iran.

In Iran, the national television still generally dubs fiction, reserving subtitles for documentaries and news. However young people and others are showing increasing favor for subtitled films on DVD or video tape. These are available for sale and hire at video clubs across the country, many of them with a label in Persian saying "For a better command of English." Viewers thus see subtitled films as a means to progress in a foreign language, which in most cases is English, as well as to enjoy the film. No research has yet been conducted to show whether or not they achieve their aim.

Despite this widespread enthusiasm and the potential it represents, it is often assumed in Translation Studies circles that subtitling audiences have a lower cultural standing than, for example, the readers of literary works in translations. There is long-standing prejudice in favor of this type of translation. Deeper consideration may well show that this is ill-founded and even suggest that viewing subtitled material is much more demanding than reading literary texts in translation.

Written material is monosemiotic, requiring readers to focus on the visual signs to interpret the verbal content signified and explore meaning. They are subject to no time constraints since they can read the text as many times as they wish, concentrate on sections they find difficult, stop and start again at will.

The tasks are not so easy for viewers of subtitled films. Time constraints are very much to the fore, since filmic material is presented in a continuous flow over which they have no control, requiring constant attention. You can only watch a film frame by frame, and this uninterrupted, immediate exposure (Gottlieb 1991: 162) entails more varied cognitive activities in decoding segments of the image and connecting these segments to the overall discourse woven into the audiovisual structure of the film.

The constraints are all the greater, as viewers simultaneously have to read one or two lines of text at the bottom of the screen in the allotted time, which is generally shorter than for the original dialog.

Subtitled films thus require a greater effort to harmonize a variety of cognitive activities and grasp the underlying idea.

We will now take a closer look at the activities involved in watching subtitled films.

Dual demands

When Herman Weinberg put the first-ever subtitles (Nornes, 1999: online) on the lower part of a film in Hollywood, he went into the theatre where his subtitled film was being screened to watch viewers' faces and see how they reacted to this medium newly-created to replace dubbing, which was then prevalent. In his words, he had "wondered if they were going to drop their heads slightly to read the titles at the bottom of the screen and then raise them again after they read the titles" (ibid) and was surprised to see that the audience simply dropped their eyes to read the subtitles.

Weinberg's prime concern, at that time, was to spare spectators excessive physical efforts (head movements) in reading the subtitles, and in this he succeeded. However, he never asked whether his viewers had to skip a portion of the movie in order to keep up with the subtitles. To seek an answer to that question, we need to look at the mental and physical processes involved in watching a film.

I list these processes separately below, noting that the order of these activities depends on the choice of viewers, as well as the conditions of presentation (montage).

1. Reading the subtitles
2. Decoding the subtitles: This involves decoding the syntactic and semantic content of the subtitles to arrive at a pragmatically acceptable interpretation. Viewers "supplement the semiotic content of the subtitles with information from other audiovisual channels—notably the image, and prosodic features in the dialogue" (Gottlieb, 1993: 273).
3. Watching the image flow
4. Deciphering the visual information: Cinema is a specific language, a language of its own, and the composition of images in cinema convey much more than what we see in real life in the street. Montage makes images speak, and the way the film is constructed adds meaning to what happens in the film (narrative, action, etc.)
5. Connecting each segment of the image flow to the underlying story
6. Listening to (or just hearing) the sound (dialogue, effects, music)
7. Guessing what is about to happen
8. Remembering what has already happened to make fresh deductions during following sequences.

Watching a film in the original entails all of these activities, except for the first two simultaneous tasks, which directly concern the subtitles. These intrude on the watching process, since reading and decoding of subtitles partly disrupt image reading. This point deserves more discussion.

Image would appear to be the component that remains intact in the subtitled product, but closer consideration shows that this is not the case, since the intruding rows of subtitles in the lower part of the screen will affect viewers' perception, one way or another. Subtitles have an impact on the whole process of image reading. A clearer account requires attention to the polysemiotic dimension of film.

Following Delabastita (1989, cited by Larsen, 1993: 213), we can say that film viewers deal simultaneously with four different types of signs:

• Verbal acoustic signs (dialogue)
• Non-verbal acoustic signs (music, sound effects)
• Verbal visual signs (written signs in the image)
• Non-verbal visual signs (what is otherwise seen in the image)

Since the first two types of sign are dependent on hearing, subtitles thus have little effect on the interpretation process. In contrast, the second two types of sign depend on sight, and subtitles (verbal visual signs) interfere with the perception of the non-verbal visual signs, making dual demands on viewers who have to divide visual attention between subtitles and image. It is not possible to keep full track of both.

Subtitles thus lead to imperfect perception of the image that is the core component of the film. A further complication is that the dialogue of one character in films is often accompanied by a different image, such as a shot of the face of another character or a landscape to provide viewers with supplementary information. If viewers neglect this shot to focus on the subtitles, they miss part of the cinematic story.

The difficulty can be even greater in scenes with a narrative commentary (2), which is common in introductory passages or for transitions from one sequence of a film to another. In such cases, the lengthy subtitles required may clog up the screen and viewers have to choose between reading and watching.

There are thus good reasons to consider that subtitles unavoidably disrupt or even halt the process of image reading.

However, this straightforward point is given little attention in theoretical accounts of subtitling. This may be partly due to the fact that subtitling was originally mainly associated with TV programs (see Gambier, 1993: 275), where disruption of the image is not as important as in cinema and viewers have grown to accept partial or complete loss of image as the price to pay for receiving the information crammed into subtitles.

One result of this flawed approach to subtitling is that the reading speed of viewers has been taken as a yardstick for the maximum number of words in writing subtitles. In the following section, I will discuss some difficulties which current theories of subtitling face in terms of their accounts of viewers' reading speed.

Reading speed: tasks for viewers

A rule commonly applied in Europe is that subtitles should consists of at most two lines of 35 characters each (70 characters in all), with viewers required to read the text within a limited time directly related to the duration of dialogues. The assumption is thus that the reading speed of viewers will determine whether or not they keep up with the flow of subtitles.

So far, few researchers have tried to come up with a reliable measure of the reading speed of typical viewers of subtitled programs. Gottlieb cites (1993: 164) a Swedish finding, dating back to early seventies, that average television viewers needed 5-6 seconds to read a two-liner of some 60-70 characters. He goes on to quote a mid-eighties Belgian study that suggested faster reading. These findings call for some caution, since a number of variables may lead studies of this kind to different conclusions.

In the first place, the media for which the subtitles are written can have a significant impact on viewers' reading speed. Logically, TV subtitles can be read faster than film subtitles, reflecting different levels of visual density.

In a TV interview, the visual components gradually lose informational relevance as we become acquainted with the participants (interviewer and the interviewee) and the location. We no longer need to pay much attention to the image since we already have it in our mind. Visual attention can thus shift away from image reading to focus on reading the subtitles. In film, in contrast, each shot provides new information, and attention to subtitles disrupts or halts the process of deciphering this information, which can thus be neglected.

Clearly, too, reading speed may vary according to the linguistic complexity of the subtitles in terms of semantics and syntax.

In addition to these factors, we must remember that watching a subtitled film is not a speed-reading competition. Movie-viewers go to theaters or sit in front of TVs to watch, not to read. Under experimental conditions, they may read one or two lines of subtitles very fast, but miss the film itself. In any event, it is clear that viewers in general do have some trouble watching a subtitled film. Their prime interest is in seeing the story in action, even when they already know what happens. However good subtitles may be, they do not cater to that desire, a fact that has sometimes been overlooked.

Gottlieb reports (1991: 167) an analysis of the Danish subtitles for Mel Brooks' film Young Frankenstein showing that only 16% of the original verbal segments had suffered a loss of information in the process of translation. But did the dialogue alone suffer a loss? Did Danish viewers lose 16% of the image too? Some revisions to the concept of loss in audiovisual translation, particularly subtitling, would appear timely.

To elaborate on this, I might recall the popular concept of equivalence of effect in subtitling, this being the principle that subtitles should be designed to ensure the same impact on target-language viewers as on source-language viewers. In practice, the concept becomes problematic if we consider the image component, since subtitles prevent viewers getting all the information from the images. It is thus not possible to reproduce the original effect.

Subtitling practitioners are in the habit of making viewers chase after their text. The maximum number of characters a viewer can take within a defined time has been set as the standard for putting lines of subtitles onto the screen. The assumption appears to be that the more characters the viewers manage to read, the more characters the subtitlers are allowed to use, and it is likely that new studies will reveal higher rates of reading speed on the part of the viewers. But this should not be taken as a license for lengthier subtitles, since, as I have already said, films are made to be watched, not read.

Instead of taking the book or the screenplay away for a comfortable read, moviegoers brave queues and crowds to watch the live version. In this context, subtitles should be designed to give viewers the gist of dialogues and let them get through the reading quickly to turn their attention to the image.

Clearly, this rises the question of how the desired result is to be achieved and in the following section, I suggest a viewer-oriented, image-bound approach to that effect. Before moving on to that, it will be useful to consider the role of dialogue in cinema more carefully.

Connectivity: the function of subtitles

The basic unit of a film is a frame, a series of frames is a shot, several shots make a scene and scenes make up the sequences that together constitute the film. The dialogue joins these components up like beads on a single thread.

In cinema, dialogues steer the image story through the flow of shots, each delivering a push to keep it on track.

Obviously, though, they do not function in a vacuum, since they are backed up by other filmic components: soundtrack, musical score, effects, the characters' tone of voice, facial expressions and body language, camera movement, distance and angles, and montage (cuts, fade-in, dissolve, etc.). Following Halliday (1976: 21), I will call this set of inter-connected elements a situation (3) in which each part makes its own contribution to the filmic discourse. Only the lexis of the dialogue is changed in the process of subtitling, while the other components of this cinematic situation remain almost untouched in the end product.

Béla Balazs' argument about the value position of dialogue in the overall structure of cinema will be helpful for our discussion. Under the heading "audible gesture speech," he describes (1970: 227) speech as an "expressive movement, which can show the most delicate shades of emotion even if we cannot understand what is being said." He downplays words as a second-class component of human speech, saying:

"all the more important is the tone in which they are said: the cadence, the emphasis, the timbre, the husky resonance, which are not intentional, not conscious. Vibrations of the voice may mean many things that are not included in the meaning of the word itself—it is a sort of accompaniment to the words, a verbal gesture." (ibid).

Balazs argues that "loquacity is often merely a vehicle for such extra-rational expression" (ibid) and goes on to describe what he calls the "micro-mimicry of the close-up," which turns speech into "an audible play of features" (ibid).

On this, two points are worth stressing: 1) there are elements of speech other than words that carry cinematic meaning, and 2) in the transition from the original dialogue to subtitles, only the words themselves are at stake, not those other elements.

Subtitles track dialogues minus their extra-linguistic features (gender, age, social class, etc), paralinguistic features (facial expression of characters, head/eye movements, gesture, etc.), pitch patterns, emotional tone, etc. All these features are intact in the end product (i.e. the subtitled material), which means that subtitles stand for only a small, albeit indispensable, fragment of the original dialogue. In view of this, any combination of words that gives the same direction to the image-story as the original dialogue may be considered an acceptable equivalent of the SL version. To put it another way, any expression that establishes the same connection between shots as the original dialogue shall be deemed as fit. An example will help to show what is meant.

In The Mummy directed by Stephen Sommers, the Egyptian guide of the British archeology team rushes to pick blue insect-like objects off a wall in the pyramid, yelling "Let's get us some treasure!" This is immediately followed by a straight cut to another location, where the leader of the American team warns his teammate away from a mysterious door, shouting "Be careful!" Both the abrupt change of the location and the staccato speech are designed to startle viewers and create dramatic tension, causing them to hold their breath.

Viewers of the original language version hear the two utterances with almost no interval, something subtitling cannot replicate. Clearly, though, the Egyptian guide's utterance has to be rendered very succinctly to allow viewers to keep track. In the circumstances, a native speaker of Persian would utter an exclamation such as "بزن بريم" (a mere 7 characters), literally "Let's hit and go!" Used as a subtitle, this would convey to Persian viewers the guide's glee over his find, giving the same thrust to the image-story, and would be short enough not to interfere with the dramatic tension produced by the straight cut.

This is an example of what I call a viewer-oriented approach, where subtitles stand in for the bare words of the original dialogue as briefly as possible to synchronize with the dialogue and blend into the flow of the image-story. In this way, viewers can grasp the original dialogue and put the rest together for themselves.

This of course recalls Seleskovitch's Theory of Sense (1978), which describes the interpreters' task in terms deverbalization or discarding words to present the underlying thought of the SL speakers in any appropriate form. The only real difference is that interpretation is non-synchronous, since it entails a lag on the speaker. Another reference that comes to mind is Halliday's Theory of Language Functions (1976) with its emphasis on the functions of utterances in situation.

Some may query whether subtitling done in this way may properly be counted as translation, but there should be no doubt about the answer, since the goal is to help TL audience access the SL material, as it is with any translation. However, unlike other forms of interlingual transfer, effective subtitling requires recognition of the constraints of the media and an approach clearly centered on the audience. This means that the final product must be designed with two fundamental considerations in mind:

1) The demands and capacities of the viewers sitting down to enjoy the film in the theater or in front of their TVs. Subtitling serves these viewers as a tool to make the image-story comprehensible and thus help them read the image. That is, the subtitles must help the viewers come as near as possible to the experience the source-language viewers have of the image-story. The subtitles are not there for their own sake, and image is certainly not there to illustrate them.

2) The specificities of the audio-visual media and, most importantly, their transitory nature (Larsen, 1993: 213), which require subtitles to adapt translation strategies and reconstitute the original audiovisual situation.

On this assessment, traditional methods and techniques of written translation are unsuited to the job and other ways must be found to achieve the desired ends. Professionals need to provide viewers with the shortest possible subtitles and spare them unnecessary shades of meaning that hinder the process of image reading. To do this, they need to identify the precise role of each segment of dialogue in moving the image-story forward and try to achieve the same end with the fewest possible words. This is no easy task, and calls for a variety of skills.

Subtitlers need to be visually literate and fully familiar with the language of cinema as well as completely fluent in the target language. Beyond that, they need talent and creative flair to encapsulate the function of each segment of dialogue in the fewest possible words. And understanding that function in the particular situation concerned often calls for repeated, attentive viewing.

In conclusion, I insist that good subtitling means first and foremost brevity to give image, and the viewer, pride of place (4). Reid, as an experienced subtitler (1987, as cited by Gottlieb, 1993: 273), defines this as "deciding what is padding and what is vital information." I would suggest that, more often than not, the image carries that vital information—and that is where we should look for it.

Notes

1 - Van de Poel reports (2001: 259-275) on a case of foreign-language acquisition by children watching television programs and discusses the benefits of subtitled programs in that context. This case of second-language learning by adults would be an interesting subject for research.

2 - An off-screen voice added to a shot to describe it or give other information related to different sequences.

3 - Halliday: "the term situation, meaning the 'context of situation' in which a text is embedded, refers to all those extra-linguistic factors which have some bearing on the text itself" (1976: 21).

4 - Pleas for brevity are increasingly common within the field. Kovacic (1993: 246) appeals to the Relevance Theory to argue for more reduction, while Gottlieb (1993: 273) calls on the "scholars of the world" to be brief.

Works cited

- Balázs, Béla (1970) 'Theory of the Film: Character and Growth of a New Art,' Translated by Edith Bone, New York: Dover.

- Gambier, Yves (1993) 'Audio-visual Communication: Typological Detour,' in Dollerup, Cay and Annette Lindegaard [eds] Teaching Translation and Interpretation 2: Insights, Aims, Visions, Amsterdam, Philadelphia, John Benjamin: 275-283.

- Gottlieb, Henrik (1991) 'Subtitling - A New University Discipline,' in Dollerup, Cay and Anne Loddegaard [eds] Teaching Translation and Interpretation: Training, Talent and Experience, Amsterdam, Philadelphia, John Benjamin: 161-170.

- Gottlieb, Henrik (1993) 'Subtitling: People Translating People,' in Dollerup, Cay and Annette Lindegaard [eds] Teaching Translation and Interpretation 2: Insights, Aims, Visions, Amsterdam, Philadelphia, John Benjamin: 261-274.

- Halliday, M.A.K. and Rugaiya Hassan (1976) 'Cohesion in English,' New York: Longman.

- Kovacic, Irena (1993) 'Relevance as a Factor in Subtitling Reduction,' in Dollerup, Cay and Annette Lindegaard [eds] Teaching Translation and Interpretation 2: Insights, Aims, Visions, Amsterdam/ Philadelphia, John Benjamin: 245-251.

- Nedergaard-Larsen, Birgit (1993) 'Culture-Bound Problems in Subtitling,' Perspective: Studies in Translation, 2: 207-241.

- Nornes, Abe Mark (1999) 'For an Abusive Subtitling (Subtitles of Motion Pictures)' [online], available from: http://articles.findarticles.com/p/articles/mi_m1070/is_3_52/ai_54731368 [accessed: June 21, 2004].

- Seleskovitch, D. (1978) 'Interpreting for International Conferences,' Washington, D.C.: Pen and Booth

- Van de Poel, Marijke (2001) 'Incidental Foreign- Language Acquisition by Children Watching Subtitled Television Programs,' Gambier, Yves and Henrik Gottlieb [eds] (Multi) Media Translation: Concepts, Practice, and Research, Amsterdam/ Philadelphia, John Benjamin: 259-273