Volume 2, No. 2  April 1998

Fotios Karamitroglou was born in Thessaloniki, Greece, on June 30, 1971.
   He entered the School of English at the Aristotle University of Thessaloniki in 1989. He received a First Degree Diploma in English Language and Literature in 1993. After serving in the Greek Army as a Reserve Cadet Officer, he moved to England to earn a Masters Degree at the University of Manchester Institute of Science and Technology (UMIST), sponsored by the Greek State Scholarship Foundation (IKY). He received his M.Sc. in Translation Studies in December 1996 and remained at UMIST for a Ph.D. research in Audiovisual Translation (to be awarded in July 1998).
    Dr. Karamitroglou is a member of the European Association for Studies in Screen Translation (ESIST) and of the British Institute of Translation and Interpreting (ITI).

Dr. Karamitroglou can be reached at:
fotios@ccl.umist.ac.uk

PhD Audiovisual Translation
UMIST, Manchester, UK
European Association for Studies in Screen Translation (ESIST)

PREFACE

Translation Studies is nowadays overwhelmed by a number of attempts to discard prescriptive standpoints and adopt new, descriptive directions. As a consequence, the research effort has been shifted from an investigation of things that should be done to an investigation of things that are being done.
   Audiovisual/Screen Translation is not left out of the trend. In Europe, current research into subtitling is oriented towards an attempt to describe the various subtitling practices around the countries of the continent rather than to dictate what practices should rather be followed. In other words, the attempt nowadays is rather to describe the various subtitling conventions being followed throughout Europe, rather than to impose new ones. However, there are a few undeniable realistic parameters that cannot pass unnoticed: a) the movement towards a United Europe necessitates the adoption of common practices that would enable the participating countries to operate as a unified body, b) new technological developments in mass media and communication (e.g. digital TV) are bound to overcome the limited physical borders of the participating countries, leading to the creation of a pan-European market audience. In such a unified framework of European mass communication, subtitling—as a means of overcoming linguistic barriers between the nations—will come to play a critical role. Large satellite broadcasting companies around the continent have already stressed the need for a unifying code of subtitling practices, a code that would enable them to reach the various individual country audiences through a unique set of subtitling standards that would not violate the already established conventions within the various countries.
   At a first glance, such an attempt or a “desire” looks rather futile. It is impossible to deviate from an already established convention without causing some turbulence amongst the subjects/recipients of the convention. This, however, does not mean that the deviation from the current convention will not be gradually accepted, replacing old practices with new ones and gradually formulating a new ruling convention, provided that the transition is smooth both in quantitative and qualitative aspects. This would mean that a limited number of the new set of suggested practices are initially introduced into a limited number of situations, steadily expanding to cover the whole set of the new suggested practices, as well as all of the applicable situations. Considering the above, I believe that what satellite broadcasting companies demand nowadays is both reasonable and feasible, especially since the already existing number of descriptive studies on human social viewing behaviour, as well as on human physiological eye movement and brain function, are sound enough to provide the basis for such an initial attempt.
   The following paper departs from such descriptive studies but goes on to adopt a clearly prescriptive perspective. Its aim is to provide a unifying formula based on thorough scientific research that could bridge the different subtitling conventions currently operating within the various European countries, to cater for the needs of the individual European viewer and address the European audiovisual audience market as a whole.

Towards a Standardisation of Subtitling Practices in Europe:
Guidelines for Production and Layout of TV Subtitles

1. General aim

The general practice of the production and layout of TV subtitles should be guided by the aim to provide maximum appreciation and comprehension of the target film as a whole by maximising the legibility and readability of the inserted subtitled text.

 
2. Spatial parameter / layout

• Position on the screen: Subtitles should be positioned at the lower part of the screen, so that they cover an area usually occupied by image action which is of lesser importance to the general aesthetic appreciation of the target film. The lowest line of the subtitles should appear at least 1/12 of the total screen height above the bottom of the screen, so that the eye of the viewer does not have to travel a long distance towards the lowest part of the screen to read it. Space should also be provided on the horizontal axis, so that, again, the eye of the viewer does not have to travel a long distance along the sides of the screen in order to read a subtitle line. To this end, image space of at least 1/12 of the total screen width should be provided to the left of the first character and at least 1/12 of the total screen width to the right of the last character, for each subtitle line. Subtitles could be positioned towards the upper part of the screen only in extreme cases where visual material (linguistic or other) of vital importance to the appreciation and the comprehension of the target film is exposed at the pre-determined part of the screen where subtitles would otherwise be inserted.

• Number of lines: A maximum of two lines of subtitles should be presented at a time. This would guarantee that no more than 2/12 of the screen image would be covered by subtitles at a time. In the case of a single-line subtitle, this should occupy the lower of the two lines, rather than the top line in order to minimise interference with the background image action.

• Text positioning: The subtitled text should be presented centered on its allocated line(s). Since most of the image action circulates around the centre of the screen, this would enable the eye of the viewer to travel a shorter distance in order to reach the start of the subtitle. An exception is the case of “double text” (i.e. dialogue turns initiated by dashes and presented simultaneously on a two-line subtitle) which should be aligned to the left side of the screen, following the conventions of printed literature that require dialogue turns introduced by dashes to be left-aligned on the printed page (see also the entry of “Dashes” in section 3 “Punctuation and letter case”).

• Number of characters per line: Each subtitle line should allow around 35 characters in order to be able to accommodate a satisfactory portion of the (translated) spoken text and minimise the need for original text reduction and omissions. An increase in the number of characters, attempting to fit over 40 per subtitle line, reduces the legibility of the subtitles because the font size is also inevitably reduced.

• Typeface and distribution: Typefaces with no serifs are preferable to fonts with serifs, since the visual complexity added to the latter results in a decrease in the legibility of the subtitled text. Typefaces like Helvetica and Arial are qualified. Proportional distribution (like the one used on the current document and on most Word Processors) rather than Monospace distribution (usually used on typewriters) saves the space required to fit the desired 35 characters into a subtitle line.

• Font colour and background: Type characters should be coloured pale white (not “snow-bright” white) because a too flashy pigment would render them tiring to the viewers’ eye. They should also be presented against a grey, see-through “ghost box” rather than in a contoured format (surrounded by a shadowed edge) since it has been proven that it easier for the eye to read against a fixed rather than a varying/moving background. In addition, the colour of the “ghost box” (grey) is both neutral to the eye and gives the impression that it does not entirely block the background image.

3. Temporal parameter / duration

• Duration of a full two-line subtitle (maximum duration): The reading speed of the “average” viewers (aged between 14-65, from an upper-middle socio-educational class) for a text of average complexity (a combination of formal and informal language) has been proven to range between 150-180 words per minute, i.e. between 2 1/2-3 words per second. This means that a full two line subtitle containing 14-16 words should remain on the screen for a maximum time of something less than 5 1/2 seconds. However, we would actually have to expand the estimate to around 6 seconds because one should also add about 1/4-1/2 of a second that the brain needs to start processing the subtitle it has traced. It should be noted that equal to the importance of retaining a full two-line subtitle for at least 6 seconds to secure ample reading time, is the importance of keeping the same subtitle not more than 6 seconds because this would cause automatic re-reading of the subtitle, especially by fast readers.
  Note: The average reading speed of children (aged 6-14) has been found to be around 90-120 words per minute. For the subtitling of children’s programmes, then, calculations regarding the duration of the subtitles on screen should be estimated accordingly.

• Duration of a full single-line subtitle (maximum duration) : Although pure mathematics would lead us to the conclusion that for a full single-line subtitle of 7-8 words the necessary maximum duration time would be around 3 seconds, it is actually 3 1/2 seconds. This happens because for the two-line subtitle it is the visual bulk of the text that signals an acceleration of the reading speed. With the single-line subtitle this mechanism is not triggered. Once again, equal to the importance of keeping a full single-line subtitle for at least 3 1/2 seconds to secure ample reading time, is the importance of retaining the same subtitle for not more than 3 1/2 seconds because this would cause automatic re-reading of the subtitle, especially by fast readers. For similar reasons of automatic re-reading, in both cases of single-line and two-line subtitles, the duration time could be calculated and shortened down to the maximum of the reading time (3 subtitled words per second or 1/3 of a second per subtitled word), if the text is lexically and syntactically easy to process and if the fast pace of the film action dictates such a reduction in the duration of the subtitles.

• Duration of a single-word subtitle (minimum duration): The minimum duration of a single-word subtitle is at least 1 1/2 seconds, however simple the word is. Less time would render the subtitle as a mere flash on the screen, irritating the viewers’ eye. Again, it should be noted that equal to the importance of retaining a single-word subtitle for at least 1 1/2 seconds to secure ample reading time is the importance of keeping the same subtitle for not more than 1 1/2 seconds because this would cause automatic re-reading of the subtitle, especially by fast readers.

• Leading-in time: Subtitles should not be inserted simultaneously with the initiation of the utterance but 1/4 of a second later, since tests have indicated that the brain needs 1/4 of a second to process the advent of spoken linguistic material and guide the eye towards the bottom of the screen anticipating the subtitle. A simultaneously presented subtitle is premature, surprises the eye with its flash and confuses the brain for about 1/2 a second, while its attention oscillates between the inserted subtitled text and the spoken linguistic material, not realising where it should focus.

• Lagging-out time: Subtitles should not be left on the image for more than two seconds after the end of the utterance, even if no other utterance is initiated in these two seconds. This is because subtitles are supposed to transfer the spoken text as faithfully as possible, in terms of both content and time of presentation and a longer lagging-out time would generate feelings of distrust toward the (quality of the) subtitles, since the viewers would start reflecting that what they have read might not have actually corresponded to what had been said, at the time it had been said.

• Between two consecutive subtitles: About 1/4 of a second needs to be inserted between two consecutive subtitles in order to avoid the effect of subtitles’ “overlay.” This time break is necessary to signal to the brain the disappearance of one subtitle as a piece of linguistic information, and the appearance of another. If no such gap is maintained, the viewers’ eye cannot perceive the change of the new subtitled text, especially if it is of the same length as the antecedent one.

• “Overlay,” “add-ons” and “cumulative text”: All these terms are synonymous for the technique of presenting a “dynamic text,” i.e. a dialogue or a briefly paused monologue, with its first part appearing first on the top line of the subtitle and the second part appearing consecutively on the bottom line of the subtitle while the first line still remains on screen. This technique is ideal for avoiding “spilling the beans,” managing to reveal “surprise” information at the time of the actual utterance. Since it is a wild-card mechanism, it should be used cautiously.

• Camera takes/cuts: Subtitles should respect camera takes/cuts that signify a thematic change in the film product and, for this reason, they should disappear before the cuts. Different camera shots, fades and pans that do not indicate a major thematic change (e.g. a change from a long shot to a close-up and back) should not affect the duration of the subtitles at all as they do not signify a thematic change.

• “Sequence dots” (or “ending triple dots”) {...}: Three dots should be used right after the last character of a subtitle (no space character inserted), when the subtitled sentence is not finished on one subtitle and has to continue over the consecutive subtitle. The three "sequence dots" indicate that the subtitled sentence is incomplete, so that the eye and the brain of the viewers can expect the appearance of a new flash to follow. The total absence of any kind of punctuation mark after the last character of the subtitle, as an alternative means of indicating the continuation of the subtitled sentence over the consecutive subtitle, does not provide such an obvious signal and, thus, the brain takes more time to process the new flash which appears less expectedly. Because of their particular function as signifiers of sentence incompleteness, the use of “sequence dots” to simply indicate ongoing thoughts or an unfinished utterance by the speaker should be considerably restricted.

• “Linking dots” (or “starting triple dots”) {...}: Three dots should be used right before the first character of a subtitle (no space character inserted, the first character non-capitalised), when this subtitle carries the follow-up text of the previous uncompleted sentence. The tracing of the three “linking dots” signals the arrival of the expected new flash of subtitle, something anticipated because of the presence of “sequence dots” in the previous subtitle. The absence of any punctuation mark as an alternative means of indicating the arrival of the remaining part of an incomplete subtitled sentence does not provide such an obvious signal and as a result the brain takes more time to process the new subtitle flash as related to the previous subtitle. Because of their particular function as signifiers of sentence continuation, “linking dots” should always be used in conjunction with “sequence dots.”

• Full stops {.}: The full stop, or period, should be used right after the last character of a subtitle (no space character inserted) to indicate the end of the subtitled sentence. This signals to the eye that it can go back to the image since there is no consecutive subtitle to anticipate. The absence of “sequence dots” as an alternative means of indicating the end of a subtitled sentence does not provide such an obvious signal and as a result the brain takes more time to process the fact that the subtitled sentence has actually been completed.

• Dashes and hyphens {-}: Dashes are used before the first character of each of the lines of a two-line subtitle (with a space character inserted each time) to indicate the exchange of speakers’ utterances, namely a dialogue, presented either in a single flash as “static double text,” or with the second speaker’s exchange as an “overlay” to the first subtitle line, i.e. as “dynamic double text.” When dashes are used to link words as hyphens no space characters should be inserted between the linked words

• Question marks {?} and exclamation points {!}: Question marks and exclamation points should be used to indicate a question or emphasis respectively, just like in printed materials, positioned right after the last character of a subtitle (no space character inserted).
  Note: For questions in Spanish, a question mark should also be inserted right before the first character (no space character inserted).

• Parentheses {( )} and brackets {[ ]}: Parentheses and brackets should be used to embrace comments which are explanatory to the preceding phrase. As the duration time for each subtitle is considerably limited and the convention of parentheses or brackets is not extremely widespread in printed materials either, they function as wild cards and, therefore, they should be used cautiously.

• Single quotation marks {‘ ’}: Single quotation marks should be used just like in printed materials, in order to embrace alleged information. For reasons similar to the use of parentheses and brackets, single quotation marks should be used cautiously.

• Double quotation marks {“ ”} : Double quotation should be used just like in printed materials, in order to embrace quoted information. For reasons similar to the use of parentheses and brackets, double quotation marks should be used cautiously.

• Commas {,}, colons {:} and semicolons {;}: Commas, colons and semicolons should be used just like in printed materials, in order to suggest a short pause in the reading pace. Unlike full stops, sequence dots, exclamation points and question marks which could all be used to close a subtitled sentence, no subtitle flash should end in a comma, a colon or a semicolon because the inevitable pause in the reading pace, as a result of the time break between the two subtitles and the necessary time for the brain to process the new subtitle, would be disproportionately long in relation to the expected short pause. Again, for reasons similar to the use of parentheses and brackets commas, colons and semicolons should be used cautiously.

• Italics: Italics on the subtitled text should be used to indicate an off-screen source of the spoken text, (e.g. when there is a voice of someone contemplating something, speaking over the phone from the other end, or narrating something). They should also be used when retaining foreign-language words in their original foreign-language version (e.g. “He’s got a certain je ne sais quoi.”).

• Quotation marks {“”} embracing text in italics: Quotation marks embracing text in italics should be used to indicate a public broadcast, i.e. spoken text coming from an off-screen source and addressed to a number of people (e.g. through a TV, a radio, or a loudspeaker). They should also be used when transferring song lyrics.

• Upper- and lower-case letters: Upper- and lower-case letters should be used just like in printed materials, as if the subtitle was to appear on paper. Subtitles typed only in upper-case letters should be used when transferring a display or a caption (i.e. a written sign that appears on the screen).

• Boldface and underline: Boldface and underline typing conventions are not permitted in subtitling.

5. Target text editing

• From a single-line to a two-line subtitle: It is better to segment a long single-line subtitle into a two-line subtitle, distributing the words on each line. This is because the eye and the brain of the viewers render a two-line subtitle as more bulky and, as a result, accelerate the reading process.

• Segmentation at the highest nodes: Subtitled text should appear segmented at the highest syntactic nodes possible. This means that each subtitle flash should ideally contain one complete sentence. In cases where the sentence cannot fit in a single-line subtitle and has to continue over a second line or even over a new subtitle flash, the segmentation on each of the lines should be arranged to coincide with the highest syntactic node possible. For example, before we segment the phrase:
  
  “The destruction of the city was inevitable.” (44 characters),

  we first have to think of its syntactic tree as follows:

  

  A segmentation on the fifth node (N5) would create the two-line subtitle

  “The destruction of the
  city was inevitable.”

  A segmentation on the second node (N2) would create the two-line subtitle

  “The destruction of the city
  was inevitable.”

  Out of the two segmentations, it is the second that flows as more readable. This occurs because the higher the node, the greater the grouping of the semantic load and the more complete the piece of information presented to the brain. When we segment a sentence, we force the brain to pause its linguistic processing for a while, until the eyes trace the next piece of linguistic information. In cases where segmentation is inevitable, therefore, we should try to force this pause on the brain at a point where the semantic load has already managed to convey a satisfactorily complete piece of information.

• Segmentation and line length: The upper line and the lower line of a two-line subtitle should be proportionally as equal in length as possible, since the viewers’ eye is more accustomed to reading text in a rectangular rather than a triangular format. This happens because the conventional text format of printed material is rectangular (in columns or pages). Taken into account the previous entry on “segmentation at the highest nodes,” this means that the segmentation of subtitled text should be a compromise between syntax and geometry. However, if we had to sacrifice the one for the sake of the other, we should prefer to sacrifice geometry.

• Spoken utterances and subtitled sentences: Each spoken utterance should ideally correspond to a subtitled sentence. The reason is that viewers expect a correct and faithful representation of the original text and one of the basic means to check this is by noticing if the number of the spoken utterances coincides with the number of the subtitled sentences. In other words, viewers expect to see the end of a subtitled sentence soon after they realise that the speaker has finished his/her utterance and before a new one begins. In this respect, merging or bridging two or more utterances into one subtitled sentence should be avoided as much as possible, unless spatio-temporal constraints strictly dictate it.

• More than one sentence on the same subtitle: No more than two sentences are allowed on the same subtitle. Following the principle of “segmentation at the highest nodes,”; they should occupy one line each, no matter whether they correspond to utterances produced by the same speaker (monologue) or by different speakers (dialogue). If they correspond to a monologue, they should be centralised like normal subtitled text. If they correspond to a dialogue, they should be left-aligned and preceded by dashes (“double text”).

• Omitting linguistic items of the original: A decision as to which pieces of information to omit or to include should depend on the relative contribution of these pieces of information to the comprehension and appreciation of the target film as a whole. The subtitler should not attempt to transfer everything, even when this is spatio-temporally feasible. The subtitler should attempt to keep a fine balance between retaining a maximum of the original text (essential for the comprehension of the linguistic part of the target film), and allowing ample time for the eye to process the rest of the non-linguistic aural and visual elements (essential for the appreciation of the aesthetic part of the target film). Categories of linguistic items that could be omitted are as follows:

  • Padding expressions(e.g. “you know,” “well,” “as I say” etc): These expressions are most frequently empty of semantic load and their presence is mostly functional, padding-in speech in order to maintain the desired speech flow.

  • Tautological cumulative adjectives/adverbs (e.g. “great big,” “super extra,” “teeny weeny” etc): The first part of these double adjectival/adverbial combinations has an emphatic role which can be incorporated in a single-word equivalent (e.g. “huge,” “extremely,” “tiny”).

  • Responsive expressions (e.g. “yes,” “no,” “ok,” “please,” “thanks,” “thank you,” “sorry”). The afore-listed expressions have been found to be recognised and comprehended by the majority of the European people, when clearly uttered, and could therefore be omitted from the subtitle. It should be noted, however, that when they are not clearly uttered or when they are presented in a slang, informal or colloquial version (e.g. “yup,” “nup,” “okey-dokey,” “tha” etc) they are not recognisable or comprehensible and should, therefore, be subtitled.

• Retaining linguistic items of the original: Linguistic items of the original that can be easily recognised and comprehended by the viewers should not only be retained if they appear in a context of unrecognisable items which blurs the meaning of the total utterance, but they should also be translated word-for-word. These items are most frequently proper nouns (e.g. geographical names like “Los Angeles,” “Africa” etc.) or items that the target language has directly borrowed from or lent to the source language or happened to have in common after they both borrowed it from a third language (e.g. the items “mathematics,” “mathématique” and “mathimatika” shared by English, French and Greek respectively). Investigations in the psychology of viewing indicate that when such linguistic items are recognised by the viewers, the exact, literal, translationally equivalent items are expected to appear in the subtitles as well. This occurs because of the constant presence of an inherently operating checking mechanism in the brain of the viewers which raises the suspicions that the translation of the original text is not “properly” or “correctly” rendered in the subtitles, every time word-for-word translations for such items are not spotted.

• Altering syntactic structures: Simpler syntactic structures (canonical forms) tend to be both shorter and easier to understand than complex syntactic structures and should, therefore, be preferred, provided that a fine balance is achieved between a) semantic aspects (maintaining the semantic load of the original), b) pragmatic aspects (maintaining the function of the original), and c) stylistics (maintaining the stylistics features of the original). Categories of complex syntactic structures could be replaced by simplified ones as follows:
  • Active for passive constructions: E.g. “It is believed by many people.” (30 characters) => “Many people believe.” (20 characters).

  • Positive for negative expressions: E.g. “We went to a place we hadn’t been before.” (41 characters) => “We went to a new place.” (23 characters).

  • Temporal Prepositional Phrases for temporal subordinate clauses: E.g. “I’ll study when I finish watching this movie.” (46 characters) => “I’ll study after this movie.” (28 characters).

  • Modified nouns for the referring relative clauses: E.g. “What I’d like is a cup of coffee.” (33 characters) => “I’d like a cup of coffee.” (25 characters).

  • Gapping for double verb insertion: E.g. “John would like to work in Germany and Bill would like to work in France.” (73 characters) => “John would like to work in Germany and Bill in France.”; (54 characters).

  • Straightforward question sentences for indicative pragmatic requests: E.g. “I would like to know if you are coming.” (39 characters) => “Are you coming?” (15 characters).

  • Straightforward imperative sentences for indicative pragmatic requests: E.g. “I would like you to give me my keys back.” (41 characters) => “Give me my keys back.” (21 characters).
    
    In certain cases, however, it is longer structures that have to be preferred because they facilitate mental processing:

  • Coherent phrase grouping for syntactical scrambling: E.g. “That a man should arrive with long hair did not surprise me.” (60 characters) => “It did not surprise me that a man with long hair should arrive.” (63 characters).

• Acronyms, apostrophes, numerals and symbols: Acronyms, apostrophes and symbols can save precious character space by abbreviating meaning signs. However, they should be used with caution and only if they are immediately recognisable and comprehensible. For example:

  • Acronyms: Use acronyms like “NATO” and “USA” but avoid acronyms like “PM” (Prime Minister) or “DC” (Detective Constable).

  • Apostrophes: Use apostrophes for abbreviations of auxiliaries like “I’d like” and “You can’t” but avoid abbreviations like “Mid’bro” (Middlesborough).

  • Numerals: For numerals, the conventions of printed materials should be followed, i.e. they should be used to indicate numbers over twelve “He is only 25” but not other numeric expressions like “1000s of times” or “the 2 of us.”

  • Symbols: Use symbols commonly used and immediately recognised on printed materials like “%” and avoid less common symbols like “&” or “@.”

• Rendering dialects: If a dialect of the target language (regional or social) is chosen to be used on the subtitled text, it should not be rendered as a phonetic or syntactic transcription of the spoken form. Only dialects that have already appeared in a written form in printed materials are allowed to be used in subtitles as well. For example, archaic or biblical forms like “thee” for “you” are allowed but sociolect forms like “whadda ya doin?” are not allowed because they are not immediately recognisable and comprehensible by the viewers’ eye.

• Taboo words: Taboo words should not be censored unless their frequent repetition dictates their reduction for reasons of text economy.

• Culture-specific linguistic elements: There is no standard guideline for the transfer of culture-specific linguistic elements. There are five possible alternatives for such a transfer: a) cultural transfer, b) transposition, c) transposition with explanation, d) neutralisation (plain explanation), e) omission. The culture specific element “10 Downing Street” (the British Prime Minister’s Residence), for example, in the expression “They were following orders from 10 Downing Street” could be transferred as follows:

  • Cultural Transfer: “They were following orders from ________,” filling the gap with the respective name of the Prime Minister’s Residence (e.g. Matignon for France, Megaro Maximou for Greece etc.)

  • Transposition: “They were following orders from 10 Downing Street”

  • Transposition with explanation: “They were following orders from 10 Downing Street, the Prime Minister’s House”

  • Neutralisation: “They were following orders from the Prime Minister”

  • Omission: “They were following orders”