The Vanishing Original

This article addresses the fundamental changes that are taking place today in many practical translation contexts and processes, namely the changed relationship between the so-called ‘original’ and its translations. This relationship has always been more problematic than translation theory has been willing to accept, but today there are so many instances in which there is no ‘original’ in the traditional sense and, even when there is one, its claim to being the ‘original’ is often elusive and doubtful. This is exempli ﬁ ed in numerous modern texts, possibly the majority of those read in the West in everyday life. They span linguistic transfers between one language pair to transfers involving numerous languages. They range from labels on food and other goods in shops and modest tourist brochures to important legal documents in the European Union.


'Simple' vs 'complex' scenarios for translation
In the present discussion, the overall movement will be from what I elsewhere termed 'simple' to 'complex' cases (Dollerup 2001: 284-285), as there is some correlation between increasing complexity and problems with 'originals' and 'translations' in traditional translation the ory. 'Simple' cases are characterised by being fairly easy to describe and by having relatively few parameters which affect the translation perceptibly. A typical 'simple' case will be that of delegates at an international conference who speak in their own language and whose speeches are then interpreted simultaneously into the target language(s) for the benefi t of those who do not understand the original. In this 'simple' case there are (a) senders, (b) source-language messages, (c) interpreters, (d) messages in the target language and (e) target-language audiences.
In a 'complex' case, numerous features and parameters interact and there are more participants than one would assume at fi rst glance. We might, for instance, consider the case of the immensely popular Harry Potter books, the fi fth of which was published in English in June 2003. 1 In early 2003, the author, J. K. Rowling handed in the manuscript to the publisher, now a multinational American publishing house (Scholastic Press) which is part of Warner Bros. The manuscript is scrutinised by the publisher's in-house editors who suggest changes in phrasing and the like -a practice which is increasing worldwide and regardless of an author's status, to ensure maximum fl uency. These in-house editors are not the same in the UK and the US and especially the American ones may adjust the text to American audience expectations and political realities: apart from the differences in spelling dictated by national orthography, the British editions have a uniform no-nonsense typography and no illustrations, whereas the American ones have small vignettes at the beginning of each chapter, all missives between the characters are set as if they are hand-written, and the main text is typeset with larger letters than the British version, thus targeting the book towards a narrower American audience than the one which I believe the author had in mind in Britain. In the very fi rst book, the American edition also introduces a "black boy" as one of the new arrivals at Hogwarts school, thus upholding the principle that there must be no discrimination on the basis of race.
Once released, the book is then translated from the English source language into numerous other languages. In that process, translators will be amused to fi nd that virtually all names in the English original are delightful puns or carry connotations, such as "Weasly", "Malfoy", "Slytherin", and "Professor Sprout." One would expect this to call for creativity in translation. But this is not so with the names of all characters, for Warner Bros also has merchandise that must sell 1 The Harry Potter books attracted my attention when (in spring 2001) Ms Bertie Kaal told me that she had read the fi rst books in English, and that she had then acquired an American version of a subsequent one. During the reading, she and her daughter had found many deviations. Much information on the Harry Potter books in this article derives from a fi ne study by Katrine Brønsted (2003) of Danish translation and American adaptation of the books. To this, I have added information about the marketing and working practices in the socio-literary system.
(videogames and fi lms). Katrine Brønsted (2003) mentions that the Swe dish translator had to sign a contract that she must not change some names. However, in her discussion of the names in translation into a variety of languages, Eirlys Davies (2003: 75;85-86) cites a number of names which are, respectively, retained and changed in various languages. There is no consistent pattern which, I believe, implies that Warner Bros do not market the merchandise in all countries in which the books are published. The matter may be more complicated, but this suffi ces to show that the translation scholar who would dare to discuss translations of the Harry Potter books according to a simple model would be disregarding many factors that have considerable impact on the British 'original' and all derivative texts.
The differences between the British and American editions are relevant because there are countries, e.g. the Philipines (as a former US col ony) which are more likely to translate the Potter books from the American rather than British editions.

The changing target text: the tourist brochure about 'AErø'
In our exploration of the relationship between the 'original' and a translation, we shall fi rst look at a case with, it seems, one author and an iden tifi able source and target language.
In Denmark, there are c. eighty inhabited islands. In order to attract tourism they publish brochures in Danish and these may be translated into one or two foreign languages and distributed free of charge. One such informative brochure, published in 2001 for the island of "AErø" south of Funen, describes the tourists' arrival to the island in Danish: "Kommer man om sommeren, vil man møde et broget og mangeartet liv på øens havne, i byerne og på de snoede veje og cykelruter samt på strandene. Kommer man i det sprudlende forår, det eftertaenksomme efterår eller vaelger man at komme ved vintertide, ja så er der mulighed for at få en snak med øboen, der af natur er interesseret i andres gøren og laden." The English translation is printed immediately after this and runs as follows: "Should you arrive in the summer you will be met with the varied, multifarious life at the qauside, in the towns, along the winding roads, and of course on the beaches. Do you arrive in the sparkling spring, the thoughtful autumn or choose the winter month for a visit, you have the chance of talking to one of the locals, who by nature is interested in other people's 'toing and froing'." Later we are given some statistics. This includes the information that the island is 9 kilometres wide and has 7,532 inhabitants. In the 2002 Danish edition, these fi gures have changed -in all likelihood because there has been a census and a geographical correction. The island is now 8 kilometres wide and it has 7,315 inhabitants. The English translation of 2002 updates the statistical fi gures and, accordingly, it cites those of the Danish brochure of 2002.
The only stylistic change in the Danish text in 2002 is that tourists may arrive in the "blomstrende forår" instead of the "sprudlende forår". This is rendered as "the bloom of spring." But in the English text this is not the only new feature. There are other words and phrases that have been changed, "you will be met with the varied, multifarious life" is replaced by "you will be met by the varied and exiting life ..." The autumn is no longer "thoughtful" but "golden", and we are told of "one of the locals, whom by nature is interested in other people" -not in their "toing and and froing." Two misspellings in English are corrected, so we get "quayside" and "winter months", and, in order to make up for these improvements, two new errors are introduced, namely 'exiting' (rather than 'exciting') and 'one of the locals, whom …' instead of 'one of the locals, who …' "The bloom of spring" may be inspired by the revision in the Danish version. The other changes are not and they do not, overall, represent an improvement. They do, however, convincingly show that the English text is breaking loose from the Danish original. We cannot argue that this is a traditional 'original'-'translation' relationship as the 'translation' is embarking on a life of its own, independent of the wording of the 'original'.

The changing original: same author and source language
Above, the Harry Potter books were cited as examples of 'complex' translational contexts. There are other cases which seem to be simpler in so far as they are literary works and it is probably because they are lite rary works with some visibility that they have been noted at all. They seem to be 'simple' cases in so far as they seem to have involved only few people. (1962) is the story of an adolescent who enjoys infl icting physical harm and who is cured by being conditioned to feel nauseous when seeing violence. The fi nal chapter of the British edition shows him a few years later, hinting that he has reached maturity and may settle down. This chapter was omitted from American editions until 1988 at, it seems, the express wish of the American publisher but with the author's acceptance (for more details, see Morrison 1996: xvii).

The British writer Anthony Burgess' A Clockwork Orange
My colleague Mr Jørgen Sevaldsen drew my attention to another case. The former British politician Mr Jeffrey Archer wrote a novel, First Among Equals (1984), about political intrigue in British politics which involved the Social Democrats in addition to Labour and the Conservatives. When the book was sold to the US, and the author was advised of intended changes, he himself rewrote the book so as to conform to the two-party American political landscape: Andrew Fraser, the leader of the Social Democratic Party is left out entirely.
The point I wish to make is that although one would normally expect books, indeed any text, to be the same when it appears on both sides of the Atlantic (and in any two or more countries speaking the 'same' language) -there may be differences which are quite substantial and which may even be introduced by the copyright holder. In some cases, the 'original' is not the same in all realisations in the source language. In a translation context, the point to note is that the 'original' may also fl uctuate. We are namely not dealing with Roman Jakobson's "intra lingual translation" which is "an interpretation of verbal signs by other verbal signs in the same language" (Jakobson 1959: 233) which thus concerns what I have termed the 'linguistic side' (Dollerup 1999: 47-48;202-213). In both the above cases, we are concerned with what I have termed the 'intentional ' and 'content' layers (1999: 47, 48-50).

The fl uctuating source languages: an installation manual
We may consider an original in the technical fi eld and pick up an installation manual for 'Heizungsumwälzpumpen' made by the German fi rm Bosch. In the installation manual seven different languages are represented.
In this case we would expect the original text to be German as is the pump. German is also given prominence by being the fi rst language in the installation manual. But on the inside of the front cover there is a 'Declaration of conformity' in all seven languages. The English version of this declaration is signed by a Dane and dated "1 st December Bjerringbro ... Technical manager". So, in this manual there are at least two source languages, namely German for the technical text and either Danish or English for the 'Declaration'. However, the declaration refers to numerous EU norms and is in itself highly standardised. This means that many features in it, if not the entire declaration, must have had many near-identical precursors in terms of words, phrases and formatting, which facts thwart or at least pose formidable barriers to our quest for an unambiguous 'original' language of this brief text. I also suggest that this declaration is a document in which various European Union languages have affected one another in the process of its creation -in the same way, albeit not on the same scale as the Manifesto for European Socialists from 1993 discussed by Christina Schäffner (1997).
What we see in the manual and in the Manifesto discussed by Schäffner are a fusion of languages that is, in all likelihood, widespread. Information from fi rms also reveal that labels, instructions and manuals in several languages are not necessarily made by the manufacturers themselves but may be outsourced to translation (or distribution) agencies operating internationally with their own network of translators.

The de-culturalised original: localisation
Today we meet with 'controlled texts' that is texts that are easy to read and translate by means of e.g. consistent terminology and relatively simple syntax.
Sometimes 'controlled texts' are combined with texts written with a view to localisation. I quote from Bert Esselink's checklist for writers of texts for localisation: "Do not use culturally-specifi c text or jargon, e.g. humour, political references, slang, references to TV shows, national monuments" etc. and "Avoid references to seasons, time zones, weather, or holidays, such as Christmas trees or Halloween pumpkins. " (2000: 28) In themselves, texts meant for localisation are, of course, normal texts, except that 'cultural features' in the source texts are deliberately toned down or done away with by authors, technical writers, and technical support staff.
The localisation process which affects the target text seems to be less of a problem since Esselink refers to fairly obvious examples, when he stresses that in proofreading it must be checked that "translations con tain no spelling or grammar mistakes", that "Company names, street addresses, zip codes ... have been adjusted according to instructions", and "Metric or currency conversions are consistent and correct" , with the overall caution that "[translators and p]roofreaders need to read translated documents from an end-user perspective." This can be illustrated as follows: Illustration 1: An illustration of 'localisation' Although the tone of the literature in the fi eld is generally optimistic, it must be stressed that, so far, 'localisation' seems only to have been successful with Indo-European languages. In addition, it is a procedure which is suited for 'impersonal' texts concerned with tangible and physical objects, such as manuals, instructions for electrical appliances, software, and the like.
I call attention to the deliberate avoidance or at least minimization of culture-bound features in the source texts. This feature sets localisation aside from the kind of approach implied in the studies of Jutta Holtz-Mänttäri (e.g. 1984), Hans Vermeer (e.g. 1983), and Christiane Nord (e.g. 1991) who, I feel, all somehow take it for granted that there is an original which is (also) targeted towards an audience in the source culture. It is not that their theories are invalid: on the contrary, they are the ones that stand up best to analytical scrutiny in descriptive translation study (including of literature -although it falls beyond the scope of this article to explore this further (but see Dollerup 1999: 323)).

The absent original: the multilingual targeted texts
It is hard in this day and age to know precisely what the long-range implications of localisation will be for translation outside the fi eld 'localisation' has defi ned for itself -and which is, furthermore, changing rapidly all the time. 2 On the other hand, it would be unwise to disregard it. This is best illustrated by a study undertaken by Tine Kristensen of brochures issued by the national Danish tourist board in six languages (Kristensen 2002).
Kristensen's model is useful for an overview of the process.
Internationalisation and localisation of the tourist brochures (Kristensen 2002: 197. Modifi ed from Dollerup 2002 Kristensen's interviews with the people responsible for the brochures yielded no information about the language of the original, but her analysis of the contents seemed to indicate that the original might have been composed in English and that the English text might thus, at least in part, also have functioned as an internationalised text. The texts differ ed in terms of the information proffered and even in terms of some of the photos in the brochures: they had been localised. Kristensen's investigation brought to light another interesting fea ture. The brochures had been inspired by six 'branding values' ra ther than the hard-to-identify original. These branding values were expressed with some slight differences in the six languages. In English, they were 'cosy' and 'unceremonius' 'oasis' and 'free', and 'design' and 'intelligent'. No matter whether there was an 'original' or not, we see a transition from a stable linguistic entity, which traditional Translation Studies schol ars would term 'the original', to a cluster of fl uid concepts which are realised and given expression as different 'real texts' in different tar get languages. This is defi nitely a far cry from the 'original'-'translation' relationship.

Recycling: Translation Memories and repeated texts
It will be understood that, so far, I have been discussing whole texts.
When we turn to fragments or strings of texts, notably those used in some kind of recycling process as we fi nd it in Translation Memories, which provide translators instantly with the target-language versions of sentences and passages that they have translated before, we are on even shakier ground. When we buy some product, such as a hair dryer, there is an instruction manual. The manual will be changed every time a new feature is added to the product or a new model is introduced. But it stands to reason that, although a translator, a technical writer or a translation agency may subject the parts of the manual which need no change to a linguistic check or revision, it will only be the new and changed information that is actually translated in the traditional sense of the word, namely rendered by reference to the source-language text.
In such translations, many source texts are made up of segments and passages already found in preceding manuals in the source language. The fi rst manual which appeared with the fi rst version of the new product was translated into one or more target languages. But this is not the case with subsequent manuals in the language of origin which are changed only as far as there may be a description of some new feature in the product. It will only be a few sentences or paragraphs that are new and need to be translated into the target-language(s) and immersed in previous translation. Therefore, it cannot be claimed that every new edition of a text dealing with much the same topic or object in the source language is a 'real original' and we cannot point to one unambiguous source text. It is true that there have been sourcelanguage texts, there has been translation, and there are target texts. But since segments and passages from translations are stored electronically against later retrieval when something similar must be translated, they reappear when this happens and may then be recycled. Today, it is not all components of a given target text that have been translated at more or less the same point in time and space. In brief, there is no one sourcetext which we can term a real, let alone 'sacrosanct', original. This, I posit, is the case with a considerable amount of translation in today's world.
In most cases, however, there will be one staple: namely that, as was the case with the installation manual discussed above, a large part of a specifi c text that was originally written in only one source language -in the Bosch case in German.

The originals and translations which are equally authoritative: EU translations
The process of recycling is not only found in Translation Memories in translation agencies but also in European Union texts, notably those which are based on previous work. 3 A directive at the EU is typically made up of the following stages: 1. There is a decision (at the Commission) to develop a directive on, for instance, methods of transport for toxic fl uids. 2. This leads to preparatory studies in the Commission (experts, Commission) 3. which conclude with a green or a white paper used for discussions etc. with the member states (translation). 4. There are follow-up discussions in the member states (in government bodies, etc.).
5. Subsequently, national delegates attend expert meetings, usually at the Commission (interpreting). 6. There are meetings to make the decisions by the Council of Ministers and the Parliament (translation into all languages). 7. They include debates (interpreting) that lead to fi nal decisions which allow for 8. the publication of the directive (translation into all languages).
Final ly there are 9. reports on its implementation (translation). 10. New initiatives incorporating this directive. 4 In order not to complicate matters unnecessarily, I shall concentrate on stages 1, 3, 4, 5, and 8.
The fi rst stage does not come out of the blue. It is based on a number of previous directives in the same fi eld. All passages that have been used in previous legally binding translations concerning the fi eld must be repeated verbatim in subsequent translations -and since legal texts make up c. 40% of the language work at the EU institutions, this means that most texts are affected. In other words, even the fi rst draft of a new initiative of a directive will already incorporate recycled texts.
In stage 3, the translators at the EU institution in question will -on their computers -install relevant documentation, including all previous translations, before they begin their work. This means that, once they reach passages previously translated and which are legally binding, they fi nd a ready-made translation from which they cannot deviate. The question we may pose now is: is the initiative of the directive real ly an 'original'? The segments previously translated have been seen numer ous times. At the EU institutions, they have been analysed by legal linguistic experts, and they have been accepted by experts in all EU countries.
These translations make it to member states and are here discuss ed in the national languages. Minutes from the meetings in the members states are, of course, in the national languages. It may be that in order to promote national views, small member states have translations of their national discussions done into English or French. However, the main point is that, whatever the means of conveying national points of view, they will sometimes introduce examples of minor linguistic interference -a feature also noted by Schäffner (1997). When, in stage 5, delegates go to the EU institutions, such subtle vestiges may or may not remain. In such cases it is obvious that even the central texts in French, English, or German are not necessarily pure examples of one sourcelanguage usage only.
Most day-to-day translation work at the European Union institutions then takes place in English and French. These two languages function as the supranational core or tool languages, but I suggest that in the process of development, the core or tool texts also become repositories of the discussion of the member states at the 'linguistic' and 'thematic' level, in addition to the 'intentional' level which is, after all, the reason why they have come into existence at all. In this way, there is no one target text which has an unambiguous relation to one specifi c 'original'. The source text is a fl uid and changeable mass of text, composed of recycled translation, new linguistic material from both the core or tool languages as well as national languages incorporated in the core languages.
The last stages of the making of an EU initiative are particularly interesting. They can be illustrated as follows: Illustration: the core or tool language(s) and legally binding texts in the EU-languages: Important reports and laws are released at the same time in all EU languages (at the time of writing (2003) eleven for fi fteen member states). The one or two in-house versions of a document (in English and French) serve as the immediate source texts for these documents. As mentioned, they are fl exible and can be changed in each round of negotiations by suggested emendations and compromises. However, when the fi nal debates have taken place in the decision-making bodies, these tool texts acquire another status: fi rst they serve as a source text for target texts in all the other EU languages. The target texts are then subject to scrutiny by legal translators to ensure that they are 'the same'.
But once all the texts have been published their status changes: All the texts are now equally valid and authoritative. As also noted by Schäffner (1997), the text in the core or tool language which served as the 'original' for the translation work in the fi nal stage loses its status as an 'original' in the act of being translated and in the fi nal approval of all the EU texts. Unlike other 'originals', it cannot be re-translated with alternative phrasings and in other words. All the EU texts have the same status: the moment it is argued that the English or French text is superior to the others, there will be a major political problem.

Conclusion
This is perhaps the best example of what I term the 'vanishing original'. Yet it is my contention that it is not a new phenomenon in translation: There is, for instance, not one Christian denomination which willingly accepts that a translation of the Bible into another language is more authoritative than its own. However, the existence of the concept of the 'vanishing original' is brought forcefully to our attention because developments in the modern world bring translation closer to sourcetext production. This makes the phenomenon so eminently obvious and it also means that we have to rethink central concepts in Translation Studies. 5 5 I am indebted to Ms Helle V. Dam for useful criticism of an earlier version of this article.