22. Translation of scientific and technical matter. Machine translation.

From P. Newmark:

Technical translation is one part of specialised translation; institutional translation, the area of politics, commerce, finance, government etc., is the other. I take technical translation as potentially (but far from actually) non-cultural, therefore

'universal'; the benefits of technology are not confined to one speech community. In principle, the terms should be translated; institutional translation is cultural (so in principle, the terms are transferred, plus or minus) unless concerned with international organisations. For this reason, in general, you translate ILO as BIT (F), IAA (G), but you transfer 'RSPCA' in official and formal contexts, but not in informal ones, where 'RSPCA' would become something like britischer Tierschutz- bund, societe britannique pour la protection des animaux.

The profession of translator is co-extensive with the rise of technology, and staff translators in industry (not in international organisations) are usually called technical translators, although institutional and commercial terms are 'umbrella' (Dach) components in all technical translation.

Technical translation is primarily distinguished from other forms of trans- lation by terminology, although terminology usually only makes up about 5-10% of a text. Its characteristics, its grammatical features (for English, passives, nominalisations, third persons, empty verbs, present tenses) merge with other varieties of language. Its characteristic format (see Sager, Dungworth and McDonald, 1980 for an excellent review of technical writing) is the technical report, but it also includes instructions, manuals, notices, publicity, which put more emphasis on forms of address and use of the second person.

TECHNICAL STYLE

Further, unless its non-technical language is jazzed up and popularised, it is usually free from emotive language, connotations, sound-effects and original metaphor, if it is well written. French medical texts are often just the contrary, and the translator's job is precisely to eliminate these features. Thus le triptyque de ce gouvernail et les d&mes longeant la radio et le sonar font saillie, one is unnecessarily translating the descriptive term ('smooth surface' - i.e., surface lisse) by a technical term (forme hydrodynamique), and eliminating the TL linguistic contrast between lisse and hydrodynamique. (See Delisle, 1982.)

Professional technical translators have a tendency to make a mystique out of their craft by rejecting any descriptive term where a TL technical term exists; a technical term (standardised language) is always more precise (narrower in semantic range) than a descriptive term (non-standardised language). It is often insisted that one must use only words that miners at the coalface, teachers at the board (!), farmers presumably at the grass roots would use - incidentally the mystique tends to ignore any distinction between the spoken and written language, which goes against good translation.

But what if the original uses descriptive terms? Take a piece on machining schedules: Dam ce cas il est tres rentable d'utiliser les machines courantes . . . sans rien crier mais en prevoyant en detail leur adaptation et leur montage. Les machines courantes could be translated as 'general-purpose machines' in anticipation of leur adaptation et leur montage. In this translation, the semi-technical term 'general-purpose' replaces the descriptive courantes of the original. More likely, courantes is in contrast with sans rien creer, and could be translated by the descriptive terms

'standard', 'normal', or 'currently in use'.

Whilst the technical term may be a translator's find (trouvaille) and will help to acclimatise the professional reader, it is I think mistaken to invariably prefer it, bearing in mind that the descriptive term in the SL text may serve other communi- cative purposes. In cases where the piece is technical and there is clear evidence (as there often is) that the descriptive, the more general and generic term is probably only being used because the narrower technical term is rare or lacking in the SL, the use of the technical term in the TL text is certainly preferable.

Conversely, where an SL technical term has no known TL equivalent, adescriptive term should be used. What to do with dismicrobismo murino? 'Microbism' being 'a state of infestation with microbes', the dis- (English: dys-) appears redundant. If one cannot risk 'murine microbism', a descriptive term such as 'acute infestation by microbes, due to rat fleas' is safer. Again rideau de terre, a technical term for a bank separating two terraced fields, is translated as a 'ridge' in La Durie's Montaillou. Little is lost in the context.

BEGINNING TECHNICAL TRANSLATION

I think that the basic technology is engineering and the basic branch of engineering is mechanical; if you want to become a technical translator, that is where you start. However, you should not specialise at the start, but, as in any post-graduate translation course, get as much practice as possible in a range of technologies, in particular the ones that are thriving (de pointe), which, at present, means computer applications in the spectrum of commerce (particularly the tertiary sector) and industry. Again, bear in mind that you are more interested in understanding the description, the function and the effect of a concept such as entropy rather than in learning laws, particularly axioms, theorems, theories, systems in some of which entropy is involved. In a sense, you are learning the language rather than the content of the subject, but, when I say of the terms that the function is as important as the description, and always easier to grasp, I am in fact bringing you back to the application of the laws and principles. When you translate a text, you have to be able to stand back and understand roughly what is happening in real life, not just, or as well as, convincing yourself that the sentence you have just translated makes sense linguistically. You mustn't write the technical equivalent of: 'The King of France is dead'; there must be a thread of action running through the passage which you can grasp at any time. Even though much scientific and technological language and terminology can be translated 'literally' and in newer subjects contains an increasing number of internationalisms and fewer false friends, you have to check the present validity in the register and dialect (viz. usually British or American English) of the terms you use. But here again, there are priorities. Technical terms that appear on the periphery of a text, say relatively context-free in a list or a foot note, are not as important as those that are central; their nomen- clature can be checked without detailed reference to their function or the descrip- tion. In a word, to translate a text you do not have to be an expert in its technology or its topic; but you have to understand that text and temporarily know the vocabulary it uses.

In science, the language is concept-centred; in technology it is object- centred: in, say, production engineering, you have to learn the basic vocabulary with the translations - e.g. 'lathe', 'clutch', 'clamp', 'bolt', 'mill', 'shaft', 'crank', etc. - in diagrams as in the Wiister and Oxford Illustrated dictionaries and obtain a clear idea of outline, composition, function and result, as well as learn the action verbs with which they normally collocate: une came tourne - 'a cam rotates'.

TRANSLATION METHOD

Both text and translation are 'thing-bound'. According to Barbara Folkart (1984)

'the latitude which the translator enjoys is subject only to the constraints of register, and possibly, textual cohesion.' This statement is questionable since the SL text is also the basis of the translation, however much the translation departs

from it: (a) owing to its different natural usage; (b) if it has to be referentially more

explicit than the original - in particular in the case of gerunds and verb-nouns, where case-partners may have to be supplied in the TL text (see Chapter 12). Thus Folkart's example: un dispositif de fixation d'un pignon d'entrainement des organes de distribution becomes: 'a device with a pinion attached to it, driving the machine parts which distribute the fuel mixture to the cylinders' (my translation); in such an example, the case-partners have to be supplied referentially; in other contexts, they can be 'recovered' from the previous sentence.

It is notable that whilst Folkart recommends a 'thing-bound' approach which appears at times to be independent of the SL text, her examples are close trans-lations modified by: (a) TL syntactic constraints; (b) appropriate explanatory reference. The point is that when a thing or a situation is 'pinned' down in a (SL) text, particularly a well-written one, it becomes precisely described. If a translator tries to set out from the object or situation, forgetting the SL verbal details, it is going to be an inaccurate translation. 'Co-writing', in which two or more copy- writers are given the description and function of a product and asked to apply them (in an advert) to local conditions, temperaments, sense of humour in various differing examples (e.g., Telecom adverts for phoning home), is fine, but any

'common' message is better translated.

When you approach a technical text (it should be useful to you at this point to refer to p. 161) you read it first to understand it (underline difficult words) and then to assess its nature (proportion of persuasion to information), its degree of formality, its intention (attitude to its topic), the possible cultural and professional differences between your readership and the original one. Next, you should give your translation the framework of a recognised house-style, either the format of a technical report adopted by your client, or, if you are translating an article or a paper, the house-style of the relevant periodical or journal.

You have to translate or transfer or, if not, account for everything, every word, every figure, letter, every punctuation mark. You always transfer the name of the publication, a periodical. You translate its reference ('Vol.1., No.5') and date, and the general heading or superscript (Mise au point, Rappel Medical, Travaux et Originaux Memoire, could become 'Medicine Today', 'Medical Review',

'Papers and Originals', 'Original Research Paper' respectively) using the standard formulae of the corresponding English periodical. For authors, delete the par or von, reproduce names and qualifications, and transfer the place of the author's appointment (e.g., Ecole des Hautes Etudes is not normally to be translated) - the reader may want to write to the author. However, you can translate and transferred word in a footnote, if the house-style permits, if you think the clients or readership would find it particularly useful, particularly if it is not a 'transparent' word.

From Hatim & Munday:

Translation in the information technology era

TRANSLATION, GLOBALIZATION AND LOCALIZATION

The volume of translation conducted worldwide has increased dramatically in the last fifty years. Even though English may have become a lingua franca of world trade, it is the increasing globalization and the advent of the internet that have meant that promotional literature, technical manuals, webpages and all ranges of other communication are being translated into other languages at a faster and faster pace.

Globalization is a multi-level term that is used to refer to the global nature of the world economy with the all-pervasive spread of multinationals. In commercial translation it is often used in the sense of the creation of local versions of websites of internationally important companies or the translation of product and marketing material for the global market (see Esselink 2000:

4). Michael Cronin’s Translation and Globalization (Cronin 2003) deals with some of the complex cultural, political and philosophical consequences of translation in the global age.

In addition, the growth of international organizations such as the United Nations and the European Union has made interpreting at meetings and translation of documentation a necessity. In the case of the European Union, the commitment to translate into all official languages of the Member States has seen the number of original pages of documentation translated by its Translation Centre in Luxembourg alone increase from 20,000 in 1995 to 280,000 in 2001 (European Communities Court of Auditors 2002). The turnover of the Translation Centre, employing 140 permanent staff and tendering out a portion of its work to commercial translation agencies, was almost 26 million euros in 2001 (European Court of Auditors 2002:

9). But this is just a small portion of the EU’s translation and interpreting costs, estimated in 2001 at 2 billion euros per year (Austermühl 2001:3).

Computer power is therefore being harnessed by the translation industry, but it still remains Computer-Assisted Translation. The goal of fully automatic or Machine Translation (MT) remains elusive although recent developments have been more promising. First a little history to put it into context:

Bar-Hillel considered that real-world knowledge was necessary for translation and that this was impossible for a machine to replicate. He felt that the goal of a fully mechanized translation on a par with that produced by a professional translator was unrealistic. In his opinion, it would be more realistic to attempt to produce machines that worked in conjunction with humans.

In a brief overview of the history of the field, Martin Kay (1980/2003) discusses some of the obstacles to successful Machine Translation including ‘words with multiple meanings, sentences with multiple grammatical structures, uncertainty about what a pronoun refers to, and other problems of grammar’. He uses a now well-known example (Example A14.1) to illustrate the problems:

Example A14.1

The police refused the students a permit because they feared violence.

The first real developments in Machine Translation (MT) took place after the Second World War, during which the first computers had been invented in the UK by Alan Turing’s team as part of the now famous code-breaking operation at Bletchley Park (Hinsley and Stripp 1993). The beginning of the Cold War in the late 1940s prompted significant investment by the US government in automatic Russian–English translation systems for the military; France, Japan, the UK and the USSR had smaller programs. These first-generation systems were known as ‘direct’ systems since they were basically word-based ‘direct- replacement’ systems; each ST word would be looked up and replaced by a corresponding TL term. As we have seen in Unit 2, word-for-word substitution is not a solid base for translation. Without significant progress, MT’s reputation fell very low in the 1960s following damning criticism by Yehoshua Bar-Hillel in his Report on the State of Machine Translation in the United States and Great Britain (1959) and in the report published in 1966 by the Automatic Language Processing Advisory Committee (ALPAC). Instead the focus shifted to more basic questions of language processing, the field that became known as computational linguistics.

From Korunets:

MACHINE TRANSLATION

Rendering of information from a foreign language with the help of electronic devices represents the latest development in modern translation practice. Due to the fundamental research in the systems of algorithms and in the establishment of lexical equivalence in different layers of lexicon, machine translating has made considerable progress in recent years. Nevertheless, its employment remains restricted in the main to scientific and technological information and to the sphere of lexicographic work. That is because machine translation can be performed only on the basis of programmes elaborated by linguistically trained operators. Besides, preparing programmes for any matter is connected with great difficulties and takes much time, whereas the quality of translation is far from being always satisfactory even at the lexical level, i.e., at the level of words, which have direct equivalent lexemes in the target language. Considerably greater difficulties, which are insurmountable for machine translators, present morphological elements (endings, suffixes, prefixes, etc.). No smaller obstacles for machine translation are also syntactic units (word-combinations and sentences) with various means of connection between their components/parts. Besides, no present-day electronic devices performing translation possess the necessary lexical, grammatical and stylistic memory to provide the required stand ard of correct literary translation. Hence, the frequent violations of syntactic agreement and government between the parts of the sentence in machine translated texts. Neither can the machine translator select in its memory the correct order of words in word-combinations and sentences in the target language. As a result, any machine translation of present days needs a thorough proof reading and editing. Very often it takes no less time and effort and may be as tiresome as the usual hand-made translation of the same passage/work.¹ A vivid illustration to the above-said may be the machine translated passage below. It was accomplished most recently in an electronic translation centre and reflects the latest achievement in this sphere of «mental» activity. The attentive student will not fail to notice in the italicized components of Ukrainian sentences several lexical, morphological and syntactic/structural irregularities, which have naturally to be corrected in the process of the final elaboration of the passage by the editing translator.