Quality or speed: that is the question. The big problem for professional translators nowadays is that our clients want both. We are required, on a daily basis, to try to do the impossible, to square the circle, to attempt to maintain a satisfactory degree of accuracy and style while meeting deadlines that at times can only be described as sadistic. Levels of translator stress have gradually risen, though sadly they have not been accompanied by a commensurate rise in translation rates. I cannot help but wonder, as I glance nervously at the clock, how we have ended up in this sorry state of calendarial bondage, in which we would hardly be surprised to hear an indignant client remark, in the immortal words of Louis XIV: ‘I almost had to wait!’
The culprit is clear: technological progress, particularly in the last 20 years. Things have not always been like this. Quite recently, no further back than the 1980s, if a client living more than 50 km away wanted a translation delivered on the same day the only options would have been to dictate the text over the telephone or to send it by telegram or courier, an exorbitant solution in all cases. Otherwise, the document would have been delivered by conventional mail. The translator could relax in the knowledge that he or she had the whole day available to finish a job, not just the morning, while the client would simply have to accept the natural barrier of distance and wait – yes, wait – until at least the next day.
In fact, until the mid-eighties the entire translation process was enviably sedate, as the only option commonly available was to make a hand-written draft followed by a neat typed version. Most translation service users were willing to accept that this procedure would take some time in view of the combination of intellectual and manual effort it involved. Then the PC and the fax machine appeared on the scene, rapidly becoming ubiquitous, and the translator's existence started its inexorable slide towards complexity and haste. With the introduction of the fax – now barely used but the latest thing 20 years ago – translations could be sent anywhere, any time, even though the text that came out of the machine was often barely legible. The impact of the PC was even greater, since it forced translators to learn a whole new range of editing and formatting skills which cut completion times in half. The paper-draft stage became redundant and translation practitioners gained the capacity to write and deliver translations in electronic format on floppy discs (which in their original 5¼ inch form truly deserved that name). This was initially an option, but it rapidly became an obligation as clients realised the advantages of having copies on disc rather than paper. Thereafter came smaller, more manageable discs, the modem (which enabled electronic transmission in digital format and started making the fax redundant) and finally the Internet, the last nail in the coffin of reasonable translation deadlines.
The new technologies have enabled an enormous range of professional and non-professional activities to be carried out in a fraction of the time that was previously required, and unfortunately translation has not succeeded in remaining on the sidelines. Quite the opposite! Deadlines are forced upon us now which we grudgingly accept but which not long ago, before the personal computing and Internet revolution, would have been rejected out of hand. Yet in essence, what a translator actually does has not changed at all. The processes of analysing the source text, choosing among the available options, consulting reference works, putting it all into the target language as best one can, are the same as ever. The fact that we have been forced to learn how to type properly has speeded up production, undoubtedly, but not to the extent that would justify the unreasonable deadlines we are now required to meet. Nevertheless, we have been swept up in a tide that neither we nor anyone else can halt. The markets in which we work have become more and more technologically advanced; speed of completion is almost universally perceived as one of the greatest advantages of technological progress; therefore, translators are expected to work faster.
Unfortunately, we are expected to produce faster but we are also required to maintain the same levels of quality as before. This may be a reasonable demand when directed at car manufacturers, with their production-line robots, or accountants, whose slide-rules and pencils have been replaced by electronic calculators. It is not reasonable when addressed to translators, however, because the advantages offered by technological progress, unlike in the case of carmakers and accountants, are not sufficient to enable us to double or triple our productivity without squeezing quality, except in very rare cases. Since there is nothing we can do to change this, though, we have no choice but to try to raise our speed to the level demanded by our markets while attempting to protect quality standards as far as we can.
At times, extreme situations can arise because clients simply refuse to budge on deadlines and force us to accept jobs in dire conditions which we would love to refuse but which we cannot, or dare not, turn down. In the case of a corporate translation manager like myself, when the job is demanded by someone of sufficiently high rank you simply cannot say no, while freelance translators dare not turn down some jobs because their clients will be obliged to search elsewhere, and might not come back. In these extreme cases, therefore, translators must be able to get jobs out extra-fast, and this may require some extra-ordinary solutions. Evidently enough, the primary factors in being able to work very fast are an excellent knowledge of one's technical field (law, medicine, computing, food, whatever), which means that time can be saved by not having to consult dictionaries every five minutes, or by not even having to think, and the ability to type properly. In addition to these fundamental skills, however, there are other tools and short-cuts which can sometimes rescue us when the rocks threaten, and these are explained below.
Computer tools
Translation memory
The best-known tool for maintaining quality while cutting production times is, of course, translation memory (TM) software, of which SDL-Trados, Déjà Vu and WordFast are the most well-known examples. I do not intend to offer here a detailed explanation of how they work, or how one program compares with another. I will instead limit myself to describing how such software can save time while helping to maintain quality and consistency, provided the conditions are right.
As I have argued in past articles and papers (much to the disgust of the true believers), TM technology is very useful when a translator's habitual workload includes high levels of repetition but otherwise, although it may be helpful to some extent for work-flow and terminology issues, it is not essential. Potential users must take into account that even the simplest of these programs takes some time to learn and may at first produce nothing but frustration. An initial investment in both time and money is unavoidable: users must therefore consider carefully whether they are going to get a good return on this investment. Having said that, and assuming there is a certain volume of repetitive text, there are numerous ways in which these tools can help us speed up our output and simultaneously maintain quality.
The most obvious is through TM's main function: recycling translations that translators can trust as being 100% reliable because they did them themselves. If you are in a hurry, there is no need to waste time revising something that you already translated satisfactorily a year ago. In addition, if we use translation memory in connection with a terminology bank such as MultiTerm (SDL), the translation memory window finds and suggests terms on the fly as the translation progresses, and again this saves time as the translator does not need to wastefully leave the work-face to look up terms in dictionaries. We can also speed things up by using other professionals' translation memories, as long as we regard them as sufficiently reliable. If we fully trust a colleague's technical capacity, it is not wholly necessary to consume precious time checking their translation. Cutting out reviewing time is, of course, a key element of emergency translating. We may not like it, but sometimes we have no choice but to ‘translate to fuzzy’ and send it out, and hope like mad that the computer made the right choices.
We can also be helped by alignment tools, which enable us to create new translation memories or enhance existing ones by reusing old documents. If properly used, they can generate fantastic time savings: all you need is to have both the old translation and the old source document. If you are in too much of a hurry to be able to properly review the legacy material as it flies past on the screen, remember the golden rule of emergency translation: if the client already accepted it once, they will accept it again.
Voice recognition software
A trusted colleague of mine told me the other day that by combining translation memory and voice recognition (VR) software, he had translated documents containing around 45,000 words in a single day. The conditions, of course, were perfect: there was a fair amount of repetition and the language involved was straightforward (informal emails). Nevertheless, this level of output is truly breathtaking, considering that the normal rate for an experienced translator is usually placed at around 500 words per hour, i.e. about 3,000 words per day.
A word of caution is needed here, however. As in the case of translation memory software, the use of voice-recognition programs requires a lengthy initial training period during which a considerable amount of patience and determination is required. The program has to be trained to recognise the translator's voice while the translator has to learn how to use the program and, to a large extent, adopt a whole new approach to his or her work. Dictation is not the same as writing. Another issue is that voice-recognition software can only be used in private. On the one hand, the translator has to enunciate loudly and clearly for the software to work properly and this is not beneficial to the concentration of nearby colleagues; on the other, background noise gets picked up and incorporated into the document as actual words which make no sense in the context, and therefore background silence is a prerequisite. This makes it unfeasible for use in an office context, but for a sole freelancer at home in relative peace and quiet, it is an interesting option. If you have a dog, however, beware: my colleague informed me that the sound of his dog barking at a neighbour outside was interpreted and typed into the translation he was working on by his Dragon VR software as ‘and and and and and’.
As is evident from the above example, if a translator makes the major effort required to personalise the software and learn the technique, voice-recognition software is a great resource for emergency translation purposes. Assuming there are no formatting problems, it enables translators to accept deadlines which otherwise would be utterly impossible. Achieving 2,000 words in an hour without TM or VR would require a Herculean effort, both mentally and physically, and a long rest thereafter while one's typing fingers recovered. With these tools, particularly with VR, 2,000 words an hour becomes cruising speed. There is one final point, however, this being that (unlike with TM) the review stage cannot and must not be sidestepped, since there is always a risk of misinterpretation of similar-sounding phonemes by the software, and even in the case of emergency translations it is not advisable to deliver gibberish to clients.
Desperate short-cuts
Of course, not all translation situations are appropriate for the utilisation of the computer aids mentioned above and sometimes we have to search for other emergency solutions to meet our clients' extravagant demands. A few suggestions in this regard are set out below.
Raiding websites
Numerous organisations nowadays include their most important corporate and financial documents on their websites, in both their local language(s) and English. These documents are often grouped together on the ‘Investor information’ page and may include annual reports for shareholders, statutory financial statements, credit rating reports, corporate governance information, etc. They are obviously public documents, so there is no restriction on their use. For a translator in a tight spot, but with a bit of time available for some preparatory work, these documents can be very useful since they may contain client-specific terminology or documents from preceding years which are similar to those requiring translation. Any terminology or text found on the client's own website is evidently kosher: if the client put it there, then it must be okay for that client. By snooping around a little, therefore, we may find terms, phrases and even whole paragraphs which we can recycle in our translation, helping us to save time, maintain quality and deliver faster.
Mining contents pages and definitions
Imagine a situation in which, as either a manager or a freelancer, you have had to subcontract a translation wholly or partly to another translator whose work you do not entirely trust or, in extreme circumstances, you do not even know. This means that you are already in emergency translation territory. At the agreed time and date, he/she sends you the job and you have, say, an hour before you need to deliver it to your own client. If it's a 50-page report or contract, you clearly do not have time to check it all, but you can revise part of it. In these cases, and obviously if they are available as part of the source document, it is advisable to make use of contents pages and lists of definitions since by their very nature they contain the essential terms that will appear throughout the text. Many lengthy reports have contents pages, while in English legal practice it is typical to include a list of definitions of the main terms and expressions at the beginning of contracts, between the opening recitals and the main clauses. This emergency method, therefore, consists of looking at the contents page or definitions list in the original, checking the equivalent terms in the translation, and correcting any errors found not only there but through the whole document by using (carefully) your word-processor's search-and-replace function. In this way, you will ensure that at least the major terms are properly translated. If you have any time left, the same process can be followed as far as you can get inside the document: every time there is a serious terminology or usage error, search for it throughout the document and replace if necessary. This rudimentary procedure can make a remarkable difference. Crass terminology errors, calques and so on tend to be spotted immediately by translation users, who may judge the whole translation on the basis of a few mistakes. By eliminating these, we at least have a chance of survival.
Lists of definitions can also be used to prepare basic glossaries in a hurry if a large contract has to be distributed among various translators in order to deliver it on time. In this way, we can ensure that at least the main terms will be translated correctly and consistently.
Conclusions
The renowned translation scholar, the late Peter Newmark, once remarked that all translations should be revised at least three times: for completeness, for linguistic correctness and for style. Doubtless he was right, but unfortunately in the modern world of technical translation, time pressure usually makes more than one read-through unfeasible. And as mentioned above, even one revision is sometimes impossible. As I have tried to stress throughout this article, the reason for which translators have been forced gradually to shift the emphasis of their efforts from quality to speed is inexorable technological change accompanied by the pervading notion that anything can be done faster if you have the right technology. This is not the case in our profession, but our explanations to this effect fall on deaf ears. We therefore have no choice but to adapt.
CAT (Computer Assisted Translation) tools can be marvellous, but must not be seen as an all-encompassing solution, and even less as a reason to offer discounts. As I have argued elsewhere, accountants were not expected to reduce their fees after the pocket calculator had been invented and there is no logical basis for translators offering reductions simply because they have computer tools to help them in their work, unless the translation concerned involves an exceptionally high level of repeated text. If a client knows that – say – five contracts are almost identical in content, they might ask for part of the work to be charged as reviewing, rather than translating. This is acceptable, but it is acceptable on the grounds of what is commercially logical and reasonable, not because TM exists. Clients were asking for such discounts 20 years ago, before TM had been invented. What is not reasonable is to ask for a discount merely because part of a text can be recycled and the translator happens to have made the investment in time and money needed to acquire and learn a specific CAT tool. In fact, it would be gratifying if CAT tools, and TM in particular, were seen primarily as a means of assuring quality and consistency rather than a means of getting through translations faster, but unfortunately manufacturers tend to place the emphasis in their publicity on cost-savings, efficiencies, etc., rather than quality control, since that is what they think their own potential customers want to hear. This is clearly illustrated by the website of SDL, the market leader in translation memory products, where the first thing visitors read is: ‘Our products accelerate delivery, reduce costs and increase the quality of multilingual content for global markets’ … in that order.
Original IBM personal computer
Louis XIV
Home display of SDL Trados Office 2011, TM market leader
In his enlightening book Mouse or Rat? Translation as Negotiation, Umberto Eco describes translation, among other things, as ‘a matter of negotiation between the translator, the reader and the original author’. Being crudely realistic, we might also add: ‘… and the party paying for the translation’. The translator's negotiating position tends to be the weaker one in this respect, due to the natural balance of power between payer and payee (the threat to withhold cash usually beats the threat to withhold a translation). We have no choice but to accept this reality, therefore, but it seems to me that we also have a professional and aesthetic obligation to try to defend quality as far as we can, no matter how frustrating our clients' demands. I hope that some of the ideas outlined above will be of assistance in this regard.■
ROSS SMITH is translation manager at the Madrid offices of a major international consulting firm. He holds a degree in English Language and Literature from Edinburgh University, is a government-appointed intérprete jurado (‘translator under oath’) in Spain and is a member of the UK Chartered Institute of Linguists. He has published extensively on a wide range of language-related subjects, including computerised resources for translators, Global English and linguistic aspects of the work of J. R. R. Tolkien. Email: ross.smith@es.pwc.com
