echnology has been reshaping the concept and practice of translation in many aspects. Until some time ago, translators were expected to be able to work solely on definite source texts with the exclusive aid of dictionaries. Specialists were called upon where research references failed or left holes, but, even in such cases, translators had the chance to develop familiarity with their source texts, becoming, in many cases, experts themselves in some fields. Textual material to be translated was basically conceptualized as having a beginning and an end, thus making contextualization of meaning easier.
The process of globalization and the technological revolution that came along with it have dramatically changed the way information is conceived and produced. According to Craciunescu et al. (2004), advances in communication have brought about a "screen culture" that increasingly tends to replace the use of printed materials, since digital information can be easily accessed and relayed through computers and allows greater flexibility for processing.
The pace of the contemporary world calls for translators to deliver their work in shorter and shorter turnaround cycles.
In addition to the growing tendency to adopt the digital format for textual production, a large part of the material translators deal with in their daily routines consists of large translation projects, whether web-based or not. Such work is usually carried out with the use of computerized tools, such as automatic translation systems and translation memory databases. These applications require the development of a new range of technical competences, from learning how to manipulate different software programs to being able to manage the translated content, whether to achieve terminological consistence (machine translation databases) or to reuse solutions to translation problems in subsequent projects (translation memory databases). As Biau Gil and Pym (2006:6) explain, in today's world,
Our translations might thus be expected to move away from the ideal of equivalence between fixed texts, becoming more like one set of revisions among many. In the fields of electronic technologies, translators are less commonly employed to translate whole texts, as one did for the books with concordances. Translation, like general text production, becomes more like work with databases, glossaries, and a set of electronic tools, rather than on complete definitive source texts.
Translation memory tools are being employed also by translators working with definitive texts, that is, materials that might be translated just once, mainly as a way to increase their database. There are many translators who work basically with web-based materials, so most part of their work might involve updating and adaptation of previously translated texts to other contexts (a common practice in localization). Whatever the situation technology might be employed, there is no denial that translators have been able to reap great benefits by achieving greater work speed and efficiency.
Nonetheless, the same tools designed to assist translators are also affecting many aspects of how their work is regarded. This is mainly due to the fact that the design of these applications seems to be based on some of the traditionally-held concepts of translation as a transfer operation of pre-established contents stored in the source text and of the translator as the one in charge of retrieving the contents the machine has failed to recover.
While seeking to investigate the basis of machine translation and translation memory programs, this work aims to analyze both the contributions and transformations arising from the contemporary concept of the translation profession through the use of those tools. The ideas presented are divided into two sections. In section one I shall examine the concept of the original text and the translation in the domain of machine translation. My attention will then turn towards the extension of the translator's responsibility in producing the final text, by examining the translator's role in the translation post-editing process. Section two looks into the application of translation memories, with focus on the extension of the translator's responsibility in creating translation databases and re-using identical or similar segments from previous translations stored in memory programs. Ultimately, I shall conclude by attempting to draw attention to the scenario posed by these technologies which, in my view, seems to raise urgent ethical questions regarding the translator's image as re-creator or editor of the final translated material.
Machine translation: the illusion of access to the source
The pace of the contemporary world calls for translators to deliver their work in shorter and shorter turnaround cycles. This fact, coupled with the search for cost reduction, seems to be one of the strongest reasons supporting the use of machines in translation. Through the perspective of the current demand for readily-done translations, the applications of machine translation programs are not seen as a break with the tradition, but as an inevitable further step in the development of the practice.
However, the growing demand for application of machine translation programs as a means to speed up translation and reduce its costs is changing the way texts are read and conceived. As Cronin (2003:22) aptly observes, "if the pressure in an informational and global economy is to get information as rapidly as possible, then the 'gisting' function becomes paramount in translation, a tendency which can be encouraged by the 'weightlessness' of the words on the screen with their evanescent existence." The generally low threshold of translation acceptability shown by many users is often justified by the argument that getting access to the informational content of a text is all that matters and that some translation, however poor, is better than no translation at all.
The prevailing idea among users is that meaning may be transported from one language to another and that machine translation programs never fail to convey a general and stable content, even though such operation may result in a roughly intelligible text. The current urgency to communicate seems endorse the notion that the content of a textual material is solidified in the source and that machine translation may provide access to the origin. As Hutchins (1999:4) claims, machine translation represents an "ideal solution" for the translation of texts for assimilation of information, that is, direct and quick access to the source, since
human translators are not prepared (and resent being asked) to produce 'rough' translations of scientific and technical documents that may be read by only one person who wants to merely find out the general content and information and is unconcerned whether everything is intelligible or not, and who is certainly not deterred by stylistic awkwardness or grammatical errors (Hutchins, 1999, p. 4).
According to this view, if the machine is in charge of recovering the content, although "awkward" and imperfect, the translator's role would be restricted to editing and stylistically adapting the translated material. As often quoted by machine translation scholars, automation should not be seen as a replacement for human translators, but as way to magnify human productivity (Kay, 1997), to supplement human translation (Melby, 1997) or even create more work for human translators (Biau Gil & Pym, 2006).
The issues regarding machine translation seem always to revolve around the descriptive views of its possible uses and the constant reminder that its applications can never supersede the abilities of human translators. However, nothing seems to be said about the extent of the translator's function in the construction of the final text that was initially translated by machine. Since original meaning recovery by the machine is often taken for granted by users, the translator's work is limited to filling out some gaps left out by the machine and stylistically adapting the translated text to the target language.
Even if the message seems to be incoherent in the "draft version" automatically prepared, there is a strong belief that the source has been recovered and adjustments are all that are left for the human translator to do.
The source-target correspondence has been a debatable issue for many years and the realization that it is at the very basis of machine translation concept brings into question the role the translator is supposed to play. If we accept the notion that the source is thus recoverable by the machine, we might have to be willing to accept that, in the post-editing job usually entrusted to human translators, the task to be carried out will be less of interpreting and reconstructing meaning in the target language, and more of allowing the automatically recovered meaning to be comprehensible through revision and adaptation.
Through this view, there is always the risk that the translator's work may remain concealed behind that of the machine, at least in most clients' eyes. Through the postmodern perspective, as the work of Brazilian Translation Studies scholar Arrojo (1997) has emphatically pointed out, "no reading can ever aspire to repeat or protect someone else's text"; therefore,
The visible translator who is conscious of his or her role and who makes the motivations, allegiances, and compromises of his or her interpretation as explicit as possible is also the translator who must take responsibility for the text he or she produces, as it is impossible to hide behind the anonymity of the ideal 'invisibility' which has allegedly been given up. (Arrojo, 1997:18)
Embracing visibility, as well as the sense of responsibility for the construction of the translated text, may be one of the most powerful ways for translators to value their work. As translators avail themselves of machine translation capabilities, whether by choice or by their client's imposition, they should likewise consider whether the speed and terminological consistence provided by the machine are worth the price of having their work downgraded as being merely of a copy editor and not as the one responsible for bringing meaning forth.
Translation memory programs: transferring translators' past solutions to present contexts
Just as machine translation, translation memories have also been imposing changes in the way translators work, many of which with further ethical implications than they might appear to have at first sight.
The basis of translation memory programs lies in accumulating and storing translation solutions that are recycled as needed through the automated use of this terminology. In addition to the investment required in the acquisition of these programs and the training needed to use them properly, time is also another factor that directly influences their performance. Far from being immediate time-savers, translation memory programs are built up as they are used; therefore, the more frequently the translator employs them, the larger the database and, consequently the more useful it will be.
Although the literature on translation memories, which basically comprises descriptive and comparative analysis of the efficiency of the programs available in the market, highlights the remarkable gains in productivity translators have been able to achieve, especially in the realm of localization (e.g. Microsoft Windows-based software localization project), little consideration has been given to the controversial ethical issues arising once a terminological database is created (Zetzsche, 2000).
Once a translator has compiled terminological options into a database as a result of previous work commissioned directly by clients or through translation agencies, it is usually expected that such database be provided along with the final translation, a usual procedure as these programs become widespread. When that database is incorporated into a larger one held by clients or translation agencies, this data will often be used as input to be provided to the same translator or other professionals working in future projects. As Biau Gil and Pym (2006) explain, whenever a translator is provided with a memory database, clients expect compliance with the terminology and phraseology of the segment pairs included in that database. Moreover, always concerned with reducing costs, companies encourage translators to seek as many matches as possible and feel disappointed when the level of text re-use reported by translators is much lower than expected (Murphy, 2000).
Just as in translation machine applications, the way translation memory programs are being designed by the industry, in a effort to achieve cost and time reductions to produce a translation, calls into mind the concept of translation as "a word-replacement activity" as Biau Gil and Pym argue, since most of the time, translators "are invited to forget about the other elements configuring the text" and concentrate on segments that might be recovered from translation databases or added to the latter. (2006:12).
The translator's interpretation of the source material and personal choices made in the formulation of the translated text might interfere with content management and consistency, even though the translator's option may at times be more appropriate for some specific context than the pre-selected options offered by the database. By reusing stored translation segments, translators might be giving the first step to relinquish the outcome of their research work as well, since clients may also require that the memory generated through a translation be provided along with the translated material.
The second step towards giving up the authorship of translation takes place whenever translators accept being paid only for what clients deem as translation per se, that is, segments that have not been translated yet. This is the result of the idea that 100% matches will keep the same meaning they had in previous texts and, for that, revision and adaptation of these words or segments are not worth being remunerated for. This situation has been met with criticism by some translators who defend that consistency does not guarantee comprehension since a text may have perfect terminological coherence, but it altogether may not make any sense to the target audience.
On the other hand, clients may not readily accept the idea that 100% matches cannot be used in a new context or even that segments that may be used inevitably gain new meaning and may still require careful revision and adjustments into the new context. For that reason, it is uncommon to remunerate the translator for corrections or adjustments in the terminological database, since it would mean accepting that previous translation work was faulty, and so unduly charged.
From that commonly adopted practice in the work with translation memories, we find an approach rather similar to that applied to machine translation. Just as there seems to be a consensus that a text translated by machine will require not much more than review and post-editing by a human translator, in the work with translation memories, reviewing also frequently includes maintaining previously translated segments. Despite the fact that segments stored in the memory may have inadequacies, they may just as well lull the translator into a false sense of belief that meaning is fixed and will not change or lead to new associations in the new contexts they have become part of.
As I hope the discussion on commonly used approaches to machine translation and translation memories have led us to consider, for the translator's role to become more, rather than less, important in the informational age, it is paramount that these professionals do not be regarded as mere content transporters or text reviewers, but as communicators fully responsible for the meanings they confer to translated texts.
Final considerations: Co-existence but on what terms?
If there is no denial translation practice has to evolve and conform to new modes of communication and work (Cronin, 2003), translators should ponder how they wish to be regarded by those who hire their services.
By conferring priority to discussions about time and cost reductions through the application of technological tools in the practice of translation, translators might be oblivious to what such tools really represent to the public in general and what consequences such representations might have in the way the profession is conceived.
The general idea, as I have argued, is that, when applying technological tools such as machine translation programs, all that is left for the translator to do is give the text the final touches to make it coherent in the translated language. The impression is that the machine is the one that does the translation work and the translator is in charge of editing the final text. As for translation memories, the widespread reuse of already translated segments, in a way, also contributes to the idea that the translator is not solely responsible for the translated text.
The illusion that the machine is able to translate may affect the way translators will be seen in the future, an impression that should be given careful consideration, mainly when we remind ourselves of the multiplicity of texts, languages and cultures that are inevitably intertwined in translation.
ARROJO, Rosemary. Asymmetrical relations of power and the ethics of translation. TextconText, v. 11, p. 5-24, 1997.
BIAU GIL, José Ramón; PYM, Anthony. Technology and Translation: a pedagogical overview. In: PYM, A., PEREKRESTENKO, A., STARINK, B. (Org.) (2006). Translation technology and its teaching. Tarragona, Spain. Available at <http://isg.urv.es/publicity/isg/publications/technology_2006/idex.htm>. Access on June 22, 2006.
CRACIUNESCU, Olivia; GERDING-SALAS, Constanza; STRINGER-O'KEEFFE, Susan. Machine translation and computer-assisted translation: a new way of translating? Translation Journal. v. 8, n. 3, jul. 2004. Available at: <http://www.accurapid.com/journal>. Access on: May 15, 2006.
CRONIN, Michael. Translation and globalization. London: Routledge, 2003.
HUTCHINS, John. Translation Technology and the Translator. Proceedings of the Eleventh Conference of the Institute of Translation and Interpreting, London, May. 1997. Available at: <http://ourworld.compuserve.com/homepages.WJHutchins>. Access on May 10, 2006.
______. The development and use of machine translation systems and computer-based translation tools. Proceedings of the International Symposium on Computer Language Information Processing. Xangai, June, 1999. Available at: <http://ourworld.compuserve.com/homepages.WJHutchins>. Access on May 10, 2006.
KAY, Martin. The proper place of men and machines in language translation. In: Machine Translation. n. 12, p. 3-23, 1997.
MELBY, Alan. Some notes on 'The Proper Place of Men and Machines in Language Translation'. In: Machine Translation. n. 12, p. 3-23, 1997.
MURPHY, Dawn. Keeping Translation Technology under Control. Machine Translation Review, n. 11, Dec. 2000, p. 7-10. Available at http://www.bcs-mt.org.uk/mtreview/11/mtr-11-7.htm. Access on Jan. 11, 2007.
ZETZSCHE, Jost. Translation memories: the discovery of assets. Multilingual Computing and Technology. v. 16 (4), n. 72, p. 43-45, 2005.