|
Opportunities for a sea change Quality, the market and translation technology
Michael Benis sees new trends in quality thinking and translation technology as pointing the way to a brighter future
Recent trends impacting on quality and the professional profile of our profession
Two frequently-discussed trends that have dominated the translation industry over the past decade are rates on the one hand, and our professional profile on the other, with the one being inextricably linked to the other, and both being perceived to have followed a downward spiral. Globalisation and translation technology are considered to have played a major role in this decline by introducing lower-priced competition from developing countries in the case of the former, and by tipping the balance of the supply-side discourse away from the skills and experience of the translator towards process efficiencies and integration within a Global Information Management (GIM) scenario. Quality has largely been relegated to considerations of appropriate terminology and terminological consistency, while aspects such as structure and style have tended to be dismissed as subjective and outside any relevant quality metrics. In many cases the inference was almost that translators could be treated as interchangeable commodities, partly because it was considered that any terminological disparities or deficiencies could be corrected easily during the quality control stages, including through the application of automatic terminology checks by project managers with little experience as translators themselves.
At the same time, with translation company margins shrinking, an increasing amount of the translation workload has tended to be outsourced, with this in many cases even including project management itself. Indeed, it has become a frequent scenario for large international translation companies to outsource to small local translation companies, often with a staff that can be counted on the fingers of one hand, and still leave several fingers spare. While this particular trend has a theoretical quality-related foundation in the advantages of outsourcing to specialist companies based in the target-language country, the choices have, however, all too often been based on purely economic considerations, resulting in mismatched teams of poorly-paid translators put together at the last minute and, as described above, revised by a project manager using translation technology to facilitate terminological consistency and quality control, while also hoping to leverage translation memories to increase productivity.
The difficulty of judging the effectiveness of a translation in a language other than one's own has allowed this trend to consolidate itself over several years, with end-clients largely accepting this quality model, not least of all since they were happy to see the enormous global costs of their information management decreasing as result. I'll be considering how this situation is changing later in the article, but first I'd like to take a brief look at how translation quality metrics and management are generally defined in the industry.
Professional standards and the current quality paradigm of the translation industry
The most significant reaction of translation companies and translators' associations to increasing competition, especially those in developed countries where rates are high, has been to attempt to develop a competitive edge through professional credentials, such as the recent introduction of the Chartered Linguist in the United Kingdom, and the European and US standards that are currently in the final phases of their development. Interestingly, and this should not be taken as a criticism, these initiatives nevertheless see quality management in the same terms as the system efficiencies approach to which they are, at least to an extent, a reaction, aiming to restore focus to translation quality as something that goes beyond the terminological "correctness" and consistency that can be measured by an automated or semi-automated system. Let's consider how both currently view the quality management process.
["Quality Pyramid" diagram - Slide 1]
In almost all cases, the quality assurance system is seen to operate like a pyramid in terms of the volume of resources required for any project, and to proceed in a simple linear fashion from start to finish. There are five basic components in the process. The first of these components is also the first stage in the quality control process, and consists of a check for completeness to ensure that the translation contains all the information present in the source text, making sure, for example, that there are no paragraphs or adjectives etc missing. The second component and generally the second stage is a check for terminological accuracy and consistency, increasingly performed by a project manager using translation technology tools that analyse the translation against the source text and an electronic glossary/termbase. Some companies will combine both these phases in a single automatic or semi-automatic check. The third component and stage is a check on file integrity, which analyses the translation against the source text to ensure that the appropriate structural elements, such as character and paragraph formatting, as well as tags in HTML and XML files, for example, are all present and used in the translation in the same way as in the source text. The fourth component is process control, which continues throughout the translation management cycle to ensure that every operation required has been performed correctly, including preparatory aspects, such as the splitting up of large translations and the assignment of the sections to the most appropriate translators, as well as the monitoring of their work and that of everyone else in the team (revisers, DTP specialists etc) to ensure deadlines are met. The fifth and final component which is essential for effective quality assurance is to have a defined and monitorable system for implementing all of the previous four components.
["A closed traditional QA model..." - Slide 2]
Failings of the current quality model
As stated, this Quality Assurance approach generally operates in a linear chronological order, proceeding from the source text through to the final translation delivered to the client. It has largely gone unquestioned in the translation industry, not least of all because it follows the traditional translation quality paradigm of fidelity to the source text that we have inherited from academia. One of the purposes of this article is to briefly challenge some of the assumptions inherent in this approach. I'd like to do this not so much by turning the concept on its head, as by turning it on its side. Although a cursory glance at the pyramid would suggest that it should be turned to the right, like an arrow that follows the direction of Western text, taking us chronologically from our source text origin to our final translation destination, this conception of Quality Assurance in fact looks backwards in every respect from the translation to the source text. The source text is, in other words, considered to be the authority for any quality metrics. I'd like to examine the validity of this assumption with a simple question: Where does that leave the client?
["Where does that leave the client?" diagram - Slide 3]
Looking at the diagram that illustrates this question, we see our champion of quality referring back to the source text as their authority, while our client looks instead to the market and has to ponder a large number of cultural variables which could significantly undermine the ability of a source-text-circumscribed communication to interface effectively between their company and their market.
Even if we consider a relatively "straightforward" technical document such as a short user manual with a brief "branding" introduction, there will be a whole series of aspects that require substantial changes to be made with respect to the source text where completeness, terminology and file integrity are concerned. For example, cultural differences could require that any colour references and metaphors used are changed. Character and paragraph formatting may also need to be altered. Some cultures make much more intensive use of bold, for example. Market and brand differences may equally require very different messages to be contained in the introduction, adding new messages and omitting others contained in the original. Industry practices could mean that the order of instructions needs to be modified. Examples illustrating the use of the product may also need to be changed because of differences in fashion and other trends between the two cultures concerned.
In many cases, therefore, the current approach to quality leaves the translation industry facing in one direction and the client in another. What this effectively means is that the industry which should be helping the client to bridge cultures is in fact failing to cross the cultural divide. As a result, recent trends in the industry risk leaving the client with potentially ineffective solutions that represent poor value, despite the significant savings that GIM efficiencies have achieved. An excellent example of what this means can be found in a case history that was presented at the ClientSide News Workshop this September in San Francisco, briefly summarised as follows:
“JD Edwards (JDE), a billion dollar software company, found itself with a 290% ROI on its optimized content and translation processes but was delivering content that customers were using as doorstops and paperweights. In order to drive value to the customer (as well as to their bottom line), JD Edwards had to shift from a cost-based optimization model to a value-based strategy.”
This analysis implicitly assigns a crucial role to a different quality model as a catalyst in achieving this shift from a cost-based optimisation model to a value-based strategy. But I'm jumping the gun a little here. I really ought to first define this quality model.
As a preface to that, I'll enumerate a number of common comments that I have encountered from trade and industry representatives in several countries around the world to underline how the current, closed quality model is failing the industry:
“It’s English, but not as I understand it. Not only does it read badly, but it's extremely patchy. In some cases the translation is completely incomprehensible.”
“I don’t expect to use it as it is – after all it’s only a translation.”
“The translation industry speaks a quality jargon based on process and terminological accuracy. That's really just the absolute minimum. It doesn’t understand our needs…”
“The translation industry is inward-looking and does not communicate effectively.”
Moving from a closed model to an open model
So, what's the alternative to this model, or rather how can we complement it and open it out, so that it's no longer inward-looking? I'd like to set out towards my definition by once again taking a technical product as an example, let's say a motor car engine. The current translation industry paradigm would measure the quality of a production line engine in terms of the extent to which all its measurable parameters correspond to the engine's blueprint and prototype, its "source text". The problem where the market is concerned is that the consumer will measure quality in terms of how the engine performs on the road: its acceleration, smoothness, fuel consumption and, perhaps above all, reliability. That's why every motor vehicle manufacturer takes great care to also monitor how their engines perform in the field, which components fail and how they can be improved, and how satisfying their customers find the engines on the road.
My contention is that the translation industry should be doing the same thing. In other words, we want to define quality not as fidelity to the original but as suitability for purpose. We should be advising our customers proactively on all the possible changes that may be needed, along the lines of those in the example of the short user manual above, and we should likewise be working with our customers to monitor how effective our communications are "on the road". And there's a very good reason for that, even if we just consider that short user manual, once again. If, for example, the technical support line serving a customer finds that consumers are continually encountering problems with using a particular function of the product on "our" market, we should work with them to consider whether this is a product usability issue or a cultural/communication problem that could be overcome by rewriting the relevant section/s of the manual. In short, we should act like every other industry, measuring quality not only during production, but also during use and in the minds of our users.
This should, at the very least, see us working with the client before the translation begins to agree and draw up a brief specifying what they need the target text to achieve. This would represent a true paradigm shift, seeing us move from the closed model of our pyramid to an open model in which we interact both with the client and their market to identify, deliver and fine-tune truly effective solutions. That's the only way we can help them move from a "doorstops and paperweights" situation, or - even worse - an ineffective user manual that erodes public confidence in the product and drives down perceptions of product quality, instead driving value to their customers through effective communication. What's more, it's a path that's likely to have the single greatest impact on raising the profile and clarifying the value added by our profession.
["A paradigm shift" - Slide 4]
The freelance translator in an open model
The scenario I have just described, portraying this open, interactive quality and indeed service model in action, offers a clear role to the freelance translator who is working as a communications consultant directly with an "end-client". In fact, specialist freelances probably account for the majority of such service providers currently working in the industry, alongside small highly-specialised companies. This same role is clearly also suited to the senior executives and account handlers of larger translation companies. Where the problems arise is when a large company attempts to provide a similar level of support for very large multilingual projects that require more than one translator per language. If we try to mesh this new approach with the current situation of freelance translators working with large translation companies, we can see that the new model won't be applicable without some radical changes even if each project is accompanied by a comprehensive brief, not least of all because this open model asks the translator to make a radical change in their working practices and the manner in which they develop and deploy their knowledge and skills. So, attempting to support this change in practices, the question this time would be: "Where's the translator in all this?"
["Where's the translator in all this?" diagram - Slide 5]
The current relationship between translation companies and freelance translators causes special difficulties due to the fact that in most cases there is very little if any feedback from the end client to the translation company and then on from the latter through to the freelance, who generally works in isolation. Additional but related difficulties concern how one can continue to gain the benefits offered by translation and GIM technologies, while at the same time keeping the translator "in the loop" about client and target market requirements, so that they can act as an effective cultural bridge, intervening to make "radical localisation" changes where required. An essential prerequisite in all this would be for the translator to play much more of a partnering role, and certainly to be included, as would any other communications professional, in the initial briefing and consultation work - something that can be achieved quite easily today, even with freelances that are located many miles apart, using Voice Over IP and Video Over IP teleconferencing technology. But this is not the only technology that helps make our new quality and service paradigm feasible for large multinationals. Translation technology and CAT in particular also have a role to play.
TM is dead! Long live CAT!
To understand this, it will be helpful to recognise that the translation industry and, indeed, TM software houses themselves, have for many years tacitly acknowledged that TM is a tool offering very limited benefits. Whereas it is a helpful technology for processing new versions of previously-translated documents, or large documents in extensive series that are put together on a modular basis, it has been found to offer much more limited benefits than had been hoped for when it comes to increasing either productivity or quality in other contexts. It is outside the scope of this article to discuss all the reasons for this, although it certainly wouldn't be the first time they have been covered in bulletin. Suffice it to say, however, that this recognition is behind the fact that every modern "TM" tool actually contains a host of other functions and features designed to support the translator in their work, ranging from electronic glossaries and functions integrating them with the TM system proper, to word processing productivity functions like AutoText and AutoCorrect, reference features such as "Concordance", "post-editing" functions like "Filter" or "Fold" (which display all instances of a given word or phrase in every file of the project being translated), and a whole host of semi-automatic quality control tools for checking numbers and terminology, as well as verifying tags. Recognising the need for close collaboration between project team members, many products also include a "Notes" or "Scratch pad" function to expedite this collaboration and enable doubts to be flagged up during the quality assurance stages.
In other words, modern "Translation Memory suites" have actually become Computer Aided Translation systems that have moved from a naive and limited notion of "perfect matches" to supporting the entire translation process. At the same time, there is abundant scope for these functions and systems to be extended further, building support for the above-mentioned feedback loops into the way they operate. A crucial element in taking the systems forward in this direction has already started to work its way through onto the market in some of the latest generation of tools, including SDL's recently-released Synergy project management system. The core enabling element here is the use of online translation memories, terminology databases and workgroup integration to enable this collaborative interaction in real time.
Taken as a whole, the major CAT and project management systems on the market already provide the bulk of the functions required to support this new model. So, although there's no CAT tool that currently provides everything we need, the solution is technically within our grasp.
|
|
First published in ITI Bulletin, 2005. |
