"Entry barriers to the field of translation and interpreting are low. The main consequence is increasingly fierce and sometimes unfair competition, as well as a decrease in prices combined with a decline in quality levels. The new EN15038 certification designed to counteract this trend ... appears to require amendments."There are several issues here. One is the decrease in prices (price stability, i.e a fall in real terms has been the norm since the mid-90's) and the implementation of a desirable pan-European quality Translation Standard as a level playing field. EN15038's drafting was an extremely painful birth, as assumptions of translation quality and translation training among EU countries were (and still are) not uniform. Some countries consider translation a career and thus have University degrees. Others only offer MA. Some others neither. The Swedish committee argued for a long time that a 2nd linguist proof-reading should not apply to them as Swedish translators offered such a high quality for a start. Even after consensus, the need for "editing" and "proof-reading" were shocking points to many companies. The whole report seems to have been put together in a rush to meet the deadline and credibility has indeed been undermined by some obvious problems. There are some visual, basic desktop publishing issues with standard font usage. There is the wrong naming of ISO9001 as 2001 a few times across the document. There are even some odd conclusions, such as the low threshold barriers to entry to our industry that should result in merger and acquisition!
"the currently fragmented nature of the language industry will continue to consolidate into larger commercial entities".M&A activity in our industry is likely to continue, but until a company really comes up with a strong case and defensible barriers to entry, acquisitions cannot possibly result in consolidation - take Lionbridge's M&A of Bowne and their recent 25% drop in the markets as an example. If history teaches us a lesson is that only technology breakthroughs and their intelligent deployment can really make a difference. This "increasingly fierce and sometimes unfair competition, as well as a decrease in prices" brings us to the point of expectations vs quality vs price ratio. The following story (from one of our ex-PMs at Pangeanic) will ring some bells to many ears.
"After introduction, business talks and the signing of the formal SLA, the client sent me quite a rather large project, considering it was the first time we ever worked together. It was the documentation which had to accompany a famous scanner. I prepared the quote and warned about the poor state of the originals, which had been machine-translated without any post-editing at all. We warned about this, but our client got the contract from the manufacturer. The instructions from the manufacturer were to "follow strictly the previous reference material". This made all translators quite uneasy, as for the first time in their lives, they were requested to follow the style of raw MT output and the terminology set by non-natives.
The team found the job extremely uneasy and two linguists quitted after the first delivery. Unexpected additions to the job were accompanied by the warning of "non-compliance with the final delivery date will carry out a % penalty of the whole job".
Of course, the basic assumption and discussion point through the process had been "discounts and adequate pricing". Of course, had we had the chance to look at the materials before hand, we would have realized that the client had been cheated the first time round and that we had been asked in as the "repair guys" or the sacrificial lambs. We could not help laughing out loud when we saw the "reference materials" and the "reference TMs", but the fear grew when we were requested to use them and our calls for editing were ignored. Obviously, the end client had been toying with machine translation in an attempt to cut costs without any background preparation and not knowing what was to the done. We later learnt they had got together a few students to clean up the uncleanable and prepare the terminology databases.
The whole job was a catastrophe. The end client (a famous electronics manufacturer advertising their products on airline and computer magazines, for example) was furious and deeply disappointed. So were we. We reported that the initial translation was not only unusable, but they possibly had paid for something very embarrassing. Even if they refused to pay the first job we were asked to follow, they had lost precious time and therefore money".
Before and after the experience, we have heard and exchanged similar experiences with European firms subcontracting “native German and Portuguese” services for US$0,03, only to have to pay rush rates after having their fingers burnt. For multilingual content generators (hardware devices, engineering, marketing and service reports, healthcare), it is necessary to really define what their quality expectations are. Machine translation may be good for gisting, even for content generation, but its implementation without a proper deployment strategy may be disastrous. Simply requesting "best" or "perfect" quality will, sooner rather than later, lead to unacceptable quality without proper peripheral terminology, TM management and post-translation QA tools. Equally important is for the language service providers (LSP) to define their optimum service levels, technology approach and the value that they provide. The report has brought a lot of good discussion around quality and adequate pricing, which in turn calls for expectation levels and the professionalization of an industry with a high turn-around of employees and low threshold entry level. These discussions will help content generators to rethink their strategies when selecting appropriate translation suppliers, and add pressure on LSP to provide more added-value services.