Making plans for the future
At the beginning of a new year, many people take a look at the past and make plans for the future. Luckily, I would say, most of these plans are never finished. On an enterprise level, the situation is similar, although usually centred more on economy than emotions. I have penned these lines about the emotions in the language industry as we experience a period of change and uncertainty.
Our industry is evolving, but not fast enough to accommodate current and new demands. Content is being generated at an incredibly fast pace, in particular on social networks. And most of that content is not being translated. Some translation companies are targeting short text translation instead of the traditionally larger projects with a start and end date. But they are certainly covering only a fraction of the information available. Reports show us that users demand content and like to read and buy in their own language, even though they might speak the source language. Only a fraction of web sites are available in more than one or even two languages. The need for linguistic services in areas like medical, entertainment, immigration and legal (to name a few) are potentially much higher than the number of current translators, interpreters and language service providers can cover.
Statistical machine translation (SMT) and its fast domain and language adaptation has brought hope to many companies and NGOs that language barriers to written language could finally be overcome in the near future. It has also brought fear and frustration to many translators as translation agencies turn to MT, imagining that they could lose their jobs, that their income may be reduced, that the quality of their work might be negatively affected and that their influence on the final results of the translation may be highly reduced. Part of this has happened, but on a much lower scale as expected and not exactly for the reasons originally argued. SMT is following a similar emotional curve to Rule-based MT before it, going from high assumptions to deception, as users and developers realise that it cannot fulfil the promise of quality instant automatic translation under most scenarios and language pairs. However, if your goals are realistic and you define your target audience expectations, SMT, hybrid MT and hybrid CAT-MT configurations properly setup and maintained can deliver surprisingly good results under certain circumstances, and may allow faster delivery and higher quality translations when combined with well-trained smart post-editors.
On the subject of quality, we are seeing exciting initiatives and studies to tackle the very complex notion of translation quality. As an example we have the QT Launchpad project, TAUS dynamic quality framework and the Quality Triangle Methodology, besides ongoing norm developments such as ISO17100 (if it ever sees the light). The problem I spot in most approaches to quality in our industry and what disappoints me most is that we seem to be more worried about how we measure quality as an isolated element, instead of how we identify and deliver the different levels of quality that the different customers require. The notion of placing the customer at the centre of the quality efforts is not new at all as you may expect. We can certainly learn from other industries or companies that have already been in our dilemma and have successfully applied methodologies like QFD (Quality Framework Deployment) and ethically implemented Corporate Social Responsibility in order to have happy stakeholders as well as sustainable enterprises. Have you ever wondered who the oldest companies in the world are, and why they have subsisted?
Outside of our industry, big data and machine-to-machine interactions are leading to a new type of researcher and the rise of analytics, but mainly for a single language. A very poor understanding of the world. And cross language information retrieval cannot provide the full answer even if it were extensively available.
I am amazed at the speed at which we are seeing all kind of new devices show up every year. New ways of interacting with these devices will allow professional translators and all kind of potential users to take better advantage of existing technologies such as voice recognition, large touch screens, eye tracking, and many other developments to increase current translation performance and make our work easier. Our wish may be to have that in place already, but we are not there yet. On the more distant horizon, all this might be replaced by direct brain to computer interaction (BCI) combined with other types of input to correct errors or to solve difficult formatting issues, which will be perhaps a much more natural way to translate.
Regarding software, we have more options than ever and the technologies are converging. Online tools, offline tools, SaaS options, smart technological platforms. It is all in the mix and this is encouraging. However, most translation tool designers and software reviewers seem to be particularly interested in the feature battle and perhaps prefer small increments in productivity instead of improving the overall user experience and including really disruptive technologies. Easier, perhaps, but the wrong way to go, in my opinion. A recipe for disappointment. There are advances in integration but are there any really significant advances in translation and localisation in the past few years compared to the changes in other industries
Despite all these technological advances, human translation and human interaction in language projects is and will continue to be a key central point for many years to come. Enjoy the future.
Please use the form opposite to get in touch
or alternatively give us a call us on:
+44 (0)1869 255820
Bignell Park Barns