Hybridity in Translation: Overgilding the lily?

Professor Andy Way, Director of Language Technology at Applied Language Solutions, argues that when it comes to MT products in the field, not all publicity is good publicity. He argues that new customers will be attracted by the quality of the tools we are producing, without needing to overgild the lily.

In a press release on 8th August, Science Applications International Corporation (SAIC) claimed to have produced the industry’s first hybrid machine translation (MT) solution. But if you go back as far as June 3rd 2009 you’ll find Systran claiming to have released “the company’s first hybrid machine translation server designed for enterprise use”. In a TAUS research report dated May 2009, Razheb Choudhury notes in relation to a debate regarding the pros and cons of rule-based MT (RBMT) vs. statistical MT (SMT) that “there is a trend towards providers developing hybrid solutions that aim to harness the benefits of both methods, such as by PROMT, Lucy Software and Services and Systran.” Clearly, then, industrial solutions using hybrid MT in some shape or form predate the recent SAIC press release by a couple of years or so.

Having crossed over from academia to the language industry, I am well aware of the pressure on companies to publicise their tools and services to attract clients. At ALS, we’ll soon be releasing our SmartMATE platform as having “the potential to vastly expand the user-base for self-serve SMT on a global basis”. That said, if we all claim that “we were the first to do this” or “the first to do that”, you have to wonder what the effect is on the clients businesses are trying to attract. For years, companies like TAUS or Common Sense Advisory have estimated the translation market as being worth several billions of dollars, yet the revenues of even the largest companies in our field amount to only a fraction of this sum.

As an academic, for over 20 years I’ve been educating students – many of whom now work in the industry – to not over-hype the capability of their companies’ tools and services. In his excellent Compendium of Translation Software, John Hutchins estimates that year on year, maybe 50% of translation companies ‘crash and burn’, attracted by the potential market size, but unable to meet the expectations of their customers. We must all act collectively to make realistic claims about what we can and cannot achieve, and to be aware when engaging in publicising recent developments in our respective companies to be aware of the rich history of developments in our field, and to ensure that claims are in synch with what has gone before, lest we alienate the very people we’re all trying to attract.

Especially now in the current global economic environment, more and more companies are looking towards automation as a solution to their increasing translation requirements. There is likely to be plenty of work for all of us, without at the same time impinging at all on the amount of human translation required; newcomers to the field will be looking for realistic expectations of what can be achieved, and their first interaction with language technology professionals is likely to be crucial. Over-hyping what we are capable of causes us to quickly lose credibility as a whole. Our translation tools are already ‘good enough’ to save clients money, or improve their throughput; there are a huge number of success stories demonstrating this already. Publicity is good, but in our field, not all publicity is good publicity.

New customers will be happy enough with the quality of the available tools, without the need for businesses to overgild the lily.

This is a guest blog written by Professor Andy Way, Director of Language Technology at Applied Language Solutions.

2 thoughts on “Hybridity in Translation: Overgilding the lily?”

  1. I totally agree with you. It is really hard to work as a translation company if you don’t meet your clients’ expectations. MT has been quite basic in the past and could not make a difference between the multiple meanings of a word. It seems that hybrid MT is a bit better and will help us translators meeting our deadlines while providing great quality translations in less time.

  2. I agree with most of what Andy says. Expectations are very high and they keep getting higher. My opinion is that MT uptake depends greatly on translator willingness to accept MT output and the change of roles. There is still plenty of negative feeling against MT among the freelancer community upon which LSPs depend. Whatever direction MT takes, there is a lot of evangelization and convincing to do with real-life scenarios and engaging translators.

    Just one “token” I received from Kirti Vashee’s blog: http://www.translationtribulations.com/2011/09/future-is-here-artful-manipulation.html

    In short, I agree with the not “over-selling” point – however difficult this may be in a commercial setting. But furthermore, little can be achieved if a service does not engage the ultimate users. The great success of CAT tools is that they actually managed to covince the translation community they were essential and productive, to the point that most translation studies at Universities include CAT modules.

    MT still needs to do that.

    “companies like TAUS or Common Sense Advisory” –> they are member-based organisations, not companies 🙂

Comments are closed.