Curva Fin Bloque
POST AGOSTO 24, 2010

Final dominance by final technology? The beginning of the MT wars

It is only a few weeks since one of the major acquisitions in our industry in recent years  (MT technology provider Language Weaver by SDL) but several trends already point in the same direction i.e. that whichever language company wants to emerge a winner or grow, it needs to include its own MT technology within its array of services. In many cases, within the buyers’ new economic mindsets, it is  becoming the most important asset and the real assurance of “cost saving”.

If web2-based management programs became the flavour of the year in the run-up to 2010 just as TM-leveraging was in the 90’scustomized MT solutions are now a must-have for any decently sized LSP. Recent reports about technology adoption by LSP’s from Common Sense Advisory (stressing how those lagging behind in technology adoption will pay very dearly) and the agreement reached by 4 companies to provide a joint service offer add to this vision. We have been witness to European LSP’s networks of collaboration in the past, particularly when tendering for large contracts but usually no higher form developed from there. It remains to be seen how a combination of 4 independent companies will work when conflicting interests are at stake, for example supplying to the members of the agreement and a competitor thereof. The manufacturing, clothing, and engineering industries –not to mention banking have plenty of such stories.

Owing a technology that is already revolutionizing the way we work may partly explain the [comparatively for this industry] high value SDL paid for Language Weaver (almost 4 times its turnover and running a $1M loss), but keeping the company management and research team intact. SDL was not acquiring another business or buying a particular, specific technology as it did with Passolo or Tridion in the past, but rather the key and dominant technology in years to come (statistical automatic translation and hybrids). After all, it will be the technology that will produce the words that SDL will sell. Shockingly high as the valuation may have seemed initially, it makes perfect sense from a defensive and an aggressive point of view. It signals the beginning of the “MT wars”. 

Strategically, the valuation and acquisition served as clear statements for it: 

  1. It signaled an acknowledgement that SDL’s own automatic translation technologies had become antiquated when compared with the flexibility of statistical methods or combination hybrids. There was not an ounce of interest in third-party rule-based or advanced leveraging technologies.
  2. It acknowledged the fact that despite wide psychological acceptance by the buyers of translation services and increasing interest by the potential users (translators turned post-editors?), there is still plenty of room for development and opportunities (and thus revenue) from MT. Most companies involved in statistical automatic translation and research have not reached full maturity yet, but translation output demands, need for immediacy and language combinations keep on growing, whilst price competitiveness, ironically for the wanted commodity, is very high. Many are purely research companies, some lack marketing or in PangeaMT’s case, technological spin-offs with a clear vision born out of the need to automate.  I mention statistical because it is the technology which has proved most successful in proving MT works satisfactorily, after decades of rule-based monopolistic dominance.
  3. Savings in the short and not too long medium-term from a vast, intensive application of automatic translation will be beyond the $42M paid – perhaps also developing client-end proprietary engines or providing them under contract terms (institutions, government, and also, why not, industry). The investment will soon be amortized with lower vendor bills.
  4. It also signals that provided you have the technological resources and will, it is essential to own, develop or at least customise your MT solution for your clients. Pressure will mount on the smaller and medium-sized LSP’s, although regional/single-language companies may be able to recycle faster in the vendor cycle.
  5. As much as it has done to shake the market and demonstrate the power of statistical methods beyond academic circles, Google cannot be the development partner. You can reach agreements for bulk use, plug-ins, but sooner or later, you will need to customize (therefore, d).

Are we facing the fact that MT technology companies cannot «live alone»? Is this the beginning of a new «integration» phase?

Next time you think languages, think Pangeanic

Follow manuelhrrnz on TwitterFollow us on