Retirement plan for superannuated TMs OR Why bigger is not necessarily better

It’s all very well deriding the inadequacies of MT and how imperfect it is and how it can never compete with humans. But how come treat our TMs with such awe and reverence? TMs are human-generated. Not infallible therefore, and getting a bit decrepit and rather too large (and slow) as time goes on. So why assume that a long-lived, clunky TM is necessarily a good thing? One that needs to be preserved and cherished?

If you ask me, it is foolhardy or naive to consider TMs as the single source of truth, containing segments that can simply be recycled ad infinitum without question. Instead they ought to be treated with increasing caution and skepticism. They are somewhat antiquated and dusty, occasionally they may contain nuggets of gold perhaps, but lots of junk, that is for certain. Full of unhelpful duplications, contradictions, silly mistakes and inelegant phrases. And have you experienced the frustration of having to live with what’s in those TMs, even though it’s blatantly wrong. But the PM has given firm instructions not to change it, because the client wants to 100% matches to be reused, and there is no budget for spending time on them.

Rings a bell?

Or are you telling me you’ve been a diligent housekeeper and cleaner of the TMs you started around the turn of the millennium? That you can really stand by all your TMs? That you are confident they only contain validated, recyclable translations? All signed off by the senior linguist and, more importantly, the client? And that they are up-to-date (think orthographical reforms, gender-neutrality, formal/informal, modernization of terms).

The pandemic was the time we all de-cluttered: our wardrobes, our kitchen and under-the-stairs cupboards, our garden sheds. It was all subsumed under the heading “working from home”.

My proposal is for the entire industry to subject the TMs currently in use to that very same radical decluttering/discarding treatment. Many of them date back to some 20 years ago. Should that in itself not be suspect? How would you feel about a fresh start, perhaps next spring? Opening a new, blank TM? With the intention of keeping it nice and pristine?

Yes, I know how hard we all find it to get rid of beloved bits and pieces when we come upon them in the attic or the corner cabinet, reminding us of bygone days and happy memories. But translation segments, really? Can we afford to be sentimental about them?

Would you not agree with me that many of them segments have long reached the end of their useful life and should be allowed to go into honorary retirement? Just think: most clients expect something fresh and attractive and modern. And we all know that language changes over time, the tone in most cultures has become less formal, more conversational and user friendly. English terms have crept into languages like Italian and German, and it does not stop with terminology. Syntax has become more relaxed, comma rules are not the same, etc. People like a more conversational, less verbose style. More modern, in other words.

If you have learnt how to write a business letter in French at school you might remember the formulaic phrases at the end (“Veuillez recevoir, Monsieur/Madame, mes salutations distinguées” OR “… l’expression de ma haute considération”. Well, you would be ridiculed if you did that now. And while, clearly, this is a rather extreme example, other, more subtle changes have taken place over the past 20 years, not least also the recently introduced requirement for gender-neutral language. This forces you to scan every segment for potential gender bias and think of a different way of formulating it (no, it is not just a case of substituting masculine forms with masculine+feminine).

To conclude: 100% matches can no longer be treated as sacrosanct. It is no longer good practice to simply accept and recycle 100% matches. Time to turn over a new leaf. Because, if we don’t do it now, it will only get worse.

So, the old credo, the reason we all bought into CAT tools and translation memories, i.e. their huge benefits when it comes to consistency and efficiency, is decreasing with time. Quite a lot of the time now, I don’t want to be consistent with a piece of material that was translated in 2002.

Everyone in the translation industry has this in common: they are keen to hang on to their TM heritage, their treasure trove, the accumulation of all effort and know-how. It gives the clients huge leverage and therefore discounts, and the translators are under strict order – much of the time – not to change anything that’s already in the TM. Often, that’s done for self-preservation, because the budget just isn’t there. What is not nice is when the analysis shows 90% of the segments as exact matches, the quote has been accepted, and you then find yourself having to change and improve those “perfect” matches.

My advice is for everyone handling TMs, doing analyses, and talking to clients, to give this whole issue of superannuated TMs a bit of thought. If your TMs go back a very long way and linguists and/or the client tends to raise doubts about the quality, engage them in a conversation about decluttering an ancient TM, starting a new one, and using the old one for reference only.