“You only get out what you put in” – My father used to mutter this phrase to me all the time back in the days when I was a competitive swimmer; moaning about yet another early get up, or the extra 100 lengths that had just been imposed on me to knock that extra second off my personal best. And unfortunately, the most frustrating thing was that he was right…and it’s a mantra that can be applied to many of life’s situations that we find ourselves in.
The quality of the input will always affect the quality of the output.
When building customised machine translation engines, we need to ensure that the corpora (data and assets going in) is clean, accurate and relevant – otherwise what chance does the engine have of outputting something decent? If a project requires Desktop Publishing, it’s so much simpler and cost-effective if the original file has been correctly formatted rather than bodged together. But perhaps one area we often overlook is the simple quality of the original text for translation.
Have you ever written your content with localisation in mind? Have you even read the text that has been passed over to you requesting translation? From experience, I think there are significant improvements we can all make here.
At Capita we have developed a ‘quality at source’ checking tool and you would be amazed at the results produced. We had one customer who had inconsistently written their own tag-line in over 50 different variations – at quick glance it looked ok – but the example below demonstrates how differently each of these key phrases had been written.
The cat sat on the mat
The Cat Sat On The Mat
The cat sat on the Mat
The cat sat on the mat!!!
Need I go on…?
And this is just one example of bad quality source language. I’ve also seen texts written by non-native speakers (god help anyone if I were to author a text in German or Spanish for translation), ambiguous terminology and meaning, sentences that go on and on for 5-6 lines, typos galore (spelling, grammar, even their own brand name), and finally completely wrong messaging. When I was a translator I even had an English text telling the user to ‘ensure that children are left unattended with the heated item’.
So you may be asking, ‘so what?’ (although I do hope not), but the effects on the localisation process are a plenty – and often very negative.
What if a translator just translated what they were asked to and conveyed this wrong message to your foreign market customers? Imagine the potential PR nightmare…
Poor quality and inconsistent source texts can also dramatically affect the cost of the translation process as the Translation Memory will not detect all the repetitions or matches that it should do – therefore not saving you pennies that are an easy win for your business. This also affects the consistency as well, and technology will not show that translator how they translated that same sentence previously – a difference in formatting, a double space, an extra capital letter – these all have a detrimental effect on both quality and cost. And then there is time – the translator won’t get the time efficiency gains from repeatedly having to translate from scratch, rather than just reviewing the translation for context suitability.
Finally, (well I’m sure there is much more to talk about on this subject), if a translator is first having to understand what the source text is trying to convey, there is a real danger of misunderstanding, and additional time requirements just to decipher what you have written in the first place. You may be amazed at how many times a translation is deemed as being bad quality by the customer, but investigation shows that it actually does accurately reflect the (poor) source language.
So – the message to take away from this – if you want a good quality, on time and cost effective localisation project – take the time and effort to ensure that the source text is of the same quality as you’d expect from the translation.
Proofreading can be applied to the original text as well, so don’t take any shortcuts. It really is true that you only get out what you put in.