Google Translate is an amazing translation engine. Along with Microsoft’s, Amazon’s, Deepl’s and other leading machine translation engines Google Translate produces better and better results both from a statistical and qualitative perspective.
Even though I operate a translation platform, the ease and simplicity with which I can translate a document within the GoogleDocs environment is very tempting to say the least.
Typically I am the better person and I resist that temptation and process the translations through Bureau Works, our translation management system. I make use of our customized terminology base and our translation memories even when I decide to translate myself.
But one Wednesday afternoon, I had a large Statement of Work that needed to get translated asap and that Google Translate temptation was greater than my urge to resist.
I hit the button and like magic, I had another version of the document instantly translated.
I had to fix certain mistranslations here and there but the overall quality was surprisingly good. After an hour or so of proofreading the document, I shipped it out to our clients and I was done.
Or so I thought.
The following week we made significant changes to our template in terms of styling and text edits. If I wanted to update the document:
- I would either have to go through the entire GoogleTranslate process again. Except now the button wasn’t anymore a source of instant gratification but one of regret. All my work, all my edits were lost. I would either have to spend a lot more time than I did originally to carry over all of edits over to the new version of the document or
- I would have to retrofit the previously translated and edited document into the new design and then implement the additional edits necessary. Both options were terrible.
I was hitting myself over the head.
Why did I not just go through the process of using our translation platform? Was it the haste? Was it the temptation of the easiest possible process? Or did I just underestimate what our technology does and how much value it does add?
I did not anticipate that I would have to update the document and I am guessing that most people that translate a document think they are translating it for a finite purpose and don’t treat it as a living document. And for most cases that may be fine, but when you do have a living document that changes over time, working within our environment with knowledge management functionalities is a must.
Had I processed the document through Bureau Works, I would have stored all of edits, all of my terminological preferences in the first run. When the second version came along all I would have to do would be reprocess the file against the knowledge base I had created and only focus on the sentences that had indeed changed. And the design would be for the most part also seamlessly carried over except for some formatting tags that I would have to address here or there.
But I did not do that, and I have not figured out how to turn back the clock. I chose to go through the process to ask my design team to re-design my originally translated file and then implemented manually the additional edits necessary. The sad thing is that we lost at least 20 work hours in this process that could have other been solved within a matter of minutes.
Here is regret #1: Forgetting about versioning and updates – they will bite
And I thought I was done with my problems but far from that. A closer look at my original translation revealed that I had made numerous inconsistent improvements that made the document look weak and less professional than originally intended. While our software ensures that repeated segments are flagged, and augments consistency through glossaries, GoogleTranslate treats the entire document as a flat document and does not offer these productivity-enhancing tools.
Which leads me to regret #2: Trying to do professional work without professional tools is not a good best practice and leads invariably to questionable results
And last but not least, I went crazy for a while trying to look for these inconsistencies. Lots of search and replace. Lots of scrolling up and down the document and it felt like a wack-a-mole. For everything I fixed it seemed new things were popping up. By that time I had already lost at least two more hours of my own time, had become frustrated and was still distant from the end result I wanted. It was clear that I would have benefitted so much more from using the right framework and tools from the beginning, And this is what I do. I build translation tools for a living. I believe in them to the bone and I still believed that this one time I could get away with a simpler more casual approach without many consequences. And I was kind of getting away with it until the versioning change ambushed me.
Which brings Regret #3: Do it the best way possible unless it’s literally impossible to do so
We are often tempted to solve things in the most straightforward, time and cost-effective way. But when it comes to language it’s not just that the words matter. They do. But much more than words, the framework matters a lot more. When you have the right framework in place you can handle surprise version changes, last-minute edits, peaks in multilingual document production relatively seamlessly while ensuring that your brand voice is consistent and intact. I know this. But I “forgot” about this for one instant and made one questionable decision that cost us thousands of dollars when you add up the time from all parties involved.
GoogleTranslate has my awe and admiration. The level of accuracy that they have managed to bring to machine translation thus far, the integration to their tools and the ease of use are astounding. But I learned it like I learn most things in life: the hard way. If you don’t have the right framework from the very beginning, even when you don’t think you need one, or especially when you think you don’t need one, that’s exactly when you can’t go without one.
July 13, 2022