This is a guest post by Suzanne Rietveld (suzarieve@gmail.com). Suzanne translates between Dutch and English; she has a background in horticulture and has worked in women’s health—these are her primary translation specializations. She is also an expert reviser and editor and is available for work in these areas.
As a beginning translator, it is quite difficult to decide which systems work and which are not helpful at all. During my Bachelor’s degree program in translation at the Institute for Interpreting and Translation in Utrecht (Netherlands), we had not been introduced to many CAT tools or other digital options at all. We all had to do an introductory course in the CAT tool Deja Vu, (DVX3) and learned how to set up a project, how to import a document, how to confirm segments, and if you were lucky, you managed to export the document as well. It took a lot of time to find the answers to my own questions, and I found out that that is not the way to learn a new system. I managed to acquire the how-to documents which were used with Deja Vu, and that is how I started to use CAT tools.
During my course, I managed to find time to start marketing for paid work as a translator. I really like the challenges that we are faced with. I registered with my first Dutch translation agency and passed their test to translate from Dutch into English, which is my preferred language combination. In the meantime, we still had to work on all school assignments; multiple translations into the two languages. I soon found out that my memory is not always up to scratch and I do not mean the translation memory in my CAT tool. It really takes me some time to think about the right word in the right place. I have always been aware of this—and there’s no chance I will ever become an interpreter—but I needed to address it in my translation business.
So, I was really focused on getting some help from machine translation and tried SmartCat for a while. I was very happy with the results because it helped me with vocabulary. In SmartCat, the source text will be translated for you with machine translation and the target text is filled in for you from the start. I found out that this had some downsides for me. With a full segment translated, the job for me as a translator had been reduced to the job of a post-editor. I am quite fond of editing – I had quite some experience as a Dutch editor-in-chief – and I did get the hang of it, but I did find out that when using SmartCat, I did not get the results that I expected, or the quality of translation that I would have translated from scratch.
That was the moment that I came across the translation tool Lilt. It is quite similar to SmartCat as a translation tool, with the biggest difference being that you will get a translation of your source text, but this appears as a suggestion and not as your target text, which you still have to type from scratch. If you choose the same word as in the Lilt suggestion, you can accept the choice by clicking Enter. If you make changes to the structure of the sentence or choose different words, the Lilt suggestion changes on the go. The system is learning while you go and that is quite apparent.
Because I was eager to use a lot of my own terminology, I started to use the translation memory options right from the start. I managed to align most of my previous translations (including a book on prostitutes in Amsterdam, a website on leadership skills and team building, and my school assignments) and uploaded them to my TM. To use your own TMs in Lilt, you have to become a PRO member, which starts at US $15 per month. With all the school work and paid jobs, this has been well worth it. I managed to translate my assignments quicker and was more confident that I used the right terminology.
It is quite easy to import different documents in Lilt, and this works great. The only thing is that documents should not be open on your screen and saved as docx in Word, but I also managed to import Excel sheets, PowerPoint slides and PDFs. Once the translation is completed, it is again very easy to export the document and save it in the place that you want. At the end of my third year in my program, I had found myself a place to gain some work experience and started to work as a translator for a publisher in floriculture. I translated 30,000 words in three months and used Lilt throughout the entire project without any regrets. In hindsight, I should have let the reviser make changes in Lilt as well. Now, I exported all my texts a little too early on in the process and still made many changes during my own revision round and afterward I accepted the reviser’s changes, also in Word. To achieve the best results, I guess the trick is to stay in your CAT tool as long as possible, but many of you probably learned this the hard way as well.
The support team at Lilt is great. There is a chat option in the right corner, and I have made use of it every now and then. The team is very helpful, and it is evident that they want to become bigger and better all the time.
There is one thing that they still have to get right: the use of capital letters and punctuation marks. In the current version of Lilt, all commas, semi-colons and periods are anchored to the target text. If you choose to use the last word (with the period attached to it) earlier in the sentence, you have to delete the period manually. This is okay-ish, but it seems to happen quite often, and with all the different types of punctuation marks, so it gets quite annoying at some point.
So, I have used Lilt during my work experience, for school assignments, and for some private customers. I trained as a horticulturist and garden designer in the past and would like to specialize as a translator in these subjects. So, I convinced myself to get some experience and translate a lot more documents on this subject into the two languages. Lilt remains a great tool to use and it is good to see how quickly it is learning and improving.
I am now in the position of having to make choices. I guess I should try my CAT tool Deja Vu again and use it with a plug-in from DeepL or something similar. At the moment, I do the odd translation job for different translation agencies. I found out that the ones outside of the Netherlands only let me translate into my mother language, which is Dutch. I do work in their web editors and do miss the help of machine translation, but during my program, it has been better for me to keep practicing without machine translation. I have stopped using Lilt for now, also because of the monthly fee and the fact that I haven’t had much use for it over the past few months.
As I said before, I am looking for the next project which will offer me the opportunity to try a CAT tool with a plugin, and I will give you all an update if this appears to be a better choice. It would be great if you can leave some feedback for me with information about other options.
Very interesting and helpful piece. For those of us who are just barely considering CAT tools: what is a CAT tool with a plugin? How does that work? And another question from the Stone Age: when you work for different agencies, do you encounter compatibility issues between different CAT tools?
Thanks again,
M
Thanks for your comment, and glad you enjoyed the piece. 1) CAT tool with a plugin means that you would run a CAT tool like memoQ, Trados, Wordfast (I put those in alphabetical order so as not to prefer one over the other!), and then as an add-on, you would “plug in” a machine translation engine, meaning that the CAT tool and the MT engine are connected. In most market-leader CAT tools, if you choose that option, you can tell the CAT tool “if you don’t find a match in the translation memory for this segment, show me a machine translation of it.” Personally I do not do that, but it’s an option. 2) Yes, if you work primarily or exclusively for agencies, you’re likely to encounter compatibility issues. You may need to purchase different CAT tools for different agencies. In theory, they are all compatible through formats like XLIFF and TMX, but in reality…not so much.
That SmartCat looks very interesting. I hope to find out a little more about that. Interesting post, thanks.
Thank you for this post. I would like to look into Lilt and see what it’s like.
I have also tried MT API with CAT Tools in the past, and was not pleased with my performance or the resulting translation either. So the idea of having the MT as a proposal, not the starting point, is exactly my style when translating.
I found the guest blog posting quite illustrative of what goes on in some corners of our profession. I’m glad she started with Déjà Vu X3, my go-to TEnT tool (I’ve worked with SDL, Trados, SDL Trados Studio, Lingotek and memoQ). I might take a look at Lilt to see how it handles specialized files such as FrameMaker’s .mif files, which are notoriously tricky to work with.
Great article, Thanks a lot for sharing such a kind of informative article. these tips will help me so much!!!
Very interesting article! Thank you!
However, the info about SmartCat it’s not really accurate.
SmartCat takes the text from the MT automatically ONLY if you choose this option.
Otherwise, the source language segments appear on the left side and the empty (target) segments on the right. The MT it’s shown on the far right (depending on how you choose to view the editor) and then you can copy/paste it into the target segment or not.
Moreover, when you start a new project you even have the option to take MT out completely – you just need to tick one box and then you won’t have MT showing up anywhere.
I have been using SmartCat for a while now and find it really easy and user-friendly. I don’t rely much on MT, but sometimes it helps to have some suggestions there. Nevertheless, I would never choose the option to automatically populate the segments with the MT.
Hope this info helps! Cheers! 🙂
Thanks very much!