CAT CORNER: MEMSOURCE

 

By Daria Toropchyn
Recommended minimum configuration for reliable performance:
CPU: Intel Core 2 Duo 1.66 GHz
RAM: 1 GB
Bandwidth: 1 Mbit/s
For pricing, see www.memsource.com/pricing

 
 
Reprinted with gracious permission from SlavFile, the Slavic language division newsletter, and the author

There are continuing debates out there over whether machine translation will ever replace human translators. Personally, I don’t see this ever happening, but there are a few by-products of artificial intelligence development that translators can benefit from in the here and now.

As a student and translator, I have had the opportunity to try various translation tools as part of my academic program. This has included guided access to Computer-Aided Translation (CAT) tools, which I can use for various projects without the pressure of having a client to satisfy.

The two main features of CAT tools that translators find most helpful are translation memory and a terminology base. A translation memory allows users to save translated segments (in most cases, sentences) and reuse them later for future projects. A terminology
base helps translators keep terminology consistent throughout an entire project.

According to ProZ, the most the most popular CAT tools today are SDL Trados, Wordfast, memoQ, Déjà Vu, Across, SDLX, and OmegaT. I would like to discuss my own experience using MemSource. This tool is not as widely known as the SDL suite, for example, but in my opinion, MemSource is a great CAT tool for certain types of work. My working language pairs are English>Russian and Czech>Russian, so I will be reviewing this tool from the viewpoint of these languages.

MemSource was launched in 2010 in Prague, Czech Republic, by the company’s founder, David Canek. The CAT tool was released for public use in 2011, and by the end of 2015 it reported 50,000 users (SDL Trados, by comparison, has 225,000 users). MemSource has both cloud- and desktop-based software versions. That means you can install it on your computer (the files will only be available there) or use the cloud-based version and have access to the files anywhere from any browser on any device, as long as you have an Internet connection. I use a cloud-based version, and I’m glad I chose that option. (Note, however, that many clients have confidentiality concerns and won’t allow you to work via a cloud-based system.)

MemSource MemSource is available as a Personal Edition (free for up to two files for translation of 10 MB each at one time) or Freelance Edition (free for 30 days, then $27 per month, can be used by multiple users, allows for unlimited files, and includes translation memory and a terminology base).

I have been using MemSource for a year now both for school and for work. MemSource has become my primary CAT tool.It supports over 50 file formats for translation. Of those, the ones I am most familiar with are .doc, .ppt, .txt, and .srt (for subtitling). Here is what the Editor looks like in the browser, where all the work is actually done:

The Editor window is very user-friendly and clean. If you know how to use Microsoft Word, you won’t have any trouble using MemSource’s Editor window. MemSource also provides short video tutorials that explain the Editor and pretty much everything else this CAT tool has to offer.

MemSource has a couple of great features that I haven’t found in other CAT tools:

Halfway through a lengthy document (over 1,000 segments long), I realized that I had already translated a similar sentence. In the top left area, “Filter Source= Text,” I typed a few words that were definitely translated previously, and MemSource gave me six results where these words were used. So, to the right of the sentence, I typed in the translation. But how could I get back to where I was before? I could either erase my filter criteria and scroll down all the way to segment #602 (where I was last), or I could leave my cursor on the filtered sentence and delete my search parameters. In the latter case, MemSource automatically brings me back to segment #602, and I can carry on translating. This sounds like a non-critical feature. But if you don’t know it is there, you can waste a great deal of time scrolling up and down.

While translating, I try not to confirm segments (“confirming” the segment means saving it in the translation memory). Because my translation memory is already big enough and has similar sentences saved in it, in the top right corner MemSource shows me source and target sentences that can be helpful. These subsegment matches are marked with pink. While they are not 100% matches, most of the time they open up new possibilities or just inspire me to continue the search for “the right word.”

The green “101” means that MemSource has found a 100% match in my translation memory; the blue “MT”  offers a machine-translated variant; and the pink “S” is a subsegment feature that recognizes a repetition within the segment and offers a translation. (The A refers to an outline heading in this case.)

Of course, MemSource is not perfect. Here are a few features that need improving or might be a reason to wonder if MemSource would be the best CAT tool for you:

  • MemSource does not replace the & symbol with its translation in an MT-offered sentence. This is not a big deal, but when you work in the English>Russian language pair, you can waste significant time just replacing this symbol (in 100% of cases, the “&” does not change between languages). As you can see, though, MemSource does offer a subsegment version of the translation, since it has “learned” that I replace “&” with “и” in my translations all the time.
  • Once I had a Czech docx file to translate into Russian, and MemSource produced so many tags (indicators of a recognition problem) that I was forced to translate the old-fashioned way, without using any CAT tool. However, this was the only Czech file of the more than 20 I have translated that caused MemSource any difficulty.

Overall I feel that MemSource is a great tool for translators who work mostly with files such as .doc, .txt, and .ppt. I like the quality of the machine-translation suggestions, which are easy to work with. And when general topics are translated, the majority of suggestions do not need serious editing. I have been able to add and edit terminology “on the go” without the need to upload a new termbase every time I have to add or edit a term. For those translators who work while traveling, the cloud-based version is a highly
useful option and would save the day if the translator’s computer died when there was a tight deadline and no backup, since it allows the use of a borrowed, Internet café computer or even a publicly available computer. How awesome is that? And $27 per month (totaling $324 per year) for a Freelance Edition is relatively affordable, without the need to pay a significant amount of money up front (such as the $825 purchase price for SDL Trados). You can access your document from any computer or laptop with any operating system (Microsoft, Apple) anywhere in the world at any time. Isn’t this the kind of freedom translators dream of?

Written by Daria Toropchyn

Leave a Reply