In a press release issued yesterday, iParadigms, the makers of both Turnitin and iThenticate plagiarism detection services, announced that they have introduced a new feature to those products, the ability to detect translated plagiarism.
According to the release, Turnitin is now “Able to take assignments written in a variety of non-English languages, translate them into English, compare them to Turnitin’s massive content databases, and highlight any matches found in the assignment.” This is specifically designed to combat plagiarism by students who do their primary work in languages other than English but often translate previously-created English content into their native language and submit it as original work.
Translated plagiarism has, historically, been a major problem for plagiarism checking services. Since all plagiarism detection services use some form of fingerprinting, which is designed to detect exact matches, translated plagiarism, which is often imprecise due to the multiple ways a work can be translated, has remained difficult to detect.
With this change, iParadigms is hoping to change that.
However, it remains to be seen how effective this system is, which is still in beta, will be. One thing that is clear is that it is attempting one of the most monstrous challenges in plagiarism detection and, even if it is only partially successful, it could be a huge step forward for the industry.
International Audiences Only
Right off the bat, the new service has one fairly significant limitation, namely that it is focused almost exclusively on international audiences. In fact, the current version is only available in the following languages: French, Spanish, German, Dutch, Swedish, Portuguese and Turkish. Also, the translated matching also works on submissions Czech, Danish, Finnish, Greek, Hungarian, Italian, Polish and Romanian.
In short, it can’t be used right now by institutions whose native language is English so, if a student in a U.S. university translates a Spanish-language work into English and turns that in, the system won’t detect it, at least not yet.
However, considering that most translated plagiarism starts with works in English that are ported over to other languages, that limitation is a fairly small one. This is especially true considering the lion’s share of iParadigm’s database is still in English.
Still, there’s little doubt that translated plagiarism is on the rise in all languages, English as well, so hopefully that will be a change on the horizon.
Waiting for Effectiveness Tests
The real test of this service will come as others begin to use and iParadigms works to improve it. While it almost certainly won’t be as accurate as detecting plagiarism and copying in the same language, if it can provide even modest results, it will be a step forward.
One particular set of tests I will be interested in is Dr. Debora Weber-Wulff’s tests of the effectiveness of this service. As part of her regular examinations of plagiarism services, Dr. Weber-Wulff tests whether or not they can detect translated plagiarism, a test every system has failed.
Since Dr. Weber-Wulff’s tests focus on the German language, one of the languages provided, it maybe be an excellent opportunity to test and see how well this service is performing. At the very least, it will be a good comparison of the performance of the service versus its performance in years past.
Until then, the effectiveness of this service outside of the lab is going to be the subject of a lot of questioning and scrutiny.
Obviously, if there were a breakthrough that enabled quick, effective and simple detection of translated plagiarism, it would be a huge step forward for plagiarism detection, closing one of the biggest “blind spots” in modern systems. However, given the way modern plagiarism checkers work and the flexible nature of translated plagiarism, it’s unlikely any such system work as well on translated plagiarism as they do verbatim plagiarism.
Still, any attempt to crack this problem and any progress in solving it is progress. Even if it only catches a small percentage of translated plagiarism, it’s still more than we were catching before, at least by these means.
In the end, I find myself cautiously optimistic about this new feature and looking forward to learning more about it and how effective it is down the road.