Quality Assessment In Translation

1222 Words5 Pages
1.Introduction With the development of the technology, the language and the way we render it as translators have been changed. Technology gives us speed, quality and ease while translating thanks to the newly-developed computer-aided translation tools. However, these CAT tools can not meet all the needs and requirements of all translators. The perfect CAT tool should have availability, interoperability, cost-effectiveness and multi-dimensional quality assessment features in one unit. Especially, in today’s world, where everyone is in a hurry, being fast is a vital point. However, speed does not always bring quality. On the contrary, according to quality evaluations, speed and quality are inversely proportional. Therefore, the aim of the perfect…show more content…
There are still some discussions about whether we can measure the quality of translation or not. Most of the translators believe that we can measure the quality of a translation but not directly. As the meaning of “quality” can be different to each people, we can compromise on what the “error” means. Therefore, we evaluate a translation by calculating the errors in it. There are several quality assessment measurements in the industry. LISA, EN 15038, BLEU, SAE J2450, TMS Classic QA Model, TAUS Dynamic Quality Framework are some of the translation quality assessment metrics. The computer-aided translation tools should have quality assessment features for a general standard for both human and machine translations. Most of the CAT tools are supporting QA feature nowadays but we still have to check our translation several times with different quality assessment tools. There should be one multi-dimensional quality assessment metric plug-in for all computer-aided translation tools. That QA metric should be evaluate both translation and source text with the grammar and syntax of the target language taken into consideration. For example; when a QA tool scans a translation which is from English to Turkish for key term mismatches, it can deliver many false-positive errors because Turkish is an agglutinating language. There is lots of assimilations or lenitions in Turkish. When there is a key term input for the…show more content…
Quality Assessment in Machine Translation As it is known, machine translations are getting popular everyday because they are time-savers and they are getting quite accurate. Therefore, controlling a machine translation is much simpler than whole human translation process. There are several quality assessment metrics for machine translation. BLEU, NIST, Word Error Rate, METEOR are some examples of these metrics. Most of these metrics are based on similar algorithm. BLEU machine translation evaluation metric evaluates the translation on a scale from 0 to 1, yet, it generally seen as percentage. The more the score closes to 1, the more the translation is parallel with the human translation. BLEU calculates how many words are overlap with the human translation of the reference text just like the other QA metric for machine translations. The reason behind the BLEU is that the human evaluation can take a long period of time and the imperfection of the human nature such as oversight or being tired or being in a bad mood. Like BLEU, the main principle of these type of quality assessment metrics should be saving time while maintaining the quality of the

More about Quality Assessment In Translation

Open Document